The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of Chicago Why Do Banks Reward their Customers to Use their Credit Cards? Sumit Agarwal, Sujit Chakravorti, and Anna Lunn REVISED December 20, 2010 WP 2010-19 Why Do Banks Reward their Customers to Use their Credit Cards? Sumit Agarwal, Sujit Chakravorti, and Anna Lunnα December 20, 2010 Abstract Using a unique administrative level dataset from a large and diverse U.S. financial institution, we test the impact of rewards on credit card spending and debt. Specifically, we study the impact of 1 percent cash-back reward on individuals before and during their enrollment in the program. We find that the marginal increase in spending per month during the first quarter of the program is $68. Average monthly payments decreased more than the marginal increase from cash-back rewards resulting in card debt increasing an average of $115 during the first quarter. Evidence from the credit bureaus confirms that consumers offset their increased spending and debt on their rewards card by lowering their spending and debt on their other credit cards. Segmenting the data by different types of cardholders, we find that cardholders who do not use their card prior to the cash-back program increase their spending and debt more than cardholders with debt prior to the cash-back program. We also find heterogeneous responses by demographic and credit constraint characteristics. Key Words: Household Finance, Credit Cards, Consumption, Financial Incentives, Rewards JEL Codes: D1, D8, G2 The authors would like to thank Gene Amromin, Marc Bourreau, Rich Rosen, Marianne Verdier and seminar participants at the Federal Reserve Bank of Chicago and the University of Paris Ouest Nanterre la Défense for helpful comments and suggestions. The views expressed in this research are those of the authors and do not necessarily represent the policies or positions of the Federal Reserve Board or the Federal Reserve Bank of Chicago. α All three authors work in the Economic Research Department of the Federal Reserve Bank of Chicago and can be reached at: sagarwal@frbchi.org, sujit.chakravorti@chi.frb.org, and anna.lunn@chi.frb.org, respectively 1. Introduction Today, rewards are routinely given by airlines, hotel operators, and credit card issuers to increase use of their products. In the case of credit cards, rewards are an effective way to attract cardholders or convince existing ones to use a specific card for their purchases and borrowing needs. In 2005, six billion reward card offers were mailed by the credit card industry. Typically these mailing are randomized and the response rates are very low. For instance, in 2005, the response rate was 0.3% (also see Agarwal, Chomsisengphet, and Liu, 2010). Card companies have pursued aggressive tactics, such as offering cash back, airlines miles, rebates and lower interest rates. The main objective of the card companies is to increase card spending that may result in cardholder’s debt in the future.1 In this paper, we study the impact of credit card rewards on spending and debt. We explore three questions. First, do consumers spend more when given rewards? Second, do consumers increase their debt because they receive rewards? Third, do consumers partially or fully offset their increases in spending and debt accumulation by reducing spending and debt on their other credit cards? We find that consumers generally spend more and increase their debt when offered one percent cash-back rewards. The impact of a relatively small reward generates large spending and debt accumulation. On average, each cardholder receives $25 in cash-back rewards during our sample period. We find that average spending increases by $68 per month and average debt increases by over $115 per month in the first quarter after the cash-back reward program starts. The greater increase in debt compared to spending suggests that average monthly payment drops more than the marginal increase in spending from the cash-back program. Specifically, we find a 1 Many websites offer tips to smartly choose the rewards programs. For example, www.rewardcreditcardsite.com suggests the following 7 tips – do not carry a balance, know what “UP TO” means, what the limit, etc. 1 reduction of payments within the first quarter of $38 of the start of the program, suggesting that the marginal increase in spending due to the cash-back reward is converted into debt along with a portion of baseline spending. Furthermore, evidence from credit bureau data confirms that consumers substitute their spending from other cards to the card with cash-back and decrease debt on their other cards. Finally, even in the long run, we find a persistent increase in spending and debt. Specifically, the average spending and debt rise during the nine months subsequent to the cash back reward is $76 and $197 per month, respectively. The reduction in payments is $83 during the same nine months period. We identify certain types of cardholders that are more responsive to the cash-back rewards program. Cardholders that do not carry debt have a larger response to the cash-back program. We find that 11 percent of inactive cardholders during the three months prior to the cash-back program used their cards to make purchases of at least $50 in the first month of the program. Specifically, inactive cardholders increase their average per month spending by $220 during the first quarter and their average per month spending only decreases to $180 during the first nine months. Their average per month increase in debt during the first quarter is $167. We find that these cardholders substitute spending and debt accumulation from other cards to the cash-back card. Cardholders react differently to cash-back rewards based on some demographic characteristics. Average per month spending increases by $55 by single cardholders and by $95 by married cardholders during the first quarter. Similarly, single cardholders increase their average per month debt by $65 as compared to $111 by married cardholders during the first quarter. We do not find significant differences between male and female cardholders. Cardholders that earn less than $40,000 increase their average per month spending by $47 as 2 compared to $74 for cardholders that earn more than $40,000 during the first quarter of the program. Those earning below $40,000 accumulate $56 additional debt on average per month versus $87 for cardholders earning more than $40,000 during the first quarter. Credit constraints also impact the response to the cash-back program. Not surprisingly, those cardholders with higher credit limits tend to spend more and accumulate more debt per month on average in response to the cash-back program. Cardholders utilizing less than 50% utilization of their credit limits tend to spend more and accumulate more debt per month. We are also able to study another tool to increase card usage and debt, albeit more costly, to convince cardholders to increase their debt: APR reductions. During our sample period, the financial institution offered certain cardholders a 10 percent APR reduction. Consistent with Gross and Souleles (2002), we find that consumers react to such a large reduction in APRs by increasing card spending and debt. However, we find that only part of this increase in spending contributes to an increase in the consumer’s balance for all her credit cards, which suggests that consumers shift spending and debt from other cards. Our paper incorporates key features from several strands of the literature in economics and finance – consumer payment choice, consumption response to income shocks, and behavioral finance. We tie our work to each of these fields and highlight our contribution. First, the literature on payment substitution argues that monetary incentives are effective in enticing consumers to use a given payment instrument over another. While the literature focuses on different types of payment instruments, our analysis suggests that these incentives are also effective in differentiating providers of the same type of payment instrument. Second, we incorporate findings from the consumption literature that study monetary payouts such as tax rebates and their impact on increased spending and debt. Our results confirm one of the main 3 findings in this literature that only a small financial incentive is required to change consumer behavior. Third, the literature on time-inconsistency suggests that at least some consumers increase their spending and debt when offered financial rewards but may incur greater debt than expected. Given our ability to study a cardholder’s overall portfolio, we are able to distinguish between increase in spending and debt on a specific card and how that affects a consumer’s overall balance sheet. In addition, our results also have policy implications. For instance, the recent regulatory and legislative actions have focused attention on the impact of rewards on consumer choice of payment instrument and who pays for these rewards. Some observers have argued that the recently passed Card Act and recent changes to overdraft access for debit cards in the United States would reduce the ability of issuers to extend rewards.2 While mandated reduction in cardholder fees and finance charges may potentially affect the level of rewards, we find that rewards have significant impact on credit card debt especially via substitution from another issuer’s credit card suggesting that rewards are an effective tool to steal customers from a financial institution’s competitors. The rest of the paper is organized as follows. Section 2, reviews the literature. Sections 3 and 4 outline the data and provide results, respectively. Finally, section 5 concludes. 2. Background During the past decade, there has been a growing literature documenting the changing nature of consumer finance due to the explosive growth of credit card usage. For instance, in 1970, credit card related consumer debt totaled $2 billion as compared to $626 billion in 2000. 2 For details about the 2010 Card Act, see http://www.federalreserve.gov/consumerinfo/wyntk_creditcardrules.htm. For details regarding recent debit card overdraft rules, see http://www.federalreserve.gov/consumerinfo/wyntk_overdraft.htm. 4 In 2007, servicing credit card debt (interest plus minimum payments) represented about 14% of disposable personal income. According to Kennickell, Mach, Bucks and Moore (2009), 29 percent of households with a percentile net worth less than 20 percent carry credit card debt, as compared to 38 percent of household with a percentile net worth above 90 percent carry debt. Credit cards serve two main purposes. First, they serve as a payment device in place of cash or checks for millions of routine transactions. Most credit cards have a grace period whereby interest charges can be avoided by paying off the outstanding balance in full at the end of the month. Second, they are the primary source of unsecured open-end revolving consumer credit, competing with bank loans and other forms of financing.3 The theoretical payment card literature focuses on how the costs of payment cards are distributed between banks, merchants and card holders through prices. These models generally conclude that banks may charge fees in excess of their costs to merchants and extend incentives to cardholders to increase card adoption and usage (Baxter, 1983; Chakravorti, 2010; Rochet and Tirole, 2002). These models focus on adoption and usage of payment cards vis-à-vis other payment instruments. The results are dependent on various model parameters including the degree of competitiveness in the market for goods and payment services along with consumer and merchant demand elasticities. For the most part, this literature does not focus the extension of credit.4 Debate continues as to who pays for credit card rewards and their social welfare implications. Some U.S. merchants have complained that financial institutions are funding their credit card rewards by extracting merchant surplus (Jacob, Jankowski, and Lunn, 2009). Theoretical models focus on other funding sources for credit cards rewards. For example, 3 There are other payment instruments that share these characteristics, e.g. checks and debit cards with overdraft protections. For more discussion about linkages between consumer payment and credit, see Chakravorti (2007). 4 Bolt and Chakravorti (2008), Chakravorti and To (2007), and Rochet and Wright (2010) are notable exceptions. 5 Chakravorti and Emmons (2003) argue that rewards are funded by those that borrow in the form of higher interest rates. More recently, Schuh, Shy, and Stavins (2010) argue that cash users subsidize these rewards because merchants are unable to separate credit card users from other payment instrument users by charging more to credit card users.5 There is anecdotal evidence from merchants suggesting that rewards are effective in convincing consumers to substitute credit cards for debit cards to reduce their payment costs. IKEA, a large furniture store operating in several countries, imposed a 70 pence surcharge on credit card transactions in their United Kingdom stores resulting in a 15% decrease in credit card usage (Bolt et. al, 2010). Given the relatively high average transaction size at IKEA, only a relatively small financial incentive was required to change consumer behavior. Some policymakers have intervened in the pricing of payment services to reduce consumer incentives to use their credit cards to make purchases especially when consumers do not avail the extension of credit. The Reserve Bank of Australia (RBA) argued that credit card rewards partially funded by fees charged to merchants distorted the efficient choice of payment instruments by consumers. The RBA (2008) estimated the benefit to consumer of using their credit cards as purely a payment device as AUS$ 1.30 for each AUS$ 100 spent. To reduce the incentive for consumers to use credit cards, the RBA mandated around a 50 percent reduction in the interchange fee (fees paid by merchants’ financial institutions to issuers that are paid for by merchants) along with other policy changes. Several empirical studies use consumer surveys to study the impact of rewards on payment instrument usage (Ching and Hayashi, 2010). Borzekowski, Kiser and Ahmed (2008) 5 In some cases, merchants are unable to charge different prices based on the type of payment instrument to make purchases. However, in cases where merchants have the ability to do so, few merchants actually set different prices. Regardless of the reason, one-price policies may result in cross-subsidies between users of different payment instruments. 6 examine survey data on debit card usage and find that financial incentives are effective in steering customers to one payment type. Twenty-one percent of survey respondents cited pecuniary reasons for substituting between payment types, and many in this group explicitly cited credit card rewards. Zinman (2009) also observes consumer price sensitivity to payment choices. Specifically, he finds that credit card users with balances are more likely to use their debit card because interest accrues immediately after credit card purchases are made. Our paper is the first to empirically test the effect of credit card rewards not only on payment choice but on change in spending and debt. The consumption literature considers permanent and transitory shocks to consumption. Giving consumers cash rewards for spending using a certain device increases their consumption because they are receiving money for purchases that they would have made without the incentives. A number of papers have studied consumers' response to a permanent predictable change in income, as a means of testing whether households smooth consumption as predicted by the rational expectation life-cycle permanent-income hypothesis. Using credit card data, Gross and Souleles (2002) find a marginal propensity to consume of 13% and for accounts that had an increase in credit limit. They also find that debt levels rise by as much as $350. Souleles (1999) finds that consumption responds significantly to the federal income tax refunds that most taxpayers receive each spring. Both of these papers find evidence of liquidity constraints.6 Aaronson, Agarwal, and French (2007) find that following a minimum wage hike, households with minimum wage workers often buy vehicles. The size, timing, persistence, composition, and distribution of the spending response is inconsistent with the basic certainty equivalent life cycle model. 6 Other related studies include Wilcox (1989, 1990), Parker (1999), Souleles (2000, 2002), Browning and Collado (2001), Hsieh (2003), and Stephens (2003). 7 There have been four recent studies, using micro data, by Shapiro and Slemrod (2003a and 2003b), Johnson, Parker, and Souleles (2006) and Agarwal, Liu, and Souleles (2007) on the 2001 tax rebates. Given the conflicting findings of the consumption literature, we cannot form a hypothesis about whether consumption will increase in response to these cash rewards. Moreover, our case becomes more complicated when we consider the relationship between the rewards and spending. While a consumer may receive cash for transferring all their spending from their debit card to this credit card, she also has an incentive to increase her spending and use the credit line attached to the credit card. Therefore, we look to the behavioral literature to find predictions about how a reward program will affect consumers’ overall debt level. The seminal paper on time inconsistency is Ausubel (1991) who finds that consumers often ignore the interest rate on credit cards because when they make purchases they fully intend to pay back but change their mind when the bill comes. Agarwal, Chomsisengphet, Liu and Souleles (2006) find that consumers both under- and over-estimate their spending on the card. More recently, behavioral economists have extended this time inconsistency feature in several directions (Heidhues and Köszegi, 2010). Laibson (1997) argues that consumer have self-control problems discounting present consumption over future consumption, describing it as “hyperbolic discounting.” This provides an explanation for the first anomaly – increased spending. A potential explanation for the second anomaly – increased debt, can be explained by the “bounded rationality” model of Gabaix and Laibson (2000). It is conceivable that the contract terms and conditions are rather complex and over time consumers forget them and use the credit card for present consumption. Ex-post consumers could even justify their mistake as financially insignificant or easily fixable since they receive several balance transfer offers on a weekly basis. 8 Therefore, there are several explanations for increased spending, but we can point to “bounded rationality” if we observe an increase in overall debt as a result of the program. 3. Data We use a unique, proprietary data set from a large financial institution that issues credit cards nationally. Account level administrative data from a financial institution has a number of advantages over consumer survey data. Relative to traditional household data sets such as the Survey of Consumer Finances, our sample is large with little measurement error. Also, because each account is observed over many months, it is possible to study high-frequency dynamics. However, using credit card data does entail a number of limitations. The main unit of analysis is a credit card account, not an individual (who can hold multiple accounts). Unfortunately, we do not observe total spending (i.e. spending via cash and checks). Our data set contains a representative sample of about 12,000 credit card accounts from June 2000 to June 2002 with monthly observations. For all card accounts, the data on the credit card transactions include monthly data from account statements, including spending, repayment, balance, debt, APR and credit limit. In addition to monthly data on credit card use, the data set also contains credit card bureau data about the other credit cards held by each account holder, in particular the number of other cards and their combined balances. Unfortunately, credit bureaus do not separately record credit card debt, spending and payments – they record only balances. The credit card issuer obtained these data from the credit bureaus quarterly. Finally, there is limited demographic data – age and marital status of the cardholders. Account holders are assumed to be married if there is a spouse also listed on the account. We provide summary statistics of all cardholders in Table 1. 9 For approximately half of the data set (6,600 accounts), we also have information regarding participation in a cash-back bonus including how much cash back is accrued and redeemed. The cash-back program begins in month 12 of our sample. The average value redeemed is around $10 and the average redemption per account is around $25. Ninety percent of cardholders redeem their cash-back rewards and 85 percent of the value is redeemed. In Table 2, we provide summary statistics for the control and treated groups for months 3-5. We also looked at the summary statistics for these two groups at other time period and we do not observe any systematic patters to suggest any selection of any particular variables. For instance, during these three months, some variables are statistically similar for these two groups such as spending, internal behavior and FICO scores, and some demographic characteristics. However, some variables such as debt on card, credit line and total overall balance are statistically different. As mentioned before, some cardholders are also part of the APR reduction program. We have also looked at the treated group without these individuals (not reported). When these cardholders are excluded, spending, debt, and credit line decrease to levels below the control group. This would suggest that the financial institution does not systematically select a group of customers for the reward program. Based on our conversations with the institution, cardholders are not selected on a given criterion to be included in the program. Moreover, large financial institutions are reluctant to prescreen cardholders for such programs due to the potential regulatory scrutiny regarding discrimination based on demographic characteristics. The additional cost to make sure such selection is legitimate is significant for issuers. Finally, if the institution had the goal of maximizing revenue, it would have given cash back to all cardholders not using their cards. Giving rewards to cardholders already using their cards with low probability of increased usage is costly. 10 In Table 3, we compare the aggregate monthly summary statistics for the treatment group during the preceding month before the cash-back program and during the first quarter after the program starts. Note that the average purchase amount increases while the average payment amount decreases. Because all consumers are lumped together across time, these summary statistics may not indicate the underlying changes in cardholder behavior. Additionally, we have information about an interest rate reduction program that is offered to certain individuals. The month in which cardholders receive reductions in APR is evenly distributed during our sample period. Over half of cardholders have promotional APRs when our panel ends. The average APR reduction is 10%. Interestingly, all cardholders that receive APR reductions are also part of the cash-back program. 4. Empirical Strategy Our empirical strategy is to quantify consumer responses to financial incentives such as cash-back and interest rate reduction programs. Our dataset allows us to study two different programs that the financial institution uses to increase card usage. In addition to studying the impact of card spending and additional debt accrued, we are able to study the impact on the cardholder’s overall balances which include additional spending and changes in debt. 4.1 Cash-back rewards We use an event window methodology to study the impact of cash-back incentives (Agarwal, Liu, and Souleles (2007)). The general structure of our OLS regressions is: Yit = f (cashbackit, account controls, demographic controls, portfolio controls) (1) Our dependent variable, Yit, represents monthly spending, change in debt, or change in overall balance on all credit cards. Our main explanatory variable, cashbackit, is an indicator variable 11 for cash back. It is one if the cardholder receives cash back for that month’s purchases and zero otherwise. We compare one month before the cash back is offered to three, six, or nine months after the cash-back program has started.7 Our account control variables are account age, realized APR, credit line and the bank’s behavioral score. Our demographic control variables are marital status, gender, income, age and age squared. Our portfolio controls are from credit bureau data and include the individual’s FICO score, sum of all credit card lines, total balance on all credit cards, number of other credit cards, and number of other credit cards with balances. All regressions are run with individual fixed effects and clustered standard errors. The expected response to a cash-back reward is to increase spending on the card. We would expect spending to increase for two reasons. First, cash back may generate additional overall spending. Second, cardholders may substitute this card for purchases made with cash, check, debit cards or other credit cards. While card issuers earn revenue from merchants indirectly through interchange fees that are paid to them by the merchants’ banks, the bulk of issuer income is earned from finance charges that accrue when cardholders carry debt. Similar to spending, cardholders may increase their overall debt or substitute credit card borrowing from one card to another. To study spending and debt substitution across credit cards, we estimate the impact of cash back on overall credit card balances. Unlike bank level data, credit bureau data combines spending and debt into one variable called total credit card balances. Thus, overall balances can increase because of increased spending and/or additional debt. Furthermore, we are cautious about our results because credit bureau data is only available for our cardholders every quarter. 7 In unreported specifications, we also tried 3 and 6 months before the program. The results do not show any spending or debt responses before the rebate and so we dropped those specification to conserve data. Additionally, in unreported specification, we also ran the same regressions for the control and treated samples together and we do not find any spending response for the control sample. 12 We are also able to compare the impact of cash-back rewards on several different subgroups. First, we compare those cardholders that carry debt and those that do not. Those that do not carry debt can be further divided into two groups—those that use their cards and payoff their balances in full every month and those that do not use their cards in the previous three months before enrollment in the program. Second, we consider different demographic characteristics such as gender and marital status along with differences in credit limits. Finally, we compare different levels of utilization of credit limits. We also study the impact of a ten percent reduction in the APR on card spending, change in card debt, and overall credit card balances. We use the same controls as in the cash-back regression along with the same event windows. Instead of the cashback indicator variable, we use an APR reduction indicator variable. The general structure of our regression becomes: Yit = f (APRit, account controls, demographic controls, portfolio controls) 5. (2) Results In this section, we report our regression results about the impact of the cash-back program and the APR reduction program. All regressions are run with individual fixed effects. 5.1 Cash Back In Table 4, we report the coefficient on the cash indicator variable for the whole treatment group. Our results indicate that spending increases significantly for all cardholders— the average consumer increases her spending by over $68 dollars per month during the first quarter of being in the cash back program. The average per month spending continues to increase at $76 per month during the first 9 months after the program is introduced. These results are both statistically and economically significant. 13 We also examine the effect of the rewards program on consumers’ incremental debt accumulation. We use the change in cardholder debt as the dependent variable to study the impact of the cash-back program. On average, a consumer increases her debt by $115 per month during the first quarter of the cash-back program. Our results confirm that cardholders not only increase spending but also their debt. The increase in monthly spending and change in debt remains relatively constant and continues during the first nine months after the beginning of the cash-back program. We show the complete regression results in appendix tables 1A and 1B. The greater increase in debt compared to spending suggests that payments drop not only for purchases due to the cash-back reward but also on spending that is not related to the cash-back rewards. Specifically, we find a average monthly reduction in payments within the first quarter of $38, suggesting that all the marginal increase in spending due to the cash-back reward is converted into debt and a part of the cardholder’s monthly baseline spending is also converted into debt. To study the overall impact on the cardholder’s total credit card spending and debt, we study the impact of cash back on the total credit card balance as reported by the credit bureaus. If the sum of spending and change in card debt is greater than the impact on overall card balances, we conclude that the cardholder has substituted some spending and debt from other cards to the cash back card. The change in overall balance only increases by an average of $40 per month during the first quarter and increases to an average of $76 per month during the first nine months. These results suggest that cardholders have not only substituted spending but also debt since their overall credit card balances are lower than both the increases in spending and debt. However, these estimates are not statistically significant. As mentioned before, the credit 14 bureau data is only available at quarterly intervals making our measurement somewhat imprecise. Cardholders differ in how they use their credit cards. Cardholders may use their cards primarily as a payment instrument by paying off their balances in full every month or make purchases on credit that they payoff over a longer time horizon. We would expect these different groups to respond to the cash-back incentive program in different ways. We separate cardholders into those that carry debt, commonly referred to as revolvers, and those that do not. In Table 5, we study the impact of the cash-back program on cardholders that carry debt from month-to-month with those that have zero balances. In the first quarter of the program, cardholders that do not carry debt increase their spending by $138 per month versus $47 per month for those cardholders that carry debt during the first quarter. During the first nine months, those cardholders without debt continue to spend more than an average $99 per month and those that carry debt increase their spending by an average of $67 per month. All of these estimates are statistically and economically significant. The effect of cash back on change in debt also differs across cardholders (Table 6). Those carrying debt, increase their debt by an average of $134 per month during the first quarter and by an average of $142 per month during the first nine months after the program starts. Those that do not carry debt increase their debt by an average of $114 per month during the first quarter and by an average of $211 per month during the first nine months. Those cardholders that do not carry debt substitute spending and debt accumulation on this card from other cards (Table 7). Those that do not carry debt do not increase their overall card balance as a result of participating in the cash back program. 15 To further investigate the impact of no debt cardholders of the cash-back program, we separate the “no debt” group into convenience users and inactive cardholders for the three months prior to being in the program. Note that in both cases, cardholders would be categorized as zero debt. About half of cardholders in the treatment group were inactive during the three months before being enrolled in the cash-back program. In Tables 8, 9, and 10, we report our results for convenience users and inactive cardholders for the previous three months. The cash back impact on spending is not statistically significant for convenience users (Table 8). However, the cash back impact on spending of inactive cardholders prior to the cash-back program is statistically and economically significant. The average per month spending increases by $220 during the first quarter and only decreases to $180 on average per month during the first nine months of being in the program. The increase in debt for inactive cardholders prior to the cash-back program is statistically and economically significant as well (Table 9). The average monthly change in debt during the first quarter is $167 and the average monthly change in debt rises to $196 during the first nine months. Furthermore, the impact of cash back on overall balances suggests that inactive cardholders substituted spending and debt accumulation from other cards (Table 10). We also find evidence that those that inactive users substituted credit card balances including spending and change in debt from other cards as they increased spending and debt on their cash-back card. In addition, we include some analysis based on demographic characteristics to study the impact of the cash-back program. In Tables 11, we report our results on certain demographic characteristics. Single cardholders increase their spending by $55 on average per month during the first quarter whereas married cardholders increase their spending by $95. Single cardholder debt increases by $95 on average per month during the first quarter and rises to $164 during the 16 first nine months after the program is introduced whereas married cardholder debt starts at $155 and rises to $262, respectively. The impact of the cash back program on the quarterly change in the cardholder’s overall balance is not statistically significant. The impact of cash back on spending and debt is similar for males and females with males being a bit more stable in their increase in spending and debt accumulation. We divide the treated sample into those that have income below $40,000 and those above $40,000 based on what the cardholder reported at the time of application. We find that those with higher income tend to spend more and accumulate more debt in response to being in the cash-back program. In Table 12, we report results from considering different levels of credit constraints. We separate our treated group into three different categories of credit limits—below $6,000, between $6,000 and $12,000, and above $12,000. Those with higher credit limits tend to spend more and accumulate more debt in response to the cash-back program. We also divide the sample by two levels of credit line utilization—below 50% and above 50%. We find that cardholders that utilize their credit lines 50% or greater spend more and borrow more especially after being in the cash-back program for nine months than cardholders who are less credit constrained. 5.2 Interest rate reduction We find that on average, consumers increase their spending by an average of $1098 per month during the first quarter following an APR reduction (Table 13). However, this sharply drops off to an average of $579 per month during the first nine months after the cardholder is in the APR reduction program. This attenuation suggests that many cardholders transferred balances or spending from other credit cards to this one at the beginning of the promotion. 17 Additionally, we find that the change in debt on average increases on average by $1059 during the first quarter but falls to $356 during the first nine months suggesting that cardholders are substituting debt from higher interest cards to this one with a lower interest rate. The coefficients of the indicator variable of the APR reduction program on the overall credit card balance suggests that there is cardholders are exchanging debt from other cards to this card. Finally, 24 percent of cardholders that did not use their cards three months prior to the APR program used their cards to make at least $50 of purchases during the first month of being enrolled in the program. 6. Conclusion Using statement level data from a large U.S. financial institution, we explored the impact of cash-back rewards on credit card spending, debt accumulation, and overall credit bureau balances. Our analysis suggests that cash-back rewards positively and significantly affects spending and debt accumulation. However, overall spending and debt accumulation measured by total credit card balances at the credit bureau remain constant or increase slightly suggesting that cardholders substitute spending and debt from other credit cards. Furthermore, the relatively small average cardholder redemption of $25 per cardholder makes such a program a cost effective tool to increase bank revenue from increased spending and borrowing by cardholders. Cash-back rewards are an effective tool to spur spending and debt accumulation by cardholders that hold the institution’s credit cards but do not use them. This group makes up about half of all the cardholders that receive the cash back offer. Furthermore, the cash-back program provides sufficient incentives to 11 percent of inactive cardholders to use their cards. The response to cash-back rewards by this group is an increase in average spending of $220 per 18 month during the first quarter and an increase in debt accumulation of close to $167 per month during the first quarter. The cash-back program is generating the greatest revenue from those that were not using this card prior to the reward. Our paper sheds light on various aspects of the consumption and payment literature. Our results support that financial incentives need not be large to generate significant shifts in consumer behavior. While not the main focus of our paper, we are unable to rule out time inconsistency issues arising from payment substitution and increases in incremental cardholder spending. A more complete view of the cardholders debt portfolio and monthly expenditures would be necessary to explore this issue further. Finally, we consider an alternative view as to why financial institutions issue rewards. Much of the theoretical payment card literature suggests that financial incentives may be necessary to gain adoption of a payment instrument. Others have suggested that credit card rewards are a form of surplus extraction. Our analysis suggests that in an extremely competitive credit card issuing market, rewards are another tool along with lower interest rates to steal customers from competitors. 19 References Aaronson, D., S. Agarwal, and E. French (2007), “Spending and debt response to minimum wage hikes,” Federal Reserve Bank of Chicago Working Paper No. 2007-23. Agarwal, S., S. Chomsisengphet, and C. Liu (2010), “The importance of adverse selection in the credit card market: Evidence from randomized trials of credit card solicitations,” Journal of Money, Credit and Banking, 42 (4), 743-754. Agarwal, S., S. Chomsisengphet, C. Liu, and N. Souleles (2006). “Do Consumers Choose the Right Credit Contracts?” Federal Reserve Bank of Chicago Working Paper 2006-11. Agarwal, S., C. Liu, and N. Souleles (2007), “The reaction of consumption and debt to tax rebates: Evidence from the consumer credit data,” Journal of Political Economy, 115 (6), 9861019. Ausubel, L. (1991), “The failure of competition in the credit card market,” American Economic Review, 81 (1), 50-81. Baxter, W. F. (1983), “Bank interchange of transactional paper: Legal and economic perspectives,” Journal of Law and Economics, 26 (3), October, 541-588. Bolt, W., S. Carbó Valverde, S. Chakravorti, S. Gorjón, and F. Rodríguez Fernández (2010), “What is the role of public authorities in retail payment systems?,” Federal Reserve Bank of Chicago Fed Letter, 280a, November. Bolt, W., and S. Chakravorti (2008), “Consumer choice and merchant acceptance of payment media,” Federal Reserve Bank of Chicago Working Paper, WP-2008-11. Borzekowski, R., E. K. Kiser and S. Ahmed (2008), “Consumers’ use of debit cards: Patterns, preferences, and price response,” Journal of Money, Credit and Banking, 40 (1), 149-172. Browning, M., and M. D. Collado (2001), “The response of expenditures to anticipated income changes: Panel data estimates,” American Economic Review, 91 (3), 681-92. Chakravorti, S. (2007), “Linkages between Consumer Payments and Credit,” in Sumit Agarwal and Brent Ambrose (eds), Household Credit Usage: Personal Debt and Mortgages, New York: Palgrave Macmillan, 161-174. Chakravorti, S. (2010), “Externalities in payment card networks: theory and evidence,” Review of Network Economics, 9 (2), Article 3. Chakravorti, S. and W. R. Emmons (2003), “Who pays for credit cards?” Journal of Consumer Affairs, 37 (2), 208-230. 20 Chakravorti, S., and T. To (2007), “A theory of credit cards,” International Journal of Industrial Organization, 25 (3), 583–595. Ching, A. T. and F. Hayashi (2010), “Payment card rewards programs and consumer payment choice,” Journal of Banking and Finance, 34 (8), 1773-1787. Gabaix, X., and D. Laibson (2000), “A boundedly rational decision algorithm” American Economic Review Papers and Proceedings, 90 (2), 433-8. Gross, D., and N. Souleles (2002), “Do liquidity constraints and interest rates matter for consumer behavior? Evidence from credit card data,” Quarterly Journal of Economics, 117 (1), 149-185. Heidhues, P. and B. Köszegi (2010), “Exploiting naivete about self-control in the credit market,” American Economic Review, 100 (5), 2279-2303. Hsieh, C. (2003), “Do Consumers react to anticipated income changes? Evidence from the Alaska Permanent Fund,” American Economic Review, 93 (1), 397-405. Jacob, K., C. Jankowski, and A. Lunn (2009), “Payments pricing: Who bears the cost?—A conference summary,” Federal Reserve Bank of Chicago Fed Letter, 266a. Johnson, David, Jonathan Parker, and Nicholas Souleles, (2006), “Consumption and tax cuts: Evidence from the randomized income tax rebates of 2001,” American Economic Review, 96 (5), 1589-1610. Kennickell, A. B., T. L. Mach, B. Bucks, and K. B. Moore (2009), “Changes in U.S. family finances from 2004 to 2007: Evidence from the Survey of Consumer Finances,” Federal Reserve Bulletin, 95 (February), A1-A55 Laibson, D. (1997), “Golden eggs and hyperbolic discounting,” Quarterly Journal of Economics, 62 (May), 443-77. Parker, J. (1999), “The reaction of household consumption to predictable changes in social security taxes,” American Economic Review, 89 (4), 959-973. Reserve Bank of Australia (2008), Reform of Australia’s Payments System: Preliminary Conclusions of the 2007/08 Review, April. Rochet, J.-C., and J. Tirole (2002), “Cooperation among competitors: Some economics of payment card associations,” RAND Journal of Economics, 33 (4), 549-570. Rochet, J.-C., and J. Wright (2010), “Credit card interchange fees,” Journal of Banking and Finance, 34 (8), 1788-1797. 21 Schuh, S., O. Shy, J. Stavins (2010), “Who gains and who loses from credit card payments? Theory and calibrations,” Federal Reserve Bank of Boston Public Policy Discussion Papers Series, 10-3. Shapiro, M. D., and J. Slemrod (2003a), “Consumer response to tax rebates,” American Economic Review, 93 (1), 381-396. Shapiro, M. D., and J. Slemrod (2003b), “Did the 2001 tax rebate stimulate spending? Evidence from taxpayer surveys,” in Tax Policy and the Economy, ed. Poterba, J., Cambridge: MIT Press. Souleles, N. S. (1999), “The response of household consumption to income tax refunds,” American Economic Review, 89 (4), 947-958. Souleles, N. S. (2000), “College tuition and household savings and consumption,” Journal of Public Economics, 77 (2), 185-207. Souleles, N. S. (2002), “Consumer response to the Reagan tax cuts,” Journal of Public Economics, 85 (1), 99-120. Stephens, M. (2003), “3rd of the month: Do social security recipients smooth consumption between checks?” American Economic Review, 93 (1), 406-422. Wilcox, D. W. (1990), “Income tax refunds and the timing of consumption expenditure,” Federal Reserve Board of Governors Working Paper 106. Wilcox, D. W. (1989). "Social security benefits, consumption expenditure, and the life cycle hypothesis," Journal of Political Economy, 97 (2), 288-304. Zinman, J. (2009). “Debit or credit?” Journal of Banking & Finance 33 (2), 358-366. 22 TABLE 1 SUMMARY STATISTICS FOR ALL OBSERVATIONS (N=261435) Mean Standard Deviation Account Characteristics Spending Debt on Card Internal Behavioral Score Account Age (years) Credit Line Realized APR Did Not Use (1=Not spending before the program) Revolver (1=Carrying debt before the program) 220 2610 712 4 8509 15.37% 790 3045 30 3 3280 6.70% 50.07% 50.00% 70.72% 45.51% Credit Characteristics FICO Score Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards 731 10007 3 5 46 13603 3 4 Demographics Income Age Gender (male=1, female=0) Married (Spouse Listed=1, No Spouse Listed=0) 58089 48 52.35% 33.49% 93789 13 49.95% 47.20% Notes: The data come from the monthly billing statement of credit card accounts, and attached credit bureaus. All values are averaged over the sample period (June 2000 to June 2002). 23 TABLE 2 SUMMARY STATISTICS BY TREATMENT AND CONTROL GROUP Months 3-5 Control (N=14209) Mean Standard Deviation Treatment (N=19430) Mean Standard Deviation Account Characteristics Spending Debt on Card Internal Behavioral Score Account Age Credit Line Did Not Use (1=Not spending before the program) Revolver (1=Carrying debt before the program) 244 2139 718 4 7494 707 2810 38 3 3205 257 3541 716 3 8569 1075 3085 24 3 2831 35.3% 47.8% 61.1% 48.8% 63.0% 48.3% 76.4% 42.4% Credit Characteristics FICO Score Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards 724 8625 3 4 56 12766 3 4 737 9916 3 5 33 12766 3 4 Demographics Income Age Gender (male=1, female=0) Married (Spouse Listed=1, No Spouse Listed=0) 57755 46 71005 13 58351 48 103033 13 52.28% 49.95% 52.68% 49.93% 32.99% 47.02% 33.83% 47.31% Notes: The data come from the monthly billing statement of credit card accounts, and attached credit bureaus. All values are averaged over the sample period (September to November 2000). 24 TABLE 3 SUMMARY STATISTICS BEFORE AND AFTER TREATMENT IN TREATMENT GROUP Months 11-14 Before (N=7037) After (N=12504) Mean Standard Deviation Mean Standard Deviation Account Characteristics Spending Debt on Card Internal Behavioral Score Account Age Credit Line 159 2716 703 4 9217 652 3048 19 3 2897 203 2772 702 4 9268 915 3099 18 3 2949 Credit Characteristics FICO Score Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards 738 10553 3 5 33 13692 3 4 739 11255 3 6 32 13819 3 4 Demographics Income Age Gender (male=1, female=0) 59104 48 100510 13 57819 49 104200 13 53.17% 49.91% 52.55% 49.94% 34.56% 47.56% 33.24% 47.11% Married (Spouse Listed=1, No Spouse Listed=0) Notes: The data come from the monthly billing statement of credit card accounts, and attached credit bureaus. All values are averaged over the sample period (May to August 2001). 25 TABLE 4 Effects of Bonus Program within Treatment Group Spending 3 Months (N=24308) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 68.41 12.73 5.37 6 Months (N=40633) Coeffic Standard T-Stat Error 80.43 10.76 7.47 9 Months (N=55989) Coeffic Standard T-Stat Error 76.14 10.18 7.48 Change in Debt 3 Months (N=21675) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 114.85 25.05 4.58 6 Months (N=35764) Coeffic Standard T-Stat Error 177.43 20.77 8.54 197.08 19.22 10.26 Change in Total Balance Across Cards 3 Months (N=24308) 6 Months (N=40633) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 40.26 56.91 0.71 9 Months (N=48487) Coeffic Standard T-Stat Error Coeffic Standard T-Stat Error 56.25 59.96 0.94 9 Months (N=55989) Coeffic Standard T-Stat Error 66.60 60.29 1.10 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau (except in equations estimated effect on change in total balance across cards), number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. TABLE 5 Effects of Bonus Program on Spending if Cardholder Carries Debt Panelists that DO NOT have Debt Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=5286) Coeffic Standard T-Stat Error 6 Months (N=8142) Coeffic Standard T-Stat Error 138.16 119.42 28.02 4.93 23.01 5.19 9 Months (N=11016) Coeffic Standard T-Stat Error 98.73 22.28 4.43 Panelists that DO have Debt Before the Program Starts 3 Months (N=19022) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 47.47 14.21 3.34 6 Months (N=32491) Coeffic Standard T-Stat Error 9 Months (N=44973) Coeffic Standard T-Stat Error 69.11 66.69 12.28 5.63 11.59 5.76 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 27 TABLE 6 Effects of Bonus Program on Change in Debt if Cardholder Carries Debt Panelists that DO NOT have Debt Before the Program Starts 3 Months (N=4002) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 133.92 28.80 4.65 6 Months (N=5868) Coeffic Standard T-Stat Error 9 Months (N=7521) Coeffic Standard T-Stat Error 137.37 142.61 23.61 5.82 21.48 6.64 Panelists that DO have Debt Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=17672) Coeffic Standard T-Stat Error 6 Months (N=29896) Coeffic Standard T-Stat Error 9 Months (N=40966) Coeffic Standard T-Stat Error 114.04 190.61 210.89 30.19 3.78 24.87 7.66 22.97 9.18 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 28 TABLE 7 Effects of Bonus Program on Quarterly Change in Total Balance if Cardholder Carries Debt Panelists that DO NOT have Debt Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=5286) Coeffic Standard T-Stat Error 6 Months (N=8142) Coeffic Standard T-Stat Error 9 Months (N=11016) Coeffic Standard T-Stat Error 293.49 122.26 371.35 129.36 2.87 466.37 136.03 3.43 2.40 Panelists that DO have Debt Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=19022) Coeffic Standard T-Stat Error 6 Months (N=32491) Coeffic Standard T-Stat Error -12.35 -20.19 64.71 -0.19 66.08 -0.31 9 Months (N=44973) Coeffic Standard T-Stat Error -18.99 68.31 -0.28 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 29 TABLE 8 Effects of Bonus Program on Spending by Convenience and Inactive Users Convenience Users Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=2113) Coeffic Standard T-Stat Error 6 Months (N=3288) Coeffic Standard T-Stat Error 9 Months (N=4417) Coeffic Standard T-Stat Error 26.86 -14.77 -28.83 38.93 44.41 0.60 40.09 -.37 -.74 Inactive Users Before the Program Starts 3 Months (N=3095) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 219.61 38.16 5.75 6 Months (N=4692) Coeffic Standard T-Stat Error 9 Months (N=6248) Coeffic Standard T-Stat Error 205.97 180.24 27.67 7.44 26.73 6.74 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 30 TABLE 9 Effects of Bonus Program on Change in Debt by Convenience and Inactive Users Convenience Users Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=1624) Coeffic Standard T-Stat Error 6 Months (N=2445) Coeffic Standard T-Stat Error 95.32 81.13 34.65 2.75 31.25 2.60 9 Months (N=3129) Coeffic Standard T-Stat Error 80.19 23.97 3.35 Inactive Users Before the Program Starts 3 Months (N=2318) Coeffic Standard T-Stat Error Cashback Indicator (Receiving Cashback=1) 166.66 45.63 3.65 6 Months (N=3312) Coeffic Standard T-Stat Error 9 Months (N=4193) Coeffic Standard T-Stat Error 186.38 196.15 34.74 5.36 33.62 5.83 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 31 TABLE 10 Effects of Bonus Program on Quarterly Change in Total Balance by Convenience and Inactive Users Convenience Users Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=2113) Coeffic Standard T-Stat Error 6 Months (N=3288) Coeffic Standard T-Stat Error 9 Months (N=4417) Coeffic Standard T-Stat Error 244.38 183.82 1.33 310.55 189.16 1.64 239.83 186.43 1.29 Inactive Users Before the Program Starts Cashback Indicator (Receiving Cashback=1) 3 Months (N=3095) Coeffic Standard T-Stat Error 6 Months (N=4692) Coeffic Standard T-Stat Error 9 Months (N=6248) Coeffic Standard T-Stat Error 276.80 368.05 182.38 2.02 594.36 197.72 3.01 159.70 1.73 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 32 TABLE 11 Only Coefficients for Cashback Indicator Variable are Reported Single Married Male Female Income<$40k Income>$40k Spending 3 6 9 55.34*** 65.06*** 59.47*** 95.06*** 110.80*** 108.05*** 47.01** 78.38*** 80.82*** 70.50*** 83.85*** 76.17*** 47.06** 56.28*** 57.74*** 73.94*** 87.03*** 81.13*** Change in Debt 3 6 9 95.05*** 150.54*** 163.61*** 155.03*** 230.87*** 261.83*** 71.28 137.58*** 170.11*** 83.85 158.41*** 177.15*** 99.96** 137.05*** 152.96*** 119.31*** 187.82*** 208.58*** 3 6 9 56.08 90.01 113.80 7.73 -3.86 -19.04 -12.79 -74.66 -22.53 169.50 248.72** 235.03 30.75 47.50 40.82 36.31 54.46 67.38 Quarterly Change in Total Balance Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau (except in equations estimated effect on change in total balance across cards), number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 33 TABLE 12 Only Coefficients for Cashback Indicator Variable are Reported Credit Limit Less than $6k Credit Limit $6-12k Credit Limit > $12k Percent Utilization <50% Percent Utilization >50% Spending 3 6 9 30.25 32.12** 23.68 69.33*** 78.05*** 77.38*** 91.72** 105.06*** 96.12*** 17.41 26.23*** 24.09*** 36.27* 94.46*** 124.64*** Change in Debt 3 6 9 68.06 94.65*** 108.78*** 108.77*** 160.66*** 166.57*** 184.97** 256.74*** 290.52*** 69.86 109.60*** 119.97*** 4.78** 129.92*** 188.35*** 3 6 9 -176.00 -226.50 -82.03 45.11 34.40 21.91 189.67 270.73* 297.52** 61.24 58.29 26.98 -68.48 -16.41 23.77 Quarterly Change in Total Balance Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau (except in equations estimated effect on change in total balance across cards), number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 34 TABLE 13 Effects of APR Reduction within Treatment Group Spending 3 Months (N=4633) Coeffic Standard T-Stat Error APR Reduction Indicator (Receiving APR Reduction=1) 1097.79 84.80 12.95 6 Months (N=7721) Coeffic Standard T-Stat Error 9 Months (N=10298) Coeffic Standard T-Stat Error 822.98 578.63 55.40 14.86 44.30 13.06 Change in Debt 3 Months (N=4022) Coeffic Standard T-Stat Error APR Reduction Indicator (Receiving APR Reduction=1) 6 Months (N=6687) Coeffic Standard T-Stat Error 1058.65 117.51 9.01 3 Months (N=4633) 660.48 636.15 432.67 7.36 Change in Total Balance 6 Months (N=7721) Coeffic Standard T-Stat Error APR Reduction Indicator (Receiving APR Reduction=1) 89.70 Coeffic Standard T-Stat Error 1.47 604.08 288.07 2.10 9 Months (N=8932) Coeffic Standard T-Stat Error 355.94 69.39 5.13 9 Months (N=10298) Coeffic Standard T-Stat Error 586.55 263.30 2.23 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the APR reduction indicator variable in equation (1) for the three regressions of spending, change in debt, and change in credit bureau balances for 3, 6, and 9 months respectively. All values are in current dollars (2000-2002). Each regression also includes a full set of controls – the quarterly change in FICO and internal behavioral scores, account age, the APR that the consumer pays, credit limit, total balance from the credit bureau (except in equations estimated effect on change in total balance across cards), number of other credit lines from the credit bureau, number of other cards with debt from the credit bureau, an indicator for married, age and age squared. The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 35 APPENDIX VARIABLE INDEX Last Balance Measure of Debt Purchase Amount Sum of Purchases made that month Payment Amount Sum of repayments made that month FICO Score FICO score from Credit Bureau Internal Behavioral Score Internal Score from bank Account Age Number of years account open APR Annual Percentage Rate Realized APR APR that the consumer faces (reflects promotional APR from bank) Credit Line Credit Available on the account Total Balance Across Cards Sum of Balances reported by the Credit Bureau Total Number of Cards with Debt Total Number of Cards with Debt reported by the Credit Bureau Total Number of Credit Lines Total Number of card held by individual reported by the Credit Bureau Income Individual Income reported by the bank Age Age of individual Age^2 Age of individual squared Gender =1 for male, =0 for female, missing if unknown Marital Status =1 if a spouse is listed on the count, =0 if spouse not listed Cash Back Indicator =1 if receiving cash back promotion from the bank, =0 if not FICO Change that Quarter Behavioral Score at t - Behavioral Score at t-1 Internal Score Change that Quarter Internal Score at t - Internal Score at t-1 Appendix - TABLE 1A Effects of Bonus Program on Spending and Change in Debt in Treatment Group 1 Month Before the Program Starts, 3 Months After Spending (N=24308) Coefficient Variable Cashback Dummy 68.41 -1.69 -0.58 24.74 -42.18 0.03 0.00 15.65 -4.46 116.73 -1.26 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 12.73 0.47 0.44 31.69 5.94 0.02 0.00 15.72 18.09 100.63 0.92 T-Stat 5.37 -3.60 -1.32 0.78 -7.10 1.74 -0.53 1.00 -0.25 1.16 -1.36 Change in Debt (N=21674) Variable Coefficient Cashback Dummy 114.85 -13.26 -0.11 104.02 -43.94 0.21 0.00 18.04 7.48 372.44 -3.53 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 25.05 1.23 0.78 51.97 8.65 0.06 0.01 25.88 27.17 168.57 1.58 T-Stat 4.58 -10.81 -0.14 2.00 -5.08 3.39 -0.12 0.70 0.28 2.21 -2.24 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for change in debt for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 37 Appendix - TABLE 1B Effects of Bonus Program on Quarterly Change in Total Balance in Treatment Group 1 Month Before the Program Starts, 3 Months After Quarterly Change in Total Balance (N=24308) Variable Coefficient Cashback Dummy 40.26 5.52 60.56 -16.16 -18.81 -0.04 1239.07 258.50 -144.76 1.33 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 56.91 3.63 4.25 205.74 25.55 0.09 268.33 131.63 521.02 5.27 T-Stat 0.71 1.52 14.26 -0.08 -0.74 -0.46 4.62 1.96 -0.28 0.25 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for change in debt for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 38 Appendix - TABLE 2 Effects of Bonus Program on Spending for those that Do Not have Debt Prior to the Program 1 Month Before the Program Starts, 3 Months After Spending (N=5286) Variable Coefficient Cashback Dummy Standard Error T-Stat (Receiving Cashback=1, Not in Program=0) 138.16 28.02 4.93 Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 -2.19 0.10 -69.87 -49.85 0.01 0.00 32.16 -30.77 17.70 -0.17 1.07 1.11 95.58 7.98 0.02 0.01 25.79 37.50 188.52 1.70 -2.06 0.09 -0.73 -6.25 0.61 -0.38 1.25 -0.82 0.09 -0.10 Effects of Bonus Program on Spending for those that DO have debt prior to the program 1 Month Before the Program Starts, 3 Months After Spending (N=19022) Variable Cashback Dummy (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Coefficient Standard Error 47.47 -1.48 -0.76 47.37 -38.25 0.03 0.00 9.50 4.19 140.81 -1.53 14.21 0.52 0.45 32.58 7.96 0.02 0.01 19.32 20.83 117.79 1.08 T-Stat 3.34 -2.82 -1.69 1.45 -4.81 1.72 -0.46 0.49 0.20 1.20 -1.42 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 39 Appendix - TABLE 3 Effects of Bonus Program on Change in Debt for those that Do Not have Debt Prior to the Program 1 Month Before the Program Starts, 3 Months After Variable Cashback Dummy Change in Debt (N=4002) Coefficient Standard Error T-Stat (Receiving Cashback=1, Not in Program=0) 133.92 28.80 4.65 Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 -6.58 -0.36 2.23 -47.16 0.01 0.01 -2.78 21.51 215.12 -1.65 1.90 1.72 83.98 9.16 0.09 0.01 19.30 21.45 242.56 2.06 -3.46 -0.21 0.03 -5.15 0.08 0.78 -0.14 1.00 0.89 -0.80 Effects of Bonus Program on Change in Debt for those that DO have debt prior to the program 1 Month Before the Program Starts, 3 Months After Change in Debt (N=17672) Variable Coefficient Standard Error T-Stat Cashback Dummy 114.04 30.19 3.78 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter -14.62 1.41 -10.37 Change in FICO Score from Previous Quarter 0.02 0.87 0.02 Account Age 130.84 59.01 2.22 Realized APR -42.37 11.71 -3.62 Credit Line 0.23 0.07 3.50 Total Balance on All Credit Cards 0.00 0.01 -0.40 Total Credit Cards with Debt 26.12 33.68 0.78 Total Number of Credit Cards 3.01 34.51 0.09 Age 400.05 199.36 2.01 Age^2 -3.82 1.86 -2.05 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 40 Appendix - TABLE 4 Effects of Bonus Program on Quarterly Change in Total Balance for those that Do Not have Debt Prior to the Program 1 Month Before the Program Starts, 3 Months After Variable Cashback Dummy Change in Total Balance (N=5286) Coefficient 293.49 20.32 63.74 746.88 49.79 0.12 1797.16 52.46 1734.26 -14.41 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 122.26 8.63 9.94 623.35 42.98 0.15 607.02 270.52 1396.63 13.79 T-Stat 2.40 2.36 6.41 1.20 1.16 0.81 2.96 0.19 1.24 -1.05 Effects of Bonus Program on Change in Debt for those that DO have debt prior to the program 1 Month Before the Program Starts, 3 Months After Variable Cashback Dummy Change in Total Balance (N=19022) Coefficient Standard Error -12.35 2.62 60.41 -149.95 -49.82 -0.07 951.42 376.15 -706.11 6.42 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 64.71 4.00 4.65 209.92 31.75 0.11 119.23 103.49 540.85 5.53 T-Stat -0.19 0.66 13.00 -0.71 -1.57 -0.63 7.98 3.63 -1.31 1.16 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 41 Appendix - TABLE 5 Effects of Bonus Program on Spending for those that Do Not have Debt and Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Spending (N=2113) Variable Coefficient Cashback Dummy 26.86 -1.94 -0.65 -49.09 -30.80 0.01 -0.01 97.62 -156.33 172.60 -1.70 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 44.41 1.30 1.21 115.12 14.61 0.03 0.01 54.32 98.99 174.06 1.61 T-Stat 0.60 -1.49 -0.54 -0.43 -2.11 0.34 -0.64 1.80 -1.58 0.99 -1.05 Effects of Bonus Program on Spending for those that Do Not have Debt and DO Not Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Spending (N=3095) Variable Coefficient Cashback Dummy 219.61 -2.75 0.65 -76.57 -56.47 0.01 0.00 4.99 24.96 -153.03 1.66 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error T-Stat 38.16 1.70 1.64 145.01 9.44 0.03 0.01 28.28 25.38 293.05 2.64 5.75 -1.62 0.40 -0.53 -5.98 0.39 0.05 0.18 0.98 -0.52 0.63 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 42 Appendix - TABLE 6 Effects of Bonus Program on Change in Debt for those that Do Not have Debt and Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Change in Debt (N=1624) Variable Coefficient Cashback Dummy 95.32 -4.12 -2.38 -90.39 -33.22 -0.04 0.01 17.77 -9.82 156.84 -1.75 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 34.65 1.85 1.56 47.73 17.72 0.15 0.01 26.15 29.49 320.51 2.77 T-Stat 2.75 -2.23 -1.53 -1.89 -1.88 -0.29 1.31 0.68 -0.33 0.49 -0.63 Effects of Bonus Program on Change in Debt for those that Do Not have Debt and DO Not Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Change in Debt (N=2318) Variable Coefficient Cashback Dummy 166.66 -9.29 0.72 63.39 -51.64 0.08 0.00 -12.47 29.70 155.98 -0.27 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error 45.63 3.28 2.54 138.35 10.70 0.11 0.01 26.44 29.52 343.55 2.94 T-Stat 3.65 -2.84 0.28 0.46 -4.82 0.72 0.41 -0.47 1.01 0.45 -0.09 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 43 Appendix - TABLE 7 Effects of Bonus Program on Quarterly Change in Total Balance for those that Do Not have Debt and Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Variable Cashback Dummy Quarterly Change in Total Balance (N=2113) Coefficient 244.38 8.83 69.28 423.11 -92.11 0.00 1348.70 215.99 3594.61 -36.44 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error T-Stat 183.82 11.91 14.48 704.02 73.89 0.18 282.79 295.81 1926.52 19.83 1.33 0.74 4.78 0.60 -1.25 0.00 4.77 0.73 1.87 -1.84 Effects of Bonus Program on Quarterly Change in Total Balance for those that Do Not have Debt and DO Not Use the Card Prior to the program 1 Month Before the Program Starts, 3 Months After Variable Cashback Dummy Quarterly Change in Total Balance (N=3095) Coefficient 276.80 30.14 62.18 984.41 98.54 0.20 1940.09 15.21 859.82 -3.73 (Receiving Cashback=1, Not in Program=0) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error T-Stat 159.70 12.43 13.47 942.20 51.72 0.20 779.58 329.80 1867.58 17.79 1.73 2.42 4.62 1.04 1.91 1.00 2.49 0.05 0.46 -0.21 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the cashback indicator variable in equation (1) for spending for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 44 Appendix - TABLE 8A Effects of APR Reduction on Spending in Treatment Group 1 Month Before the Program Starts, 3 Months After Spending (N=4633) Variable APR Reduction Dummy Coefficient 1097.79 -15.74 2.43 -387.62 -12.65 0.05 0.01 -69.78 4.78 6.84 -3.86 (Receiving APR Reduction=1) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error T-Stat 84.80 1.88 2.33 155.90 6.47 0.08 0.01 71.12 66.09 622.06 6.08 12.95 -8.37 1.04 -2.49 -1.96 0.65 0.47 -0.98 0.07 0.01 -0.64 Effects of APR Reduction on Quarterly Change in Total Balance in Treatment Group 1 Month Before the Program Starts, 3 Months After Variable Quarterly Change in Total Balance (N=4633) Coefficient APR Reduction Dummy 636.15 14.22 98.92 97.99 -0.12 0.36 59.94 385.37 1243.31 -6.33 (Receiving APR Reduction=1) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 Standard Error T-Stat 432.67 8.70 21.24 395.68 50.01 0.28 628.03 481.07 2040.60 23.91 1.47 1.63 4.66 0.25 0.00 1.28 0.10 0.80 0.61 -0.26 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the APR indicator variable in equation (2) for change in credit card balances at the credit bureaus for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 45 Appendix - TABLE 8B Effects of APR Reduction on Change in Debt in Treatment Group 1 Month Before the Program Starts, 3 Months After Change in Debt (N=4022) Variable Coefficient APR Reduction Dummy (Receiving APR Reduction=1) Change in Behavior Score from Previous Quarter Change in FICO Score from Previous Quarter Account Age Realized APR Credit Line Total Balance on All Credit Cards Total Credit Cards with Debt Total Number of Credit Cards Age Age^2 1058.65 -26.81 2.79 -336.92 -16.02 0.10 0.00 -34.43 -35.98 -263.55 -1.27 Standard Error 117.51 2.72 2.84 163.22 8.43 0.11 0.02 88.32 84.86 775.01 7.60 T-Stat 9.01 -9.84 0.98 -2.06 -1.90 0.87 0.06 -0.39 -0.42 -0.34 -0.17 Notes: This table reports the coefficient value, the standard error, and the t-statistics for the APR indicator variable in equation (2) for change in card debt for the period 1 month before and 3 months after the program starts. All values are in current dollars (2000-2002). The results control for individual fixed effects and clustered standard errors that are adjusted for heteroscedasticity across individuals and correlation within. 46 Working Paper Series A series of research studies on regional economic issues relating to the Seventh Federal Reserve District, and on financial and economic topics. Risk Taking and the Quality of Informal Insurance: Gambling and Remittances in Thailand Douglas L. Miller and Anna L. Paulson WP-07-01 Fast Micro and Slow Macro: Can Aggregation Explain the Persistence of Inflation? Filippo Altissimo, Benoît Mojon, and Paolo Zaffaroni WP-07-02 Assessing a Decade of Interstate Bank Branching Christian Johnson and Tara Rice WP-07-03 Debit Card and Cash Usage: A Cross-Country Analysis Gene Amromin and Sujit Chakravorti WP-07-04 The Age of Reason: Financial Decisions Over the Lifecycle Sumit Agarwal, John C. Driscoll, Xavier Gabaix, and David Laibson WP-07-05 Information Acquisition in Financial Markets: a Correction Gadi Barlevy and Pietro Veronesi WP-07-06 Monetary Policy, Output Composition and the Great Moderation Benoît Mojon WP-07-07 Estate Taxation, Entrepreneurship, and Wealth Marco Cagetti and Mariacristina De Nardi WP-07-08 Conflict of Interest and Certification in the U.S. IPO Market Luca Benzoni and Carola Schenone WP-07-09 The Reaction of Consumer Spending and Debt to Tax Rebates – Evidence from Consumer Credit Data Sumit Agarwal, Chunlin Liu, and Nicholas S. Souleles WP-07-10 Portfolio Choice over the Life-Cycle when the Stock and Labor Markets are Cointegrated Luca Benzoni, Pierre Collin-Dufresne, and Robert S. Goldstein WP-07-11 Nonparametric Analysis of Intergenerational Income Mobility with Application to the United States Debopam Bhattacharya and Bhashkar Mazumder WP-07-12 How the Credit Channel Works: Differentiating the Bank Lending Channel and the Balance Sheet Channel Lamont K. Black and Richard J. Rosen WP-07-13 Labor Market Transitions and Self-Employment Ellen R. Rissman WP-07-14 First-Time Home Buyers and Residential Investment Volatility Jonas D.M. Fisher and Martin Gervais WP-07-15 1 Working Paper Series (continued) Establishments Dynamics and Matching Frictions in Classical Competitive Equilibrium Marcelo Veracierto WP-07-16 Technology’s Edge: The Educational Benefits of Computer-Aided Instruction Lisa Barrow, Lisa Markman, and Cecilia Elena Rouse WP-07-17 The Widow’s Offering: Inheritance, Family Structure, and the Charitable Gifts of Women Leslie McGranahan WP-07-18 Incomplete Information and the Timing to Adjust Labor: Evidence from the Lead-Lag Relationship between Temporary Help Employment and Permanent Employment Sainan Jin, Yukako Ono, and Qinghua Zhang WP-07-19 A Conversation with 590 Nascent Entrepreneurs Jeffrey R. Campbell and Mariacristina De Nardi WP-07-20 Cyclical Dumping and US Antidumping Protection: 1980-2001 Meredith A. Crowley WP-07-21 Health Capital and the Prenatal Environment: The Effect of Maternal Fasting During Pregnancy Douglas Almond and Bhashkar Mazumder WP-07-22 The Spending and Debt Response to Minimum Wage Hikes Daniel Aaronson, Sumit Agarwal, and Eric French WP-07-23 The Impact of Mexican Immigrants on U.S. Wage Structure Maude Toussaint-Comeau WP-07-24 A Leverage-based Model of Speculative Bubbles Gadi Barlevy WP-08-01 Displacement, Asymmetric Information and Heterogeneous Human Capital Luojia Hu and Christopher Taber WP-08-02 BankCaR (Bank Capital-at-Risk): A credit risk model for US commercial bank charge-offs Jon Frye and Eduard Pelz WP-08-03 Bank Lending, Financing Constraints and SME Investment Santiago Carbó-Valverde, Francisco Rodríguez-Fernández, and Gregory F. Udell WP-08-04 Global Inflation Matteo Ciccarelli and Benoît Mojon WP-08-05 Scale and the Origins of Structural Change Francisco J. Buera and Joseph P. Kaboski WP-08-06 Inventories, Lumpy Trade, and Large Devaluations George Alessandria, Joseph P. Kaboski, and Virgiliu Midrigan WP-08-07 2 Working Paper Series (continued) School Vouchers and Student Achievement: Recent Evidence, Remaining Questions Cecilia Elena Rouse and Lisa Barrow Does It Pay to Read Your Junk Mail? Evidence of the Effect of Advertising on Home Equity Credit Choices Sumit Agarwal and Brent W. Ambrose WP-08-08 WP-08-09 The Choice between Arm’s-Length and Relationship Debt: Evidence from eLoans Sumit Agarwal and Robert Hauswald WP-08-10 Consumer Choice and Merchant Acceptance of Payment Media Wilko Bolt and Sujit Chakravorti WP-08-11 Investment Shocks and Business Cycles Alejandro Justiniano, Giorgio E. Primiceri, and Andrea Tambalotti WP-08-12 New Vehicle Characteristics and the Cost of the Corporate Average Fuel Economy Standard Thomas Klier and Joshua Linn WP-08-13 Realized Volatility Torben G. Andersen and Luca Benzoni WP-08-14 Revenue Bubbles and Structural Deficits: What’s a state to do? Richard Mattoon and Leslie McGranahan WP-08-15 The role of lenders in the home price boom Richard J. Rosen WP-08-16 Bank Crises and Investor Confidence Una Okonkwo Osili and Anna Paulson WP-08-17 Life Expectancy and Old Age Savings Mariacristina De Nardi, Eric French, and John Bailey Jones WP-08-18 Remittance Behavior among New U.S. Immigrants Katherine Meckel WP-08-19 Birth Cohort and the Black-White Achievement Gap: The Roles of Access and Health Soon After Birth Kenneth Y. Chay, Jonathan Guryan, and Bhashkar Mazumder WP-08-20 Public Investment and Budget Rules for State vs. Local Governments Marco Bassetto WP-08-21 Why Has Home Ownership Fallen Among the Young? Jonas D.M. Fisher and Martin Gervais WP-09-01 Why do the Elderly Save? The Role of Medical Expenses Mariacristina De Nardi, Eric French, and John Bailey Jones WP-09-02 3 Working Paper Series (continued) Using Stock Returns to Identify Government Spending Shocks Jonas D.M. Fisher and Ryan Peters WP-09-03 Stochastic Volatility Torben G. Andersen and Luca Benzoni WP-09-04 The Effect of Disability Insurance Receipt on Labor Supply Eric French and Jae Song WP-09-05 CEO Overconfidence and Dividend Policy Sanjay Deshmukh, Anand M. Goel, and Keith M. Howe WP-09-06 Do Financial Counseling Mandates Improve Mortgage Choice and Performance? Evidence from a Legislative Experiment Sumit Agarwal,Gene Amromin, Itzhak Ben-David, Souphala Chomsisengphet, and Douglas D. Evanoff WP-09-07 Perverse Incentives at the Banks? Evidence from a Natural Experiment Sumit Agarwal and Faye H. Wang WP-09-08 Pay for Percentile Gadi Barlevy and Derek Neal WP-09-09 The Life and Times of Nicolas Dutot François R. Velde WP-09-10 Regulating Two-Sided Markets: An Empirical Investigation Santiago Carbó Valverde, Sujit Chakravorti, and Francisco Rodriguez Fernandez WP-09-11 The Case of the Undying Debt François R. Velde WP-09-12 Paying for Performance: The Education Impacts of a Community College Scholarship Program for Low-income Adults Lisa Barrow, Lashawn Richburg-Hayes, Cecilia Elena Rouse, and Thomas Brock Establishments Dynamics, Vacancies and Unemployment: A Neoclassical Synthesis Marcelo Veracierto WP-09-13 WP-09-14 The Price of Gasoline and the Demand for Fuel Economy: Evidence from Monthly New Vehicles Sales Data Thomas Klier and Joshua Linn WP-09-15 Estimation of a Transformation Model with Truncation, Interval Observation and Time-Varying Covariates Bo E. Honoré and Luojia Hu WP-09-16 Self-Enforcing Trade Agreements: Evidence from Antidumping Policy Chad P. Bown and Meredith A. Crowley WP-09-17 Too much right can make a wrong: Setting the stage for the financial crisis Richard J. Rosen WP-09-18 4 Working Paper Series (continued) Can Structural Small Open Economy Models Account for the Influence of Foreign Disturbances? Alejandro Justiniano and Bruce Preston WP-09-19 Liquidity Constraints of the Middle Class Jeffrey R. Campbell and Zvi Hercowitz WP-09-20 Monetary Policy and Uncertainty in an Empirical Small Open Economy Model Alejandro Justiniano and Bruce Preston WP-09-21 Firm boundaries and buyer-supplier match in market transaction: IT system procurement of U.S. credit unions Yukako Ono and Junichi Suzuki Health and the Savings of Insured Versus Uninsured, Working-Age Households in the U.S. Maude Toussaint-Comeau and Jonathan Hartley WP-09-22 WP-09-23 The Economics of “Radiator Springs:” Industry Dynamics, Sunk Costs, and Spatial Demand Shifts Jeffrey R. Campbell and Thomas N. Hubbard WP-09-24 On the Relationship between Mobility, Population Growth, and Capital Spending in the United States Marco Bassetto and Leslie McGranahan WP-09-25 The Impact of Rosenwald Schools on Black Achievement Daniel Aaronson and Bhashkar Mazumder WP-09-26 Comment on “Letting Different Views about Business Cycles Compete” Jonas D.M. Fisher WP-10-01 Macroeconomic Implications of Agglomeration Morris A. Davis, Jonas D.M. Fisher and Toni M. Whited WP-10-02 Accounting for non-annuitization Svetlana Pashchenko WP-10-03 Robustness and Macroeconomic Policy Gadi Barlevy WP-10-04 Benefits of Relationship Banking: Evidence from Consumer Credit Markets Sumit Agarwal, Souphala Chomsisengphet, Chunlin Liu, and Nicholas S. Souleles WP-10-05 The Effect of Sales Tax Holidays on Household Consumption Patterns Nathan Marwell and Leslie McGranahan WP-10-06 Gathering Insights on the Forest from the Trees: A New Metric for Financial Conditions Scott Brave and R. Andrew Butters WP-10-07 Identification of Models of the Labor Market Eric French and Christopher Taber WP-10-08 5 Working Paper Series (continued) Public Pensions and Labor Supply Over the Life Cycle Eric French and John Jones WP-10-09 Explaining Asset Pricing Puzzles Associated with the 1987 Market Crash Luca Benzoni, Pierre Collin-Dufresne, and Robert S. Goldstein WP-10-10 Does Prenatal Sex Selection Improve Girls’ Well‐Being? Evidence from India Luojia Hu and Analía Schlosser WP-10-11 Mortgage Choices and Housing Speculation Gadi Barlevy and Jonas D.M. Fisher WP-10-12 Did Adhering to the Gold Standard Reduce the Cost of Capital? Ron Alquist and Benjamin Chabot WP-10-13 Introduction to the Macroeconomic Dynamics: Special issues on money, credit, and liquidity Ed Nosal, Christopher Waller, and Randall Wright WP-10-14 Summer Workshop on Money, Banking, Payments and Finance: An Overview Ed Nosal and Randall Wright WP-10-15 Cognitive Abilities and Household Financial Decision Making Sumit Agarwal and Bhashkar Mazumder WP-10-16 Complex Mortgages Gene Amromin, Jennifer Huang, Clemens Sialm, and Edward Zhong WP-10-17 The Role of Housing in Labor Reallocation Morris Davis, Jonas Fisher, and Marcelo Veracierto WP-10-18 Why Do Banks Reward their Customers to Use their Credit Cards? Sumit Agarwal, Sujit Chakravorti, and Anna Lunn WP-10-19 6