Entrepreneurial firms that were recipients of venture capital financing were initially identified from the 1987-1989 editions of the Venture Capital Journal. Data regarding the relationship between entrepreneurs and their VCs were collected via a survey instrument in late 1989 and early 1990. The data for the dependent variable was then derived by noting the status of the VC-funded ventures in 1995. These ventures were then categorized in the following manner: (1) no longer in business [30 ventures], (2) private and/or marginally profitable ("living dead") [84 ventures], (3) merged or acquired [24 ventures], and (4) publicly traded or Initial Public Offerings (IPOs) [58 ventures]. The status of these ventures was ascertained by searching the following sources: Compact Disclosure, Million Dollars Directory, Ward's Business Directory, CorpTech Directory, Mergers and Acquisitions Journal, and F&S Predicast. If the venture's performance could not be determined from these sources, inquiry was made with the VCs who funded these ventures via fax or by telephone. These procedures resulted in the classification of 196 out of 205 firms that had received first round funding from VCs. While we suspect that most of these unidentified firms are no longer in business, they were omitted from further analyses.
Data for the independent variables were obtained from the survey data collected in 1989-90 from firms that had received at least first round venture capital financing. Eight hundred thirty-seven ventures were identified in the 1987-1989 editions of Venture Capital Journal. We received responses from 235 firms, for a response rate of twenty-eight percent. Thirty of these firms indicated that their financing was used for seed money or for a leveraged buy-out, and therefore, were omitted from further analyses. These procedures yielded a final sample of 205 firms. Additional data was gathered from the Venture Capital Journal to check for a possible response bias. However, no significant differences were found between respondents and non-respondents in regard to the amount of funding reported by VCs [F (1,688) = 1.53; p = .22] nor by the stage of funding [C2 (5, N = 779) = 7.672; p = .17). Respondents in this study were recipients of equity investments from VCs who's syndicate size averaged 3.1 firms. Average amount of first round funding was just over $3 million (s.d. = 2.87).
For the dismissal variable, respondents were asked "How many employees in your firm have VCs played a major role in dismissing?" Because dismissal is usually an event versus something that occurs incrementally, we dichotomized this variable for analysis purposes to indicate whether or not one or more dismissals had occurred. We the 196 observations used for hypothesis testing, there were 69 cases of dismissal.
The measures of business management advice, operational assistance, and procedural justice emerged from a set of 12 questions that asked entrepreneurs about the type and extent of advice they receive from VCs. Responses to these variables were received via a five part Likert-type scale ranging from strongly disagree (1) to disagree (2) to neutral (3) to agree (4) to strongly agree (5). The business management advice and operational assistance variables were the same as those reported by Barney et al. (1996). All indicators had factor loadings greater than 0.50. The first factor reflected business management advice (a = 0.88), the second factor reflected operational assistance (a = 0.76), and the third factor comprised the procedural justice (a = 0.71)..The procedural justice variable was designed to be a measure of global justice to examine the overall perception of fairness in the VC-NVT relationship as perceived by the NVT. Our objective was to tap the three central dimensions of procedural justice: formal characteristics, explanation or information offered, and interpersonal treatment (Greenberg, 1990). We used the same scale reported by Busenitz, Moesel, Fiet, and Barney (1997). The responses for each of the three factors above were totaled and then divided by the number of items in the factor.
There are other factors in addition to the hypothesized relationships that may impact the long term status of a venture. First, the payoff for investing in high growth potential firms is somewhat irregular (Bygrave and Timmons, 1992). More specific to this study, the flow of money to VCs is clearly cyclical. Consequently, we felt it was important to control for some of these broader, macro factors impacting VC financing. Since each firm in our sample received first round funding during the 1984-1989 era, we first develop five dichotomous variables based on the year of funding. Only variables representing the 1988 and 1989 calendar years were significant. To simplify our analysis, the 1988 and 1989 variables where collapsed into one making up our first round funding era variable. Ninety-six of the firms in our data set received first round funding during this era.
Our measure of technological differentiation was made up of three items using 5-point Likert scales ranging from "strongly disagree" (1) to "strongly agree" (5): (1) Our products/services have many unique features; (2) Our product/services are not protected by patents; and (3) We utilize R&D to differentiate our products (* = 0.725).
For our new venture team skills (NVT Skills) variable, managers were asked "How easy would it be for key mangers in your company to apply the same managerial skills in other organizations?" Another question in the same format asked about the transferability of their "technical/engineering skills." Responses to these two questions ranged from "All could easily be transferred" (1) to "None could be easily transferred" (4). These two items were then summed together and divided by two.
Some research has asserted that the top VC firms may add more value than others (Sapienza, 1987; Rosenstein et al., 1993). It has been noted that the top 13% of the VC firms managed 57% of the total pool of venture capital. However, instead of dichotomizing this variable, we left VC Firm Size as a continuous variable. Based on data gathered from the 1990 Pratt's Guide to Venture Capital Sources, the total amount of capital under the management of the lead VC was used as our VC firm size variable.
The VC board seats variable reflected the proportion of directorships controlled by the VC(s). This variable was derived by dividing the total number of seats held by VCs at the time of first round funding by the total number seats on the venture's board of directors (VC controlled board seats / total board seats).
Given the nominal nature of the dependent variable, we first show an analysis of the means across the four categories. Table 1 shows these results for each of our four hypothesized variables. The difference in means between the "still private" and "merged or acquired" classes remains remarkably close across these four variables. The most noticeable contrast across the four categories of the dependent variable involves the procedural justice and dismissal variables.
Because the dependent variable (venture status) is nominal with four reasonably
distinct classifications (out of business, still private, merged or acquired, and IPO),
and Menard (1995) suggests that generalized logits should be formed along with multiple
logits per subpopulation.
These multinomial logistic regression models are an extension of the dichotomous logistic model (Menard, 1995). Such procedures allow for the examination of the overall effect of the independent variables on the dependent variable as well as the closer examination of the reference class to the other dependent variable classifications. Since all firms in our sample were private at the beginning of the study period, we chose the "still private" category as our reference class against which we specifically contrasted the out of business, merged or acquired, and IPO firms.
The multinomial logit analyses reported in Table 2 has two parts: First, Panel A contains the Wald chi-square analysis for the overall effect of each independent variable in the model. These test are analogous to those found in a conventional ANOVA table for a balanced ANOVA design after controlling for all other effects. After controlling for the other effects, the data indicate that operational assistance (Hypothesis 1) and business management advice (Hypothesis 2) received by entrepreneurs from their VCs does not have a significant effect on long-term venture status. However, support was found for Hypothesis 3. The perceived sense of fairness in the VC-NVT relationship does have positive long term implications for the venture. Another important finding of this study evolves from the dismissal data. The dismissal of NVT members by VCs has a significant and overall negative impact on the long term status of the venture. A dismissal event increases the probably that the venture firm is not be successful in the long run.
The second part of Table 2 (Panel B) contains the
"Analysis of maximum-likelihood estimates" which gives the coefficient estimates
for each variable in the model. The three individual logits reported in Panel B of Table 2 compare the reference class (firms that are still
private) with each of the other three categories (out of business, merged or acquired, and
IPO). This procedure allows for a more fine-grained analyses for examining what
contributed to the change in status from private to out of business, being merged or
acquired, or to go through an IPO for many of the firms in our data set.
Maximum-likelihood estimates of the model parameters and directionality can be observed
from this part of the analysis.
Because the analysis reported in Panel A indicates solid overall support for procedural justice and dismissal, we can now probe the more specific contrast reported in Panel B for these two variables. The procedural justice variable is significant and positive in the out-of-business / still private contrast as predicted. However, the differences are not large enough to be significantly different in the contrast with merged or acquired firms and with those that have gone public (these results are consistent with the differences in means reported in Table 1). Taken together, these findings indicate that a sense of fairness in the VC-NVT relationship does have a significant impact on preventing a firm from experiencing unsuccessful results.
Further examination of Panel B also confirms earlier indications that dismissal of NVT members by VCs is a negative predictor of venture success. The contrast with the out of business firms is significant and positive specifying that dismissal events increase the likelihood that a firm may go out of business. The contrast with IPO firms is also significant at the 0.10 level, but negative indicating that the absence of dismissal by the VCs increases the probability of an IPO for a venture. In further probing this finding, we checked for possible interaction effects with venture income the year after they received first round funding. However, neither the main nor interaction effects with dismissal were significant (due to some missing data with the income variable, the fuller model was preferred and reported in the above tables). In sum, the above evidence suggests that a dismissal event is, on average, a negative indicator regarding the longer-term status of the venture.
© 1997 Babson College All Rights Reserved
Last Updated 06/05/98