APPENDIX I: DATA CONSTRUCTION

This appendix describes the procedure used to construct my data set, including the method used to construct annual capital stocks and stocks of debt that are not available in the original survey data.

Data Construction

The data has been taken from the annual surveys on manufacturing establishments conducted by the Central Bureau of Statistics since 1975. An additional data set, which proves to be very useful because it contains data on capital stocks and exports, is the 1986 Census of Manufacturing Establishments. The number of establishments in the annual survey varies from 8,300 establishments in 1975 to around 14,000 in 1988, and the 1986 census has 5,830 establishments with complete capital stock data.

I select a sample of firms from those two sources as follows. As there is no data available on financial sources prior to 1981, a sample period which runs from 1981 to 1988 is used. The 1981–88 survey data include 4,400 firms with complete data for at least three sequential years of output. The census data include 5,430 firms. Merging the 1981–88 survey with the 1986 census set brings the total number of firms to 3,192. I then construct a capital stock estimate by back-casting and forecasting the capital stocks, using the capital stock from the 1986 census data as a benchmark (see below for details). Leaving out establishments that have negative and zero capital stock and outliers, and keeping firms that have at least one year of positive investments, leaves us with 2,970 establishments (data set I). The frequency of each category of firms is given in Table 1 of this appendix. This data set is used to calculate the tables in Chapter 2 and the summary statistics in Chapter 4, section 4.

A very large number of firms report zero investment for many years. I am unable at this time to determine whether reporting of zero investment is in fact a non-response or represents a real observation of very low investment. I run a logit estimation of investment where it takes a value of 1 if investment is positive, and zero otherwise, the results of which are given in Table 2 of this appendix.