محتوای اطلاعات استفاده از ظرفیت برای بهره وری کل عوامل

In the production function approach, an accurate output gap assessment requires a careful evaluation of the total factor productivity (TFP) cycle. We build a common cycle model that links TFP to capacity utilization and we show that, in almost all of the pre-enlargement EU countries, using information about capacity utilization reduces both the total estimation error and the revisions in real-time estimates of the concurrent TFP cycle compared to a univariate decomposition. We also argue that relaxing the constant drift hypothesis in favour of a non-linear specification helps to offset a general tendency to underestimate the TFP cycle in the last decade.

Strong criticisms have been directed at the output gap concept due to measurement uncertainty. According to Orphanides (2003), the underestimation of the output gap was responsible for the US's high inflation rate in the 1970s. Nelson and Nikolov (2003) reach a similar conclusion for the UK. Both studies attribute the output gap mis-measurement to an undetected slowdown in productivity growth. In the production function framework, which is the standard method used in institutions such as the IMF, the OECD, and the European Commission (EC; see D'Auria et al., 2010), the output gap combines the cycles of labour and of TFP. In addition to secular growth, TFP contains a strong cyclical component that is indeed difficult to measure in real-time, also because of data revisions. For almost all pre-enlargement EU countries, we show that the information included in the capacity utilization (CU) indicator improves the accuracy of concurrent TFP cycle estimates in real-time and reduces the revisions that these estimates incur when new vintages become available.

The Cobb–Douglas production function gives rise to a common cycle model for TFP and CU. Formal testing favours this common cycle model in 11 EU countries, especially with the most recent TFP vintages. Our CU indicator combines a measure of capacity utilization in industry with two business sentiment indicators for services and the building sector. We show that the use of CU data for detrending TFP yields important gains in accuracy in real-time: the standard error of the concurrent cycle estimates for 2000–2011 decreases by a proportion of between roughly 10% and 40% in the 12 pre-enlargement EU countries considered. Also the four-step revision error in concurrent TFP cycle estimates shrinks by 10–40% when CU data are used, with the exception of LU. Due to a general tendency of TFP growth to decline over the last decade for 10 European countries specifying a non-linear trend model that allows the drift to switch or change at random points in time yields a further reduction in the revision error.