Department of Economics, UCSD

In the 2001 U.S. Survey of Consumer Finances (SCF), 27% of households report simultaneously revolving significant credit card debt and holding sizeable amounts of liquid assets. These consumers report paying, on average, a 14% interest rate on their debt, while earning only 1 or 2% on their liquid deposit accounts. This phenomenon is known as the “credit card debt puzzle”, as it appears to violate the standard no-arbitrage condition. In this paper, I quantitatively evaluate demand for liquidity as an explanation for this puzzle: households that accumulate credit card debt may not pay it off using their money in the bank, because they expect to use that money in situations where credit cards cannot be used. Using both aggregate and survey data (SCF and CEX), I document that liquid assets are a substantial part of households’ portfolios and that consumption in goods requiring liquid payments appears to have a sizeable unpredictable component. This would warrant holding positive balances in liquid accounts both for transactions and precautionary purposes. I develop a dynamic heterogeneous-agent model of household portfolio choice, where households are subject to uninsurable income and preference uncertainty, and consumer credit and liquidity coexist as means of consumption and saving/borrowing. The calibration of the model parameters is based on the simulated method of moments. The calibrated model accounts for between 85% and 104% of the households in the data who hold consumer debt and liquidity simultaneously, and for between 56 and 62 cents of every dollar held by a median household in the puzzle group. Thus the transactions and precautionary demand for liquidity appear to be a significant factor in accounting for the credit card debt puzzle.