《宏观经济学 Macroeconomics》课外读物：Money, Credit, and Prices in a Real Business Cycle.pdf

Money, Credit, and Prices in a Real Business Cycle By robert G. KING AND CHARLES I PLOSSER* An important recent strain of macroeco- the analysis is on the banking system, build- nomic theory views business cycles as arising ing on the earlier work of James Tobin(1963) from variations in the real opportunities of and Eugene Fama(1980). In our real busi the private economy, which may include ness cycle model, monetary services are shifts in government purchases or tax rates privately produced intermediate goods whose as well as technical and environmental con- quantities rise and fall with real economic ditions. These models are often viewed as developments incomplete or wrong because they do not In the absence of central bank policy re generate the widely emphasized, but not sponse, the model predicts that movemen easily explained, correlation between the in external money measures should be uncor quantity of money and real activity related with real activity. Some preliminary This paper integrates money and banking empirical analysis (using annual data from into real business cycle theory. The result is a 1953 to 1978) provides general support for class of models that can account for the our focus on the banking system since the correlation between money and business correlation between monetary measures and cycles in terms that most economists would real activity is primarily with inside money. label reverse causation. The main focus of Our proposed explanation of the corre lation between money and business fluctua- Department of Economics and Graduate School of tions stands in sharp contrast to traditional theories that stress market failure as the ke Rochester, NY 14627. A preliminary version of this to understanding the relation and interpret paper was presented at the Seminar on Monetary The- monetary movements as a primary source of ory and Monetary Policy, Konstanz, West Germany, impulses to real activity. Given the con- Robert barro, Herschel Grossman, John Long, benne troversies surrounding the main contending McCallum, anonymous referees, and participants of hypotheses concerning money and busines cycles-the incomplete information frame- Pennsylvania, Harvard, and Princeton. The National work of Robert Lucas( 1973)and Keynesian Government Policy and Business of the University of Sticky wage models as revitalized by Stanley Rochester have supported this research. The above indi- Fischer(1977)-it seems worthwhile to con viduals and institutions should not be regarded as neces- sider alternative hypotheses. 3 sarily endorsing the views expressed in this Robert Lucas(1980)provides an ov general equilibrium approach to business cycles work by Fynn Kydland and Edward Prescott (1982 )and tions in money demand including those arising from g and Plosser(1983)illustrate ns in real activity = an mimic key elements of business cycles. in In our view, there are good reasons for dissatisfac tterns of persisten with existing macro c ea that monetary quantities are endogenous is ities, from the textbook reliance on exogenous values to models typically rely on implausible wage or price rigid old one. but has received little the recent more sophisticated effort of Fischer (197 find it useful to categorize earlier stories into two broad that relies on existing nominal contracts. As Barro(1977) points out, a key feature of the explanations that stress central bank policy odel is that agents select contracts that do not I fully or example, James Tobin (1970) provides an ploit potential gains from trade. In addition, Costas em shasizes central bank policy response. Tobins determin- ment contract perceived monetary dis- tic treatment involves the turbances do not Recent analyses of monetary nonneutrality that stress ence-aggregate demand. In Fischer Black's(1972) expectation errors based on "imperfect information analysis, external money passively responds to all varia- ( Lucas, 1977, provides a summary of this viewpoint 363

364 THE AMERICAN ECONOMIC REVIEW JUNE 1984 The organization of the paper is as fol- the output of the financial-banking indu lows In Section I we describe a simple model is an input into production and purchase of that is capable of generating real business final goods. This view is consistent with the cycles. The model is used to discuss corre- general focus on prod duced inputs and sec- lations between an internal monetary quan- toral interactions that is the hallmark of real tity and real activity. In Section II, with fiat business cycle model money included in the model, we analyze the elation between monetary quantities, out A. Final Goods Industry ut,and the price level in both an unregu lated and regulated banking environment. In The single final product (y)is produced Section Ill we discuss some of the empirical by a constant returns to scale production plications of the theory and provide a process that uses labor(n), capital(k), and preliminary analysis of the postwar U.S. ex- transaction services(d)as inputs. The pro- duction technology is summarized by I. The Real economy In this section we describe a simple model where k, is the amount of capital, nyr Is the economy in which business cycles arise as a amount of labor services, and w in the onsequence of the intertemporal optimizing amount of transaction services used in the ehavior of economic agents. Our model has final goods industry. Capital services are two productive sectors with one intermediate measured in commodity units llocated to and one final good. The output of the final production at time t, labor services are hours goods industry is stochastic and serves as worked, and transaction services can be ther a consumption good or as an input viewed as the number of bookkeeping entries into future production. The output of the made( described more fully below ) We also financial industry is an intermediate good make the standard assumptions of positive called transaction services that is used by and diminishing marginal products to each services arises because these services econo- dated by the time of their realization ct to firms in the final goods industry and by y factor. The production process is subj households. The demand for transaction two random shocks, p, and 5,+1 mize on time and other resources required to Transaction services in (1)are viewed as accomplish hange of goods. an intermediate good purchased by final good Recent real general equilibrium theories of producers from the financial industry (to be the business cycle(such as Finn Kydland described below ). Although not involved di- stress produced inputs and interrelations be- same sense as labor and capital, trans tween sectors as central to understanding the services are part of a cost-reducing ac persistence and comovement of macroeco. similar to other organizational and co nomic time-series. The simple model econo- inputs ny that we study has only one final product The sequences() and(5) are assumed and thus does not possess such a rich set of to be strictly positive stationary stochastic dynamics or sectoral interactions. Neverthe- processes that are mutually and serially inde less, the framework embodies our view that pendent with E(D)=E(5)=1. The roles played by the two shocks are quite different similarly rely on an apparent failure in the market for At this point it is sufficient to note that p information. For example, information on monetary alters expected time t +l output and affects tatistics is cheap and readily available. King(1981) time t input decisions by altering intertem demonstrates that in Lucas'(1973)model, real outp poral opportunities. On the other hand, 5, ole monetary information. John boschen and herschel represents the basic uncertainty of the pro- rossman(1982)empirically investigate this proposition duction process by altering output in an nd find that it is rejected by the data. unexpected manner. The multiplicative na

VOL,74 NO. 3 KING AND PLOSSER: MONEY CREDIT AND PRICE 365 ture of the randomness in total production financial industry has a supply curve that is implies a technological neutrality of the horizontal at a particular rental price, P* shocks with respect to individual factors of Although at this stage of our analysis we production. Alternatively, different stochas- focus on the flow of transaction services, the tic elements could be associated with particu- transaction(banking)industry typically(but lar factors not necessarily) provides these services in Production is assumed to be under super- conjunction with portfolio management or ision of identical competitive firms. Firms intermediary services. It is convenient te by selling claims against the future imagine, therefore, that the financial industry output and using the proceeds to purchase holds claims(shares)on the probability dis- factors of production. Labor, capital, and tribution of output and issues other claims transaction services are rented at rental prices (deposits). In the process of market ex- qr, and p,, respectively. Each firm is change, the claims that individuals and firms assumed to sell one unit of claim for each hold on the bank's portfolio(deposits)are unit of expected output as determined by altered through simple bookkeeping entries f(kw, nun, du), which amounts to defining a Banks pass on to depositors the return to the share" in the firm. If the market price of portfolio of assets less a fee for transactio claims is u, the firm faces a static maximiza- services tion problem involving the choice of inputs The structure of the financial industry im- hat maximizes profits, U(kyn, dur) lies that the direct cost of bookkeeping w,nyr-q, kyr-p,dyr. The assumption of con- services, P,, does not depend on the char- stant returns to scale implies that the firm acter or composition of the bank's portfolio the price u*, corresponding to minimum unit there is no reason to expect homoge has a supply of claims that is horizontal As discussed by Fama(1980), it follows tha cost at prices qr, w, and p deposits in an unregulated financial indust More generally, this conclusion holds so long B. Financial Industry as the respective portfolio costs and transac tion services are borne by portfolio holders The financial industry provides accounting and transaction users ervices that facilitate the exchange of goods by reducing the amount of time and other resources that otherwise would be devoted to market transactions. The production of this The individual households in the model intermediate good which we call transaction e consumers, suppliers of labor services ervices, is summarized by the production and capital goods, purchasers of transaction function(2)in which nd, and kdr are the services, and ultimate wealth holders. The amounts of labor and capital allocated to the representative individual is assumed to be financial sector: infinite lived and possess the intertemporal utility function h ∑B(x,+, This instantaneous production structure em odies the hypothesis that production of where B is a fixed utility discount factor and ransaction services requires less time than u( is a single period utility function that production of the consumption-capital good. depends on consumption(x,+) and leisure Technological innovation in this industry is (n-ni+)with n indicating the total hours captured by Ar which is assumed to be a available in each period. The utility maxi- strictly positive stochastic process with a mand is the expected utility measure E, U mean of one. Finally, we assume(2)repre- where et denotes the conditional (rational) sents a constant returns to scale structure so expectation based on all information avail that, at given factor prices w, and qu, the able at time 4

THE AMERICAN ECONOMIC REVIEW JUNE 1984 The representative agent arrives at date t D. Equilibrium Prices and t th total wealth equal to the sum of current realized output(y)and the depreciated value Analysis of dynamic, stochastic general of the previous period's capital stock(k equilibrium models is a difficult task. One Sk-1). The agents current decisions involve strategy for characterizing equilibrium prices the selection of the levels of consumption and quantities is to study the planning prob- (x, )and of total effort (n, )as well as alloca- lem for a representative agent(see Lucas tion of effort to market and nonmarket activ- 1978, or Long and Plosser). This procedure ities. These decisions imply a level of saving is valid so long as the competitive equi that then must be efficiently allocated, along librium is Pareto optimal. The planning with current wealth, to purchases of invest- problem can also be used to generate specific ment goods (i,)and financial assets (for equilibria if explicit functional forms for example, real bonds, shares, etc. preferences and technologies are assume Households are assumed to combine time We do not pursue this strategy in detail as and transactions services to accomplish pur- our objective is more modest. Instead, w chases of consumption and investment goods. make a number of simplifying assumptions In particular, the time required for this non- regarding the general framework proposed market activity is above that allow us to highlight the condi (4)n21=r(dn:/(x1+1)(x1+i), tions necessary to obtain certain business cycle comovements in general equilibrium The state of the economy at date t is where T'<0,T"<0. Our individual selects summarized by the values of four variables an amount of transactions services dht so as y,(1-8)k-1,P,, and A,. The first is a measure of national income the second is + p dhr. So long as hours are freely variable, the current stock of depreciated capital, , is w, is the opportunity cost of effort, and a technical factor affecting current opportu this minimization problem can be treated nities to transfer resources intertemporally separately from the household,'s general and A, is a technical factor influencing the allocations.( However, efficiently selected production of transaction services.The ansactions patterns will have wealth and agents vector of decisions variables is(nyr substitution effects on desirable household ndr, n,, dn, dyn,ky, kd) allocations.) In order to simplify the problem, we make Minimizing the total cost of transactions three assumptions that are sufficient to re- activities implies a derived demand for duce the state vector to two elements and the purchases of transaction services of the form decision vector to two elements while pre- dhi=(P/w)(x,+i,), where g=(T")I< serving the essential features of the model 0. Similarly, hours allocated to transactions First, we assume a depreciation rate of 100 activities are proportional to expenditures, percent, eliminating(1-8)k-I as a state taking the form n*=T(g(p,/w))(x,+iD) variable. Second, we assume that transaction The presence of transaction costs for the services are produced determi urchase of consumption and investment 1, for all () and depend only on labor input goods implies that the total cost of a unit of (d, =hon dt ) Deterministic production of 1+[w,T(g(p /x greater than unity (i.e. variable and the simplified production tech- ection nology implies that the competitive price is of an optimal pattern of consumption(x,), p*=w, ho. This implies that households(and folio allocations involves the usual sort of tion services in fixed proportiong d transac- total effort (n, =nur+nar +nr), and port- firms below)use time and purcha intertemporal efficiency conditions with the The third assumption is to restrict the final exception of this modification. Fischer(1982) goods production function to employ finan- provides an interpretation of the altered cial services in a manner that is symmetric to efficiency conditions in a similar context