Are Tax Refunds Used Wisely?

Transcription

1 Are Tax Refunds Used Wisely? A Look at How Americans Spend their Refunds Aron Szapiro Policy and Finance Expert, HelloWallet February 2016

2 Are Tax Refunds Used Wisely? A Look at How Americans Spend their Refunds Every spring, Americans reconcile their taxes, determining how much they actually owed to the IRS and how much they paid throughout the year. Typically, Americans pay more taxes than they actually owe. In fact, the IRS sent refunds to 83% of taxpayers in 2015, averaging $2,800. Although Americans may enjoy the moment when they get their refund, does lending money to the IRS (at no interest) throughout the year and collecting it every spring hinder workers from achieving their financial goals? In this paper we find that: Only 37% of workers saved their tax refunds. The rest spent their refunds on a mix of durable or nondurable goods, or used them to catch up on bills or credit card debt. About 38% of workers used their refunds to make unusually large purchases. The larger the refund, the more of it workers spent. This result suggests that workers regard their tax refunds as a windfall that entitles them to spend additional money. It also suggests that workers deliberately overwithheld their taxes, in some cases to save for big purchases. About one quarter of workers allocated a large share of their tax refunds to paying down credit card debt. These increases in payments were substantial, as the workers that used their tax refunds to pay off credit card debt increased their payments by slightly more than $2,000 after they received their tax refunds. Workers could reduce interest payments from credit card balances and bolster their retirement savings by reducing their tax withholdings. Tax refunds seem to reduce workers financial security by encouraging them to spend refunds rapidly on nondurable goods instead of saving them. By reducing the value of each paycheck, overwithholdings also may increase the likelihood that workers will start to accrue credit card debt. Are Tax Refunds Used Wisely? 2

3 Introduction 01 Contents Methodology 02 Findings 03 Conclusion 06 Endnotes 09

4 Introduction Every spring, Americans reconcile their taxes, determining how much they actually owed to the IRS and how much they paid throughout the year. Typically, Americans pay more taxes than they actually owe. In fact, the IRS sent refunds to 83% of taxpayers in 2015, averaging $2, Although Americans may enjoy the moment when they get their refund, does lending money to the IRS (at no interest) throughout the year and collecting it back every spring hinder workers from achieving their financial goals? The permanent income hypothesis suggests that Americans should avoid tax refunds in favor of paying no more than they owe. Instead, people should (the theory goes) want to smooth their consumption over their lifetime, and certainly their consumption over the year. 2 Despite what this theory suggests, 4 in 5 taxpayers defer much of their income until the calendar year after they earn it when they get their tax refund. By overpaying the IRS throughout the year, these workers forgo interest they could earn on these funds if they saved them. And, as extra tax payments cannot be refunded until filing season, these workers have shifted their savings to a completely illiquid place which is a problem because most workers have insufficient emergency savings. 3 On the other hand, some workers may overpay their taxes as a deliberate savings strategy; in other words they precommit to saving money by giving it to the IRS to hold for them until filing season. 4 For these workers, the fact that they cannot access their money is a feature rather than a bug of overpaying their taxes, precisely because they lose access to their money and are forced to save it. While this might be an easier way for households to save, it could be quite costly (and ineffective) if taxpayers subsequently take on more debt throughout the year. Another reason workers may overwithold is that some workers might find it too difficult to navigate complex tax rules to set their withholdings in line with their actual taxes. The IRS has guidance for filing its W-4 form (the form workers submit so employers can withhold the correct federal income tax from the worker s pay) to roughly match withholdings to likely tax liability. But precise matching requires a more-detailed calculation of projected taxes. Because there is a penalty for underpaying, workers may be reluctant to attempt to align their withholdings to their actual taxes for fear of making an error. Although the form s instructions suggest that workers should consider completing a new Form W-4 each year and when [their] personal or financial situation changes, inertia may be a key reason why many workers overwithhold. 5 Workers may also be reluctant to adjust their withholdings because they re used to receiving tax refunds and enjoy getting them. Additionally, workers who are eligible for the Earned Income Tax Credit or the Child Tax Credit may have no choice but to collect their refund yearly when they file. This is because workers can collect a refund even if they have no federal income tax liability. But, there is no mechanism to get these refunds other than when filing. Almost 28 million people are eligible for the EITC, with credits totaling more than $66 billion, so this issue is significant. Empirical work on tax refunds has largely examined the marginal propensity to consume these refunds and whether or not the permanent income hypothesis holds, given an increase in predictable income. In other words, theory would predict that people would save almost all of their tax refunds, since these refunds would simply increase lifetime income. Researchers have examined to what extent refunds have actually been consumed. For example, using the Consumer Expenditure Survey, Souleles found that 64% of refunds are spent and that spending increases on both durable and nondurable purchases. He also found that liquidity-constrained households households that cannot easily borrow increase spending on nondurable items more than households that can borrow, although both types of households do spend their refunds. In sum, he rejects the permanent income hypothesis based on the fact that spending does change in response to predictable tax refunds. A key reason that Americans may consume their tax refunds instead of using them to increase lifetime in- Are Tax Refunds Used Wisely? 01

5 come is mental accounting, meaning they perceive tax refunds differently from other sources of income. For example, Thaler argues that people think about small gains as income and spend them, but larger gains (like many tax refunds) are added to asset accounts and are not spent as much as income would be spent. When evaluating this claim, however, Souleles did not find strong evidence in his study that the higher the refund, the more likely it was to be saved. Another finding is that people build up credit card debt and use tax refunds to pay it down. In a study using credit card data, Agarwal, Souleles, and Liu found that consumers used tax rebates in 2001 (which were $300 to $600) to pay down credit card debt. 6 They also found that many households then rebuilt debt on their credit cards in the months after receiving their rebates. These results have the advantage of being based on actual credit card data instead of self-reported survey data, yet they mirror the survey data: About half of respondents indicated they would use their tax rebates to pay down debt. 7 Other studies of tax refunds have focused on the EITC, which is the largest antipoverty program in America. 8 Many of these studies explore whether the EITC encourages work, but there has been little published on the practical effects of giving low-income workers a substantial portion of their refunds as a lump sum. Tach and Halpern-Meekin argue that EITC-recipients use the refunds to save for larger purchases than they would otherwise be able to save for if they had access to their money throughout the year. In this paper, I find that although some workers use their tax refunds as a forced savings device, many spend the refunds quickly on nondurable goods the month after they receive it. This suggests that workers would be better off forgoing refunds and collecting more of their salary throughout the year. I also find that many people build up credit card debt throughout the year and then pay it off with their tax refunds, leading to unnecessary interest charges. In sum, workers could benefit from guidance on how to optimize their tax withholdings to best support their financial goals. This guidance could help them save money on credit card interest and set more funds aside for long-term financial goals such as retirement rather than spending their refund quickly. Methodology In this paper, I analyzed HelloWallet user data to investigate spending before and after users received their 2014 tax refunds. HelloWallet is a financial application that helps users understand their finances by aggregating data from credit cards, checking accounts, and investment accounts. Companies that subscribe to HelloWallet distribute it to their workers. HelloWallet membership is a little younger and higher-income than the general public, so the results in this paper may not be representative of the U.S. population. (Workers in our tax-refund sample have a median income of $75,000 as opposed to around $50,000 for households in America; most workers in our sample are under age 50.) Nonetheless, our study of this group reveals interesting behavioral insights and complements existing work on tax refunds. HelloWallet software can often categorize transactions (e.g., credit card charges to McDonald s would be categorized as restaurants). When our software cannot do so, users often categorize the data themselves. Exploiting this unique data set helps add to the literature on tax refunds because I can analyze transactions across different credit cards, checking accounts, and savings accounts. By examining the financial memoranda associated with their tax refunds, I was able to identify exactly when 3,055 users received direct deposits of their 2014 refunds. I was also able to identify changes in spending across different categories such as restaurants, transportation, and home goods. Table 3 in the appendix shows the change in spending in different areas and the highlevel categories I organized these transactions into for presentation in this paper. I examined spending transactions two months before and after workers received their tax refunds. In both cases, I defined a month as a 31-day period so that monthly charges, such as rent or car payments, would Are Tax Refunds Used Wisely? 02

6 be captured without risking undercounting these cyclical expenses. I also examined how much of the tax refund a worker saved by comparing a worker s total average liquid account balance in the months before and after the tax refund. I used average balances to cope with monthly expenses that could have otherwise led to spurious results. To identify whether the size of the refund was associated with an increase in spending, I ran the following ordinary least squares regression, using robust standard errors on the users for whom I could identify a tax refund: Spending = β 0 + β 1 income + β 2 liquid + β 3 taxref + U Spending is the change in the month before and after the tax refund; income is the user s income; liquid is the sum of savings, checking, and money market assets; and tax refund is the size of the tax refund. I ran similar regressions for increases in spending on the largest item a worker purchased to determine whether workers used tax refunds to make unusually expensive purchases. Figure 1: Key Uses for Tax Refunds Source: Author s analysis of HelloWallet data. I identified a number of workers who increased their credit card spending. By searching for credit card memos labeled as interest charges, I was able to identify some credit cards that had revolving balances, but it is likely an undercount that failed to capture every card with a revolving balance. I used this information to examinewhether users built up credit card debt prior to their refunds. Findings Only 37% of Workers Saved Their Tax Refunds Only 37% of workers saved their tax refunds; the rest spent their refunds or used them to pay down debt. After receiving their refund, spenders spent more on almost every category HelloWallet tracks, as shown in Table 3 in the appendix. Fully 25% of workers spent some of their tax refunds to pay down credit card debt. In terms of using their tax refund as a forced savings device, about 38% of workers used their refunds to make a large purchase. Workers who used their refunds did so very quickly. In fact, workers spent 80% of their refunds (or used them to pay down debt) within a month of receiving them. Workers seemed to treat tax refunds as bonuses to be enjoyed: average total spending among those who did not save their refund shot up almost 73% in the month following the refund. 9 It seems likely that some workers mentally accounted for tax refunds as extra money to be used for more-frivolous spending. Or, perhaps a sudden windfall made workers feel wealthier, and therefore more inclined to spend. In general, workers spent their refunds on a mix of durable and nondurable goods, as well as on a variety of transactions that are harder to identify, as shown in Figure 2. (See Table 3 in the appendix for a crosswalk of HelloWallet s transaction categories to the meta categories I used in this analysis.) Average cash withdrawals went up $1,200 in the month following the refund. Although it seems likely that this spending was largely on restaurants or entertainment, we don t Are Tax Refunds Used Wisely? 03

7 Figure 2: Average Change in Spending in Key Areas After Refunds Within Different Time Periods (Among Workers who did not Save their Refund) Source: Author s analysis of HelloWallet data, restricted to users who increased spending following receipt of their tax refunds. Note that loan payments are excluded from this figure. know what workers actually bought with this cash. Average spending on nondurable purchases that we can track (such as groceries, restaurants, clothing, and travel) also increased sharply by $590. Note that all changes from one month before to one month after refunds were received are statistically significant by at least the 1% level. Bill payments went up about $320 on average, which indicates that some workers were behind on their bills and used the refunds to settle up with their utility companies. Many workers likely perceived tax refunds as bonus money, because they increased spending on nondurable items such as clothing and dining out. The largest category of spending is uncategorized, which likely includes durable, big-ticket items such as a new car and nondurable, big-ticket items such as a vacation. Therefore, the increase in nondurable spending did not necessarily exceed the increase in durable spending. Later in the paper, I examine the relationship between refunds and large purchases. The Larger the Refund, The More of it Workers Spent Contrary to Thaler s theory, I found that the larger the absolute size of the refund, the more of it a worker spent. Using an ordinary least squares regression (with robust standard errors), we can see that every dollar increase in the tax refund was associated with an average increase of 71 cents in spending, holding liquid assets and income constant, as shown in Table I also used the ratio of the tax to income as a regressand along with liquid assets and I found no statistically significant relationship between the size of the refund relative to income and the amount users spent. This result suggests that workers regard their tax refunds as a windfall that entitles them to spend additional money. It may also suggest that workers deliberately overwithheld their taxes, in some cases to save for big purchases. Quantile analysis reveals that workers who spent the most were also the most likely to increase their spending in response to an increase in their refund. When we focus on workers in different quantiles (as shown in Table 2), the strongest effect of an increase in tax refunds on spending was for the distribution at the 90th percentile and the 75th percentile of increased spending. This may indicate that people who spent the most did so because they intentionally overwithheld their taxes in order to make a large purchase and planned to spend their entire tax refund, regardless of its size. Table 1: OLS Regression of Ordinary Least Squares Regression of Change in Spending and the Size of the Tax Refund (Controlling for Income and Liquid Assets) Change in Spending After Receiving Refund Coefficient t Score Tax Refund 0.714** (3.15) Income (1.28) Liquid (0.61) Constant (0.72) R N 2, * p < 0.05 ** p < 0.01 Source: Author s analysis of HelloWallet data. Are Tax Refunds Used Wisely? 04

8 Table 2: Quantile Regression of Change in Spending and the Size of the Tax Refund (Controlling for Income and Liquid Assets) Quantile Quantile Regression Estimate Coefficient t Score ** (3.89) ** (8.94) ** (14.46) ** (16.16) ** (8.27) N 2, * p < 0.05 ** p < 0.01 Source: Author s analysis of HelloWallet data. A substantial portion of workers 38% used their tax refunds, at least in part, to make a big purchase. The mean increase in large transactions before and after tax refunds was $4,109 and the median increase was $927. Tax refunds added a boost to existing savings, as some workers used the tax refund to support purchases that were much larger than their tax refunds. After receiving their tax refund, 25% of users made unusually large purchases that were more than twice as large as their refund. For example, a worker might contribute a tax refund to their savings for a car, using the tax refund to buy a nicer car or to defray some of the cost. About One Quarter of Workers Allocated a Large Share of Their Tax Refunds to Paying Down Credit Card Debt As discussed earlier, 25% of the sample increased payments on their credit cards in response to receiving their tax refunds. These increases in payments were substantial, as shown in Figure 3. In fact, the average credit card payment increased by slightly more than $2,000 in response to receiving a tax refund. (Note that all month-to-month changes in Figure 3 are statistically significant at the 1% level.) This pattern of using tax refunds to pay down credit card debt, which is usually high-interest debt, indicates that deferring income until refund season is not a good financial strategy for some of the workers in our sample. These individuals would be better off collecting their income as they earned it to avoid building up debt.f3 It is important to keep in mind that the workers who paid off credit card debt with their refunds are not low-income or liquidity-constrained. They had a median income of $75,000 and a mean income of $91,162. These workers held median liquid assets of $7,806 and mean liquid assets of 19,825. Due to limitations in our software, it is hard to know for certain whether Figure 3: Average Change in Credit Card Payments Before and After Receipt of a Refund Among Those Who Used Their Tax Refund to Address Credit Card Debt Interestingly, the behavior of these workers seems to violate permanent income hypothesis, as most of them had sufficient liquidity to make similar purchases without tax refunds. People who spent on big-ticket items had a median liquid savings of $7,920. Just 30% of these spenders needed the tax refunds to make their purchases; the rest had sufficient liquid assets to make a similar purchase without a refund. Workers would have had a median balance of $4,880 after making their purchase even without the tax refund helping them. In other words, tax refunds don t help people buy big-ticket items so much as they make it seem financially prudent to buy a big-ticket item. Source: Author s analysis of HelloWallet data. Are Tax Refunds Used Wisely? 05

9 the credit card payments were to pay down revolving debt. In more than half of cases, however, I can see that these workers had revolving debt. Given that spending did not generally increase in the month prior to the refund, it seems very likely that the payments were for revolving debt, rather than to account for increased spending in the previous month. Workers Could Find the Money to Reduce Unnecessary Interest Payments and Bolster Their Retirement Savings by Reducing Their Tax Refunds Tax refunds seem to reduce workers financial security by encouraging them to rapidly spend these funds on nondurable goods instead of saving them. It also encourages workers to take on credit card debt unnecessarily. Given that 25% of workers spent an average of $2,000 to cope with credit card debt, I estimate they could have saved about $160 in interest over the course of the year by reducing their withholdings to better match their tax liability and by paying their debt throughout the year. 11 Workers might base their spending on previous years refunds, which can be a dangerous strategy. Tax refunds vary based on life circumstances and factors workers may not be aware of. Children age and are not eligible for the child tax credit. Incomes increase. Mortgage interest declines as people pay off their homes these are just a few common reasons why tax refunds can decline. Workers who spent their refunds quickly, particularly on nondurable items, increased their standard of living for a brief period of time in response to a mismatch of their withholdings and actual tax liability. These spenders, who increased spending on nondurable goods by $600 on average, would be better off reducing their refunds and retaining those funds throughout the year. The spenders could have bought the same goods without the refund, since most had more than $600 in liquid assets. Instead, people experience a wealth effect that makes them spend more than they normally would after receiving their tax refunds. Though $600 may not seem like a lot of money, workers could reduce their refunds by changing their withholdings, then put this money in a retirement account without affecting their day-to-day standard of living. For young workers, this small adjustment could add up to about $34,700 (in constant dollars) of extra assets in retirement, if the money achieved historically modest 4% real returns. One place to look for a boost to retirement income might well be adjusting tax withholdings. Even the people who use their tax savings as a forced savings device to make a large purchase would be better off receiving their money throughout the year and automatically allocating it to a savings account. Tax withholdings are completely illiquid and are not available for emergencies. Unless an individual has done a detailed calculation in advance, it s impossible to know the precise refund amount, making it difficult to save for a particular large purchase. If the tax refund is greater than what the worker expects, he or she may be tempted to spend more than usual or spend the excess frivolously. Conclusions When policymakers add credits and deductions to the tax code, they should remember that many workers are not well-served by withholding more money and receiving large tax refunds. As I have shown, people often do not use their tax refunds rationally. Many spend their refunds relatively quickly, and much of it on nondurable goods. Some people do use these refunds as a forced savings device for large purchases the best use of a tax refund while other users build up credit card debt throughout the year and then pay this debt down with their tax refunds. In an ideal world, workers would pay taxes they owe as they owe them, allocating extra money for savings or living expenses. The workers who use their tax refunds to pay off short-term debt would be much better served by keeping the money throughout the year so they could avoid building up this debt. (The same is true for workers who deferred paying their bills and used tax refunds to catch up, presumably while incurring some penalties.) Workers who use their tax refunds as a forced savings device, but have demon- Are Tax Refunds Used Wisely? 06

10 strated the ability to save without such a device, could also benefit from having liquid savings rather than being forced to wait for a refund. Employers could help employees financially by providing better counseling on how to fill out their W-4 forms so they don t defer income if that doesn t work well for them. Policymakers should also consider the likely response to increasing tax refunds by adding in new credits and consider ways to help people collect this money throughout the year rather than as a lump sum. A tax refund is not bonus money, just money that was overwithheld. Because workers live without this money until they receive a refund, many could learn to use it more effectively to meet other goals beyond a brief spending binge every spring. Are Tax Refunds Used Wisely? 07

11 Appendix Table 3: Crosswalk of HelloWallet Categories to Spending Categories in this Paper and Average Spending One Month Before and After Among Those Who Spent Their Refunds in These Categories HelloWallet Category Spending Category for this Paper One Month Before One Month After P Value for One Month Before and After Bank Fees Bills $74.45 $ Cable/Int/Phone Bills $ $ Car/Transportation 1 Nondurable $ $ Cash Cash $ $1, Children/Family Durable $58.41 $ Clothing Nondurable $ $ Coffee Shops Nondurable $18.65 $ Credit Card Payments Credit Card Payments $1, $1, Donations Other $21.71 $ Education Other $59.01 $ Entertainment Nondurable $ $ Gifts Nondurable $23.16 $ Groceries Nondurable $ $ Health Nondurable $ $ Home Durable $ $ Income Excluded $31.04 $ Insurance Bills $ $ Job Expenses Excluded $59.77 $ Loan Payments Loan $ $ Mortgage/Rent 2 Other $ $ Other Bills Bills $ $ Personal Care Nondurable $54.24 $ Reimbursements Excluded $15.05 $ Restaurants/Bars Nondurable $ $ Taxes Excluded $84.68 $ Transfers Excluded $5, $6, Travel Nondurable $ $ Uncategorized Other $1, $3, Utilities Bills $ $ This category mostly includes gas and spending on public transportation, not the purchase of new vehicles. 2 For many users, we cannot track these payments as they are made by check. Source: Author s analysis of HelloWallet data. Are Tax Refunds Used Wisely? 08

12 End Notes 1 Internal Revenue Service. July 20, Friedman, Milton Windfalls, the Horizon, and Related Concepts in the Permanent-Income Hypothesis (Department of Economics of the Graduate School of Business of the University of Chicago). 3 Lusardi, Annamaria, Daniel J. Schneider, and Peter Tufano Financially Fragile Households: Evidence and Implications. National Bureau of Economic Research, Working Paper No (May): 4 Souleles, Nicholas S. September The Response of Household Consumption to Income Tax Refunds. The American Economic Review. Vol. 89, No.4, p Jones, Damon Inertia and Overwithholding: Explaining the Prevalence of Income Tax Refunds. National Bureau of Economic Research, Working Paper No (May): 6 Agarwal, Sumit, Chunlin Liu, and Nicholas S. Souleles The Reaction of Consumer Spending and Debt to Tax Rebates Evidence from Consumer Credit Data. National Bureau of Economic Research, Working Paper No (December): 7 Ibid. 8 Center for Budget and Policy Priorities. July 20, This spending figure is the total change in spending from the 31 days before the refund was received to 31 days after, including increases in debt payments. 10 Note that we have incomplete data on user demographics and income data, leading to a sample for this analysis of just 2,712 individuals. 11 I assumed a 18% APR on a hypothetical credit card, with debt accrued evenly throughout the previous year. Are Tax Refunds Used Wisely? 09

13 For More Information Sales Team Research Team 2121 Ward Court NW, 3 rd Floor Washington, DC