This paper develops an analytically tractable Bewley model of money featuring capital and financial intermediation. It is shown that when money is a vital form of liquidity to meet uncertain consumption needs, the welfare costs of inflation can be extremely large. With log utility and parameter values that best match both the aggregate money demand curve suggested by Lucas (2000) and the variance of household consumption, agents in our model are willing to reduce consumption by 7% ~ 10% (or more) to avoid 10% annual inflation. In other words, raising the U.S. inflation target from 2% to 3% amounts to roughly a 0:5 percentage reduction in aggregate consumption. The astonishingly large welfare costs of inflation arise because inflation tightens liquidity constraints by destroying the buffer-stock value of money, thus raising the volatility of consumption at the household level. Such an inflation-induced increase in the idiosyncratic consumption-volatility at the micro level cannot be captured by representative- agent models or the Bailey triangle. Although the development of a credit and banking system can reduce the welfare costs of inflation by alleviating liquidity constraints, with realistic credit limits the cost of moderate inflation still remains several times larger than estimations based on the Bailey triangle. Our finding not only provides a justification for adopting a low inflation target by central banks, but also offers a plausible explanation for the robust positive relationship between inflation and social unrest in developing countries where money is the major form of household financial wealth.
