Systematic Risk of Sample Firms


Hypothesis 1 calls for the analysis of the systematic risk (beta) of the firms involved in the ranking. Beta is a measurement of the sensitivity of a company’s stock price to the overall fluctuations in the stock market, proxied here by the Standard & Poor’s 500 Index Price for Industrial Companies. For example, a beta of 1.5 indicates that a company’s stock price tends to rise (fall) 1.5% with a 1% rise (fall) in the index price. Beta is calculated here for a 5-year (60-month) time period, ending in the ranking month. If less price history is available, beta is calculated for as few as 24 months. Month-end closing prices, including dividends received, are used in the calculation. The resulting average betas for each event month around the ranking date are presented in exhibit 3. Beta is shown to hover in the 1.25 to 1.43 range, significantly greater than the market average of 1.00. These figures support hypothesis 1 and, by direct implication, the use of risk-adjusted returns in complement to the simpler market adjustment, which would be upward-biased in this case.


Monthly Abnormal Returns


The next step in the analysis consists in analyzing the monthly abnormal returns around the respective ranking dates, using the various adjustments mentioned in the methodology section. An examination of the data indicates that a large proportion of the monthly returns appear to be on the positive side following the rankings, but inferences are not particularly easy from this type of statistic. Instead, cumulative returns should be used over the post-ranking period. These results are presented in exhibit 4.


Individual stocks are tracked for 36 months post-ranking. Two features are particularly striking in the long-term performance of the shares. First of all, the Inc.100 rankings are followed by a significant "dip", or a period of abnormally negative returns. This temporary "underperformance" is particularly evident when using the risk-adjusted returns and appears to last for about 10 to 12 months following the rankings. This, in and of itself, would tend to support the notion that a share appearing in the Inc.100 rankings may be "overbought" by investors chasing the next "hot" company, resulting in prices that are not sustainable over time. The market correction results in the negative abnormal returns observed through month 10.


Second, following these negative returns, firms tend to experience statistically significant positive abnormal returns. These returns would tend to support an alternative interpretation: that the market is actually underestimating the future growth potential of the firms listed in the rankings or overestimating their risk and, accordingly, systematically underpricing them. Strategies consisting of purchasing stocks listed in the Inc.100 rankings in the month of the ranking and holding them over 36 months thus generate raw returns of approximately 68% over the period, or some 40% in excess of what would have been expected given the level of risk assumed in the strategy. The t-test results for the most conservative of the adjustments, the market-based correction, indicate that the returns for the first 36 post-ranking months are significant at the 1% level of confidence.


These cumulative abnormal returns indicate the apparent inability of the market to properly price supergrowth stocks, leaving ample opportunities for arbitrage profits, either short-selling the list over a short horizon (about 6 months) or buying the stocks and holding them over the long term (up to 36 months).


A critical factor to consider in implementing such simple arbitrage strategies is whether or not the pattern of long-term underpricing is robust with respect to the year of the ranking. To formally test for such factor, long-term cumulative abnormal returns were first analyzed on a year-by-year basis for the same sample of Inc.100-ranked firms. The results of the analyses using unadjusted raw returns indicate that even though the pattern seems very much present in almost all ranking years, the 1985, 1986 and 1987 cohorts would not have been such great investments over the 36-month horizon considered here. These three cohort years actually include the effect of the broader 1987 stock market crash in their long-term performance.


As mentioned earlier, raw returns are not a proper measure to account for the systematic risk of these supergrowth firms. Exhibit 5 presents the same information but on a market-adjusted basis. Once again, the cohort years 1984, 1985, 1986 and 1987 seem to have been affected by the 1987 stock market crash.


Regression Analyses


The empirical analyses performed above pertained to determining to what extent simple strategies based on public information, in this case the Inc.100 ranking of the fastest growing public companies in America, could be used to generate returns in excess of what a normal risk/return relationship would require. Such deviations are interpreted as supporting the inefficiency of the market, i.e. its inability to properly price stocks characterized by high systematic risks (beta) and extremely large historical growth rates.


A final step involves the investigation of the factors that may explain the abnormal returns observed in a classic regression methodology. Possible explanatory factors for the cumulative abnormal returns over the 20 months post-ranking include the growth in sales over the previous 5 years (GSALES5), the growth in net income (GNI5), the growth in the number of employees (GEMPLOY5), the market-to-book ratio at the time of the ranking (MKBKR0), the price-earning ratio at the time of the ranking (PEM0), the owner’s salary (SALARY) or equity ownership (EQUITY) in the firm, and whether the current CEO is also the company founder (CEOFD).


Before running these regressions, it is important to determine to what extent these possible explanatory or control factors are actually correlated in order to detect the existence of a multicolinearity problem. The Pearson correlation coefficients between the factors outlined above are detailed in exhibit 6 with the degrees of confidence for rejection of H0 : the coefficient of correlation is equal to zero. As anticipated, a number of the variables appear to be correlated significantly, such as GSALES5, GNI5, and GEMPLOY5. On the other hand, SALARY and EQUITY appear to be less correlated. Both market-to-book and price-to-earning ratios have little correlation with the other factors and have not been included in the table.


With this multicolinearity problem, it will be difficult to evaluate the real explanatory power of each of the correlated variables on the long-term performance of the supergrowth firms in the sample. None of the regressions performed, with either single explanatory factors or combinations thereof, indicate significant relationships. In other words, none of the growth-related historical factors, such as past growth in earnings, sales, net income, employment, or time of ranking appear to significantly determine the stock price performance following the Inc.100 rankings. The only two forward-looking variables (market-to-book and price-to-earning ratios at the time of ranking) usually appear in the regressions with negative coefficients, indicating that the firms with the largest ratios actually did worse on the long-term performance variable, but none of these coefficients are statistically different from zero. Hypothesis 3 does not seem to be supported.



The initial objective of the study was to determine the extent to which the market is able to properly price stocks characterized by very high historical growth rates, what is referred to here as "supergrowth" stocks in deference to the more commonly known "growth" stocks. A number of arguments can be made to support possible market inefficiencies under these limit conditions. The finance literature focuses on the importance of growth rates and systematic risk in pricing shares, both factors which are likely to be difficult to evaluate for firms having experienced explosive growth in the last five years. The psychology literature, and in particular, its subset studying human inferences and its biases, highlights the tendency for individuals to diverge from pure rationality, for example by attributing larger probabilities than deserved to events relatively close in time to the present. In other words, humans may not be perfect Bayesian updaters, letting a number of biases taint their inferential processes.


In order to test market efficiency in these conditions, the returns to simple investment portfolio strategies based on public information are investigated. The portfolios consist of shares in the firms listed in the Inc.100 Ranking of the Fastest Growing Public Companies in America. These portfolios are assembled when the rankings are published and held for various periods of time. The analyses conducted here indicate that significant abnormal returns are generated for these strategies in excess of what would normally be required to compensate for the level of risk incurred. Although the tests could potentially be affected by a form of survivorship bias, supplementary cohort analyses indicate that this is unlikely to be the case here. Cross-sectional regressions were not able to single out significant explanatory factors for the long-term performance of these investment strategies.


The implications of this initial study are enormous, for both investors and issuers. If indeed the market is not able to properly price high-growth entities, a fact long theorized by growth and new venture specialists, then significant abnormal returns could be earned by simple trading rules. From a company standpoint, such inefficiencies essentially indicate that "windows of opportunity" indeed exist in the market for issuing new shares, a view again long-supported by investment bankers and issuers alike. In other words, periods of overpricing and underpricing of shares exist, justifying the recourse to, respectively, new issuances or stock repurchases.

The existence of "pockets" of inefficiency in the market in its high-growth segments puts a serious cap on the generally accepted concept of efficiency as a whole. If indeed the market is efficient under "reasonable" conditions, deviations from that norm (such as those resulting from explosive growth, bankruptcies, liquidations, major catastrophes, etc.) seem to quickly stretch the ability of the market to analyze and incorporate the new information into the prices. These delayed responses or mispricings open up the door to strategic behavior by issuers and investors alike, something most financial actors have long supported but could not be accommodated by the classical market efficiency paradigm.

Top of Page
Previous Page | Next Page
Return to 1996 Topical Index

1997 Babson College All Rights Reserved
Last Updated 4/12/97 by Cheryl Ann Lopez & Dennis Valencia

To sign-up for the Center for Entrepreneurial Studies' publication lists,
please register with the
Entrepreneurship WebTeam.