With three essays, this dissertation investigates risk and profitability of consumer credit cards. In the first essay (Chapter 2), I look at the customer profitability within the affinity program context. Specifically, I investigate whether affinity cardholders generate more profit for the card issuer. Using propensity score matching to control for the selection bias, I show that, contrary to conventional wisdom, affinity card customers are no more profitable than non-affinity cardholders. I also examine profitability for different types of affinity cards and show that sports-based programs are the least profitable type of affinity cards. In the second essay (Chapter 3), I analyze banks and colleges' affinity program contracts to determine how they reach to final contract terms considering the bargaining power of both parties. I use a generalized Nash bargaining solution to model the equilibrium conditions and the optimal split of contract surplus between bank and college. I also collect data on more than 300 affinity program contracts between banks and colleges and universities to empirically test the hypotheses resulted from the theoretical model. I use aggregate discrete-choice methodology to model the card adoption behavior of students and alumni and estimate the demand for affinity card. The supply side equation also results from the equilibrium model. Through simultaneous estimation of demand and supply models, I empirically determine the effect of bank and college bargaining power factors on the affinity contract terms. In the third essay (Chapter 4), I focus on the impact of consumer repayment and spending behavior on his/her credit risk. I develop a hidden Markov model to dynamically estimate the unobserved risk state of customers and the transitions between the states. I find that higher risk states can be predicted based on spending in discount stores and pawn shops, excessive spending on entertainment such as gambling and dating services. In contrast, spending on necessity items such as gas and groceries is associated with lower risk. I also use cluster analysis to group customers based on their risk evolution pattern and find four groups of customers based on their initial and final risk states. I show that each cluster has its distinct purchase categories and that potential profitability and risk are considerably different across the four clusters.