Baxter, M., and D.D. Farr, 2001, “The Effects of Variable Capital Utilization on the Measurement and Properties of Sectoral Productivity: Some International Evidence,” NBER No. 8475.
Chen, E.K.Y., 1997, “The Total Factor Productivity Debate: Determinants of Economic Growth in East Asia.” Asian Pacific Economic Literature 18, pp. 313-328.
Cheung, Y.W. and K.S. Lai, 1993, “Finite Sample Sizes of Johansen’s Likelihood Ratio Test for Cointegration.” Oxford Bulletin of Economics and Statistics 55, pp. 313–328.
Conway, P., V. Janod, and G. Nicoletti, 2005, “Product Market Regulation in OECD Countries: 1998–2003,” OECD Economics Department Working Papers No. 419.
Corrado, C., and J. Mattey, 1997, “Capacity Utilization,” The Journal of Economic Perspectives, Volume 11, Issue 1 (Winter), pp. 151–67.
Dupaigne, M., 1998, “Capital Operating Time and Economic Fluctuations,” Recherches Economiques de Louvain Vol. 64, No. 3, pp. 243–67.
Elliott, G., T.J. Rothenberg, and J. Stock, 1996, “Efficient Tests for an Autoregressive Unit Root,” Econometrica 64, pp. 813–36.
Everaert, L. and F. Nadal De Simone, 2006, “Improving the Estimation of Total Factor Productivity Growth: Capital Operating Time in a Latent Variable Approach,” Empirical Economics, forthcoming.
Everaert, L., and W. Schule, 2006, “Structural Reforms in the Euro Area: Economic Impact and Role of Synchronization Across Markets and Countries,” IMF Working Paper 06/137.
Finn, M.G., 1995, “Variance Properties of Solow’s Productivity Residual and their Cyclical Implications,” Journal of Economic Dynamics and Control 19, pp. 1249–81.
Hornstein, A., “Towards a Theory of Capacity Utilization: Shift Work and the Workweek of Capital,” Federal Reserve Bank of Richmond quarterly 88, 2, pp. 65–86.
Lucas, R.E., Jr., 1970, “Capacity, Overtime, and Empirical Production Functions,” American Economic Review, Proceedings, Vol. 60 (2) pp. 23–7.
Rumbos, B., and L. Auernheimer, 2001, “Endogenous Capital Utilization in a Neoclassical Growth Model” Atlantic Economic Journal Vol. 29, No. 2, pp. 121–34.
Shapiro, M.D., 1996, “Macroeconomic Implications of Variation in the Workweek of Capital,” In Brookings Papers on Economic Activity 2, pp. 79–133.
Shimotsu, K. and P.C. Phillips, 2005, “Exact Local Whittle Estimation of Fractional Integration,” The Annals of Statistics, 33, 4, pp. 1890–1933.
Prepared by Francisco Nadal De Simone. I thank Robert A. Feldman for his insights, the CPB, De Nederlandsche Bank, and the Ministry of Social Affairs and Employment for their helpful comments, Mostafa Tabbae for kind assistance with the data, and Susan Becker for her able data management support.
TFP growth in Zhou (2003) and Bell (2004) was calculated by applying growth accounting to a Cobb-Douglas technology with constant returns to scale, but without adjusting the stock of capital for the intensity in its use as is done, and later described, in this paper.
DNB (2005) concluded that, to bring the fall in productivity growth to a standstill, several measures should be taken, including enhancing the knowledge and training level of the labor force and further market liberalization.
This inconsistency prompted Lucas (1970) to argue that one possible reconciliation could lie in allowing for cyclical variation in the utilization of the capital stock.
This study uses the most recent national accounts available. In the period 1977–2004, average annual growth is about 0.2 of a percentage point higher in the most recent set of national accounts.
That movements in the (utilization-adjusted) capital stock can be characterized by a relatively long “cycle” has been observed in several other advanced economies, e.g., France (Nadal De Simone, 2003), and is most likely associated with specific technical factors such as the average obsolescence of the capital stock and time-to-build arguments (Prescott, 1982).
The reason why it can be surprising is that time series for individual countries often show less variation than cross-country data. Here, because estimated TFP is a “statistically clean” series, this may be less of a problem.
There are alternatives to this specification. For example, some specifications assume that labor and capital utilization do not move in tandem, as when capital is subject to a variable depreciation rate (e.g., Greenwood, Hercowitz, and Huffman, 1988).
There is an average difference of about 0.4 of a percentage point between measuring labor services by using hours worked and measuring labor services by using labor participation, structural employment and labor force growth. Thus, the estimated trend growth could be viewed as an upper bound, with the lower bound about 0.4 of a percentage point relatively lower.
Changes in the rate of capacity utilization with new hiring proxies changes in the use of the capital stock when the number of machines brought on stream changes—even though there may be no changes in the duration of the workweek or the organization of labor. The rate of capacity utilization without new hiring proxies changes in the duration of the workweek over the cycle, through overtime and temporary layoffs. Even though the latter measure would be preferred to adjust the capital stock—as it does not include changes in the workforce—it is unfortunately unavailable in the Netherlands.
A CES production function specification was not attempted. While more general than the Cobb-Douglas production function, the econometric tests and the quality of the results obtained with the later simpler specification is reassuring.
This specification assumes, like in the majority of empirical studies, a constant work effort.
As indicated above, caution is necessary when using official capital stock series, which normally assume a constant depreciation rate (Burnside and others, 1995). Ultimately, however, the estimation results have a natural benchmark, i.e., that the estimated factor shares match national accounts’ factor shares.
The unit root test is the augmented Dickey-Fuller test proposed by Elliott, Rothenberg, and Stock (1996). Including a deterministic trend in the test is important when working with trending data.
The residual includes measurement errors, as discussed above in the text, but also production function specification errors and estimation inefficiencies due to multiple causes such as possible simultaneity.
See Färe and others (1994) for a detailed survey on production frontiers work; and Temple (1999) for a related survey on the empirical research on growth.
TFP growth was also estimated as an AR(2) process, and as an ARMA(1,1) process. The coefficients of the AR(2) process or the MA(1) were statistically insignificant. Importantly, using the adjusted capital stock, the sum of the autoregressive coefficients of the AR(2) process was not statistically different from 0.98, the estimated value of the coefficient of the preferred AR(1) process. Finally, neither the estimated values of factor shares nor their standard errors were affected by the choice of the order of the AR process.
As a reference, for the period 1990–98, Nicoletti and Scarpetta (2003) estimate TFP growth at an annual average of 1.6 percent. In this study, for the same period, TFP growth based on adjusted capital is 1.8 percent.
A constant added to the AR(1) process was statistically insignificant, and thus dropped. The autoregressive coefficient estimate is very similar to estimates obtained in empirical work that allows for variable labor effort or changes in physical capital utilization (e.g., Bils and Cho, 1994).
To increase confidence in the estimated TFP growth series, a test of TFP growth exogeneity proposed by Hall (1989) was performed. The results of the test based on either changes in real government military spending in isolation, or together with changes in oil prices, concluded that TFP growth is uncorrelated with variables known to be neither the causes of productivity shifts nor to be caused by productivity shifts. In contrast, there is some weak evidence of correlation between the unadjusted Solow residual and military spending. These results provide further support to the constant returns-to-scale outcome of the estimation using the adjusted capital stock.
This filter is not affected by leakage from the zero frequency component of nonstationary series. In contrast to Baxter and King’s band-pass filter, the filter does not involve the loss of observations at either end of the series, and it is consistent. The standard Hodrick-Prescott filter was not used because it is bound to introduce spurious cycles, and it suffers from end-of-sample bias (Cogley and Nelson, 1995).
The application of the filter to estimated total factor productivity growth left the series practically unchanged, as it already contained no significant cyclical component.
While some factors such as education and ICT can be important, their effects are embodied in the factor input variables.
TFP growth was regressed on a constant, the ratio of the minimum wage to the median wage, union density, gross replacement rates, and different measures of the tax wedge. The R2 of the regression is 84 percent and the residuals show no evidence of serial correlation.
That indicator was not available in the time series form of the others used in the econometric exercise performed in this paper.
A regression of the same labor market institutional indicators on the adjusted capital stock and on labor services shows a positive significant impact of the replacement rate on the former and a negative significant impact of the replacement rate on the latter. Regression results are available upon request.