ABSTRACT
We adopt a hedonic regression approach to capture the importance of the specific risk of cardholders (demand side) and credit card attributes (supply side) in determining the price of credit card plans. The specification of the empirical model is as follows:
APRit = α + γ xriskit + βxcc_attributesit + φxmarketit + fi + Tt + εit . (1) APR is a vector of the most common interest rate charged by a specific credit card i in year t. xriskit accounts for six different indicators of the risk of cardholders (three standard risk measures comprising FICO credit score, outstanding credit card debt, unpaid debt in the previous 12 months, and three additional risk indicators computed from a combination of the standard indicators, explained in Section 4) living in the regions where the credit card plan i is offered. By using each risk indicator, we are able to estimate up to six different empirical models. xcc_attributesit is a vector of credit card attributes such as regional coverage, type of credit card provider, grace period, reward programmes, fees, network brand, platinum, gold, student, classic card, etc. xmarketit includes two control variables, comprising the one-year CD interest rate and the deposit market share of the bank offering credit card plan i. The model also includes issuer-fixed effects (fi) to capture unobserved issuer characteristics that are constant over time, a time dummy for technological change (Tt), and a constant term α (which captures an interest rate mark-up). εit is a stochastic error term.
