One of the best-known propositions in textbook monetary economics is that concerning the long run neutrality of money, first shown by Sidrauski (1967). Yet there is growing empirical evidence that long-run changes in the level of inflation do in fact have real effects. A small increase in the rate of money growth in economies with initially low inflation rates is found to increase the long-run levels of capital stock (Kahn et al., 2006, Loayza, et al., 2000) and output (Bullard and Keating, 1995). In order to reconcile this apparent gap between traditional monetary theory and empirical evidence, a number of papers have re-evaluated the hypotheses under which long run inflation neutrality holds. Potential long-run real effects of monetary policy have been considered via inflation’s redistribution of seigniorage rents across households (Grandmont and Younès, 1973; and Kehoe et al., 1992) or across generations (Weiss, 1980; Weil, 1991), and inflation’s distortionary effect on capital taxation (Phelps, 1973 and Chari et al., 1996 among others) or labor supply (Den Haan, 1990).

This paper proposes a new channel for the non-neutrality of money transiting via borrowing constraints. If households can use both fiat money and capital to partially self-insure against individual income shocks, they may substitute away from real balances towards financial assets when inflation rises and the return to money falls. However, if there are asset market imperfections, borrowing-constrained households will not be able to undertake such portfolio adjustment and will adjust their money holdings differently compared to unconstrained households. Inflation thus triggers endogenous intra-period heterogeneity in money holdings when borrowing constraints are binding, providing incentives for unconstrained households with positive income shocks to increase their savings in order to smooth consumption between periods. Hence inflation may affect aggregate capital and output in the long run. Since the tightness of borrowing constraints is a well-established empirical fact (Jappelli 1990, Budria Rodriguez et al., 2002), this new channel may well account for a sizeable quantitative impact of inflation on the real economy and household welfare.