This paper explores the business cycle dynamics of nominal money growth, inflation, nominal and real interest rates and the velocity of money. Accounting for the observed relationships among these variables has proved to be difficult in a variety of monetary models, such as cash-in-advance models, models with sticky prices or with segmented markets (Hodrick, Kocherlakota, and Lucas (1991), Cooley and Hansen (1995), King and Watson (1996)). In particular, accounting for the negative correlation of inflation and nominal interest rates with nominal money growth, for the high volatility of money velocity and the weak correlation of inflation and nominal interest rates is still a challenge. Moreover, there are large and persistent deviations of the model-predicted money demand relationship from its counterpart in the data. Certainly, a monetary model, which can successfully account for these empirical observations, would raise the confidence in the conclusions drawn from policy experiments.
I show that it is possible to overcome these shortcomings in a model with strong liquidity effects (Increases in nominal interest rates decrease real money demand and increase real interest rates). I find that these liquidity effects imply that the estimated model can closely match the business cycle facts that standard models have not been able to replicate. An important assumption in the theoretical model is that households are hit by idiosyncratic preference shocks, which determine their demand for money in a model with cash and credit goods. This assumption generates a significant precautionary demand for money and I demonstrate that it also induces strong liquidity effects.
The literature suggests that the failure of standard models can indeed be traced back to the absence of substantial liquidity effects. First, without these liquidity effects, real money and real interest rates are almost invariant with respect to monetary policy. The Fisher equation – the nominal interest rate equals the real interest rate plus the inflation rate – then implies that inflation, money growth and nominal interest rates move almost one-for-one in response to the observed persistent changes in nominal interest rates. Liquidity effects break this linkage and can thus potentially reconcile the predictions of an economic model with the data. Second, the lack of a strong precautionary demand for money also explains why velocity is not volatile enough (Hodrick, Kocherlakota, and Lucas (1991)): households almost always hold the right amount of money to purchase the desired amount of cash goods. Changes in nominal interest rates therefore hardly affect the decision to acquire real balances, leaving velocity and real interest rates almost unchanged. Finally, rationalization of observed Federal Reserve monetary policy requires the existence of liquidity effects (see Ohanian and Stockman (1995)). Along the same lines, Alvarez, Lucas, and Weber (2001) argue that the observed practice of increasing nominal interest rates to lower inflation would contradict the quantity theory of money in the absence of substantial liquidity effects.
A crucial model feature which contributes to finding strong liquidity effects is the presence of a commercial bank (`a la Diamond and Dybvig (1983)). Households have to acquire money and bonds and make deposits before they learn their individual preference shock to cash goods. The bank diversifies this risk, providing partial insurance against this source of uncertainty. Households deposit money with the bank before they learn the realization of their shock, but can withdraw money after the shock realization. In equilibrium, house-holds are willing to accept a lower real return on their deposits if the bank allows them to withdraw more money ex-post. The bank therefore faces a trade-off between providing more real money, which is costly because of positive nominal interest rates, and paying a higher real return on deposits.
There are three reasons to add banking to the model. First, the bank’s simple trade-off leads to strong liquidity effects of monetary policy. In response to an increase in nominal interest rates, the bank provides households with less money but pays a higher real return on deposits, establishing the presence of liquidity effects. From a reduced-form perspective this mechanism renders the model equivalent to a model where both bonds and money enter the utility function. Changes in money then change the marginal value of holding bonds and not only the marginal utility of consumption as in models with only money in the utility function. I demonstrate that this modification of the utility function leads to strong liquidity effects. In particular, I show that the volume of liquidity and not just changes in liquidity (as in models with money in-the-utility) affect real interest rates. For example, an increase in the steady state level of real money leads to lower steady state real interest rates if the volume of liquidity matters, but has no effect if only first differences matter.
Second, the finding that households are willing to substitute a lower real return on their bank deposits for more money provided by the bank is useful for measuring households’ valuation of liquidity. This substitution effect reveals households’ precautionary demand for money and thus puts discipline on the choice of the unobservable idiosyncratic shocks. The substitution effect, together with the bank’s response to it, is consistent with the data as it implies a negative relationship between real deposit rates and real money. Third, adding a bank takes seriously the fact that households receive non-negligible interest rates on their deposits, instead of adopting the standard, but counterfactual, assumption that monetary aggregates (M1 or M2) are non-interest-bearing assets.
