American Economic Review: Papers & Proceedings 2009, 99:2, 374–379 http://www.aeaweb.org/articles.php?doi=10.1257/aer.99.2.374 Did the 2008 Tax Rebates Stimulate Spending? By Matthew D. Shapiro and Joel Slemrod* In an effort to bolster economic performance I. 2008 Rebate: Survey Evidence in light of a looming downturn in economic activity, on February 13, 2008, President George The survey evidence is based on a rider on W. Bush signed the Economic Stimulus Act of the University of Michigan Survey Research 2008. More than two-thirds of the $152 billion Center’s Monthly Survey, also known as the bill consisted of economic stimulus payments to Survey of Consumers. The survey provides a be sent beginning in May to approximately 130 representative sample of US households. The million households. To qualify, recipients had to survey’s core content contains the questions have a valid Social Security number, an income about expectations of economy-wide and fam- tax liability, or at least $3,000 of “qualifying ily economic circumstances that are the basis of income” that included earned income and some the University of Michigan Index of Consumer benefits from Social Security, Veterans Affairs, Sentiment. Each month, the survey includes or Railroad Retirement, and have filed a 2007 about 300 new respondents selected by random federal tax return. For the most part those who digit dial and 200 respondents reinterviewed filed as single individuals received between $300 from six months earlier. and $600, while those who filed jointly received The rider to the survey was included each between $600 and $1,200. Additionally, par- month from February through June 2008. In ents with children under 17 who were eligible the first three months the questions were asked for a Child Tax Credit and had a Social Security before the rebate payments, while in the next number received an extra $300 per child. The two months the questions were asked while tax rebate phased out at higher levels of income, households were in the midst of receiving rebate as the payment was reduced by 5 percent of the checks. The tax rebate survey module begins by amount of adjusted gross income over $75,000 briefly summarizing the tax policy change and for singles, or over $150,000 for married cou- the rebate, and then addresses the household ples. The Treasury remitted payments either by response to the rebate.1 Specifically, direct deposit, mainly in the first half of May, or by mail, mainly from mid-May through mid- Under this year’s economic stimulus pro- July. gram tax rebates will be mailed or directly The effect of these stimulus payments depends deposited into a taxpayer’s bank account. on how much was spent. This paper reports new In most cases, the tax rebate will be six hundred dollars for individuals and twelve survey evidence on the propensity of consumers hundred dollars for married couples. Those to spend the 2008 rebate. It also relates the sur- with dependent children will receive an vey evidence to aggregate data and to evidence additional three hundred dollars per child. about spending from the 2001 tax rebate. Individuals earning more than seventy- five thousand dollars and married couples earning more than one hundred fifty thou- * Shapiro: Department of Economics, University of sand dollars will get smaller tax rebates Michigan, Ann Arbor, MI 48109-1220 (e-mail: shapiro@ or no rebate at all. Thinking about your umich.edu); Slemrod: Department of Economics, University (family’s) financial situation this year, will of Michigan, Ann Arbor, MI 48109-1220 (e-mail: jslemrod the tax rebate lead you mostly to increase @umich.edu). We are grateful to Richard Curtin for advice in the design of the survey instrument, to Jonathan Parker and Nicholas Souleles for valuable discussion, and to Christian Broda and Jonathan Parker for providing a tab- 1 Shapiro and Slemrod (1995, 2003a) use a similar ulation based on research in progress. Shapiro gratefully survey to examine the spending from two earlier policies acknowledges the funding of National Institute on Aging meant to provide stimulus by putting cash into the hands of grant 2-P01-AG10179 for partially supporting the data consumers: the 1992 temporarily reduced income tax with- collection. holding and the 2001 tax rebates. 374 VOL. 99 NO. 2 DiD thE 2008 tAx REBAtEs stimuLAtE sPENDiNg? 375 Table 1—Responses to 2008 Rebate Survey Number of responses Percent Mostly spend 447 19.9 Mostly save 715 31.8 Mostly pay off debt 1,083 48.2 Will not get rebate 212 Don’t know/refused 61 Total 2,518 100 source: Survey of Consumers, February 2008 through June 2008. spending, mostly to increase saving, or that one-fifth of the 2008 rebate recipients mostly to pay off debt? mostly spent the rebate translates into an aggre- gate MPC of slightly less than one-third. This Table 1 gives the basic results of the survey MPC of about one-third should be used in com- about spending from the 2008 rebates. Because paring the survey evidence to econometric evi- there were no significant differences between dence based on the response of consumption to the answers provided in anticipation of receiving rebates, or for calibration of a macroeconomic a rebate (February to April) and those provided model of the effects of the rebate.2 during the rebate remittance (May and June), The survey contains a number of covariates we aggregate the answers over all five surveys. that provide further implications of the spending The first column of numbers gives the number plans of households and can also serve to vali- of responses to the survey (unweighted). Of the date the survey responses. Table 2 shows how 2,518 individuals asked the rebate question, 61 the fraction reporting mostly spending the rebate respondents (2.4 percent) either answered they varies by age. There is a powerful correlation of did not know what they planned to do with the the fraction spending the rebate with age. Those rebate or refused to answer. This low level of over 65 are more than 11 percentage points more item nonresponse suggests that individuals likely to report mostly spending the rebate than did not have difficulty with the question. Two those younger than 64. Overall there is a clear, hundred twelve respondents (8.4 percent) said monotonically increasing relationship between they would not get the rebate. The last column age and spending, which is what a life-cycle of Table 1 gives the fraction reporting that the model would predict for a temporary windfall. rebate led them to mostly spend, save, or pay off The policy discussion prior to the pas- debt. Of those households receiving the rebate, sage of the 2008 rebates presumed that low- 19.9 percent report that they will spend the income households are particularly likely to rebate, 31.8 percent report that they will mostly increase their spending as a result of a rebate. save the rebate, and 48.2 percent report that they The Congressional Budget Office (2008, 7), in will mostly pay debt with the rebate. Hence, the its analysis of the options for stimulus in early most common plan for the rebate is to use it to 2008, states, “Lower-income households are pay off debt, and only one-fifth plan to mostly more likely to be credit constrained and more spend the rebate. likely to be among those with the highest pro- Of course, households who will mostly save pensity to spend. Therefore, policies aimed at the rebate will do some spending, and vice lower-income households tend to have greater versa. Hence, the fraction of households who mostly spend does not necessarily equal the 2 marginal propensity to consume (MPC) in the Survey answers might not reflect households’ actual behavior. On this issue we are reassured by recent work by population. Shapiro and Slemrod (2003b) infer Christian Broda and Jonathan Parker that tracks cumula- the aggregate MPC from a reasonable param- tive purchases recorded by scanners made during the five eterization of the distribution of the MPC and weeks after household receipt of the 2008 rebate. Based on a survey question similar to ours, they find that spending the presumption that individuals will respond to increased by twice as much among those saying they had the survey that they will mostly spend if their mostly spent the rebate compared to those who said they individual MPC exceeds one-half. The finding mostly saved or paid down debt. 376 AEA PAPERs AND PROCEEDiNgs mAY 2009 Table 2—Spending the 2008 Rebate, by Age Age group Percent mostly spending 29 or less 11.7 30–39 14.2 40–49 16.9 50–64 19.9 Age 64 or less 17.0 Age 65 or over 28.4 Table 3—Spending the 2008 Rebate, by Income Income group Percent mostly spending $20,000 and under 17.8 $20,001–$35,000 21.0 $35,001–$50,000 16.6 $50,001–$75,000 18.7 $75,001 and over 21.4 Refused to state income 23.9 Total 19.9 stimulative effects.” A Hamilton Project paper II. 2008 Rebate: Aggregate Evidence on stimulus options (Douglas W. Elmendorf and Jason Furman 2008, 20) reaches the same con- Because of the electronic disbursement of a clusion. Table 3 examines the spending rate from substantial fraction of the rebates, a very large the 2008 rebate based on the survey responses. amount of extra income reached households in a It shows that the spending rate is not strongly few months. Over 80 percent of the rebate was related to income; indeed, the point estimate of disbursed in May and June, so there was a sharp the spending rate for the lowest-income group is increase in disposable income in the second smaller than the average. quarter of 2008. According to calculations by the Because the spending rate bounces up and Bureau of Economic Analysis, the rebates added down as a function of income, the correct infer- $331.6 billion at an annual rate to income in the ence from these data is that there is no dis- second quarter of 2008 and $61.4 billion at an cernible difference in spending propensity by annual rate in the third quarter. These amounts are income. Instead, the survey paints a picture of 2.2 percent and 0.4 percent of GDP in the second low-income individuals who use a cash wind- and third quarters of 2008, respectively. Given fall to pay off debt. Of those earning less than the magnitude of the stimulus payments, even if $20,000, 58 percent planned to use the rebate only a modest fraction was spent the payments to mostly pay off debt. In contrast, 40 percent would have given a boost to growth in the middle of those with income greater than $75,000 of the year and then depressed growth toward the planned to mostly pay off debt. This behavior is end of the year as their effects dissipated. consistent with the CBO’s suggestion that low- Of course, quarter-to-quarter movements in income consumers are liquidity constrained, time series have multiple causes. For example, but it suggests that low-income consumers face in the second quarter of 2008 oil prices were a liquidity constraint that will also be binding skyrocketing, affecting both nondurable spend- in the future. Hence, they place a premium on ing (including on gasoline) and depressing the using the rebate to improve their balance sheet. demand for new automobiles. We pursue the Put differently, low-income individuals are survey evidence precisely because of the diffi- needy today, but because they also are likely to culty of using time-series analysis to understand be needy in the future, they do not necessarily the impact of one-time events such as the stimu- use the windfall for current consumption. lus payments. VOL. 99 NO. 2 DiD thE 2008 tAx REBAtEs stimuLAtE sPENDiNg? 377 6 Personal saving Rebate Percent of disposable income 5 4 3 2 1 0 Jan Feb Mar Apr May Jun Jul Aug Sep 2008 Figure 1. Personal Saving and the 2008 Rebate Nonetheless, the official aggregate data on other significant events in financial and credit personal saving are broadly consistent with markets during the year, it is hard to distinguish most of the rebate being saved. Figure 1 shows the effect of the rebate from other shocks. the personal saving rate in the first nine months of 2008. The solid line shows the official data III. Looking Back at the 2001 Rebates as published by the Bureau of Economic Analysis. After hovering just above zero for As part of the ten-year tax cut bill passed by the first part of the year, the personal saving Congress in the spring of 2001, the Treasury rate spiked sharply in May, when the rebate mailed tax rebate checks of up to $300 for single program began, and through July it remained individuals and up to $600 for households from much higher than in previous months. Figure late July through late September 2001. Shapiro 1 also shows the rebate payments as a per- and Slemrod (2003a, b) report on the results of cent of disposable income. Because we do not a survey conducted in August, September, and know the counterfactual—what saving would October 2001. Only 21.8 percent of households have been absent the rebate—the figure does reported that the tax rebate would lead them to not establish what fraction of the rebates was mostly increase spending. As in the 2008 survey, spent. Yet, that personal saving jumped by there was no evidence that the spending rate was nearly as much as the increase in the stimu- higher for low-income households. lus payments provides striking circumstantial The aggregate data in 2001 show a spike in evidence for little contemporaneous spending the saving rate precisely at the same time the from the rebates. tax rebates were mailed in July, August, and We have also examined aggregate data on September 2001. The situation becomes much consumer credit for evidence of using the rebate more complex beginning in September 2001. to pay debt. There were distinct slowdowns in the The saving rate remained high in September growth in credit in 2008, but the timing of the due to a reduction in spending while the nation’s slowdowns does not exactly match the timing of attention was riveted on the terrorist attack. the rebate. Revolving credit outstanding slowed October saw a recovery in spending in all cat- in April; total credit fell in August. Given all the egories of consumption, but especially for 378 AEA PAPERs AND PROCEEDiNgs mAY 2009 automobiles in response to the zero-percent an increase in debt, though not a statistically financing and other incentives offered by auto- significant one.3 motive companies. Both the CEX and credit card studies find Two studies of the 2001 tax rebate episode use effects of the 2001 rebate on impact that are the random timing of mailing the rebate to iden- broadly consistent with the 2001 survey’s find- tify the response to the rebate in microdata. David ings. Both the survey and the CEX suggest that Johnson, Parker, and Nicholas Souleles (2006) the MPC over the first quarter after receipt of measure the change in consumption caused by the rebate is in the range of 30 to 40 percent. the receipt of the rebate using a special module The credit card evidence on debt repayment of questions added to the Consumer Expenditure supports the survey’s finding that some of the Survey (CEX). The CEX module asked house- rebate went to debt repayment, but it suggests holds when they received the rebate checks and that the debt repayment reversed after three the amount of any rebate. The authors find that months. Like the CEX findings, the credit card total expenditures, including durable expendi- evidence suggests a lagged increase in spending tures such as auto and truck purchases, did not following the rebate. respond to the timing of the check receipt in a Our surveys also provide evidence on lagged statistically significant way. When expenditures responses. Six months after the 2001 survey, are restricted to exclude durable purchases, they we added follow-up questions to a rider of the do find a significant effect. Their results show Survey of Consumers. Those who in the follow- that, during the three-month period in which up survey said that the rebate led them to mostly the rebate was received, expenditures on non- save or pay off debt were asked whether they durable goods increased by 37 percent of the would use the additional savings to make a pur- rebate check amount. They also investigate the chase later this year, or alternatively try to keep longer-run responses of spending by looking at up the higher savings (or lower debt) for at least the effect in the second and third three-month a year. The response was overwhelmingly the periods after the receipt of the check. The two- latter, with 85.3 percent and 93.4 percent plan- quarter cumulative response to the rebate was ning to maintain higher savings or lower debt, estimated to be 69 percent. This estimate fig- respectively (Shapiro and Slemrod 2003b). The ured prominently in the policy discussion prior 2008 survey asks of those who said they would to the 2008 rebate (see Elmendorf and Furman mostly save the rebate, “Will you use the addi- 2008; Congressional Budget Office 2008). The tional savings to make a purchase later this year, standard error on this estimate is 26 percent, so or will you try to keep up your higher savings for the two standard deviation confidence interval at least a year?” A parallel question was asked on it runs from 17 percent to 121 percent. Hence, of those who would mostly pay off debt. Most the Johnson, Parker, and Souleles estimate of the respondents reported that they would stick to response of nondurable expenditures in the first their plans to save or pay off debt. Of those who quarter after the receipt of the checks is broadly initially reported mostly saving, only 18.7 per- consistent with what the survey responses in cent said they would spend later. Of those who Shapiro and Slemrod (2003a, b) suggest—an initially report mostly paying off debt, only 7.8 MPC of about one-third. What differs is the percent said they would spend later. Taking these suggestion that consumption responses persisted departures from the initially reported intentions into the second, and even third, quarter after the as spending, the ultimate “mostly spend” rate receipt of the checks. from the rebate increases to 29.5 percent. Note Sumit Agarwal, Chunlin Liu, and Souleles that the survey gives respondents an opportunity (2007) also make use of the timing of the to amend their answer from either saving or receipt of the rebate checks to study payment paying off debt to spending, but not vice versa. of debt using a large sample of credit card Because of this asymmetry, the follow-up accounts from a financial institution. They find that, on average, the consumers in their sample initially increased their payments and thereby 3 Translating the magnitude of the responses they find to paid down debt. The initial decline in debt is an MPC is not feasible because the spending (and saving) statistically significant. Debt stays lower for they can measure—one credit card account—is only one three months, after which the pattern shifts to component of total spending (and saving). VOL. 99 NO. 2 DiD thE 2008 tAx REBAtEs stimuLAtE sPENDiNg? 379 questions perhaps overstate the magnitude of buck” as economic stimulus. Putting cash into the shift in the direction of mostly spend. the hands of the consumers who use it to save or In summary, the survey evidence and the evi- pay off debt boosts their well-being, but it does dence from microdata on actual spending and not necessarily make them spend. In particular, debt repayment tell very consistent stories for low-income individuals were particularly likely the contemporaneous, first-quarter, response to to use the rebate to pay off debt. We speculate the 2001 rebate: the MPC was between 30 and that adverse shocks to housing and other wealth 40 percent. The conflict between the analyses of may have focused consumers on rebuilding microdata on actual spending and debt repay- their balance sheets. Given the further decline ment and the survey evidence concerns only the of wealth since the 2008 rebates were imple- lagged effects. While survey respondents report mented, the impetus to save a windfall might be only modest incremental spending, the CEX even stronger now. Hence, those designing the and credit card data suggest otherwise. We note, next economic stimulus package should take though, that in both studies the point estimates into account that much of a temporary tax rebate for cumulative spending come with substantial is likely not to be spent. uncertainty. Thus, even with large datasets such as the CEX or the credit card data, and the inno- REFERENCES vative research design of Johnson, Parker, and Souleles and Agarwal, Liu, and Souleles, it is Agarwal, Sumit, Chunlin Liu, and Nicholas S. difficult to make precise inferences about the Souleles. 2007. “The Reaction of Consumer magnitude of the lagged effects of the rebate. Spending and Debt to Tax Rebates—Evidence from Consumer Credit Data.” Journal of Polit- IV. Conclusion ical Economy, 115(6): 986–1019. Congressional Budget Office. 2008. “Options for Only one-fifth of survey respondents said Responding to Short-Term Economic Weak- that the 2008 tax rebates would lead them to ness.” Washington, DC: Congressional Budget mostly increase spending. Most respondents Office. said they would either mostly save the rebate Elmendorf, Douglas W., and Jason Furman. 2008. or mostly use it to pay off debt. The most com- if, When, how: A Primer on Fiscal stimulus. mon plan for the rebate was debt repayment. Washington, DC: Brookings Institution. The survey estimates imply that the marginal Johnson, David S., Jonathan A. Parker, and Nich- propensity to spend from the rebate was about olas S. Souleles. 2006. “Household Expenditure one-third and that there would not be substan- and the Income Tax Rebates of 2001.” Ameri- tially more spending as a lagged effect of the can Economic Review, 96(5): 1589–1610. rebates. Nonetheless, the aggregate amounts of Shapiro, Matthew D., and Joel Slemrod. 1995. the rebates were large enough that they would “Consumer Response to the Timing of Income: have had a noticeable effect on the timing of Evidence from a Change in Tax Withholding.” GDP and consumption growth in the second American Economic Review, 85(1): 274–83. and third quarters of 2008, even if only one- Shapiro, Matthew D., and Joel Slemrod. 2003a. third of the rebates were spent. Growth in the “Consumer Response to Tax Rebates.” Ameri- second quarter was stronger and growth in the can Economic Review, 93(1): 381–96. third quarter was weaker than they would have Shapiro, Matthew D., and Joel Slemrod. 2003b. been absent the rebate. “Did the 2001 Tax Rebate Stimulate Spend- Because of the low spending propensity, the ing? Evidence from Taxpayer Surveys.” tax rebates in 2008 provided low “bang for the Policy and the Economy, 17: 83–109.
Pages to are hidden for
"2008 Tax Rebates"Please download to view full document