“Tipping Points” Symposium — June 9, 2016
Ray Boshara, Federal Reserve Bank of St. Louis
Thank you to all the authors, discussants, and participants of the Tipping Points Symposium. The roundtable revealed a number of fascinating and sometimes counterintuitive research findings that we hope will inform future research, private and non-profit practice, and public policy.
The Center for Household Financial Stability at the St. Louis Fed was founded on the idea of studying the entire family balance sheet of struggling and middle-class families, not just their savings and assets, which was the prime focus of the asset-building field for most of its 25 years of existence. Our view is that the debt side of the balance sheet has been relatively understudied but increasingly recognized as critical to the financial well-being of families and overall performance of the U.S. economy—a view long held by Richard Vague of the Governor’s Woods Foundation and our excellent symposium partners at the Private Debt Project.
As my colleagues William Emmons and Lowell Ricketts observe in their paper, debt ratios are higher now than before the boom. And as my colleagues Don Schlagenhauf and Lowell demonstrate in the preview of our new report, The Quarterly Debt Monitor, consumers have rapidly increased borrowing for non-housing related debt, especially in the form of auto and student loans.1
In addition, our partner Sherle Schwenninger of the Private Debt Project observes that the U.S. economy has, over time, become more dependent on debt to fuel economic growth: “American households, in particular, have become dependent on debt to maintain their standard of living in the face of stagnant wages. Rising levels of private debt have also fueled consecutive investment asset bubbles, whose bursting not only caused the Great Recession but also left a large and burdensome debt overhang that is still being dealt with today. The entirety of America’s debt build-up from the 1990s to 2008 was the result of a dramatic increase in private debt, not public debt.” 2
While many challenges remain in understanding household debt—including challenging theoretical and empirical questions, some of them presented in an excellent paper by Jonathan Zinman 3—our symposium goal was, of course, to make some progress on the notion of a “tipping point” in household debt. Some questions we considered include: At what point does debt move from being wealth-building and productive for families and the economy to being wealth-depleting and destructive? Is this just an economic analysis, or are non-economic factors at play as well? How do decisions at the micro level impact the macro level, and vice-versa? It’s my sense that we did, in fact, succeed in starting to address some of these and other questions in many areas of household debt.
Let’s briefly review the papers to see how each advances the idea of a tipping point in household debt. Emmons and Ricketts, using a demographic lens to examine the effects of income shocks on delinquencies, find that the tipping point signified by a loan delinquency is more likely to be reached among younger, less-educated and non-white families—but then they raise the interesting question of whether that’s because they make riskier financial choices, or because structural, systemic forces (e.g., socio-economic factors, discrimination) largely shape financial behavior. The implications are stark, and diverging: If risky behavior largely drives financial choices, then we have an immense financial education and financial capability challenge; if, however, systemic forces primarily drive those choices, then we face a public policy challenge that must view delinquencies as one barometer of a larger problem of families regularly living with financial risk.
In their paper, Carlos Garriga, Ricketts and Schlagenhauf suggest the likelihood of a household reaching a tipping point should be based on the monthly debt payment-income ratio, as distinct from the more commonly used monthly income-debt ratio. This conclusion is based on an analysis of the household budget constraint, which includes available assets, the liquidity of those assets, the ability to secure additional debt, and the structure of that debt. Further, they argue that this broader measure of leverage could help predict tipping point outcomes such as bankruptcy and foreclosure. Using a large consumer level data set, they document how household debt payment-income ratios behave before and after a financial default, as well as evaluate how this measure performs in predicting individual debt outcomes. Garriga et al. thus leverage unique data to create a dynamic metric which is both illuminating and predictive of household “tipping points,” and thereby advance our understanding of family fragility.
These first two papers, which examine income shocks and a more robust measure of household leverage, then nicely set up the paper by Daniel Alpert and Robert Hockett, who discern a tipping point in the effects of household debt on consumer behavior and the macro economy. Alpert and Hockett observe that the composition of household debt has shifted in the last several years away from mortgage debts and towards student loans and consumer credit, especially auto loans, and that the pattern of consumer spending since 2013 closely tracks access to these forms of credit. This household dependence on debt is a continuation of a much longer trend that began several decades ago. Their data show that, starting with the era of debt-fueled growth in the mid-1980s—and, especially since the Great Recession—rising levels of household debts should be viewed as “a canary deep in the coal mine of (a) strapped households borrowing to make ends meet, and (b) less-strapped households borrowing to maintain their standards of living.” Compounded by weak accumulation on the asset side of family balance sheets, economic growth then precariously relies upon ongoing growth in household debt—a reliance, they observe, that led to “bubbles, busts, and subsequent debt-deflation in the first place.”
Daniel Cooper, Barry Cynamon, and Steve Fazzari, like Garriga et al., also attempt to construct a household tipping point measure beyond debt-income ratios. They employ balance sheet and income data from the PSID to project lifetime resources and compare those resources with household consumption to assess what they call “household financial sustainability.” To them, a household is sustainable if current consumption “can be maintained through its remaining working and retirement years without leading to negative net worth in its terminal year.” Their preliminary results show that “while the vast majority of American households were sustainable in the mid-1980s, sustainability declined sharply through the early 2000s and remained low until the eve of the Great Recession. Sustainability improved modestly in the aftermath of the crisis, but remains well below levels from two decades ago.” Surprisingly, and conflicting in some ways with the findings of Alpert and Hockett, Cooper et al. find that the impact of eliminating all debt from family balance sheets is small. Also surprisingly, they find that additional education had little impact on sustainability, largely because those with higher levels of education have both more resources and more expenses; across education levels, no group was spared the decline in sustainability. Their prime concern, given widespread financial unsustainability, is retirement security.
Neil Bhutta and Benjamin Keys shed new light on the concept of a tipping point by examining the understudied area of home equity extractions, which they report totaled nearly $1 trillion between 2002-2005 (not counting using extracted funds to move into more expensive homes or to buy second homes). This level of extraction—averaging around $40,000 among those households that actually extracted equity—was fueled by both historically low mortgage interest rates (monetary policy) and rising home prices and, unexpectedly, led to little debt consolidation among extractors. Bhutta and Keys also found that extractors were more likely to become delinquent on their mortgages, even after controlling for credit score and many other risk factors. Remarkably, those who extracted toward the end of the housing boom in 2006 were more than twice as likely to become severely delinquent on their mortgage debts over the subsequent four years, and they were also almost 40 percent more likely to become delinquent on non-mortgage debt as well. Their paper accordingly reveals two interesting tipping point observations: One, a family’s largest source of collateral for borrowing becomes a tool for wealth accumulation (such as to finance a small business) for some, but a means to wealth destruction (such a through higher delinquency rates and possible loss of assets) for others. And two, changes in monetary policy affect changes in equity extraction and thus financial risk in the household sector, especially when the risk of a house price correction is salient.
Christopher Foote, Lara Loewenstein and Paul Willen, also looking at the housing sector, find that, despite popular perceptions, the large increase in mortgage debt before the financial crisis was not sparked by higher rates of mortgage borrowing among lower-income communities: Between 2001 and 2006, the richest quintile of zip codes accounted for about $1.5 trillion of new mortgage debt compared to about $320 billion for the lowest quintile, and defaults across income categories rose in rough proportion as well. In their view, overall mortgage debt rose in the early 2000s not because borrowing constraints for marginal borrowers were relaxed, but rather because households across the income distribution thought housing was a good investment. In other words, a tipping point (of excessive mortgage borrowing leading to a crash) was fueled not by a financial indicator (the amount of income needed for a mortgage), but by a psychological one (the expected increase in future housing prices). This finding reminds me of an observation by economist Chris Carroll who, in a 2013 keynote address for our Center at the St. Louis Fed, argued that household deleveraging was also primarily driven by the psychological notion of expectations—that deleveraging is over when people think it is over. 4
And finally, Zachary Bleemer, Meta Brown, Donghoon Lee and Wilbert van der Klaauw examine the tipping point that resides at the intersection of college costs and homeownership in early adulthood. Average enrollment-weighted college tuition and fees per school year, measured at the level of the state, rose by $5,385 between 2001 and 2009. How are recent cohorts absorbing this surge in college costs, and what effect is it having on their post-schooling consumption? In aggregate analysis of tuition, educational attainment, and debt patterns for nine youth cohorts across the states, Bleemer et al. find that the 2001-2009 tuition hike is associated with a $4,021 increase in average student debt per capita among 24-year-olds from 2003 to 2011. However, they estimate no meaningful response to tuition in the state-cohorts’ graduation rates. The estimates are consistent with American youth having absorbed tuition shocks not by foregoing schooling, but instead by amassing more debt. Further analysis demonstrates that the $5,385 real tuition climb, despite leaving graduation unchanged, can explain between 1 and 4 of the observed 7-percentage point decline in age 28-30 homeownership over 2007-2015 for these same nine cohorts. The results suggest that states that increase college costs for current student cohorts can expect to see a response not through a decline in workforce skills, but instead through weaker spending among young consumers in the years to come. Naturally, given the rise of student loan balances documented by several authors in this volume, as well as research by William Elliott and IlSung Nam as well as Richard Fry of the Pew Research Center (among others) showing lower levels of net worth—due to lower levels of homeownership, retirement savings, family formation, small business ownership, etc.—among college graduates with loans compared to those without loans, the findings of Bleemer et al. suggest that multiple or cascading tipping points are triggered by rising college costs.5 6
Beyond the papers, the discussion yielded some other perceptive observations as well. One, echoing Foote et al., as well as Emmons and Ricketts, was that we must go well beyond the balance sheet to understand the notion of a tipping point in household debt. Systemic, demographic and sociological mechanisms may propel lower-income and disadvantaged families to a negative tipping point, while other institutional mechanisms may drive a more advantaged family to a positive one. There’s also an aspirational element: when a young adult does not pursue her goal of homeownership because of student debt, an aspiration is forgone; this, in turn, may affect the perception of other debt, wealth-building or not. Finally, there’s a generational element: a debt- and risk-averse young millennial may have been spooked by the consequences of easy mortgage and consumer credit on his Generation X parents, who may still be recovering from the Great Recession; he may instead aspire to own or owe nothing, simply getting by in the “gig” economy. Each of these observations, of course, is fodder for further research.
Given stagnating wages and weak income gains over the last generation; greater income volatility and vulnerability to income shocks across the income spectrum; growing wealth and income inequality; a global savings glut; millions of households still recovering from the wealth losses of the Great Recession; and lagging public and household spending by policymakers, overall household debt levels are likely to remain high and possibly rise, making tipping point discussions potentially even more important. We look forward to our next symposium, which will have more of an emphasis on the effects of household debt on macro-economic performance.