The authors have put together a whole mass of data for six East European countries and have presented various values relating to technical change, capital and labor elasticities, returns to scale and factor elasticity of substitution for the total economy, for industry and its various branches both on a time series and cross-section basis and also by combining the time series and cross-section data. As the authors themselves emphasize, the estimated values do not indicate any conclusive pattern.
My main objection to the paper relates to the reliability of the estimates. In the majority of the results, the values of the R2 statistics are very low. This is especially disturbing in the case of the time series analysis. On the other hand, the values of the Durbin-Watson statistics are quite high and in quite a few cases, statistically significant. The combination of these two facts seems to indicate that if the data were plotted, they would indicate a wide, and in some cases, an erratic scatter. Such a preliminary investigation of the data would indicate that much of the subsequent computation work could have been and should be eliminated.
Kyns made the aims of their paper very clear and were cautious in formulating them. There would be, therefore, no point in re-raising the general issue of meaningfulness of macroeconomic production functions. I consider this a interesting exercise and its results in terms of the levels of parameters obtained (if not their precise values) are of a considerable importance, even though for some of the countries not quite conclusive.
What I particularly liked about the approach was the endeavor to obtain macroeconomic production functions from disaggregated data. The way of using the cross-section data in order eventually to combine them with time-series analysis is a very interesting contribution and though there are problems of aggregation. I felt after very careful reading that there was no logical flaw in the construction of the method, at least not to my mind. The one question that arose for me was related to the assumptions on technical change. Five such assumptions are tried out in the paper, and two of them are based on the adoption of two rates of technical change: one for recession years and one for non-recession years (these rates being either constant or linear functions of time). I am not very clear about the use of the concept of recession years with regard to the centrally planned economies. I think that its use must imply very arbitrary choices which, therefore, raise doubt as to the meaningfulness of this approach. The other minor point is that I suggest that the term "imputed income shares" for elasticities might better not be used in the case of socialist economies, as they really do not represent incomes.
I thought that it might be wise to devote some attention to one basic problem at the "ground floor" level - that of the impact of the data used on the outcomes of the analysis. It is the real problem, as we all know, that the technical subtleties are not capable of helping us to avoid of the biases contained in the basic data.
Output in Kyns' paper is measured in two ways: by gross values of output and - in the case of Czechoslovakia - also by GNP estimates. The latter way looks doubtful for the reason that these estimates were made by using i.a. the capital and labor inputs. I wonder why they were used in spite of this. With regard to GVO the authors are right in pointing out that because of its nature it tends to give relatively high rates of technical change. I would like to add to it, that also - and this may be of considerable importance in evaluating the final results of the analysis - it should bring about everywhere a distinct slowdown of these rates because of the declining shares of material inputs in the gross material product. As this slowdown is one of the major observations resulting from the analysis, one wonders to what extent it results from the way in which output is measured.
7One would feel happier if the effect of "inflation of constant prices" would be more elaborated, as it is conceivable to think of ways of neutralizing it to some extent. I am, on the other hand, not quite clear about the way of eliminating the price distortion effects.
With regard to labor inputs either the number of employees or manhours were used as units of measurement. Particularly in the case of Hungary the use of numbers instead of manhours may have had an important effect (adverse) because of the change in the length of the working week in 1968.
Finally, the question of adjusting capital inputs for the degree of utilization raised some doubts in my mind, It would he important if capital inputs include unfinished investment, but this, as I understand, is not the case here. The use of the shift coefficient may, therefore, distort the picture rather than help, because in all cases when it is above lit represent increased productivity of capital.