The sample used for this study contains 128 non-financial companies that initiated their IPO in 1988. A total of about 250 firms filed securities registrations with the Securities and Exchange Commission (SEC) to conduct an IPO, and the list was pared down to 170 by deleting those that were listed as closed-end funds, real estate investment trusts, and other firms not producing a good or service. Upon receipt of the 170 prospectuses, those firms in the "not producing a good or service" category were dropped. As a result, the final sample, after deleting cases based on missing data, includes a total of 128 non-financial companies. The year, 1988, was chosen because it allowed for five years of follow-up data, and because it had a 60% survival rate as of 1993, permitting a comparison between survivors and non-survivors.
Data were gathered from the prospectus of each firm. The prospectus is the document provided to the Securities and Exchange Commission (SEC) prior to the public offering, and it is also the document circulated by the underwriter to assess demand for the firm's stock. The SEC requires firms to follow strict guidelines in the format. The document itself is usually written by members of the management team, investment bankers, and lawyers for both parties; it is then scrutinized by other lawyers and accountants. While the potential for positive bias exists in the prospectus, the firm is liable for any information that might mislead investors (O'Flaherty, 1984). The Securities and Exchange Act of 1934 (with amendments) sets the requirements for the prospectus, thus assuring consistency in the type of information that is included in the document. The SEC also requires that the prospectus be accurate to the best knowledge of management. Given this requirement and the fact that the SEC requires a tremendous amount of detail regarding company operations, the prospectus is a useful data source (Marino, Castaldi, & Dollinger, 1989).
Data Collection and Coding
Prospectuses from 1988 are not readily available in public sources, therefore, copies were obtained from Disclosure, which is a data clearing house for the Securities and Exchange Commission. Data were coded using a two-step process with two coders who were unaware of the survival status of the companies as they coded. First, a five-page summary of each prospectus was constructed. Given the fact that the prospectus is not a traditional data source, this first step allowed for careful reading of each document, cross checking at the second stage of coding, and notation of any unusual firm characteristics. The second step involved numerically coding each five-page summary for specific information. Researchers cross-coded a sample of companies (two people coded the same prospectus), and they switched companies for the second stage of coding. Any questions about codes were resolved through group consensus, which involved meetings with the coders and an additional researcher.
The average age of the firms as of 1993 was 11.17 years, with a standard deviation of 12.12. The median was 8 years. Half the firms employed fewer than 110 workers, however, 20% had 700 or more employees. The firms had an average of 6 executive officers and 3 outside directors. The businesses were located throughout the United States but were most heavily concentrated in the Pacific states. The sample includes companies from numerous industries, ranging from food service retailing to biotechnology to steel minimills. As of 1993, 77 firms still survived (60% survival rate).Independent Variables
CEO ownership is coded as the percentage of the company owned by the CEO after the initial public offering. The mean is 15.64 percent, with a standard deviation of 17.79, and the median was 10. The percent ownership ranges from 0 to 89. The number is logged in order to correct for skewness. Top executive team ownership is coded by including the total percentage ownership of all individuals listed in the prospectus as part of the top mangement team, and this number includes CEO ownership.1 Given that ownership proliferation is the focus of the study, this variable shows the extent of total management team ownership. The mean for executive team ownership is 38.33 percent, with a standard deviation of 22.39. The median is 38.90 per cent, and the range is between 0 and 89 per cent.
All employee ownership is coded as a dichotomous variable, and companies that have incentive stock options for all employees (including the CEO and management team) or employee stock ownership plans for all employees are coded as a 1, and those without such plans are coded as a zero. The mean is .61 with a standard deviation of 49 (See Tabel 1).
All firms still in business at year-end in 1993 were coded as survivors. Survival status is not easily determined, therefore, several steps were taken to assure correct identification of survivors. First, an on-line data base of current public firms was searched to find current information on the companies. Supplemental information was gathered from Disclosure, a data clearing-house for the Securities and Exchange Commission (SEC). Disclosure was able to identify many of the active and inactive companies. The Directory of Obsolete Securities (1994) also was searched to identify bankruptcy, name changes, recapitalization, and mergers. In addition, phone calls were made to the numbers provided in the prospectus. Fourteen firms (10%) changed their names. These firms were called to find out whether the name changes were cosmetic, or whether the businesses had undergone other major transformations. Mergers were considered nonsurvivors under the logic that the firm, as coded in 1988, had been joined with another set of management and organizational culture (Aldrich & Marsden, 1988; Kalleberg & Leicht, 1991). In addition, the stock price of the mergers was tracked, and 7 out of 8 mergers had stock prices that had decreased prior to the merger. Eight mergers (6%) occurred among the IPO firms. Overall, 77 companies, or 60% of the sample, were coded as survivors.
Each companys year end stock price (from 1989 to 1993) was obtained from Disclosure. Given that these companies went public at different times (with some finalizing in 1989), year end 1990 was chosen as the first year for analysis in an effort to equalize all firms.Control Variables
Based on a review of the initial public offering and strategic human resource management (due to risk hypotheses) literatures (e.g. Beatty & Zajac, 1994; Huselid, 1995) several control variables were used in the analysis. The total number of employees, logged to correct for skewness, was included as a measure of size. Net profit at the time of the IPO was added as a performance measure. Net profit was used as a measure of performance because many firms had net losses reported, which indicated variance in performance. Other measures of performance, such as sales, presented the problem of zero sales for many organizations in the sample. A dichotomous measure coded "1" for service industry and "0" for manufacturing was used to control for industry. This code was limited to manufacturing versus service firms in the analyses reported due to sample size restrictions. For the stock analysis, an additional control variable, year end stock price 1989, was included.
A final control variable indicated the level of risk of the firm. Each prospectus contains a section listing all risk factors faced by the firm, which must be disclosed to meet the requirements of the Securities and Exchange Commission. Prior research on initial public offering firms found that this measure was a useful way to code risk (Beatty & Zajac, 1994; Rasheed & Datta, 1994). The presence of the following risk factors were included in this measure: technological obsolescence, new product, few or limited products, limited number of years in operation, inexperienced management, technical risk, seasonality, customer dependence, supplier dependence, inexperienced underwriters, competition, legal proceedings against company, liability, and government regulation . The summated risk measure ran from 1 to 11, with a mean of 4.18 and a standard deviation of 1.80. Table 1 includes the means and standard deviations for all of the dependent variables and control variables. Table 2 includes the correlation matrix for all variables included in the analyses. (See Table 2)
1Separate analyses were run for CEO and executive team ownership, and the results are noted in Table 4.
Previous Page | Main Menu | Next Page
© 1997 Babson College All Rights Reserved
Last Updated 4/27/97 by Germaine Wong
To sign-up for the Center for Entrepreneurial
Studies' publication lists,
please register with the Entrepreneurship WebTeam.