Our study was conducted in the Greater Vancouver area of British Columbia, Canada, in the summer and fall of 1995. We sampled firms from records obtained from the Greater Vancouver Contacts Target Marketing database for 1994 and 1995. Because we were interested in gender differences in business practices-the major focus of our study-we sampled industries from three categories: male-dominated, female-dominated, and mixed. To obtain enough women owners, we selected all of the female-owned firms and a random sample of male-owned firms from these industries. Because we were interested in human resource issues, we chose firms that had 6 employees or more, to ensure that our questions about hiring, promotion, and other human resource practices would be relevant to the firms. Firms were excluded from our sampling frame if they were franchises, branches or subsidiaries. This resulted in a total of 574 eligible firms. Of these, 229 owners agreed to participate, resulting in a response rate of 40%. Of the 229, 141 were men and 88 were women; 204 were Caucasian, 15 were Chinese, and the rest were of other ethnicities.
Information was obtained in personal interviews and typically took between 50 and 80 minutes to complete. In the final interviewed sample, the industry distribution was as follows: manufacturing firms, 19.6%; transportation services, 4.8%; wholesale trade, 18.3%; retail trade, 13.5%; finance, insurance, and real estate, 3.0%; personal services, 6.1%; business services, 31.4%; and other industries, 3.0%.
Because we are interested in how firms' hrm practices have changed over the course of the current owners' experience, we analysed only the 145 firms that were founded by the owners that we interviewed. In our regression analysis, we analysed only the 138 that had hired employees since they were founded.
We constructed three indicators of hrm practices: two for recruiting
and one for jobs and training. Respondents were asked about their
general hiring practices, and asked to respond to a list of possible procedures.
We separated 8 of the responses into two indices: (1) formal and
impersonal hiring practices: using newspapers, employment agencies,
and school or university programs; and (2) informal and personal practices:
using recommendations from current employees,other employers, family members,
friends, and favouring people already known to the respondent. For
each index, respondents received a "1" for each practice they engaged in;
the mean was 1.89 (s.d. 0.92) for formal practices and 2.78 (s.d. 1.34)
for informal practices.
For our index of formalised job and training procedures, we used four questions: have employees been given specific job assignments, written job descriptions, a training program, and does the firm keep written personnel records. Out of range of 0 to 4, the mean was 2.86 (s.d. 0.98).
Our two major independent variables are business age and workforce size. Age was measured as the number of years since start-up, with a mean of 12.36 (s.d. 8.06). The youngest business was less than 6 months old, and the oldest had been in business 43 years. Because we felt that founders' propensities for learning would decline over their years in business, we used a logarithmic transformation of business age.
We separated employment into four categories: family members employed, full- and part-time, and non-family members employed, full- and part-time. As we discuss below, the number of family members was never very large. However, most businesses added a sizeable number of non-family employees after start-up, resulting in a somewhat skewed distribution. We also expected the effects of employment size on formalization to diminish gradually with increasing size, and so we used a logarithmic transformation of the number of non-family employees.
We measured business experience through questions on: (1) managerial experience in previous businesses, (2) business experience in the same industry, (3) previous ownership of a business, and (4) current ownership of another business. All were measured as dummy variables. We included four demographic characteristics: (1) sex, (2) education, (3) age, and (4) marital status married or unmarried.
We first describe the ownership and employment composition of the sampled firms at startup, and then examine their current composition. Next, we use multiple regression to test our two hypotheses concerning changes over time in hrm practices.
Most new ventures begin on a small scale, with owners contributing a great deal of the labor to the enterprise. In our sample, the average number of owners on the startup team was 2.25, a figure quite consistent with other studies of business startups. Of the 145 firms, 36 percent began with only one owner. Forty percent had two owners, 11 percent had three, and 13 percent had four or more. Many of the owners were only part time, however, with 8 percent having no owner initially working full time in the firm and another 50 percent having just one full time owner. Family members were represented on the founding team in only 17 percent of the ventures, and in only 5 percent of the firms was the startup team entirely from the founder's family. Thus, from the beginning, these small startups were organized by founders who were not from the same family.
Employees also were mainly not family members, at startup. Only
46 percent of the ventures began with any paid employees, and only 21 percent
of the firms with employees hired any family members. About 18 percent
of all firms had one employee, 12 percent had two, and only 16 percent
had three or more employees. Only 6 percent of all startups began
with a labor
force consisting of 50 percent or more family members. By contrast, 91 percent of the firms with employees hired at least one non-family full-time paid employee, and 80 percent of all the firms with employees hired only non-family members.
Family members were not hired as part-time workers at the start, as only 3 of the 144 firms employed any part-time family members. By contrast, 14 percent of the firms hired at least one part-time non-family member. Although we had expected to find some unpaid family members working in the startups, only 4 used any unpaid family labor, and all 7 people thus employed were working part-time. Thus, from the very start, most founders were managing a paid workforce of non-relatives
Current Workforce Composition.
At the time of our interviews, in 1995, the average age of the 145 firms was 12.4 years, and most had grown since they were founded. In contrast to startup, when firms had, on the average 1.3 paid full time employees, the average number of employees per firm was now 15.4. About 92 percent of all firms now had paid employees, in contrast to only 46 percent at startup, and two-thirds had five or more employees. In the interim, the firms had hired more family members, and 29 percent now employed at least one full-time family member, in contrast to only 10 percent at startup. More part-time family members were also employed, as 17 percent of the firms now employed such workers, compared to only 2 percent at startup.
Despite the increased employment of family members, non-family members were much more important than at startup, as many more were hired over the firms' lifetimes. Currently, only 7 percent of the firms employed more than one full-time family member, and only 4 percent employed more than one part-timer. Because so many more non-family members were hired, the proportion of the workforce that was related to the owners dropped substantially. About 29 percent employed at least one full-time family member, but in only 14 percent of those firms did family members constitute half or more of the labor force. Only 4 of the 145 firms now employed just family members in full-time positions. The percent of firms hiring part-time family members changed only slightly, from 14 to 16 percent, with part-timers constituting half or more of all part-time employees in only 9 percent of the firms.
Thus, over the life of the firms in our sample, family members were a minor portion of the startup team and the workforce. Although founders increased the number of family members they hired, as their firms grew, they hired many more non-family members and by the time of our current observations, the great majority of their employees were non-family members. Of the 42 firms with a full-time family member on the payroll, only 13 had more than one family employee, full- or part-time. We turn now to an examination of what implications employment growth had for hrm practices.
We regressed our three dependent variables on 9 independent variables:
8 indicators for number of family employees and number of non-family employees,
full- and part-time, at startup and currently; and business age.
Because we have separate measures of family employment, these equations
also permit us to test hypothesis 2's predictions regarding the effects
of a "family" orientation.
In the regression model for formalized recruiting practices, shown in Table 1, only one of the 4 indicators for family employment is statistically significant, as the number of current full-time family employees has a negative effect on formalization, as we expected from hypothesis 2. Two of the 4 indicators of non-family employment are statistically significant, as the number of startup and current full-time employees both raise formalization. None of our indicators of part-time employees have any effect, family or non-family. Finally, business age has a significant positive impact on formalization, as we predicted.
The regression model for informal recruiting practices is less powerful than the previous model. Only two of the 8 indicators of employment size have a statistically significant effect on informal procedures, and one has the opposite sign from what we expected. Number of part-time startup family employees has a very strong negative effect on current informal recruiting, a result with potentially interesting implications, but because it departs so strongly from our expectations, we resist the temptation for an ad hoc interpretation. Number of part-time non-family employees at startup also has a negative effect on current informal recruiting. Results from this second regression suggest that informal recruiting is not at all sensitive to numbers of full-time employees, family or non-family. Business age has a significant negative effect, as we expected. An inspection of the individual items found that use of personal networks was fairly constant across the entire age range, with the exception of relying on current employees the youngest firms have none and using people you know the youngest firms were more likely to recruit that way.
Our third regression model is for the extent of formalization of hrm
practices regarding jobs. The power of this equation stems entirely
from measures of the current workforce, rather than from conditions at
startup: none of the 4 indicators of workforce size at founding are
statistically significant. By contrast, 3 of the 4 current size indicators
are significant, and all 4 have the expected sign. The greater the
number of current full-time family employees, the lower the level of formalization,
as predicted by hypothesis 2. The greater the number of full-time
and part-time non-family employees, the higher the level of formalization.
Our expectations regarding business age are not supported, unlike in the
two previous regressions, because age has no effect on these hrm practices.