Data, Methodology, and Testable Hypotheses

managerial ownership the relationship between managerial ownership and Tobin’s Q is positive in support of an alignment effect; at high levels of managerial ownership the relationship is negative in support of an entrenchment effect. The findings are consistent with the arguments of Jensen and Meckling 1976 and Stulz 1988, and shed light on the empirical research of Morck, Shliefer and Vishny 1988, McConnell and Servaes 1990, and Chung, and Jo 1996 after both analyst coverage and managerial ownership effects are included within the same model of valuation. 2. The percentage of managerial ownership is a nonlinear determinant of the number of equity analysts. We argue the result is explained by a diminishing substitution effect and a diminishing marginal value for managerial ownership that becomes negative at a sufficiently high percentage of managerial ownership. The explanation of this relationship runs parallel to the explanation for the causal relationship from managerial ownership to Tobin’s Q. Furthermore, the inflection point in the rela- tionship between managerial ownership and analyst coverage is impressively close to the inflection point in the relationship between managerial ownership and Tobin’s Q. This result serves to extend the empirical research of Moyer, Chatfield, and Sisneros 1989 and Chung and Jo 1996 by offering more insight into the interaction between internal and external monitoring forces. 3. Analyst coverage is a negative determinant of managerial ownership. This result is consistent with a substitution effect and a diminishing marginal value to external monitoring in the form of analyst coverage. The result serves to extend the empirical research of Crutchley and Hansen 1989 and Jensen, Solberg, and Zorn 1992, which do not consider the impact of analyst coverage on managerial ownership. 4. Analyst coverage, managerial ownership and firm valuation are jointly determined. 5. Our conclusions are robust to the inclusion of institutional ownership effects and to alternative methods for measuring the financial variables. The remainder of the paper is structured as follows: in Section II, the data, method- ology, and testable hypotheses are presented; in Section III, the empirical results are reviewed; concluding remarks are made in Section IV.

II. Data, Methodology, and Testable Hypotheses

Data and Models We study a sample of firms as of December 1994. The sample of firms used in the study includes all NYSE and AMEX firms which have relevant financial data on Compustat, CRSP, Analyst Concensus Estimates ACE, and Compact Disclosure databases. The total number of firms meeting these data requirements is 824. The following abbreviations are used to represent the variables employed in the study: Q 5 Tobin’s Q; NANL 5 Number of analysts making earnings estimates for a particular firm; LNANL 5 Log of the number of analysts; OWN 5 Percentage of managerial ownership defined to be the shares owned by officers and directors as reported in the firm’s proxy statement and the compact disclosure data base divided by the total shares outstanding. OWN2 5 Percentage of managerial ownership squared; Tobin’s Q, Managerial Ownership, and Analyst Coverage 369 LTA 5 Log of the total firm value measured as the book value of total assets; LEQTY 5 Log of the equity value measured as the stock price times the shares outstanding; DA 5 Total debt divided by total assets; DISP 5 Dispersion of the analyst’s consensus growth estimates as measured by the standard deviation of the estimates; RDA 5 Research and development expense divided by the total assets in the firm; ROA 5 Net Income divided by the total assets in the firm; GRTH 5 Analysts’ consensus growth rate forecast; 1P 5 The inverse of the stock price per share; SD 5 Total risk of the firm’s equity measured as the standard deviation of the market returns; NYSE 5 1 if the firm is listed on the New York Stock Exchange, 0 otherwise. INST 5 The percentage of institutional ownership 5 We develop a simultaneous-equation model with analyst coverage, managerial own- ership, and Tobin’s Q jointly determined within the model. 6 The three-equation model can be represented as: LNANL 5 ƒ OWN, OWN2, Q, LEQTY, RDA, GRTH, SD, NYSE, 1P, 7 1 OWN 5 ƒ LANL, Q, LEQTY, DA, RDA, SD, DIV, 2 Q 5 ƒOWN, OWN2, LANL, LTA, DA, DISP, RDA, ROA. 3 The variables included in the models consider prior research and model identification. Analyst Coverage Equation Managerial ownership OWN is included to capture the substitution effect between internal and external monitoring. Moyer, Chatfield, and Sisneros 1989 consider this relationship using a linear specification. However, our model includes the square of managerial ownership OWN2 to capture the possibility of a diminishing substitution effect and the possibility of a turning point in the relationship consistent with a decreasing value for managerial ownership McConnell and Servaes 1990. Tobin’s Q is included based upon Chung and Jo’s 1996 argument that Q is a measure of the quality of the firm and higher quality firms are easier to market. The exogenous variables included are: size LEQTY, stock price 1P, risk SD, growth potential GRTH, research and development RDA, and New York Stock Exchange listing NYSE. 8 LEQTY is expected to be a significant, positive determinant 5 We later include institutional ownership into the model as an examination of the robustness of the results. 6 The use of a simultaneous equations methodology has a been employed by a number of previous studies in finance. See, for example: Bathala, Moon, and Rao 1994, Jensen, Solberg, and Zorn 1992, Jalilvand and Harris 1984, Peterson and Benesh 1983, McCabe 1979, and Dhrymes and Kurz 1967. 7 We assume a log-liner functional form for variable NANL number of analysts. Since most economic variables exhibit diminishing marginal returns, we conjecture that a diminishing monitoring effect is reasonable if the marginal contribution to monitoring by analysts decreases as their numbers increase also see Chung and Jo, 1996 for similar arguments. We also assume a log-liner functional form for variables EQTY and TA as they are the most popular one for these two variables found in many prior studies. 8 The exogenous variables are the same as those employed by Chung and Jo 1996. 370 C. R. Chen and T. L. Steiner of LNANL. Presumably a large equity value will be associated with a higher trading volume that justifies the cost of the additional information acquired by the analysts. Brennan and Hughes 1991 develop a theoretical model with empirical support for an inverse relationship between the share price and analyst following. They argue that stock splits reduce the relative share price and at the same time stock splits signal a brighter future for the firm which attracts more analysts. Bhushan 1989 contends that risk increases the value of the analyst’s supply of information. Research and development RDA is included to capture the possibility that RD intensive firms are more likely to be followed by more analysts because they are perceived as higher quality firms. Chung and Jo 1996 offer arguments supportive of this reasoning. On the other hand, high RDA firms may diminish the need for external monitoring in the form of analyst coverage consistent with the free cash flow hypothesis of Jensen 1986. More specifically, if research and development serves to lower the level of free cash flow, and therefore agency problems associated with free cash flow, then the value of analyst coverage as an external monitoring force may be reduced yielding a lower level of analyst coverage. Chung and Jo 1996 similarly contend that NYSE listed firms are better known firms and perceived as being higher quality firms, therefore these firms are more likely to receive more analyst coverage. Finally, Moyer, Chatfield, and Sisneros 1989 report a positive relationship between firm growth GRTH and analyst coverage. They argue that high growth firms require the additional monitoring from more analysts. Managerial Ownership Equation We postulate that analyst coverage determines the percentage of managerial ownership. Consistent with the arguments of Jensen and Meckling 1976, who contend that both managerial ownership and analyst coverage serve as monitoring forces in the firm, we argue that a higher level of external monitoring in the form of analyst coverage will reduce the need for internal monitoring provided by managerial ownership. Moreover, we expect the magnitude of this substitution effect to decrease as the number of analysts increase. This is expected if the marginal value of an additional analyst diminishes and, conse- quently, the substitution effect is retarded. Tobin’s Q is also included as a determinant of managerial ownership. We anticipate that higher quality firms, as measured by a higher Tobin’s Q, will inspire higher percentages of managerial ownership. Consistent with the self-interest of managers, we would expect a manager’s decision to commit financial capital, as well as human capital, to a firm to be a function of the quality of the firm. The exogenous variables in this equation include size LEQTY, debt DA, research and development RDA, risk SD, and dividend policy DIV. Each variable has been used in previous research to model managerial ownership. 9 9 Although not exactly the same as managerial ownership, Demsetz 1983 and Demsetz and Lehn 1985 make persuasive arguments regarding the determinants of the ownership concentration. They argue that the opportunity for shirking should encourage the formation of a more concentrated ownership structure. However, as the firm size increases, increasing percentages of wealth are needed to achieve the same percentage ownership; therefore, as the size of the firm increases the concentration should fall. They also argue that firms which operate in risky markets are more difficult to monitor externally; as such, the higher the risk, the greater the need for a concentrated ownership, and the higher the level of concentration. However, they also contend that at sufficiently high levels of risk, a negative relation between risk and ownership concentration may result due to risk aversion. Consistent with their arguments, Demsetz and Lehn 1985 report empirically that size is negatively related and risk is nonlinearly related to ownership concentration. Tobin’s Q, Managerial Ownership, and Analyst Coverage 371 Crutchley and Hansen 1989 use the direct ownership by officers and directors as reported by Spectrum to measure managerial ownership. They find that this measure is positively related to risk SD, negatively related to size LEQTY, and inversely, but not significantly, related to research and development expense RDA. Jensen, Solberg, and Zorn 1992 find insider ownership, as reported by Value Line, to be negatively related to size LEQTY while they also argue that insider ownership is a function of the amount of dividends DIV and the level of debt DA. Size is expected to negatively impact managerial ownership because wealth constraints prevent managers from obtaining a large percentage of equity as the firm size increases. Debt is also expected to be a negative determinant of managerial ownership because higher debt firms are more intensively exposed to the monitoring process of the market. Demsetz and Lehn 1985 contend that a firm’s control potential is directly associated with the noisiness of the environment in which it operates. Their arguments suggest a more risky firm may require a higher level of managerial ownership or inside monitoring due to asymmetric information. Further- more, the higher the level of research and development expense and the higher the dividend, the lower the firm’s free cash flow and, thus, the lower the need for managerial ownership to control agency problems associated with free cash flow Jensen, 1986. Tobin’s Q Equation Managerial ownership OWN is expected to nonlinearly impact firm valuation consistent with the alignment and entrenchment effects of Jensen and Meckling 1976 and Stulz 1988 and consistent with the empirical findings of Morck, Shleifer, and Vishny 1988 and McConnell and Servaes 1990. We expect the number of analysts LNANL to positively cause Q consistent with the arguments of Jensen and Meckling 1976 that outside monitoring reduces agency costs. This is also consistent with the empirical findings of Chung and Jo 1996. Firm size, measured by total assets LTA, research and development expense RDA, profitability ROA, the dispersion of analysts’ forecasts DISP and analyst coverage LNANL were found to be significant determinants of Tobin’s Q in Chung and Jo 1996. A negative relation between LTA and Q may be expected due to the tendency for firms with higher total assets to be more diversified which, in turn, has been found to negatively impact firm valuation Lang and Stulz 1994. Both profitability ROA and research and development RDA are expected to positively cause firm valuation. In addition to Chung and Jo 1996., these findings are also reported by Hirschey 1982, Cockburn and Griliches 1988, Morck, Shleifer, and Vishny 1988, McConnell and Servaes 1990, and Hall 1993. The dispersion of analysts’ forecasts, which serve as a proxy of ex ante risk see Farrelly and Reichenstein, 1984, is expected to be a negative determinant of Tobin’s Q. We estimate Equations 1–3 simultaneously using a nonlinear-three-stage-least- squares estimation procedure. The procedure yields efficient, and unbiased parameter estimates Green, 1997. Hypotheses The simultaneous equations model allows us to test four primary hypotheses: Hypothesis 1. The number of analysts following a firm is a nonlinear function of the percentage of managerial ownership consistent with a substitution effect. At lower levels 372 C. R. Chen and T. L. Steiner of managerial ownership, managerial ownership serves to align the interests of manage- ment and outside equity holders allowing for a substitution effect between managerial ownership and the amount of external monitoring through equity analysts. At higher levels of managerial ownership, managerial ownership serves to entrench management which leads to a higher level of external monitoring from equity analysts. Hypothesis 2. The percentage of managerial ownership is an inverse function of the number of equity analysts following a firm. This is because the number of equity analysts serves as a substitute for managerial ownership in monitoring the firm. We contend that agency costs are reduced if a firm has significant monitoring by equity analysts. Conse- quently, the firm may find it less important to institute policies which motivate managerial ownership. Hypothesis 3. Both managerial ownership internal monitoring and analyst coverage external monitoring reduce agency costs and thus impact firm value. The percentage of managerial ownership is a nonlinear determinant of the valuation of the firm after controlling for the level of analysts coverage. Over low levels of managerial ownership, we expect a positive relationship between managerial ownership and valuation in support of an alignment effect. Over higher levels of managerial ownership, we expect a negative relationship in support of an entrenchment effect. We further expect the number of analysts to be a significant positive determinant of firm valuation after controlling for the percentage of managerial ownership as the monitoring effect of analysts serves to enhance valuation. Hypothesis 4. Analyst coverage, managerial ownership, and firm valuation are jointly determined.

III. Empirical Results