Ebook Learning in the Credit Card Market

Economists believe that learning through experience underpins optimization and generates technological progress. Large literatures measure learning dynamics in the lab, and in the field.

However, because of data limitations, relatively few papers measure learning in the field with micro-level (household) panel data. Among such household studies, most show that households learn to optimize over time. For example, Miravete (2003) and Agarwal, Chomsisengphet, Liu and Souleles (2007) respectively show that consumers switch telephone calling plans and credit card contracts to minimize monthly bill payments.

Moreover, a few papers are able to identify the specific information flows that elicit learning. For instance, Fishman and Pope (2006) study video stores, and find that renters are more likely to return their videos on time if they have recently been fined for returning them late. Ho and Chong (2003) use grocery store scanner data to estimate a model in which consumers learn about product attributes. They find that the model has greater predictive power, with fewer parameters, than forecasting models used by retailers.

In this current paper, we study the process by which individual households learn to avoid add-on fees in the credit card market. We analyze a panel dataset that contains three years of credit card statements, representing 120,000 consumers and 4,000,000 credit card statements. We focus our analysis on credit card fees late payment, over limit, and cash advance fees since some observers argue that new customers do not optimally minimize such fees. We want to know whether credit card holders pay fewer fees with experience.

We find that fee payments are very large immediately after the opening of an account. We find that new accounts generate direct monthly fee payments that average $15 per month. However, these payments fall by 75 percent during the first four years of account life. To formally study these dynamics, we estimate a learning model with the Method of Simulated Moments. The data reveal that learning is driven by feedback. Making a late payment and consequently paying a fee reduces the probability of another late payment in the subsequent month by 44 percent.

These learning effects may be driven by many different channels. Consumers learn about fees when they are forced to pay them. Alternatively, consumers may pay more attention to their credit card account when they have recently paid fees. Through these many channels, card holders learn to sharply cut their fee payments over time.

We find that the learning dynamics are not monotonic. Card holders act as if their knowledge depreciates i.e., their learning patterns exhibit a recency effect. A late payment charge from the previous month is more influential than an identical charge that was paid a year ago. The monthly hazard rate of a fee payment increases as previous fee payments recede further into the past (holding all else equal). We estimate that this knowledge effectively at a rate of between 10 and 20 percent per month. At first glance, this finding seems counter intuitive. But there are actually several examples of papers that have found such forgetting effects. For instance, Benkard (2000) finds evidence for both learning and forgetting that is, depreciation of productivity over time in the manufacturing of aircraft, as do Argote, Beckman and Epple (1990), in shipbuilding.

Our findings imply that learning is very powerful, but that knowledge depreciation partially offsets learning. Nevertheless, the net effect of learning is clear. Learning generates a substantial net reduction in fee payments.

We organize our paper as follows, Section 2 summarizes our data and presents our basic evidence for learning and backsliding. Section 3 presents a model for those patterns. This model is estimated with the Method of Simulated Moments in section 4. Section 5 discusses alternative explanations for our findings. In Section 6, we draw some conclusions.