Working Paper: NBER ID: w3665
Authors: Elizabeth Kremp; Jacques Mairesse
Abstract: In the present study, we have taken advantage of the wealth of information provided by the French annual survey of market services to construct a panel sample of data on about 2300 large firms, from 1984 to 1987, in nine selected service industries (at the four digit level of the industrial classification) . We have contrasted the average performances of firms across industries, in terms of labor productivity ratios and profitability margins, both in levels and in growth rates. We have compared these averages indicators for more or less inclusive sample definitions, going from the survey of all firms to a 'balanced' and "cleaned' panel data sample of large firms, and for the two kinds of averages usually considered in macro and micro-analyses. We, then proceeded to show that the differences across industries in average productivity and profitability are usually small when compared to the range of individual differences within industries, and have investigated to what extent the extreme variability in individual performances could be accounted for by other heterogeneity factors, besides the industry effects.
Keywords: firm performance; service industries; productivity; profitability; panel data
JEL Codes: L25; L80
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| average productivity and profitability of firms influenced by industry effects (L25) | individual firm performance varies significantly within these averages (L25) |
| individual firm characteristics and external factors significantly influence performance (L25) | average performance of firms in terms of labor productivity and profitability varies across industries (L25) |
| intrinsic factors and industry effects contribute to extreme variability in individual firm performances (L25) | extreme variability in individual firm performances (L25) |
| differences in average productivity and profitability metrics across selected service industries are less pronounced than range of individual performances (L25) | intra-industry variability is substantial (L19) |
| importance of controlling for outliers and ensuring quality of data (C80) | identification strategies involve cleaning the dataset (C55) |
| potential confounders such as firm size and service industry (L25) | affect observed performance metrics (D29) |