Output Changes and Employment Changes
To illustrate the close connection between changes in employment and changes in real output, we fitted a series of multiple regressions. In annual data extending from 1971 to 2002, the percent change in aggregate LFS employment was regressed on the current and lagged percent change in real GDP. Positive coefficients were expected on both output change variables while the intercept was expected to be negative, reflecting the effects of productivity growth. Relationships were fitted for all countries in the OECD-20 group and for nine East and South Asian countries.
For most countries (25 of 29) the estimation was based on at least 29 observations. Only New Zealand, Indonesia, and Malaysia had fewer than 25 observations (16, 18, and 21, respectively). Table B.1 shows the estimation period for each country. Several countries had one major break in the employment series, explaining why the estimation periods skip one year.
The principal data source for the OECD-20 countries was the OECD publication Labor Force Statistics (OECD 2001b). Earlier editions were used to extend the data series back to 1970. In several instances the OECD has adjusted these data for breaks. For the Asian countries, the data came principally from information developed at the U.S. Bureau of Labor Statistics (BLS), Division of Foreign Labor Statistics. A file was developed at BLS that extended from 1970 to 1995 for several countries. These data were updated with information taken directly from country LFS publications extending through 2002.
Alternative data were also explored. For both the OECD-20 and Asian countries the ILO Web site had LFS employment data that can extend back to 1969. The member countries report these data in different ways, and there are more breaks in these data than in the OECD data. The BLS Division of Foreign Labor Statistics supports files of annual LFS data for 10 OECD countries. These data are available from 1959 using the country's own LFS concepts and using U.S. concepts.