The Global Journal of Finance and Economics, Vol. 10, No. 2, (2013) : 139-149 THE RELATIONSHIP BETWEEN CAPACITY UTILIZATION AND INFLATION: THE CASE OF BAHRAIN Ashraf Nakibullah * and Bassim Shebeb ** ABSTRACT Capacity utilization constraint is blamed as one of the reasons of recent higher inflation in the GCC countries, including Bahrain. This paper investigates the relationship between capacity utilization and inflation in Bahrain. For the sample period 1984 2012 utilization rate did not contribute much to inflation in Bahrain; however, capacity utilization constraint did contribute to higher inflation in the last decade. JEL Classification: D24, E30, E31 eywords: Dual cost approach, shadow price, utilization, inflation, ARDL model INTRODUCTION In pursuant of the monetary policy goal of price stability of the GCC countries, the central banks of these countries have been more or less successful in maintaining the stable price level in the last three decades. However, these countries, in a period in the last decade (especially from 2004 to 2008), experienced higher inflation compared to the recent past. The exchange rate pass-through, oil price induced increased government and private spending and money supply, increased wages and rent (especially in Qatar and the UAE), worldwide rise in food prices, change in inflationary expectations, capacity utilization (supply bottlenecks) and other factors have been identified and blamed for the recent inflation (Sturm et al., 2008). Recent studies have considered the role of some of these factors in the determination of inflation in the GCC countries (andil and Morsy, 2009, and Hasan and Nakibullah, 2014). This paper considers the role of capacity utilization in recent inflation in one of the GCC countries, namely, Bahrain. We only concentrate on Bahrain due to data limitation. As pointed out in Dotsey and Stark (2005), there is a common belief since the days of eynes that when the capacity is underutilized (labor and capital are not fully employed), the economy can expand without an increase in inflation. Opposite is also true; when the capacity * Department of Economics and Finance, University of Bahrain, P. O. Box 32038, Bahrain, E-mail: anakibullah@uob.edu.bh; anakibullah@yahoo.com *** Branch Director, Arab Open University, Bahrain Branch, P. O. Box 18211, Bahrain, E-mail: bshebeb@aou.org.bh; bshebeb@yahoo.com This paper is the project number 2011/25 of the Deanship of Scientific Research (DSR) at the University of Bahrain. We would like to thank the DSR for financial support.
140 Ashraf Nakibullah and Bassim Shebeb is over utilized, economy expands with accelerated inflation. In other words, traditional theory predicts that the capacity utilization rate and inflation rate move together. This traditional view is also reflected in the Taylor rule of how the Fed determines the intermediate target for the nominal Fed funds rate (Taylor, 1993). However, this traditional view is not universally accepted; that is, it is also observed that utilization and inflation move in opposite directions (Finn, 1996, Corrado and Mattey, 1997, and Dotsey and Stark, 2005). Instead of going through this debate, this paper simply tries to understand whether capacity utilization played any role in recent inflation of Bahrain. However, this is not an easy task because capacity utilization indexes are not computed and published by any country in the GCC area. On the other hand, advanced economies publish their capacity utilization data regularly. For example, the Federal Reserve Board publishes data for the US capacity utilization. Thus, the purpose of this paper is two-fold: first, we show how we can construct a capacity utilization index for Bahrain using microeconomic theory and then, relate it to inflation. Rest of the paper is organized as follows: Section 2 introduces a theory of dual cost measure of capacity utilization developed by microeconomic theorists. This section also applies this theory to construct an empirical measure of capacity utilization for Bahrain economy. Section 3 relates the measured capacity utilization for Bahrain to Bahrain s inflation. Section 4 concludes the paper. MEASURING CAPACITY UTILIZATION FOR BAHRAIN The traditional approach uses a ratio of actual level of output (Q) to a sustainable maximum or capacity output (Q * ) as a measure of capacity utilization. That is, the traditional approach constructs an index from the ratio (Q/Q * ) as a measure of capacity utilization. However, sustainable maximum output (Q * ) is easy to define but not easy to measure that probably explains why most countries do not publish their capacity utilization data. The Federal reserve in the US, for example, defines sustainable maximum output as the greatest level of output a plant can maintain within the framework of a realistic work schedule, after factoring in normal downtime and assuming sufficient availability of inputs to operate the capital in place (Dotsey and Stark, 2005, p. 9). Since we do not have this measure available for Bahrain, we create capacity utilization index from theoretically more sound dual cost approach suggested by Morrison (1985, 1988, and 1989). Theoretical Background The dual cost approach explicitly takes into account the fixity of different inputs that may occur in the short-run production process; it also determines the firm s optimal responses under the fixity of these inputs. In this approach, fixity is the key factor that causes capacity not to be fully utilized in the short run. Basically, this approach is based on the short-run specification of cost structure and Morrison (1985, 1988) has defined the capacity utilization as the ratio of the shadow cost, C (.), to the observed cost, C( ). If inputs are under-utilized, more output can be produced at a lower cost because the shadow price of the underutilized inputs would be below its market price.
The Relationship between Capacity Utilization and Inflation: The Case of Bahrain 141 To understand how this approach creates a measure of capacity utilization, we begin with a general form of the short-run total cost function, C( ), as: C( P, Q, ) V ( P, Q, ) P, (1) i i where P i is the price of the i th variable input, Q is the observed output level, is the capital (quasi-fixed input), P is the price of capital, and V( ) is the variable cost. For practical point of view, in equation (1) we have assumed capital () is the only quasi-fixed input, otherwise we could easily modify the equation (1) by adding any finite number of quasi-fixed inputs P Z Z instead of P where Z is the k th quasi-fixed input. The shadow price of the quasi-fixed input capital () can be defined as: P V(.) /. Therefore, the short-run shadow cost function can be: C ( P, Q, ) V( P, Q, ) P. (2) i i The level of capacity utilization can be determined by the difference between C (.) C( ). Full utilization of capacity, in other words, will be recognized in the short-run if P P. However, P may not be equal to P we have cases of under or over-utilization as: If P P C (.) C(.) C (.) / C(.) 1 Quasi-fixed input is under-utilized. If P P C (.) C(.) C (.) / C(.) 1 Quasi-fixed input is over-utilized. Morrison (1985, 1988) has shown how a dual cost measure of capacity utilization (CU) can be derived by exploiting the relationship between the elasticity of cost with respect to the quasifixed input and with respect to output. For our purpose, we only concentrate on the elasticity of cost with respect to the quasi-fixed input and define the elasticity of cost ( C ) with respect to the fixed or quasi-fixed input capital () as: and ln C(.) C(.) C. (3) ln C(.) Then by exploiting the definition of shadow price (cost) of the fixed input, the dual cost measure of capacity utilization (CU) can be written as: C (.) V(.) P C(.) ( P P ) ( P P ) CU 1 1 C (.) V (.) P C (.) C (.) C. (4) Obviously, equation (4) is based on the existence (or the assumption) of constant returns to scale. Equation (4) shows that the dual cost measure of capacity utilization may be greater or less than unity.
142 Ashraf Nakibullah and Bassim Shebeb Econometric Specification For an econometric measure of the CU given in equation (4), we need a measure of shadow price of capital which is obtained from an estimated variable cost function. Thus, capital utilization rates data are created using an estimated variable cost, V( ), function. To understand how this approach provides a measure of capacity utilization, following Shebeb (2002, 2005) we have estimated a log-linear form of Cobb-Douglas variable cost function as in equation (5): V AP P Q (5) b1 b2 b3 b4 L M where V is the short-run variable cost, A is a positive constant representing state of technology, P L is the price of variable input labor, P M is the price of variable intermediate input. Based on the estimated log-linear variable cost function in equation (5) the following steps are taken to obtain a measure of capacity utilization (CU): P Pˆ bˆ ( Vˆ / ) (6) 4 ( P ˆ P ) CU 1. (7) C Equation (6) provides the estimated value of the shadow price of capital. Thus, first we estimate a variable cost function using equation (5), then the estimated variable cost function along with the estimated parameters, capacity utilization data are created using equations (6) and (7). Data: Measurement and Sources All time series data used in this study are obtained from the Central Statistical Organization (CSO), the official data sources in Bahrain. The time period covered in this study is from 1984 to 2012 (the latest data available). This time period has been chosen mainly due to the availability of data. Cost-function based capacity utilization studies make use of aggregated output (Q) and inputs which are generally identified as capital (), labor (L), and intermediate materials (M). The level of input prices, output, variable cost, and capital (quasi-fixed) stock are constructed as follows. Gross Output (Q) For all economic performances, output is measured in physical or real values. Physical (quantity) data are not often readily available, but the value data usually exist. Thus, these value data are separated into their quantity and price. We have adjusted the value data for changing prices using the appropriate price index. Thus, the output (Q) data we have used are real value of production. Capital Stock () No data on capital stock is available. However, both investment and depreciation are available. The data set on capital investments includes capital expenditures on new buildings, other
The Relationship between Capacity Utilization and Inflation: The Case of Bahrain 143 construction, machineries and equipments, land and other fixed assets. An average annual capital depreciation rate of 10 per cent in 1984 is assumed to create a benchmark (initial) capital stock for 1984. Then using annual net investment (at fixed prices), an aggregate capital stock is constructed. The perpetual inventory method is employed in accounting for capital stock with adjustment for the change in prices and depreciation rates. Labor Input (L) For labor input, the real value of compensation is used as a measure of labor input (L) to take into account the difference in skills among workers assuming that there is a strong relationship between wages and the labor s level of skills and experience. The compensation consists of all payments, both in cash and kind, and the supplement to wages and salaries. Then the price of labor input (P L ) is derived as the implicit wage deflator. Intermediate Input (M) In this study, the other-inputs (M) are defined as equal to the real value of the purchases of materials and supplies for production that include all inputs other than labor and capital. In other words, other-inputs represent the cost of all production input excluding the cost of labor and capital inputs. Empirical Measure of Capacity Utilization of Bahrain The short-run Cobb-Douglas variable cost function (5) was estimated using annual data from 1984 to 2012 for Bahrain. Since we are mainly interested for creating data for CU, we just report only parameter estimates along with their p-values in brackets. The estimated variable cost function ( ˆV ) is: ln Aˆ 1.755 (0.108); bˆ ˆ ˆ 1 0.497 (0.079); b2 0.373 (0.001); b3 0.975 (0.00); bˆ Vˆ P P Q 4.497.373.975.234 0.234 (0.018) 5.7834 L M. The estimated variable cost function in equation (8) satisfies most of the regularity conditions. Monotonicity in input prices requires the cost-share equations to be greater than zero; and the necessary and sufficient condition for the monotonicity in output is that the partial derivative of the cost function with respect to output is non-negative. This monotonicity in output is satisfied. However, the monotonicity with respect to the price of labor is not satisfied which may be due to high multicollinearity with other input prices and probably due to the special nature of Bahrain economy which depends on abundant expatriate workers. Using equations (6) (8) our measure of capacity utilization for Bahrain is presented in table 1. Results in table 1 shows that quasifixed input (or capacity utilization) in Bahrain had been severely underutilized in the 1990s as the estimated values of CU had been less than 0.5. This is not surprising. Like other GCC countries all economic activities (production, consumption and other activities) of Bahrain evolve around the oil and gas sector and fluctuate with the world oil price. During 1990s world oil prices were more or less depressed compared to 1980s or to the recent past and capacity utilization rates were very low. (8)
144 Ashraf Nakibullah and Bassim Shebeb Table 1 Capacity Utilization of Bahrain, 1984 2012 Year CU Year CU Year CU 1984 0.5667 1994 0.4302 2004 0.6169 1985 0.5535 1995 0.4247 2005 0.6651 1986 0.5403 1996 0.4258 2006 0.6218 1987 0.5614 1997 0.4295 2007 0.5689 1988 0.5498 1998 0.4305 2008 0.5788 1989 0.5439 1999 0.4331 2009 0.5718 1990 0.4309 2000 0.4469 2010 0.5643 1991 0.4197 2001 0.4560 2011 0.5159 1992 0.4433 2002 0.6011 2012 0.5301 1993 0.4240 2003 0.5889 One interesting feature of the CU is revealed in figure 1 where CU along with its smoothed long-term trend component are plotted using the HP Filter. As we see from figure 1, capacity utilization rates were lower than its long-term trend for most of the sample period. However, utilization rates had been above its long-term trend in recent times (especially during 2003 2008) when inflation rates soared in all the GCC countries including Bahrain. Figure 1: CU Along its Long-term Trend (HP Filter) UTILIZATION RATES AND INFLATION IN BAHRAIN The traditional eynesian theory postulates that when economy produces les than full employment GDP, the policy induced increase in aggregate demand would not raise wages and as a result cost of producing more output would not require prices to increase. However, with increase in aggregate demand, more and more industries would approach to full employment or capacity output and with further increase in demand with the full employment of labor, capital and other inputs prices would increase at an accelerated space.
The Relationship between Capacity Utilization and Inflation: The Case of Bahrain 145 The GCC countries, including Bahrain, had experienced a higher inflation in recent years when the average inflation for the region increased from 1.5 per cent in 2003 to about 7 per cent in 2007. Capacity utilization is identified (without evidence) as one of contributing factors of the recent inflation (Sturm et al., 2008). Figure 2 plots the CU and inflation rates (measured by CPI) of Bahrain for the period 1985 2012. Figure 2 shows a somewhat clear movement of utilization and inflation in the same direction only early few years of the sample period and the periods of the last decade. The relationship between these two measures is not clear in the 1990s when Bahrain experienced both a low inflation (about 3% on average) and severe underutilization of capacity. Even when the two series moved in the same direction in the last decade, we see periods when inflation reached highest levels in 2007 2008, capacity utilization did not follow in the same pace. Inflation increased from 2% in 2006 to 3.3% and 3.5% in 2007 and 2008, respectively whereas utilization rate fell by 5% in 2007 and increased by only 1% in 2008. Figure 2: CU and CPI Inflation of Bahrain, 1985 2012 Some studies also compared utilization rates with inflation measured by GNP or GDP deflator (Gittings, 1989). Figure 3 examines the relationship between utilization and inflation of Bahrain measured by the GDP deflator. Comparing two figures the similar pattern emerges, but the relationship seems to be more discernible using the CPI inflation in figure 2. This is also understandable. Inflation measured by GDP or GNP deflator is not an accurate measure of inflation for oil-based economies. Oil price dominates the GDP deflator for oil-based economies and with its wild swings inflation changes without much attention to utilization rates. Thus, for Bahrain we concentrate on CPI inflation for the rest of the paper. Empirical Evidence The traditional or original theory postulates a relationship between the price level and utilization. However, the modern theory links inflation with utilization (Dotsey and Stark, 2005). Following
146 Ashraf Nakibullah and Bassim Shebeb Figure 3: CU and Inflation Measured by GDP Deflator, 1985 2012 Dotsey and Stark (2005), we ran a Granger-causality test for the entire sample based on the following equation: n c CU t i t i j t j t i 1 j 1 n (9) where inflation rate ( t ) is defined as t ln( CPIt / CPIt 1) *100 and capital utilization (CU t ) is also in percentage. We ran regressions using equation (9) with different lags and settled with two lags for both inflation rate ( t ) and utilization rate (CU t ) based on Akaike information criteria (AIC). Standard errors are corrected for heteroskedasticity and autocorrelation using the methodology of Newey and West. Thus, standard errors reported are heteroskedasticity and autocorrelation consistent (HAC) standard errors. Results are presented in table 2. The coefficients, 1 and 2 for CU t 1 and CU t 2, respectively are significant at 5% and 10% levels, implying that capacity utilization helps to predict inflation for the entire sample. Standard Granger causality test that utilization does not Dependent Variable = Inflation rate ( t ) Table 2 Estimates of Equation (9) Variable Estimate HAC Standard Error t-statistic P-Value Intercept 0.188 1.058 0.178 0.861 t 1 0.991 0.288 3.440 0.003 t 2-0.351 0.193-1.818 0.083 cu t 1-0.078 0.036-2.177 0.041 cu t 2 0.090 0.046 1.944 0.065 R 2 = 0.637 2 (2) = 4.750 (0.093)
The Relationship between Capacity Utilization and Inflation: The Case of Bahrain 147 Granger cause inflation produced test statistic 2 (2) = 4.75 (p-value equal to 0.093) for the hypothesis 1 = 2 = 0. Thus, we reject the hypothesis that utilization does not cause inflation only at the 10% level. Thus, there seems a weak evidence that utilization rates helps to predict inflation of Bahrain for the entire sample. Figure 1 shows that utilization rates had been above its long-run trend for the most part of the last decade starting from 2002. Figure 2 also shows that the two series utilization and inflation mostly trended together in the last decade whereas such trend is not clear in the 1990s. Such pattern or even the movements of these two series in opposite directions for some periods are also observed for other countries (Finn, 1996). The traditional eynesian theory explains why the effect of utilization on inflation might vary over time. Referring to the suggestion of the basic eynesian theory, Dotsey and Stark (2005, p. 13 14) point out that the weakest link between capacity utilization and inflation occurred at very low utilization rates, while the strongest link occurred at very high utilization rates. For the former, we would expect that when utilization was below some threshold, utilization rates would rise with no change in inflation. For the latter, we would expect that when utilization rates were above some threshold, changes in aggregate demand would bring about big changes in inflation. To test this theory especially for the period 2002 to 2012 when Bahrain experienced a higher inflation, we estimated a form of the following autoregressive distributed lag (ARDL) model: n c CU u. t i t i j t k t i 1 k 0 n (10) Again in choosing the number of lags for both inflation rate ( t ) and utilization rate (CU t ), we tried with different lags and settled with ARDL(1,0) based on AIC and BIC criteria. Standard errors reported are again heteroskedasticity and autocorrelation consistent (HAC) standard errors. Results are for the whole sample periods 1985 2012 and 2002 2012 presented in table 3. Though the estimated coefficient 0 for CU t is positive for the whole sample but its value is negligible and most importantly it is not statistically significant at any standard level of Dependent Variable = Inflation rate ( t ) Sample Period 1985 2012 Table 3 Estimates of Equation (10) Estimate HAC Standard Error P-Value R 2 Intercept 0.301 1.194 0.803 0.512 t 1 0.833 0.194 0.000 CU t 0.011 0.018 0.562 Sample Period 2002 2012 Intercept 9.526 4.338 0.059 0.537 t 1 0.660 0.269 0.039 CU t 0.169 0.072 0.046
148 Ashraf Nakibullah and Bassim Shebeb significance. However, coefficient 0 for the recent period of 2002 2012 is positive and it is statistically significant at the 5% level. It indicates that a one per cent increase in utilization rate would increase inflation rate about 0.17% in the short run. These results do support the basic eynesian theory mentioned above. We may also conclude that capacity utilization did somewhat contributed to the recent higher inflation in Bahrain. CONCLUSION The relationship between capacity utilization and inflation is based on a simple idea that when capacity is unused, competition among producers would be low and that would hold prices down; on the other hand, prices increase when capacity constraints are reached and the competition among the producers are high. Such traditional ideas led some studies to presume (without empirical evidence) that capacity utilization constraint, among other factors, has contributed to the recent higher inflation in the GCC countries. Due to data limitation for other GCC countries, this paper provides empirical evidence only for Bahrain. One of the main problems of conducting such study for the GCC countries is, unlike other industrial countries, utilization data are not readily available for the GCC countries. Thus, in this paper we first construct capacity utilization data for the period 1984 2012 using the dual cost measure of capacity utilization for the economy at large. The approach of dual cost measure is based on sound microeconomic foundation and is applied in some industries. Data for Bahrain show that capacities were severely underutilized in 1990s but increased and trended in the same direction with inflation during the periods in the last decade when Bahrain, along with other GCC countries, experienced higher inflation. Granger causality test indicates that utilization rates help to predict inflation of Bahrain for the entire sample. However, our main empirical result is that the capacity constraint contributed to the recent higher inflation in Bahrain which also supports the basic eynesian theory that the link between capacity utilization and inflation is the strongest when utilization rates are high. References Corrado, C., and J. Mattey (1997), Capacity Utilization. Journal of Economic Perspectives, 11, 1, Winter, 151 67. Dotsey, M., and T. Stark (2005), The Relationship Between Capacity Utilization and Inflation. Federal Reserve Bank of Philadelphia Business Review, Q2, 8 17. Finn, M. G., (1996), A Theory of the Capacity Utilization/Inflation Relationship. Federal Reserve Bank of Richmond Economic Quarterly, 82, 3, Summer, 67 80. Gittings, T. A. (1989), Capacity Utilization and Inflation. Federal Reserve Bank of Chicago Economic Perspectives, May/June, 2 9. Hasan, M., and A. Nakibullah (2014), Price Level and Inflation in the GCC Countries. Working Paper No. 2011/26, DSR, University of Bahrain. andil, M. and H. Morsy (2009), Determinants of Inflation in GCC. WP/09/82, IMF, Washington DC. Morrison, C. J. (1985), Primal and Dual Capacity Utilization: An Application to Productivity Measurement in the U.S. Automobile Industry. Journal of Business and Economic Statistics, 3 (4), 312 324.
The Relationship between Capacity Utilization and Inflation: The Case of Bahrain 149 Morrison, C. J. (1988), Capacity Utilization and Productivity Measurement: An Application to the U.S. Automobile Industry. In A. Dogramaci (ed.) Studies in Productivity Analysis, Boston: luwer Nijhoff Press, 163 193. Morrison, C. J. (1989), Unraveling the Productivity Growth Slowdown in the U.S. and Canada, and Japan: The Effects of Subequilibrium, Scale Economies and Markups. NBER Working Paper No. 2993. Shebeb, B. A. (2002), Productivity Growth and Capacity Utilization in the Australian Gold Mining Industry: A Short-Run Cost Analysis. Economic Issues,7 (2), 71 81. Shebeb, B. A. (2005), Measuring Capacity Utilization Using A Short-Run Cost Function: An Application To Bahrain Economy. Journal of the Gulf and Arabian Peninsula Studies, 31(118), 31 51. Sturm, Michael, Jan Strasky, Petra Adolf, and Dominik Peschel (2008), The Gulf Cooperation Council Countries Economic Structures, Recent Developments and Role in the Global Economy. Occasional Paper Series No. 92, July, European Central Bank. Taylor, J. (1993), Discretion Versus Policy Rules in Practice. Carnegie-Rochester Conference Series in Public Policy, 1993, 195 214.