Manufacturing sectorial level data and variable constructions

Pake (1996), Hagemejer (2006) and the stochastic frontier and DEA analysis offered by Cullmann and Von Hischhausen (2006). It is worth mentioning that the partial productivity measures can lead to misrepresentation of firm and sector performance (see Kathuria, Raj and Sen, 2011). For instance, improvement in labour productivity could be caused by changes in scale economies (see Mahadevan, 2004).

An investigation on the empirical productivity studies for Central Eastern European Countries (CEE) which include in their country sample Poland, reveals this not to be an intensively researched area (13 studies).²² A majority of the empirical studies employed labour productivity measures and focused on examining the effect of catch-up, trade liberalization and FDI effects.

In sum, this review indicates that there are just few studies investigating productivity in Poland and most of them apply partial productivity measures or the parametric frontier approach.

4.4. Methodology and Data for sectorial level measures analysis

4.4.1 Manufacturing sectorial level data and variable constructions

Manufacturing sectorial level data

Data was obtained from the Statistical Yearbook of Industry from the Polish Central Statistical Office (CSO). According to the Polish Classification of Activities and NACE rev.

1.1 the industry is divided into main three areas: Mining and quarrying, Manufacturing and Electricity, and Gas and Water supply. Manufacturing alone is divided into 22 sections (Table 1 in Appendix 4.3). This research was conducted between 1995 and 2007. Selecting this period allows for consistency of data as Polish manufacturing data analyses since 1995 have been made to comply with EUROSTAT‘s “Nomenclature des Activités de Communauté Européenne –NACE rev. 1.1” through a decree of the Polish Council of Ministers. Also 2007 was selected because the Polish Statistic Office

22 Sector level analyses include Monnikhof and van Ark (2002), Van Ark Bart (1999), Havlik (2004), Piatkowski and Van Ark (2004), Stephan (2004). On the other hand, firm-level analysis consider Torlak (2004), Majcen, et al. (2003), Damijan, et al. (2001), Gersl, et al. (2007), Wziatek-Kubiak, et al. (2004), Claessens, et al. (1997), Zukowsa-Gagelmann (2001), Tonini and Jongeneel (2006)

105

continue to correct and update their data two years after the publication of each study.²³ Furthermore, the Polish Statistic Office changed manufacturing data aggregation in 2008 and it is not possible to merge this data at the sectorial level. as definitions for the value of output, capital, intermediate inputs consumption and investments, comparable over time and across industries. The definition of variables and the deflator applied are provided later in this section.

Output

There are essentially three indicators to describe the output in the Polish manufacturing sector at the CSO.

This measure includes products designated for increasing the value of own fixed assets, and the changes in inventories of finished goods and work in progress. Also sold production includes the value of finished products sold (regardless of whether or not payments due were received from them), semi-finished products and parts of own production, the value of paid work and services rendered, lump-sum agent fees in the case of concluding an agreement on commission terms and full agent fees in the case of concluding an agency agreement. This variable in expressed in million zlotych.

Notes: Gross output definition according to CSO is slightly different than commonly understood. As Coelli, Rao, O’Donnell, and Battese (2005, p.156) gross output is defined as the value of the total outputs of all the firms belonging to a particular sector. Source: Statistical Yearbook of Industry from Polish Central Statistical Office (CSO) (various years)

23 However, these corrections can happen for periods even longer than 2 years. Manufacturing data for the period between 1995 and 2005 was available in hard copy which made the corrections more visible.

106

Even, if there are three different measures of output, it seems show that it is rational to adopt Gross Value Added (GVA), given that this measure is the most common in the empirical literature for Poland; which allows us to obtain a possible comparison of the TFP results (see Gradzewicz and Kolasa, 2004; Czyżewski, 2002;

Kolasa and Żółkiewski, 2004). In the general literature of TFP, there are supportive voices to use GVA as measure of output (see Diewert, 2002, Hossain and Kaunahara, 2004).

Input quantities

This subsection provides the definitions and explores concerns arising from these definitions about input quantities in the categories of capital (K) and labour (L). If output is measured as gross value added, then we are going to analyse two types of inputs such as capital and labour.

Capital

The measure of capital input has been widely discussed in the theoretical and empirical literature. As mentioned in subsection 4.2.4, this discussion has not lead to agreement with a correct measure of capital. Most studies in the Polish manufacturing sector do not use the Perpetual Inventory Method (PIM) as the Polish CSO does not provide information on the accumulated depreciation of capital on the manufacturing sector level. Hence, PIM method could not be used in the TFP computation. In this case, inputs of physical capital (K) are defined by a gross value of fixed assets (GVFA) in thousand zloty. Data for this category is available for the period between 1995 and 2007 for all divisions, sections and sectors.

Labour

According to the Polish CSO, at the sector level this labour measure might be defined in three ways: the average salary of full-time paid employees (APE), employment, as of 31 XII (EM31_) and average number of employed persons in the industry (EM) (see Table 4.5).

107 Table 4.5: Description of labour indicators

Notation Variable description

Average salary of full-time paid

employees (APE) This measure includes seasonal, temporary and part-time paid employees in terms of full-time paid employees (expressed in thousands )

Employment, as of 31 XII (EM31_)

Full-time paid employees (including seasonal and temporary employees) and part-time paid employees in their primary job without converting them into full-time paid employees (expressed in thousands )

An average number of employed persons in the industry (EM)

This measures is obtained after converting part-time paid employees to full-time paid employees excluding employers own-account workers and agents (expressed in thousands ) and Jakubiak 2002, 2006). To obtain TFP results which can be comprisable it was decided to use the average salary of full-time paid employees as the measure of labour.

Price deflator

The CSO’s dataset did not provide a consistent measure for input deflator. The common deflator was implemented as price indices of sold production of industry.

According to the CSO’s database this deflator is classified in line to NACE classification (one-digit, two-digits and three-digits). As Greenstreet (2007) and Eberhardt and Helmers (2010) pointed out, the single deflated price indicator is an inappropriate measure for all parameters such as output, investments and physical capital.²⁴

4.4.2. Sector productivity methodology

This methodology describes TFP measures on level. As the results suggest a potential problem of misrepresenting the performance of firms in the manufacturing sector through partial productivity measures, this analysis focuses on TFP measures (see Coelli et al, 2005).

24 All variables were deflated to 1995 prices level and were expressed in logarithm form for production estimation. For instance, ln (GVA) is expressed as ls. The same rule was applied to other variables.