Economics of Household Saving
July 21, 2012
Greg Kaplan, University of Pennsylvania and NBER, and Giovanni Violante, New York University and NBER
A wide body of empirical evidence, based on randomized experiments, finds that 20-40 percent of fiscal stimulus payments (for example, tax rebates) are spent on nondurable household consumption in the quarter that they are received. Kaplan and Violante develop a structural economic model to interpret this evidence. The model integrates the classical Baumol-Tobin model of money demand into the workhorse incomplete-markets life-cycle economy. In this framework, households can hold two assets: a low-return liquid asset (such as cash or a checking account) and a high-return illiquid asset (such as, housing or a retirement account) that carries a transaction cost. The optimal life-cycle pattern of wealth accumulation implies that many households are "wealthy hand-to-mouth": they hold little or no liquid wealth despite owning sizeable quantities of illiquid assets. They therefore display large propensities to consume out of additional income. The authors document the existence of such households in data from the Survey of Consumer Finances. A version of the model parameterized to the 2001 tax rebate episode generates consumption responses to fiscal stimulus payments that are in line with the data.
Claus Kreiner, David Lassen, and Soren Leth-Petersen, University of Copenhagen
Basic consumption theory predicts that households facing tight credit constraints, and therefore a high price on liquidity, respond more strongly to a fiscal stimulus. Kreiner, Lassen, and Leth-Petersen test this hypothesis directly using a unique dataset on household-specific marginal interest rates, computed from third-party-reported administrative records of individual loans/deposits, in combination with a fiscal stimulus reform that transformed the illiquid pension wealth of Danish households in 2009 into liquid wealth available for consumption. The authors find substantial variation in the price of liquidity across households. This variation in credit constraint tightness is a strongly significant predictor of the marginal propensity to consume. The observed differences in the price of liquidity are strongly correlated with the level of financial asset holdings more than a decade before the stimulus, which suggests that differences in the tightness of credit constraints across people in the data are the result of heterogeneity that persists beyond what can be explained by short lived shocks appearing within the duration of a typical business cycle. Differences in the price of liquidity therefore seem to be driven by differences in the demand for liquidity, implying that credit constraints are self imposed.
Henrik Cronqvist, Claremont McKenna College, and Stephan Siegel, University of Washington
Analyzing identical and fraternal twins matched with data on their savings propensities, Cronqvist and Siegel find that genetic variation explains about 33 percent of the variation in savings behavior across individuals. Parenting effects on savings behavior are strong for those in their twenties but decay to zero by middle age -- that is, parents do not have a lifelong non-genetic impact on their children's savings. The family environment when growing up and an individual's socioeconomic status later in life moderate genetic effects, so that more supportive environments result in a stronger genetic expression of savings behavior. The authors also find that savings behavior is genetically correlated with income growth, smoking, and body mass index, suggesting that the genetic component of savings behavior reflects innate time preferences and lack of self-control. In a world moving towards individual retirement savings autonomy, understanding the deeper origins of individuals' savings behavior is becoming increasingly important.
Lorenz Kueng, Northwestern University
Although theoretical models often emphasize fiscal foresight, most empirical studies neglect the role of news, thus underestimating the total effect of tax changes. Measuring the path of expected future tax rates from the yield spread between taxable and tax-exempt bonds, Kueng finds that the consumption of high income households increases by close to 1 percent in response to news of a 1 percent increase in expected after tax lifetime income, which is consistent with the basic rational expectations life-cycle theory. Using novel high-frequency bond data, Keung then develops a model of the term structure of municipal yield spreads as a function of future top income tax rates and a risk premium. Testing the model using the presidential elections of 1992 and 2000 as two natural experiments, he shows that financial markets forecast future tax rates remarkably well in both the short and long run. Combining these market-based tax expectations with consumption data from the Consumer Expenditure Survey, he demonstrates that households with lower income, less education, or that are more credit constrained respond less to news. However, the same households also respond one-for-one to large news shocks, consistent with rational inattention. Overall, the results in this paper suggest that ignoring anticipation effects biases downward the estimates of the effect of fiscal policy .
William Gale, Brookings Institution; Michal Grinstein-Weiss, Clinton Key, and William M. Rohe, University of North Carolina, Chapel Hill; and Mark Schreiner and Michael Sherraden, Washington University in St. Louis
Gale, Grinstein-Weiss, Key, Rohe, Schreiner, and Sherraden examine the long-term effects of a randomized experiment in Tulsa, Oklahoma conducted during 1998-2003. During that period, Individual Development Accounts offered low-income households 2:1 matching funds for housing down payments. Prior work has shown that among households who rented in 1998, homeownership rates increased more through 2003 in the treatment group than for among the control group. These authors show that control group renters caught up rapidly with the treatment group after the experiment ended. As of 2009, the program had an economically small and statistically insignificant effect on: homeownership rates, the number of years respondents owned homes, home equity, and foreclosure activity among baseline renters.
Ralph Koijen, University of Chicago and NBER; Stijn Van Nieuwerburgh, New York University and NBER; and Motohiro Yogo, Federal Reserve Bank of Minneapolis
Koijen, Van Nieuwerburgh, and Yogo develop a pair of risk measures for the universe of health and longevity products including life insurance, annuities, and supplementary health insurance. Health delta measures the differential payoff that a product delivers in poor health, while mortality delta measures the differential payoff that a product delivers at death. Optimal insurance choice simplifies to the problem of choosing a portfolio of health and longevity products that replicates the optimal exposure to health and mortality delta. For each household in the Health and Retirement Study, the authors calculate the health and mortality delta implied by its ownership of life insurance, annuities including private pensions, supplementary health insurance, and long-term care insurance. For the median household aged 51 to 58, the lifetime welfare cost of market incompleteness and suboptimal insurance choice is 17 percent of total wealth.