In search of price rigidities : recent sector evidence from Argentina

POLICY RESEARCH WORKING PAPER 155 8 In Search of Price Rigidities Monetaryand exchange rate policies have different effects on relative prices among (Recent Sector Evidence economic sectors - and thus from Argentina) significanUy influence the real side of the economy. Jacques Morisset Cesar Revoredo The World Bank Latin America and the Caribbean Country Department I Country Operations Division December 1995  POLICY RESEARCH WORKING PAPER 1558 Summary findings The hypothesis that the price adjustment to nominal almost instantaneously (although not one-for-one). The shocks is instantaneous has been part of the monetarist response in nontradable activities depends on indirect approach explaining the inflationary process in effects and whether prices are indexed to a foreign Argentina. currency, likely when transactions are conducted in a But Morisset and Revoredo argue that monetary and foreign currency. exchange rate policies have had different effects on Because understanding this is essential for effective relative prices and thus have a significant influence on policymaking, Morisset and Revoredo analyze price the real side of the economy. The existence of rigidities behavior of four economic sectors - agriculture, has prevented full and instantaneous price adjustments. industry, (retail) commerce, and services - in Argentina Recent work on inflation in imperfectly competitive from 1981-94. (The two nontradable sectors account for markets explains rigidities as a consequence of firms' most GDP and employment in Argentina.) strategic responses to nominal shocks, which in turn The econometric analysis shows large differences in depend on the market structure and demand elasticities the price behavior across sectors. Firms do not respond faced by firms. uniformly to changes in production costs, foreign prices, Price rigidities emerge when firms facing changes in and demand conditions. aggregate demand behave collusively, and there are costs The conclusions have obvious policy implications. The for customers to switch between suppliers. The higher response of individual prices reflects the distribution of the collusive behavior, the higher the possibility for these adjustment costs across sectors in the case of nominal firms to maintain or eventually increase their prices shocks. This is most evident when, facing a recession, during recession. In contrast, when the costs for some sectors are able to maintain their margins through customers to switch between suppliers are low, firms are collusive behavior, while others have to reduce them to obliged to adjust their prices to new demand conditions, retain their customers. To maintain social and political otherwise they will lose their customers. stability, the government's challenge is to minimize Changes in foreign prices affect domestic prices divergence across sectors. Increasing competition appears depending on the degree of foreign competition and the to be a crucial element of this strategy since monopolistic price formation mechanism in each sector. As expected, power is frequently associated with the existence of price price rigidities are minimal in tradable sectors where rigidities. firms react to these changes by changing their prices This paper - a product of the Country Operations Division, Country Department I, Latin America and the Caribbean - is part of a larger effort in the department to understand goods and services market behavior in developing countries. Copies of the paper are available free from the World Bank, ]818 H Street NW, Washington, DC 20433. Please contact Nan.c Cuellar, room Q7-116, telephone 202-473-7892, fax 202-522-3131, Internetaddress ncuellar@worldbank.org. December 1995. (26 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issujes. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own and should not he attributed to the World Bank, its Executive Board of Directors, or any of its member countries. Produced by the Policy Research Dissemination Center  IN SEARCH OF PRICE RIGIDITIES: Recent Sectoral Evidence from Argentina 1 by Jacques Morisset and 'esar Revoredo The World Bank LAlCO I We are grateful to A. Estache and A. Izquierdo for their helpful comments and suggestions.   ii Table of Contents Introduction 1 1. Nominal Shocks and Relative Prices Variability in Argentina: Some Evidence 2 2. Price Behavior across Economic Sectors: An Econometric Analysis 6 Industry 9 Agriculture 12 Services 15 Commerce 17 3. A Summary: The Differentiated Effects of Changes in Aggregate Demand and Foreign Prices on Sectoral Prices 19 4. Concluding Remarks 21 Bibliography 24 Appendix 1 26   1 Introduction "Even if the ultimate effects of monetary changes on different prices iwere to be, at long last, uniform, this would still be less important for many purposes than the initial variability'. [.. .] The fact that monetary changes do not affect all prices in the same wtaY, in the same degree, or at the same time, is what makes them significant. It is thle divergence bethween the movements of different price levels which are at once the test and thle measure of the social disturbances which are occurring. " J. M. Keynes. "A Treatise on Money: The Pure Theory of Money," pp.83-84. Traditionally, the monetarist approach in which all prices adjust instantaneously to nominal shocks has been used to explain the inflationary process in Argentina. Indeed, the variations in the price level have generally been the consequence of frequent changes in monetary and exchange rate policies. Still, this paper argues that these policies have had differentiated effects on relative prices and, thus, have influenced significantly the real side of the economy. The existence of rigidities has prevented full and instantaneous price adjustments. The existence and effects of rigidities in an economy can be understood in the light of recent work on inflation in imperfectly competitive markets (see Blanchard and Fischer (1989) for a survey). The main lesson is that the response of firms to nominal shocks is based on strategic considerations, which in turn depend on the market structure and demand elasticities. This mechanisimn must be clearly identified to be used for policy. That is what we do here, by examining the inflation behavior of four economic sectors in Argentina over the period 1981-1994; agriculture, industry, (retail) commerce, and services. The case of Argentina is particularly interesting because it offers a unique combination of large nominal shocks and disproportionate variations in the inflation rate over the past decades. Furthermore, unlike most existing empirical studies dealing with this issue, the analysis will not be restricted to industrial products. The inclusion of retail commerce and services will allow us to analyze the price behavior of the non-tradable sector, which accounts for the major share of GDP and employment in Argentina.  2 This approach has obvious policy implications. We subscribe to the view that the most important aspect of inflation consists in its asymmetric effectz on economic sectors. Traditionally, this aspect has been examined by emphasizing the costs and benefits of inflation on income distribution, with the wealthiest being able to protect themselves by using indexed financial instruments and strong wage bargaining power (see Khan and Morisset (1993)) In this paper, an alternative approach is taken that stresses the heterogeneous effects of (homogenous) nominal shocks on the prices of economic sectors. Large and unexpected divergences would have important real effects on the allocation mechanisms, which ultimately would affect social stability and the willingness of different groups to adhere to the existing social contract. The paper proceeds as follows. In the first section, we briefly review the principal explanations advanced in the recent literature to explain price rigidities, emphasizing their merits and limitations. We also present evidence on price rigidities in Argentina over the last decade. In the second section, we turn to the question of which variables influence the price behavior in Argentina. Because these variables and, above all, the magnitude of their effects on the inflation rate are likely to differ across sectors, the analysis is done at a sectoral level by distinguishing four main economic sectors. In the third part, we try to put the pieces together by summarizing the implications of a change in economic activity or/and foreign prices on the adjustment of prices in each economic sector in Argentina. The last section contains our principal conclusions as well as policy recommendations. 1. Nominal Shocks and Relative Prices Variability in Argentina: Some Evidence In the Keynesian tradition, nominal shocks such as changes in monetary and exchange rate polices are not expected to be neutral on the structure of relative prices because of the existence of rigidities. This issue is well known and several arguments coexist in the literature. First, price rigidities emerge when there are (menu) costs of changing prices such as fixed-term contracts or uncertainty created by asymetric  3 information. Under these conditions, some price-setters may decide not to adjust their prices in response to small nominal shocks (see Blanchard and Fischer (1989) or Andersen (1993) for a summary). Second, the differentiated response of individual firms to nominal shocks may also be caused by the availability of external funds. Liquidity-constrained firms may be forced to increase their prices more rapidly than others in periods of recession (or high interest rates) in order to raise internal funds (Chevalier and Sharfstein (1995)). Others popular arguments are that price rigidities are created by the existence of inventories (Blinder (1982)), the attachment of customers to their supplier because it is costly to shift between suppliers (e.g. the cost of research), or the collusion between firms in oligopolistic markets (Rotemberg and Woodford (1991). It also generally takes time for consumers and suppliers to react to a nominal shock or to realize that a shock occurred. Another related explanation is that prices are more rigid downwards than upwards, leading to dramatic changes in the structure of relative prices. This explanation, the so-called "kinked demand curve" has a long history in the industrial organization literature and has been applied to a macroeconomic context extensively (see Stiglitz (1984)). Finally, even if the initial monetary disequilibrium, for example, is eliminated by allowing an increase in prices so as to bring money in real terms back to its initial level, the transfer of wealth that will have taken place during the adjustment process will affect the relative price adjustment. As emphasized by Keynes (1930), this transfer will depend on the differential effect of the price movement on cash holders, on owners of fixed interest paper and on creditors, compared with on debtors. The existence of rigidities can be illustrated by the effect of nominal shocks on the structure of relative prices in Argentina over the last decade. The case of Argentina is attractive because the inflationary process in this country has been traditionally explained by the frequent changes in monetary and exchange rate policies (see Dornbusch and de Pablo (1988) among others). The close association between the growth in monetary aggregates and the inflation rate is depicted in Figure 1, and the role of the exchange rate is best evidenced by the adoption of a fixed rate in March 1991 --the so-called Convertibility Law-- that produced a virtually unprecedented  4 reduction in the inflation rate from about 30 % monthly in early 1991 to international levels in mid-1993.2 Therefore, price variability across sectors, if any, can be certainly interpreted as the result of nominal shocks rather than (real) supply shocks. Figure 1: Inflation and Ml, 1980-94 I 5.2 4.2 3.2 22.2 1.2 -0.8 a) a^) a) l) X) 0 0N 0| 0C 0o 0o O 00 0) 0) 0) Quarters -inflation Ml There exists substantial empirical evidence that nominal shocks have not been neutral with respect to the structure of relative prices in Argentina. Perhaps the best indicator is that the (quarterly) general consumer price inflation rate averaged about 33 % between 1981 and 1994. but ranged from only 28 % in industry to over 50 % in retail 2 There is no strong empirical evidence, however, that variations in inflation were associated with movements in monetary aggregates or the exchange rate during periods of moderate and low inflation such as the Convertibility Plan. For example, the exchange rate was fixed during this period, and the contemporaneous correlation between the inflation rate and MI only averaged 0.35 between the third quarter of 1991 and the fourth quarter of 1994 (versus over 0.90 for the period 1980-94).  5 commerce. Similar differences can be depicted in shorter periods, such as the Convertibility Plan. Table 1 reports that the variability of the CPI inflation rate across economic sectors averaged 410 % during the period 1981-94, with the highest peak in the 1986-91 years. 3 Interestingly, as in Blejer (1983), the average level of inflation appears to be related positively to the variability of relative prices Table 1: Price Variability in Argentina: 1981-94 a/ In % Period Inflation Rate b/ Variability c/ 1981.1-1994.1 33.2 410.7 1980.2-1985.2 44.4 244.3 1985.3-1986.2 d/ 8.5 71.5 1986.3-1991.1 47.5 879.8 1991.2-1994.1 e/ 1.9 30.7 Notes: a/ Based on 9 economic sectors: agriculture, industry, comunerce services, transports, construction, hotels, electricity, and mines. b/ Simple arithnmetic average c/ see footnote 3 dl So-called Austral Plan e/ So-called Convertibility Plan In spite of the above evidence, the effect of nominal shocks on relative prices has been widely ignored in Argentina. Most authors have assumed that these shocks have no real effect on the real economy. Nevertheless, the above evidence suggests 3 The variability of relative prices is calculated as follows: VR =Z w,(Ap,, - Apt)2 where Ap,, is the inflation rate of sector i, Apt is the average inflation rate, and w, the share of i in the average inflation rate. VR increases monotically with the differences betwveen individual price movements. and its lower bound is zero, when all the prices change proportionally. 4 The exceptions are Blejer (1983), and R. Frenkel in many papers.  6 that further attention must be devoted to this aspect of the inflationary process in Argentina. That is what we do in the next section. 2. Price Behavior Across Economic Sectors: An Econometric Analysis We now turn to the question of why price-setters have had heterogeneous responses to homogenous nominal shocks in a country like Argentina. Although there exist multiple causes for the variations in relative prices --as described earlier--, many of them are brought about because of the existence of some degree of imperfect competition. Microeconomists have long recognized that sticky prices and perfect competition are incompatible, because in a competitive market, a firm does not set its price, but accepts the price quoted by the Walrasian auctioneer. The existence of monopolistic power seems to be compatible with the empirical evidence in many goods and service markets in Argentina as discussed further in the text. Using the approach defined by Blanchard and Muet (1993), the inflation behavior is based on an error correction mechanism. The short-term dynamics are partially determined by the response of price setters to changes in costs, foreign prices, and demand conditions and partially by the error correction mechanism itself. That is, if actual prices are above both unit costs and foreign prices, we expect inflation to fall in the next period, which brings prices closer to cost and foreign prices.5 Also, the model assumes the existence of nominal rigidities --lags in the adjustment of prices to wages --and real rigidities such as lags in the response of wages and prices to demand conditions. This is consistent with the idea that rigidities prevent full and instantaneous price adjustments in Argentina. 5 The error component specification captures the idea that agents alter their behavior according to "signals" that they are out of equilibrium. It can be justified theoretically (within finite or infinite horizon) by quadratic costs of adjustment framework (see Hendry and Von Ungern-Sternberg 1981).  7 Equation 1 Ap,, = ao + a,Ac, + a2Ap* i + CAzx. + a4Ay-, - a5[p, i8C, (1-8) p*] with a,a2,a5 and,8 /0; a3,a4 >- or -< 0 where pi, is the price of goods/services in sector i a time t, c; unit cost, p*t the domestic price of foreign goods, yi, current excess demand and xit the expected value of future excess demand. All variables are expressed in logarithms (with A equivalent to percentage change). It is worth underscoring that the changes in monetary policy are captured by the excess demand in the goods/services market rather than the variations in monetary aggregates because the latter cannot be discriminated by economic sectors. Such a discrimination is important because an exogenous variation in money is unlikely to spread evenly and proportionally over all economic sectors. Moreover, differences are likely to emerge between variations in current and future excess demand as argued by Benabou (1992) and Rotemberg and Woodford (1991). According to these authors, the signs of the coefficients associated with the current excess demand (a3) and with the expected value of future demand (a4) depend on the market structure: customer market versus collusive 6 market. Although their arguments have been presented elsewhere , their basic contents are briefly summarized below. The customer market theory predicts that temporarily high current demand leads firms to raise prices without fear of customer loss, while the expectation of high future demand leads them to compete harder now, to attract a clientele that can be 7 profitably exploited later. Hence inflation should increase with current excess demand (CC3 > 0) and decline with the expected value of future excess demand (a4 < 0). 6 See Rotemberg and Woodford (1991), Blanchard and Fischer (1989) or Andersen (1993). 7 Put broadly, lower prices are a form of investment in market share, assuming that the demand is relatively elastic and that the capacity constraint will not prevent any further (short-term) increase in production.  8 * Alternatively, the collusive market theory predicts that when current demand is high relative to future demand, the incentive for any firm to cut its price rises because it becomes more valuable to capture current sales than to maintain collusion in the future. The inflation rates should now decrease (c3 < 0) in the short-nin with current excess demand but increase in the longer-run (a4 > 0) as captured by the expected value of future excess demand. In the longer-run, the price behavior is determined as a dynamic equilibrium (if all nominal variables are growing at the same rate and there is an equilibrium on the good market):8 Equation 2 pi. - Ci= (P *i - _ ') - al - a2) APi' Rather than the price level, we prefer to present the markup of price over costs in the long-run because it allows us to illustrate the differentiated effects of the sectoral inflation rate as well as of foreign prices on markups in each economic sector. The above approach has been applied to four economic sectors using quarterly data over the period 1984-93: industry, agriculture, services, and commerce. The data was extracted from Argentina's National Accounts published by the Ministry of Economy, Argentina's National Institute of Statistics (INDEC), and the World Bank. The sectoral inflation rates were defined using the wholesale price and the consumer price indexes at three digit levels (see the statistical appendix for detail). The current excess demand in each economic sector has been defined as the deviation from sectoral GDP trend, while the expected value of future excess demand was calculated by the methodology used previously by Benabou and Rotemberg and Woodford.9 Unit costs have been proxied by the average s For simplicity, the long run equilibrium excludes the feedback effect from wages to prices. 9 The expected value of future demand, xt = E(Yj=o &iYt+j