« Back to Results

Debt Drivers Late in the Life Cycle

Paper Session

Friday, Jan. 5, 2018 2:30 PM - 4:30 PM

Marriott Philadelphia Downtown, Grand Ballroom Salon L
Hosted By: American Economic Association
  • Chair: Karen Dynan , Harvard University

Debt and Financial Vulnerability on the Verge of Retirement

Noemi Oggero
George Washington University
Annamaria Lusardi
George Washington University
Olivia S. Mitchell
University of Pennsylvania


Debt and Financial Vulnerability on the Verge of Retirement
Annamaria Lusardi, Olivia S. Mitchell, and Noemi Oggero
Older Americans today are more likely to enter retirement in debt compared to previous
decades, with potentially serious implications for retirement security and the macroeconomy.
While the existing literature offers useful insights regarding the overall rise in debt, much less has
been done regarding older Americans’ debt patterns. Moreover, the little research that does exist
has not looked at the determinants of indebtedness close to retirement. In this paper, we evaluate
the factors associated with older American’s debt and debt management practices, to determine
whether and how these patterns have changed over time, and to evaluate whether these practices
leave people especially vulnerable in old age.
We examine older individuals’ indebtedness using the Health and Retirement Study (HRS)
and the National Financial Capability Study (NFCS). The HRS is a unique dataset with both
longitudinal/panel and cross-cohort features. The NFCS complements the HRS data in two ways:
First it provides quite recent information, and second, it includes a rich set of questions about debt
and debt management unavailable in other surveys.
Using the HRS, we compare three different cohorts of people on the verge of retirement
(age 56-61) as well as people slightly older (age 62-66). Using the 2012 and 2015 NFCS, we
explore detailed information on debt and debt management among the same age groups 56-61 and
62-66, highlighting many signs of financial distress among individuals who should be close to the
peak of their wealth accumulation. We are also able to examine the determinants of financial
fragility and overindebtedness on the cusp of retirement.
Our main finding is that recent cohorts took on substantially more housing debt and hence
face more financial insecurity than prior cohorts. Factors associated with greater financial
vulnerability include having had more children, being in poor health, and experiencing unexpected
large income declines. Thus shocks do play a role in the accumulation of debt close to retirement,
yet having resources is insufficient to protect older Americans against financial frailty. Our paper
motivates additional analysis on how debt and debt management practices can inform policy.

In Debt and Approaching Retirement: Tap Into Your Social Security or Work Longer?

Nadia Karamcheva
U.S. Congressional Budget Office
Barbara Butrica
Urban Institute


Americans’ indebtedness increased dramatically since the 1980s—a trend likely to have important implications for retirement security. On one hand, higher indebtedness might compel individuals to keep working and delay Social Security benefit claiming so they can pay off their financial obligations, but on the other hand, it may induce those who are cash-strapped and unable to service their debt to claim their benefits as soon as possible. Using data from the Health and Retirement study, we find that the incidence as well as the absolute and relative value of debt among older households has been increasing. Further our empirical findings suggest that even when controlling for other factors, having debt is associated with higher propensity to work and lower likelihood of receiving Social Security benefits. Older individuals with outstanding mortgages, as well as other debt are more likely to postpone the initial claim of their benefits and delay fully retiring from the labor force.

Portfolio Allocations of Older Americans: The Role of Cognitive Ability and Preference Parameters

Maria Casonova
California State University-Fullerton
Marco Angrisani
University of Southern California


Portfolio Allocations of Older Americans: The Role of Cognitive Ability and
Preference Parameters
Marco Angrisani and Maria Casanova
Understanding heterogeneity in household portfolios has been a longstanding aim of the financial
literature (Campbell, 2006). In particular, the low rate of stock market participation and lack of
diversification among older adults has raised concerns about the adequacy of retirement savings
(Clark, Lusardi and Mitchell, 2014). Indeed, given the gradual shift from defined-benefit to
defined-contribution pension plans, more investment responsibility has been placed on
households. As a result, retirement income security nowadays hinges as much on sound portfolio
allocation strategies as on the level of accumulated wealth.
This paper investigates the role of cognition on portfolio composition for older Americans. Our
contribution to the existing literature is twofold. First, unlike previous studies that have exclusively
focused on the relationship between cognitive ability and stock market participation (Christelis et
al., 2010), we consider a broad range of financial assets spanning the entire spectrum of
instruments available to private investors, including checking and savings accounts, CDs, Treasury
Bills, bonds, bond mutual funds, stocks, and stock mutual funds. These financial instruments differ
both in terms of their associated risk and complexity, with the least risky assets, such as checking
and savings accounts, being also the least sophisticated.
The fact that more risky financial products tend to also be more complex raises the identification
issue of disentangling the effect of cognition from that of risk preferences. Experimental studies
have shown that higher levels of cognitive ability are systematically linked to a higher willingness
to take risks (Benjamin et al., 2006; Burks et al., 2009). At the same time, risk tolerance and
cognition show the same decreasing pattern with age (Albert and Duffy, 2013). Therefore, the
correlation between cognition and portfolio allocation may be partly driven by individuals with
different levels of cognitive ability systemically exhibiting different risks attitudes.
Disentangling the two mechanisms is crucial for policy prescriptions. If poor cognition translates
into difficulties in gathering and processing financial information and learning about different
financial products, then interventions aimed at removing such barriers could be welfare-improving.
Alternatively, if the portfolio choices of those with poor cognition are determined by intrinsically
different attitudes towards risk, there is no obvious need for intervention, even when their
conservative allocation strategies result in lower retirement savings.
Explicitly accounting for the role of risk preferences is the second contribution of this paper. The
standard measures used to elicit risk preferences in survey questions, based on hypothetical
gambles, may be inadequate for this purpose. A particular concern is that respondents'
comprehension of the questions may be correlated with their level of cognition: Those with poor
cognitive skills may be less able to evaluate the expected returns and the risk associated with the
hypothetical lotteries. We propose a novel approach to deal with this issue by considering financial
outcomes for which the effects of cognition and attitude towards risk act in opposite directions,
such as health insurance and life insurance holdings.
Using data from the Health and Retirement Study, we show that older adults with low cognitivetests
scores participate less in the stock market, invest less in bonds and bond funds, and keep a
significantly larger fraction of their wealth in checking and savings accounts. At the same time,
they are also less likely to hold life insurance policies or to be covered by health insurance plans.
Our results indicate that the degree of cognitive sophistication required to access complex financial
products represents a severe barrier for less skilled individuals, and that simple preference
heterogeneity is not sufficient to explain the correlation between cognition and portfolio allocation.

The Role of Cognitive Decline on Retirement Decisions: A Mendelian Randomization Approach

Amal Harrati
Stanford University
Mark Cullen
Stanford University


The Role of Cognitive Decline on Retirement Decisions:
A Mendelian Randomization Approach
Amal Harrati and Mark R. Cullen
Stanford University Medical School
There are many reasons to believe that cognitive change is a driver of retirement
decisions. As the number of knowledge economy jobs grows in comparison to manual
labor, cognitive capacity may become more important determinants of job productivity
and job satisfaction. There is already an extensive literature of the effect that this change
in occupational structure has on the perceived and real productivity of older workers (see
Borsch-Supan, 2008 for a review of this literature) but little work exploring subsequent
retirement decisions. Secondly, cognition is closely associated with financial literacy and
levels and composition of savings, which have clear implications on the ability and desire
to retire. Finally, changes in cognition can impact preferences by changing the relative
utility of work and leisure. Changes in cognition may make work less enjoyable; or,
continued intellectual challenges from work may work to slow or reverse cognitive
The instrument is selected based on a growing body of evidence in several
scientific and medical fields that have identified specific genetic markers which possess
significant associations with specific diseases and health behaviors. While there has long
been scientific evidence suggesting that the association between genetic factors and
health is substantial (Culter and Glaeser, 2005; Di Chiara and Imperto, 1988), the recent
collection of genetic markers in large-scale, respondent-based surveys—including the
HRS--has made the integration of genetic material into social science research
possible.We exploit facts from Mendelian randomization that dictates that any given gene
is both randomly assigned from one parent to offspring, as well being independent of the
assignment of any other gene (known as Mendel’s Second Law).
Using the Health and Retirement Study survey and genetic data, we make use of a
genetic risk score for Alzheimer’s Disease in an instrumental variables framework to test
the causal role of cognitive decline on retirement decisions. After carefully testing the
validity of the genetic risk score as an instrument, we find a modest and significant role
of cognitive decline on age of retirement.
Karen Pence
Federal Reserve Board
Courtney Coile
Wellesley College
Brigitte Madrian
Harvard University
Daniel Benjamin
University of Southern California
JEL Classifications
  • I0 - General
  • Z1 - Cultural Economics; Economic Sociology; Economic Anthropology