Comovements in Aggregate and Relative Prices: Some Evidence on Neutrality

International Finance Discussion Papers Number 285

July 1986

COMOVEMENTS IN AGGREGATE AND RELATIVE PRICES: SOME EVIDENCE ON NEUTRALITY

by

B. Dianne Pauls

NOTE: International Finance Discussi circulated to stimulate disc publications to Internationa acknowledgment by a wri material) should be cle

on Papers are preliminary materials ussion and critical comment. References in 1 Finance Discussion Papers (other than an ter that he has had access to unpub lished

ared with the author or authors.

ABSTRACT

This paper develops an alternative test of the neutrality of anticipated monetary policy. A multi-good equilibrium model along the lines of Barro and Hercowitz is used to derive a neutrality proposition for anticipated movements in the aggregate price level and to demonstrate econometrically its equivalence to the exogeneity of relative prices with respect to

the aggregate price level. Multivariate causality tests provide a

basis for testing these restrictions. The empirical results provide mixed evidence for the equilibrium models, while the variation in the findings across industries suggests a role for supply-side disturbances

ir. explaining comovements in aggregate and relative prices.

Comovements in Aggregate and Relative Prices: Some Evidence on Neutrality

by B. Dianne Pauls*

Introduction

Within a wide class of macroeconomic models, including both Keynesian and classical rational expectations models, the channel for monetary policy to affect real activity is through changes in actual or perceived relative prices and wages. As a result, if anticipated monetary policy is to have real effects, these should appear in the relative wage and price structure. For example, suppose anticipated monetary policy impinges on the real sector through changes in the real rate of interest. In that case, any Mundell-Tobin effects will be reflected in the cross-sectional data as changes in the relative wages and prices of capital-intensive versus labor-intensive industries. Therefore, it is appropriate, in gathering evidence on the debate over the neutrality of anticipated monetary policy, to examine directly comovements between relative prices and relative wages on the one hand, and nominal variables on the other.!

This paper takes one class of models in which anticipated monetary policy is neutral, equilibrium models with misperceptions, and

derives a set of testable restrictions on the joint time-series behavior

* The author is a staff economist in the International Finance Division,

Board of Governors of the Federal Reserve System. This paper is the revision of an earlier paper entitled, “On the Causal Implications of Inflation and Relative Price Dispersion.” The author gratefully acknowledges helpful comments from Peter K. Clark, Stanley Fischer,

Robert Litterman, Christopher Sims, Robert Solow, and Stuart Weiner.

The views expressed herein are the sole responsibility of the author

and not of the Board of Governors of the Federal Reserve or its staff.

1. Similar tests of the invariance of the relative wage structure with respect to anticipated monetary policy appear in B.D. Pauls, “Relative Wages, Wage Determination, and Inflation.”

of aggregate and relative prices. The reasons for concentrating on these particular modeis are twofold. First, it is important to see how far we can push the competitive paradigm in explaining the covariation in the data. And second, this is the class of models which generates the greatest concern over the potentially destabilizing effects of macroeconomic policies. Specifically, in a world where markets clear but information is incomplete, monetary disturbances may create distortions in relative prices and thereby decrease the allocative efficicucy of relative price signals.

The paper begins with a description of a prototype multi-market equilibrium model, which follows along the lines of Hercowitz (1981). The major innovation in the model is the introduction of dynamics into the demand and supply decision rules. Using this model as an example, it is shown that the characteristic assumption of equilibrium models, that prices are flexible, implies a kind of neutrality proposition. Given the past history of a spanning set of relative prices, today's relative price structure is independent of anticipated movements in the general price level. Since, under rational expectations, these restrictions are equivalent to the econometric exogeneity of relative prices, Granger-3ims causality tests provide a natural method for testing this proposition,

The empirical methodology is similar to that used by Sargent (1976), Barro (1977), and Sims (1980). Rather than directly examine comovements between relative prices and nominal variables, these studies focus on the implications of an equilibrium model for the exogeneity of aggregate quantity variables, such as GNP and unemployment, with respect

to nominal variables. Although a given equilibrium model may restrict:

-3-

both aggregate quantities and relative prices to be exogenous, statistical inference in the two contexts will not necessarily produce equivalent results because these tests only examine restrictions on the reduced

form. Any particular set of reduced form restrictions is potentially generated by a number of structural models.2 Since the set of models

that implies the statistical exogeneity of aggregate quantities and the set cf models that restricts relative prices to be statistically

exogenous are not identical, the likelihood that the equilibrium misperceptions paradigm provides a useful approximation is increased if both sets of restrictions hold. In interpreting this literature, it is important to recognize that all of these tests are direct tests of the dichotomous structure of these models--that is, that a real sector representation exists that does not include lagged values of nominal variables. However, because dichotomy is more readily generated from a model in which anticipated monetary policy is neutral, the tests also provide indirect evidence on neutrality.

The empirical results provide mixed evidence in support of the equilibrium models. In roughly half of the industries examined, the exogeneity of relative prices is rejected. The pattern of rejections across industries suggests a different dynamic representation for the shocks, rather than deviations from the competitive paradigm or rational expectations as an explanation for the failure of the model. Specifically, supply-side disturbances, which are transmitted across sectors over time, may play an important role in accounting for comovements in aggregate and 2. For a general discussion of the issue of observational equivalence as

it pertains to this class of empirical work see the exchange between McCallum (1979) and Sargent (1979a).

-4~

relative prices during the 1966 to 1980 period. Furthermore, the results of the exogeneity tests that include proxies for the supply-side stocks tend to support this conjecture. Viewed in this light, these findings suggest that the aggregate price level may add significant predictive power because it is a proxy for omitted real variables, and, therefore, the results do not pose any conclusive evidence leading to a rejection of the equilibrium model. I. An Illustrative Model

Previous studies have adopted the Phelps~Lucas model of informationally separate markets to a multi-good world and used it as a vehicle for explaining relative price behavior. (Parks (1977), Cuikerman (1979), and Hercowitz (1981)). A modified version of Hercowitz's (1981) multi-market model is used for illustrative purposes. The economy consists of a continuum of markets, each trading a different good. At each date t, any individual agent transacts in only one market. It is assumed that markets are competitive so that agents take their own local prices, pt(z) as given. Both consumers and producers locate across markets based on perceived relative prices. The basic tenet of the equilibrium misperceptions theory is that agents observe their own local prices more readily than they observe prices in other markets, and hence more readily than they observe the aggregate price level. Consequently, agents evaluate relative prices given current knowledge of their own market price and their conditicnal expectation of the aggregate price level.

Indexing markets by z, supply and demand in the various markets are represented by the following stylized set of decision rules, expressed

in log-linear form:

s yr (2)

d yt(z)

where:

-5s s Ss = a(z)[p_e(z) — ECpe|Tp(2))] + A )ye-1(2) + A2M)ye-1

+ 5(L) (pea (2) - Peet) + ve + e¢(2) (1)

d = -o(z)[pe(z) — ECpe|Te(z))] + BME - E(pe Te (2))]

d d d d + 044(L)yp¢-1(z) + AQ(L)y_-1+ 43(L)(pe-1(2) - Pe-1) + er (z) (2)

s d ye(z), ye(z) = log of supply, and demand, respectively, in market z. pr(z) = log of price in market Zz.

E(pe|I¢e(z)) = expectation of average economy-wide log of price conditional on information available in market z~-- It(z). s d a(z), a(z) = relative price elasticity of supply, and demand in market z, a(z)>0, alz)>0.

Vt = economy-wide average real disturbance.

s d e¢(z), €¢(z) = relative disturbance to supply, and demand in market z.

8 = elasticity of demand with respect to real balances, 8 >0. Me = log of the money supply.

s s s d d d 441), AQ(L), A3(L), and Ai(L), AQ(L), 43(L), represent one-sided polynomials in non-negative powers of the lag operator.

According to equations (1) and (2) the distribution of demand

ard supply across markets depends on the perceived distribution of relative

s prices, together with the realizations of sector-specific shocks — e¢(z)

d and €¢(z). In the “islands” paradigm of Lucas(1973), a single good is

traded.

Perceived relative prices are the primary determinant of local

supply and demand decisions because they represent arbitrage opportunities.

Ia a multi-good context, agents' willingness to change markets in response

to perceived relative prices will depend upon the substitutibility of the

-6-

goods in consumption and production. Allowances for intermarket differences in the degree of substitutability are made by introducing relative price elasticities that vary across markets.

Each market is treated as a microcosm of the economy, in that market supply and demand schedules are expressed as subcomponents of the aggregate demand and aggregate supply schedules for the economy as a whole. Thus, the real balance effect enters each market demand function directly. Real shocks to aggregate supply, such as an exogenous change in produc— tivity, are represented as a common shift in the supply function in each market ——vt.

The basic modification made to Hercowitz's model is the introduction of dynamics into the demand and supply decision rules. This alteration is necessary if the real variables are to display cyclical behavior. The dynamic specification represents intertemporal substitution possibilities in production and consumption, both of which depend upon expected future as well as current relative prices. Based on work by Hansen and Sargent (1979), the dynamic decision rules include lagged values of the decision variable together with any information that helps predict future relative prices. Since economic theory generally does not dictate the information set used by firms and consumers for predicting relative prices, a general specification, consisting of relative ‘orices and market output relative to aggregate output, was adopted. In keeping with the spirit of the equilibrium model, the intramarket dynamics, which are incorporated in the A parameters, are constrained to be the same

across markets.

-7-

There are three sources of disturbances in the economy--nominal aggregate disturbances or innovations in money growth, real aggregate disturbances and relative disturbances. Assuming the rate of growth of money is serially uncorrelated, the money supply is given by:

Me = Me-1 + Ut Because equilibrium models with misperceptions provide no channel for predictable components of money growth to affect relative prices, no gererality is lost by assuming this form for money growth.

Real disturbances in this model are decomposed into an economy wide average shock to supply--vy-~and relative shocks to supply and

s d demand in the various markets-- €¢(z) and e¢(z). Since vt represents the aggregate real disturbance that is common to all markets, the relative disturbances include both market-specific components of aggregate real disturbances and such relative shocks as a redistribution of demand between two markets. Thus, the relative disturbances net out by construc— d s tion: JSez(z) dz = 0, where €+(z) = €¢(z) - €¢(z) is the net excess demand disturbance in the market z. The distributional assumptions regarding the disturbances are: 2 ut ~ NCO, Oy) 2 Vt ~ NCO, Oy) 2 In addition, it is assumed that each type of disturbance is serially uncorrelated and that all disturbances are mutually orthogonal for all t.3 The 3. In terms of the model developed here, it is sufficient that the aggregate disturbances and the relative disturbances be uncorrelated at

nonzero lags and leads. However, imposing the additional structure simplifies the exposition.

-8-

first condition, that aggregate disturbances be uncorrelated with all lagged relative disturbances seems plausible given relative disturbances are defined to net out in the aggregate. This property follows from the assumptions that prices are flexible and relative shocks are drawn from the same distribution with a mean of zero. Thus, there is, for instance, no basis for the policy that determines aggregate demand to respond to relative excess demand or supply in a particular sector. The second condition, that relative disturbances be uncorrelated with lagged aggregate disturbances, does not rule out any non-neutrality per se. Rather, it only limits the channel whereby lagged aggregate disturbances affect market-specific behavior, so that they impinge upon particular sectors either through lagged output—-both sectoral and aggregate-—-or lagged prices and not through the relative disturbance term. Assuming markets clear instantaneously, we can equate (1) and (2) and solve for the market price under partial information: Pe(z) = [1-By(z)] E(pe|Ie(z)) + By(z)Mp-7- y(z)Aq(L)y¢-1 (z) ~ Y(Z)AQ(L) ye-1 - ¥(z)A3(L)(pe-1 (2) - pe-1) + By(z)ug - y(z)v_e + y(z)ee(z) (3) where: y(z) = 1/(a8(z) + a4(z))

Aq(L) = 4 (L) - Su) for i=1,2,3, and each of these newlydefined lag distributions is assumed to be one-sided and to have a one-sided inverse.

The model is solved using the method of undetermined coefficients. Conjecturing the reduced form for the aggregate price level gives: Pt = Mey + 1y(L)yg-y + 1o(Bug- v_) (4) Given agents observe local prices more readily than they observe

prices in other markets, they face an optimal filtering problem. Individuals

-9-

observe a composite of the three disturbances contained in the current market price and have to infer the aggregate and relative components. The information set of agents in market Zz, I_(z), consists of current local prices, pe(z), lagged values of market prices and output in all markets, the parameters of the model, and the first and second moments of the probability distributions of the random disturbances. If expecta~ tions are rational then agents update their forecasts of the aggregate price level based on equation (4).

E(pelIe(z)) = Mena + Ty(L)ye-1 + WoECbue — velTe(z)) (5) The optimal updating procedure uses agents’ knowledge of the historical sources of variability in the economy to filter the composite disturbance contained in the current market price, But ~ Vt + e¢(z). Substituting

the optimal linear filter into the expectation formula gives: 4

22 B Oy + on, + O¢

2 E(pel|Ie(z)) = Me-1 + M1(L)ye-1 + 12/8 on + 4 | (Bug - ve + €¢(2)) (6) Replacing the expectation in (3) with (6) yields the reduced

form for the market price:

pe(z) = Me-1 + (01 - By(z))Ny(L) — y(z)A2(L) lye-1

y(z)Azy(L)ye-1(2) = ¥(z)A3(L) (pe~1 (2) -Pe-1)

+

22 2 (1 - By(z))12(B oy + Sy) + y(z)( (Bug — ve + €¢(2)) (7) 22 2 2 B Oy + Oy + Se

Integrating over markets under the assumption that y(z), the reciprocal ee 4. The coefficient on the composite disturbance is that proportion of the

variance in market price innovations which has been attributed historically to aggregate price innovations.

-10-

of the elasticity of net excess demand, is distributed independently of €4(z), and solving for the undetermined coefficients gives the following expressions for the aggregate price level and the undetermined

coefficients:

Pe = Me-y ~ [QA](L) + AQ(L))/Blye-z + [6/8 + y (1 - 0)] (Bug - ve) (8)

i]

where: y = Sy(z)dz = average of the reciprocal of the relative price elasticity of excess demand across

markets.

3 2 2 3 2 2 — 2 (B oy + Boy)/(B oy + Boy + (1/y)a¢)

QO = y(L) = ~ (A4(L) + A2(L))/8 Ig = 6/6 + y(1 - 6)

Substituting these coefficients into (7), the relative price in

market z is given as: Pte(z) — pe = — y(z)ay(L)(ye-1(z) - ye-z) — y(z)A3(L)(pe-1(z) - pe-1) + (y(z) - Y)(1 - ©)(Bug - ve) + [8/8 + y(z)(1 - 0)] e¢(z) (9)

In order for equation (9) to be a rational-expectations solution, the information set that was used in projecting future relative prices should be consistent with the description of the relative price process contained in equation (9). Comparing equations (1) and (2) with equation (9), the solution meets this criterion.”

According to equation (8), innovations in the aggregate price level are a function of But - ve. Thus, equation (9) indicates that, while innovations in relative prices and innovations in the aggregat:e price level are contemporaneously correlated, aggregate prices do not Granger-cause relative prices. Although this implication was derived in

5. However, the information set used is not unique in providing an internally consistent solution. ,

-1ll-

the context of a particular, highly stylized example, it embodies a kind of classical dichotomy property that is a general feature of equilibrium models. The characteristics of these models, that expectations are rational and wages and prices are flexible, imply that only the “surprise” component of the nominal variables affects the real variables of the system. While the dynamic form of the decision rule allows the effects

of agents' past forecast errors to persist, ina market-clearing framework, their effects on current relative prices will be mediated through lagged relative prices and output shares. Thus, all the relevant history for predicting the future path of the real variables is embedded in the past realizations of the real variables, including, in the more general case where market dynamics vary across sectors, relative prices in other sectors of the economy. © As a result, these models dichotomize in expected values. Econometrically, this property implies that an appropriately defined set of relative prices and other real variables--in this case relative output shares--is jointly exogenous with respect to the aggregate price level. Thus, the causality tests directly examine the dichotomy proposition outlined above, and bear indirect evidence on the neutrality issue in the sense that dichotomy is more readily generated from the

class of models in which money is neutral.

6. In addition, there is the technical requirement that the relevant vector autoregressions for predicting relative prices be one-sided in lagged observations. (Geweke (1983)). This assumption of the model is

tested jointly with the implications of market-clearing and rational expectations.

-12-

Il. Empirical Results

Tests of the hypotheses that aggregate prices do not Grangercause relative prices were conducted for a sample of seven four-digit: industries.’ The sample consists of monthly observations for the period 1965 to 1979. Because this period spans the wage-price controls of August 1971 to April 1974, a natural question arises concerning the stability of the reduced form equations. The hypothesis of no structural change versus the alternative of three distinct subperiods--the periods before, during, and after wage-price controls--was tested in the context of the null model. Based on equation (9), the regressors included in the null specification were: lagged relative prices, lagged relative output shares, a constant, seasonal dummies, and a time trend. All variables are in logs. To allow for the possibility of breaks in the level and trend, wage-price control dummies and additional time trends were included in the second and third subperiods. The tests reject subsample stability at the 99 percent confidence level in five out of seven industries and reject stability in the sixth industry at the 90 percent confidence level.

Having rejected the hypothesis of no structural change across all three subperiods, the wage-price control episode was omitted and a test of whether the data permit pooling the pre and post wage-price control periods was conducted. In all but one industry--glass containers-—— the tests fail to reject subsample stability at a marginal significance level of 0.05 percent. For the one industry where stability was rejected, 7. The number of industries in the sample is restricted by the availa-

bility of conformable price and output indices. For a description of the data, see Appendix A.

ire 2 enna eee

-13-

the Granger causality tests were performed separately on each portion of the sample.

The tests that relative prices are not Granger-caused by the aggregate price level entail F tests on lags of the aggregate price level in equation (9). However, in order for the computed test statistic to have an F distribution, the residuals must be independently and normally distributed. Consequently, after pretesting for the appropriate lag length for the relative prices and relative output shares for each industry, Lagrange multiplier tests were conducted to verify that there was no remaining serial correlation in the residuals. Because pretesting for the appropriate lag length for the causal variable--in this case the aggregate price level--alters the significance level of the exogeneity tests, the Granger test statistic is computed for alternative lag specifications for this variable using six- and twelve-month lags.

Tables 1 and 2 present an illustrative relative price equation based on one industry and a summary of the F statistics for the exogeneity tests for all the industries in the sample. The equation shown in Table 1 represents the null specification corresponding to equation (9). This equation is only meant to be illustrative, as the specification of the lag length in the null model varies across industries. Due to differences in lag length and data availability across industries, the estimation period varies somewhat among the different industries, although it generally covers the period 1966 to 1980, omitting the wage-price control observations.

The exogeneity test results displayed in Table 2 are mixed. The

exogeneity of relative prices with respect to the aggregate price level can

-14-

be rejected at the 95 percent confidence level in three out of the seven industries examined. These are cigarettes, synthetic rubber, and brick.

A consistent interpretation of the results requires an explanation of' the differences in the findings across industries, in addition to a rationale for the observations that are at variance with the model. On this basis, the failure of rational expectations is not a likely explanation for the rejection of the exogeneity restrictions because it is difficult to see why expectations should be formed rationally in some industries and not

in others. Potential explanations do include departures from the competitive paradigm and non-market clearing. Regarding non-competitive behavior, there is no correspondence between concentration ratios and causal orderings across industries.

To determine whether deviations from market-clearing might account for the results, the industries in the sample are classified according to whether they can be primarily characterized as customer or auction markets. Among the industries examined, cigarettes, wines and brandy, and, possibly, glass containers are likely auction markets, g:iven they are primarily final products. The remaining industries—~synthet:ic rubber, man-made fibers, brick and aluminum—largely consist of internediate goods and building materials. As such, they are apt to be customer markets in which there is some contractual relationship between the supplier and the industry purchasing the intermediate input. According to this framework, rejections of the model reflecting departures from market~clearing are expected to be concentrated in customer markets. However, contrary to this prediction, rejections of the exogeneity

restrictions cut across both categories of markets, and in many

-15-

industries characterized as customer markets, the data are consistent with an equilibrium model.

It is a general property of the causality tests that differences in the dynamic properties of the shocks from those assumed in the model can lead to the failure of exogeneity. In particular, it is assumed that the market-specific shocks to demand and supply, eu(z) and er (z) are serially uncorrelated. However, as these disturbances include unobserved shocks to tastes and technology, there is no reason to presume that they might not be serially correlated. In that case, as Sargent (1979a) demonstrates, provided that the real variables that are used as regressors are not sufficient to span the space containing the real shocks, lagged nominal variables can add predictive power in an equation describing the real variables because they help predict the unobserved shocks to tastes and technology.

Another possible difference in the dynamic properties of the shocks is that the aggregate real shocks may have a variable timing across sectors of the economy. Moreover, the model was developed in the context of aggregate shocks that have a coincident impact on the various sectors of the economy. Although shocks to nominal aggregate demand are apt to exhibit this behavior, supply shocks, which were prevalent in the 1970's, tend to have an uneven impact across industries, both in terms of timing and magnitude. Differences in magnitude are allowed for in the market-specific shock, but variable timing is not. In an environment where aggregate disturbances have a variable timing across sectors of the econony, information about the transmission of shocks among industries

becomes relevant. As a result, the reduced form needs to be amended to

-16-

capture these interdependencies between lagged sectoral prices. In terms of the above tests, the exogeneity restrictions may be rejected in a subset of the industries examined because the aggregate price level is proxying for the omitted variables, either lagged relative prices in other sectors or real shocks to tastes and technology.

At least one of these explanations, that asymmetries in the dynamic effects of supply shocks contributed to the failure of exogeneity, can be tested. Taking the lagged relative prices of food and energy as representing the supply shocks, the exogeneity tests for the industries with rejections were rerun including these variables among the regressors. It should be noted that, to the extent that the energy shock induced changes in technology, the relative price of energy may also proxy for real shocks to technology. The results based on this specification, shown in the lower portion of Table 2, fail to reject the exogeneity restrictions in all but one industry, suggesting that intersectoral differences in the timing of supply shocks and/or serially correlated real shocks to technology may have accounted for the previous evidence against the model.

III. Conclusion

This paper proposes an alternative way of addressing the question of whether anticipated monetary policy is neutral. Rather than confronting the theory with observations on aggregate quantities and nominal variables, the tests focus on comovements in relative and aggregate prices. This approach cuts to the heart of the neutrality issue because relative price changes are the mechanism for monetary policy to affect real activity

in most macroeconomic models. The tests of neutrality examine a kind

-l7-

of classical dichotomy proposition that characterizes a major class of models in which anticipated money growth is neutral--equilibrium models with misperceptions. The proposition maintains that, given the past history of an appropriately defined set of real variables, today's relative price structure is independent of expected movements in the general price level. Under the condition that expectations are formed rationally, these restrictions are equivalent to the econometric exogeneity of relative prices with respect to the aggregate price level. Granger-Sims causality tests are then used to examine this empirical implication of the model.

The evidence from a cross-section of industries provides mixed results for the equilibrium models. The exogeneity of relative prices with respect to the aggregate price level is rejected in three out of the seven industries examined. Attempts to explain the cross-sectional variation in the results in terms of the failure of a particular assumption of the model find no correspondence between either the degree of market power or the proximity to an auction market and the exogeneity test results.

It is suggested that the dynamic representation of the shocks in the model is too restrictive to capture adequately real disturbances-—— either shocks to tastes and technology or commodity price disturbances. To the extent that changes in technology are induced by changes in relative factor costs, and supply-side disturbances are transmitted across sectors over time, the reduced form for relative prices needs to be amended to encompass these interdependencies. In the absence of an extensive set of lagged relative prices in other sectors, the aggregate price level may be

proxying for these interindustry disturbances. The results of the

-18-

exogeneity tests that include the lagged relative prices of food and energy among the regressors tend to support this conjecture. . Thus, although these findings imply that the aggregate price index contains information that is relevant for predicting future relative prices i1 the context of a limited set of relative price information, they do not imply that an aggregate price index would add significant predictive power to forecasts based on a complete vector of relative prices, and, hence, do not present any conclusive evidence leading to a rejection of the equili-

brium misperceptions model.

- 19 -

Table 1: Illustrative Relative Price Equation Null Specification

Dependent Variable: Relative price of cigarettes (S.1I.C. 2111) Period: 1965:7 - 1971:8 and 1974:10 - 1980:6

mmm

Variable (Lag (-)) Coefficient Standard Error eee Constan: -.021 -026 Dummy -020 -020

(1974: 10-1980: 6)

Time -00009 -00014 (1965:7--1980:6)

«

Time -.00002 -00016 (1974: 10-1980:6)

Relative Price (-1) 844 .091 (-2) -.072 -120 (-3) -.072 122 (-4) -094 — 121 (-5) -.030 122 (-6) 071 091 Relative Output (-1) -030 .028 (-2) | -028 -029 (-3) -.013 .031 (-4) -.047 -030 (-5) 022 -030 (-6) -.053 028

eee Note: Seasonal dummies were also included in the regression, but are not reported here.

- 20 -

Table 2: Exogeneity Test Results Tests of Causality from the Aggregate Price Level to Relative Prices

Marginal

Industry Significance (S.1.C.) Period F-Statistic Level 2084 1966:1 - 1971:8 & F (6, 44) = 1.327 - 266

1975:5 - 1975:12 F (12, 38) = 1.419 - 200 2111 1966:1 - 1971:8 & F (6, 96) = 5.109 -0001

1975:5 - 1980:6 F (12, 90) = 3.433 -0004 2822 1966:1 - 1971:8 & F (6, 90) = 2.776 .016

1975:5 - 1980:6 F (12, 84) = 1.596 .108 2824 1969:6 - 1971:8 & F (6, 28) = .459 -833

1975:10 - 1980:6 F (12, 22) = .563 -848 3221 1966:1 - 1971:8 F (6, 39) = 1.160 -347

F (12, 33) = .953 -510 1975:5 - 1980:6 F (6, 33) = 2.169 .071 F (12, 27) = 1.548 - 168

3251 1966:1 - 1971:8 & F (6, 88) = 3.603 -003

1975:5 - 1980:6 F (12, 82) = 2.830 -003 3334 1968:2 - 1971:8 & F (6, 56) = 1.235 - 303

1975:6 - 1980:6 F (12, 50) = .759 688 ADDENDA:

Causality Tests with Relative Prices of Food and Energy Included Among the Regressors

2111 1966:1 - 1971:8 & F (6, 84) = 3.836 -0020 1975:5 - 1980:6 F (12, 78) = 2.324 0134

2822 1966:1 - 1971:8 & F (6, 72) = 1.733 .126 1975:5 - 1980:6 F (12, 66) = .887 +564

3251 1966:1 - 1971:8 & F (6, 68) = .716 -638 1975:5 - 1980:6 F (12, 62) = 1.175 321

ne LL

-21-

Data Appendix

The relative price and output share data were constructed using conformable price and output indices at the four-digit industry level of disaggregation. The industries in the sample were:

SIC CODE

2084 - Wines and Brandy

2111 - Cigarettes

2822 -— Synthetic Rubber

2824 - Man-made Fibers

3221 - Glass Containers

3251 - Brick

3334 - Aluminum

The market-specific prices were constructed from the Industry- Sector Price Indexes published by the Bureau of Labor Statistics. These indices are based on regroupings of the Wholesale Price Index to correspond to indiustries rather than the end use of the product. The sample period generally begins in 1965 with the availability of these series and ends in 1930, when the weighting procedure used to compute the indexes was substantially revised. Relative prices were calculated as the logarithmic difference of the Industry-Sector Price Index and the Wholesale Price Index for all manufacturing, which was taken to be the aggregate price level,

The industry shares of output were constructed from corresponding subcomponents of the Industrial Production Index published by the Board of Governors of the Federal Reserve System. These were computed as the logarithmic difference of industrial production for the industry and indust:rial production for all manufacturing.

For the supply shock variables, the relative prices of food and

fuel, the logarithmic differences between the WPI for farm products, and

-22-

fuels and power, and the WPI for all commodities were used. All data were seasonally unadjusted and seasonal dummies were included in the

estimation.

-23- References

Barro, R., "Rational Expectations and the Role of Monetary Policy",

Journal of Monetary Economics, vol. 2, (January 1976), pp. 1-32.

» “Unanticipated Money Growth and Unemployment in the U.S.", American Economic Review, vol. 67, (March 1977), pp. 101-115. nomic keview

——

Cuikerman, A., “Relative Price Variability, Inflation, and the Allocative

Efficiency of the Price System", Journal of Monetary Economics, vol. 9, (March 1982), pp. 131-162

Geweke, John, “Inference and Causality in Economic Time Series Models", in Handbook of Econometrics, Griliches, Z., and M. Intrilligator, (eds.) North Holland, New York, 1983,

Hansen, 4. and T, Sargent, “Formulating and Estimating Dynamic Linear

Rational Expectations Models", Journal of Economic Dynamics and Control, vol. 2, (February 1980) 7-46.

» PP.

Hercowitz, Z., “Money and the Dispersion of Relative Prices," Journal of Political Economy, vol. 89, (April 1981), pp. 328-356,

» “Money and Price Dispersion in the United States," Journal

of Monetary Economics, vol. 10, (July 1982), pp. 25-38,

Lucas, Robert E. Jr., and Thomas J. Sargent, Rational Expectations and Econometric Practice, — te tactice

University of Minnesota Press, 1981,

ee

pp. xi-xxxiv.

__» “Some International Evidence on Out put-Inf lation Trade-offs", American Economic Review, vol. 63, (June 1973), pp. 326-335, nomic Review

McCallum, Bennett, "On the Observational Inequivalence of Classical and

Keynesian Models” Journal of Political Economy, vol. 87, (April 1979), pp. 395-402.

Parks, R., "Inflation and Relative Price Variability", Journal of Political Economy, vol. 86, (February 1978), pp. 79-96.

Pauls, B. Dianne, “Relative Wages, Wage Determination, and Inflation", (unpublished) Dartmouth College 1981,

Sargent, T.J., "A Classical Macroeconometric Model of the United States”, sfournal of Political Econom » VOl. 84, (April 1976), pp. 207-238,

_» “Causality, Exogeneity, and Natural Rate Models: Reply to C. R.

Nelson and B. T, McCallum", Journal of Political Economy. , vol. 87, (April 1979), pp. 403-411.

-24- , “Interpreting Economic Time Series,” Journal of Political Economy, vol. 89, (April 1981), pp.213-248. , Macroeconomic Theory, Academic Press, New York, 1979.

Sims, C., "Macroeconomics and Reality”, Econometrica, vol. 48, (January 1980), PPpe 1-48.

, "Money, Income and Causality”, American Economic Review, vol. 62, (September 1972), pp. 540-552.

IFDP NUMBER 285 284 283

282

281

280

279

278

277

276

275

274

273

Please address requests for co Papers,

Division of International Finance,

International Finance Discussion Papers ee i cusSi0n Papers

TITLES

1986

Comovements in Aggregate and Relative Prices: Some Evidence on Neutrality

Labor Market Rigidities and Unemployment: the Case of Severance Costs

A Framework for Analyzing the Process of Financial Innovation

Is the ECU an Optimal Currency Basket?

Are Foreign Exchange Forecasts Rational? New Evidence from Survey Data

Taxation of Capital Gains on Foreign Exchange Transactions and the Non-neutrality of Changes in Anticipated Inflation

The Prospect of a Depreciating Dollar and Possible Tension Inside the EMS

The Stock Market and Exchange Rate Dynamics ©

Can Debtor Countries Service Their Debts? Income and Price Elasticities for Exports of Developing Countries

Post-simulation Analysis of Monte Carlo Experiments: Interpreting Pesaran's (1974) Study of Non-nested Hypothesis Test Statistics

A Method for Solving Systems of First Order Linear Homogeneous Differential Equations When the Elements of the Forcing Vector

are Modelled as Step Functions

International Comparisons of Fiscal Policy: The OECD and the IMF Measures of Fiscal Impulse

An Analysis of the Welfare Implications of Alternative Exchange Rate Regimes: An Intertemporal Model with an Application

the Federal Reserve System, Washington, D.C. 20551.

—25-

AUTHOR(s)

B. Dianne Pauls

Michael K. Gavin Allen B. Frankel Catherine L. Mann

Hali J. Edison

Kathryn M. Dominguez

Garry J. Schinasi

Jacques Melitz Michael K. Gavin Jaime Mar quez

Caryl McNeilly

Neil R. Ericsson

Robert A. Johnson

Garry Schinasi

Andrew Feltenstein David Lebow Anne Sibert

pies to International Finance Discussion Stop 24, Board of Governors of

IFDP

NUMBER

272

271

270

. 269 268

267 266

265 264

263

262 261

260

International Finance Discussion Papers

TITLES

1985 (partial listing) Expected Fiscal Policy and the Recession of 1982

Elections and Macroeconomic Policy Cycles

Assertion Without Empirical Basis: An Econometric Appraisal of Monetary Trends

Hones in ... the United Kingdom by Milton Friedman

and Anna J. Schwartz Canadian Financial Markets: The Government's Proposal for Reform

Was It Real? The Exchange Rate Interest Differential Relation, 1973-1984

The U.K. Sector of the Federal Reserve's Multicountry Model: The Effects of Monetary and Fiscal Policies

Optimal Currency Basket in a World of Generalized Floating: An Application to the Nordic Countries

Money Demand in Open Economies: Substitution Model for Venezuela

A Currency

Comparing Costs of Note Issuance Facilities and Eurocredits

Some Implications of the President's Tax Proposals for U.S. Banks with Claims on Developing Countries

Monetary Stabilization Policy in an Open Economy

Anticipatory Capital Flows and the Behaviour of the Dollar

Simulating Exchange Rate Shocks in the MPS and MCM Models: An Evaluation

-26-

AUTHOR(S)

William H. Branson Arminio Fraga Robert A. Johnson

Kenneth Rogoff Anne Sibert

David F. Hendry Neil R. Ericsson

Garry J. Schinasi Richard Meese Kenneth Rogoff

Hali J. Edison

Hali J. Edison Erling Yardal

Jaime Mar quez Rodney A. Mills

Allen B. Frankel

Marcus H. Miller Arnold Kling

Arnold Kling