DF~IF~ SEPTEMBER/OCTOBER 1995 Peter S. Yoo is an economist at the Federal Reserve Bank of St Louis. Richard II. Taylor provided research assistance. producer price index (PPI) finished goods- Capacity based measures of inflation. Since inflation is an aggregate phenomenon, their focus is Utilization and undoubtedly justified. Yet, the economic analysis that links inflation to capacity uti- Prices Within lization should apply to any product market, regardless of its size. Therefore, the relation- Industries ship between price and capacity use should also be evident in industry level data—per- haps more so. Peter S. Woo In this paper, I use two-digit standard industrial classification (SIC) industry mea- tt%tjtthe strength of the economic expansion sures of capacity utilization to explore the ~IIduring the past two years has renewed robustness of the relationship between I fears of accelerating inflation. As these capacity utilization and prices. The results fears have grown, people have turned to suggest that such measures do not have a various statistics to substantiate any signs consistently strong and simple relationship of rising inflation. Commodity prices, with each industry’s price data. wages, sales-to-inventory ratios, civilian unemployment rates and capacity utilization T;HE RELATIONSHIP rates are some of the statistics commonly used to predict the future path of inflation, BETWEEN PRICE AND These measures embody the basic idea C.APACITY UTILIZATION of supply and demand: As the demand Economists typically have used two for scarce goods increases, their prices frameworks to estimate the relationship must also increase. between prices and the strength of economic The staff of the Board of Governors activity. First is the supply curve, a relation- of the Federal Reserve System measures ship between prices and quantities. Shea capacity utilization as the ratio of industrial (1993) finds that the supply curve of several production to industrial capacity’ Since four-digit SIC industries is upward sloping: the denominator in this ratio normalizes Any increase in demand is met by a combina- industrial production by a measure of the tion of additional output and higher prices. potential industrial output of the economy Over some moderate time frame in which the ratio provides a cyclical measure of firms have finite and fixed capacity any industrial output. The Boards measure increased production then implies higher of capacity assumes that a firm’s or an indus- rates of capacity utilization, which creates try’s production capacity is fixed over some a positive relationship between price moderate time horizon, usually due to the changes and capacity utilization. quantity of the available plant and equipment The second and more common frame- stock. When firms attempt to produce beyond work is a forecasting relationship between their “normal” levels, the cost of producing capacity and inflation, Such studies include the additional output becomes increasingly Garner (1994), Mcfilhattan (1978, 1985) expensive if the firm’s production process and Finn (1995). Garner and Finn estimate exhibits diminishing returns-to-scale. The simple linear equations in which the current higher cost then translates into higher prices. rate of inflation is a function of previous See FederalReserve Measures of Most of the empirical researchers on this periods’ inflation and total industrial capacity Capacity and Uviizatioe (1978) subject use total industrial capacity utiliza- utilization rates. McElhattan assumes there is and Shapiro (1992) far discussions tion and the consumer price index (CPI) or a boundarypoint of total industrial capacity aheut the canstructian af the se/es. FEDERAL RESERVE BANK OF St. LOUIS 15 II1YIIE~ SEPTEMEER/OCTOSEB 1995 utilization, beyond which inflation increases rates to forecast price changes within or decreases, a concept analogous to the non- the context of a simple linear relationship. accelerating inflation rate of unemployment. Current price changes are functions of past Therefore, she regresses changes in inflation price changes and capacity utilization rates on previous changes in inflation and on in forecasting equations: lagged capacity utilization rates. All three of these studies find a statistically significant relationship between total industrial capacity where sr, is the monthly percentage change utilization rates and inflation. in an industry’s net output price level, the The accompanying figures show the are lagged price changes, and the ete,~s relationship between price changes and are that industay~current and lagged capacity capacity utilization for 23 two-digit indus- utilization rate, Unlike in Shea~study esti- tries and three aggregate groups: total, mates of the above relationship cannot be mining and manufacturing industries.2 The interpreted as supply curves, because capacity price changes in the figures are monthly per- utilization and price changes are equilibrium centage changes im each industry~snet output values determined by the intersection of the price level without their seasonal components. demand and supply schedules, This causes (I used regressions with 1.2 monthly dummy an identification problem because it is impos- variables to remove the seasonal component sible to determine whether prices increased from each industay~smonthly percentage because the demand schedule shifted out or price changes.) The finished goods producer because the supply schedule shifted in. Still, price index is the price index associated with many people estimate such relationships and total industrial capacity utilization rates. use capacity utilization rates as sufficient The sample covers the period of 1987-94. indicators of future price changes. Indeed, The figures yield mixed signals about the media and other popular sources of busi- the relationship between capacity utilization ness news usually promote the idea that high and prices. Total industrial and manufacturing current rates of capacity utilization indicate capacity utilization rates seem to track price imminent price pressures. changes from late 1990 to early 1993, but Most macroeconomic data series have otherwise show no obvious relationship. persistence, that is. current and past values The mining aggregate shows volatile price are significantly related, Therefore, a regres- changes, but with little connection to changes sion that attempts to estimate the relation- in capacity utilization, The 23 two-digit ship between capacity utilization and price industries show similar ambiguity Some changes should include lagged values of industries, such as paper and fabricated price changes to account for their persistence metals, show an extremely close relationship rather than attributing it all to movements between capacity use and percentage price in capacity utilization, Including past price changes. The figures for these two industries changes then allows one to estimate the indicate that capacity utilization rates and marginal information contained in capacity price changes moved in tandem from 1987 utilization about current and future price to 1994. Other industries, such as the leather changes. industry show no discernible relationships Unfortunately determining the number between capacity constraints and price of lags to include in a regression is a problem. changes. Still others, like stone, clay and Including too many lags can reduce the pre- glass products, show signs of positive co- cision of the estimated coefficients or yield movements for a portion of the sample period spurious significant correlations, whereas 11 do eat use the term icdiaien hut not for the entire sample period. using too few lags will not capture all of the persistence in the data, The Schwarz infor- when referring to iadustry data because inilotiar is ue ircrease mation criterion provides a way to capture in the overall pice level, while the amount of persistence in price changes. an increase ia ua indestry’s price I now turn to linear regressions to It weighs the gains in explanatory power level is net. examine the ability of capacity utilization against the number of additional variables FEDERAL RESERVE RANK OF St. LOUIS 16 II [~IF~ SEPTEMBER/OCTOBER 1995 included in the regression, analogous to changes are positively and significantly related an adjusted R’ measure, I use this criterion (at the 5 percent level) to previous price because Geweke and Meese (1981) found changes, with a percentage-point increase in that it outperformed most others in the the previous month’s price change associated consistency of lag-length selection. with 0.38 and 0.47 percentage-point increases I therefore estimate a linear equation in current prices, respectively The same two in which current price changes are functions groups also show positive and statistically of: previous price changes; capacity utiliza- significant relationships with contemporane- tion rates using monthly percentage price ous capacity utilization, The estimates indicate changes; and capacity utilization rates that that a percentage-point increase in capacity have had their seasonal components removed.’ utihzation is associated with a 0.04 percentage- The sample starts in 1986 and extends through point increase in prices in the current period 1994. To determine the number of lags of and just over a 0.06 percentage-point increase price changes and capacity utilization for in the long run. While the effect is significant each regression, I use the Schwarz informa- and has the correct sign, the size is an