LABOUR AND CAPITAL PRODUCTIVITY IN POLAND

Session Number: 1
Session Title: Factor Productivity and Technological Change
Paper Number: 2
Session Organizer: Barbara Kondrat and Derek Blades
Discussant: Derek Blades

Paper prepared for the 26^th General Conference of
The International Association for Research in Income and Wealth
Cracow, Poland, 27 August to 2 September 2000

LABOUR AND CAPITAL PRODUCTIVITY IN POLAND
Leszek Zienkowski

For additional information please contact:

Author Name(s) Leszek Zienkowski
Author Address(es) 03-943 Warsaw, Szczuczyńska 11 ap 12
Author E-Mail(s) ZBSESEK@stat.gov.pl
Author FAX(es) (22) 617 88 84
Author Telephone(s) (22) 617 88 64

This paper is placed on the following websites: www.stat.gov.pl
www.econ.nyu.edu/dept/iariw

Leszek Zienkowski

Labour and capital productivity in Poland

This paper is composed of three parts. In the part one productivity of labour and total factor productivity in Poland by sections of NACE are estimated and analysed. Productivity is estimated as a ratio of gross value added by sections to the inputs of factors of production (employment, capital). In the second part productivity is estimated and analysed by branches of manufacturing industry. Productivity is estimated as a ratio of gross output (or more precise: value of sales) to the inputs of factors of production (employment, capital). In the third part labour productivity is estimated in a different way. It is defined as a ratio of final demand (consumption, capital formation export) in different branches to the full (direct and indirect) input of labour calculated on the base of input output tables.
Before main parts of the paper few words will be said on international comparisons of the productivity of labour. The level of labour productivity in Poland was and is distinctly lower than in EU countries. This is a decisive factor of lower per capita GDP in Poland (almost 3 times lower than in EU).

Nevertheless the gap in labour productivity narrowed in the last 15 years. Roughly speaking (the result of ICP comparisons which have been the base for estimates should be treated as rough approximation) the average labour productivity in EU was about 3 times higher than in Poland in 1985 and about 2,5 higher in 1998 (see graph 1). The difference in labour productivity between Poland and EU was lowered especially in the middle of nineties. It was the time when first results of economic reforms known as Balcerowicz program started to be visible. It should be noted that the increase in labour productivity was to large extend due to the partial elimination of hidden unemployment in enterprise sector and due to changes in the economic institutions.

Part I. Productivity by sections (1985 1998)

In order to analyse the changes in productivity in Poland in the period 1985 1998 we had to estimate the value of GDP (and gross value added) by sections as well as the value of capital stock and the level of employment according to NACE for the period 1985 1992 since the data for these years which are at the disposition of CSO have been compiled according to Polish classifications (KGN) different from NACE. A substantial statistical work was needed for this purpose.

Since statistical data relating to the stock of capital, especially capital at constant prices, are of relative low reliability (this is true not only for the period 1985 1992) we concentrate analysis on labour productivity and total factor productivity (the results of the estimates of gross capital productivity are also presented in tables and graphs). Total factor productivity was estimated as the ratio of an index of GDP/gross value added to the indexes of labour and of capital with weights equal to their average participation in value added (0,6 and 0,4 respectively) in the period 1992 1998. There is no information for 1985 1991 of those shares according to NACE but even if such data were available they should not be in my opinion used for analysis. The share of labour and of capital in value added resulted in those years from administratively regulated prices which did not show economically meaningful relations. One can question from that point of view even 1992 prices but they were already near to market clearing prices.

We are of the opinion that estimates of productivity should relate rather to the gross value added total for the economy than to GDP (the results of calculation at constant prices of the value of elements constituting the difference between GDP and gross value added seem to be always doubtful). Furthermore we share the opinion that in principle the analysis of the productivity of labour and of capital should be restricted to market sectors of the economy only. The very method of calculation of the production of non-market services at constant prices makes the results not suitable for such type of analysis (constant level of labour productivity or certain percentage of growth arbitraly attributed). This is an argument to exclude the value of non-market services from the value of GDP when labour productivity is estimated.

As far as Poland is concerned there are also certain doubts if meaningful calculation of the level and of changes in time of labour productivity in agriculture can be made. First, it is difficult to estimate the level of employment in agriculture since a number of persons work both in agriculture and outside agriculture. The results of estimates overcoming this problem are of low reliability. Second, the level of hidden unemployment in this section is so high and persistent that the results of the estimates of changes in labour productivity in agriculture should be interpreted in the different way than in other sections of the economy (it should be however noted that hidden unemployment is present also in non-agricultural sections of the economy).
Finally we have decided to show in this paper the results of the estimates (indices of real change in time) of labour productivity, productivity of capital and total factor productivity for the following categories of production:

Gross value added total for the economy (including non-market services and agriculture)
Market sections of the economy excluding agriculture
Manufacturing industry.

5. For all three aggregates, total economy, non-agricultural market sections, manufacturing industry, several periods can be distinguished.
First period cover years 1986 1988 when labour productivity was growing on the average at the level of ca 3% per year (from 2,7 to 3,3 depending on the aggregate) and total factor productivity was growing at the level of 1,2 to 1,7 depending on the aggregate. At that time there were only minor changes in the level of employment and a moderate growth of capital.
In the next period, 1989 1991 when the economic reforms started, the average growth rates of labour productivity in the economy was equal almost to zero. Labour productivity increased in non-agricultural market sections treated as total but declined in the manufacturing section where the process of restructuring was most needed and in fact was the strongest. Total factor productivity declined in that period independently an the aggregate. In manufacturing this process started already before 1989, thus before the transition period. Declining productivity in 1989 - 1991 was accompanied by declining employment and by constant moderate increase in the stock of capital (at a lower rate than in the previous period).
In the years 1992 1994 both, labour productivity and total factor productivity were increasing rapidly with the highest growth in manufacturing where the decline of productivity (both labour and total factor productivity) in 1989 1991 was higher than in other non-agricultural market section of the economy. The increase of production in those years was possible mainly as the result of the use of unutilised capacity in those enterprises, which were restructured in the years 1990 1991 as the response to the demand shocks. Since growing production was accompanied by declining employment (reduction of hidden unemployment) and only moderate increase in the stock of capital the growth rates of productivity were exceptionally high.
The situation changed in the years 1995 - 1998. High growth rates of total factor productivity (especially in the years 1995 1997) were based, as before, on relatively high growth of labour productivity, much higher than in the pretransition period but distinctly lower than in the previous period (1992 1994). In the years 1995 1998 there was a moderate growth of employment and relatively high growth of the stock of capital (high growth rates of fixed capital formation).
It seems to be obvious that one can not expect that growth rates of labour productivity of the level of ca 15% per year (in manufacturing) can last long. The possibilities of the simple use of non-utilised capacities as well as reduction of hidden unemployment which transmitted themself into very high growth rates of productivity in the years 1992 1994 became very limited after 1994.
Nevertheless the increase in production and in productivity in the years 1994 1998 was much higher than in pretransition period (for market non-agricultural sections growth of labour productivity 3.3% as compared to 2.6% and of total productivity 2.9% as compared to 1.2%). There is no statistical proof for such statement but it is almost certain that high growth of productivity was mainly the result of changes in the general economic environment, privatisation and new managerial spirit of entrepreneurs (the so called non material factors). Preliminary data for 1999 and forecast for 2000 show that in spite of certain slow down of the growth rates of production the growth of productivity is going on.
One additional point seems to be worth mentioning. Certain distortion in the results of calculation of productivity growth rates can be attributed to the fact that a number of foreign owned enterprises report losses while in fact they are profitable. In order to transfer profits out of Poland they report purchases of material inputs from abroad (e.g. from a mother enterprise) at artificial high prices. Thus not only profits but also gross value added at constant prices is affected. It seems that it my be said that growing share of such enterprises in the last years leads to underestimation of the growth rates of gross value added and of productivity growth rates in Poland. If such statement is accepted it means that actual growth rates of productivity in Poland in the period 1995 1998 were slightly higher than reported in this paper (in the national accounts those transactions could be treated as a kind of hidden economy; intermediate consumption could be corrected downwards and operating surplus upwards).

Part II. Productivity by branches in manufacturing (1991 1998)

In this part of the paper I should like to present the results of studies the aim of which was to find an answer to the question Why the growth rates of productivity differ between branches of manufacturing industry?. Productivity was defined in these studies as the ratio of the value of sales (output) to employment (labour productivity) and to the inputs of production factors (labour and capital = total factor productivity). In the latter case a very simplified solution was chosen and the same ratio of labour to capital (weights) was used for all branches. Data for the years 1991 1998 were compiled according to NACE (23 branches) and expressed (sales, capital stock) at constant 1992 prices.
The period for which calculation was performed is the period of economic growth after first years of recession connected with economic and political transformation. From 1992 onwards manufacturing production started to grow. The average growth rate of manufacturing output in the period 1992 1998 was equal to ca 9.8%.
The average growth rate of labour productivity in manufacturing measured according to the above mentioned definition was equal in this period to ca 10.7%, ranging from 37% to 4.4% depending on a branch, while total factor productivity was increasing at the rate of 8.6% ranging from 32% to 0.5% depending on a branch.
As in the first part of this paper the period 1992 1998 can be divided into two subperiods: 1992 1994 and 1995 1998. The growth rates of productivity were distinctly higher in the first subperiod than in the second one.

Manufacturing: growth rates

1992 1994

1995 1998

Labour productivity

13.2

8.9

Total labour productivity

9.8

7.7

We tried to verify the following hypotheses relating to the factors, which determine the differences between growth rates of productivity of different branches of manufacturing:

growth rates of productivity are dependent on the growth rates of capital per employee; positive correlation of the growth rates of productivity with the growth rates of the ratio of capital per employee,
growth rates of productivity are dependent on the scale of modernisation of capital (as a proxy for modernisation of capital we decided to take the ratio of the sum of investment outlays with two years lag to capital stock at the end of the period); positive correlation of the above ratio with the growth rates of productivity,
growth rates of productivity are dependent on the scale of modernisation of machinery and equipment (as a proxy the same ratio as before but relating only to machinery and equipment); positive correlation of the above ratio with the growth rates of productivity,
growth rates of productivity are dependent on the share of private sector in the value of sales; positive correlation between the average share of private sector in 1993 and in 1996 and the growth rates of productivity (for subperiods respectively data on private sector in 1993 and 1996),
growth rates of productivity are dependent on the growth rates of employment; negative correlation between growth rates of employment and of productivity (reduction of hidden unemployment and rationalization of employment increase productivity),
growth rates of productivity depends on the volume of R+D expenditures over a period of time (data on R+D available for years 1994 1998 only and only for 14 branches); positive correlation between growth rates of productivity and the ratio of the sum of the volume of R+D expenditures 1994 1998 to capital stock in 1998 and respectively between growth indices of productivity and the sum of the volume of R+D expenditures 1994 - 1998 per employee in 1998.
The results of the calculations of correlation coefficients were rather disappointing (see table).

Table. Correlation coefficients

Growth rates Factors	Growth rates of labour productivity			Growth rates of total factor productivity
Growth rates Factors	1992-1998	1992-1994	1995-1998	1992-1998	1992-1994	1995-1998
1) growth rates of the ratio capital/employment	0.184	0.446	-0.179	-0.091	-0.163	-0.427
2) scale of moder-nisation of capital	0.199	-0.011	0.283	0.192	-0.040	0.250
3) scale of moder-nisation of machinery and equipment	x	x	0.25	x	x	0.23
4) share of private sector	0.267	0.168	0.374	0.403	0.282	0.496
5) growth rates of employment	-0.606	-0.829	-0.051	-0.497	-0.761	0.102
6a) ratio of R+D to capital stock	x	x	0.74	x	x	0.69
6b) ratio of R+D to employment	x	x	0.601	x	x	0.484

Nevertheless some tentative conclusions can be drown from the results of calculation.

ˇ Differences in the growth rates of productivity between branches of manufacturing industry can not be explained satisfactorily by the factors, which we expected could give answer to our question. One can only speculate that factors which were responsible for interbranch differences in growth rates of productivity relate to technological change not connected with the modernisation of capital (mainly machinery) but to the interbranch differences in entrepreneurship and in abilities to absorb imported know-how and new technologies. The profile of managers by branches could perhaps help in finding an answer.
ˇ In the years 1992 1994 certain role in the differences in the growth rates of productivity can be attributed to the differences between branches in the speed of the reduction of hidden unemployment or, in another words to the differences in the growth rates of employment. The lower growth rates or the higher decline in employment the higher growth rates of productivity. That factor was no more important in the second subperiod, it means in the years 1995 1998. In that period certain influence on interbranch differences in the growth of productivity can be attributed to the differences in the share in production of private sector. However in the former and in the latter case the influence was rather limited (see graphs .).
ˇ The positive relations between R+D expenditures and growth of productivity (see graphs ) can be interpreted only indirectly. It should be added that expenditures on R+D were included in the calculations despite the fact that the level of such expenditures in manufacturing enterprises (growing in relation to gross value added in the last years) is very low and it is difficult to believe that those expenditures can in fact influence the growth of productivity. However if the relative level of R+D expenditures, which has been very low in Poland, could be regarded as a proxy for modern, effective entrepreneurship abilities, positive relations of R+D expenditure with growth rates of productivity confirm to same extend our reasoning presented above.

Part III. Total labour productivity 1992 - 1999

In this part of the paper I should like to present the result of the calculations of the growth of total labour productivity in the period 1993 1999 (taking 1992 as a base = 100,0). By total labour productivity we understand the ratio of the value of final demand for domestically produced commodities by divisions to the total (direct and indirect) labour inputs by divisions. For the calculations of labour productivity input/output tables 1992 and 1999 were used. To be more precise, following the approach of O. de Juan and E. Febrero total labour productivity for Poland, 1992 and 1999 has been computed as follows:

direct labour productivity has been calculated by the formula:

,

where: P _j - direct labour productivity,
            q_j - gross output,
            l_j - direct labour coefficient
            j - index of industry

total labour coefficients have been calculated by the formulae:

where: l^*= [l₁^*, , l_n^*] vector of total labour productivity by industry,
l= [l₁, , l_n] - vector of direct labour productivity by industry,
(I A)^-1 - Leontief inverse matrix.

total labour productivity by industry has been calculated as:

where: P _j^* - total labour productivity by industry.

Direct labour productivity has been computed as a ratio of gross output (in basic prices) and employment (as of end of the year). PPI price indices 1999/1992 have been used to convert 1992 output into 1999 prices.
Input-output matrices 1992 and 1999 have been used which were taken from the following sources:
Input-output matrix 1992 i/o matrix, as published by the Central Statistical Office (GUS) was adjusted by: (i) aggregation to the format consistent aggregation level of 1999 i/o matrix, (ii) import matrix has been estimated and subtracted from the original GUS matrix,
Input-output matrix 1999 i/o matrix 1995, published by the Central Statistical Office has been projected to 1999 using partial information on output, value added, exports, imports, investment and consumption, and adjusted by: (i) aggregation to the format consistent aggregation level of 1992 i/o matrix, (ii) import matrix has been estimated and subtracted from the projected matrix.

The results are presented in the table on page in which direct and total labour productivity indices are presented. It may be seen that in majority of divisions the ratio of the total to direct labour productivity decreased in 1999 in comparison with 1992 and indices 1999/1992 of direct labour productivity were higher than indices of total labour productivity. That is probably mainly the result of the growing role of imported inputs in the Polish economy. By the way this may be an argument against the use for analysis of direct labour productivity (ratio of gross output to direct labour input). However it may be noted that the relative differences between divisions do not change dramatically if we use input-output coefficients productivity from 1992 table instead of the coefficients from 1999 when calculating total labour productivity.

To analyse the impact of structural change, measured by input-output coefficients, on pattern of total labour productivity evolution the following additional computations have been performed:

hypothetical total labour coefficients have been calculated by the formulae:

where: l*^*= [l₁^*, , l_n^*] vector of hypothetical total labour productivity by industry,

l₉₉= [l₁, , l_n] - vector of direct labour productivity by industry, as of 1999

(I A)₉₂^-1 - Leontief inverse matrix, as of 1992

hypothetical total labour productivity by industry, assuming direct labour productivity as of 1999 and production structure as of 1992, has been calculated as:

where: P _j^** - hypothetical total labour productivity by industry.

In such a way we calculated the impact of the changes in the input/output coefficients (domestic inputs) on the total labour productivity. It has occurred that those changes played different role in different divisions and only slightly positive role for manufacturing as a whole (average index 241 and 237 respectively). The results of these calculation are presented in the last two columns of the table.

	Manufacturing: growth rates
	1992 1994	1995 1998
Labour productivity	13.2	8.9
Total labour productivity	9.8	7.7