Credit Demand and Interest Rate Transparency

Transcription

1 Credit Demand and Interest Rate Transparency Bruno Ferman Department of Economics, MIT Preliminary - please do not quote or cite without permission. Abstract How effectively can consumers process information? In this paper, I test whether firms are able to exploit clients limited attention by changing the relative visibility of the information presented in credit solicitations. Using a randomized field experiment, I estimate how credit demand is affected by prices, interest rate disclosure, and default options for a sample of credit card clients in Brazil who were offered a menu of installment plans to pay off their balances. For the average consumer, there is no evidence that clients fail to consider information that is not emphasized. On average, clients are sensitive to prices even when this information is hidden, and take-up rates and price sensitivity are not significantly different when the interest rate is emphasized. However, there is one important exception. High-risk clients are price insensitive when the interest rate is hidden, but they react to prices when the interest rate is more visible. Therefore, while information disclosure regulations might have a limited effect for the full population, they might be relevant for high-risk clients. Take-up rates are not affected by changes in the suggested payment plan, which suggests that clients are able to consider options that are not emphasized. However, the suggested plan has an important effect on determining which payment plan clients actually choose. This is consistent with clients using the suggested plan as a default option in order to avoid making a decision about which payment plan to choose. This effect is weaker (though still significant) when the stakes are higher. 1 Introduction Are consumers misled by credit solicitations? There is a growing body of evidence suggesting that consumers have difficulties in making optimal decisions (for example, Stango and Zinman (2009), Agarwal et al. (2009), and Gross and Souleles (2002)), and that firms exploit consumers behavioral biases (for example, Stango and Zinman (2011) and Barr et al. (2008)). This evidence has provided support for policy makers to propose regulations in consumer financial markets, either 1

2 by restricting consumers options or by regulating the information that lenders must present to their clients (Campbell et al. (2011)). In this paper, I test whether firms are able to exploit clients limited attention by changing the relative visibility of the information presented in credit solicitations, and whether information disclosure regulations would be effective in changing clients behavior. In a randomized field trial conducted with a credit card company in Brazil, I study how credit demand is affected by prices, interest rate disclosure, and default options. These clients were offered a menu of installment plans to pay off their credit card balances with randomly assigned interest rates. The clients also received a randomly assigned advertisement layout of the payment plans, with varying visibility of the interest rate information (whether this information was emphasized or was presented in a footnote), and a different payment plan assigned as the suggested payment plan among the menu of options. Unlike the price changes, the proposed changes in the advertisement layout only affect the relative visibility of different features of the offer, holding constant both the information content that is provided to the clients and their real options. However, these changes can affect clients behavior if they have limited attention, so that they put more weight on information that is more visible to them when making their decisions. The hypothesis that individuals are able to process and consider in their decision making all the available information has been challenged since Simon (1955). Barr et al. (2008), have pointed out the importance of limited attention in consumer credit 2

3 settings. They argued that if clients have limited attention, creditors can easily evade interest rate disclosure regulations by leading their clients to focus on the low monthly payment in the contract, yet still complying with disclosure regulations. The two changes in the advertisement layout will affect which information is more visible to the client. If the contract interest rate is hidden, then clients are induced to focus mainly on the monthly payments; when the interest rate is emphasized, clients are induced to consider both the monthly payment and the interest rate. Also, changing the suggested payment plan can lead the client to focus on either a lower monthly payment (longer maturity) or a higher monthly payment (shorter maturity) contract. Even if clients are able to consider all of the available options, changing the payment plan that is emphasized can affect their behavior if they use the suggested plan as a default option in order to avoid making a decision about which payment plan to choose. Bertrand et al. (2010) studied the importance of the advertisement content of credit offers, and found that frames and cues have significant impact on take-up rates. The evidence on how interest rate disclosure affects credit demand, however, is mixed. In one of their treatments, Bertrand et al. (2010) estimate the impact of interest rate disclosure on credit demand. They do not find a significant effect of interest rate disclosure on take-up rates. Bertrand and Morse (2009) also study the effects of information disclosure in credit markets using a randomized field trial, and find that information that helps people aggregate the costs of payday loans over time has a significant effect in reducing take-up rates. However, these two papers do not study the interaction between prices 3

4 and information disclosure. As proponents of disclosure regulations often argue, better disclosure regulations would make consumers more interest rate elastic, enhancing competition among lenders (Kroszner (2007)). Therefore, understanding how information regulation affects demand price elasticity is crucial to evaluating the potential of these policies to enhance competition in financial markets. Stango and Zinman (2011) studied the equilibrium effects of weakening enforcement of APR disclosure in the Truth-in-Lending Act; they find that it increased the gap in interest rate paid by clients who underestimate the interest rate more and clients who underestimate it less. This paper provides no evidence that, on average, clients fail to consider information that is not emphasized in making their decisions. Overall, clients are sensitive to prices, even when the interest rate information is hidden. The interest rate elasticities estimated in this experiment are in line with recent estimates from randomized or quasi-experiments in both developed and developing countries (Karlan and Zinman (2008), Gross and Souleles (2002), Huang and Tan (2009), and Attanasio et al. (2008)). Even more interesting, the take-up rates and interest rate elasticities do not differ significantly when the interest rate is emphasized. However, there is one important exception. While low- and medium-risk clients are interest rate elastic, and their demand curve for payment plans is not affected by the visibility of the interest rate, high-risk clients are not sensitive to prices when the interest rate is hidden, but they become interest rate elastic when the interest rate is more visible. These results are consistent with high-risk borrowers being less attentive to the details of the contract, so the visibility of the contract s interest rate does have an impact on 4

5 their borrowing decisions. The results on changes in the suggested payment plan also do not indicate that clients only consider the payment plan that is emphasized more in the advertisement s layout. It is not possible to reject that take-up rates are constant with respect to changes in the maturity of the suggested plan. In particular, there is no evidence that emphasizing payment plans with lower monthly payments (longer maturity) will attract more clients. However, the suggested maturity does have a strong influence on which payment plan is chosen. Most of the clients choose either the shortestterm payment plan or the suggested one. This suggests that clients use the suggested payment plan as a default option in order to avoid making a decision about which payment plan to choose. This is consistent with the evidence that having many options may create feelings of conflict and indecision (for example, Shafir et al. (1993), Bertrand et al. (2010), and Iyengar et al. (2004)). The probability of following the suggested plan is lower, however, when the interest rates are higher and when the emphasized plan has a longer maturity. This suggests that clients are less likely to simply follow the suggested plan when its cost relative to the shortest-term plan is higher. Therefore, while clients are susceptible to nudges, these results provide evidence that nudges are relatively less important when the stakes are higher. 5

6 2 Market Setting From 2000 to 2009, the number of credit cards issued in Brazil increased from 28 to 136 million (an increase of 378%), and the widespread use of credit cards is changing the way that households borrow in Brazil. In June 2000, almost 60% of the personal borrowing in Brazil corresponded to current account overdrafts and less than 5% of it corresponded to credit card lines of credits; those numbers were 34% and 26% in May These changes reflect the increasing importance of the credit card in household borrowing decisions in Brazil, despite the high interest rates charged: revolving interest rates usually are higher than 10% per month (214% per year). In the last decade, access to credit cards in Brazil began to reach low and middle class consumers, especially through credit card companies associated with retail stores. The major retail stores in Brazil offer their clients the opportunity to apply for store credit cards so that they can finance their purchases in installments. In some cases the store credit cards can be used only at the originating retail store; in other cases they can be used elsewhere. The credit card company that participated in this study is associated with a major retail store in Brazil; its clientele are mostly low- and middle-income consumers. These credit cards can be used not only at the associated retail store but also as an ordinary credit card. At the time of the experiment, this company had more than 5 million active clients. Credit card companies in Brazil usually offer two borrowing alternatives if clients do not pay 6

7 their balance in full. The most common alternative is to use the credit card s revolving line of credit: the client pays an amount equal to or greater than the minimum required payment, but smaller than the credit card balance. The remaining balance plus interest accrued is carried over to the next billing period. For the clients in this study, the revolving rate ranges from 11.89% to 15.99% per month, and the minimum payment is equal to 15% of the credit card balance. To the consumer, the main advantage of this source of credit is that it is pre-approved, and therefore it is easier to access than other borrowing alternatives. Also, the client has the flexibility of choosing the amount he wants to, or can, pay (assuming it is equal to or greater than the minimum payment so he does not default). The other alternative that credit card companies offer to their clients is the possibility of paying off their debt with installment payment plans. In this case, the client is offered a menu of plans, with different monthly payments (M n ) and a fixed number of installments (n), which implies an interest rate (r n ). Given a balance (B), the number of installments(n), and the monthly interest rate (r n ), the monthly payment (M n )is 1 : M n = B (1+rn)n 1 r n (1+r n) n 1 If the client does not make any new purchases, then after n monthly payments of this amount, he will have repaid his credit card debt. The number of installments usually ranges from 4 to 24 months, and the interest rate may be equal to or lower than the revolving rate. 1 The amount of the monthly payments is defined such that, given the contract interest rate (r n), the present value of the stream of payments is equal to the credit card balance (B). 7

8 This type of credit is also pre-approved and easily accessible. To enroll in a payment plan, the client simply has to pay the exact amount of the monthly payment of the plan he has chosen. In doing so, the client automatically enrolls in the chosen plan, and is charged the remaining installments on his credit card statement for the n 1 following months. For example, if the credit card statement presents a payment plan offer of 6 installments with a monthly payment of R$183.38, then the client simply has to pay exactly R$183.38, and he will be charged that amount on the next five following billing cycles. There is no penalty for canceling the payment plan. If the client wants to cancel the plan, he simply has to call customer service and the present value of the unpaid installments will be charged on his next billing cycle. However, many clients may not be completely aware of the possibility of canceling these payment plans, and they might be afraid of facing additional fees and having to to deal with a bureaucracy to do this. These concerns might prevent some clients from choosing such longer-term credit offers. In fact, the number of clients who cancel these payment plans is extremely low. Under a payment plan contract, clients need to pay the monthly installments in full in order to stay current. The credit card minimum payment will be equal to the monthly installment of the chosen payment plan, plus 15% of any new purchases made with the credit card. Therefore, if the client follows the payment plan and does not make new purchases, he will pay off his credit card balance in full after n months, whereas it would take much longer to pay off his balance if he makes 8

9 only the minimum payment every month. Other borrowing alternatives include personal loans from banks (average monthly interest rate of 4.79%), checking accounts overdraft (average monthly interest rate of 7.40%), personal loans from finance houses (average monthly interest rate of 9.87%), or informal loans. In case of default or delayed payments with the credit card, the interest rate and fees are substantial. If a client does not pay on time, then in addition to the revolving interest rate (which can be up to 15.99% per month) he is charged a 2% late payment fee over his full balance, plus interest charges on future installments. The client also will have his credit card blocked after 11 days, and his name will be reported to credit bureaus after 20 days. If the client stays in default for more than 70 days, then his credit card will be canceled, and he will never be able to apply for this particular credit card again. 3 Borrowing Decisions and Interest Rate Transparency When deciding whether to enroll in a payment plan or to use a revolving line of credit, the client faces a trade-off between lower interest rates and flexibility in the stream of payments: interest rates on payment plans are usually lower than revolving credit card rates, but payment plans have less flexibility in terms of the stream of payments the client must make. To enroll in a payment plan, the client could not pay more than the first monthly payment: for example, with a 6-month 9

10 payment plan and an interest rate of 11.89%, the client would have to pay only 21.67% of his credit card balance. In addition, because clients might be unaware of the option to cancel these contracts, or if they believe that it would be costly to do so, then a payment plan would imply a commitment to borrow for a longer period. In both cases, that inflexibility implies that clients might have to distort their optimal consumption streams if they choose such plan. It might instead be optimal for clients to use their revolving line of credit to pay off a higher fraction of their balances, or to pay off their debt in a shorter period, even if this means borrowing at a higher interest rate. Past data reveals that in the absence of payment plan offers clients paid more than 30% of their balances in almost 92% of the cases. Also, only 12.59% of clients used the revolving line of credit for six consecutive months. In other words, even though payment plans usually have lower interest rates, enrolling in such plans might imply borrowing more with their credit cards than usual for most clients. The lower the interest rate, the more attractive the payment plans should be for the clients, but figuring out whether a payment plan has an attractive interest rate may not be a straightforward task. Contract interest rates usually are hidden by the credit card companies, increasing the computational cost of comparing payment plans and alternative options. Information disclosure regulations in Brazil require that credit card companies provide information on the interest rate charged in financial contracts. However, firms evade the spirit of these regulations by presenting the information only in a footnote, leading clients with limited attention to focus on the low amount of 10

11 the monthly payments in payment plan offers (Barr et al. (2008)). Although a rational consumer would be able to calculate the interest rate on payment plans given the number of installments and monthly payments, limited financial literacy and cognitive biases may prevent clients from correctly evaluating contract interest rates (Stango and Zinman (2009)). A policy that requires lenders to state interest rates transparently would reduce the computational costs of comparing different borrowing alternatives. Under that hypothesis, interest rate disclosure could lead clients to put more weight on the interest rate of the contract when deciding among different borrowing alternatives, thus implying that the demand for payment plans would become more interest rate elastic, as argued by Kroszner (2007). A second important decision in this setting is the maturity choice of the payment plan. Clients usually receive a menu of payment plan options, varying the number of installments and the amount of the monthly installments. When deciding among the possible payment plans, the client will be trading off the duration of his debt and the monthly payment burden needed to pay off his debt. The higher the interest rate, the more costly it is to choose a long-term relative to a short-term payment plan. Therefore, it would be expected that clients choose shorter-term contracts and make more careful decisions when the interest rates are higher. Firms usually emphasize the longestterm payment plan as a suggested maturity, possibly in an attempt to attract clients focused on lower monthly payments and because, given the high interest rates, longer-term contracts are more profitable. 11

12 4 Experimental Design and Implementation 4.1 Treatment Dimensions In a first treatment dimension, clients were randomly assigned to groups which the monthly interest rate of the payment plans was 3.99%, 7.49%, or 11.89%. The objective of this first treatment dimension was to estimate the price sensitivity of demand for payment plans. Given the payment plans interest rate, clients had four different payment plan options, with the number of installments varying from 6, 8, 10, or 12 months. The Table below displays the monthly payments for each contract, assuming a balance of R$ A client with an assigned interest rate of r n would be able to choose among the contracts in the corresponding column. Along with their credit card statement, clients received a one-page advertisement where they were informed of the payment plan offers. In a second treatment dimension, clients were randomly chosen to receive one of two different advertisement layouts. The standard advertisement (with a hidden interest rate) stated that the client could pay off his balance with payment plans with a special interest rate. The advertisement 2 The exchange rate during the experiment was US$1 R$

13 then displayed an example of one of the payment plan options, saying you can pay off your balance of B in n installments of M n. Figure 1 presents the one-page advertisement for the payment plans, along with the credit card statement. In that advertisement layout, the interest rate of the contract is not emphasized and is only present in a footnote. Therefore, the client is induced to focus on the monthly payments (M n ) of the contract. Clients also were shown a table with all four payment plan options at the top of their credit card statement. This table had a suggested payment plan (the same as in the one-page advertisement), and the interest rate of the payment plans was presented in a small font size next to the table. Although the interest rate on the contracts was not emphasized in this advertisement, the client easily could find out this information if he decided to do so. However, the advertisement layout was designed to induce clients to focus on the monthly payments of the suggested plan instead. The alternative layout for the one-page advertisement (which emphasized the interest rate) was exactly the same as the standard ad, except that the interest rate information was more visible (Figure 2). This layout stated that the client could pay off his balance using payment plans with a special interest rate of XX.XX% (in large font size). Clients who received this layout also had the same table at the top of their credit card statement as those who received the standard layout. Even though both layouts presented the same information, the different layouts could affect clients decisions because of limited attention, since the value of the interest rate was more visible in the alternative layout (DellaVigna (2009)). This second treatment dimension attempts to estimate how 13

14 demand varies with information disclosure and, especially, how price demand sensitivity varies when the interest rate of the contract is hidden versus when it is emphasized. Finally, a third treatment dimension has the suggested maturity in the advertisement (6, 8, 10, or 12 months) randomly assigned among clients, independent of the interest rate and the advertisement layout randomization. The standard default option used by the firm is the 12-month contract. The firm expected that clients would focus on the monthly payments of the suggested plan and thus be more attracted if a longer-term/smaller monthly payments plan was emphasized. 4.2 Sample Frame and Randomization Balance For this experiment, 28, 340 credit card clients were selected to be offered a one-time menu of payment plans, in either July or September Clients who had not used their revolving line of credit in the previous six months were excluded from the sample selection 3. However, the credit card company only offers payment plans to clients with credit card balances greater than R$100.00, and if they are not on default. Also, some clients received the payment plan offers but were selected to receive a different advertisement (not related to this experiment). Ultimately, the experimental sample had 19, 690 clients who received payment plan offers and who received the standard or the alternative advertisement layouts as presented in Figures 1 and 2. Table 1 shows the sample size in each treatment cell. Table 2 presents the baseline characteristics 3 This was a requirement from the credit card company, since these clients already were participating in another experiment. 14

15 of the final sample. During the experiment, the exchange rate was US$1 R$1.75. The average credit card balance was R$ In around 30% of the cases, clients used the revolving credit line, and they revolved 14.7% of their balances on average. This is less than the proportion of households with outstanding credit card balances in the US, which might be because revolving interest rates are much higher in Brazil. Clients made late payment in 18% of the cases. In sum, clients were charged R$20.6 per month in revolving interest charges and late payment fees. Columns 2 to 4 of Table 2 show baseline characteristics separately for each risk category group. Risk categories are defined by the credit card company based on credit bureaus and own data information. On average, credit card balance is slightly higher for low-risk than from high-risk clients (R$678.0 vs R$551.5), while the credit card limit for low-risk clients is much higher than for high-risk clients (R$ vs R$675.9). This implies that high-risk clients use a much larger proportion of their credit card limit. The proportion of clients who use the revolving line of credit, the probability of late payments, and the monthly interest and fees charged are all higher for highrisk clients. Table 2 also reports the 8-month probability of default for each category group 4. As expected, low-risk clients have a lower probability of default (2.8%) than high-risk clients (13.1%). The results in Tables 3 and 4 present the randomization balance across the treatment dimensions. Table 3 shows the averages for baseline variables in the final sample in each interest rate x advertisement layout treatment cells. Table 4 shows the same statistics for suggested maturity x 4 This is the probability that, conditional on being current in the base month, a client does not make the minimum payment for 70 days at some point within the following 8 months. These calculations are based on an outside sample. 15

16 advertisement layout cells. The p-values of the test that each of these variables has the same mean across the different treatment groups suggest that the sample is well balanced. In the Appendix Tables A1 to A3, it is shown that the sample is also well balanced when low-, middle- and high-risk clients are analyzed separately. 5 Methodology Exploiting the fact that the interest rates on payment plans were randomly assigned, it is possible to estimate how credit demand varies with the interest rate by using the following linear probability models (logit marginal effects are also presented): Y i = α + δ emphasized i + β 1 (r i r)+β 2 (r i r) emphasized i + ε i (1) where Y i is equal to one if client i chose to enroll in a payment plan, r i is the interest rate offered, r is the average interest rate, and emphasized i is a dummy variable equal to one if client i received the alternative advertisement layout. The base group is clients who received the standard (hidden interest rate) advertisement layout. Because interest rates were randomly assigned, ˆβ 1 yields consistent estimates of the price-sensitivity of payment plans demand. Also, because clients were randomly assigned into receiving an advertisement with a hidden or an emphasized interest rate, ˆδ yields consistent estimates of the average effect on take-up of emphasizing the interest rate, while ˆβ 2 16

17 yields consistent estimates of how emphasizing the interest rate affects price-sensitivity of demand. The interest rate elasticities under the two different advertisement layouts then will be calculated as: Ê r,hidden = ˆβ 1 r hidden Ȳ hidden and Êr,emphasized =(ˆβ 1 + ˆβ 2 ) r emphasized Ȳ emphasized (2) where r layout and Ȳlayout are, respectively, average interest rate and take-up rates when the advertisement layout had either a hidden or emphasized interest rate. The standard errors of Êr,hidden and Êr,emphasized are bootstraped, which takes into account the fact that the average take-up rates also are estimated. Again, because of the random assignment of advertisement layouts, the differences between Êr,emphasized and Êr,hidden yield the effect of making the interest rate more transparent on the interest rate elasticity. Similarly, we can estimate demand sensitivity with respect to suggested maturity using the following model: Y i = α + δ emphasized i + γ 1 (n i n)+γ 2 (n i n) emphasized i + ε i (3) where n i is the suggested maturity offered to client i, and n is the average suggested maturity. The suggested maturity elasticities are calculated as: 17

18 Ê n,hidden =ˆγ 1 n hidden Ȳ hidden and Ên,emphasized =(ˆγ 1 +ˆγ 2 ) n emphasized Ȳ emphasized (4) 6 Results 6.1 Interest Rate Elasticity and Interest Rate Transparency The first set of results shows the sensitivity of payment plans demand to prices when the interest rate is hidden (that is, under the standard advertisement layout). Table 5, column 1, presents the payment plans take-up rates for each interest rate offer when the interest rate is hidden. The results suggest that payment plans demand is affected by interest rate changes. The average take-up rate is 2.4% when the payment plans interest rate is 11.89%, and it goes up to 4.5% when the interest rate falls to 3.99%. The test that payment plans take-up is equal for all interest rate values rejects this hypothesis, with a p-value smaller than 1%. The interest rate sensitivity estimated using model (1) is presented in Panel i of Table 6. The interest rate coefficient ( ˆβ 1 ) is negative and statistically significant at the 1% level, confirming that the demand for payment plans is negatively sloped. Panel ii of Table 6 reports logit marginal effects of this model. As expected, interest rate sensitivity is similar for the two methods. The implied interest rate elasticity, estimated from (2), is (s.e ). Column 2 of Table 5 presents the payment plans take-up rates when the interest rate is empha- 18

19 sized. Column 3 shows the difference in take-up rates when the interest rate is emphasized versus when it is hidden. If clients have limited attention, then making the interest rate of the payment plan contract more visible might lead them to put more weight on this information. In such a case, demand should increase when the payment plan interest rate is low and decrease when the payment plan interest rate is high. The average effect of interest rate disclosure on the take-up rate is economically small, and not statistically different from zero. We estimate an increase of only 0.1 percentage points in the take-up rate when the interest rate is emphasized. However, such a small change might reflect the countervailing effects of emphasizing the interest rate when the it is high versus when the it is low. In fact, emphasizing the interest rate results in increased take-up rates when the interest rate is 3.99%, and decreased take-up rates when the interest rate is 11.89%. These differences, still, are not statistically different from zero. This is also shown in figure 3. As presented in Column 1 of Table 7 clients are slightly more interest rate elastic when the interest rate is more visible than when it is hidden (the elasticity goes from to 0.829). However, it is not possible to reject that these two elasticities are equal. The results for the full sample suggest that clients are sensitive to prices and that changing the visibility of the contract interest rate has a limited effect on clients behavior. However, these results hide an important heterogeneity when clients are classified according to their default risk. Columns 4, 7, and 10 of table 5 show payment plan take-up rates separately for low-, medium- 19

20 , and high-risk clients. These risk categories are defined by the credit card company based on information from credit bureaus and their own data. The demand for payment plans is increasing with the risk profile of the clients. The average take-up rate is 2.5% for low-risk clients, 4.1% for medium-risk clients, and 5.9% for high-risk clients. More importantly, payment plans take-up rates are fairly constant across different interest rates for high-risk clients. A joint test fails to reject the hypothesis that that take-up rates are equal for all interest rates for these clients, with a p-value of Columns 2 to 4 of Table 7 show the estimated interest rate elasticity for each risk group when the interest rate was hidden. Interest rate elasticity is equal to (s.e ) for the low-risk clients (p-value< 0.01) and (s.e ) for medium-risk clients (p-value< 0.10). For the high-risk clients, though, the estimated interest rate elasticity is equal to (s.e ), which is both small and statistically equal to zero. Such heterogeneity in interest rate elasticity across the different risk categories might reflect the fact that high-risk clients have fewer outside borrowing opportunities, or that these clients are more likely to default. Alternatively, high-risk clients might be less attentive when deciding among different borrowing alternatives. In that last case, clients should become more interest rate elastic when the interest rate is emphasized. Comparing take-up rates and interest rate elasticities when the interest rate rate is hidden versus emphasized for the low- and medium-risk clients, we find that making the interest rate more visible 20

21 has no impact on the behavior of these clients. These results are reported in tables 5 and 7. For the high-risk clients, though, emphasizing the interest rate value will strongly reduce payment plans demand when it is equal to 11.89% (from 5.7% to 2.9%, statistically significant at 5%), and will increase payment plans demand when it is equal to 3.99% or 7.49% (although these differences are not statistically significant). As reported in Column 4 of Table 7, the interest rate elasticity for this high-risk group changes from (s.e ) when the interest rate is hidden to (s.e ) when it is emphasized, becoming statistically different from zero at the 1% level. The p-value is in a test that the interest rate elasticity under these two advertisement layout is equal. These results are consistent with low- and medium-risk clients being more careful when choosing different borrowing alternatives, so that they look for the value of the interest rate of the contracts even when this information is not emphasized on the advertisement layout, and high risk borrowers being less attentive to the details of the contract, so that the salience of the interest rate has an impact on their borrowing decisions. 6.2 Suggested Maturity and Maturity Choice Columns 1 and 2 of Table 8, respectively, show the take-up rates by suggested maturity when the interest rate is hidden and when it is emphasized. Although all clients have the same menu of options, this treatment could affect consumers demand for payment plans if they focus only on 21

22 the emphasized payment plan. If clients focus mostly on the value of the monthly payments, then they would be more attracted when longer-term/smaller monthly payments plans were emphasized. However, it is not possible to reject the hypothesis that take-up rates are uniform across different maturity suggestions. The suggested maturity elasticities presented in Column 1 of Table 9 also reject the hypothesis that clients are more likely to enroll in a payment plan when the emphasized payment plan has a lower monthly payment. These elasticities are (s.e ) when the interest rate is hidden and (s.e ) when the interest rate is emphasized; both are small and not statistically significant. The lack of demand response with respect to the suggested maturity does not imply, however, that clients are indifferent with respect to the maturity of their payment plans, nor that the payment plan suggestion has no effect. Information on actual maturity choices shows that clients have strong preferences for short-term contracts. Of the clients who chose a payment plan, more then half of them chose the 6-month plan (the shortest available maturity choice). Figure 5 shows the distribution of payment plan choices by suggested maturity, and reveals that clients choose either the shortest available plan (6 months) or the suggested plan. For example, when the suggested maturity was 6 months, more than 80% of the clients chose this plan. When the suggested maturity was 12 months, nearly half of the clients chose the 6-month plan, and around one third of them chose the suggested plan. These results suggest that even when one option is emphasized, clients will consider other 22

23 alternatives if the suggested plan is not attractive. Therefore, take-up rates are not affected by changes in the suggested payment plan. However, the emphasized payment plan has a strong influence in determining which payment plan the clients choose. This may reflect the fact that clients are marginally indifferent among the different maturity options. Assuming that clients discount rates are smaller than the interest rates, then the cost of extending the maturity of the payment plans is increasing with the interest rate. Thus, clients should be more likely to select among the different alternatives rather than simply following the maturity suggestion when the interest rate is higher. Column 1 of Table 10 shows the probability of not following the maturity suggestion when the interest rate is 3.99%, and how this probability changes when the interest rate raises to 7.49% and 11.89%. All of these regressions in Table 10 are conditional on payment plan enrollment. When the interest rate is low, around half of the clients choose a payment plan that is different from the suggested one. This proportion is around 10 percentage points higher when the interest rate of the payment plans increases to 7.49% or 11.89%. However, it is not possible to reject the hypothesis that the probability of following the maturity suggestion is the same when the interest rate increases from 7.49% to 11.89%. These results should be viewed with caution. Because clients were interest rate elastic, and the sample is conditional on enrollment, there is presumably some selection bias in estimating the effects of changing the interest rate of the payment plans on maturity choices. Column 2 of Table 23

24 10 includes as covariates all of the baseline variables presented in Table 2. The results remain the same. In order to further explore the hypothesis that clients are more likely to choose a payment plan that is different from the suggested one when the cost of following the suggested plan is higher, in column 3 of Table 10 we show the correlation between the probability of not following the suggested plan and the cost of the suggested plan relative to the 6-month plan. The relative cost variable is constructed as the total nominal cost of the suggested payment plan divided by the total nominal cost of the 6-month plan. Clients with a suggested plan equal to 6 were excluded from this analysis. The cost variable attains its lowest value (1.038) when the suggested plan is the 8-month plan, and the interest rate is 3.99%. It reaches its highest value (1.325) when the interest rate is 11.89% and the suggested plan is the 12-month plan. The coefficient on this variable is positive and equal to 0.99 (p-value< 0.01), meaning that a change from the lowest to the highest relative cost of the suggested plan will increase the probability that the client will not simply follow the suggested plan by 30 percentage points. Although the regression presented in Column (3) includes covariates, it is possible that some omitted variables bias might still exist. To deal with this problem, the model presented in Column (4) instruments the cost of the suggested plan variable using suggested plan dummies. The idea is to explore the changes in the suggested plan cost that are driven by the changes in the suggested payment plan, not by changes in the interest rate. The estimated effect of the relative cost of the 24

25 suggested plan on the probability of not choosing the suggested plan is remarkably similar to the OLS estimate. Columns (5) to (8) presents the same results using the probability of choosing the shortest-term plan as a dependent variable, and the results are similar. These results suggest that clients use the suggested payment plan as a default option in order to avoid making a decision about which payment plan to choose. This is consistent with the evidence that having many options may create feelings of conflict and indecision (for example, Shafir et al. (1993), Bertrand et al. (2010), and Iyengar et al. (2004)). Interestingly, however, the use of the suggested plan as a default option is less relevant (although still significant) when the relative cost of the suggested plan is higher. 7 Conclusion This paper finds that, overall, clients are able to consider information that is not emphasized in credit solicitations. Demand is price sensitive even when the interest rate is hidden, and a treatment that emphasizes the interest rate does not have a significant effect on that sensitivity for the full sample. This suggests that clients are able to consider the price of the contracts even when it is not emphasized. Also, there is no evidence that changing the suggested payment plan (while holding the menu of options constant) affects payment plan take-up rates. That suggests that clients consider other options in the menu of payment plans when the suggested payment plan is not attractive. In 25

26 particular, many clients choose the shortest maturity plans, even when a payment plan with longer maturity/smaller monthly payments is suggested. However, the heterogeneous effects of prices and advertisement layout across risk categories demonstrate an important exception. High-risk clients are not sensitive to prices when the interest rate is hidden, but they do react to price changes when the interest rate is emphasized. Therefore, while information disclosure regulations might have little effect on low-risk (and possibly more sophisticated) consumers behavior, it can be relevant for high-risk consumers. Because high-risk borrowers are likely to be prone to making financial decision mistakes, and likely to face financial difficulties, my results suggest that information regulation in the consumer credit setting might be relevant. In addition, even though clients are able to consider payment plans that are not emphasized in the advertisement layout, as evidenced by the lack of response to changing the suggested payment plan on take-up rates, it seems that the suggested plan has a strong influence on the payment plan chosen among the menu of offers. In particular, most clients choose either the shortest payment plan available (even when a longer maturity plan with lower monthly payments is emphasized as a suggested plan) or the suggested plan. These results are consistent with clients using the suggested payment plan as a default option in order to avoid dealing with the decision about which payment plan to choose. Therefore, if a policy maker wants to induce consumers to pay off larger fractions of their credit card debts, then it might be sufficient to require credit card companies to offer standard 26

27 payment suggestions with higher than the minimum payments (in addition to the standard options of paying in full or paying the minimum payment), as was proposed in the Credit CARD Act of This effect is weaker (but still significant), however, when the stakes are higher. References Agarwal, S., P. M. Skiba, and J. Tobacman (2009, May). Payday loans and credit cards: New liquidity and credit scoring puzzles? American Economic Review 99 (2), Attanasio, O. P., P. K. Goldberg, and E. Kyriazidou (2008, 05). Credit constraints in the market for consumer durables: Evidence from micro data on car loans. International Economic Review 49 (2), Barr, M., S. Mullainathan, and E. Shafir (2008). Behaviorally informed financial services regulation. New America Foundation White Paper. Bertrand, M., D. Karlan, S. Mullainathan, E. Shafir, and J. Zinman (2010). What s advertising content worth? evidence from a consumer credit marketing field experiment*. Quarterly Journal of Economics 125 (1), Bertrand, M. and A. Morse (2009). Information Disclosure, Cognitive Biases and Payday Borrowing. SSRN elibrary. Campbell, J. Y., H. E. Jackson, B. C. Madrian, and P. Tufano (2011). Consumer financial protection. Journal of Economic Perspectives 25 (1), DellaVigna, S. (2009, June). Psychology and economics: Evidence from the field. Journal of Economic Literature 47 (2),

28 Gross, D. B. and N. S. Souleles (2002, February). Do liquidity constraints and interest rates matter for consumer behavior? evidence from credit card data. The Quarterly Journal of Economics 117 (1), Huang, D. and W. Tan (2009). Estimating the Demand for Credit Card: A Regression Discontinuity Approach. SSRN elibrary. Iyengar, S., G. Huberman, and W. Jiang (2004). How much choice is too much? Contributions to 401(k) retirement plans. Oxford University Press. Karlan, D. S. and J. Zinman (2008, June). Credit elasticities in less-developed economies: Implications for microfinance. American Economic Review 98 (3), Kroszner, R. (2007). Creating more effective consumer disclosures. Speech at George Washington University. May 23.. Shafir, E., I. Simonson, and A. Tversky (1993). Reason-based choice. Cognition 49, Simon, H. A. (1955, February). A Behavioral Model of Rational Choice. The Quarterly Journal of Economics 69 (1), Stango, V. and J. Zinman (2009, December). Exponential growth bias and household finance. Journal of Finance 64 (6), Stango, V. and J. Zinman (2011). Fuzzy math, disclosure regulation, and market outcomes: Evidence from truth-in-lending reform. Review of Financial Studies 24 (2),

29 Figure 1 Standard advertisement Layout (hidden interest rate) 29

30 Figure 2 Alternative advertisement Layout (emphasized interest rate) 30

31 Figure 3 Take-up rates by Interest rate x advertisement layout 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Interest Rate Graphs by Advertisement layout Full Sample 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Interest Rate Graphs by Advertisement layout Low-Risk Medium-Risk 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE High-Risk 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Graphs by Advertisement layout Interest Rate Graphs by Advertisement layout Interest Rate 31

32 Figure 4 Take-up rates by maturity suggestion x advertisement layout 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Maturity Suggestion Graphs by Advertisement layout Full Sample 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Maturity Suggestion Graphs by Advertisement layout Low Risk 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Maturity Suggestion Graphs by Advertisement layout Medium Risk 1. HIDDEN INTEREST RATE 2. EMPHASIZED INTEREST RATE Maturity Suggestion Graphs by Advertisement layout High Risk 32

33 Figure 5 Maturity choice by maturity suggestion Payment Plan Distribution By Maturity Sugestion Suggested Maturity=6 Suggested Maturity=10 Suggested Maturity=8 Suggested Maturity=12 33

34 ./012 )3)'( )3(#- A1=818D65=<08D16=8K121!D58L A1!8MN*++9++O381!=865D5<G5=80A58701!=16=8/680A !10<G581=G560<75?5!08 ;<==5! %&'!"#$ (%) 4! "8-9)): -3')-!"%$ %'#!"%$ %*( -3*)-!"*+$ %+*!"*+$ (%,!"*'$ %'&!"*'$ %+' 4! "8(9&): 4! "8**9%): -3-*' -3-''!"#$ %--!"#$ %++!"%$ %,,!"%$ %+- -3'')!"*+$ %'%!"*+$ %+(!"*'$ ()#!"*'$ %*)!"#$ ()(!"#$ %#(!"%$ %--!"%$ %+' -3-&*!"*+$ %#&!"*+$ %-&!"*'$ %'%!"*'$ %-%

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