Abstract
Undoubtedly, many researchers have sought to analyze the complexities of industry dynamics, but in many cases have been hindered by the lack of comprehensive data which track firms and corresponding wages over a significant time period. In fact, data sets which track new firms, employment and wages have only become more available to researchers in the past several years.19 Since these data quite often contain confidential information such as social security numbers and employer sensitive non-public information, public officials are sometimes reluctant to release them. For this reason, the data used here for the empirical analysis, namely the Unemployment Insurance or ES-202 data from the state of Georgia, are quite unique and virtually unexploited in a dynamic context where the unit of observation is the average wage paid by a firm. Rather than give a complete and full description of all of the data sets used, this chapter focuses predominantly on the ES-202 data. The other data sets used for this study are briefly described in Table 4–1 along with a short description of the ES-202 data.20
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
Notes
Hamermesh (1999) emphasizes the important role of the recently available linked employer-employee (known as LEE) data in learning more about the functions of both sides of the labor market and furthering our understanding of policies. Leonard (1999) stresses the point that such new data will widen our perspective regarding the role of the firm in the determination of wages.
The descriptive statistics for the relevant variables created from the other data sets [used in the regressions in the following chapters] are available in tabular format upon request from the author.
One employment level report per year may occur at a time where a firm is beginning a new fiscal period. If a firm traditionally makes employment adjustments at the beginning of this new fiscal period, which coincides with the point in time in which the firm’s employment is reported, then this may lead to biases in firm-level employment or more simply stated, a misrepresentation of the firm’s true level of yearly employment.
It is important to re-emphasize that the research here focuses on firms, not establishments. Thus, all the data on wages and employment is the summation over all plants in the state of Georgia. However, it is also equally important to emphasize that a new establishment or new plant in the state could be part of a pre-existing out-of-state firms, and would thus be counted as a new firm in this study.
Another method, which was used first to find cohorts of industry startups, grouped firms by SIC code each quarter and checked for the presence of the firm’s id number within each industry for that quarter. This method however, did not detect firms which switched SIC categories, which seems to happen with frequency for new firms in the ES202 data Thus, the firm is counted as a death when in fact it exists within the confines of another market. This may likely have posed a problem for past researchers who examined changes in aggregate industry data without information on a specific firm. Thus, to limit this bias, a group startups in one year, was merged with each quarter of the ES-202 data that had the universe of firms (as many as 250,000), and if the id number was not found then the firm was counted as a death.
An additional measurement issue that may occur is with firms, which exists outside of the state, but then introduce a plant within the state in a given year. When the firm enters the state it is assigned a new identification number and thus is included as a new firm when in fact it is a new establishment or plant. In short, while this research is focused on studying new firms, unavoidably, some new plants of existing out-of-state firms are picked up as well.
A more detailed description (SIC classifications) of all the industries comprising the particular knowledge and non-knowledge cohorts used for the study is found in the Appendix.
All wages are in real 1997 dollar figures. The CPI explicitly designed for the South, constructed by the Bureau of Labor Statistics, was used to adjust all wage figures and series in this study.
A comparison of the per capita wage paid by all proprietors in Georgia versus the United States is shown in the Appendix.
Troske (1996) gives a thorough empirical analysis of the growth and decay of firm size over the entry and exit process.
It is also worth mentioning here that these growth rates for firm level employment do not suffer from the regression fallacy detailed and discussed by Davis, Haitiwanger, and Schuh (1996), Friedman (1992), and Leonard (1986). This fallacy occurs when firms are reclassified each year into an appropriate size class. Thus, if a firm with under 100 employees in the first year grows to 200 in the second year it is reclassified as a medium sized firm. However, as Davis, Haltiwanger and Schuh (1996) point out, firms which are classified as large, for example, in the base year are likely to have just (relative to the time of measurement) experienced a recent transient increase in employment. Thus, taking a “snapshot,” this firm now appears in the data as a large firm, however, these temporary fluctuations in employment reverse themselves and the same firm is likely to experience a subsequent decrease in employment in the next period, which would push it down into the next size class. Conversely, some firms that are measured as “small” firms, are more likely to have just experienced a decrease in firm size from the last period which subsequently puts them into the new size class of small firm. Again, since these temporary decreases reverse themselves, these firms are likely to experience an increase in employment the next period which would bring the firm back to its long-run equilibrium level of employment, but this firm which has always been in the large firm-size class registers as an increase in size for small firms, and subsequently adds to the measure of small firm growth biasing overall small firm growth upward. This bias never occurs in this study because the same firm is tracked through time, and once it is considered a new small firm it remains in that cohort throughout the analysis whether or not it grows or shrinks. Thus it is never re-classified. Once a firm is classified as large or small, it stays in that size class since the purpose of this study is to track the behavior of a cohort of firms through time. The only time a firm exits the cohort is if it fails.
The peculiarity of all startups exceeding the growth rate of knowledge and non-knowledge in the first two years may be due to the influence of growth rates from other new and small firms that did not fall into either category. This result suggests that the all-new-and-small cohort includes firms, which had high growth rates but were not selected for the knowledge and non-knowledge categories.
One may raise the issue of the lack of data concerning the difference in hours worked in large and small firms. Of course, this research is based on the implicit assumption that hours worked across firm size are constant. This is not unreasonable given the lack of conclusive evidence regarding the relationship between hours worked and firm size. In fact, Abraham, Spletzer and Stewart (1999) have recently noted that data and research have not made it clear if “workers at younger establishments work more or fewer hours than their counterparts at more established concerns” (38).
Entry-level wages are computed from the year 1985 in order to capture the average wage paid in the first full year of operation for most firms. Since some firms enter in the last quarter of 1984, for example, rather than use a measure based on one quarter of observation to compute the entry-level wage, a full year of data in the following year is used to develop a more precise measure of the wage associated with that firm.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2002 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Korobow, A.K. (2002). Measurement: The Data Base. In: Entrepreneurial Wage Dynamics in the Knowledge Economy. Springer, Boston, MA. https://doi.org/10.1007/978-1-4615-1121-2_4
Download citation
DOI: https://doi.org/10.1007/978-1-4615-1121-2_4
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4613-5404-8
Online ISBN: 978-1-4615-1121-2
eBook Packages: Springer Book Archive