Knowledge that Transforms

is asserted to be reducible to the question of whether a key reduced-form coefficient P5 iS positive or not. I argue that Stein has failed to work out fully the dynamic implications of his model. It has led him to the incorrect proposition that a positive P5 coefficient implies a permanent positive effect of fiscal policy on the equilibrium rate of growth of nominal income. This comment shows that the correct conclusion is that fiscal policy will, with the exception of a very special case where Stein's result holds, produce over the long run a rate of growth of nominal income that either attains an equilibrium value equal to the rate of growth of the

In a recent article in this Review (1981), James Buchanan and Robert Tollison used a truncated neoclassical model to deduce the allocational effects of forced equal-pay schemes, including equal pay for (a) workers of the same trade, (b) workers in the same firm or industry, (c) workers who perform the same work, and (d) the minimum wage. The model that Buchanan and Tollison used is one in which firms hire a single unit of each type of input, each may be hired in competitive markets where their prices differ, each has the same marginal physical product within the firm, and there is within limits among inputs of different types. They also assume only human inputs, since they are interested in equal-pay schemes. The purpose of this comment is to show that, depending on their meaning of substitution within limits, their model is either inconsistent or incorrectly specified. The key to understanding the BuchananTollison model appears to lie with identifying a production function that corresponds to substitution within limits. The early masters of marginal productivity theory recognized that two separate assumptions could be made about production coefficients. They could be assumed fixed or variable. The assumption of fixed coefficients meant that there could be no among inputs in a production process. The assumption of variable coefficients was used to illustrate cases in which a change in the relative prices of inputs would prompt entrepreneurs to alter their production methods in order to substitute inputs whose relative prices had fallen for those whose relative prices had risen. If one assumed variable coefficients, the equilibrium conditions and the various marginal equalities could be easily traced. If one assumed fixed coefficients, the marginal conditions could not be expressed. As a substitute for marginal productivity, however, Marshall and others introduced the concept of net productivity. This was not satisfactory to mathematical purists. Moreover, it left the door open to the possibility of haggling over a surplus; since product exhaustion would not occur even at the margin in the typical case. The assumption of fixed coefficients has sometimes been expressed in terms of lumpiness or indivisibility.' It seems clear that Buchanan and Tollison do not use the term within limits to mean variable coefficients. Suppose that, by chance, a firm that faced variable coefficients hired only one unit of each input of different types and that these inputs had identical marginal value products. Suppose further that the firm, industry, and economy were in general equilibrium. Now let there be a general reduction in the wages of one type of input, perhaps as a consequence of a preference by owners of that input for less leisure. The relative price of the input would fall. In neoclassical theory, the firm would attempt to substitute more of this input for other inputs and would alter its production methods to do so. In Buchanan-Tollison's model, however, this is impossible by assumption, since only one unit of each input is hired. One must conclude that within limits does not mean variable coefficients. Now consider whether they mean fixed production coefficients. Although this appears to be their meaning, it will be shown that it is inconsistent with their analytical framework. Consider their graphical representation of a firm, which provides the reference for their verbal discussion. The relevant parts are reproduced in Figure 1. Note that inputs of different types are arranged in order of ascending supply prices. An input of one

In a recent issue of this Review, Jeffrey Sachs considers the early history of inflation. In the first section of the paper he concludes: evidence is rather striking. For mild contractions, downward price flexibility seems to have ended with the pre-World War II period. For moderate and severe contractions, similarly, the response of wages and prices have fallen significantly since 1950 (p. 81). It is the contention here that those conclusions follow from the choice of a measure of price inflation. There is also a data error that slightly weakens the conclusion. My last point is that Sachs overlooked a conclusion that is pertinent to the choice of time period for his estimation of Phillips curves. The data error is first. Sachs classifies the business cycle downturn of 1918-19 as a strong contraction (Table 2, p. 81). Given that, it would be convenient to have the annual percentage change of the BLS Wholesale Price Index from peak to trough for 1918-19 be -5.5 as stated in his Table 1. Unfortunately, it is not so. The error is the sign. The observation for 1918-19 for change in price inflation in his Table 2 becomes -5.4 instead of 16.4. My second point is the choice of the measure of the general price level. Sachs writes: empirical displayed in this section are for the WPI and indexes of average hourly compensation. The calculations have also been made for a number of other price and wage series, with very similar results (p. 79).' Why choose the Wholesale Price Index? It is now quite out of style as a measure of the general price level. There is some difficulty with choosing a Consumer Price Index because there is no one series that covers the entire period. The obvious alternative to have used is the Implicit Deflator. The question is the significance of the choice. Table 1 duplicates Sachs' Table 1, but uses the Implicit Deflator rather than the Wholesale Price Index as the measure of price inflation. Sachs concludes from his table: For the WPI, the rate of inflation declined during every downturn from 1890 to and for three of the five recessions after 1948, and Almost every contraction from 1890 to 1927 produced a sharper deceleration in price change than did later recessions. Only the deceleration in 1949 is of similar magnitude with the earlier cycles (p. 80). Those statements must now be altered. The rate of inflation declined in seven of eleven downturns from 1890 to 1948. In one of these, the decline in the rate of inflation is less than one percentage point. The result is not altered for the six recessions after 1948.2 The choice of the Implicit Deflator causes the pattern to become much more complex and the 1949 deceleration much more conspicu-

In recent years, implicit theory has grown following by Martin Baily (1974) and Costas Azariadis (1975). A major concern in much of this literature has been problem of enforceability of implicit contracts. The problem is that for any to trade labor services, or some other good at some future time t, there will always be motivation for one of contracting parties to breach whenever future spot price at t deviates from contractual price. In absence of some formal enforcement mechanism (i.e., courts), or informal mechanism (such as concern for reputation or front-end loading), contracts would never be fulfilled. In an important contribution to this literature, Clive Bull (1983) proposed a model under which implicit contracts would become partially enforceable due to what might be called a package contract for two distinct labor services, which he called labor and Bull argued that inability to trade effort separately due to nonexistence of a market for such would lead to implicit contracts that were partially enforceable (i.e., in at least some states of world). His conclusion was that the very aspect of economy that gives rise to implicit contracts [is]. ... absence of a complete set of (p. 668). It is theme of this note that absence of markets for some aspects of labor is unnecessary in order to achieve Bull results of partial enforceability. While such enforceability is possible when some labor markets are missing, it is no less possible when labor markets are complete (although contingent claims markets must remain incomplete). To see this, let us follow Bull and imagine that a worker is supplying two distinct labor services, which will be called work and effort. Each of these may be marketed separately to different purchasers, or two may be marketed jointly in a multiple contract.1 Now future prices of (or Xt) and of effort (or Y,) are unknown before t, and no contingent claims or enforceable forward markets for them exist. In absence of formal legal enforcement or informal enforcement (such as concern for reputation), no for supplying either labor service separately would be fulfilled. A to supply at price Xt would be broken by workers as soon as future spot price rose above that, and would be breached by employers as soon as it fell below Xt. The same would hold for separate contracts to purchase But what about joint or multiple contracts to supply both and effort simultaneously? Let indirect utility function of worker be i3tU(Xt, Yt), where Pt is discount factor for future period t. Let indirect utility function of employer or purchaser be /3tW(Xt, Yt). For both U and W functions, first partials with respect to both arguments are positive and second partials are negative. Both U and W are continuous, differentiable functions and Inada conditions hold. Let us assume there is a to jointly trade and effort at promised prices of

Robert Eisner has recently entered the debate on the relative potency of monetary and fiscal actions. He demonstrates the ineffectiveness of the 1968 tax surcharge in checking inflation, then goes on to assert that tight money would be similarly ineffective. This paper considers Eisner's analysis as it pertains to the inadequacy of monetary policy. First, it is shown that his conclusions do not necessarily follow from his own model. Second, using parameter estimates representative of other studies, it is demonstrated that Eisner's conclusions are not substantiated by the empirical evidence.

This study was conducted to evaluate George Kaufman's extension of the Friedman and Meiselman technique for an empirical definition of money. This method defines as money that financial aggregate which satisfies two criteria: 1) it exhibits the highest correlation with GNP, and 2) the correlations between GNP and each of the components considered separately do not exceed that between GNP and the aggregate. The components are thus substitutes the public alters the composition of its portfolio due to changes in supply conditions, while keeping its portfolio size constant relative to GNP. (See Friedman and Schwartz, ch. 2; and J. R. Hicks, p. 49.) The set of assets heretofore considered include liquid financial assets. Friedman and Meiselman discovered that the dual criteria were best satisfied by the sum of currency and all privately held deposits at commercial banks (pp. 182-84). Kaufman examined the proposition that if money is a factor in determining GNP, its effect may be delayed by as long as a year. From correlations between GNP and various financial aggregates which led GNP by +4 to -2 quarters, he found that the best definition of money depends on the number of quarters by which the financial measure leads or lags GNP. In general, the broader aggregates perform better when they are observed two or more quarters before income while the narrow definition performs best when observed concurrently with income (see Kaufman, pp. 86-87, Tables 1 and 2). Kaufman examines the impact of changes in the monetary aggregate on changes in income in a particular current or future quarter, a procedure which is appropriate if money's effect on income occurs with a discrete time lag. The present study, extends the Kaufman analysis, allowing the effect of money on GNP to be distributed over several quarters by examining regressions of the following form: