Liquidity, transaction costs, and real activity.

• Author: Zhang, Junxi

1. Introduction

   In recent years, there has been growing interest in studying the effects of unanticipated monetary shocks. It is well understood that such effects are mingled by two opposing forces, known as a liquidity effect and an anticipated inflation effect. The former suggests that the extra injected money pushes down interest rates and stimulates economic activity, while the latter suggests that the extra injected money could raise the public's expectations on inflation rates and eventually push interest rates up, which may depress economic activity. The widespread view among economists is that the liquidity effect is stronger than the anticipated inflation effect, at least in the short run.

   To date, there exist, by and large, two distinct strands of research in the literature: One investigates empirically whether a dominant liquidity effect can be found in the data, and the other, by assuming that a liquidity effect does exist, seeks to incorporate it into general equilibrium models. As for the empirical study, economists appear to have a general consensus (e.g., see the recent discussions in Cook and Hahn [1989], Romer and Romer [1989], and King [1991]). As for the second line, theoretical work to capture such a dominance in general equilibrium models did not emerge until recently (see Lucas 1990; Christiano 1991; Fuerst 1992, among others).(1) Although these models have shown some promise of success, questions remain as to whether or how general this result is with respect to alternative models and behavioral assumptions.

   The difficulty in understanding the economic impact of a money growth shock is that it does not impact equally on all economic agents, as noted by many authors (Grossman and Weiss 1983; Rotemberg 1984); that is, in order to induce a stronger liquidity effect, there must exist an asymmetry of monetary injections. In most of the existing models, where financial intermediation is endogenous, the kind of asymmetry is usually embedded in a number of behavioral assumptions, such as that agents' loan decisions or firms' investment decisions are made prior to when the state of the economy is observed. In addition, these studies adopt a standard practice of modeling the role of money by introducing a cash-in-advance (CIA) constraint that limits the amount of money available for use in loan or securities markets. Both features are shown to be important for a dominant liquidity effect.

   Lately, the assumption of CIA constraints is under scrutiny. Dotsey and Ireland (1995) argue that this specification places infinite transaction costs on flows of funds across segmented markets. While it might be appropriate for investigating the behavior of asset prices on a daily or weekly basis, it is certainly not innocuous to studying the effects of monetary policy at business cycle frequencies. While retaining the same degree of asymmetry, they relax the CIA constraint and allow agents to rearrange their portfolios at a finite cost after observing the monetary disturbance and conclude that the liquidity effect disappears.(2) As a result, their finding casts doubt on the usefulness of this class of models for studying the liquidity effect at business cycle frequencies.

   The purpose of this paper is thus modest and limited. It is intended to demonstrate that the liquidity effect can be displayed in a similar general equilibrium model if alternative monetary specifications and behavioral assumptions are made. For the monetary specification, we follow Sims (1989) to present a pecuniary transaction-cost model that takes the transmission of economic fluctuations into consideration.(3) Unlike conventional transaction-cost models, which adopt a specific shopping-time technology (e.g., Saving 1971), the model proposed below specifies a transaction-cost function that expresses the cost as simply a proportion of real output. This set-up of money demand recognizes the fact that people demand real money balances not only as part of their wealth holdings but also as a means of payment that reduces transaction costs. The channel through which monetary injections produce real effects here is via affecting transaction costs that individuals engage in costly activities.

   To retain a necessary channel of asymmetry regarding the response to monetary injections, we first modify the basic transaction cost model by making an alternative behavioral assumption - sluggish money demand: Households cannot continuously revise their decisions for holding money each period.(4) It seems intuitive why this modification should induce the needed stronger liquidity effect. With ex ante money demand decisions, an unanticipated monetary injection will create an excess supply of money. The extra money then exerts a downward pressure on the nominal rate of interest, which in turn creates an upward pressure on economic activity. Under plausible parameter values, it is shown that this is indeed the case.

   As a confirmation on the model's ability to generate the dominant liquidity effect, another natural modification proposed by Christiano (1991) is investigated - sluggish firm investment: Firms' investment decisions cannot be revised continuously and hence they do not respond instantaneously to a money shock. Once again, we find that the liquidity effect swamps the anticipated inflation effect. However, as documented in related studies (e.g., see Hansen [1985] for a nonmonetary model and Christiano [1991] for a monetary model), the monetary model used here remains unable to account well for some features of the U.S. business cycles, notably the volatility of labor effort and the nominal interest rate.

   The rest of the paper is organized as follows. Section 2 presents the basic transaction-cost model in order to investigate its ability to account for the dominant liquidity effect, while section 3 modifies the basic model by considering two variants. A brief discussion of the solution method and procedure is also given. In section 4, we report the empirical findings for the three models. Finally, section 5 offers some concluding remarks and possible extensions.

2. A Basic Transaction-Cost Model

   The model economy is populated by a finite number of infinitely lived, identical, and perfectly competitive households. Households supply labor and purchase the output of the firms, which is limited to a single storable good in this model. Households also own three paper assets in the economy: money, bonds, and equities.

   Money, of which the per capita quantity is denoted by [M.sub.t], is a liquid paper asset used as a medium of exchange. It is issued by the government, and monetary policies are conducted through open market operations. The second asset is a variable-coupon bond, of which the per capita quantity is denoted by [B.sub.t], is also issued by the government, and the expected nominal rate of return associated with holding this asset is [R.sub.t]. The third paper asset consists of equities, which are issued by firms to finance investment. In fact, equities are the financial counterpart of physical capital. For simplicity, firms presumably issue no bonds and retain no earnings, so all investment is financed by issuing equities. Households hold, buy, and sell equities, and firms are prohibited from trading equities. Since there are no risks involved here, households regard equities and bonds as perfect substitutes, thereby leading to the same expected yields in equilibrium.

   At period zero, a typical household seeks to maximize the expected discounted value of its lifetime utility, described by

   [E.sub.0] [summation of] ([Beta])[prime] U([C.sub.t], [L.sub.t]) where t = 0 to [infinity] (1)

   where [C.sub.t] denotes consumption of the market produced good, [L.sub.t] denotes hours worked (0 [less than or equal to] [L.sub.t] [less than or equal to] 1), [E.sub.09] is an expectation operator conditional on all information available in...