Knowledge that Transforms

THE comparison of prices and price levels is an important ingredient in many economic forecasts, policy prescriptions and empirical academic studies. For example, prices charged by a firm for domestic sales are compared with those charged on exports to substantiate charges of dumping. Differences in national price levels are also used in place of real foreign exchange rates as more stable and appropriate translators for international comparisons of such macroeconomic indicators as gross national product or per capita income. On the domestic side, temporal comparisons of individual prices and their contributions to a high and volatile inflation rate have been cited by advocates of wage and price controls and an oil consumption tax, and comparisons of prices, price levels, and exchange rates are at the center of academic controversies over the validity of the law of one price and the purchasing power parity theory. The fundamental assumption underlying these investigations is that the goods or services under comparison are identical. This assumption has proven troublesome because differences in product characteristics such as quality, service support, and delivery availability are difficult to identify and value; they are also generally specific, in size and character, to a product or industry. Further, it may be difficult to see the relationship between these qualitative characteristics and macroeconomic variables, a relationship that, once established, might simplify the identification and measurement of product differences. As a result, the effect of differences in product quality on comparisons of prices is commonly ignored or assumed random or stable. However, one product quality-extended trade credit-can be measured easily because it is neither random nor stable. It also has the desirable trait of being systematically related to several macroeconomic variables: the cost of extending trade credit is a function of interest rates and expected inflation-a correlation that allows us to recognize the importance of trade credit in some price comparison applications. It is possible to use this relationship to adjust prices and price levels for the extension of trade credit. More importantly, some price comparisons may be substantially affected by the adjustment. This paper describes a model linking transactions prices and trade credit. It demonstrates that the prices of identical goods will differ, even in perfectly efficient markets for goods and capital, if credit terms differ. As a result, apparent arbitrage profits may disappear when prices are adjusted for the value of trade credit extended to finance the goods. One major component is the length of time for which trade credit is extended. Within a single economy different explicit or implied credit maturities translate into different credit costs. In turn, these costs are reflected in price differences. A good example is the recent use of merchant discounts for cash payments as a disincentive against credit card payment. The amount of the discount is directly related to the merchants' cost of funds. When prices are compared internationally, another element is added. Variations in the cost of credit extended in different currencies may reflect differences in interest rates and expected inflation. When compared at the current exchange rate the national prices of goods sold on credit should differ by the variation in the cost of providing that credit. The reflection of trade credit in the relationship between prices, exchange rates, and interest rates has several interesting implications. For example, the increased rate of inflation normally expected after a devaluation may not present itself if it has been reflected in transactions prices prior to devaluation; when devaluation occurs the domestic inflationary response may seem unReceived for publication June 12, 1980. Revision accepted for publication February 18, 1981. * Duke University. I am grateful for helpful comments from Jacob Frenkel, John Hekman, Richard Levich, Stephen Magee, David Mullins, Rachel McCulloch, Aris Protopapadakis, and William White. I received additional suggestions from the International Trade Workshop of the Harvard Economics Department. Support for this work from the Associates of the Harvard Business School is gratefully acknowledged.

In this comment to a paper by Levhari and Levy published in this journal on February 1977 we provide evidence that shows that their conclusion does not hold if one applies the model to a large sample of common stocks.

source and the extent of lags in the response of prices to changes in various explanatory variables. Knowledge of the relevant lag patterns is of obvious importance for forecasting movements in prices and for formulating policies that are designed to reduce the rate of inflation. In the recent empirical literature on prices, the model most commonly used to specify price equations is the static neo-classical model of price formation.1 This model presumes that the representative firm behaves as if it were a simple monopolist that faces competitive factor input markets. The firm is assumed to set its price so as to maximize current profits. The model yields the proposition that the firm's long-run optimal price depends on expected normal factor input prices and variables, e.g., expected real income, that shift the firm's demand curve. Lags are introduced into the pricing process of this model in two ways. Some authors2 focus on expectation formation lags whereby normal factor input prices or variables that shift the firm's demand curve are related through distributed lags to past values of the variable being forecast. Often, the distributed lag is assumed to be

T HE international monetary events of the 1970s-the collapse of Bretton Woods, the introduction of generalized floating exchange rates, the appearance of significant inflation differentials among the major industrialized nations (e.g., Goldstein and Young, 1979), and the recent threat of rate wars -have renewed interest among international economists in the role real rates of return play in determining the international pattern of capital flows (Mudd, 1979) and exchange rates (Frankel, 1979). In this paper I investigate the empirical relationship between long-term portfolio capital flows and the real rate of interest for three European countries and the United States. The flows are disaggregated according to asset and liability categories and are measured on a quarterly basis; I examine only long-term portfolio flows into and out of the private sectors of the United Kingdom, West Germany, Italy, and the United States. The methodology represents an extension and mathematical redefinition of the stockadjustment approach to capital flow modeling developed by Branson (1968) and applied by Miller and Whitman (1970) to analyze long-term foreign portfolio investment. Real rates of interest are explicitly calculated through the computation of (weakly) rational expected inflation variables (Kreicher, 1979), which are then composed with long-term nominal rates in a manner consistent with a long-run, relative Purchasing-Power Parity theory of exchange rate expectations. The results reported here support the stockadjustment approach and provide fairly convincing evidence in support of the hypothesis that international investment decisions are sensitive and responsive to inflationary and exchange rate expectations and real interest rates.

Artus, Jacques R., 'The Behavior of Export Prices for Manufactures, IMF Staff Papers 21 (Nov. 1974), 583608. Artus, Jacques R., and John H. Young, Fixed and Flexible Rates: Renewal of the Debate, IMF Staff Papers 26 (Dec. 1979), 654-698. Detomasi, D. D., The Elasticity of Demand for Canadian Exports to the United States, Canadian Journal of Economics 2 (Aug. 1979), 416-426. Grimm, Bruce T., An Analysis of the Lagged Determinants of United States Import and Export Components, Ph.D. thesis, University of Pennsylvania (1968). Houthakker, Hendrik S., and Stephen P. Magee, Income and Elasticities in World Trade, this REVIEW 51 (May 1969), 111-125. Junz, Helen B., and Rudolf R. Rhomberg, Price Competitiveness in Export Trade among Industrial Countries, American Economic Review 63 (May 1973), 412-418. Magee, Stephen P., Currency Contracts, Pass-Through, and Devaluation, Brookings Papers on Economic Activity (1: 1973), 303-325. , Import Prices in the Currency-Contract Period, Brookings Papers on Economic Activity (1: 1974), 117-168. Marston, Richard C., Income Effects and Delivery Lags in British Import Demand: 1955-67, Journal of International Economics 1 (Nov. 1971), 375-399. Miller, Joseph S., and Michele Fratianni, The Lagged Adjustment of U.S. Trade to Prices and Income, Journal of Economies and Business 26 (Spring 1974), 191198. Murray, Tracy, and Peter J. Ginman, An Empirical Examination of the Traditional Aggregate Import Demand Model, this REVIEW 58 (Feb. 1976), 75-80. Wilson, John F., and Wendy E. Takacs, Differential Responses to and Exchange Rate Influences in the Foreign Trade of Selected Industrial Countries, this REVIEW 61 (May 1979), 267-279. Yadov, Gopal, A Quarterly Model of the Canadian Demand for Imports, 1956-72, Canadian Journal of Economics 8 (Aug. 1975), 410-422.

T HIS paper is concerned with differences among Canadian manufacturing industries in productivity relative to their U.S. counterparts. Data for two years, 1963 and 1972, were utilized to enable tests of constancies and changes over time. Some findings are quite familiar, some less so. The most important and robust novel finding is that U.S. counterpart industry technological activity is negatively related to Canada/U.S. relative productivity in 1972, but not 1963. The body of the paper has four parts: In the first two the expected impacts of structural and behavioural variables on relative productivity are discussed. Then the empirical results are presented and analyzed. A general discussion concludes the body of the paper. An appendix, separately available from the author, explains the sample selection and variable construction.

IN the field of industrial organization, the concept of of is of fundamental importance. Large minimum optimum scale (MOS)1 and significant unit cost reduction up to MOS size together suggest the possibility of high industry concentration and significant barriers to entry.2 However, a recent paper in this REVIEW (Caves, Shirazi and Porter (1975)) has pointed to the severe difficulties encountered by students of industrial organization in empirical analyses of these relationships. The chief problem is that objective measures of MOS (i.e., engineering estimates)3 and of the cost disadvantage of suboptimal scale production are available for only a small minority of all U.S. manufacturing industries at, say, the 4-digit level. Since there are more than 400 such industries, it is most unlikely that there ever will be much change in this situation. This means that in large-scale cross section statistical analyses various proxies for scale economy influences must be used. This is also the case in other fields of economics in which the influence of industrial organization has been recently felt-in the investigation of sectoral patterns of direct foreign investment and the commodity composition of international trade, for example. Unfortunately, little has been achieved in terms of the development of reliable proxies for scale economies or minimum optimal scale. This relative poverty in the data seriously jeopardizes the empirical analyses undertaken in all the above-mentioned fields. The present paper is a first step at overcoming this deficiency: it offers a critical analysis of the proxies currently in vogue, and it suggests an alternative technique for generating estimates of MOS that appear to be substantially superior to them.

LIKE any livestock industry, the Canadian beef and dairy cattle industry is characterized by a cyclical pattern in terms of the number of cattle on farm, the number of cattle slaughtered and the number of cattle exported. Traditionally, the analysis and forecast of cattle stocks are based either on econometric models which often include the biological life cycle of cattle or on the pure biological nature of cattle.' In this paper, we investigate the behavior of the cattle industry through the use of a third approach-the Markov chain technique. The Markov chain technique is by itself a very mechanical procedure, but one which can incorporate economic justifications. It can then, in our view, provide a very fruitful view of the industry. Basically, the Markov chain technique allows us to construct flow matrices of beef and dairy cattle according to their biological sequences. For instance, a male calf born during any time period t can, in the same period, be slaughtered, exported, die or remain on farm as a calf. The decision to retain a calf as a steer (for future slaughter) or as a bull (for future reproductive purposes), to export the calf, or to slaughter the calf (for veal), is basically an economic decision. The outcome of such decisions is translated into the elements of the Markovian transition flow matrix. Based upon the biological sequences of the different categories of cattle and the structure of the beef and dairy cattle industry, we can set up transition matrices for Western and Eastern Canada. Table I indicates the structure of such matrices for Western Canada. Cells representing possible flows are identified by numbers while cells representing impossible flows are left blank.2 Transition probability matrices of cattle movement can be constructed by dividing each row element in the matrix by its corresponding row total. These probabilities reflect the probabilities of cattle moving from one category to another. It is also through the use of such probabilities that we will carry out our simulation analysis. This paper is divided into seven sections. In the second section a model of demand, supply and inventory for beef and dairy cattle is presented. Section III discusses some general empirical results based on the transition probability matrices. Section IV discusses the procedures for simulation using the conditional transition probability matrices. Section V presents the results of the historical simulation while section VI presents the results of some sensitivity analysis experiments. The last section is for concluding remarks.

D ESPITE acute interest in Monetary Approach to Balance of Payments (MBOP) over past several years, both as an explanation of worldwide inflationary epidemic of late 1960s and early '70s (Johnson, 1972b) and as a simple and manageable theoretical basis for policy recommendations by groups like IMF (Rhomberg and Heller, 1977), no general consensus has been reached as to validity of monetary model. One reason for this is that task of interpreting results of plethora of empirical tests of monetary approach' has been complicated by controversy over accuracy of exogeneity assumptions underlying MBOP that are implicit in those studies. These include exogeneity with respect to reserve flows of determinants of demand for nominal money balances-the domestic price level, interest rate and level of real income-as well as domestic credit component of a country's money supply. Exogeneity makes task of interpretation difficult, for if MBOP assumptions regarding exogeneity are incorrect, then, to borrow Geweke's (1978, p. 163) words (and to add studies in MBOP literature to which they would apply), the otherwise identifying restrictions imposed on structural equations may not be sufficient to identify those equations (Argy and Kouri, 1974; Genberg, 1976), estimation procedures will be inconsistent (e.g., Bean, 1976; Zecher, 1976), and model cannot adequately portray dynamics of system it seeks to describe. This last problem transcends any issue of estimation (although it is not unrelated) for it implies that theoretical specification of MBOP is incorrect to begin with, thus invalidating putative reduced forms of MBOP that have come to represent approach and form basis for policy recommendations alluded to earlier. Historically, exogeneity was simply assumed as something not amenable to testing. Recently, however, Geweke (1978) has demonstrated that specification of exogeneity is a hypothesis with testable implications. The purpose of this paper is to test formally exogeneity hypothesis underlying empirical tests of MBOP specifically, and theoretical specification of model in general, using Geweke's methods. It is found that hypothesis that variables mentioned above are in fact exogenous can be rejected for each and every country in sample. This finding casts some doubt on theoretical validity of MBOP and even more doubt on accuracy of much of empirical evidence garnered in its support. The plan of paper is as follows. In section II a MBOP model is briefly presented and its exogeneity restrictions are highlighted. In section III formal test of exogeneity is discussed, and then in section IV results of tests using data from Australia, France, Germany, Norway and Sweden are presented. Section V contains some conclusions.