Are Opt-out Retirement Savings Programs Robbing Peter to Pay Paul?

stock image of coinsBy James Choi, Professor of Finance, Yale School of Management

  • Opt-out retirement savings programs have much greater enrollment than opt-in programs
  • Greater retirement savings do not appear to increase credit card debt or harm credit scores
  • Increasing retirement savings does lead to higher first mortgage and auto loan balances

Economics Nobel Laureate Richard Thaler recently declared, “Retirement savings is probably behavioral economists’ greatest success story.” Over the past two decades, retirement savings policy around the world has been changing significantly in response to findings from behavioral economics research. The key insight has been about the power of automaticity.

I was fortunate to have been able to contribute to the early research in this area. My collaborators and I showed that changing from opt-in to opt-out (i.e., automatic) 401(k) enrollment dramatically increases 401(k) participation rates (Choi et al., 2004). Recent data from 1,900 Vanguard-administered retirement savings plans (Vanguard, 2017) bears out the original findings. Among employees with less than one year of tenure, the average participation rate is 38% under opt-in enrollment. This rises to 78% among employees with ten or more years of tenure. In contrast, the average participation rate under automatic enrollment among those with less than one year of tenure is 50 percentage points higher (88%), and it remains 17 percentage points higher (95%) among those with ten or more years of tenure.

These kinds of dramatic results, coupled with concern that many households are not saving enough for retirement, have spurred widespread adoption of automatic enrollment. In 2016, 45% of Vanguard plans featured automatic enrollment, up from only 15% in 2007. Abroad, automatic enrollment has become an integral part of the pension systems of the U.K., Turkey, and New Zealand. Even in Afghanistan, automatic enrollment has been shown to have effects of similar magnitude as in the U.S. (Blumenstock et al., 2017).

But our knowledge about the effects of automatic enrollment has an important blind spot. Although it is indisputable at this point that automatic enrollment can be used to increase contributions to retirement savings plans, we have limited visibility into its effects on the rest of the household balance sheet.

The hope is that the additional retirement account contributions are being funded by reduced current consumption, but if those marginal contributions are instead being funded by withdrawals from other accounts or increased debt, then automatic enrollment may not be accomplishing its intended purpose of increasing economic security in retirement.

The principal obstacle to answering the question of automatic enrollment’s total balance sheet impact is a lack of data that contains information on both retirement savings accounts and the rest of the household balance sheet. But in a recent paper (Beshears et al., 2018), we were able to perform a part of this linkage. We obtained data on contributions to the Thrift Savings Plan—the 401(k) equivalent for U.S. federal government employees—and linked it to credit reports for civilian employees of the U.S. Army. Despite their employment by the military, these employees looked rather similar to typical U.S. workers along many dimensions.

James Choi, professor of finance at the Yale School of Management

James Choi, professor of finance at the Yale School of Management

In the research, we exploited the natural experiment created when the Army adopted automatic enrollment for civilian employees hired from August 1, 2010 onwards. The default contribution rate was 3% of income, the most common 401(k) savings rate default in the private sector. Importantly, employees hired before August 2010 were never subjected to automatic enrollment. Therefore, we can compare employees hired shortly before August 2010 to employees hired shortly after August 2010 to identify the effect of automatic enrollment.

We first confirmed that automatic enrollment has the same kind of effect on retirement plan contributions among Army civilian employees as has previously been documented elsewhere. After controlling for macroeconomic shocks that affect both the pre- and post-automatic enrollment cohorts equally, we found that automatic enrollment increases cumulative contributions to the Thrift Savings Plan by 3.9% of income on average at four years of tenure. This positive average effect comes entirely from the left tail of the savings distribution: while automatic enrollment causes those who wouldn’t have otherwise saved in the TSP to start contributing, it has no effect on those whose contribution rates are at or above the median.

We used the same statistical method to identify the effect of automatic enrollment on debt. We defined three different measures of debt. The first measure is all debt excluding auto loans and first mortgages. The second measure adds auto loans to the first measure. The third measure encompasses all debt. The reason we treated auto loans and first mortgages differently is that these types of debt are typically used to obtain new assets. Since both the assets and liabilities of the balance sheet increase, the net worth impact of increased auto or first mortgage debt is considerably muted. In contrast, credit card debt incurred to purchase a nondurable good decreases net worth dollar for dollar.

Ultimately, we found no evidence that automatic enrollment increases debt excluding auto loans and first mortgages. So the good news is that the worst-case scenario—automatic enrollment funding retirement contributions through expensive credit card debt—does not seem to be true. Nor does automatic enrollment have any significant effect on credit scores. On the other hand, we found that at four years of tenure, automatic enrollment increases auto loan balances by 2.0% of income, and first mortgage balances by 7.4% of income.

What should we make of these increases in secured debt? The increases in auto debt are probably the most worrisome for long-run wealth accumulation. Whether employees take out larger auto loans because they have less non-TSP wealth available or because they are buying a more expensive vehicle (a rapidly depreciating asset), the increase in auto loans is likely to have a negative impact on net worth. Larger first mortgages have a more ambiguous impact. Since TSP balances can be borrowed against in order to finance a down payment, the TSP assets generated by automatic enrollment may have eased a down payment constraint for employees, allowing them to buy a larger home. Homes often appreciate in value, and the payment schedule of a mortgage may force households to save more than they otherwise would have, so the long-run net worth impact of a larger first mortgage may actually be positive.

Like much research, this latest paper raises as many questions as it answers. At the very least, it hints at the possibility that automatic enrollment may have unintended consequences that were invisible to the first generation of research that had data only on retirement savings accounts. I hope in the future to get data on which vehicles and homes employees subject to automatic enrollment purchase, as well as data on their other financial accounts, so that we can continue get a broader and broader sense of how this important policy affects household welfare.

Dr. James Choi currently works as a Professor of Finance at the Yale School of Management. Dr. Choi is a two-time recipient of the TIAA Paul A. Samuelson Award for Outstanding Scholarly Writing on Lifelong Financial Security. He is a TIAA Institute Fellow and has also served as the Associate Director of the Retirement Research Center at the National Bureau of Economic Research and as a member of the FINRA Investor Issues Committee. With research interests such as behavioral finance, behavioral economics, household finance, and sociology, his work on default options has led to changes in 401(k) plan design at many U.S. corporations and has influenced pension legislation in the U.S. and abroad.


Practitioners may find the following resources valuable for more information on this topic:

Clark, Jeffrey W., and Jean A. Young, 2018. “Automatic Enrollment: The Power of the Default.” Vanguard Research.

Koenig, Gary, 2012. “The Case for Automatic Enrollment in Individual Retirement Accounts.” AARP Public Policy Institute.


Blumenstock, Joshua, Michael Callen, and Tarek Ghani, 2017. “Why Do Defaults Affect Behavior? Experimental Evidence from Afghanistan.” American Economic Review, forthcoming.

Beshears, John, James Choi, David Laibson, Brigitte Madrian, and William Skimmyhorn, 2018. “Borrowing to Save? The Impact of Automatic Enrollment on Debt.” Working paper.

Choi, James, David Laibson, Brigitte Madrian, and Andrew Metrick, 2004. “For Better or for Worse: Default Effects and 401(k) Savings Behavior.” In David Wise, editor, Perspectives on the Economics of Aging. University of Chicago Press, Chicago, pp. 81–121.

Vanguard, 2017. How America Saves 2017: Vanguard 2016 Defined Contribution Plan Data.

Leave a Reply

Your email address will not be published.

Skip to toolbar