AEA Poster Session
Poster Session
Saturday, Jan. 2, 2027 1:30 PM - 9:00 PM (EST)
Sunday, Jan. 3, 2027 7:30 AM - 6:00 PM (EST)
Monday, Jan. 4, 2027 7:30 AM - 6:00 PM (EST)
Tuesday, Jan. 5, 2027 7:30 AM - 1:00 PM (EST)
A Model of Currency Crises under a Flexible Exchange Rate Regime (F3, F4)
Abstract
This paper develops a novel model of currency crises for developing economies operating under a flexible exchange rate regime. While there is no commitment to commit exchange rate reserves to defend an exchange rate peg in that case, foreign reserves can be insufficient to purchase imports and service foreign currency debt. The crisis is set off by excessive depreciation, which is possible with a flexible exchange rate, and is a feature which isn't present in most previous models of currency crises. There are two policies that governments can implement to delay or prevent a crisis. The first is to run budget surpluses, which will reduce imports and improve the current account. The second is for its central bank to follow the Taylor Principle, raising real interest rates in response to an inflationary shock. We assume that domestic currency public debt is only held by domestic residents, and so lower real interest rates lead to a portfolio allocation to foreign currency assets, which makes the government unable to issue domestic currency debt, which then forces the issuance of more foreign currency debt, which then must be serviced out of foreign reserves. We discuss the case of the 2022 currency crisis in Ghana as an archetype of this type of crisis, as well as other countries which have suffered similar crises recently. We also discuss the case of other countries, like Brazil, which have avoided similar crises by following policies consistent with what our model would predict. Closed-form solutions and phase diagrams are used to show the effects of budget surpluses, a more hawkish monetary policy, and other policy changes.A New Model of Trend Inflation Using Disaggregates, Survey Expectations, and Uncertainty (E3, C3)
Abstract
This paper develops a new empirical model that estimates trend inflation by combining modeling features that have advanced the literature on trend inflation over the past two decades. These features include incorporating information about long-term inflation expectations from surveys in a flexible way, modeling aggregate inflation via sectoral data (goods and services), allowing for stochastic volatility (SV) in the shocks to the trend and transitory components of inflation, allowing for a time-varying price Phillips curve, and allowing for time-varying uncertainty effects on the level of inflation. We estimate the model using state-of-the-art Bayesian methods. We document the competitive properties of the new model compared to variants that include only a subset of the above features. The new model provides a more interpretable historical decomposition of inflation data than the models it extends. The decomposition suggests that uncertainty effects play a greater role than cyclical effects in explaining inflation fluctuations.A Pinch of Salt, Measurable Costs: The Socioeconomic Effects of Soil Salinity in India (O1, Q5)
Abstract
Soil salinization is a gradual but underexplored form of land degradation with growing implications for agricultural activity, household livelihoods, and economic well-being. This paper provides new evidence on its socioeconomic consequences in India by combining nationally representative panel data from the India Human Development Survey (IHDS) with satellite-based measures of soil salinity. Using household- and individual-level fixed-effects models, I exploit within-unit variation over time to estimate the impact of changes in salinity exposure on agricultural income and key adjustment margins: labor supply, consumption, and human capital, while controlling for seasonal climate conditions and household characteristics.The results show that rising salinity significantly reduces rural agricultural earnings, primarily through declines in crop income. While aggregate labor participation remains broadly stable, responses are heterogeneous: rural women experience a decline relative to men, consistent with gender-differentiated constraints in labor reallocation. Despite income losses, total household consumption appears largely smoothed, but granular evidence at the food-item level reveals reduced reliance on common salt-sensitive crops.
This apparent short-run resilience masks important effects on human capital accumulation. School attainment declines for children of compulsory age in rural areas, with effects concentrated among girls. These patterns suggest that salinity-induced income shocks lead households to adjust educational investments along gender lines, widening existing gender gaps, with potential implications for future development.
By linking large-scale household data with spatially detailed environmental indicators, this paper highlights how gradual land degradation affects income, household welfare, and human capital outcomes in agrarian economies.
A Projection-based Control Function Approach for Reducing Aggregation Bias in Demand System Models (C1, C4)
Abstract
In both academic research and industry practice, demand models are widely used to evaluate subsidy effects and analyze taxation impacts through estimated price elasticities. Theoretically, simulations using demand models should incorporate every independent product affected by the policy, typically represented by Universal Product Codes (UPCs). However, a significant empirical challenge arises due to dimensionality. For instance, a single grocery store can offer over 200 UPCs for sugary beverages alone, but most demand models cannot converge when handling more than 50 products, limiting the applicability of demand models for policy simulations. To address this issue, the current approach is to carry out demand analysis at some level of aggregation across commodities, validated by aggregation tests, such as the Hicks-Leontief Composite Commodity Theorem (CCT) and the Generalized Composite Commodity Theorem (GCCT) Tests. However, these tests often fail, prompting practitioners to aggregate products under the assumption that approximate estimates are better than none, despite resulting in biased elasticity estimates. Our research introduces a projection-based control function approach to reduce bias in demand models when aggregation tests are not satisfied. Using simulated consumption data for theoretically non-aggregable elementary products, we compare estimated price elasticities from ground-true values, direct aggregation, and our proposed control function method. Our findings indicate that the control function approach reduces bias in elasticity estimates by an average of 56% compared to direct aggregation, providing more accurate policy simulations and elasticity estimates closer to true values.A Promise in the Name of God (O0)
Abstract
How do sellers signal product quality in weak contractual environments? Using hundreds of audio-recorded trade negotiations in Afghanistan, we document that sellers often invoke God to reassure buyers, particularly when product quality is difficult to verify and when they lack a fixed selling location, limiting reliance on reputation. We argue that these divine invocations function as a signaling device, as making a false claim under them is believed to entail cosmic costs. Evidence from field experiments matches detailed predictions of a signaling model. Our findings indicate that religious beliefs can be actively employed to strengthen the credibility of verbal commitments.Aggregation Bias in Quantitative Trade and Spatial Economics (F1, R1)
Abstract
We characterize the bias from data aggregation in quantitative trade models that use exact hat algebra. We derive six results: a zero-bias condition requiring identical within-aggregate primitives; an exact decomposition into trade share and elasticity channels; I-O amplification through the Leontief inverse; a three-layer GE decomposition where wage feedback dominates; an upper bound with zero violations across 155 checks; and sufficient conditions for safe aggregation. Application to the Caliendo and Parro (2015) NAFTA counterfactual reveals 58-65% welfare sign reversals, driven primarily by baseline parameter inconsistency in naively aggregated models. Spatial aggregation of the China shock studied by Caliendo, Dvorkin, and Parro (2019) produces sign reversals. Random concordance analysis confirms the bias is structural. Bias correction from aggregate data alone proves infeasible.An Estimate of the Infrastructure Multiplier in the Global South (E6, F3)
Abstract
We estimate the macroeconomic effects of public infrastructure investmentin 119 emerging and developing countries. For identification, we exploit
shifts in aggregate investment credit within China’s Going Global Strategy,
and historical sensitivities of the debtor countries to these shifts. We find
that one US-dollar of additional infrastructure investment has modest short-
run effects on output, with a cumulative multiplier of 0.6, but substantial
medium-run effects, with a multiplier around 2 five years after the shock.
We document a crowding-in of private consumption, and that the less
developed and more closed economies in the sample have larger multipliers.
Aspiration-Weighted Influence (D9, D8)
Abstract
Celebrities, experts, and wealthier peers publicly display consumption over a broad range of alternatives, many of which their aspiring followers cannot afford. Existing models of social influence have focused on settings where the influencer and the decision maker (DM) choose from the same choice set. This paper instead studies how exposure to an influencer's consumption involving unattainable goods reshapes choice among feasible ones.We propose the Aspiration-Weighted Luce Model (AWLM). In this framework, the DM forms a convex combination of her idiosyncratic Luce preferences and the influencer's full consumption distribution, and then projects this "attempt target" onto her strictly smaller feasible set. This simple "mix then project" structure has a distinct and testable behavioral prediction: aspirational dampening. Holding the within-feasible composition of the influencer’s choices fixed, shifting exposure toward infeasible, aspirational alternatives attenuates the influencer's leverage over the DM's feasible choices. In other words, an influencer who heavily features luxury items exerts a weaker influence on a follower's affordable consumption than one who showcases mostly accessible goods.
This framework introduces a new source of variation for recovering DM's underlying preferences. Standard revealed preference approaches typically require variation in the DM's choice set. In contrast, the AWLM achieves point identification of both the influence strength and idiosyncratic preferences from as few as two distinct exposure regimes on a fixed feasible set (i.e., a single shift in the influencer’s consumption pattern suffices). By exploiting the parallel geometry between shifts in aspirational exposure and shifts in feasible choice, this approach offers a new identification strategy for empirical research. We provide an axiomatic characterization based on proportional response and leverage restrictions, and demonstrate how this identification strategy naturally yields testable overidentifying restrictions for GMM estimation.
Asset Market Participation, Liquidity, and Monetary Policy (E5, G5)
Abstract
How does monetary policy affect households’ decision to participate in the asset market? In this paper, we propose an underexplored mechanism: the liquidity channel. Using a New Monetarist model, we show how monetary policy affects both the opportunity cost of holding money, and the ease with which different assets can be used in transactions. Through this mechanism, monetary policy changes the liquidity value of competing assets based on their relative liquidity and, in turn, households’ incentives to participate in asset markets.To motivate our analysis, we present empirical evidence from the Survey of Consumer Finances (SCF). We highlight two observations: (1) changes in the Federal Funds Rate (FFR) have little effect on the average rate of asset market participation; (2) however, as FFR goes up, the variation of asset market participation rate increases significantly.
Our baseline model highlights both (i) a direct effect where increases in the nominal interest rate increases the cost of holding money, encouraging households to switch to other forms of assets; and (ii) an indirect effect where a higher nominal interest rate increases the liquidity premium of those assets, which discourages participation. When nominal interest rates are low, there is no indirect effect due to liquidity abundance. However, as nominal interest rates increase the indirect effect becomes operational, making the total effect ambiguous, consistent with the empirical evidence we observe from the SCF data.
Next, we allow asset liquidity to be endogenous and show that multiplicity can emerge due to the strategic complementarities between households’ decision to participate in the asset market and sellers’ decision to accept the asset as a means of payment. Furthermore, hysteresis can occur if it takes time for sellers to develop that ability to accept the less liquid asset, implying that temporary shocks can permanently change households’ asset market participation decisions.
Auditing Environmental Services under Information Asymmetry: Experimental Evidence from a Credence Goods Framework (Q4, D4)
Abstract
The transition to a low-carbon economy is significantly constrained by an Energy Efficiency Gap, where households and firms fail to invest in cost-effective green technologies. One driver of this gap is informational asymmetry inherent in the market for energy retrofits (e.g., insulation, HVAC systems, solar installation, etc.). These services have credence goods properties, meaning that consumers often cannot verify if they are strictly necessary or if the installed materials indeed meet the promised efficiency standards. We report on a laboratory experiment to test the effectiveness of market interventions in mitigating this problem. We compare a baseline control to treatments introducing reputation systems, direct punishment, a combination of both, and reputation with ostracism enforced by Mystery Shoppers. Our results show that while reputation and punishment individually reduce fraud, their combination is the most effective. Furthermore, introducing the threat of ostracism (market exclusion) via mystery shopping audits significantly enhances reputational effects.Behavioral Effects of Tax Enforcement on Non-Compliant Business Taxpayers: Evidence from Administrative Tax Data (H2, H3)
Abstract
We examine how enforcement interventions affect the compliance behavior of business taxpayers. We focus on the most important pillar in the collection of taxes in the United States - income and employment taxes withheld in trust by employers. How to better motivate employers, especially delinquent employers, to comply with the tax law? The study analyzes the period 2010-2019, when the IRS experienced declining budget appropriations, and adopts a natural experiment approach.An important question addressed in this paper is: how effective and how quick are the impact of the IRS enforcement programs? We examine the behavioral effects of tax treatments on a quarterly basis, to compliment the business tax reporting and filing cycle. Rather than focusing on a single type of tax treatment, we examine the deterrence effects of a broad range of tax treatment programs across different stages of the tax lifecycle.
We find strong and robust direct effects, even within a short time frame. Our regression analysis shows that tax filing and payment enforcement programs have significant direct impacts on securing delinquent business tax returns in a timely manner. Among these tax treatments, notices with deterrence messages such as Notice of Intent to Levy, which emphasize the penalties to be applied for continuing noncompliance, are more effective for collecting outstanding tax liabilities.
We find mixed evidence regarding the indirect effects of tax treatments. We find that IRS notice treatments have a persistent indirect effect on compliance, leading to reductions in tax debt. However, these local spillover effects are weaker than the direct effects. Our findings confirm the multiplier impact of the tax interventions and are consistent with the observation verified in Boning et al. (2020) that in general the increased enforcement deters evasion through changing taxpayers’ perceptions of the probability that evasion will be detected and punished.
Behavioral Games and Water-Saving Technology Adoption in Uzbekistan (O3, Q1)
Abstract
Water-saving technologies can reduce irrigation demand by 30–50%, yet adoption remains low in developing countries despite substantial subsidies. This paper uses behavioral games to identify why. We modify the standard Irrigation Game to include an explicit technology adoption choice—purchasing drip irrigation during gameplay—and conduct 38 sessions with 190 farmers across four regions of Uzbekistan. Players are randomly assigned to upstream and downstream positions along a simulated canal, with positions reshuffled between sessions to separate positional effects from individual characteristics. The experiment proceeds through three phases: a baseline irrigation game, the same game with a drip irrigation purchase option, and a water deficit condition with declining supply simulating climate-driven scarcity. We find three main results. First, downstream players facing water scarcity are significantly more likely to adopt drip irrigation, consistent with positional disadvantage driving private adoption incentives. Second, upstream adoption generates far larger positive spillovers: each additional upstream adopter increases downstream water availability by 0.407 units and raises peer adoption probability by 6.4–21.5 percentage points. This creates a fundamental misalignment—farmers with the weakest private incentives to adopt (upstream) generate the highest social returns, while those who adopt most readily (downstream) produce limited spillovers since few players sit below them. Third, real-world farmer characteristics strongly predict in-game behavior—farmers with more downstream plots, higher irrigation costs, and connections to agricultural clusters adopt more in the game, while real-life drip irrigation users consume 0.775 fewer water units in-game, validating the game as a measurement tool. The results imply that uniform subsidy programs are suboptimal: targeting upstream farmers yields higher social returns than subsidizing downstream farmers who would adopt anyway. More broadly, technology adoption acts as a substitute for collective action—by aligning individual and social incentives, it mitigates the distributional conflict inherent in sequential irrigation systems without requiring sustained cooperation.Bias in the Level, Trend, and Cyclicality of the U.S. Unemployment Rate (J6)
Abstract
The unemployment rate is among the most cited economic statistics, yet it is derived entirely from household surveys that suffer from substantial measurement error. We document that these errors meaningfully distort the level, trend, and cyclicality of measured unemployment. Using tax records, we identify a subset of the truly unemployed – recipients of Unemployment Insurance (UI) on Form 1099-G – and link them to their retrospective reports of unemployment in the Current Population Survey Annual Social and Economic Supplement (CPS ASEC). We find that 49% of UI recipients in 2010 reported no unemployment in the same year they received UI, with comparable misreporting in the Survey of Income and Program Participation despite its shorter reference period and design features intended to reduce measurement error. Unemployment errors decrease as UI benefits increase: 80% of recipients in the lowest ventile of benefits reported no unemployment, compared to 30% in the highest ventile. This gradient suggests that salience plays a central role in reporting accuracy and that UI non-recipients may be particularly likely to be missed by the official measure. The CPS ASEC allows us to construct a retrospective unemployment rate – the share of weeks in the labor force spent unemployed – which we show tracks the official unemployment rate closely. We develop methods to correct the retrospective unemployment rate for three sources of errors: false negatives, false positives, and errors in reported duration. Given the extremely close and stable relationship between the retrospective and official unemployment rates, we argue that these corrections are informative about bias in the official rate. Finally, within-survey inconsistencies – the share of survey-reported UI recipients who report no unemployment – suggest that misreporting is procyclical and has doubled from the early 1990s to 2022. These trends raise concerns about the accuracy of current unemployment statistics.Bootstrapping Smoothed Factor-Augmented Quantile Regression Models (C1, C6)
Abstract
Factor-augmented regression models are widely used for forecasting in data-rich environments, but most applications focus on conditional mean forecasts. To better capture tail dynamics and uncertainty, this paper studies the smoothed factor-augmented quantile regression (FAQR) model, which uses estimated latent factors to model conditional quantiles. Valid inference in this framework faces two main challenges. First, the standard quantile regression objective function is non-smooth, which complicates Hessian-based asymptotic inference. Second, existing theory treats factor estimation error as asymptotically negligible, but this result relies on restrictive conditions on the relative growth rates of the cross-sectional dimension ($N$) and the time dimension ($T$).This paper addresses these issues by proposing a residual-based smoothed wild bootstrap procedure. The contribution is threefold. First, I extend smoothed quantile regression to the FAQR framework and establish valid inference under the general asymptotic regime $\sqrt{T}/N \rightarrow c \in [0,\infty)$. I show that when $c>0$, factor estimation error induces a non-negligible asymptotic bias, which can be characterized through density-weighted population moments. Second, I provide theoretical justification for the residual-based smoothed wild bootstrap and show that it reproduces the limiting distribution of the estimator, including the bias term. Third, Monte Carlo simulations show that the studentized smoothed wild bootstrap substantially improves finite-sample coverage relative to analytical asymptotic methods. These improvements remain robust under heavy-tailed, asymmetric, and heteroskedastic error distributions. Finally, I revisit the predictive relationship between systemic risk factors and macroeconomic outcomes and show that the proposed method provides a more reliable evaluation of their predictive power.
Breaking Patronage in the Courtroom: Judicial Independence and Bias in Petty Corruption Cases (K4, D7)
Abstract
Petty corruption, though small in monetary value, is widespread and imposes substantial economic and governance costs. This paper constructs a novel dataset of judicial decisions in petty corruption cases in China from 2014 to 2021, based on administrative judgments collected from China Judgments Online. We show that local political leaders, especially municipal Party secretaries, influence courts to favor lower-level government officials connected to them through shared hometown ties. These connected defendants receive systematically lighter punishments, including shorter prison sentences and lower financial penalties. We then examine whether greater judicial independence can curb such patronage in the courtroom. To do so, we exploit a major judicial reform in China that recentralized authority over local courts by transferring their financial and personnel control from local to provincial governments. Using the staggered rollout of the reform, we find that it significantly reduces the judicial advantages enjoyed by lower-level government officials connected to local political leaders. To explore the underlying mechanism, we conduct a textual analysis of court judgments. The results suggest that greater judicial independence leads to a more consistent application of legal provisions, more standardized reasoning across cases, and less discretionary language in written decisions. At the same time, the reform does not fully eliminate unequal treatment, as defendants connected to higher-ranking officials above the municipal level continue to enjoy substantial advantages. Overall, our findings suggest that judicial independence helps reduce patronage in petty corruption cases, but its impact is limited where political influence remains deeply rooted.Bridging Labor Market Dualism: Employment Security and Fertility in South Korea (J1, J4)
Abstract
A growing literature emphasizes that fertility increasingly depends on whether employment is compatible with family life, and flexible jobs are often seen as one way to ease this trade-off. Yet in dualized labor markets, temporary jobs may offer flexibility without security: protection gaps between temporary and permanent workers are large, and transitions into permanent employment are limited. Little is known about whether reducing this divide affects marriage and childbearing, especially in settings such as South Korea, where ultra-low fertility coexists with pronounced labor market dualism. This paper examines whether improving employment security by narrowing the divide between temporary and permanent jobs affects family formation. I study South Korea's 2017 public-sector contract conversion reform, which converted eligible fixed-term workers into open-ended contracts. Using newly linked administrative employer-employee data and an event-study design that exploits variation in reform exposure across public entities, I find that greater reform exposure significantly raises the probability of first marriage and first childbirth. To investigate mechanisms, I complement the administrative data with evidence from the Korean household panel survey. The results show that transitions from fixed-term to open-ended contracts substantially improve perceived job security and increase the probability of marriage. Among a broad set of job attributes, perceived job security emerges as the main margin of change. These findings suggest that in dualized labor markets, the family-formation effects of temporary employment depend critically on whether temporary jobs provide a pathway to secure employment. Policies that reduce the protection gap between temporary and permanent jobs may therefore offer an underappreciated lever for addressing fertility decline — one that operates through labor market structure rather than direct family subsidies.Bringing Services Closer to People? Distance to Administrators and State-Citizen Relations in Uganda (H7, D7)
Abstract
Looking to improve public service delivery, many developing countries have increased the number of subnational administrative units over the past three decades. A key dimension of this strategy is spatial: smaller administrative units reduce the physical distance between households and local administrators. Leveraging an episode of district proliferation in Uganda, this paper studies whether geographical proximity to local government administrative centers affects the provision of public goods and services. To establish causality, I exploit a policy implementation rule that introduced exogenous variation in households' proximity to district headquarters. The main findings show that households living closer to new administrative centers have significantly better access to electricity, roads, health facilities, and water sources. These improvements are accompanied by a shift from contentious to participatory engagement with the state and lower direct experiences of corruption. Despite these gains, proximity does not translate into higher trust in government or increased democratic perceptions. These findings suggest that in hybrid regimes, bringing the state closer to citizens can improve services, engagement with the government, and accountability while leaving the underlying political relationship between citizens and the state unchanged, a pattern that holds even in politically aligned districts where engagement is stronger.Buying Stability: Affirmative Action and Elite Co-optation in Imperial China (N4, P0)
Abstract
How do autocracies stabilize elite politics when they cannot rely on representative institutions to share power with elites? We argue that one tool is place-based quotas in elite recruitment. This paper studies the Ming 1427 regional quota reform in the civil service examination of imperial China, which sharply reallocated access to bureaucratic office. Using a prefecture-level difference-in-differences design, we show that quota expansion broadened entry into the bureaucracy and reduced social unrest, with especially large declines in elite-led rebellions and no comparable effect on peasant uprisings. However, these gains were politically selective. Additional quotas were disproportionately captured by historically strong prefectures and incumbent local elites, suggesting that the reform operated less as egalitarian redistribution than as targeted elite co-optation. We also show that quota expansion lowered selection quality and weakened the long-run advancement of newly incorporated candidates. The findings suggest that autocratic regimes can stabilize rule by selectively widening access to elite office, revealing a central tradeoff in state-building between defusing elite conflict and preserving bureaucratic quality and limiting uneven rent allocation.Can Institution-Rooted Conflicts be Treated by Policy Interventions? Evidence from a Nationwide Program in Myanmar (D7, O2)
Abstract
Weak and exclusionary institutions are a central driver of civil conflict in developing countries by eroding inclusive local governance and dispute-resolution capacity, especially in areas marked by ethnic divisions. To study whether such institution-rooted conflict can be mitigated by policy interventions, we examine the impact of Myanmar’s national community-driven development (CDD) program during the large-scale civil war after an unexpected coup in 2021. Exploiting the discontinuity of treatment along administrative boundaries, we estimate that CDD enrollment causally reduced inter-ethnic conflicts by 36.9%. Our mechanism analysis suggests that this peace dividend stems from improvements in local governance and cooperative behavior, rather than from economic development. Treated villages exhibit stronger local governance in both quantity and quality, greater female participation, and more effective responses to public challenges, e.g. the COVID-19 pandemic. However, we find no evidence of broader economic gains beyond infrastructure delivery.Can the Social Safety Net Catch Lead? The Moderating Effect of WIC on Prenatal Lead Exposure. (I1, Q5)
Abstract
While decades of policy have successfully limited sources of lead, research shows that there is no safe level of exposure. Concern has largely been directed at children, but recent attention has shifted to prenatal exposure, where a pregnant individual can pass inhaled lead through the placenta to the developing fetus. Quasi-experimental evidence has linked low levels of prenatal lead exposure to adverse birth outcomes such as lower birth weight and preterm birth. The medical literature indicates that nutrition, particularly calcium, plays an important role in limiting lead absorption in the body. Access to the social safety net is one avenue through which people receive nutrition and care that might insulate them from the dangers of lead. I provide the first study to analyze social insurance channels as a moderating force against environmental hazards. To do this, I combine spatial variation in geographic access to federal nutrition programs such as WIC with spatial and temporal variation in lead exposure via the openings and closings of lead-releasing facilities documented in the Toxic Release Inventory. Using restricted birth certificate data from North Carolina, I categorize pregnant individuals by (1) exposure to airborne lead and (2) geographic proximity to WIC clinics. I use a triple differences model to estimate the effect of shocks to prenatal lead exposure across residences geographically near versus far from WIC clinics. I find that proximity to WIC offsets the adverse health effects of a lead shock during pregnancy, essentially insulating the individual from the damages of lead exposure. These results are largely driven by non-white individuals and individuals on Medicaid. With racial and socioeconomic disparities in both exposure to pollution and food access/security, evidence that informs policymakers on how to best address exposure among these populations could have substantial welfare benefits.Carbon Intensity and Global Capital Flows: Evidence from Mutual Fund Rebalancing (G1, F3)
Abstract
We uncover a new channel of international capital reallocation—carbon intensity—shaping cross-border flows. Using global mutual fund allocation data, we find that the fund flows are negatively related to changes in the countries’ carbon intensity. This relation is stronger in developing countries and countries with stringent environmental regulations, lower shares of renewable energy in energy production, and a more business-friendly regulatory environment. We leverage the staggered ratification of the Paris Agreement and show that funds increase their portfolio allocations to countries after ratification. Finally, fund investors react to funds’ exposure to country carbon risk.Changes in Time Allocation Due to Unilateral Divorce, Evidence from Mexico (D1, J0)
Abstract
Unilateral divorce may affect how households allocate their time through changes in bargaining power. This paper uses data from Mexico to analyze the impact of unilateral divorce on married couples. Exploiting exogenous variation in the timing of unilateral divorce implementation across states and data from the ENUT time-use survey, we find that married women with children increase the time they dedicate to housework activities. Other interventions do not drive this effect. Future policies should account for the potential backlash effects of such reforms in the context of traditional gender norms.City of Shame: Crime Signaling and Inverse Deterrence in a Weak State (K4, O1)
Abstract
Do high-profile crimes deter future offending or signal weak enforcement? We study the effects of the highly publicized rape and murder of a seven-year-old girl in Kasur, Pakistan, in 2018 on child sexual abuse. Using district–month data from 2015–2024 and a synthetic difference-in-differences design, we compare Kasur to a constructed counterfactual from other districts. We find a large, persistent increase in reported child abuse, about ten additional cases per month, concentrated in opportunistic offenses and weakly supervised settings. The increase emerges most clearly following subsequent institutional responses and coincides with judicial congestion, consistent with inverse deterrence driven by institutional strain rather than transient reporting effects.Climate-Related Financial Policy and Systemic Risk (G2, Q5)
Abstract
We examine the relationship between climate-related financial policies (CRFPs) and banks’ systemic risk. Using a sample of 458 banks in 47 countries over the period 2000-2020, we document that more stringent CRFPs are detrimental to overall financial stability and contribute to increased system-wide distress, where excessive regulatory constraints may impose burdens on banks. Decomposing systemic risk shows that stricter climate-related financial policies raise bank-level volatility but not interbank correlation, indicating that higher systemic risk stems from increased individual bank fragility rather than stronger synchronization. We investigate the bank-level transmission channels through which climate-related financial policies may contribute to higher systemic risk. Tighter policies are associated with slower loan growth, lower profitability, an increase in non-performing loans and compressed net interest margins, as funding costs rise faster than lending rates. At the same time, capital adequacy ratios decline, indicating mounting balance sheet pressures. However, the implementation and ratification of the Paris Agreement, more robust adaptation strategies to cope with climate shocks and a higher incidence of natural disasters and a larger number of people affected by extreme climate events may counteract the amplifying effects of CRFPs on systemic risk. Moreover, banks with stronger environmental, social, and governance (ESG) commitments experience less systemic distress when exposed to green financial policies.Cognitive Origins of Extrinsic Sentiment: Prevailing Ideology and Business Cycle (E7, E3)
Abstract
I develop a microfounded account of extrinsic sentiment ("sunspot") fluctuations in a New Keynesian DSGE model by linking expectation dynamics to the propagation of non-fundamental ideologies. The core mechanism adapts the spreading-activation theory of semantic memory (Collins and Loftus, 1975): beliefs are formed through associative activation, allowing salient ideological content to spill over into economic expectations. Within the model, an ideological shock operates as a persistent demand disturbance, distorting aggregate consumption and output endogenously. I operationalize this framework using sexism as a representative ideology, quantifying its evolution over a 75-year news archive through large language models. I discipline the model in two steps. First, I establish the causal effect of ideological salience on household financial sentiment by exploiting Dobbs v. Jackson (2022) in a difference-in-differences design. Second, I trace the manifestation of such non-fundamental ideology on non-rational belief at the macro level, exploiting both the dynamic response of professional economic beliefs to ideological innovations in a VAR and the systematic explanatory power of ideological innovations for forecasters' forecast errors. Evidence consistently supports the activation of this associative channel: negative non-fundamental ideological shocks systematically distort economic expectations in the pessimistic direction, echoing the demand disturbance mechanism formalized in the model.Communication and Innovation – Evidence from Open Source (M5, O3)
Abstract
Innovation and knowledge creation often stem from collaborative efforts that require communication and coordination. Yet it is unclear how communication and coordination affect innovation due to limited observability and endogeneity issues. We study how communication affects knowledge creation by exploiting a platform-wide feature change on GitHub — the world's largest open-source software platform — that sharply increased the salience of communication. On March 23, 2011, GitHub activated notifications for @-mentions: tagging a username in a comment began generating alerts and subscribing the tagged user to the thread. Before this date, @-mentions were plain text with no notification function.Using production-level microdata from the top 1% projects on GitHub, we leverage the plausibly exogenous timing of the feature change to study how communication affects innovation in a difference-in-differences framework, comparing activities with and without at-mentions, before and after their introduction. We also leverage the sudden drop in the cost of salient communication and consequent increase in at-mentions use in an instrumental variable framework to study how communication volume impacts innovation. We study the effects of communication in both early stages of the innovation process (ideation via issues) and late stages (implementation via pull requests).
We find that salient communication slows down convergence during early ideation stages, decreasing the daily likelihood of resolution (closure) by 30% and 44% when directed at people with and without context-relevant knowledge (project members vs. external contributors), respectively. Conversely, effective communication more than doubles contribution success (pull request merging) and speeds up convergence in the implementation stage, but only if directed at people with context-relevant knowledge (project members). Yet we also observe congestion effects at this stage: the marginal responsiveness of receivers that were not already involved in a discussion declines as mentions accumulate, and the marginal merge gains decline as senders escalate targeting.
Compete or Concede? Incumbent Pricing and Product Decisions Following Tesla’s Entry (L6, F2)
Abstract
Trade and foreign direct investment (FDI) in green energy generate climate benefits but also raise concerns about technological dependence, forcing countries to balance openness with protection. Does opening domestic markets to technologically advanced foreign firms spur domestic upgrading (the “catfish effect”) or crowd out incumbents through intensified competition (the “shark effect”)? How do such entries affect product variety, prices, and welfare? I study these questions through Tesla’s 2019 entry into China’s electric vehicle (EV) market, where it built a large-scale gigafactory in Shanghai as the first fully foreign-owned automaker, creating a significant competitive shock with annual capacity exceeding 750,000 vehicles and extensive supply-chain localization.I develop a structural model of consumer demand and firm behavior. Consumers choose differentiated vehicles in a random-coefficients logit framework, with heterogeneous preferences over price, fuel type, driving range, and other attributes. On the supply side, firms choose product portfolios and compete in prices under Bertrand competition. I allow for marginal cost heterogeneity linked to supplier overlap with Tesla's localized production. Following Eizenberg [2014], I infer bounds on fixed costs of product entry by treating observed portfolios as Nash equilibrium outcomes. This framework captures both competitive effects through markups and pricing, and technological spillovers through supply-chain linkages.
Preliminary demand estimates show strong price sensitivity (mean own-price elasticity of approximately −3.6) and substantial heterogeneity in preferences for EVs. Supply-side estimates reveal economically meaningful cost differences correlated with Tesla supplier overlap. Counterfactual simulations will quantify market outcomes absent Tesla's entry, under alternative entry timing and different conduct assumptions.
This paper contributes to work on endogenous product positioning and entry in differentiated product markets and to studies of firm upgrading in developing countries. It differs from prior work by modeling both price and product-variety responses to an industrial-policy-driven FDI shock, while incorporating supply-chain spillovers from highly localized production.
Competing for Its Own Sake: Experimental Evidence on the Welfare Effects of Competition (D9, C9)
Abstract
Economists often view competition as a means to motivate effort, improve efficiency, and therefore enhance welfare. However, this instrumental perspective may overlook that competition itself can directly influence individuals’ welfare. This paper investigates how competition affects utility derived solely from the act of competing, independent of material outcomes. I conduct a series of experiments which show that competition affects utility through two opposing channels: a belief channel, in which competition lowers expectations of success and reduces utility; and a preference channel, through which individuals derive enjoyment directly from competing. The preference channel dominates, resulting in a net positive impact on utility. I further show that these welfare effects influence future choices: competition induces attribution bias, leading individuals to misattribute the enjoyment of competition to the underlying task and increase their willingness to engage in it again, even in the absence of competition. These effects also extend to social interactions, reducing post-competition zero-sum thinking and fostering altruism.Constrained Bargaining (C7, D7)
Abstract
In many real-world negotiations, counteroffers are often constrained by procedural rules, institutional norms, or limits on how quickly underlying outcomes can be adjusted. We study a variation of the Rubinstein bargaining game with a common discount factor $\delta$, in which a player’s counterproposal following a rejection must lie within a distance $b$ of the previous proposal. For generic $(\delta, b)$ we prove that there is a unique subgame perfect equilibrium, and we show that constrained bargaining weakly increases the first mover’s payoff. In contrast to the standard Rubinstein model in which agreement is reached immediately, we find that for sufficiently patient players (high $\delta$) and tight constraints (low $b$), equilibrium agreement is delayed.We then extend our framework to a dynamic spatial bargaining model in which two players alternate controlling the evolution of a state variable in $\mathbb{R}^2$. In each period, the player in control can move the state by a vector of bounded length, and the game ends once the state reaches a fixed radius $R>0$ from the origin. Payoffs depend on the alignment of the terminal state with each player’s preferred direction and are discounted by the time to agreement. This model captures negotiations between two agents with heterogeneous preferences over the trajectory of systems that can only be adjusted gradually. As the process evolves, feasible adjustments become increasingly limited, generating endogenous path dependence and making large deviations from the current trajectory progressively more difficult.
Consumption Peer Effects and the Transmission of Negative Product Information (D8, D1)
Abstract
Consumers are more likely to purchase a product after one of their peers has already purchased that same product. In this paper we examine the reverse situation; whether a negative product quality experience of one peer reduces consumption by other peers. Evidence on this reverse relationship is challenging to provide, for both theoretical as well as econometric reasons, despite there being a long-standing interest in this reverse relationship.Our key innovation in this paper is to use data on warranty claim filing by a single peer, as an indicator of that peer's negative consumer experiences with the claimed product. We empirically examine the effect of that single warranty claim by that individual (the sending peer) on the future consumption of that product by other peers (the receiving peers). Our use of the warranty claim event allows us to identify the causal flow of information between peers, and also to overcome the "visibility bias” problem.
We perform our analyses on data provided by a major Canadian consumer durable goods retail store chain. This data includes six million sales transactions across over 200 brands along with 95,000 warranty claim filings. To estimate the causal effects of warranty claimants' negative experiences on peers' purchasing behavior, we use the precise geographic location of consumers and claim and purchase dates. We conduct a hyper-local difference-in-differences (DID) empirical methodology. The treatment group is defined as neighbors who live within 100 meters and the control group is defined as neighbors who live between 100 and 200 meters from the claimant.
Our main finding is that a negative quality experience with a brand by the sending peer (as indicated by a warranty claim), significantly lowers subsequent purchases of that brand by the receiving peers by 7%.
Corporate Debt Composition, Access to Credit, and Monetary Policy (E5, G3)
Abstract
In both the U.S. and the euro area, the share of market finance in aggregate corporate credit has grown over time. To study the implications of the corporate debt structure for the transmission of monetary policy, we develop a New Keynesian dynamic stochastic general equilibrium model in which firms are heterogeneous in productivity and have access to different forms of external finance. Our setup makes both the corporate debt composition and firms' credit access endogenous and dependent on aggregate economic conditions. We find that following a monetary policy contraction, aggregate corporate debt contracts and credit access tightens. At the same time, however, there is substitution from bank loans toward bond finance, as loan supply contracts due to a squeeze in bank equity. We show that the model qualitatively replicates empirical impulse responses to monetary policy shocks in the euro area. Our results lend support to the relevance of the bank lending channel of monetary policy transmission.Corporate Misconduct, #MeToo, and Gender Equity: Evidence from LinkedIn (J7, Z1)
Abstract
Whether public exposure of workplace misconduct reshapes firm behavior and advances gender equity remains an open question. Using LinkedIn career histories for over 10 million workers matched to 418 sexual harassment and discrimination scandals between 2010 and 2023, this paper estimates the causal effect of scandal revelation on firm-level gender composition, promotions, and pay. Scandals are identified from federal court records, EEOC filings, and news archives with precise revelation dates. Identification exploits staggered variation in scandal timing in a heterogeneity-robust difference-in-differences design, using never-accused firms in the same industry and size class as the control group.At S&P 500 firms, scandal exposure increases the female senior employee share by 0.54 percentage points and the female leadership share by 0.46 percentage points, while male senior hiring falls by 0.60 percentage points. In the broader Russell 3000 sample, female promotion rates rise by 0.48 percentage points with no effect for men. An individual-level salary analysis using 16.3 million person-year observations finds persistent wage declines at scandal-hit firms, reaching 1.1 percent for women and 1.4 percent for men five years after revelation.
These results show that reputational shocks modestly improve female seniority while broadly depressing wages. Future work will track individual career trajectories to examine how workers sort across firms following a scandal, whether female workers disproportionately exit or advance at affected firms, and how the effects vary by firm size, industry, and pre-scandal gender composition. The results contribute to a growing literature on how external accountability pressures shape workplace gender inequality.
Courts, Costs, and Crises: The Role of Penalties for Sovereign Default Expectations (F3, H6)
Abstract
This paper studies the causal effect of default costs on sovereign default probabilities. I exploit high-frequency changes in litigation-induced default costs around legal news-shocks from creditor lawsuits. A one ppt increase in the default-penalty-to-GDP ratio leads to a 1.38 ppt drop in default probability, indicating that direct economic costs are a central reason creditors expect repayment. I develop these results in a model of international borrowing with endogenous default and heterogeneous bond contracts. Here, similarly, adopting more punitive debt portfolios reduces default probabilities and risk premiums. However, the same mechanism can also lead shortsighted governments to pursue high-debt-high-default-cost regimes, heightening financial instability.Delaying Cancer: The Effect of Education on the Age at Cancer Diagnosis (I1, I2)
Abstract
This paper studies whether education affects not only whether serious disease occurs, but also when it occurs over the life course. We examine the causal effect of schooling on the age at first cancer diagnosis using Austria’s 1962 compulsory schooling reform, which increased the required schooling from 8 to 9 years. The reform shifted some students from apprenticeship-based vocational training into more school-based education, generating quasi-experimental variation in educational attainment.We combine a regression discontinuity design with newly linked administrative data covering the full Austrian population, including schooling records and cancer registry. We first estimate the effect of additional schooling on the probability of a first cancer diagnosis, and then use accelerated failure time and discrete duration models to examine whether education shifts the timing of diagnosis.
We find that additional education reduces the incidence of behaviourally linked cancers, particularly lung cancer, and delays the age at cancer diagnosis among men. We find no comparable effects for women or for cancers less clearly related to behaviour. We provide evidence that this effect operates in part through occupational sorting: the reform shifted men into occupations with lower exposure to risky behaviours and workplace hazards, and it also delayed the age at which such exposures began.
These findings suggest that education influences not only the likelihood of serious disease, but also the timing of its onset. By shifting cancer to later ages, education may extend healthy working lives and help explain persistent socioeconomic gradients in health, age at diagnosis, and subsequent labor-market trajectories.
Departure from the Dust (Q5, R1)
Abstract
Environmental shocks are universal, but the capacity to adapt is not. This paper studies how recurrent dust storm exposure shapes internal migration in China. Using satellite-based dust measures linked to population census migration flows, we estimate the causal effect of dust storms on out-migration via panel fixed effects and a wind-alignment instrumental variable that isolates exogenous downwind exposure. Dust storms generate substantial and robust out-migration responses. Adaptive capacity is sharply stratified: agricultural hukou holders respond at rates five times those of non-agricultural residents, and responses decline steeply with age. We document two opposing policy margins: road infrastructure amplifies migration by lowering relocation costs, while the Three-North Shelterbelt Program dampens it by reducing dust incidence at source, with benefits spilling over to downwind counties. Embedding these estimates in a quantitative spatial equilibrium model, we show that road expansion raises aggregate welfare through spatial reallocation but concentrates gains among mobile households, whereas afforestation generates more equitably distributed improvements. The contrast reveals a fundamental efficiency-equity trade-off between adaptation through exit and mitigation through exposure reduction.Determining the Structure of Dynamic Factor Models (C3, C1)
Abstract
We propose two procedures for determining the number of dynamic factors q, extending Bai and Ng [2002] and Ahn and Horenstein [2013] to dynamic factor models with filter length m > 1. As an intermediate step, we develop a simple and computationally efficient alternating least squares (ALS) algorithm that directly estimates the dynamic factors, rather than their static representations. By working with these direct estimates, our approach enables joint determination of the filter length m and the number of factors q, which is not feasible in Bai and Ng [2007] and Amengual and Watson [2007]. We apply our procedures to estimate the number of primitive shocks in a large panel of US macroeconomic time series.Disclosure Control as a Real Option: Contractual Governance in Clinical Trials (O3, I1)
Abstract
Clinical trial publications are the primary channel through which drug-efficacy evidence reaches physicians, shaping prescribing decisions and patient outcomes. To bolster credibility, pharmaceutical firms outsource trials to independent investigators. Yet the majority of these trials still include embargo clauses granting sponsors manuscript-review and publication-delay rights. We frame the embargo as a real option that grants firms disclosure discretion when results disappoint. Linking over 18,000 U.S. clinical trials to peer-reviewed publications, we show embargo adoption concentrates where option value is highest—when fewer co-sponsors reduce contracting frictions, larger enrollment facilitates statistical power for specification search, and adverse results would cause larger market-value losses (Phase 3 trials). Our core finding concerns statistical reporting: embargoed trials are 60% more likely than non-embargoed trials to report primary outcomes at a significance coined by the FDA as "very statistically persuasive" (p < 0.0001)—a level which can qualify their drug for approval without additional replication. Such "very persuasive" significance is moderated by the number of primary outcomes tested in the trial, consistent with selective outcome reporting, and does not appear in non-embargoed trials. Furthermore, among marginally insignificant trials, embargoed studies also show 33 percentage points more positive spin in published abstracts, eliminating 82% of the negative tone associated with marginal significance for non-embargoed trials. We document stratified sophistication: top-tier journals and the FDA penalize embargo-enabled manipulation, whereas lower-tier journals and capital markets do not, allowing manipulated findings to flow into the medical literature that guides prescribing and patient care.Discrimination of Toddlers: A Nationwide Field Experiment (J1, I2)
Abstract
We provide causal evidence on whether toddlers face discrimination, independent of parents' race. Identifying discrimination against toddlers is methodologically difficult because most signals available to gatekeepers, such as names, speech, or behavior, are tied to parental characteristics. We exploit natural variation in children's skin tone within mixed-race parents, which is plausibly orthogonal to parental traits. We use this variation in a nationwide field experiment involving the near-universe of more than 25,000 German childcare facilities. Each facility receives a standardized email inquiry accompanied by an AI-generated family photograph in which parental appearance is held fixed while we gradually vary the toddler's racial features on a Black-white scale. We find a non-linear effect of race on response rates, information quality, and sentiment, with the Black toddlers showing the lowest response rate independent of parental race. We further examine how discrimination varies with local level variables, such as right-wing vote shares, competition over childcare slots, and exposure to migrants.Disentangling the Net: Evidence on the Production Effects of Fisher Cash Transfer Programs in Mexico (Q2, O1)
Abstract
Considerably little attention has been paid to income support programs for fishers, which are tools for addressing issues of equity, food security, and development. Mexico’s fishing industry provides a compelling case study, as several fisheries subsidy programs overlap in time, location, and commercial scale. This study evaluates the production impacts of PROPESCA/BIENPESCA, a direct cash transfer program for fishers while accounting for the impact the other subsidies and environmental variables may have on the production.The challenge in aggregating and transforming administrative data is that fish quantities are not linked directly to subsidy participants for any subsidy, bringing up the issue of a "tangled" effect. We used a novel geographic information system, connecting offices (where catch/harvest is reported) to a fishers' locality by the nearest Euclidean and driving distance and to incorporate the environmental controls; monthly sea surface temperature, wind speed, rainfall, and chlorophyll concentration in the water were recorded through rasters across the coast of Mexico and large bodies of water. Program "double dippers" are identified by name and geographic location. Most importantly, to distinguish the impact of this specific subsidy, we target fisheries that are predominantly found among PROPESCA participants and low-scale fishers.
Following the recently developed two-stage DID estimator, we can estimate (conditionally on parallel trends) the average treatment effect on the treated that is robust to heterogeneity in treatment time and effect, while controlling for time-invariant local characteristics and common temporal shocks. Our findings show insignificant results for aggregate catch but a 4,627 kg increase on average for mullet fish catch, supporting the use of target fisheries. This paper contributes to the literature on the role of subsidies in resource-dependent sectors by providing a comprehensive and causal analysis of a direct cash transfer, while explicitly addressing environmental confounding factors and the overlap of program participation.
Do All Inputs Matter? Export Market Expansion, Capital Goods Adoption, and Upgrading (F1, O0)
Abstract
How do firms respond to expanding export markets? Firms in developing countries may upgrade by accessing high-quality inputs, skilled workers, and advanced technologies used in industrialized economies. While the effects of accessing intermediate inputs (e.g., raw materials) have been well documented, little is known about the impact driven by capital goods (e.g., machinery and equipment). This paper fills this gap by emphasizing capital goods adoption as a key driver of both product and process upgrading. I exploit the removal of the Multi-Fibre Arrangement quotas in 2005 as a significant positive export demand shock for Chinese textile and apparel exporters. Using Customs data from 2001 to 2015 and detailed industry information, I identify direct machine imports at the firm level. The difference-in-differences analysis shows that: (i) firms with higher exposure to export market expansion are more likely to import high-end machines; (ii) machine-adopting firms are found to subsequently upgrade their products: exporting higher values and wider varieties of products, at higher prices and with improved quality, to more high-income countries; (iii) an increasing relative labor-capital share is observed, particularly among machine-adopting firms; (iv) these results are primarily driven by the intensive rather than the extensive margin. To rationalize these empirical findings, I develop a theoretical framework incorporating firms' capital goods adoption decisions and endogenous technological change. The model emphasizes the presence of fixed costs and adjustment frictions involved in adopting high-end machines. Welfare gains from trade may be underestimated if the unique role of capital goods in determining process upgrading is not taken into account.Do Commonly Held Firms Coordinate Their R&D? Evidence from Patents (L1, L4)
Abstract
A growing literature documents that common ownership by institutional investors softens price competition and boosts aggregate R&D spending. However, the fundamental question of where firms choose to innovate has remained unanswered. We provide evidence that common ownership reshapes the strategic direction of innovation, not merely its scale.We develop a theoretical framework that identifies two opposing forces. A spillover effect pushes commonly owned firms toward similar technologies, as shared investors help internalize cross-firm technological benefits. A differentiation effect pulls competing firms apart, as common owners profit when rivals carve out distinct product-market niches and avoid head-to-head innovation races. The net effect depends critically on how close competitors are: Common ownership increases technological similarity among distant firms but drives technological divergence among close rivals.
We test these predictions using a dataset linking patent portfolios for over 2,400 U.S. public firms from 2000 to 2017. We measure patent similarity using a state-of-the-art language model trained on patent text, capturing the technological proximity of firms' entire innovation portfolios. To address the concern that institutional investors sort into firms with similar technologies, we instrument common ownership with joint inclusion in S&P stock market indices, which generates plausibly exogenous variation through the mechanical behavior of index-tracking funds.
We show that common ownership increases patent similarity among firms in different industries but reduces it within narrow industry groups. This differentiation effect for same-industry firms is significantly stronger for product patents than for process patents, indicating that common owners most aggressively redirect innovation away from direct product-market confrontation.
Our findings reframe the policy debate around common ownership. Beyond softening price competition, common ownership may quietly determine the technological trajectory of competing firms.
Does Medicaid Expansion Lead to Earnings Adjustment? Evidence from Monthly SIPP Data (H3, I3)
Abstract
This paper studies labor supply responses to the Affordable Care Act (ACA) Medicaid expansion. Despite a sharp eligibility cutoff at 138 percent of the federal poverty line (FPL), prior studies find limited effects on earnings and employment. Using monthly data from the Survey of Income and Program Participation (SIPP) and a regression discontinuity design, I show that these null results mask within-year income adjustments that are not captured in datasets based on annual income measures.I introduce a novel behavioral mechanism, the “dip-a-toe” strategy, in which individuals reduce earnings in a single month to qualify for Medicaid despite higher annual income. Using prior-year lowest monthly earnings as the running variable, I find that childless adults in expansion states reduce their lowest monthly earnings by approximately 39 percentage points of the FPL relative to similar households just below the cutoff. These responses are strongest between 2014 and 2016, when the individual mandate penalty was highest, and attenuate after the penalty is eliminated. Additional evidence shows increases in zero-earnings months and reductions in working hours, indicating adjustments along both extensive and intensive margins.
These findings reconcile the gap between theory and prior empirical evidence and highlight the importance of measurement in evaluating behavioral responses to means-tested programs. More broadly, the results show that eligibility rules based on monthly income can generate substantial labor supply distortions
Does the Timing of Income Matter? Evidence from Delayed Tax Refunds (I3, H2)
Abstract
Alcohol and tobacco consumption impose substantial social costs, yet traditional policies such as sin taxes and bans are often regressive and may have reached their limits. This paper examines whether the timing of income transfers, rather than their level, affects consumption of harmful goods among liquidity-constrained households in the United States.I exploit the Protecting Americans from Tax Hikes Act of 2015 as a natural experiment. Beginning in the 2017 tax filing season, the IRS was prohibited from issuing Earned Income Tax Credit refunds before February 15, creating an exogenous delay of four to six weeks. The average transfer is about 4,272 dollars, nearly two months of take-home pay for households with limited liquid assets. Using the Nielsen Homescan Consumer Panel, I construct a household-level weekly panel of alcohol expenditures and estimate a triple difference model that exploits variation across state EITC generosity, household education, and the timing shock.
The results show that the delay reduced weekly alcohol expenditure by about 27 percent among low-education households in high supplement states. The effect persists over time and is not concentrated around refund periods, suggesting it is not driven by short-term liquidity. Instead, the evidence supports a permanent income revision mechanism. Timing uncertainty leads households to adjust long-run income expectations and reduce consumption of harmful goods.
These findings show that the timing of transfers, not just their size, is a powerful and underused policy tool for improving household welfare.
Duration as Information: Judge Speed, Belief Updating, and Appeal Decisions (D8, K4)
Abstract
Does adjudication duration affect litigants' appeal decisions? Standard litigation models treat court delay purely as a cost. We propose instead that duration serves as a public signal: litigants observe how long a case took and update their beliefs about judgment stability and the likelihood of appellate correction. Using Chinese court records on first-instance contract disputes, we exploit quasi-random judge assignment and instrument duration with leave-one-out judge speed. A 30-day increase in duration raises appeal probability by 1.65 percentage points. Consistent with the signal interpretation, longer duration also predicts higher appellate reversal rates conditional on appeal. Six heterogeneity tests support the mechanism: effects are stronger for larger claim-award gaps, higher stakes, greater complexity, and in low-trust provinces. Competing explanations based on case complexity, strategic delay, and judge heterogeneity are ruled out. Our findings highlight the informational role of judicial procedures and suggest that reducing delay alone may overlook how procedural features shape litigants' beliefs and relitigation behavior.Dynamic Network Competition in the Index ETF Market (L1, G2)
Abstract
Direct network effects — where a product becomes more valuable as its user base grows — are a natural source of market power. Yet network advantages must be earned, and firms actively compete to attract users and build their installed base. We develop and estimate a dynamic oligopoly model that captures this force: firms lower current prices, sacrificing current margins to attract more users, which strengthens their network and raises future demand. These dynamic incentives can help resolve the longstanding regulatory dilemma between market power and fragmentation — they discipline pricing from within, especially in fast-growing markets, without requiring entry that dilutes network quality.We apply this framework to the U.S. exchange-traded fund (ETF) market. ETFs tracking the same index offer identical risk-return profiles but differ in expense ratios and secondary-market liquidity, the latter operating as a direct network effect: more trading activity improves liquidity, which in turn attracts further inflows. We document concentrated market structures, episodes of aggressive introductory pricing, and a negative relationship between market growth and markups — patterns consistent with dynamic network competition. We estimate within-index demand using a nested BLP model, addressing the reflection problem by instrumenting liquidity with exogenous shocks to dealers' operating costs, and recover supply-side primitives following Bajari, Benkard, Levin (2007).
Counterfactual exercises show that dynamic incentives substantially discipline pricing relative to a static benchmark, with the effect stronger in fast-growing markets. We evaluate antitrust policies through this lens: mergers that consolidate small firms can benefit consumers by strengthening competition against the dominant firm, but operating multiple product lines without consolidation weakens dynamic competition and harms welfare. Reducing entry costs strengthens the disciplining role of potential entrants, though realized entry may fragment networks. Targeted entry subsidies, such as lead market maker programs, can alleviate fragmentation by accelerating new entrants' network formation.
E-Cigarette Flavor Bans, Food Demand, and Nutrient Intake (I1, D1)
Abstract
E-cigarette flavor bans have been widely adopted to curb youth vaping, yet their broader behavioral and welfare effects remain understudied. This paper examines whether such bans generate spillovers into food consumption and diet quality.We propose a two-stage mechanism. First, restricting access to flavored e-cigarettes shifts household demand across food categories, as consumers substitute toward products that provide similar sensory attributes. Second, these changes in food demand translate into differences in nutrient intake, particularly calories and added sugar.
We exploit the staggered implementation of flavor bans across U.S. states and cities as a natural experiment. Using NielsenIQ household scanner data linked to UPC-level nutrition information, we construct detailed measures of both food purchases and nutrient consumption. Our empirical strategy employs a staggered triple-differences design comparing households with prior flavored e-cigarette use to non-users across policy regimes. We first estimate the impact of bans on food category demand and then quantify the implied changes in nutrient intake.
We find that flavor bans shift consumption toward sugar-intensive food categories and increase added sugar and total caloric intake among affected households. Additional analyses show stronger effects among high-intensity users and provide evidence consistent with substitution mechanisms, including changes in cigarette and sugar-sweetened beverage consumption.
These findings highlight an unintended consequence of tobacco regulation and suggest that policies targeting one domain of health behavior may have offsetting effects in another. Accounting for such cross-domain responses is important for evaluating the welfare implications of public health policies.
Economic Convergence, Health Convergence, the Internet and Cell Phones (O4, F6)
Abstract
Economic convergence, the idea that poorer countries can grow faster than “more advanced” economies, seems obvious to many non-economists, but the evidence, for at least two hundred years, is that there has been a huge divergence between economies throughout the world. However, recently several studies have shown that beginning in the 1990s this pattern has changed, as since then, albeit at a slow pace, economies in the world appear to be converging. This study contributes further evidence, based on Penn World Table version 11, of economic convergence starting in the 1990s and more so from the first decade of the 21st century. The key question is then what happened to change a pattern that existed for more than two hundred years?In this study we propose and test two hypotheses that might explain the change from economic divergence to economic convergence. One, in addition to economic convergence, there has also been health convergence with smaller disparities in life expectancy in countries throughout the world almost continuously starting from the 1960s. Could it be that the increase in life expectancy in low-income countries, which on average finally reached fifty years in the end of the 1980s, became sufficient to enable the economies of low-income countries to finally start to converge with more developed economies? The second hypothesis is that the advent of the internet and cell phones in the 1990s led to greater information in low-income countries about producing goods and services, and also significantly improved the ability to communicate in low-income countries, which led to more opportunities to produce goods and services. Initially, access to the internet and cell phones was mostly in the developed economies, but over time their use has spread throughout the world and maybe this has led to economic convergence.
Effects of Incarceration on Employment and Earnings: The Mediating Role of Education (K4, J6)
Abstract
This research studies the relationship between education, incarceration, and post-release labor market outcomes. Leveraging discontinuities in Minnesota’s criminal sentencing guidelines, I estimate the treatment effects of incarceration on reoffense, employment, wages, and post-sentencing educational attainment. I use a rich set of case-specific linked administrative data including sentencing information, case characteristics, defendant demographic information, education history, and wage and employment records. These data allow me to build upon existing causal evidence regarding incarceration’s effects on employment and wages, by understanding the role of education prior to interactions with the criminal justice system and carceral education. This research therefore contributes to our understanding of the role of education in breaking the incarceration cycle. Additionally, I am able to leverage both a control function approach and marginal treatment effects to understand the relationship between judicial departures from recommendations and heterogeneous treatment effects of incarceration. Finally, I contextualize these estimates in a cost-benefit analysis, to understand the fiscal impacts of marginal changes to criminal sentencing, carceral education programs, and high school completion initiatives.Empowering Women Digitally: A Randomised Controlled Trial on Digital Financial Literacy and Women's Economic Empowerment in Rural Pakistan (C9, D1)
Abstract
This study examines the effectiveness of a digital financial literacy intervention aimed at improving financial knowledge, confidence, and behaviour among rural women in Pakistan. Using a randomized controlled trial conducted in two selected villages in the Rawalpindi district, women were assigned to receive digital financial literacy training either individually or jointly with a male household member. The intervention, delivered in person and via mobile phones, focused on core topics including budgeting, saving, and secure digital transactions. The training substantially improved women's financial knowledge, digital confidence, and self-efficacy. The intervention also increased the use of mobile wallets, greater engagement with formal savings mechanisms, and encouraged more consistent budgeting practices. When male household members participated alongside women, the intervention further enhanced women’s financial autonomy and promoted more active joint decision-making over household finances. These findings demonstrate the potential of contextually grounded digital interventions to expand women’s financial inclusion and highlight the value of household engagement in reinforcing women’s economic agency.Empowerment for Whom? Inheritance, Gender Gaps, and Growth (E0, O1)
Abstract
Countries with larger gender inheritance gaps are also poorer. Would enforcing equal inheritance promote growth? I study the development and distributional effects of equalizing parental bequests among children using a quantitative model of land inheritance, disciplined by novel empirical evidence from Benin’s 2004 reform granting women equal rights to parental bequests. I show that equal inheritance increases production and welfare, yet family behavior responses toward inheritance change widen gender gaps in education and labor supply. Parents compensate men's reduced bequests through increasing education, which strengthens men’s comparative advantage in wage job and facilitates participation. This expands output but crowds women, who don't have physical advantage, back to agriculture, thus raising misallocation. Despite long-run welfare gains, women incur short-run welfare losses due to their dual roles: empowerment benefits women as individuals; but as parents, banning gender-biased inheritance constraints women from allocating bequests according to their preferences.Energy Shocks, Consumption Inequality & Fiscal Policy Design (E5, E3)
Abstract
Supply-driven energy price shocks generate substantial consumption inequality. Using household expenditure data and instrumental variable local projections, we show that a one percentage point energy price increase reduces consumption of households in the lowest income decile by 0.5 percent, while consumption of those in the highest decile remains unchanged. This differential response persists for up to one year and reflects both higher energy expenditure shares and tighter liquidity constraints among low-income households. A two-agent New Keynesian model with two production sectors (energy and non-energy) and non-homothetic household preferences replicates these patterns and provides a framework for policy evaluation. At comparable fiscal cost (0.08 percent of GDP over two years), targeted cash transfers fully offset the consumption inequality increase, while energy subsidies reduce it by only half. Cash transfers dominate because they enable constrained households to allocate resources optimally across all consumption categories, whereas subsidies are restricted to energy expenditure. These findings inform the design of fiscal responses to relative price shocks affecting necessity goods.Equity vs. Subsidy: The Rise of Government Venture Capital in China’s Industrial Policy (O2, O1)
Abstract
We compare government venture capital (VC), an equity-based industrial policy instrument, with traditional subsidies. We exploit China’s 2014 Budget Law reform, which tightened local fiscal constraints, limited local governments’ ability to sustain subsidies, and induced a shift toward government VC. Combining this policy shock with cross-city variation in pre-reform subsidy intensity, we study how switching from subsidies to government venture capital affects firm dynamics, innovation, and private capital participation. Using a newly constructed dataset covering the universe of VC firms, funds, and investment deals in China, we document three facts on the landscape of government VC growth episode in China. First, government VC fundraising and investment increased sharply after 2014 and eventually outpaced private VC. Second, government VC and private VC exhibit similar industry allocation patterns, but government VC is more concentrated in manufacturing. Third, government VC invests in larger and more mature firms than private VC, consistent with a more conservative risk profile. Employing a continuous Difference-in-Differences design, we confirmed that cities with higher pre-reform subsidy intensity experienced larger post-reform increases in government VC investment, consistent with more exposed local governments substituting away from subsidies and toward equity-based support. This increase was broad-based, spanning both early- and late-stage investments as well as syndicated and standalone deals, suggesting that governments used VC not only to replace direct transfers but also to leverage private capital and investment expertise. Evaluating firm dynamic outcomes, we find that compared to subsidies, government VC reduced firm entry but increased entrant quality, as measured by registered capital. It also improved innovation outcomes, especially patents granted and citation-weighted patents, and crowded in private VC investment. These findings show that the form of industrial policy matters: compared with subsidies, equity-based interventions are more selective and generate distinct effects on entry, innovation, and private capital mobilization.Exposure to Wildfire Smoke during Pregnancy and Birth Outcomes: Evidence from Mexico (Q5, I1)
Abstract
This paper examines the effects of wildfire smoke exposure on birth outcomes in Mexico. I combine satellite-based smoke plume data with municipality-day birth records from 2012 to 2023 to estimate the impact of in utero exposure. I find that an additional smoke-exposed day per month during pregnancy reduces birth weight by 3.5 grams, birth height by 0.015 cm, gestational age by 0.14 days, and Apgar scores by 0.0022. I further find that these effects concentrate in exposure to light smoke and in the third trimester of pregnancy. These patterns are consistent with behavioral avoidance on medium- and heavy-smoke days, such as remaining indoors, and with the relatively low frequency of such days in the data, which may limit statistical power. The stronger effects in the third trimester align with this period’s critical role in fetal development. These findings highlight the importance of policies to mitigate the health impacts of climate-related air pollution on pregnant women.FDI Liberalization and Allocative Efficiency : Evidence from Multi-Product Firms in India (D2, F1)
Abstract
This paper studies how foreign direct investment (FDI) liberalization affects allocative efficiency within multi-product firms. I exploit India’s staggered industry-level liberalization in the 2000s as a quasi-natural experiment and combine CMIE firm accounts with detailed firm–product data on prices and quantities. This allows me to estimate product-level markups and marginal costs using the multi-product framework of De Loecker et al. (2016), and to directly observe how resources are reallocated across product lines within firms.The analysis is guided by the Baqaee and Farhi (2020) framework, which shows that aggregate productivity depends on the covariance between wedges (markups) and economic activity, rather than on average markups alone. In this setting, allocative efficiency improves when resources shift toward high-markup (high marginal revenue product) units and when markup dispersion declines.
Using a difference-in-differences design, I document three main findings. First, FDI liberalization leads to a broad expansion in real activity: firm–product quantities, sales, and inputs increase, while prices decline on average. Second, liberalization induces systematic within-firm reallocation: product lines with higher pre-reform markups expand disproportionately relative to low-markup lines, consistent with a relaxation of financial constraints that allows high-return activities to scale. Third, markups compress—especially among initially high-markup products—leading to a decline in dispersion.
Taken together, these results provide causal evidence that liberalization improves allocative efficiency along both margins emphasized by Baqaee and Farhi (2020): reallocating activity toward high-wedge units and reducing wedge dispersion. Importantly, average markups remain largely unchanged, implying that standard aggregate measures would miss these efficiency gains.
The paper contributes to the literature on misallocation, market power, and FDI by providing the first firm–product evidence on how policy reforms reshape within-firm allocation and the joint distribution of markups and activity, offering a new mechanism linking liberalization to aggregate productivity.
Feeling the Heat: How CEO Climate Imprinting Drives Corporate Green Innovation (Q5, O3)
Abstract
This paper investigates how executives’ physical climate experiences shape corporate green innovation. Using Large Language Models to track the spatiotemporal trajectories of CEOs in Chinese listed firms, we exploit exogenous abnormal heat shocks in a dual-temporal framework that contrasts early-career exposure with contemporaneous shocks. Early-career heat exposure produces a persistent imprinting effect: firms led by imprinted CEOs see approximately 1.5 additional high-quality green patents annually, a 121% increase, whereas contemporaneous shocks trigger only transient adaptation. Imprinted CEOs exhibit heightened climate risk awareness and stronger “Nature-Loving” preference. These psychological shifts translate into action: imprinted CEOs are more likely to be listed as inventors and to mobilize sustained R&D investment toward innovation. Heterogeneity analyses indicate that market competition and managerial ability condition these effects. Our findings provide novel evidence that decision-makers’ sensory experiences can serve as important endogenous drivers of corporate transitions toward sustainability.Femicides in the Age of Social Media. Are They a Copycat Crime? (K4, E7)
Abstract
Femicides and violence against women (VAW) are considered a social and economic emergency throughout the world. Despite the awareness of the seriousness of the phenomenon and the effort to find effective policies to combat violence against women, some gray zones still persist. In the age of social media, it seems worth investigating whether the media dissemination of news can help explain the sharp increase in this criminal behaviours.The starting point for research is that in Italy femicides and crimes against women have grown more in comparison with other kind of crimes. The sharp increases of this kind of crimes against women has not found justification in the detrimental economic condition. Three kinds of data were used to measure femicides and VAW. The official statistics made available from Italian National Institute of Statistics (ISTAT). The data made available from Italian nonprofit organization (NPO). Media coverage of this type of crime, measured extracting data from websites and newspapers.
This is the first analysis on femicides that uses the imitation theory approach to perform and econometric analysis of this phenomenon.
Based on the empirical findings it is possible to say that some imitation effects exist for femicides and crimes against women. To counter the spread of such crimes, it would be better to use social media campaigns that promote examples of positive behaviour, rather than expensive policies to repress crimes. Finally, the approach followed in this research could be useful for preventing and predicting femicides and other forms of violence against women.
Financing Corporate Risk Management: The Role of Purchase Obligations (G3, E2)
Abstract
This paper studies the role of purchase obligations in corporate risk management and financing. We argue that purchase obligations—legally binding, non-cancelable commitments to procure inputs—serve as an operational hedging device by stabilizing input costs, while simultaneously crowding out debt capacity by competing for collateral. We develop a general equilibrium model in which firms choose capital, borrowing, and purchase obligations under financial constraints and input price uncertainty. The model generates a debt capacity crowding-out effect and highlights a trade-off between hedging benefits and financing costs. Empirically, we exploit staggered mandatory disclosure of purchase obligations, collateral shocks, and exogenous supply chain disruptions. The evidence supports the model’s predictions: purchase obligations reduce borrowing capacity but decrease in response to input price risk for financially constrained firms.Fiscal Expansion and Households’ Income Inequality Expectations: A Survey Experiment (E6, D8)
Abstract
Advanced economies face rising government spending needs—driven by shocks such as COVID-19 and the energy crisis—as well as longer-term challenges like population aging and climate change. At the same time, income inequality and political polarization have increased, while trust in governments has declined, including in Germany. Although the economic and redistributive effects of fiscal expansions materialize only gradually, citizens may adjust their perceptions of inequality in the short run. The direction of these adjustments is theoretically ambiguous and likely depends on prior attitudes toward government policy.This paper examines how individuals’ perceptions of current and future income inequality respond to expansionary fiscal policy. We conduct a randomized controlled trial on a representative sample of the German population, providing respondents with information about fiscal expansion. We find that combining fact-based (numerical) information with a narrative about fiscal policy reduces inequality expectations among individuals who are ex ante dissatisfied with the government’s economic policy. The effects are particularly strong among respondents without a college degree and those with low political interest, underscoring the importance of narrative communication for groups that may benefit most from such information.
To shed light on the underlying mechanisms, we analyze revisions in respondents’ macroeconomic and personal expectations. The results indicate that individuals primarily update their beliefs about future economic growth and their own risk of job loss, suggesting that improved economic outlooks and reduced perceived vulnerability drive the observed changes in inequality expectations.
Flight to Safety: Evaluating Stablecoin’s Role as Safe Haven-Asset in DeFi Markets (G2, G4)
Abstract
This study examines the impact of Tether (USDT) on systemic liquidity across the Ethereum and Bitcoin markets, utilizing an event study approach that integrates on-chain wallet data, pricing, and financial metrics. By analyzing cryptocurrency market responses to key protocol and market-moving events, augmented by nonlinear volatility models, we identify distinct, chain-specific flight-to-safety behaviors. Our results show that USDT acts as a primary liquidity lifeline for Ethereum holders during stress, particularly among retail investors, whereas its role for Bitcoin holders is more muted and stabilizing. Notably, we find stronger flight-to-safety evidence in wrapped Bitcoin (Ethereum-based) compared to native Bitcoin, highlighting that USDT’s function is network-dependent. These findings underscore that effective crypto regulation must move beyond a monolithic approach, requiring tailored frameworks that reflect diverse investor behavior, specific network dynamics, and evolving market conditions.For Whom the Bell Thole? Employee Lawsuits and Corporate Pension Policies (K2, G2)
Abstract
This paper examines how employees’ ability to monitor defined benefit (DB) pension plans through litigation affects corporate pension policies. While litigation is commonly viewed as an important mechanism for disciplining fiduciaries, it may also constrain efficient managerial discretion. We exploit the Eighth Circuit’s decision in Thole v. U.S. Bank, which restricts employees’ legal standing to sue employers for alleged breaches of fiduciary duty. Using a difference-in-differences design and granular pension plan-level data from Form 5500 filings, we find that reduced litigation-based employee monitoring leads employers to shift pension portfolios toward equities and reduce reliance on external investment advisors. Cross-sectional tests indicate that the main effect is stronger when alternative monitoring mechanisms are weaker, the value-enhancing benefits of risk-taking are greater, and employer–employee preference divergence is larger. Moreover, we demonstrated that restricting litigation-based employee monitoring improves employer welfare, with no evidence of adverse effects on employees. Overall, our findings highlight litigation as a costly and imperfect monitoring channel and suggest that, in some settings, a more permissive regulatory approach may improve the economic efficiency of delegated asset management without harming beneficiaries.Foreign Direct Investment Selection and Costs in Monopolies (F2, F6)
Abstract
How do differences in firm productivity influence the selection of firms into foreign direct investment (FDI) in monopoly markets? How do the type of cost savings—additive (unit) versus multiplicative (proportional)—affects which firms are attracted to FDI?Standard theories, mostly under monopolistic competition, predict that the most productive (lowest marginal cost) firms engage in horizontal FDI to avoid trade costs, while less efficient firms export or serve only the domestic market. However, this paper shows that in monopolies, the selection of firms into FDI crucially depends on the form of trade or production cost savings.
With unit (additive) trade costs, the lowest-cost monopolists choose horizontal FDI, aligning with the standard prediction. In contrast, with proportional (iceberg) trade costs, higher- or intermediate-cost monopolists may be the ones who engage in horizontal FDI as these cost savings rise with marginal cost. Analogously, for vertical FDI, if foreign production reduces marginal costs by a fixed amount per unit (compared to domestic production), the lowest-cost firms select into FDI; whereas if these cost savings are proportional, higher- or intermediate-cost firms can be attracted into vertical FDI first. Thus, the ability of sufficient proportional cost savings from FDI to reverse to order of heterogeneous firms choosing FDI extends to both horizontal and vertical FDI decisions.
The paper develops a monopoly model to analyze these distinctions and demonstrates the impact of the form of trade costs and production cost reductions (unit versus proportional) on the FDI decision, with implications for FDI incentives. The findings suggest that policy makers should carefully consider the structure of cost savings when designing incentives, as they can attract different types of multinational firms depending on whether the cost reductions are per unit (additive) or proportional. Further work (in a companion paper) extends these results to oligopolies.
Friends and Firearms: Social Ties and Gun Demand After Distant Mass Shootings (R2, H8)
Abstract
Gun ownership is central to debates over household security and public safety, yet research on its determinants remains limited. This paper examines how social networks shape the decision to own a firearm. Using gun-store visits as a proxy for gun demand and Facebook social connectedness to measure network ties, we show that local gun demand significantly increases after a mass shooting in socially connected but geographically distant communities. Specifically, gun-store visits in counties that are more socially connected to shooting locations rise by 1.0–1.9\% after distant shootings, with the effect that peaks around week five before decaying. Mechanism analysis indicates that the effect operates through an information-salience channel: socially connected areas exhibit larger increases in attention to shootings and react more strongly in gun-store visits when the focal areas have no prior shooting events. Finally, heterogeneity tests reveal a stronger effect in states with looser gun-related control policies and in counties with more women, older residents, families with children, and Democrats. These findings highlight how gun ownership decisions are shaped through social networks in response to distant traumatic events.From Branches to Deserts: How Local Bank Access Shapes Credit and Economic Mobility (G5, G2)
Abstract
In recent decades, the US bank branch market has been exposed to several waves of closures, resulting in concerns about branch presence and consumer access to financial services. This paper studies the effects of banking deserts on local access to credit using banking desert data and a matched difference-in-difference research design. Compared to the matched bank-tracts with similar economic and demographic conditions, small business lending in the treatment tracts increases 0.108 million (equivalent to 2.2%) after banking desert formation. Overall, results suggest that while some localized reductions in branch access occurred, the significant reduction in the number of branches did not result in significant decreases in access to credit for households or businesses.From GDP to Ideology: The Evolution of China's Political Tournament (P0, R5)
Abstract
This paper studies how the content of China’s political tournament changed under Xi. We develop a behavioral measure of ideological loyalty based on whether local officials undertake costly policy actions that align with the central leadership’s agenda, even when doing so runs against conventional GDP-oriented incentives. Our setting is the urban land market. Using transactions near future subway stations, where local leaders possess privileged information and substantial control over allocation, we identify below-market transfers of valuable land to local government-controlled firms for welfare-oriented development as signals of ideological loyalty.We show that such loyalty signaling rises significantly under Xi. At the same time, growth-oriented behavior does not disappear: measures of upward land-price manipulation remain prevalent and, in some specifications, increase. The shift is therefore not a simple replacement of growth incentives by loyalty, but a reweighting of political incentives.
We then examine promotion outcomes. The political return to loyalty signaling rises sharply under Xi, while the return to growth-oriented signaling weakens substantially. Instrumental-variable evidence yields the same qualitative pattern. The results suggest that the local political tournament did not vanish in the Xi era; rather, it persisted while rewarding a different type of effort.
More broadly, the paper contributes to the political economy of authoritarian governance by showing that loyalty need not be measured only through pre-existing personal ties. In a centralized system, loyalty can also take the form of costly, observable policy behavior. Our findings point to a shift toward a more centralized and ideology-based form of political screening in contemporary China.
From One LATE to Another: Machine Learning and the External Validity of IV Estimates (C2, C5)
Abstract
Instrumental variable (IV) estimates identify local average treatment effects (LATEs) for complier populations, raising a central external validity question: how can a LATE estimated in one population be used to predict the LATE in another? This paper studies LATE-to-LATE extrapolation when complier composition differs across settings and asks whether flexible machine-learning methods can improve external validity by reducing functional-form misspecification and mitigating the effects of covariate shift in IV extrapolation. We compare interacted 2SLS with Double/Debiased Machine Learning (DML) and derive a bias decomposition that separates prediction error into in-sample bias from first-stage misspecification, slope error magnified by covariate shift, and a residual term due to nonlinear treatment effect heterogeneity. For baseline OrthoIV, DML eliminates the first component and reduces the second, while the linear final stage leaves the third. We then study nonlinear final-stage extensions, including spline-based specifications, which can reduce this remaining component at the cost of greater variance or weaker extrapolation stability. In Monte Carlo and Card-calibrated simulations, DML substantially lowers prediction error relative to interacted 2SLS in most designs, especially when the first stage is nonlinear, and the target population differs from the estimation sample. In an empirical application to cooking fuel use across six developing countries, DML improves out-of-sample LATE predictions in four settings. The results clarify when machine learning improves the external validity of IV estimates and when its gains are limited by nonlinear heterogeneity or by failures of the external unconfoundedness.Geopolitical Pressures and the Economic Cost of Sovereign Default: Evidence from Mexico (F4, F1)
Abstract
This paper examines how geopolitical pressures shape sovereign default risk and asset prices, exploiting Donald Trump’s rhetoric and actions toward Mexico as a natural experiment. I provide causal evidence that these geopolitical shocks produced sizable and persistent increases in Mexico’s perceived default probability. A textual analysis of President Trump’s social-media posts shows that trade-related threats—rather than migration-related comments—were the main drivers of this rise in sovereign risk. A one-basis-point Trump-induced increase in default risk led to a 5.3-cent nominal depreciation of the peso and a 0.11 percent decline in Mexican stock returns, with the tradable and services sectors most affected. At the macro level, the heightened risk perception reduced aggregate consumption by roughly 0.32 percent during 2017–2018, and filtering estimates imply that consumption falls by 0.02 percent within two weeks of a one-basis-point rise in default risk.GLP-1 Receptor Agonists and the Transformation of Food Demand (I1, D1)
Abstract
This paper provides new empirical evidence on how Glucagon-like peptide-1 (GLP-1) receptor agonists reshape household food demand and nutritional intake in the United States. While early evidence suggests these medications reduce aggregate caloric intake, the product-level substitution patterns and the underlying mechanisms remain under-explored. We leverage a novel, high-frequency dataset linking the NielsenIQ Consumer Panel with the NielsenIQ GLP-1 Panel Survey, covering approximately 40,000 households. This allows us to map transaction-level records to detailed UPC-level product attributes, including package size, NOVA processing classifications, and Syndigo Nutrition Facts labels. To address the non-random adoption of GLP-1 therapies, we employ a staggered difference-in-differences (DID) design, utilizing nearest-neighbor matching to construct a counterfactual group of non-adopting households. Our preliminary findings indicate that GLP-1 adoption leads to a significant reduction in overall food expenditures, particularly within high-calorie categories. Crucially, we document significant intra-category heterogeneity: the decline is concentrated among products with larger package sizes, while the probability of purchasing items with health-related claims increases. These results suggest that GLP-1 medications induce dietary improvements through both inter-category reallocation and the selection of superior nutritional attributes within categories. This study contributes to the literature on the determinants of food demand by offering some of the first product-level evidence of how medical shocks to obesity influence consumer substitution. By identifying previously unobserved shifts in package size preferences and health-claim responsiveness, we provide essential insights for understanding the broader effects of weight-loss pharmacotherapy on the food industry and public health.Gun Violence in School and Teacher Workforce: Impacts on Mobility, Retention, and Teacher Quality (H0, J0)
Abstract
Teacher shortages remain a critical policy concern in the United States, with an annual attrition rate of approximately 8%. Although existing literature has examined how factors such as low compensation, poor working conditions, and student discipline contribute to teacher turnover, the effect of school insecurity, particularly school shootings, on teacher mobility remains largely understudied. This paper fills that gap by estimating the causal impact of school shootings on teacher retention and mobility in North Carolina public schools.The study combines two primary data sources: school-level shooting incident data from the Center for Homeland Defense and Security (CHDS), spanning 1966 to 2023, and individual-level longitudinal teacher employment records from the North Carolina Education Research Data Center (NCERDC), covering 1995 to 2023 and including teacher characteristics, mobility, and working conditions.
The empirical strategy employs a difference-in-differences and event-study framework, treating school shootings as exogenous shocks to the teacher workforce. Each treated school is matched to its two nearest comparable schools of the same level outside the treated district, mitigating potential spillover effects. Year and district fixed effects are also included to account for systematic differences in district characteristics, such as poverty levels.
The analysis finds that schools exposed to shootings experienced an average reduction of approximately 6.37 full-time equivalent teachers relative to matched non-exposed schools, capturing both teacher attrition and mobility to other schools. Event-study estimates further reveal that teacher mobility begins rising two years post-shooting and does not recover within seven years. Heterogeneous analysis indicates that these effects are concentrated among the most severe incidents involving deaths or injuries.
Future work will examine heterogeneous effects by school type, poverty level, and teacher demographics, as well as potential spillover impacts on neighboring schools, with the goal of informing school security and teacher shortage policy.
Heat as a Disamenity in Urban India: A Rosen-Roback Spatial Equilibrium Approach (R1, O1)
Abstract
Extreme heat events pose serious risks in India, a country whose population is rapidly urbanizing. We quantify urban Indian residents' marginal willingness to pay for heat abatement using a Rosen-Roback framework with mobility costs. We use hourly weather station data to generate high-frequency metrics of heat stress, then combine this with Indian census data on wages and housing costs to estimate general equilibrium effects of higher temperatures. We estimate the model across income quartiles to highlight heterogeneous responses by households. In counterfactual exercises, we estimate households' ability to adapt to higher temperatures by purchasing air conditioning, as well as other technologies aimed at reducing urban heat island effects.Helping Hands, Healthier Infants: The Effect of Medicaid Doula Coverage Mandates on Birth Outcomes (I1, H5)
Abstract
Medicaid finances a large share of U.S. births, making state Medicaid policy an important potential lever for improving infant health. This paper studies whether Medicaid doula coverage mandates improve birth outcomes. Doulas are trained non-clinical birth supporters who provide emotional, informational, and physical support during pregnancy, delivery, and the postpartum period. To date, 26 states have adopted Medicaid doula coverage, with staggered implementation across years creating useful policy variation for identification. Using CDC WONDER natality data from 2016–2024, I construct a state-year panel covering 51 states, 107,254 aggregated observations, and more than 33 million births. The main outcomes are low birth weight (LBW) and very low birth weight (VLBW).I estimate weighted difference-in-differences models with state and year fixed effects and a one-year treatment lag. Event-study estimates show pre-treatment coefficients close to zero and broadly similar pre-trends between treated and never-treated states, supporting the empirical design while also indicating limited precision in the aggregate effects.
The most meaningful results appear in subgroup analyses. Heterogeneity estimates suggest larger benefits for disadvantaged populations, especially Black mothers with lower education. For example, the estimates imply about a 2.7 percentage-point decline in low birth weight and a 0.7 percentage-point decline in very low birth weight for Black mothers relative to the omitted group. These findings suggest Medicaid-financed doula care may help reduce disparities in birth outcomes even when average aggregate effects are modest. I have also recently been approved to access restricted individual-level natality data, which will allow richer covariates and additional outcomes, including infant mortality, to strengthen and extend the analysis.
Here are my recommendations: Paper Title: The Speed of Aging and the Asset Composition Trap: Housing, Pensions, and Fiscal Sustainability (H5, R2)
Abstract
Developing economies are aging faster than advanced economies did, and their households hold the majority of wealth in housing rather than diversified financial assets. I develop a 55-period overlapping generations model with endogenous labor supply, a pension system with government debt, and household portfolio choice between productive capital and housing to study how the speed of demographic transition interacts with wealth composition to shape optimal pension policy and fiscal sustainability. Housing enters as both a consumption good and a store of value, with a construction sector that generates the Glaeser and Gyourko (2005) durability asymmetry: shrinking household-formation cohorts depress housing prices because the durable stock cannot shrink quickly. This creates a "double exposure": the public pension faces a rising dependency ratio while the dominant private savings vehicle simultaneously loses value. I establish four results. First, a two-period analytical model shows that the effective return on funded pension savings converges toward the population growth rate as the housing portfolio share increases, eliminating the diversification benefit of shifting from pay-as-you-go to funded accounts. Second, the speed of aging amplifies this channel nonlinearly, generating a "window of vulnerability" in which compressed transitions produce welfare losses that gradual transitions do not. Third, I characterize the fiscal cost of smoothing the demographic transition through government borrowing: maintaining stable tax rates and benefits requires increasing debt-to-GDP ratios during the transition, with the burden falling disproportionately on post-transition cohorts. The debt accumulation is larger in housing-heavy economies because depressed asset prices erode the tax base and reduce the capital stock available for future output. Fourth, calibrating the model to the United States and Colombia, I show that the standard recommendation of expanding funded individual accounts reverses in fast-aging, housing-heavy economies: a minimum public pension floor is welfare-improving precisely where fiscal space is most constrained.Heterogeneous Preferences and the Insurance Effect of Income Redistribution (E2, H2)
Abstract
This study evaluates how income redistribution, implemented through progressive income taxation and the social security system, functions as an insurance mechanism against income risk. In particular, we examine the extent to which redistribution mitigates fluctuations in lifetime income arising from idiosyncratic income shocks. A central challenge in measuring this insurance effect is separating income variability driven by differences in individual preferences from variability due purely to stochastic shocks. Ignoring such heterogeneity may lead to an overstatement of the insurance role of redistribution.To address this issue, we develop a two-earner life-cycle model that incorporates heterogeneous preferences over time and work effort. In the model, households make dynamic decisions regarding consumption, labor supply, and asset accumulation, while facing income uncertainty and opportunities for human capital investment. We estimate the model using data from the Japan Household Panel Survey (JHPS/KHPS), which uniquely contains survey-based measures of individual preferences. These data allow us to identify the distribution of time preferences across individuals. The model is estimated using the Simulated Method of Moments to match key empirical moments related to income, labor supply, and preference measures.
Using the estimated model, we conduct counterfactual simulations that vary the degree of tax progressivity while holding total tax revenue constant. Our results show that the reduction in the interquartile range (IQR) of lifetime income within preference types is only about half of the reduction observed in the overall (unconditional) distribution. This finding implies that conventional approaches, which ignore preference heterogeneity, substantially overestimate the insurance effect of redistribution. Furthermore, we find that the welfare impact of increased tax progressivity is heterogeneous: individuals with lower discount rates and lower disutility of work benefit less from the insurance provided by redistribution. These results highlight the importance of accounting for preference heterogeneity when evaluating redistributive policies.
Higher Education and Adult Health: Evidence from China’s College Entrance Exam Suspension (I1, I2)
Abstract
While the correlation between education and health is well-established, causal evidence on the role of higher education in producing health in developing countries remains scarce. This study exploits a natural experiment in China, where high school seniors between 1966 and 1969 were denied timely access to higher education due to the suspension of the national college entrance examination (NCEE), to study the causal effects of college completion on health. Employing a fuzzy regression discontinuity design, I find that the NCEE suspension resulted in a 29.3% reduction in college completion for the disrupted cohorts by age 50 compared to unaffected cohorts. This lost educational opportunity led to significant and lasting adverse health consequences. Estimates imply that a one percentage-point decrease in the cumulative college completion rate increases the likelihood of smoking and drinking in later life by 1.1 and 2.2 percentage points, respectively. Further analyses suggest that these effects are partly driven by lost labor market returns, are concentrated in provinces where college disruption was most acute, and are absent for later cohorts who faced milder restrictions on access to higher education. Together, the results point to higher education as a critical determinant of population health, with implications for both education policy and health equity in resource-constrained settings.Homeland Echoes: The Persistent Grip of Origin Culture on Immigrants (J1, D1)
Abstract
Do origin-country policy changes echo in the behavior of immigrants who live under a different legal regime? We study Chinese immigrants in the United States around China's staged relaxations of birth limits between 2011 and 2016, culminating in the universal Two-Child Policy. Using complete U.S. Vital Statistics for 2004--2019 and difference-in-differences models, we compare births to Mainland Chinese mothers with births to immigrants from Hong Kong, Macau, and Taiwan, and we contrast first-generation and U.S.-born Chinese women. Births to first-generation Mainland Chinese mothers rise sharply in the reform windows, with pronounced increases at second-order births, earlier timing at first-order births, and shorter birth intervals. U.S.-born Chinese Americans also respond, but more modestly and mainly at first-order births and in places with larger Chinese communities. Our results provide a concrete example of how a salient, nonbinding policy shift at origin can alter the behavior of migrants living under a different legal regime, in this case the fertility intention and timing. These findings imply that evaluations of major origin-country reforms in settings with large diasporas may need to account for behavioral responses among migrants abroad, not only domestic outcomes.Housing Shocks and Local Elections: Evidence from Short-Term Rentals in Italy (R2, D7)
Abstract
This paper provides causal evidence on how platform-induced tourism reshapes housing markets, inequality, and electoral outcomes. We study the expansion of short-term rental platforms such as Airbnb across Italian cities and show that platform growth generates both housing-market pressures and politically consequential distributional changes. Combining granular data on Airbnb activity with administrative records on housing, income, and precinct-level voting, we exploit quasi-experimental variation in local exposure to identify these effects. We conceptualise short-term rentals as a dual shock: a reallocation of housing from long-term use to tourist accommodation, and a redistribution of income toward property owners. To address endogeneity, we implement a shift–share (Bartik-style) instrumental variable strategy that interacts pre-determined tourist attractiveness (monument density) with national variation in Airbnb demand. Preliminary results show that greater Airbnb penetration increases housing prices and raises top income shares. These effects are accompanied by significant electoral responses: turnout declines, while support rises for anti-establishment and right-wing parties, particularly the Five Star Movement and Fratelli d’Italia, with smaller increases for social-democratic parties. These findings highlight how platform-driven changes in urban economies can amplify inequality and reshape political behaviour.How Do Incentive Policies Affect the Adoption of Electric Vehicles in the Ride-hailing Industry? Evidence from Chicago (R4, Q5)
Abstract
Transportation Network Company (TNC) drivers, with annual mileage roughly three times the national average, represent a particularly cost-effective target for vehicle electrification. Yet most EV incentive programs treat all buyers identically, and little is known about how these high-utilization commercial drivers respond to broad-based rebates. This paper examines how the Illinois EV Rebate Program—a $4,000 purchase rebate introduced in July 2022—affected electric vehicle adoption among Chicago ride-hailing drivers, the third-largest U.S. TNC market.I use administrative data from the City of Chicago covering active TNC vehicles in Chicago, collected monthly from 2015 to 2025. A regression discontinuity in time (RDiT) design estimates a statistically significant 0.85 percentage-point increase in BEV share at the rebate's introduction, a 62% jump relative to the pre-policy baseline of 1.36%. This provides evidence that the rebate accelerated electrification beyond the pre-existing secular trend. To move beyond the reduced-form evidence and recover the structural parameters needed for policy evaluation, I estimate a random-coefficients logit (BLP) demand model over almost 1,000 differentiated vehicle models, using monthly entry flows to construct market shares and BLP-style instruments to address price endogeneity.
The recovered preference parameters would enable counterfactual simulations comparing TNC-targeted rebates of varying generosity against the current untargeted program, as well as alternative policy designs. For each scenario, I assess cost-effectiveness by comparing program expenditures (including transfers to inframarginal buyers) to the social value of induced emissions reductions. I also evaluate equity implications, given that the current program prioritizes low-income applicants and that operating cost savings from EVs accrue disproportionately to high-mileage drivers. The findings inform the design of electrification policies for ride-hailing fleets across jurisdictions and, more broadly, illustrate how incentive design interacts with vehicle utilization intensity, with implications for cost-effective decarbonization of urban transportation beyond the ride-hailing context.
How Does the Climate Change Affect Household Consumption? Evidence from China (Q5, D1)
Abstract
This study examines climate change’s effects on Chinese household consumption by using four waves of China Family Panel Survey(CFPS) data(2016–2022) and meteorological data from the U.S. NOAA. Our findings indicate that climate change significantly reduces household spending, with greater effects on rural households, those with poor health, lower education, and lower wages. This mainly caused by diminishing household income, and dampening future expectations, but adaptive behaviors help mitigate it. Heterogeneity analysis shows that persistent climate change exacerbates consumption declines, thoughgreater adaptive capacity alleviates its impact. Additionally, population concentration intensifies climate-related risks through heat island and rain island effects, whereas industrial clustering mitigates these negative impacts.
How Shareholder Voice Shapes Board Agendas (G3)
Abstract
Board voting offers an incomplete picture of director influence, overlooking board agenda-setting and board–shareholder meeting interactions. Using unique corporate disclosures in China, we examine the full decision process—from board agenda formation to shareholder voting. Minority blockholder-appointed directors connect board decisions to shareholder voting by bringing minority investor voting power into board deliberations. Exploiting a reform that increases dissent visibility at shareholder meetings, we find stronger shareholder voice amplifies these directors’ influence on agenda setting. We document that tunneling proposals reach board agendas less frequently, especially when shareholder approval is required. As public dissent at board meetings becomes costlier, directors increasingly monitor through agenda-setting rather than by board voting.Human–AI Collaboration as a Dynamic Bayesian Game (C7, D8)
Abstract
We introduce a game-theoretic framework for modeling human–AI collaboration as a principal–agent problem with incomplete information and dynamic feedback. A human principal delegates decision-making to an AI agent but must design incentives and decision rules to align the AI’s actions (effort exertion and information reporting) with the human’s objectives. We formalize this interaction as a Bayesian game in which the AI (agent) privately knows its competence and chooses an effort level and whether to report a prediction with confidence or to abstain, while the human (principal) commits to a mechanism consisting of a payment/penalty contract and an acceptance (delegation) rule. The mechanism employs proper scoring rules to incentivize truthful probability reports, penalties for over-confident errors to discourage AI ”hallucination,” and an abstention option with a baseline reward to encourage the agent to defer when unsure. We characterize the equilibrium of the one-shot game, showing that under the proposed mechanism the AI’s best response is to report its uncertainty truthfully and to choose an optimal effort level balancing score gains and effort costs, while the human’s acceptance rule (e.g. threshold on predicted risk) optimizes delegation decisions. Extending the model to a repeated interaction, we incorporate a reliability index and dynamic contract terms (e.g. reputation-weighted bonuses and a budget for evidence requests) that evolve based on the AI’s performance over time. We discuss how this dynamic game can sustain long-term incentives for the AI to remain well-calibrated and refrain from strategic manipulation. Our framework bridges mechanism design, principal–agent theory, and recent ideas in cooperative AI, illustrating how classical incentive techniques (Myerson’s revelation principle, Holmstr¨om’s effort incentives) can be applied to modern human–AI systems. We provide examples connecting our model to prior work in uncertain forecasting, AI delegation (e.g. off-switch games), and learning to defer to humans.Identifying Demand with Network Externalities (L1, C5)
Abstract
This paper studies the identification of demand models with network externalities. In many markets, a product's appeal depends on the size of its user base: users adopt messaging platforms to communicate with others, join dating websites to access potential matches, and may avoid popular transit modes due to congestion. Such network externalities shape competition, welfare, and the effects of regulation and mergers.The article's main contribution is methodological: I show that demand models with network externalities are generally not identifiable with market-level data alone, as the direct effects of product characteristics on market shares cannot be disentangled from feedback effects owing to network externalities. One solution involves exploiting across-market variation in market size, which may determine the scale of network externalities, although this approach requires assuming the market size does not directly affect preferences. Micro data linking consumers' choices and characteristics offers a more credible path to identification: within-market variation in consumer characteristics permits identification of patterns of substitution between alternatives whereas across-market variation in market shares permits identification of network externalities. I characterize the instrumental variables for endogenous market shares that may be used to identify network externalities, highlighting the potential of Waldfogel instruments exploiting across-market differences in local demographics.
Guided by the identification analysis, I estimate the model on the US dating websites industry using browsing micro data. The estimates imply substantial network effects, explaining variation in market shares across cities despite limited variation in site characteristics. I use the model to study mergers that make co-owned websites interoperable, thus allowing users of one site to interact with users of another. Under interoperability, a price reduction on one site attracts users who become accessible across all co-owned sites, generating positive network spillovers. Joint pricing leads the merged firm to internalize these spillovers, encouraging price reductions that benefit consumers.
Identifying Hidden Mathematical Talent: Experimental Evidence from Rural China (I2, J2)
Abstract
The under-production of STEM talent in rural areas represents a severe misallocation of human capital. While the literature frequently attributes this gap to informational frictions, we investigate whether deeper structural barriers prevent the realization of mathematical potential. We conduct a large-scale field experiment with 11,419 middle school students across 22 schools in rural China. The sample was bifurcated by prior achievement: low-achievers received factual data on the wage premium for quantitative skills, while high-achievers were randomized into a 2x2 factorial design varying exposure to effort requirements, institutional benefits, and role models.We observe precise null effects across both experiments; intensive informational nudges failed to shift academic aspirations or behavioral engagement with high-stakes mathematical challenges. This indicates that information is not the primary roadblock. Instead, by deploying an objective mathematical diagnostic scaled via Item Response Theory (IRT), we isolate latent cognitive ability from noisy teacher-assigned administrative grades to reveal a profound structural failure of talent identification.
We find that over 40% of the 447 students demonstrating elite (top-quartile) mathematical logic remain "hidden" from the traditional tracking system. This misclassification is driven by severe institutional and psychological frictions: 20.1% of elite scorers are penalized by low administrative grades ("Institutional Underdogs"), and 30.6% suffer from a profound lack of self-efficacy (the "Confidence Gap"). Most critically, we identify a subset of 33 "Invisible Geniuses" who suffer a double penalty of both institutional and psychological misclassification. We conclude that capturing rural human capital requires a policy shift from high-stakes tracking to objective, low-stakes diagnostic platforms.
Imperfect Information and Slow Recoveries in the Labor Market (E3, E7)
Abstract
The unemployment rate remains elevated long after recessions, a persistence standard search-and-matching models fail to explain. I show that noise shocks—expectational errors from noisy signals about productivity—constitute a novel channel generating this sluggishness. Using a structural VAR, I find unemployment would have recovered six quarters earlier absent such shocks. To interpret this evidence, I introduce imperfect information in a search-and-matching model, which successfully replicates the observed recoveries. Persistence arises through two channels: slow learning amplifies the effects of persistent productivity shocks, while noise shocks provide an additional, independent source of sluggishness—highlighting information frictions as an important driver of slow recoveries.Imprecise Beliefs About Malaria Infection Risk and Vaccine Protection in Ghana (O1, D8)
Abstract
This paper studies how mothers respond when a new multidose vaccine changes both expected protection and uncertainty about that protection. I use door-to-door survey data from 4,383 mothers in Ghana’s Oti Region during the malaria-vaccine rollout. For four vaccination states, current conditions, no doses, one dose, and all doses, respondents report a best guess of malaria infection risk and, if they are not fully certain, the lowest and highest risk they consider possible. This allows me to measure both central beliefs and belief imprecision.I relate these belief objects to three outcome families: general malaria preventive behavior, malaria-vaccine engagement, and booklet-recorded timing outcomes. Three results emerge. First, non-zero belief ranges are common: roughly half of respondents report a non-zero range in a given scenario. Second, greater belief imprecision is strongly associated with planning, schedule tracking, and other forms of vaccine engagement, but less clearly associated with booklet-recorded follow-through. Third, when I compare adjacent dose transitions, perceived reductions in malaria risk matter more for behavior than reductions in uncertainty alone. The evidence points more strongly to perceived protection gains than to a dominant role for declining uncertainty.
The paper contributes to work on subjective beliefs, health behavior, and vaccine uptake by showing that imprecise beliefs are empirically important in a low-income-country health setting and can be measured directly during a live multidose vaccine rollout.
Individualism and Economic Development: Evidence from Traditional Songs (Z1, O1)
Abstract
This paper studies whether cultural individualism affects economic development. We construct a new measure of individualism from traditional song structure, drawing on systematically coded musical features that capture social coordination, conformity, and individual expression across ethnolinguistic groups worldwide. Because traditional songs are culturally transmitted and largely predate modern economic development, this measure provides a novel and historically anchored proxy for inherited cultural values. We merge the song-based individualism index with high-resolution satellite nighttime luminosity to study economic activity at the pixel level. In cross-sectional and within-country analyses, more individualistic regions exhibit significantly higher levels of economic development, even after controlling for geography, population density, subsistence patterns, and political hierarchy. To address endogeneity concerns, we exploit discontinuities in individualism at ethnolinguistic boundaries within countries using a spatial regression discontinuity design. Areas that differ sharply in inherited individualism but are otherwise geographically similar display discontinuous differences in economic development. The results provide the new a within-country, quasi-experimental evidence that inherited individualism is a fundamental cultural determinant of long-run economic development.Infectious Diseases and Inequality (E1, I1)
Abstract
We develop a dynamic general equilibrium model in which heterogeneous agents face an endogenous health shock governed by epidemiological dynamics. Unlike standard productivity risk, this shock affects both labour income and utility directly and can be partially insured through private prevention. Individuals self-insure along two margins — asset accumulation and direct risk reduction behaviour. These decentralized choices feed back into the aggregate evolution of risk. Thus, the idiosyncratic risk differs from the one in canonical heterogeneous-agent macroeconomic models which cannot be affected by individual choices. Estimating the model to match key UK moments, we show that, absent short-run mitigation policy, pandemics can compress the wealth distribution even as they worsen health and income inequality. The mechanism is a two-margin insurance problem: wealthier households primarily insure through prevention, whereas low-wealth households respond through precautionary saving, leading to faster asset accumulation at the bottom of the distribution. In transitional dynamics, whether pandemics compress or amplify wealth inequality depends on how strongly mitigation policies disrupt production. This mechanism helps explain cross-country differences in post-pandemic wealth inequality dynamics.Inflation Expectations Volatility and Trade Balance Dynamics (E3, F4)
Abstract
Inflation expectations are central to monetary policy, yet their stability has received limited attention beyond domestic outcomes. This paper shows that inflation expectations volatility is an important driver of trade balance (TB) volatility and identifies the dominant transmission channel through competing theoretical mechanisms and panel evidence. Two channels generate opposing predictions. The credibility channel predicts that unanchored expectations compel aggressive monetary tightening, generating interest rate and exchange rate volatility that destabilizes trade flows. The exposure channel predicts that well-anchored expectations deepen trade integration and reduce precautionary savings, increasing the economy's sensitivity to external shocks. The paper tests these predictions using a quarterly panel of 30 economies over 1990–2019 and finds that the credibility channel dominates. A one-standard-deviation expectations volatility shock raises TB volatility by 5% of its mean and generates a persistent trade deficit of 0.23 percentage points of GDP. To address endogeneity, the paper exploits the adoption of inflation targeting(IT) as an exogenous stabilization episode. Reduced-form local projections show post-adoption TB volatility declines by approximately 25%. This is consistent with the credibility channel, as IT anchors expectations and attenuates the monetary transmission that destabilizes trade flows. Heterogeneity analyses stratifying by development level, exchange rate regime, and exporter type broadly support the credibility channel. To discipline the transmission mechanism, the paper employs a small open-economy New Keynesian model with a stochastic Taylor rule coefficient. The credibility channel is isolated by sweeping policy aggressiveness: more hawkish policy simultaneously anchors expectations and stabilizes TB. The exposure channel is tested by varying the import share, proxying the deeper trade integration that stable expectations are assumed to encourage. The exposure channel fails structurally: greater import share amplifies CPI pass-through, triggering a stronger endogenous Taylor rule response that offsets the direct exposure effect. These results establish central bank credibility as a stabilizer of external balances.Inspections as Soft Signaling: Central–Local Learning in China’s Environmental Campaigns (Q5, H7)
Abstract
China’s nationwide environmental inspections were designed to strengthen local environmental enforcement through high-profile central scrutiny. This paper argues that when inspections are not backed by consistent punishment or durable follow-up, they instead operate as soft signals: local governments and regulated firms learn that compliance can be temporary and optimally reduce effort in subsequent rounds. I formalize this mechanism in a repeated central–local signaling game with learning, in which observed sanctions, legal actions, and post-inspection pollution rebound update beliefs about the Center’s effective “toughness.” A simple calibration, disciplined by observed enforcement intensity and the persistence of pollution changes, provides quantitative benchmarks for the level of punishment needed to sustain compliance.Empirically, I construct a city-week panel of ambient air pollution from China’s national monitoring network, aggregated to the city level and validated against satellite-based pollution measures. I match these outcomes to the staggered timing of the original inspection rounds (2015–2017) and the “lookback” round (2018). I estimate dynamic impacts using a stacked event-study design indexed in weeks relative to the on-site inspection window, with city fixed effects and calendar-week fixed effects. To test the learning mechanism, I relate lookback responses to measures of first-round enforcement and follow-up, including rectification orders, shutdowns, filed cases, fines, and cadre discipline. I additionally examine whether the Center’s strategic reorganization in later rounds—shifting oversight and punishment toward selected targets—restores credibility and yields more persistent pollution reductions. Finally, I link inspection intensity to downstream political and legal consequences, including local officials’ career outcomes and formal environmental law actions.
Interlocking Directorates, Innovation, and the Flow of Ideas (L4, O3)
Abstract
Interlocking directorates, the practice of firms sharing board members, are illegal in the United States for competing companies. Yet, a large fraction of public firms in the US engaged in this practice in the past due to weak antitrust enforcement. In this paper, I quantify how horizontal interlocks affect innovation. To achieve this, I propose a new identification strategy that relies on selection on observables after estimating and controlling for board member ability through an AKM model. I find that Section 8--violating horizontal interlocks are associated with higher innovative output. However, this extra output is not matched by higher-value patents or larger R&D spending. I also show event study evidence consistent with knowledge transmission: when two firms interlock for the first time, they cite one another’s patents more often thereafter. Sensitivity analyses imply that unobserved factors would need to be much stronger than the rich set of observed controls to overturn the results, making omitted variable bias an unlikely driver of these results. Collectively, the results suggest that horizontal interlocks operate primarily as a conduit for idea flow between competitors rather than broad expansions in R&D spending, and that these flows do not translate into higher-quality innovation.Inventories and Merger (In)efficiencies: Evidence from the Petroleum Industry (L4, Q4)
Abstract
Despite the prevalence and importance of inventory in many markets, it is largely ignored in competition policy. In this paper, I develop a dynamic model and a corresponding static approximation of the steady-state in which refineries optimally choose inventories of both inputs and outputs, production, and wholesale sales quantities in the presence of crude oil price uncertainty and demand seasonality. The model highlights how vertical integration reshapes intertemporal tradeoffs: it not only eliminates double markups (the standard efficiency argument) but also the need for costly precautionary crude oil storage, generating a new inventory channel for merger efficiencies. To test the predictions of the model, I compile a novel database of refinery-product-level inventories, production, and sales for Texas from 2014-2025 by digitizing regulatory filings with large language models. I document that refineries build up crude oil inventories when expected prices are higher and oil prices are more volatile, use gasoline inventories to smooth demand across seasons, and draw down output inventories following unplanned outages. Following vertical mergers, refineries decrease crude oil inventories by 9.3 percent, implying annual cost savings of 0.41-1.10 billion dollars through the inventory channel alone. This accounts for 25-50 percent of the reported cost efficiencies from large, vertical mergers. I estimate a structural model of wholesale petroleum demand and refinery decisions in which firms hold costly inventories. Preliminary estimates suggest there are convex storage costs. The model can be used to quantify the welfare implications of vertical mergers, evaluate passthrough of oil price levels and volatility, and consider alternative policies, such as inventory mandates and subsidies, to alleviate dynamic inefficiencies due to insufficient inventories.Is ESG Assurance a Genuine Signal of Sustainability？ (M1, Q5)
Abstract
Sustainable investment is a growing trend, and yet it is a challenge to identify truly sustainable firms, firms that exert real efforts to improve their sustainable performance (doing good) and to avoid negative impacts (avoiding bad). We address this challenge by introducing novel measures based on assurance of material and immaterial issues in CSR reports. These assurances are provided by third-party auditing companies. To construct these measures, we extract ESG assurance data from STOXX Europe 600 firms from 2017 to 2022. We classify assured issues using the Sustainability Accounting Standards Board (SASB) framework as an external benchmark and develop firm-level ratios that capture the extent to which assurance is concentrated on material versus immaterial sustainability topics. We find that firms with higher material assurance ratios (Rm) are granted more green patents and face fewer violations of sustainability-related regulation, whereas immaterial assurance (Ri) is associated with the opposite pattern. These results are robust after controlling for ESG scores and greenwashing indicators, and to alternative constructions of the assurance measures and alternative innovation outcomes. We further find that higher material assurance ratios predict subsequent improvements in environmental performance. Finally, we examine what shapes how firms allocate assurance across material and immaterial issues and find that stakeholder engagement plays an important role in both Rm and Ri, while sustainability-linked compensation is more weakly related to material-focused assurance. Overall, our findings suggest that the content of ESG assurance carries incremental information beyond the simple presence of assurance and helps distinguish substantive from symbolic sustainability practices. Our measures, therefore, provide new insights into how firms signal sustainability, as well as a practical tool for identifying sustainable firms and avoiding capital misallocation to firms that engage in symbolic sustainability.Kingmaking: Venture Capital in the AI Era (G1, G3)
Abstract
This paper develops a theory of valuation dynamics in venture capital markets characterized by megafund concentration and winner-take-all technology competition. Adapting the Inelastic Market Hypothesis to private capital markets, I derive deployment pressure from the following constraints: agency problems between limited partners and general partners, reputation concerns that create time-inconsistency in optimal contracting, and coordination failures among institutional investors. These constraints create inelastic demand curves for startup equity, potentially inflating valuations above fundamental value. I formalize the \textit{kingmaking} phenomenon (in which companies receive valuations exceeding one billion dollars pre-revenue) and distinguish contemporary patterns, where kingmaking occurs at inception, from historical precedents where it occurred during commercialization. The model generates discriminating cross-sectional predictions: under the inefficiency hypothesis, deployment pressure correlates with capital overhang and predicts risk-adjusted underperformance; under the efficiency hypothesis, it reflects rational pricing of winner-take-all option value and carries no predictive power after controlling for systematic risk. A winner's curse extension establishes that heterogeneous beliefs and deployment pressure are complements in generating overvaluation. Welfare analysis derives conditions under which megafund concentration dominates fragmented capital allocation: when computational requirements exceed the syndication capacity of smaller funds, concentrated capital is welfare-enhancing despite high failure rates. The framework generates testable predictions that distinguish market failure from efficient adaptation to technological change.Labor Market Power and Innovation (O3, L1)
Abstract
This paper examines how labor market power shapes how firms innovate and grow. We develop an endogenous growth model where firms optimize R&D spending to increase their future productivity while facing an upward-sloping labor supply curve, generating monopsony power. This creates two opposing distortions: (1) monopsonistic firms have stronger incentives to innovate and grow as they enjoy larger profits, but (2) firm growth increases (infra-)marginal labor costs by pushing firms up the labor supply curve, which reduces the returns to productivity-enhancing innovation. Theoretically, the first effect dominates for small firms, while the second is stronger for large firms. We test these predictions using rich firm-level data from the German manufacturing sector (1995-2018) to estimate firms' productivity and labor market power. Empirically, we find that, conditional on size, labor market power negatively correlates with R&D investment. Furthermore, small (large) firms in high-monopsony-power regions exhibit relatively high (low) R&D spending, compared to competitive labor markets, which aligns with our model's predictions. When applying our model to the data, we find that East Germany's higher labor market power can explain 25% of the persistent productivity gap between East and West Germany and depresses overall GDP growth by 0.3% p.a.Leaving Out the Elderly: Mistargeting of Bangladesh’s Old Age Allowance (D3, H5)
Abstract
Social protection programs are designed to reduce poverty and vulnerability (Coady et al., 2004), but the decentralized nature of such programs in developing countries make beneficiary allocation susceptible to discretion, political interference, and capture (Bardhan & Mookherjee, 2006; Niehaus et al., 2013; Panda, 2019; Asri, 2019; Asri et al., 2020; Banerjee et al. 2024). With weak institutions, an autocratic (former) government, a large informal labor force, and a rising elderly population, provision of social protection for the elderly is crucial in Bangladesh but is vulnerable to corruption and political manipulation (ODI, 2018; Transparency International Bangladesh, 2023). I examine Bangladesh’s Old Age Allowance (OAA) program, and find that approximately 45% of current beneficiaries are ineligible. I find political connections are predictive of program receipt. By contrast, reciprocity is not a predictor, as hypothesized would be the case in settings with weak democratic institutions (Feitosa, 2020; Giommoni, 2021). Furthermore, I provide evidence that mistargeting is regressive, implying potential welfare gains from reallocating transfers toward the excluded. To answer this, I do a welfare exercise to see whether giving the OAA to those deemed most eligible leads to a better outcome. I find this approach leads to an overall reduction in welfare, potentially highlighting the tradeoff between targeting on impact versus deprivation (Haushofer, 2025). I further extend the analysis to examine whether the 2021 digitization of OAA applications reduced exclusion among the previously excluded elderly, and whether patterns of mistargeting are shaped by local political incentives using national parliamentary and upazila election results.Licensing and Innovation Regimes in Pharmaceutical R&D (L1, I1)
Abstract
We study how licensing affects the allocation of innovation in pharmaceutical R&D. We develop a model in which projects differ in both quality and innovation regime, distinguishing between incremental and novel innovations. Information precision is higher for incremental projects and lower for novel ones, generating different equilibrium dynamics in the market for technology. The model predicts that licensing sustains positive selection and competitive return equalization for incremental innovation, while novel projects may exhibit weaker screening consistent with lemons-type frictions. Using product-level data and Double Machine Learning methods, we test these predictions across success probabilities and monetary returns. We find that in-licensed products exhibit 3.8–4.3 percentage point higher success probabilities overall, yet generate substantially lower net returns for incremental projects, a pattern consistent with competitive return equalization. For novel projects, the return penalty is statistically indistinguishable from zero despite the success probability advantage, signalling lemons-type frictions at the innovation frontier. Causal estimates from the DML-IV analysis confirm this heterogeneity: rushed licenses for incremental projects carry neither a success probability advantage nor a return penalty, preserving competitive non-dominance; rushed licenses for novel projects yield no success probability gain but a significant return loss of approximately $64M, confirming that information frictions are exacerbated under pipeline pressure precisely where signal precision is lowest. Our results reconcile evidence on both competitive efficiency and information frictions in markets for technologies, showing that market performance depends systematically on the type of innovation being transacted.Lines of Credit and the Bank Liquidity Rebalancing Channel: Evidence from Bank Regulatory Data (G2, G3)
Abstract
This paper provides theory and evidence that banks reallocate credit-line supply ex ante to manage liquidity risk as interest rates move. A simple model predicts that, when rates fall, banks rebalance their loan portfolios away from rate-sensitive borrowers that are more likely to draw. The mechanism is driven by a shadow cost of providing credit-line liquidity, which increases under tighter liquidity regulations. Using confidential supervisory data, we find associations consistent with this mechanism, and use a within-borrower design to establish a causal interpretation. The results are strongest for banks with larger liquidity imbalances tied to undrawn exposures, consistent with a higher shadow cost of liquidity. Rebalancing activities spill over to the real economy, depressing investment and asset growth in rate-sensitive sectors. These results uncover a liquidity-rebalancing channel of monetary policy operating through pre-committed credit lines and highlight its interaction with liquidity regulation.Litigation or Foreclosure? The Effects of Judicial Efficiency on Lending Markets (K4, D8)
Abstract
The efficiency of judicial enforcement significantly influences credit market outcomes. We examine a bilateral lending relationship with opportunistic borrowers and imperfectly efficient courts, where lenders choose between costly legal proceedings and non-judicial foreclosure to enforce debt contracts. We endogenize legal enforcement by modeling litigation as a contest, and show that lenders’ enforcement choices depend on the relative effectiveness of courts and non-judicial recovery, with judicial efficiency determining the lender’s prospects in court. More efficient courts allow lower upfront interest rates while increasing expected recovery on each loan. Although improvements in judicial efficiency reduce credit rationing, they can adversely affect credit access, disproportionately benefiting wealth-constrained borrowers, and potentially tightening credit availability for asset-rich borrowers. We further show that informal enforcement mechanisms, such as reputational penalties, complement judicial enforcement in deterring defaults. However, greater reliance on penalties lowers both contractual interest rates and lenders’ expected returns. Collateral is most relevant for lower-quality investments and serves as an effective repayment incentive only when its value exceeds the outstanding debt. These predictions are supported by SBA 7(a) loan data. Taking Operation Lone Star as an exogenous shock to judicial capacity in Texas statutory county courts, we find that credit access worsened in affected counties, with lower loan volumes and smaller average loan sizes.Locked In: Housing Market Response to Interest Rate Increases (E5, R2)
Abstract
This paper develops a dynamic equilibrium directed search model with mortgage lock-in effects and price stickiness to explain the response of the U.S. housing markets to the 2022-23 interest rate increase, where prices fell only modestly, while transaction volumes collapsed. The model incorporates fully amortizing mortgages, explicit distributional heterogeneity in outstanding mortgage rates and endogenous refinancing decisions. It shows that mortgage lock-in creates heterogeneous sellers who face different incentives based on their outstanding mortgage terms, while price stickiness generates gradual adjustment in list prices. Along with an empirically estimated lock-in parameter, these features produce realistic housing market responses that imply a 14.5 percent reduction in selling propensity per percentage point of rate differential translating into aggregate housing supply reduction of approximately 27 percent, equivalent to roughly 1.35 million fewer annual home sales. The model shows that mortgage lock-in acts as a negative supply shock, providing 14-16 percentage points of price support at the peak while causing an additional 30 percent volume decline beyond demand effects alone. The result is a two-tier mortgage market where locked-in homeowners benefit from low rates while new buyers face high rates and reduced inventory.Market Dynamics in Cryptocurrency: The Role of Crypto Whales as Influencers in Decentralized Finance (G4)
Abstract
This study examines the impact of cryptocurrency whale transactions on investor composition and short term price volatility across the Bitcoin and Ethereum markets, introducing a novel empirical approach that links Whale Alert notification signals directly to on chain blockchain transaction data. By matching each whale alert to all on chain transfers occurring in the minutes that follow and classifying non whale, non exchange addresses into small, medium, and large investor categories within rolling six month windows, we construct a granular measure of how different investor classes respond to whale activity in real time. This linkage between public whale signals and address level blockchain data enables direct observation of leader follower dynamics that prior literature has only inferred from aggregate price or volume data. Our results show that Bitcoin whale alerts substantially reset investor participation, with pre alert composition explaining as little as 4 percent of post alert investor shares and small and medium investor participation rising by approximately 10 and 9 percentage points respectively. Ethereum exhibits near unity compositional persistence under Proof of Work, whereas the transition to Proof of Stake following the September 2022 Merge significantly amplifies the disruptive effect of whale alerts on investor composition, highlighting that whale influence is consensus mechanism dependent. We further document that whale transactions elevate Bitcoin volatility at the 15 minute horizon with significant decay at longer windows, while cross market spillovers are asymmetric. Notably, Wrapped Bitcoin whale alerts produce volatility effects on Bitcoin prices roughly double those of native Bitcoin alerts, underscoring that whale influence operates across network boundaries. These findings suggest that effective cryptocurrency regulation must account for heterogeneous investor responses to whale activity, network specific dynamics, and the structural consequences of consensus mechanism transitions.Market-Based Surplus Food Redistribution and Household Food Access: Evidence from Too Good To Go (Q1, I3)
Abstract
Nearly one-third of food produced in the United States goes unsold or uneaten, yet millions of households report not having enough to eat. While public programs address this gap through direct provision, an emerging alternative—market-based surplus food redistribution—remains unevaluated. This paper provides the first causal evidence on whether Too Good To Go (TGTG), a platform connecting consumers with restaurants and grocery stores selling surplus food at discounted prices, improves household food access.We exploit TGTG’s staggered rollout across 15 major metropolitan areas between 2020 and 2024, combining proprietary store-level participation data with over 1.4 million household observations from the Census Bureau’s Household Pulse Survey. Using the heterogeneity-robust difference-in-differences framework of de Chaisemartin and D’Haultfœuille (2024), we estimate dynamic treatment effects under staggered adoption with validated parallel pre-trends. We distinguish between preference-aligned sufficiency (having enough of desired foods) and preference-unaligned sufficiency (having enough food but not preferred types).
Following TGTG’s entry, preference-aligned sufficiency increases by 1.6 percentage points (2.3%), while preference-unaligned sufficiency and moderate insufficiency decline by 0.9 and 0.6 percentage points, respectively, with no change in severe insufficiency. Exploiting variation in store-level participation intensity, we find that food access improvements scale proportionally with the number of participating stores. The effects vary across the income distribution: low-income households see reductions in insufficiency, middle-income households gain preference alignment, and high-income households show null effects. We rule out contemporaneous confounders and find that TGTG reduces affordability-driven insufficiency without displacing public programs.
This paper makes three contributions: (1) first causal evidence that a market-based surplus food platform improves food access; (2) distinguishing between the quantity and quality dimensions of food sufficiency—not just whether households have enough, but whether they access foods they prefer; (3) novel use of proprietary store-level data to validate treatment timing and measure participation intensity.
Marketing, Demand Accumulation, and Exporter Dynamics (L1, F1)
Abstract
Firms face substantial demand barriers in foreign markets, as suggested by the slow export growth after first entry. This paper investigates the costs of demand barriers and how marketing mitigates them, thereby boosting export performance. We first document four stylized facts using data from Chinese manufacturing firms. First, marketing investment increases substantially following the firm's first export attempt. Second, greater marketing investment is associated with stronger export growth. Third, marketing payoff increases after the firm begins exporting. Fourth, post-export expansion is driven primarily by increases in quantity rather than by price changes. Using unique data on output prices and quantities from the Chinese textile industry, we separate firms' demand advantage from productivity. This allows us to directly evaluate the impact of marketing on demand. We find that demand plays a more important role than physical productivity in explaining sales variation in both domestic and export markets. Marketing boosts demand in both markets, but has a limited impact on markups. Counterfactual analysis shows that demand is a major barrier to firms' expansion in export markets. On average, marketing investment increases export sales by 8.4%. This study highlights the role of demand-side barriers and marketing in shaping post-export performance.Markups in Food Retail Within a City and Transportation Improvement (L1, R4)
Abstract
This paper studies the distribution of shop-level markups in restaurants and grocery stores and the adjustment of markups in response to transportation improvement. Using rich anonymous consumer–shop-level card transaction data, we find that (i) shops have a large market share in their customers' transactions and (ii) shops differ on this margin systematically across locations.Motivated by these findings, we develop a single-sector partial equilibrium quantitative model that delivers heterogeneous markups at the shop level through oligopolistic competition. In our model, the product of average transaction size and an HHI-style customer concentration index serves as a sufficient statistic for markup at the shop level. The sufficient statistic does not depend on functional form assumptions and is driven by two standard assumptions: additively separable indirect utility and Gumbel taste shocks. We estimate the model using variation in transportation costs from the opening of a subway line.
We find that median markups are lower downtown by 12\% (0.8 p.p.) for restaurants, but not for grocery stores.
We use the model to study the impact of the opening of the subway line on shops and consumers. When transportation improvement happens nearby, there are two counteracting effects on a shop's markup. Firstly, the shop becomes more attractive to some of its customers, and this pushes its markup up. Secondly, some of the shop's competitors become more attractive to some of its clients, which pushes the markup down. Quantitatively, we find that the first effect dominates. For the most affected consumers, markup adjustments reduce welfare by 0.5–2\% of the net welfare effect of subway expansion and markup adjustment.
Overall, this paper shows that even in a seemingly competitive environment, some shops have large market shares in their customers' transactions, and endogenous markup responses partially offset consumer gains from transportation infrastructure improvement.
Maturity Walls (G3, E4)
Abstract
Maturity walls occur when a majority of a firm's debt comes due within a short period (1-2 years), increasing rollover risk. Despite this, 47% of non-financial firms have them. This paper understands why firms adopt maturity walls and its implications for the aggregate economy. Using Mergent FISD data, I provide evidence that firms incur substantial fixed costs in bond issuance. I develop a dynamic model where firms decide each period the level and dispersion of their debt payments. The main trade-off is rollover risk from maturity walls in the presence of costly equity injections, versus the lower issuance costs incurred from infrequent rollovers. I estimate the model to match both aggregate and distributional moments of firms' debt payment schedules. Maturity walls increase credit spreads by 21% (36 bps) and default rates by 25% (30 bps). Lowering issuance costs reduces the adoption of maturity walls, but increases firms credit risk. Moreover, omitting maturity walls could underestimate the transmission of a credit market freeze up to 60%.Menstrual Stigma and Human Capital: Experimental Evidence from Madagascar (O1, I3)
Abstract
Menstrual stigma affects adolescent girls worldwide, yet its impact on human capital development remains largely unexamined. We use a field experiment in 140 schools in Madagascar to evaluate interventions designed to reduce menstrual stigma and promote hygiene behaviors (N=2,250). Teacher-led sensitization on stigma and hygiene, menstrual products, and sanitation infrastructure together substantially improve girls’ learning outcomes on standardized tests (+0.2 SD). These gains do not operate by improving school attendance or health, the channels typically invoked to justify menstrual hygiene programs. Instead, the improvements appear to arise from psychosocial mechanisms, including reduced menstrual stigma (measured using lab-in-the-field exercises, enumerator observations, and self-reports) and reduced stress (lower heart rate). We also test a novel approach for norm change by identifying “positive deviants” – girls within schools willing to openly challenge menstrual stigma. Selecting and training these positive deviants to serve as peer ambassadors for norm change produces significant additional improvements in self-reported stigma and hygiene behavior. The results demonstrate that addressing gender-specific psychosocial barriers can substantially improve girls’ education outcomes in highly deprived contexts, while highlighting both the promise and limitations of leveraging positive deviance for social norm change.Mid Life Crisis: HRT Safety Warning and Economics of Menopause (I1, J1)
Abstract
How do health information shocks affecting gender-specific biological transitions shape labor supply? This paper studies the impact of the 2002 Women’s Health Initiative (WHI) announcement—a major negative shock to beliefs about hormone replacement therapy (HRT)—on the labor-force participation of midlife women in the United States. Prior to WHI, HRT was widely used to manage menopausal symptoms and was believed to prevent chronic disease. The unexpected evidence linking HRT to increased risks of cardiovascular disease and breast cancer led to rapid and sustained decline in its use, plausibly increasing untreated symptom burden among affected women .I combine data from the Study of Women’s Health Across the Nation (SWAN) and the Current Population Survey (CPS) from 1996 to 2008 to estimate labor-supply responses to this shock. Because menopausal status is not observed in CPS, I construct a continuous measure of treatment intensity by predicting the probability of being peri- or post-menopausal using pre-WHI SWAN data and mapping these probabilities to CPS respondents based on demographic characteristics; results are robust to alternative machine learning-based prediction methods. I then estimate event-study models that exploit the timing of WHI announcement.
Women with higher predicted exposure to menopause experience large and statistically significant declines in labor-force participation following the WHI announcement. At mean exposure levels, participation falls by approximately 10–18 percentage points among peri-menopausal women and 10–12 percentage points among post-menopausal women, with more persistent effects among the latter group . These effects are sizable relative to baseline participation rates and robust across specifications.
The findings establish menopause as an economically meaningful determinant of labor supply and show that negative medical information shocks can reduce employment by constraining access to symptom-relieving treatments. More broadly, the results highlight how gender-specific health risks interact with information and treatment access to shape labor-market outcomes over the life cycle.
Mobile Money and Informal Risk Sharing After Flood Shocks (O1, G5)
Abstract
Rural households in developing countries use labor migration as a strategy to insure themselves against exogenous shocks. With the introduction of mobile money, remittances from migrants are now available to these households at a cheaper and faster rate. Previous literature finds that mobile money does not fully insure against shocks, but there is no quantitative evidence as to why this insurance is incomplete. In this paper, I quantify the causal effect of mobile money on smoothing consumption after large unexpected shocks, using a quasi-experimental design. I use a three period household panel representative of rural Bangladesh from 2011 to 2019, and collect satellite based inundation data, to identify mobile money use and exogenous flood shocks at the household level. I also construct a shift-share measure for network connectivity at the district level by tracking baseline bilateral migration from districts. The difference-in-difference estimations indicate that mobile money use alone cannot offer full insurance against shocks, and the distribution of networks who also use mobile money can explain the incompleteness. I also find evidence of negative spillover effects of mobile money use, as households who fail to adopt the technology are likely to be exempt from existing informal risk sharing contracts after shocks. I confirm that remittances are the main mechanism through which insurance against flood shocks transmit, with sizable differences in the amount of remittances coming from household and non-household members after shocks. I test for parallel pre-trends to ensure households who adopt mobile money in the future are not systematically different in smoothing consumption after shocks compared to non-adopters. The empirical evidence has important implications for how disruptive technology can alter shock coping mechanisms and informal risk sharing behavior among rural households.Monopsonistic Wage-setting and Inflation Dynamics (E3, J3)
Abstract
Motivated by evidence documented in labor economics, we introduce firms' monopsonistic wage-setting in an otherwise standard dynamic stochastic general equilibrium (DSGE) model. Our model identifies shocks to the wage markdown as labor demand shocks, a feature absent in standard models with households' monopolistic wage-setting. We estimate our model as well as the standard counterpart model using full-information Bayesian methods on US macroeconomic time series. With both labor demand and supply shocks, our model empirically outperforms its standard counterpart. Monopsonistic wage-setting leads real unit labor cost to be decomposed into not only real marginal cost but also the wage markdown. The refined measure of real marginal cost enhances the Phillips curve's ability to describe inflation dynamics. The model attributes fluctuations in the inflation rate to a broad mix of shocks that affect real marginal cost and transmit to the inflation rate through the price Phillips curve, while assigning only a marginal role to price markup shocks in the Phillips curve. This result stands in contrast to that obtained in the estimated counterpart model, which equates real marginal cost to real unit labor cost and attributes inflation fluctuations largely to price and wage markup shocks.Moral Hazard in Natural Disaster Insurance Markets: Evidence from the NFIP (Q5, R2)
Abstract
This paper evaluates the tradeoff between voluntary insurance coverage and incentives to reside in areas exposed to higher flood risk, which poses a central tension in policymaking around insurance rate-setting and mitigation of natural disaster hazards. I leverage spatial and temporal variation in flood insurance prices, due to discount changes and reverification cycles in the Community Rating System, to assess how flood insurance takeup and housing prices respond to changes in insurance prices. I estimate heterogeneous effects for properties that are subject to distinct risk levels and regulatory environments. Within high-risk areas, households residing in flood-adapted properties are found to be more price sensitive in insurance demand than those in non-adapted properties. Effects on housing prices are consistent with full or over-capitalization of insurance price changes and imply a high degree of demand substitutability between high-risk and lower-risk areas. These design-based estimates are used to estimate a model of flood insurance demand and residential location choice with respect to flood risk. A key mechanism that allows the model to match data patterns is sorting over unobserved heterogeneity in biased flood risk beliefs. The model can be used to quantify welfare under alternative approaches to pricing flood insurance and incentivizing insurance coverage.Multidimensional Missing Data in IV Models (C1, C2)
Abstract
Missing data is a pervasive challenge in empirical economics, particularly in survey and panel datasets where non-response often occurs sequentially across variables. Traditional approaches such as complete case analysis or simple imputation frequently yield inconsistent estimates when data are not Missing Completely at Random (MCAR), while standard Missing at Random (MAR) assumptions may be too restrictive for many empirical settings. This paper proposes a generalized Inverse Probability Weighting (IPW) GMM estimator to address multidimensional missingness—where instruments and endogenous variables are simultaneously missing—under a Sequential Missing at Random (SMAR) assumption. The SMAR framework relaxes the standard MAR condition by allowing the missingness of a variable to depend not only on always-observed covariates but also on the observed values of other sometimes-missing variables, provided these follow a sequential ordering consistent with the data collection process. I establish consistency and asymptotic normality of the proposed estimator under regularity conditions. Monte Carlo simulations demonstrate that the IPW-SMAR estimator converges to the true parameter value across varying sample sizes, whereas both the complete case estimator and the conventional MAR-based IPW-GMM estimator produce inconsistent estimates when the underlying missingness mechanism is genuinely sequential. Notably, the proposed estimator incurs no efficiency loss relative to the standard IPW-GMM estimator when MAR does hold. The method is applied to the 2020 Health and Retirement Study (HRS) to evaluate the causal impact of social isolation on cognitive decline among the elderly, exploiting COVID-19-related variables as instruments in a setting where sequential non-response across survey modules makes the SMAR assumption particularly well-suited.Multinational R&D Location and Knowledge Spillovers: Evidence from the Semiconductor Industry (F0, F0)
Abstract
Multinational firms conduct research and development (R&D) activities across mul-tiple countries, generating local knowledge spillovers that they do not fully internalize.
Empirically, we study the semiconductor memory (DRAM) industry from 1974–2004
using detailed firm-level production, capacity, plant, and patent data. We document
that innovation is concentrated among multinational firms operating R&D in multiple
countries, that these firms replicate their internal technology structures across locations,
and that R&D entry increases local knowledge spillovers. These facts raise a quantita-
tive question: Do multinational firms under-expand geographically in R&D relative to
an efficient industry allocation, and how do industrial policies such as the U.S. CHIPS
and Science Act affect the global distribution of innovation? To answer this question, we
develop a multi-country endogenous growth model in which heterogeneous firms choose
both the scale and geographic distribution of R&D. Innovation generates local spillovers
that depend on the presence and size of foreign R&D sites. The model delivers a sim-
ple characterization of equilibrium spatial expansion and shows that geographic R&D
allocation may be inefficient when spillovers are not fully internalized. In particular,
when small R&D sites generate substantial externalities, equilibrium under-expands
geographically relative to the planner’s allocation. Using the estimated parameters,
we evaluate how industrial policy reshapes the global allocation of R&D and affects
industry growth, aggregate industry productivity, and the global stock of knowledge.
Narratives Shape the Term Structure of Inflation Expectations (E5, G1)
Abstract
How does central bank communication shape the term structure of inflation expectations? We study how inflation-related language in Federal Reserve post-meeting statements and Chair press conferences reprices breakeven inflation (BEI) yields and forwards from two to ten years. Using a domain-adapted transformer to extract monetary policy stance and inflation-narrative indices, we embed these measures in an event-study design that conditions on target-rate surprises and complements daily estimates with intraday BEI changes in narrow announcement windows.Four findings emerge. First, inflation language in statements uniformly lowers BEI compensation across maturities, consistent with markets reading the committee-vetted text through the policy reaction function—a pattern we call narrative co-entailment. Second, the same language in press conferences raises long-horizon BEI forwards, consistent with the Chair conveying incremental information about medium-run inflation risks beyond the statement. Third, decomposing press-conference narratives into Delphic (outlook) and Odyssean (commitment) components reveals sharp maturity segmentation: Delphic language reprices long-horizon forwards upward, while Odyssean language compresses belly-of-curve forwards over the stabilization window—providing a direct term-structure test of the forward-guidance typology. Fourth, intraday identification shows a sign reversal: communication indices raise BEI during the press-conference window, whereas the statement window drives the negative daily effect, yielding the first within-day decomposition of how the two communication objects contribute to net announcement-day repricing.
These results demonstrate that inflation compensation repricing depends jointly on communication format, narrative content, intraday timing, and maturity. More broadly, institutional design—not just content—governs what markets learn from central bank language.
Network Effects, Market Power, and Fuel Prices: A Structural Analysis of Sustainable Aviation Fuel Mandates in Airline Industry (L9, Q4)
Abstract
This paper studies how marginal-cost shocks and climate policy reshape competition in a network industry. I focus on the airline industry, where jet fuel prices are volatile, and fuel is the largest operating expense, accounting for 20 to 30 percent of total operating costs. Yet little is known about how fuel prices affect pricing, flight frequencies, and route networks. I develop a structural model of the airline industry incorporating market power, network effects, and fuel prices. In the first stage, airlines choose route networks and flight frequencies. In the second stage, they compete in prices for nonstop and one-stop products. The model captures three sources of network interaction: demand-side benefits and cost-side interactions through network connectedness, and the endogenous creation of one-stop itineraries from linked nonstop flights. Jet fuel prices enter both the marginal cost of serving passengers and the fixed costs associated with flight frequencies. I apply the model to study climate policies that raise fuel costs, including sustainable aviation fuel mandates and carbon taxes, with a focus on SAF policies such as ReFuelEU in the EU. In an industry characterized by thin profit margins and high fuel-cost shares, such policies raise fuel costs and hence may reduce entry, lower service frequencies, and intensify market power, with these effects amplified by the network structure. The model quantifies the effects of such policies on fares, flight frequencies, market structure, emissions, and welfare. More broadly, the results can guide the design of SAF policies that reduce emissions while limiting distortions to competition and network connectivity. Preliminary results suggest that a one-dollar increase in fuel price per gallon reduces route-entry probabilities by 10 to 15 percent, with effects varying across carrier types and route centrality within the network. The estimates also indicate significant network effects on both the demand and marginal-cost sides.Nkie or Nike: A Counterfeit Trade Market Analysis (K4, F1)
Abstract
Counterfeit goods have become an enduring issue in the global trade landscape, continuing to have significant impacts. Counterfeit and pirated goods accounted for up to 2.5% of world trade in 2013, equating to USD 461 billion, and rose to 3.3% in 2016, or USD 509 billion.This study investigates the economic and institutional drivers of global counterfeit trade, with a focus on its implications for the US market. The research examines how governance quality, trade, and economic development influence counterfeit activity. By analyzing these dynamics, the study aims to provide insights that enhance consumer safety by mitigating risks posed by counterfeit goods while also strengthening protections for legitimate producers whose brands, revenues, and innovations are under threat. Ultimately, this work seeks to inform policy solutions that safeguard both consumers and businesses, promoting secure and transparent trade networks.
Our paper uses bilateral data which extends the focus of the analysis to a multitude of origin and destination countries. Applying a two-way fixed effects regression model, our results reveal that economies with mid-range GDP per capita are more likely to engage in counterfeit trade. This finding aligns with prior research showing that as countries develop economically, demand for high-value counterfeit goods rises alongside the sophistication of counterfeit production and distribution. However, as economies move into higher-income brackets and strengthen their institutional frameworks, counterfeit activity diminishes. The regression results also highlighted the significance of trade intensity, with economies more integrated into global trade networks showing greater involvement in counterfeit activity. Surprisingly, corruption showed limited statistical significance, suggesting that while governance weaknesses contribute to counterfeiting, other factors—such as regulatory capacity and consumer demand—may play a larger role.
Nonemployer Entrepreneurs: Their Dynamics and Evolution (L2, D2)
Abstract
Nonemployers (i.e., businesses with no paid employees) account for almost 80 percent of all businesses in the United States and in the last decade, their growth has greatly outpaced that of employer firms. However, their contribution to the economy and their dynamics remain less well understood than their employer counterparts. Using a novel linkage methodology between the nonemployer and employer business universes, we study the evolution of nonemployer business dynamics from 1995–2023, following startup cohorts 2003-2016 over a 7-year period . We find that the share of nonemployer startups migrating to employer status within seven years fell from 3.2% to 2.4%, driven by a surge in low-revenue startups - consistent with the emergence of the gig economy. Crucially, we find that the employment migrants generate (as measured at their seventh-year after start-up) has increased over time. Migrants’ share of seventh-year cohort employment rose from 9.6% to 11.1%, driven by a nearly 40% increase in employees per migrant. These results show that nonemployer migrants have gradually grown in importance as contributors to employment – in contrast with the declining dynamism among employer startups. Understanding the reasons for the rising employment rate is crucial to inform and tailor policies, but currently those reasons remain to be understood.Not So Fast: Disaster Relief Timing and Political Incentives (H1, Q5)
Abstract
Prompt relief is a central determinant of how quickly economies recover after natural disasters strike, yet the speed of government response may be shaped by political incentives. This paper studies variations in approval timing of Federal Emergency Management Agency (FEMA) disaster declarations in the United States. Using the universe of FEMA disaster declaration requests and approvals from 1980 to 2024, we construct a long-run dataset of approval times and analyze how these vary with the electoral calendar and partisan alignment between state and federal executives. We find that elections and party matching significantly affect the speed of disaster declaration approvals. In presidential election years, approvals are accelerated by more than two months relative to baseline timing, whereas in midterm years they are delayed by a similar amount. We also find small but significant evidence of partisan bias, with approval delayed by about two days when a state’s governor is from a different party than the President. Across individual Presidents, estimated effects are significant but vary substantially, sometimes in opposing directions. This suggests that approval timing operates as a discretionary political instrument, with Presidents using it heterogenously according to personal incentives or preferences. At the same time, there is a long-run trend toward faster approvals overall in coexistence with persistent and increasing political effects. The findings show that even relatively early stages of disaster response are politically mediated, with likely important consequences for recovery, public trust, and the distribution of federal assistance.Nothing Stops a Bullet like a Job? The Long-Term Impact of Summer Youth Employment Programs on Criminal Justice Involvement (K4, J4)
Abstract
Summer Youth Employment Programs (SYEPs) have been shown to reduce youth violence in the near-term, yet little is known about how to sustain these impacts. To carry out this analysis, we make use of an embedded lottery design to study a cohort of 4,219 youth who applied to the Boston SYEP during the summer of 2015. We link these data to information provided by the Boston Police Department (BPD) from 2014 to 2019 on both juvenile and adults arrests as well as noncustodial police contacts, or “field interrogation and observation” reports (FIOs). This allows us to observe involvement with the criminal justice system during the 17 months prior, and up to 4.5 years after, SYEP participation for both juvenile and adult participants and at a much more granular level than prior studies. We find that being offered a job through the Boston SYEP leads to lower criminal justice involvement for up to 4.5 years, but only among youth who are more likely to come into contact with police. Among males there is a 28 percent decrease in violent crime arrests over the entire 4.5 years, with 18-24 year olds and high-risk youth experiencing even larger reductions in violent crime arrests of 53 percent and 79 percent respectively during the full follow-up period. Delving further reveals that the program reduces co-offending by 30 percent and youth who were randomly selected to participate the following summer experienced even larger drops in criminal justice involvement that also included a reduction in felonies. These findings run counter to the narrative that SYEPs primarily reduce violent crime among the general population of youth. Instead, we find that SYEPs are successful in engaging youth that are at high-risk of repeat violence, suggesting that program impacts might be sustained through targeting and/or repeat participation.Optimal Carbon Taxation and Solar Geoengineering for Managing the Tipping Point Risk of Coral Reef Ecosystems (Q5, C6)
Abstract
Solar geoengineering, also known as Solar Radiation Modification (SRM), provides rapid cooling that can reduce temperature-driven climate tipping point (CTP) risks, but it may simultaneously trigger carbon-driven CTPs by weakening incentives for carbon emissions abatement. We endogenize the risk of coral reef ecosystem collapse, an urgent climate tipping point jointly driven by warming and acidification, in a stochastic version of the Dynamic Integrated Climate-Economy Model (DICE-2023), and examine the optimal mix of carbon taxation and SRM. We solve the model using the simulated certainty equivalent approximation (SCEQ) method, a novel algorithm that enables fast and stable computation of numerical solutions. The results show that: (1) under the "Abatement-Only" strategy, ignoring collapse risk ultimately leads to ecosystem collapse. The optimal carbon tax rises from about $50/tCO2 initially to roughly $500/tCO2 by 2100 and the probability of collapse exceeds 50% before 2030. Once collapse risk is internalized, the optimal carbon tax becomes more than four times higher and net-zero emissions are reached before 2060. Even so, collapse may still not be fully avoided, depending on coral reef sensitivity to ocean acidification. (2) Under the "Abatement + SRM" strategy, ignoring collapse risk only delays collapse until after 2045, rather than preventing it. When collapse risk is accounted for, however, "Abatement + SRM" can avoid collapse and improve economic output, with 2040–2100 serving as the key window during which additional SRM deployment is required to prevent collapse. Overall, although SRM is a double-edged sword for coral reef ecosystems, it appears to be both effective and necessary for preventing collapse.Optimal Climate Policy under Distortionary Fiscal Constraints: The Innovation-Learning Dilemma (Q5, O3)
Abstract
I study optimal climate policy in a model with distortionary fiscal constraints, learning-bydoing, and directed technical change. The framework gives rise to an innovation–learning
dilemma: because research subsidies must be financed through distortionary taxation,
the planner cannot simultaneously support the optimal trajectory of research and the
accumulation of expertise. I show analytically that learning-by-doing amplifies the effect
of carbon taxation on research incentives through an expertise channel—sector-specific
experience raises the relative returns to innovation—which may lower the need for research
subsidies. In a calibration to the United States, I assess how these forces jointly shape the
optimal policy mix to implement an emission target. I find that carbon should be taxed
heavily, at a rate exceeding the social cost of carbon by about 3 percent throughout the
transition. While this policy promotes green learning, it implies a greater misallocation
of research effort: higher carbon taxes induce too rapid a shift of researchers from fossil
to renewable energy technologies. A welfare decomposition shows that learning-by-doing
increases the welfare losses associated with this misallocation, corroborating the presence
of an innovation–learning dilemma under distortionary fiscal constraints.
Overlapping Calendars: Colonial Legacies and School-Agriculture Misalignment (I2, O4)
Abstract
Do colonial legacies shape the timing of schooling in the Global South—and at what cost? Using cross-country data, we document systematic overlap between academic calendars and local harvest seasons in countries in southern latitudes and show that these patterns are strongly associated with colonial institutional legacies rather than local climatic and agricultural conditions. Exploiting this variation, we estimate the causal effect of school–agriculture calendar overlap and find that greater overlap significantly increases school dropout in former colonies. We argue that this relationship operates through child labor: in agrarian economies, seasonally concentrated agricultural work allows households to reallocate children’s time away from schooling when labor demand peaks. To test this mechanism, we exploit a natural experiment in Colombia in which a reform exogenously shifted school calendars, increasing overlap with local agricultural cycles. Using administrative school census data and a synthetic difference-in-differences design, we show that increased overlap reduces enrollment through higher dropout and lower grade progression, with effects concentrated in rural areas and accompanied by increases in child labor, as documented using household survey data. Our findings highlight school calendar timing as an important and underexplored institutional determinant of human capital accumulation, suggesting that better alignment with local economic conditions may offer a low-cost policy lever to improve educational outcomes.Overreacting to Information Events: Composite Beliefs and Two-Step Updating (G4, D8)
Abstract
Beliefs about composite events are central to economic decisions, yet most evidence on Bayesian updating concerns simple, single-state probabilities. This study examines whether departures from Bayesian posteriors are amplified when the object of belief is a composite event and whether the measured deviations depend on the elicitation procedure. We run an incentivized urn-learning experiment in which participants face three possible data-generating states: one produces uninformative signals (an even mix of two colors), while two produce informative signals with opposite skews. Within each trial, participants observe repeated draws with replacement and report, after each draw, the probability that the underlying state is informative (a composite event). Deviations are defined as the difference between reported beliefs and the Bayesian posterior implied by the objective prior and the observed signal history. To identify the role of intermediate conditional reasoning, we compare two elicitation treatments that hold the information structure and incentives fixed. In a one-step treatment, participants directly report the composite probability. In a two-step treatment, participants first report the conditional probability of one informative state versus the other (conditioning on being in the informative set) and then report the composite probability. We also vary the prior environment (baseline versus equal priors) and include a benchmark treatment that elicits beliefs about a single informative state rather than a composite event. Three results emerge. First, composite-event beliefs drift upward relative to the Bayesian posterior over the course of a trial. Second, making the intermediate conditional judgment explicit changes the time path of deviations, with effects that depend on the prior environment. Third, the single-state benchmark exhibits distinct dynamics, indicating that aggregation across states is an important source of systematic error. The findings provide causal evidence that composite-event beliefs are not invariant to elicitation design and are consistent with models of sequential inference.Ownership and Electricity Prices under Regulation: Evidence from Hawai'i (L9, Q4)
Abstract
How do governance structures shape adaptation in regulated monopoly electricity systems? This paper studies a rare institutional transition: Kauaʻi’s 2002 electric utility conversion from investor-owned to member-owned cooperative governance. Hawaiʻi provides a useful setting because its major islands operate as isolated, vertically integrated monopoly electricity systems under a common regulatory framework, though some rules differ by ownership form. Using monthly sectoral retail electricity prices from 1990–2025 and an augmented synthetic control approach, the analysis constructs a counterfactual for Kauaʻi from Oʻahu, Maui County, and Hawaiʻi Island. The results show no immediate price break at conversion. Instead, Kauaʻi’s retail prices gradually diverge downward from the synthetic control over the post-conversion period, with the gap emerging only over time and widening during periods of elevated fuel costs. The pattern appears across residential, commercial, and industrial sectors and is more consistent with dynamic adjustment to cost shocks and long-lived investment decisions than with a one-time level effect at conversion. Descriptive evidence suggests that the cooperative period was associated with a different generation and procurement path, lower petroleum exposure, and weaker pass-through from global fuel-cost shocks into retail prices. The findings suggest that the shift from investor-owned to cooperative governance, together with ownership-form-specific regulatory differences, altered long-lived procurement and investment decisions in ways that changed exposure to fuel-price risk and response to falling renewable costs, shaping the evolution of retail rates over time.Paying You to Spend: The Effects of Credit Card Cashback on Household Food Demand (D1, H3)
Abstract
This paper examines the effects of credit card cashback on household food demand using Discover’s rotating 5% cashback program as a natural experiment. Using NielsenIQ scanner data and a difference-in-differences design, we exploit exogenous variation in grocery cashback eligibility across quarters to estimate how cashback incentives influence food expenditures and price elasticities. Our findings show that cashback significantly increases household grocery spending, with heterogeneous effects across food categories: non-essential and less healthy foods exhibit larger demand responses than staples and healthier alternatives. These results highlight the potential of market-driven price incentives to shape consumer behavior and offer insights for designing targeted food assistance programs that leverage price-based interventions to promote healthier food choices.Political Impact of Curtailing Migrant Workers: Evidence From The Bracero Program (J6, N4)
Abstract
A central question in U.S. immigration policy is how a reduction in low-skilled immigrant labor shapes the economy and politics. This paper examines the political impact of restricting the supply of immigrant workers under the Bracero Program, a U.S.–Mexico agreement that provided Mexican farm labor to address postwar shortages. This restriction occurred in two phases: first, a 1962 minimum wage increase for Mexican workers and second, the program’s official termination. Using cross-county variation in exposure to Mexican farm workers, I employ a difference-in-differences model to compare electoral outcomes in high- and low-exposure counties after 1962. In the short run, these policies led to a 2.8 percentage point increase in the vote share for the Republican party, which had opposed the program’s termination. This effect is likely driven by voter reaction to higher agricultural prices and by the mobilization of voters through Republican-affiliated media.Powering the Workforce: The Role of Community Colleges in Energy Transformations (I2, J2)
Abstract
How do workforce training institutions respond to structural shifts in labor demand? As rapidly growing electricity consumption and the clean energy transition require a massive expansion of the skilled trades workforce, understanding the supply side of human capital is critical. Public community colleges, enrolling nearly 30% of U.S. postsecondary students, have vocational mandates intended to align with local labor markets. Yet, empirical evidence on whether they can adapt curricula dynamically remains scarce. I examine how community colleges respond to energy-driven labor demand shocks using the U.S. shale oil and gas boom as a natural experiment. Shale’s fixed geography and the technological shocks that enabled extraction make the boom exogenous to college location and program choice. Exploiting the staggered timing of these localized shocks, I employ stacked difference-in-differences and event study designs using IPEDS annual institution-level data covering the universe of public community colleges. Treated colleges expand energy-related programs by 26% relative to their pre-period means, generating a sustained 100% increase in energy completions driven entirely by new program creation. However, institutional adaptation involves significant tradeoffs. Non-energy fields stagnate, yielding much larger absolute effects. Treated colleges see 20% fewer programs and 44% fewer completions than expected absent the boom, consistent with increasing opportunity costs of enrollment and reallocation of institutional resources. Responsiveness varies substantially across colleges. Baseline reliance on state and local funding and administrative pay are positively associated with adaptation, while higher faculty salaries are associated with slower responses. My results demonstrate that community colleges can serve as intermediaries between labor demand and supply when shocks are large and sustained, but stagnation in non-energy fields raises concerns about overspecialization. Ongoing work extends the empirical framework to solar and wind shocks to assess whether these findings generalize to the clean energy transition and how community colleges can help ease labor bottlenecks.Predicting Household Purchases Using GPTs (C4, D5)
Abstract
Can large language models predict household behavior based on information available at the time of decision? Using the New York Fed Survey of Consumer Expectations, we ask whether ChronoGPT, a chronologically disciplined large language model, can predict realized household purchase decisions from lagged one-year-ahead inflation expectations, household characteristics, and macroeconomic information. We compare a penalized logistic regression on tabular covariates with a penalized logistic regression on prompt-conditioned ChronoGPT embeddings. The logit benchmark outperforms the embedding-based specification out of sample, suggesting that textual macroeconomic information does not improve prediction of household purchases. Placebo exercises strengthen this result: embeddings built from the macroeconomic information set do not outperform even time-only prompts and are comparable to embeddings of shuffled macroeconomic text. These findings suggest that predicting revealed behavior remains difficult for current prompt-conditioned GPTs.Pricing Wildfire Risk in Residential Real Estate: Invasive Grass-Fire Cycle (R3, Q5)
Abstract
We develop an ex-ante, ecology-based measure of wildfire risk and examine whether this forward-looking risk is capitalized into U.S. residential real estate prices. Using nationwide pixel-level data on invasive vegetation and wildfire occurrence together with roughly 40 million residential property transactions from 1991 to 2021, we first estimate wildfire probability as a function of lagged ecological conditions. We show that invasive grasses significantly increase the likelihood of subsequent wildfires, consistent with an invasive grass-fire feedback in which ecological degradation raises expected hazard over time. We then aggregate these predicted wildfire probabilities to the county-year level and link them to housing transactions in a hedonic framework with rich property controls and fixed effects. A 10-percentage-point increase in predicted wildfire risk is associated with a 2.6 percent decline in transaction prices, implying an average loss of about $9,000 for a median-priced home. The results indicate that housing markets price expected ecological wildfire risk, not only realized wildfire losses. More broadly, the findings show how ecological change can affect long-run asset values by altering the probability of future disasters before those disasters occur. By introducing a transparent and policy-relevant measure of wildfire exposure grounded in ecological conditions, the paper contributes to the literatures on climate risk, housing markets, and environmental finance. The results also suggest that interventions targeting invasive vegetation may generate economic benefits by reducing underlying wildfire risk, with implications for housing valuation, insurance design, and public investment in climate adaptation, vegetation management, and land-use policy. More generally, our approach provides a market-based framework for quantifying how slow-moving ecological deterioration can be incorporated into asset prices before extreme events are realized, thereby linking environmental change, expected disaster risk, and the valuation of long-lived household assets in a unified empirical setting.Printing Morality: Uncle Tom’s Cabin and the Antebellum Political Realignment (N0, P0)
Abstract
Can a literary work shape mass politics? In the 1850s, the United States experienced one of the most consequential political realignments in its history: the collapse of the Whig Party, the fragmentation of the Democratic coalition, and the rise of a Republican North. This paper argues that Uncle Tom’s Cabin played a pivotal role in this transformation by diffusing antislavery ideas across the northern states. Using advertisement records extracted from digitized historical newspapers, I construct a county-level exposure to the novel based on travel times along the 1850 transportation network. Difference-in-differences estimates show that counties with greater exposure experienced larger declines in Democratic vote shares and a stronger shift toward the emerging Republican Party during the 1850s. Mechanism evidence suggests that the novel reshaped moral sentiment toward slavery, amplified antislavery discourse, and helped translate that response into partisan mobilization.Private Financing in U.S. Public-Private Partnerships: Evidence from a New Dataset (H5, L3)
Abstract
Despite growing policy interest in public-private partnerships (P3s) and billions in federal subsidies, there is virtually no systematic evidence on how U.S. P3s are actually financed. I bridge this gap by constructing the first comprehensive, project-level dataset tracking private capital in U.S. infrastructure, covering nearly 1,400 P3 and traditional procurement projects across all major sectors from 1990 to 2022. The dataset disaggregates each project's capital stack by funding source while simultaneously recording concession terms, delivery modes, and risk allocation. Using this data, I empirically characterize the economic determinants of private investment. First, federal credit instruments---Private Activity Bonds and TIFIA loans---are critical to securing financial close and are associated with higher private leverage, whereas direct federal grants exhibit near-complete crowd-out of private capital. Second, consistent with incomplete contracts theory, private financing intensity strongly predicts greater task bundling, confirming that financial commitment and operational control are complements. Furthermore, longer concession horizons enable significantly greater private capital shares. Finally, project risk shapes the debt-equity mix: construction risk reduces leverage while government-backed availability payments increase it, reflecting the core principles of project finance.Productivity Externalities of Working from Home: Welfare and Policy Implications (R0, D0)
Abstract
I study how the socially optimal level of onsite work differs from the market equilibrium. I develop a general equilibrium model in which workers decide how much to work onsite and work from home. Productivity spillovers can occur within and between onsite and remote workers. The model predicts that the balance between onsite and remote productivity spillover effects affects the gap between the socially optimal and the market equilibrium level of onsite work. I measure these spillovers by matching the model to U.S. survey data from 2022 to 2024 at the city-sector-work mode level. I find that, on average, a social planner could improve welfare by 2% by increasing hybrid workers' share of onsite time by 3% and increasing the number of fully onsite workers by 2%. This could be accomplished by offering a subsidy for onsite work equal to 11% of hybrid workers’ gross income. Without the remote productivity spillovers, a similar level of welfare improvement would require larger changes: hybrid workers’ share of onsite time would need to increase by 5%, and the number of fully onsite workers would need to increase by 3%. The subsidy would cost 15% of hybrid workers' gross income.Prosumers (D5, D1)
Abstract
Motivation and Research QuestionThe rise of "prosumers"—agents who both produce and consume the same good—is transforming sectors like renewable energy, agriculture, and digital content. Despite this shift, economic theory has largely relied on traditional models that separate production and consumption. This paper develops a unified price-theoretic framework to analyze how prosumption opportunities alter market equilibria, efficiency, and the effectiveness of policy interventions.
Methodology
We extend the standard supply-demand framework to incorporate prosumption costs and benefits. The model evaluates equilibrium outcomes under two critical informational regimes: symmetric information and asymmetric information regarding product quality. By endogenizing the decision to prosume, we identify the market conditions—such as transaction costs and technological accessibility—under which a prosumer class emerges and remains sustainable.
Key Findings
Our analysis reveals a striking departure from classical welfare economics. We find that prosumption activities can eliminate the traditional Harberger deadweight loss triangle. Because prosumers can bypass market frictions and transaction costs through self-provision, they recoup value typically lost in standard trade environments. However, under asymmetric information, prosumers may face unique "lemons" problems that require specific regulatory designs to maintain market efficiency.
Policy Implications
We assess the impact of taxes, subsidies, and regulatory controls on prosumption-heavy markets. Our results suggest that traditional "one-size-fits-all" subsidies for green energy may be suboptimal if they do not account for the prosumer’s dual role. Instead, policy should target the marginal cost of self-production versus market participation. This study provides a foundational framework for economists and policymakers navigating the transition toward decentralized, prosumer-driven economies.
Quantifying Macroprudential Policy in a Small Open Economy with Safe and Risky Debt (E4, F3)
Abstract
This paper studies macroprudential policy in a small open economy with safe and risky debt under collateral constraints. I develop a quantitative DSGE model in which households borrow from international lenders using two instruments: safe debt denominated in domestic currency and risky debt denominated in foreign currency. Risky debt carries an exogenous interest rate spread that depends on its level but provides borrowing capacity advantages through a collateral constraint that assigns a weight theta to risky debt. In the decentralized equilibrium, agents fail to internalize how borrowing affects collateral values and tightens future borrowing limits, generating both excessive borrowing and inefficient allocation between safe and risky debt.The model is calibrated using Mexican TFP data from FRED over 1980-2019, with aggregate income shocks following an estimated AR(1) process. Quantitative results indicate that, relative to the constrained social planner, the decentralized equilibrium generates both excessive borrowing and an inefficient allocation between safe and risky debt. Households rely excessively on safe debt to avoid the interest rate spread on risky debt, while the planner allocates more borrowing to risky debt when it provides borrowing capacity advantages through the collateral constraint. As the collateral weight theta increases, both total borrowing and debt composition change systematically. As the borrowing capacity advantage of risky debt declines with higher theta, the social planner allocates progressively less risky debt and eliminates risky borrowing when theta = 1.
Comparing macroprudential policies, a uniform tax on both safe and risky debt delivers the largest welfare gains and the lowest crisis probability, closely approximating the social planner allocation. A tax on risky debt performs moderately well, while a tax on safe debt is the least effective policy, with the smallest welfare gains and the highest crisis probability, because risky borrowing generates larger externalities due to the spread it carries.
Reducing the Supply of Unused Opioids: An Evaluation of a Community Health Center Opioid Return Program (I1, C9)
Abstract
More than one-third of opioid-related deaths are attributed to secondary users of prescription opioids with two-thirds obtaining painkillers from family and friends. Using a randomized control trial, we test the effectiveness of a community-based medication return program. We focused on an economically disadvantaged, largely Hispanic population receiving medical care at one of five community health centers located in a mid-sized city in Massachusetts1 with one of the highest opioid overdose rates in the state as well as the nation. Patients in the treatment group were counseled by pharmacy staff according to a pre-specified protocol that provided information about the addictive nature of opioid medication, the potential for misuse, and how to safely return any unused medication. We further varied the intervention across locations using a text reminder, gift card, mail-back envelope, or provider consultation and examined whether these features increased the opioid return rate. Our primary outcome measure was the rate at which patients in the treatment group returned their unused medication relative to that of the control group. We also explored heterogeneous treatment effects across demographic groups and prescription characteristics. Treated patients were 3.0 percentage points more likely to return medication relative to the control group, yet these effects remained small even when coupled with a text reminder, financial incentive, mail back envelope, or clinician involvement. Moreover, the amount of medication prescribed for a given condition varied widely and surprisingly, patients who were prescribed larger amounts were less likely to return their medication, perhaps due to greater severity of the medical condition for which it was prescribed. our results suggest that while medication return programs are useful in removing unused opioids from a small number of households, they are unlikely to yield sizeable reductions in the supply of opioids for secondary use at the community level.Regulatory Spillovers and the Interstate Flow of Crime Guns (K4)
Abstract
This paper develops a unified theoretical and empirical framework to study the interstate flow of crime guns in the United States. We propose a model in which individuals and networks choose source states based on acquisition costs shaped by gun regulations and geographic distance. The model predicts that firearm flows decline with distance, decrease when source states tighten regulations, and increase when destination states impose stricter laws. Restrictions on legal access in a destination state do not eliminate demand but instead shift sourcing toward external markets. Using bilateral firearm tracing data from 2009--2023, we estimate a gravity equation exploiting variation in state gun laws. The empirical results find that distance and regulatory differences systematically shape interstate firearm flows. We then structurally estimate the model and conduct counterfactual exercises to quantify how changes in gun regulations alter the national pattern of firearm flows. The results highlight the central role of geographic frictions and regulatory differences in shaping the interstate network of crime guns.Return To Party (R0, D1)
Abstract
We study the implications of post-pandemic nightlife for cities. We document a sharp decline in evening socialization: relative to the aughts, staying at home all evening has increased by 19%, a phenomenon that is present across all ages, genders, and many other demographics. Present day 18-24 year olds stay at home as much as 35-49 year olds in the aughts, and present-day 25-34 year olds stay at home as much as 50-64 year olds. Increased isolation at home has occurred simultaneously as a decline in restaurant visits and visits to friend's houses, declines which vastly outpace both the pandemic work-from-home boom and the recession-era increase in social isolation. Leveraging cell phone tracking data, we study the implication for cities. We rank cities by their post-pandemic nightlife and study the correlates that explain nightlife recovery.Rise of Platforms and Aggregate Market Power (L1, O3)
Abstract
This paper examines the rise of platform firms and its implications for aggregate market power in the U.S. economy. We develop a classification framework leveraging large language models to construct a continuous, firm-year level platform score from 10-K filings for the universe of publicly listed firms from 2000 to 2020. We document three facts. First, the sales-weighted platform share rises from 6.3% to 17.6%, driven primarily by the reallocation of market share toward existing platforms. Second, platform firms persistently exhibit markups 1.36 times those of non-platform firms and profit rates 6.4 percentage points higher. Third, the expansion of platforms accounts for 3.3 times the total increase in aggregate markup, as the non-platform contribution declines.To rationalize these patterns, we develop a model of oligopolistic competition with heterogeneous network externalities. Platform firms enjoy demand-side economies of scale, which incentivize price cuts to attract users, but the resulting larger market shares drive up markups through oligopoly power. The model generates a self-reinforcing loop: network effects increase market share, and for sufficiently large platforms, the oligopoly size effect dominates, producing high markups consistent with the data. We calibrate the model to match the joint distribution of platform scores, markups, and sales shares, and conduct counterfactual exercises to quantify the contribution of platformization to the rise in aggregate market power and to evaluate the welfare trade-off between higher productivity from network effects and greater deadweight loss from market power.
Rising Waters, Falling Spreads: Climate Adaptation and Banks’ Repricing of Disaster Risk (G2, D0)
Abstract
How does risk pricing change in credit markets after observing a nearby catastrophicevent? We combine detailed geospatial data on ex-ante flood risk of German firms with
credit register data and show that after a major flood in 2021, loan rates decrease for
high-flood risk firms that were not directly affected. We provide evidence that this
decrease is an endogenous reaction to post-flood climate adaptation measures. First,
we use novel insurance data to show that interest rate decreases are strongest, when
firms purchase additional insurance after the flood. Further, in regions where local
governments invest heavily in climate protection measures after the flood, we observe
the largest decreases in interest rates. Banks also decrease their disaster risk exposure
by increasing securitization. We thus highlight a novel tradeoff between credit risk and
other measures of mitigating climate risk.
Risk Exposure in Convenient Assets (G0, E0)
Abstract
We study the term structure of convenience yields across asset classes that provide nonpecuniary services to their holders. Using Chilean sovereign green bonds and U.S. Treasuries as leading examples, we document a common empirical regularity: convenience yield term structures are downward-sloping, implying that convenient assets carry greater interest-rate risk exposure than otherwise identical securities. We rationalize these facts through a preferred-habitat term structure model in which convenient assets arise to satisfy arbitrageurs balance sheet ratio constraints. This assumption captures a broad class of institutional frictions, regulatory constraints, collateral requirements, and investment mandates, that are denominated in market-value terms. Because long-duration bonds have greater price sensitivity to discount-rate shocks, convenience services on these bonds are more volatile, generating a risk premium that compresses convenience yields at longer maturities. Our results suggest that the declining term structure of convenience yields is a general feature of convenient assets, with implications for optimal issuance strategy and portfolio risk management.Risk-Return Trade-off Analysis of Novel Biocontrol Use for Bacterial Spots on Bell Peppers in Florida – Policy Implications (Q1, M2)
Abstract
Experts conduct experimental research using biocontrol to address concerns in agriculture, as an alternative to chemicals that are supposedly harmful to the environment. Since farmers operate under unpredictable climate and market conditions that often strongly affect yields, this volatility leads to a random pattern of agricultural incomes over time. To adopt novel products, profitability must be proven. Since the expected agricultural outcome drives substantial risks for decision-makers who must consider the risk-return tradeoff, rigorous analysis is required. Monte Carlo simulations rely on classical statistical distributions that do not align with the small sample sizes from experimental research. We use a triangular random distribution to conduct Monte Carlo simulations in drawing useful conclusions from novel biocontrol experiments aimed at addressing diseases on bell peppers at the farm level in Southwest Florida. In addition to the triangular random distribution, we used partial budgeting to compare risks across 11 spray programs designed to reduce the severity of bacterial spots in pepper production. We identified profit variability within and across programs, as well as specific factors affecting yield variations. Three biocontrol programs generate profits exceeding $2,000; however, only one offers an efficient risk-return trade-off. There are avenues to design agricultural and environmental policies to promote biocontrol programs in farm pest management.Secrecy, Spectrum, and Certification: Evidence from Wireless Electronic Devices (O3, L5)
Abstract
This paper examines the strategic decisions of firms to withhold information in innovation-intensive markets, highlighting that secrecy contains distinct temporal dimensions. The study illustrates the thesis using wireless products, the majority of which utilize unlicensed spectrum. Using all FCC equipment filings from 2001 to 2021, we analyze firms' choices between long-term confidentiality (LTC), which permanently conceals technical information such as block diagrams, and short-term confidentiality (STC), which temporarily delays the public release of marketing information, including user manuals. We argue that the former acts to preserve ``trade secrets,'' the traditional role of IPR. In contrast, the latter acts as ``commercialization insurance,'' a previously unrecognized role for secrecy to ensure against launch and coordination risks in the commercialization process. We examine evidence consistent with this view. We find that the determinants for LTC and STC differ sharply. LTC use reflects appropriability concerns and institutional complexities—technically complex products and competitive markets are 12 to 35% more likely to adopt it. Traditional forms of IPR, such as patenting, are a substitute for LTC, especially in the case of WiFi. By contrast, STC is driven by the commercialization timing of branded products by large firms. Large, domestic, patent-owning firms with regulatorily complex products are more likely to use STC. Technical aspects of the product do not predict STC usage. Usage is consistently higher among incumbents for STC, while it quickly converges for LTC, even after controlling for all other variables. An exposure-based event study further shows that firms with a high ex-ante propensity to use STC bring products to market faster after the policy becomes available, with no corresponding changes in product complexity or volume. Together, these results show that permanent and temporary confidentiality serve distinct strategic purposes.Self-Detrimental Avoidance of Rest (D9, I3)
Abstract
Across many cultures, resting instead of working is viewed as a barrier to higher earnings. This belief is also reflected in many canonical economic models. Recent empirical evidence highlighting the productivity benefits of rest challenges this belief. Yet, existing work tends to ignore individuals’ demand for restful activities and whether it aligns with their returns. In the context of an online labor market experiment in South Africa, we explore whether workers capitalize on the returns to short rest periods. After eliciting demand for rest, we estimate returns to rest for the same individuals and find that mandated rest boosts productivity by 0.3 standard deviations, thus making up for forgone earnings from resting. At the same time, only 19% of workers voluntarily choose to rest. Contrary to the notion of selection on returns, workers with high financial returns to rest do not select into rest. We provide suggestive evidence that misperceived financial returns are driving the disconnect between demand for and returns to rest. Our results provide proof-of-concept evidence that individuals may be misallocating effort between resting and working and could reach higher overall utility by working less. This highlights the importance of understanding misperceptions around rest, especially in light of the economic burden of long-term costs of overworking such as burnout.Smarter Money, Slower Growth (E2, G2)
Abstract
This paper studies how the composition of investors shapes capital allocation, firm dynamics, and aggregate productivity. We document two empirical facts using mutual fund holdings and firm-level data. First, institutional investors disproportionately allocate capital to larger firms along both intensive and extensive margins. Second, conditional on size, they tilt toward firms with higher marginal product of capital (MRPK), suggesting more efficient capital allocation relative to retail investors. Motivated by these patterns, we develop a tractable model of heterogeneous firms with borrowing constraints and multiple sources of external finance. Institutional investors differ from retail investors in two key dimensions: they require a lower risk premium due to superior information and scale, but face liquidity constraints that induce a preference for larger firms. This generates a trade-off between efficiency and allocation bias. Quantitatively, a reduction in the required premium improves aggregate productivity by reallocating capital toward high-productivity firms. However, the size-dependent bias of institutional capital leads to misallocation by disproportionately favoring large firms, even when smaller firms have higher marginal returns. As a result, “smarter money” can paradoxically slow growth. The model highlights how changes in investor composition and liquidity regulation can have unintended real effects through capital misallocation.Social Security and the Long-Run Evolution of the Racial Wealth Gap (D3, H5)
Abstract
Using detailed Social Security calculations and the historical and modern waves of the Survey of Consumer Finances, I construct the first time series of white-to-Black per capita Social Security wealth ratios extending back to 1950, adding roughly four decades to the existing literature. I show that the racial Social Security wealth gap narrowed steadily and substantially over the 1950-2019 period. This decline coincided with the complete closing of the racial coverage gap, a sizable but less pronounced reduction in life expectancy differentials, and a slight widening of the racial earnings gap. Adding Social Security wealth to market wealth meaningfully alters the long-run evolution of racial wealth inequality: the augmented wealth gap begins at a markedly lower level and declines considerably more than the market wealth gap.Sovereign Credit Risk, U.S. Monetary Policy, and the Role of Financial Intermediaries (F3, G1)
Abstract
International asset prices are fundamentally interconnected, with common sources of variation driving global financial markets (Miranda-Agrippino and Rey, 2020) and sovereign credit markets (Longstaff et al., 2011). A significant challenge in the literature is explaining the economic nature of this common variation and linking it to macroeconomic and financial market fundamentals.We show that U.S. monetary policy stands out as a critical driver of sovereign credit spreads and that the health of global intermediaries serves as a key amplification mechanism. Using various measures of intermediary stress (He et al., 2017), we demonstrate that sovereign CDS spread sensitivity to U.S. monetary policy increases substantially when intermediary health deteriorates.
Regulatory data on dealer-country CDS positions provide granular evidence: when dealers show reduced risk capacity before FOMC announcements, evidenced by decreased sovereign position underwriting, CDS sensitivity to policy shocks increases significantly. Importantly, dealers' aggregate cross-country portfolio movements better indicate constraints than individual country positions.
We rationalize these findings with an international general equilibrium asset-pricing model where a large, advanced economy lends to emerging markets through financial intermediaries whose assets are constrained by their net worth (Gertler and Karadi, 2011; He and Krishnamurthy, 2013). Following U.S. monetary policy shocks, emerging economies adjust output, borrowing, and default risk, affecting credit pricing.
Consistent with our empirical evidence, the model generates substantial sovereign credit spread responses to U.S. monetary policy, particularly when intermediaries are constrained. These responses primarily reflect an intermediation premium tied to bank balance sheet health, affecting both low and high-risk countries. Additionally, the adverse reaction of emerging market output to U.S. monetary tightening contributes significantly to default risk and credit spread responses.
Our findings highlight that intermediary constraints are crucial for understanding how U.S. monetary policy affects global sovereign credit markets, emphasizing the central role of financial institutions in international shock transmission.
Spatial Frictions and Market Segmentation: Evidence from China's Electric Vehicle Charging Networks (L1, R4)
Abstract
How do infrastructure investments interact with local protectionism to fragment emerging markets? We study the severe spatial segmentation in China's electric vehicle (EV) industry, where local governments and automakers deploy proprietary, incompatible charging networks. By constructing a novel Charging Market Access index from granular spatial data, we quantify brand-specific infrastructure moats across cities. Reduced-form estimates reveal that these physical network disparities explicitly explain nearly 39.4% of the observed local protectionism premium, fundamentally reshaping the traditional subsidy-driven market boundaries. To quantify the welfare and competitive dynamics, we estimate a structural model of vehicle demand and oligopoly pricing. By disentangling hardware costs from implicit infrastructure and political subsidies, we show that premium automakers rely heavily on infrastructure moats for market power, while budget brands depend on traditional local protection. Counterfactual simulations of a mandated universal interoperability policy reveal substantial but highly asymmetric consequences. While eliminating infrastructure silos generates a Pareto improvement—increasing aggregate consumer surplus by up to 2.66%—it democratizes network externalities. Long-tail brands disproportionately free-ride on the expanded network, completely diluting the spatial competitive advantage of early innovators. These asymmetric gains trap local planners and incumbents in a prisoner's dilemma, explaining the persistence of fragmented networks and underscoring the necessity of federal interoperability standards to unlock aggregate welfare.Special Economic Zones as Institutional Laboratories: Crisis, Regulatory Experimentation, and Policy Diffusion (F5, K2)
Abstract
Special Economic Zones (SEZs) have become one of the most widely adopted policy instruments in development strategy, yet their role in facilitating broader institutional reform remains insufficiently understood. This paper examines how economic crises create political and institutional conditions that enable governments to experiment with new regulatory regimes through SEZs and how reforms developed within these zones diffuse into national economic policy frameworks. Focusing on the period following the 2008 global financial crisis, the paper argues that crisis conditions legitimize the creation of bounded regulatory spaces in which governments can pursue market-oriented reforms while limiting political and institutional risk.Rather than implementing comprehensive nationwide reforms, policymakers increasingly deploy SEZs as contained governance environments in which regulatory experimentation can occur under controlled conditions. Drawing on comparative institutional analysis and policy materials from Latin America and South Asia, the paper conceptualizes SEZs as institutional laboratories for economic governance. The analysis identifies three mechanisms through which zones facilitate policy experimentation and institutional change. First, SEZs create territorially bounded jurisdictions in which governments adjust tax policy, trade rules, labor regulation, and investment regimes in order to attract capital and stimulate export-oriented production. Second, zones embed administrative flexibility through delegated rulemaking authority, allowing regulatory frameworks to be adjusted iteratively in response to economic outcomes. Third, zones generate policy feedback mechanisms through which successful regulatory practices diffuse beyond the zone through legislative codification, administrative adoption, and policy emulation.
The paper argues that SEZs function as strategic instruments of crisis governance that allow states to reconcile pressures for economic liberalization with domestic political constraints. By conceptualizing SEZs as adaptive regulatory platforms rather than isolated export enclaves, the study contributes to research on institutional change and development economics by demonstrating how bounded regulatory experimentation can shape broader national economic policy reform.
Still Echoing: The Intergenerational Effects of Welfare Reform on Noncognitive Skill Formation (I3, D1)
Abstract
What are the long-run developmental consequences of redesigning the social safety net? This paper studies whether the 1996 U.S. welfare reform altered the noncognitive skill (NCS) formation of children exposed during early childhood. While prior work documents impacts on maternal employment and household income, evidence on children’s socioemotional development into adulthood remains limited, despite its central role in human capital formation and intergenerational mobility. I fill this gap by linking the PSID Main Interview, Child Development Supplement (CDS), and Transition to Adulthood Supplement (TAS), which together allow intergenerational tracking of children exposed to welfare reform into young adulthood. Using plausibly exogenous cross-state variation in AFDC waiver and TANF implementation, alongside differences in birth timing and family composition, I estimate a cohort-based triple-difference (DDD) model to isolate causal effects on validated measures of NCS. Exposure from in utero to age five significantly improves NCS among children whose mothers had at most a high school education, the group most likely to be affected by the reform, with effects largest at the earliest ages and attenuating with later childhood exposure, underscoring the importance of early-life timing for skill formation. The effects are concentrated in less stringent reform environments, larger for white than nonwhite children, and more pronounced among girls. These findings are robust under extensive specification checks confirming they are not statistical artifacts. A back of-the-envelope calculation further indicates that the skill gains translate into higher employment probability and earnings in adulthood. The results indicate that welfare reform influenced not only short-run behavior but also the long-run socioemotional development of children, with implications for how policy design shapes intergenerational mobility.Stop Believing in Haircuts (G2, E4)
Abstract
We examine how frictions in repurchase agreement (repo) markets shape the setting of haircuts and affect market outcomes. Repo haircuts—intended as a key risk management tool—should reflect counterparty and collateral risk. However, using confidential transaction-level data from the euro area’s bilateral dealer–client segment collected by the ECB, we document systematic deviations from these predictions. Haircuts are frequently set sub-optimally, suggesting that they often fail to serve their intended risk-mitigating role.We show that three core frictions—dealer balance-sheet constraints, collateral heterogeneity, and market power—jointly drive these patterns. Balance-sheet constraints, particularly those arising from leverage regulation, lead dealers to adjust haircuts in ways that reflect capital costs rather than risk. Collateral characteristics, such as convenience yields, further influence haircut setting, especially in collateral-driven transactions. In addition, the segmented and relationship-based structure of bilateral repo markets allows dealers to exercise market power, resulting in relationship-specific and often sticky haircut levels.
To rationalize these findings, we develop a theoretical model of dealer competition with capacity constraints. The model highlights how balance-sheet costs, collateral scarcity, and bargaining power interact to determine equilibrium haircuts and explains why observed haircuts can deviate from risk-based benchmarks.
Finally, we show that suboptimal haircut setting has important consequences for market functioning. When haircuts do not adjust efficiently, repo rates and trading volumes become more volatile, and monetary policy transmission weakens. Our results provide new insights into the role of financial frictions in short-term funding markets and carry important implications for financial stability and regulatory design.
Subjective Beliefs and the Choices of Community College Students (D8, I2)
Abstract
Economists are increasingly modeling decision-making with limited information and subjective beliefs. Education choice has been a fruitful area for this research, but despite making up 40% of US undergraduates, the decisions of community college students have been largely neglected. I present novel survey data from a large and diverse Maryland community college, offering the first systematic look at the subjective beliefs of community college students about outcomes related to transfer to 4-year school. Beliefs deviate significantly from rational expectations: students are optimistic about the probability of graduation after transfer but pessimistic}about the cost of 4-year school, to ambiguous net effect. Using an information experiment, I show that students update beliefs in sensible directions when exposed to new information on earnings, tuition costs, and dropout risks. I estimate a dynamic model of community college student decisions, where beliefs data allow me to avoid imposing expectations assumption and the information experiment identifies unobserved heterogeneity in tastes for schooling. Counterfactual experiments suggest a small elasticity of transfer with respect to tuition beliefs and a moderate elasticity with respect to dropout beliefs. A policy which seeks to encourage transfer with a 2-year degree as insurance against dropout faces a targeting problem: the intended targets of the policy, who are transferring without first earning the 2-year degree, have pessimistic beliefs about the returns. Instead, the policy moves students who were previously entering the workforce after earning the 2-year degree. Future work will focus on incorporating benchmark rational expectations distributions to evaluate the welfare implications of correcting biases and to simulate realized student outcomes.Tariffs, Prices, and Consumers: Unpacking the Pass-through Puzzle (F1, D4)
Abstract
This paper investigates the impact of the Trump administration’s tariffs on U.S. consumer prices, focusing on both the extent of tariff pass-through and the underlying sources of pass-through heterogeneity. By linking the U.S. Consumer Price Index (CPI) to tariff schedules through a multi-stage concordance, this paper successfully extends tariff pass-through estimates to the retail level. The results indicate that tariff-induced increases in retail prices were incomplete but economically significant. Moreover, this paper finds substantial cross-sector variation in tariff-induced consumer price responses and identifies three structural forces that jointly explain this heterogeneity. These forces—supply-chain cost absorption, consumer demand elasticity of substitution, and retail market concentration—provide a coherent framework for understanding the transmission of import tariffs to consumer prices.Tax Multipliers Across the Business Cycle (E6, E3)
Abstract
We estimate the impact on output of a shock to aggregate and different types of taxes across the business cycle. We do so for a panel of advanced economies using a new dataset of harmonized narrative-identified exogenous tax changes and a smooth transition local projection model. The output response to an aggregate tax shock is significant only during economic expansions. In recessions, the aggregate tax multiplier is insignificant, both in the short- and medium term. The state dependence of the tax multiplier is also found for income and indirect taxes, and is less pronounced for corporate taxes. Our results hold for both positive and negative tax shocks. We rationalize our findings using a model with an occasionally binding borrowing constraint that shows a stronger response of labor supply to (income and indirect) tax shocks during recessions that dampens the contraction in output.Term Structure, Interest Volatility, and the Failure of the UIP Condition (F3, G1)
Abstract
The uncovered interest parity (UIP) condition predicts that the higher-interest-rate currency should depreciate so as to equalize expected returns across currencies. Yet despite its central role in international macro models, this condition is often rejected in the empirical data. This paper argues that UIP coefficient estimates are inherently regime-dependent, a latent feature that becomes visible once one studies their term structure. Using historical Eurocurrency/LIBOR deposit rates at the 1-, 3-, 6-, and 12-month maturities, I document three empirical facts. First, rolling-window UIP coefficient estimates are negatively related to the volatility of interest rate differentials, and this monotonic relationship evolves gradually over time rather than shifting discretely around the 2008 Global Financial Crisis. Second, both the sign and the magnitude of UIP coefficients vary across maturities: under normal times, short-maturity currency trades earn larger premia in high-volatility states, whereas near the zero lower bound the premium shifts toward longer-maturity deposits. Third, these maturity patterns are difficult to reconcile with a pure ``delayed-overshooting'' explanation based solely on sluggish exchange rate responses to monetary shocks, but they are also hard to explain using a standard finance-based ``risk premium'' argument alone. Instead, the term-structure evidence from UIP regressions points to the coexistence of both mechanisms, with time-varying risk premia---shaped in part by the signs of yield curve slopes---playing a central role in the UIP puzzle.Text Sentiment about Monetary Policy (E4, C3)
Abstract
This paper uses text data fromFOMC meeting transcripts to estimate the reference levels of fullemployment, inflation, and financial conditions perceived by voting members,
and to uncover time variation in the Taylor rule parameters. We develop topic dictionaries for
economic slack, inflation, and financial markets, and infer their reference levels fromFOMC members’
sentiment using a state-space framework. The estimated employment reference level indicates that
FOMC voting members generally perceived the labor market as tighter than implied by the CBO’s
estimates between the mid-1980s and early 2000s, whereas the two measures align closely during
the Great Recession and its subsequent recovery. The perceived inflation target varies widely in the
1970s and 1980s, trends downward in the 1990s, and stabilizes slightly below two percent thereafter.
The estimated Taylor rule exhibits shifting policy weights over time—stronger emphasis on inflation
stabilization before the mid-1990s, greater responsiveness to employment deviations thereafter, and
renewed emphasis on the inflation trend following
The Double-Edged “Operation National Sword”: How Did China’s Waste Import Ban Affect Global Plastic Recycling Innovation? (O3, Q5)
Abstract
With only 9% of plastics being recycled, plastic pollution is currently one of the grand challenges facing humankind. China, owing to its abundant labor, imported and processed about 45% of world recycled plastic waste over 1992-2017. However, due to increasing environmental concerns, the Chinese government, under the initiative of “Operation National Sword,” banned plastic waste import on January 1, 2018, causing serious repercussions in the global plastic recycling industry. This study examines how this ban has affected recycling innovation both within China and in former waste-exporting countries (particularly the United States), and how any resulting directed innovation has influenced plastic recycling globally.Patent data are obtained from the China National Intellectual Property Administration, the U.S. Patent and Trademark Office, and the European Patent Office. The Cooperative Patent Classification (CPC) is used to identify patents related or unrelated to technologies for plastic recycling. Our first measure of innovation is the number of granted patents by CPC group and year. We also collect the full text of the patents and use text search to further study the focuses of the patents. A conceptual framework is developed to demonstrate forces driving recycling innovation under an import ban: the factor-supply effect and the market-size effect. Difference-in-differences (DID) and synthetic control methods are employed to quantify the impact of the import ban on innovation in different countries. Preliminary results show that the US, one of the most-affected countries by the ban, responded by increasing innovation in plastic recycling by 28% over 2018-2023 when compared to the counterfactual where the ban had not occurred, despite the output market has shrunk due to the ban. This study enriches the innovation literature by offering insights on the determinants of innovation in the context of waste recycling and circular economy.
The Dynamics of Trade Fragmentation: a Network Approach (F4, F6)
Abstract
We analyze the economic consequences of trade fragmentation between major geopolitical blocs using a novel dynamic general equilibrium model that integrates multi-sector and multi-country production and investment networks. Our framework uniquely incorporates a newly constructed investment network that captures the propagation of shocks through the supply and demand of capital goods, expanding traditional network models that focus solely on intermediate goods. Our findings highlight several important insights. First, we demonstrate that the economic consequences of trade fragmentation are amplified through investment networks, as increased costs for capital goods reduce investment rates and lower capital accumulation over time. This is especially relevant given the significant role that capital goods play in international trade and their durable nature. Unlike intermediate goods, the negative effects of trade disruptions in capital goods accumulate gradually as firms adjust their investment levels in response to rising costs. This novel mechanism adds a crucial dimension to the understanding of trade fragmentation, as traditional models focusing only on intermediate goods may underestimate its long-term consequences. Second, we provide new insights into the dynamic adjustment process by highlighting two key factors: the short-term rigidity in switching suppliers and the long-term impact on capital accumulation. The combination of these factors explains both the immediate negative effects and the delayed recovery dynamics following trade disruptions. Our results underscore the need for policymakers to account for the distinct investment network effects when assessing the economic impact of geopolitical tensions and designing strategies to mitigate the risks of trade fragmentation.The Economic Value of Precision Phosphorus Management (Q1, Q2)
Abstract
Fertilizer recommendations are typically based on field-level averages, potentially masking substantial within-field heterogeneity in crop response. This paper evaluates the economic value of precision phosphorus (P) management when yield responses vary at fine spatial scales.We use subfield-level experimental panel data from multi-year corn trials in Kentucky, a major U.S. corn-producing state providing a representative setting for precision phosphorus evaluation, where plots are subdivided into treated and control units with detailed soil characteristics. To capture heterogeneous treatment effects, we estimate a Bayesian hierarchical model that allows responsiveness to P application to vary across subplots as a function of soil conditions, including pH, buffer pH, and spatially correlated nutrient measures.
We first document that conventional soil-test-based recommendations perform well at the field level but fail to predict yield responses at the subplot level. The estimated model reveals systematic heterogeneity in returns to P application, with consistently positive effects in high-pH conditions and smaller, more variable responses elsewhere.
We then evaluate the economic implications of this heterogeneity through counterfactual simulations comparing uniform application rules to precision targeting based on posterior predictions. Precision management increases farm profits by 33–54% relative to conventional practices while reducing total P use.
These results highlight the importance of accounting for fine-scale heterogeneity in input allocation decisions. More broadly, they demonstrate how precision management can improve input efficiency and generate substantial economic gains in agricultural production.
The Effects of China's Double First-class Initiative on Research Output and Coauthorship Network (I2)
Abstract
This paper examines how China’s “Double First-class Construction” (DFC) initiative affects research output and coauthorship networks in Economics. Using difference-in-differences, we find that DFC universities publish more economics papers after treatment than non-DFC universities, driven by a larger number of authors publishing in top journals. However, research quality, measured by citations, does not improve. Using dyadic treatment effect estimation, we also find more coauthorship links involving DFC universities, including more collaborations between DFC and non-DFC institutions. Overall, the policy appears to increase the publication presence of selected DFC institutions.The Financial Impacts of Pregnancy and Childbirth (G5, J1)
Abstract
Pregnancy and childbirth represent a major yet foreseeable shock to household finances. Despite being one of the most common life events experienced by families, surprisingly little is known about how household financial outcomes evolve around the transition to parenthood. This paper provides new evidence on this question using a novel dataset that links mothers in California birth records to their quarterly consumer credit reports, creating what is, to our knowledge, the first population-scale panel connecting childbirth events to detailed consumer financial outcomes.We adapt the event study framework from the child penalty literature to credit report data and examine the evolution of four categories of financial outcomes: credit card spending, credit access, financial distress, and overall creditworthiness. Because we observe outcomes at quarterly frequency, we can distinguish mothers' financial behavior during the pregnancy period from the post-birth period.
Our results reveal two key patterns. First, mothers exhibit a deliberate "rationing" response, reducing credit card utilization sharply at the onset of pregnancy. This decline is not accompanied by changes in credit access, suggesting a voluntary demand-side adjustment consistent with a precautionary motive. Second, despite this rationing, mothers experience broad-based increases in financial distress after birth that persist over time, indicating that precautionary adjustments are insufficient to absorb the financial shock of childbirth.
We further examine heterogeneity across the income distribution by comparing mothers who use Medicaid at delivery to those who use private insurance. While privately insured mothers drive the full-sample deterioration in financial health, Medicaid mothers experience improvements after birth, consistent with expanded safety net eligibility offsetting the direct financial strain.
In ongoing work, we examine the robustness of our results to alternative specifications and event study methodologies, and extend the analysis from mothers to both parents.
The Hidden Tax in Export Rebates: VAT Credits and Formalization (H2, O1)
Abstract
Export VAT rebates are the world's most widely used export promotion tool and universally understood as an export subsidy. We show they can work as a hidden tax that compresses exports and drives firms toward informality. When input tax credits cannot be fully offset against output tax liabilities, the excess accumulates as credits carried forward, which is costly. Rebate rate increases deepen this cost. Using China's 2015 HS8-level rebate adjustments among approximately 1,300 products matched with National Tax Survey and customs data from 2013–2016, we find that all firms reshuffle exports toward treated products by 3.5% within their portfolio, but the total effect on treated products diverges sharply: flexible manufacturers — firms that can fully offset their credits — increase exports by 2%, while trapped wholesale firms — those burdened with excess credits — reduce exports by 6.3%. The policy also drives informalization across the board, through distinct channels. Flexible firms shift toward informality as the policy relaxes their need for formal invoices, releasing them from costly compliance. Trapped wholesale firms are hit eight times harder: they contract exports, abandon formal procurement, and retreat into informality. At the industry level, untreated firms partially fill the market gaps left by treated firms, but the offset is much smaller than the direct effects. Wholesale industries become more informal and more polarized overall. Our findings suggest that imperfect VAT credit refunds — a friction present in most developing countries — are a first-order determinant of how trade policy transmits to firm formalization.The Impact of a Voluntary Housing Buyout Program on Property Values: Evidence from the Kīlauea Eruption in Hawai‘i (R3, Q5)
Abstract
Following the 2018 Kīlauea volcanic eruption on the Island of Hawai‘i, the Hawai‘i County Community Development Block Grant Disaster Recovery section launched the Voluntary Housing Buyout Program in 2021 with funding from the U.S. Department of Housing and Urban Development. The program aimed to assist impacted residents, particularly low-to-moderate-income households, while acquiring properties in high-risk areas to maintain them as permanent open space.This paper examines the impact of this voluntary buyout program on real estate prices. Conceptually, the program could affect property values in opposing ways. First, buyouts may serve as highly visible signals of elevated disaster risk and potential lava inundation, depressing nearby property values. Conversely, buyouts could generate positive externalities by reducing neighborhood density and increasing open space, thereby supporting property values.
To examine the impact of the program, I employ a spatial difference-in-differences approach, comparing properties in close proximity to buyout sites against a set of comparable controls. Preliminary results indicate that property prices decline in the presence of nearby buyouts, suggesting that the negative signaling effect outweighs any positive amenity effect.
This study contributes to the nascent literature on voluntary housing buyouts and housing prices in two ways. First, while prior work has focused on urban coastal areas, it evaluates a program in a rural, geographically isolated housing market. Second, it examines a geological hazard with permanent damage, providing a unique setting to understand how markets price enduring environmental risk.
The Impact of Decreasing Pretrial Detention on the Criminal Case Outcomes of Violent Offenders (K4, K1)
Abstract
A growing body of literature shows that pretrial detention worsens defendants’ criminal justice outcomes, yet relatively little is known about how reforms that expand access to bail for defendants charged with violent offenses affect their case outcomes. This paper examines Virginia’s bail reform, which eliminated broad presumptions against bail and required judges to consider release for defendants previously subject to near-automatic detention. Using administrative court records from the Virginia Court System and a difference-in-differences design, I estimate the reform’s impact on pretrial release decisions, conviction and plea outcomes, and subsequent criminal re-offending. The results show that the reform increased pretrial release and, in turn, reduced the likelihood of conviction and guilty pleas. At the same time, rates of future offending rise, consistent with the higher baseline risk of recidivism among the affected population. These findings highlight a central policy tradeoff - expanding pretrial release can improve defendants’ immediate legal outcomes, but may come at the cost of increased public safety risks, highlighting the importance of balancing fairness and risk in bail policy design.The Incidence of Diversity: Labor Market Consequences of DEI Corporate Pledges—Evidence from LinkedIn (J7, J2)
Abstract
Corporate diversity pledges have become a central feature of U.S. hiring markets, with major firms collectively committing over $90 billion to workforce diversification predominantly aimed at advancing underrepresented minorities. To evaluate the equilibrium effects of these commitments, I construct a novel Pledge Intensity Index scoring 500 major US firms on commitment specificity and enforcement. I then link this index to comprehensive LinkedIn employment histories. Pledging firms employ over 10 million workers, representing 33% of total Fortune 500 employment. Using an event study design on the full workforce panel, I first document that pledging firms successfully altered their hiring composition, differentially increasing their nonwhite hiring share after 2020, concentrated at the entry level, with effects monotonically increasing in pledge intensity. A triple-difference specification reveals this directly narrowed the racial salary gap at these firms by $2,400. To estimate the causal impact on individual career trajectories, I narrow the focus to a panel of 2.58 million entry-level placements from 473 universities between 2014 and 2024. Exploiting predetermined university-to-firm placement networks as a Bartik shift-share instrument, I document a zero-sum demographic substitution: Black and Hispanic graduates from highly exposed universities gained access to pledging employers at the direct expense of White graduates. This displacement fell almost entirely on White male graduates, who lacked the absorptive margin available to White women via parallel gender diversity goals, fundamentally restructuring elite labor market pipelines.The Information Cost of Payoff Sensitivity in State-Contingent Sovereign Debt (F3, G1)
Abstract
GDP-linked bonds have become a recurring feature of emerging market debt restructurings, most recently in Sri Lanka and Zambia in 2023, joining earlier issuances by Argentina, Greece, and Ukraine. These instruments make debt service contingent on economic performance, reducing default risk and providing automatic fiscal stabilization. Yet every issuance has been followed by persistent trading premia over plain vanilla bonds from the same sovereign, premia that cross-default provisions cannot explain. We show that these premia reflect a structural information cost of the GDP-contingent payoff: a plain vanilla bond requires only a judgment about whether GDP will exceed the default threshold, while a GDP-linked bond requires beliefs about the entire distribution of GDP conditional on no default, making its price sensitive to volatility uncertainty that the plain vanilla bond filters out. We build the argument in two steps. First, in a binomial tree model, we show that the GDP-linked continuation value is convex in the volatility parameter, so its conditional payoff variance loads on volatility uncertainty through a channel absent from the plain vanilla bond; this result is preference-free. Second, in a dynamic model with Bayesian filtering and parameter learning calibrated to the sample countries, we show that risk aversion amplifies this asymmetry: volatility-learning shocks widen the GDP predictive distribution, but the stochastic discount factor overweights the left tail near the default boundary, reducing the risk-neutral mean of GDP and raising GDP-linked valuation dispersion even as state-uncertainty contraction simultaneously reduces plain vanilla dispersion. Bid-ask spread data from all four issuing sovereigns confirm the predicted opposite-signed pattern, significant at the one percent level throughout. Ghana, which experienced comparable distress but issued no GDP-linked bonds, shows a null result. The cost scales with macroeconomic volatility and GDP data sparseness, the conditions characterizing the emerging market sovereigns that adopt these instruments.The International Transmission of Fiscal Policy through Global Production Networks (F1, E6)
Abstract
This paper studies how U.S. government demand shocks propagate internationally through global production networks. I combine newly constructed data on federal procurement with bilateral trade flows and international input-output linkages to trace the transmission of demand shocks to foreign partner countries. Exploiting plausibly exogenous variation in defense procurement across industries, I estimate dynamic responses of partner-country exports to the United States. Preliminary results suggest that procurement shocks increase imports from foreign partners, with stronger responses in upstream, intermediate-input sectors and in countries more exposed to U.S. industries through global production networks. These findings point to a network-based channel for the international transmission of fiscal policy.The Market Value of Reproductive Rights: Evidence from U.S. Housing Markets (I1, R3)
Abstract
We estimate the causal effects of total abortion bans enacted after the Supreme Court's 2022 Dobbs decision on U.S. housing markets. Using a synthetic difference-in-differences design, we compare housing outcomes in the 13 states that implemented total bans to 25 states that maintained or expanded abortion protections. We draw on three data sources: Zillow's Observed Rent Index at the county level, Zillow's Home Value Index at the county level, and the Census Housing Vacancy Survey at the state and MSA level.We find that total abortion bans reduced rents by approximately 2.2 percent on average over the post-Dobbs period, with effects growing to 4.0 percent in the most recent year. Rental vacancy rates increased by 1.0 percentage point on average and 1.8 percentage points in the most recent year. Home values declined by 1.8 percent on average, though these estimates are not statistically significant. The combination of rising vacancies and falling rents is consistent with a negative housing demand shift, and the implied vacancy-rent relationship aligns with standard benchmarks in the rental adjustment literature. Results are robust to controlling for housing permits, suggesting that differential construction activity in ban states does not drive our findings.
Our findings are consistent with a companion paper showing that total bans increased net population outflows concentrated among single-person households, who are disproportionately renters. These results suggest that access to reproductive healthcare is a valued local amenity that capitalizes into housing markets, contributing to a growing literature on the economic consequences of abortion policy.
The Non-Wage Premium in Occupational Choice: Evidence from Job-to-Job Transitions (J6)
Abstract
Using individual career histories from LinkedIn, linked to O*NET task codes, we ask how workers value different jobs and how those valuations shift as the economy evolves. Standard measures rely on wages, but wages capture only part of what makes a job desirable. Non-wage amenities, career prospects, and working conditions remain largely invisible in existing data.We apply a revealed-preference framework to recover the total employment value of each occupation directly from workers' choices. When a worker voluntarily moves from one occupation to another, they reveal which offers greater total value. Aggregating millions of such choices produces a ranking that reflects everything workers care about, not just what employers pay.
We document striking reranking within the technology sector over the past decade. Workers increasingly moved toward roles requiring specialized infrastructure and reliability skills. DevOps Engineer and Quality Assurance rose from the bottom quartile to the top of the revealed-preference ranking across 14 million transitions and 1,497 occupations. The pattern suggests that specialization within technology, not technology broadly, drove the shift in occupational desirability. Part of this premium is non-wage and therefore invisible to standard labor market measures.
We extend the analysis in two directions. First, we link occupational choices to task content via O*NET codes to identify which skill bundles workers most actively seek out or avoid. Second, we decompose employment values at the firm-role level into firm, occupation, and match components to measure whether workers are primarily chasing employers or occupations.
These results extend the revealed-preference approach from firms to occupations, complementing wage-based evidence on occupational polarization with a measure of total employment value that includes non-wage components.
The Pass-Through of Sales Tax Exemptions: Evidence from Menstrual Hygiene Products (H2, J1)
Abstract
This paper examines how sales tax exemptions for menstrual hygiene products affect prices, using NielsenIQ Consumer Panel Data from 2004 to 2022. I estimate the pass-through of sales tax repeal using two complementary designs. First, I implement a staggered difference-in-differences framework that exploits variation in policy timing across 18 states. Second, I estimate a triple-differences specification that additionally exploits variation across products by comparing period products with deodorants, whose tax treatment is unchanged. I find that the pass-through rate ranges from 94% to 100%, implying that sales tax exemptions are slightly less than fully, or fully, passed through to consumer prices. These findings indicate that removing sales taxes on menstrual products can increase affordability and help reduce period poverty.The Patient Capital Catalyst: Government Support and Global Green Real Estate Projects (R3, G2)
Abstract
This study addresses a significant gap in the literature by exploring the role of government-related “patient capital” in financing green real estate projects globally. Despite recognized economic and social benefits of green buildings, the market faces a persistent under-supply due to high upfront costs and long payback periods. This study uncovers a critical catalyst for overcoming this inertia government-related “patient capital.” Leveraging a global project-level dataset, we demonstrate that patient capital increases the probability of a real estate project being green by 4%-5%, significantly enhances project resilience (reducing project cancellations), and improves financing outcomes (increasing funding success rates and leverage). Crucially, we find that direct financial de-risking instruments (loan or equity participation) achieve these effects more effectively than indirect support (government-affiliated sponsors but no direct project-level financial support), and that government-backed green projects access finance on terms comparable to conventional projects, a finding that signals highly efficient risk mitigation. By shifting the scholarly focus from the value to the supply of green buildings, we provide policymakers with a scalable blueprint for using public capital to strategically crowd in private investment and accelerate the transition to a sustainable built environment.The Price of Permanence: Out-of-Pocket Costs and Male Sterilization Uptake (I1, J1)
Abstract
In 2015–2017, 18.6% of women aged 15-49 relied on female sterilization for contraception, while 4.0% of men aged 18-44 had undergone vasectomies (NSFG). Despite its prevalence, sterilization has received little attention in the economics literature as compared to the extensive work on oral and long-acting reversible contraceptives. The Affordable Care Act's contraceptive mandate requires coverage of female sterilization but imposes no equivalent requirement for vasectomies, leaving men facing substantial out-of-pocket costs. Since contraception is often a joint household decision, this coverage gap may distort which partner bears the medical burden of permanent contraception. Motivated by this policy asymmetry, this paper uses coverage variation across plans, states and years to estimate the effect of price on male sterilization uptake using a nested logit model and Marketscan claims data on 30 million households from 2015–2017. Counterfactual simulations reveal that mandated vasectomy coverage would increase annual uptake by about 148–170%, with substitution primarily from female sterilization and the outside option. Additionally, older, parous households show a strong preference for sterilization, regardless of the method. These findings carry fiscal and health implications: relative to female sterilization, vasectomies are less costly, less invasive, and have lower rates of unintended pregnancy, suggesting that expanding the mandate could reduce both the cost and medical burden of permanent contraception.The Role of Industries and Occupations in the Evolution of Wage Inequality (J3, J0)
Abstract
This paper examines the evolution of wage inequality in Germany from 1985 to 2020 using matched employer-employee data. We build on existing literature that highlights the role of industry premiums and sorting of workers to industries but does not account for the occupational structure of employment. We bring industries, occupations, and firms into a unified decomposition framework allowing us to capture three layers of heterogeneity: (i) between cells, (ii) within cells but across firms, and (iii) within firms but across workers. We then trace how each layer contributed to the rise in wage inequality. We distinguish worker wages across narrowly defined industry-occupation cells, yielding more than 10,000 cells that can be consistently observed over the 35-year period.Our key findings are threefold. First, we document a substantial increase in wage inequality of about 9 log points, driven primarily by increased wage dispersion between industry-occupation cells. Second, this between-cell component is highly concentrated: more than half of the increase in between-group dispersion and nearly 40 percent of the overall rise in inequality are accounted for by just 39 cells. These cells represent around one third of total employment in 2020 and cluster in two groups: non-routine abstract occupations in high-paying industries and non-routine manual occupations in low-paying industries. Third, a decomposition of premiums attached to cells shows that sorting of occupations to industries explains 40 percent of the cell premium, followed by occupation premiums relative to other occupations within industries (26 percent) and pure industry premiums (18 percent).
Taken together, our results highlight that the evolution of wage inequality cannot be fully understood by focusing solely on firms or industries. Instead, the interaction of industries and occupational structures provides the key to understanding the sources of rising dispersion.
The Secret Handshake: The Labor Market Effects of Being a Freemason (J2, M5)
Abstract
This study investigates how membership in the Freemasons, the Western world’s oldest social network, affects career outcomes in Brazil, home to the world’s second-largest Masonic network. By matching a hand-collected dataset of over 10,000 Freemasons with administrative employer-employee records and using Propensity Score Matching, we compare labor market outcomes of network members to non-members with similar observable profiles based on age, education, industry, and firm size. We include a rich set of employer and employee controls and time fixed effects in all our models. Our findings show that Freemasons enjoy significantly better career dynamics than their matched peers. They experience higher annual wage growth, faster promotion rates, and reach higher ranks within their company's salary structure. These advantages stem from accelerated advancement rather than simply staying at a firm longer. Furthermore, membership acts as a safety net during recessions. Freemasons face a lower risk of job termination and, when unemployed, secure new jobs more quickly. Ultimately, our study highlights how social networks drive both upward mobility and job resilience during economic downturns.The Value of the Feed (D9, L8)
Abstract
How much do consumers value access to digital platforms, and what drives these valuations? We address these questions using incentive-compatible methods to elicit willingness-to-accept estimates for temporary platform deactivation, combined with a novel privacy-preserving measurement pipeline that employs vision-language and large language models to continuously classify on-device screen content. This pipeline measures usage intensity, advertising exposure, and engagement patterns at unprecedented granularity. We document substantial valuations only partially explained by direct consumption utility, with network effects and social pressure playing quantitatively relevant roles. We then implement a randomized intervention with two treatment arms: one providing users with precise information about their own opportunity costs of platform use, and another correcting misperceptions about peers' actual usage and valuations. These treatments allow us to decompose the wedge between private valuations and consumer surplus from digital platforms.The VIP Effect in Medicine: How Patients’ Insider Knowledge, Social Ties, and Organizational Rank Shape Clinical Decisions (I1, J4)
Abstract
Very Important Persons (VIPs) often receive preferential treatment across many settings. We study how patients' privileged traits, including insider knowledge, social ties to physicians, and organizational authority, drive physicians’ clinical decision-making. We leverage a policy reform that alters physicians' financial incentives without affecting patient-physician matching. Using 100% insurance claim data from a major Chinese provincial capital city, we separately examine how physicians’ responses vary by patients’ privileged traits. We show that a policy eliminating physician profits from drug sales reduces drug utilization but increases the use of other forms of care, raising overall costs without improving patient health. These responses are strongest for non-insider patients (those not working in a healthcare institution) and attenuate by comparable magnitudes at each step along the social proximity gradient: from non-insiders to insiders outside the treating hospital and to insiders within the treating hospital. However, the responses diminish only modestly between low-rank and high-rank colleagues within the same hospital. These findings suggest that preferential access to efficient care operates primarily through insider knowledge and social ties, rather than organizational rank.To Comply or Not To: A Study of Property Taxes in an Indian Municipality (H7, O2)
Abstract
Property taxes revenues are crucial to the state capacity of urban municipalities, however tax compliance is under-studied in low-enforcement settings of developing countries. In this paper, I estimate a function of property tax compliance elasticities by exploiting a set of geographical discontinuities arising from a feature of the property valuation formula. To do this, I collect, digitize and geocode 15 years of tax payment data for over 80,000 households, in the municipality of Kolhapur. I then compare the compliance and tax payment behavior of households that are located very close to each other, but fall on opposite sides of the discontinuity, or in other words, are located in different `circles'. Each circle has a different 'circle rate', and the circle rate feeds into the property valuation formula, leading to a difference in tax liabilities. I exploit all such discontinuities in my data to estimate a function of compliance elasticities. This allows me to isolate the effect of a change in tax rates on compliance and accounts for any bias arising from people with an affinity for paying taxes, selecting into better neighbourhoods. I then check for the existence of a property-tax Laffer Curve, purely due to evasion. I find that the compliance function turns flat as rates increase, suggesting that an increase in rates will unequivocally lead to an increase in revenues. I also find some evidence of partial compliance, suggesting that households who evade in one year, pay their tax in the next, but do not pay the penalty. This has large policy implications especially for municipalities looking to increase revenues in order to improve public service deliveryTraining Calibrated Simulacra Using Omissions and Chain-of-Thought (C6, C4)
Abstract
Silicon sampling methods condition large language models on demographic personas to simulate human respondents, but producing well-calibrated aggregate distributions requires Monte Carlo simulation over alternative persona configurations. We formalize this computational bottleneck as the Finetuning for Controls (F4C) problem: given a prompt and a partial control string, recover the Bayes-optimal mixture over full controls consistent with the observed fragment. We prove that training with random control omissions, computing loss only on action tokens, recovers population calibration if responses are single-round. This objective can be implemented with supervised fine-tuning, replacing brute-force simulation at inference time. Sequential settings can similarly be sped up using generative modeling of omitted controls and chain-of-thought at inference time. The resulting method yields substantial computational speedups and enables simulacra-based inference in settings where exhaustive persona enumeration previously was intractable. We validate on synthetic calibration tasks and Pew ATP political opinion data.Trump 1.0 and the War on AIDS in Africa: Evidence from the Unexpected Expansion of the Mexico City Policy (P0, I1)
Abstract
U.S. foreign aid is increasingly shaped by domestic ideological signaling rather than needs-based allocation, yet the health consequences for recipient populations remain poorly quantified. This paper examines how the 2017 expansion of the Mexico City Policy (MCP) under the Trump administration affected HIV incidence in sub-Saharan Africa. The expansion conditioned all U.S. global health funding for non-U.S. NGOs -- including HIV/AIDS assistance -- on agreeing not to provide or promote abortion-related services, with noncompliance triggering total funding loss.Using a dose difference-in-differences design that exploits sub-national variation in pre-expansion exposure to non-U.S. NGO, we find that a one-standard-deviation increase in non-U.S. NGO reliance raises adult HIV incidence by 8.1 percentage points, with significant increases for both men and women. Neonatal mortality rises by 3.5 percentage points among exposed birth cohorts.
We trace the mechanism through three linked channels. First, NGOs in high-exposure regions experienced significant funding cuts, with non-U.S. NGOs bearing disproportionate losses. Second, the cuts reduced HIV testing and treatment enrollment. Third, erosion of women's sexual autonomy increased risky sexual behavior and HIV stigma. Crucially, the MCP penalized providers based on ideological compliance rather than performance, and we show that the most effective NGOs were disproportionately defunded. This efficiency inversion explains the large HIV incidence increases we document.
The results stand in sharp contrast to evidence-based aid allocation, which redirects resources away from underperforming providers. Framed as a "pro-life" policy, the MCP expansion instead cost thousands of lives. This paper demonstrates how ideology-driven aid institutions generate systematic inefficiency and impose substantial human capital costs on recipient populations.
Unemployment Insurance and Wage Dynamics in the Post-Pandemic Labor Market (E2, J3)
Abstract
This paper studies how the unprecedented expansion of unemployment insurance during the Covid pandemic affected wage dynamics and their passthrough to inflation. Using microdata from the Current Population Survey (CPS), we show that the post-pandemic increase in aggregate wage growth was driven primarily by wage gains among job stayers rather than by job-to-job reallocation. At the same time, wage growth was disproportionately concentrated among lower-wage workers, leading to a temporary compression of the wage distribution. Complementary evidence from the Survey of Consumer Expectations (SCE) indicates that reservation wages rose sharply during the period of expanded UI benefits and declined in states that terminated these programs early. These findings point to a central role for UI in shaping workers’ outside options and firms’ wage-setting behavior. We interpret the evidence through the lens of a general equilibrium wage-posting framework with incomplete markets, in which more generous UI raises reservation wages and induces firms to increase wages within ongoing matches, particularly for lower-wage workers. Ongoing work uses this framework to quantify the contribution of different UI policies to wage growth, labor reallocation, and inflation during the post-pandemic recovery.Urban Informality in the Wake of Floods: Evidence from Mumbai (Q5, R3)
Abstract
I study a new question in a growing literature on the economic response to climate change: the impact of climate change on slum growth. In India, among areas with slums, waterlogged slums grow the fastest and are increasing in size, especially in areas of high, anomalous rainfall. Traditional drivers of slum growth exist in the Indian context –– urbanization, slower GDP per capita growth, and urban-to-urban migration are all associated with faster slum area growth –– but climate-induced changes in flooding and rainfall are also relevant. A one percentage point increase in the share of waterlogged slums is associated with 27% faster annual slum area growth between 2002 and 2012; a one standard deviation increase in the rainfall anomaly is associated with 48% faster growth. Building on this reduced form evidence, I estimate a structural model to quantify the direct impact of flooding on slum growth along with the spillovers and spatial distribution of changes throughout Mumbai. I employ a quantitative spatial model with high- and low-skilled individuals, formal and informal firms, and formal and informal housing. I incorporate flooding into the economic decisions of agents directly: through residential choice, the building development decision, and available land supply. With flooding, the value of land for formal use falls and slums sprawl into areas of high flood risk. Using the model to endogenize slum rents, I estimate a formal and informal rental price gradient by distance to flooding. I find that though flooding reduces developable land, rental prices fall with increased flooding leading to increased informality, and I calculate overall welfare under a range of flood rate scenarios. Finally, I evaluate and rank water infrastructure investments for the city.What Are Job Candidates Looking For: Prestige or Abundance? Evidence from Civil Service Exams (J3, M5)
Abstract
The recent job search literature has addressed how job posting characteristics influence candidates' application decisions. However, there is little evidence about how job candidates respond to the number of openings available for each position and their level of prestige, in terms of how difficult it is to obtain that position for those who apply for it. This project fills this gap by separating the effects of prestige and the number of openings available for a particular position and analyzing how these two aspects can influence job candidates' preferences.I use administrative data on job preferences elicited by a sample of high-performing job applicants who take centralized civil service exams to estimate the willingness-to-pay (WTP) for a higher level of prestige and for a higher number of openings in a specific job posting. I leverage quasi-experimental variation from individuals' exam scores and from job posting-specific quota rules for candidates who are eligible for openings reserved for affirmative action. This allows me to construct instruments for the individual's probability of being matched to each job and for the number of openings for each job posting. To estimate the WTP, I apply a two-stage residual inclusion and a control function selection correction to the revealed preference approach from the school choice literature in a high-stakes environment. Combining data on applicants’ rank-ordered lists and public bodies’ official job descriptions, this approach allows me to separate the effects of prestige and abundance of openings and compute how much money job applicants are willing to pay, on average, for a more prestigious job or a position with more openings.
I find that candidates with high exam scores prefer jobs that are more prestigious and have more openings, which indicates that the effects of the two characteristics move in the same direction for top-performing candidates.
What Drives Expected Stock Returns? Long-Run Risk and Endogenous Leverage (G1, E4)
Abstract
This paper studies how leverage and default decisions shape the pricing of macroeconomic risk. We develop a consumption-based structural model with endogenous corporate policies in which firms face transitory shocks to aggregate consumption (short-run risk, SRR) and persistent fluctuations in expected growth (long-run risk, LRR). Macroeconomic risk endogenously reshapes financing and default policies, primarily through long-run risk: firms optimally reduce leverage and delay default, yet these adjustments do not offset its pricing implications. Long-run risk remains the dominant source of compensation, accounting for 80–85% of firm-level equity premia and most of their time variation. In the cross section, SRR and LRR interact through endogenous leverage: greater exposure to one source raises its premium but lowers compensation for the other, as reduced leverage dampens equity’s sensitivity to the alternative risk channel. Idiosyncratic volatility is not directly priced, yet reshapes both the level and composition of expected returns by altering optimal capital structure.When a Child Falls Ill: Effects of Shared Temporary Parental Leave on Parental Caregiving and Labor Market Outcomes (J2, I1)
Abstract
This study focuses on a reform to Israel’s Temporary Parental Leave (TPL) policy, which provides paid leave to parents caring for a sick child. Before the reform, the TPL compensation scheme created costs for fathers and mothers who wished to share caregiving days. The reform modified the TPL compensation, which is now calculated jointly for both caregivers, enabling parents to share the paid leave. Using data from Israel’s Labor Force Survey and a difference-in-differences approach, I find that the reform leads to a more equitable allocation of TPL, increasing fathers’ engagement in sick child care. I also find that the reform has no effect on employment or working hours for parents already in the labor force. However, mothers in households where fathers are affected by the reform became more likely to work outside their home district, suggesting that TPL sharing policy decreases mothers’ implicit commuting costs.When School Poverty Measures Change: The Effects of the Community Eligibility Provision on School Funding (H3, I2)
Abstract
For decades, states have used students’ eligibility for free and reduced-price lunch (FRPL) as a proxy for poverty to allocate funding for disadvantaged students. However, the expansion of universal free school meal programs under the Community Eligibility Provision (CEP) has substantially altered the informational content of FRPL data, with potential implications for state K–12 funding allocation. Exploiting variation in CEP adoption across districts, this paper examines the fiscal and spending effects of CEP with linked administrative data. I find that CEP adoption increases federal food service revenue, with little impact on state or local funding sources prior to 2020. Additionally, there is no evidence of spillover effects on instructional spending on average, although substantial heterogeneity exists across states.When Women Run: Double-Edged Effects of Political Representation (P4, H3)
Abstract
This paper examines how candidates’ donations received during their political campaign, and potentially their legislative voting pattern later in Congress, change depending on their opponent’s gender. Using a regression discontinuity design that exploits close primary races, I estimate the impact of women running for the U.S. House of Representative’s office between 1980–2014 on the donations to the competing candidates. I find that Republican candidates receive more support from socially conservative groups when they run against Democratic women versus men. This impact lasts beyond the election and we see suggestive evidence that the legislators continue to vote more conservatively on reproductive rights in Congress. Democratic candidates show no change in campaign finances but vote more liberally on women’s issues after running against Republican women versus men. These results suggest that the presence of women on the ballot may lead to backlash against the very policy issues they stand for. Running but failing to win office may be costly for women, making the effort for political representation a double-edged endeavor.Where Did the Money Go? Transit Project Selection Under ARRA (R4, H4)
Abstract
Despite increasing federal investment in public transit, most American urbanized areas face stagnant service levels and ridership. This may be due to federal funds being allocated through formulas independent of demonstrated need. To study spending behavior resulting from this arrangement, I exploit the American Recovery and Reinvestment Act of 2009 (ARRA), which produced an unexpected, one-time 40% increase in federal capital funding for transit. Using a large language model, I extract structured project data from unstructured grant award descriptions and combine it with a long panel of transit outcomes. Despite the open-ended nature of ARRA awards, only about 25% of funded projects were expansionary. Project selection reveals structural heterogeneity across urbanized areas: high-demand, capital-stressed systems used ARRA to address maintenance backlogs, while lower-demand systems with newer infrastructure could afford to expand. Neither type of spending, however, produced measurable gains in service or ridership. To explain these patterns, I develop a model in which equilibrium transit usage is demand-limited rather than supply-driven. The model implies that top-down capital infusions cannot generate ridership gains without stimulating demand. Need-blind formula funding, in turn, tends to stimulate expansion where demand is weakest — the opposite of what an efficient allocation would target.Who Holds the Conditional Cash Transfer? A Structural Evaluation of Shifting Benefit Recipients from Mothers to Adolescents (D1, I2)
Abstract
Conditional cash transfer (CCT) programs traditionally target mothers to promote child welfare. However, as children reach adolescence, their growing autonomy raises questions about whether mothers remain the optimal recipients. This paper investigates how reassigning CCT payments from mothers to adolescents affects schooling, intrahousehold bargaining, and individual welfare. Using a randomized pilot within Mexico’s PROSPERA program, I develop and estimate a three-member collective household model involving a father, a mother, and an adolescent. I first provide a theoretical analysis showing that the recipient change affects schooling propensity through two competing channels: a shift in intrahousehold bargaining power and a reallocation of household resources. I then estimate the structural model to recover the members' underlying preferences and Pareto weights. The results allow me to identify the treatment effect on latent structural outcomes, including shifts in decision-making power and changes in individual welfare. Finally, counterfactual simulations evaluate alternative transfer schemes that optimize enrollment and reduce gender gaps in adolescent schooling while accounting for government budgetary costs.Who Invests Matters: FDI Origins and Global Value Chain Integration (F1, F4)
Abstract
When a multinational opens a factory abroad, the host economy may gain far more than assembly wages — or almost nothing beyond them. Which outcome obtains depends, we argue, on who is investing. We develop a general equilibrium model in which the source country's organizational compatibility with host-country suppliers endogenously determines the cross-ownership sourcing matrix of the domestic production network. The model yields a sharp prediction: the macroeconomic return to FDI — propagated through the Leontief inverse of the host's input-output structure — is governed not by the volume of inward investment but by its bilateral organizational quality. We test this prediction by combining ownership-partitioned input-output tables (OECD AAMNE), bilateral FDI stocks (World Bank HBFDI), and gravity variables (CEPII) for 77 countries, 41 sectors, and the period 2005–2020. Three findings emerge. First, foreign presence reduces local sourcing intensity on average, but the effect varies sharply by origin: European FDI is associated with significantly deeper backward linkages than Asian or US FDI in the same host-sectors. Second, local sourcing exhibits strong positive spatial spillovers — a country's domestic supplier intensity closely tracks that of its neighbors — suggesting regional supply-chain ecosystems that amplify or attenuate the enclave tendency. Third, production-network analysis reveals that foreign MNEs occupy systematically less central positions than domestic firms in host-country input-output networks; the more central a foreign affiliate becomes, the less it sources locally, consistent with hub-and-spoke enclave structures. Together, these results imply that developing countries should design FDI attraction strategies around bilateral organizational compatibility, not investment volume alone.Who Pays for Technology De-Risking? Venture Capital, Industrial Policy, and Experimentation at the Semiconductor Frontier (L5, O3)
Abstract
Venture capital is a proven driver of innovation in high-technology industries, financing experimentation, accelerating firm growth, and enabling the commercialization of new ideas. However, it does not support all forms of innovation equally. Over the past four decades, venture-backed activity has shifted away from capital-intensive technologies such as semiconductors and toward software and other domains where uncertainty can be resolved quickly and at low cost. This paper asks whether semiconductor industrial policy can reverse that pattern by altering the allocation of experimentation.The empirical setting is the rollout of CHIPS-supported semiconductor investments following the CHIPS Act of 2022. I construct a project-level dataset of CHIPS-supported semiconductor investments using official awards and Preliminary Memoranda of Terms and aggregate these to the state-year level to measure local investment intensity. The empirical strategy exploits cross-state variation in the timing and magnitude of these investment shocks in a difference-in-differences framework. I examine whether states receiving larger CHIPS-induced investment experience increases in hardware venture capital deal activity, which serves as a proxy for independent experimentation. Software venture activity is used as a placebo outcome to distinguish semiconductor-specific effects from general changes in local venture ecosystems. I complement this analysis with measures of local innovation, including semiconductor patenting.
Preliminary results show no evidence that CHIPS-induced investment increases hardware venture capital activity, despite large increases in capital deployment. Software venture activity also does not respond, suggesting that the estimates do not reflect general economic expansion. These findings are consistent with a model in which semiconductor de-risking remains internal to incumbent firms that control critical complementary assets, rather than being diffused through venture-backed entry. The paper contributes by distinguishing between expanding production capacity and enabling independent experimentation.