Debit cards have surpassed credit cards to become the most common form of Visa point-of-sale (“POS”) transaction in the United States (Visa 2002). Overall, debit was used for over 15.5 billion POS transactions totaling $700 billion in the year 2002 (CPSS 2003). This represented about 35% of electronic payment transaction volume and 12% of POS noncash payments (Gerdes and Walton 2002). Debit’s ascension has been sudden, with 47% of households using it by 2001, up from 18% in 1995 (Table 1). Industry observers predict continued strong growth for debit, while forecasting relatively weak growth in credit card charge volume.
Despite debit’s growth and prominence, the determinants of debit use have largely escaped academic scrutiny. The introductory quotes belie that fact that there are actually potentially important, pecuniary cost-based reasons for using debit. Principally, the 53% of credit card users who revolve balances incur interest costs to charge purchases on the margin (i.e., they don’t get the float), and hence might rationally choose to use debit rather than credit in order to minimize transaction costs.This motive holds even for the “small” (Laibson et al. 2003) fraction of consumers who simultaneously hold nontrivial stocks of low-yielding liquid assets and expensive credit card debt. Debit use might also be rational for consumers lacking access to a credit card or facing a binding credit limit.
But perhaps 29% of debit users lack any price-based reason for doing so (Table 2). Accordingly we should take seriously the popular notion that debit use serves as a form of commitment device against the type of “overspending” with credit cards posited by Ausubel (1991), Prelec and Simester (2001), and Bertaut and Haliassos (2002).Such hypotheses seem plausible in part because it would be relatively cheap to use debit in this fashion— I show below that the pecuniary cost of “incorrectly” choosing debit rather than credit is perhaps $12 per month. This cost is comparable, in present value terms, to the estimated $2,000 that consumers with sophisticated quasi-hyperbolic preferences should pay to commit themselves not to borrow on credit cards (Laibson et al. 2004). Other stylized facts also fuel the intuition that one or more “behavioral” explanations drive debit use. Most provocative is that debit tends to be used for smaller transactions involving instantaneous consumption, with credit cards used to purchase larger, more durable items (Reda 2003)— a pattern consistent with the mental accounting model in Prelec and Loewenstein (1998). Yet neither behavioral nor more traditional explanations for debit use have been put to the test.
Identifying the correct model(s) of debit use has implications for high-frequency intertemporal consumer choice more generally. Validation of the puzzle outlined in Table 2 would add to the growing list of consumer behaviors in financial markets that that have proven difficult to explain with straightforward applications of canonical models.More specifically, validation of a spending control motive for debit use that operates via mental accounting would bear on the existence and welfare implications of time-inconsistent preferences. On the other hand, evidence that consumers respond strongly to the pecuniary marginal cost of payments, in the face of small stakes, would be compelling support for traditional rationality Miravete 2003).
The primitives of the debit vs. credit choice also have implications for modeling and regulating the industrial organization of payments networks. Specifically, a growing theoretical literature finds that the relative efficiency of alternative pricing practices, merchant acceptance rules, and governance arrangements depends critically on the elasticity of consumer demand for payments services (Chakravorti 2003).
Accordingly this paper puts competing consumer choice models of debit use to the test. It starts by developing a standard (“canonical”) consumer choice model that focuses on consumer sensitivity to the (implicit) relative price of electronic payments at the POS. The lack of consumer-level data on explicit transaction fees is not much of a constraint, as I show that the first-order theoretical determinant of relative payments price is often whether the consumer has been revolving balances on her credit card and hence must “borrow-to-charge”. One thus can use widely available household data to test whether consumer behavior is consistent with joint optimization over payment options. Specifically, a canonical model of consumer choice generates the following testable predictions: 1. consumers who revolve credit card balances (“revolvers”) should be more likely to use debit than those who don’t (“convenience users”); 2. revolvers facing binding credit constraints should be more likely to use debit than revolvers who don’t; 3. convenience users should be less likely to use debit than those without credit cards (since they should exploit the float and minimize low-yielding transaction balances). I find statistically and economically significant support for these predictions, and show that the observed empirical relationships are likely driven by causal effects of the credit card variables of interest on debit use.
Download
PDF Ebook Why Use Debit Instead of Credit? Consumer Choice in a Trillion-Dollar Market
