Capital-labor substitution in a developing country: The case of Greece: Comments and some new results

Abstract

In an article in EER, Professor T. Lianos (1975) derived estimates of the elasticity of substitution (0) between capital and labour in Greek manifacturing. The purpose of this note is to point out certain sources of errors in Lianos' estimates and present some new results regarding the value of 0 in Greek manufacturing. Assuming a CES production function homogeneous of degree one and perfect competition in all markets, Lianos derived two estimable equations [see his eqs. (5) and (6) on p. 132]. Hereafter we shall refer to these equations as model (1) and (2), respectively. Using data for the period 1958-69, Lianos obtained two sets of time-series estimates for 0. For the reader's convenience we reproduce these estimates in table 1 below. To facilitate the evalution ofthe results we also calculated the standard errors associated with the values of 0 derived from model (2) and the confidence intervals of 0 for both sets ofestimates.' By simple inspection of table 1 we observe that the differences between the two sets of estimates are quite substantial. Most values of 0 derived from (2) are very poor and erratic. On the basis of these findings one may conclude that in the case of Greek time-series data the application of (2) leads to very unstable estimates of (1. The numerical values of 0 obtained from (1) are always lower? than those derived from (2). It follows that the choice between (1) and (2) is more than academic. In particular model (2) leads to very unstable results when it is estimated with time-series data.? In spite of this very fact, Lianos does not discuss the problem of choice at all. For a number of two-digit industries the unusual values of (1 are due to the use of wrong data. In 1963 the National Statistical Service of Greece made a reclassification offour-digit industries among branches. As a result, for a number