Share performance following severe decreases in analyst coverage.

Fortin, Rich ; Roth, Greg

INTRODUCTION

Earlier researchers, such as Chang, Dasgupta, and Hilary (2006), argue that security analysts likely help to mitigate information asymmetry between managers and outside investors by: (a) synthesizing complex information for less sophisticated investors; and (b) making private information available to the public. (Examples of private information include that gained from firm visits and, prior to enactment of Regulation Fair Disclosure, discussions with top managers.) Chang, et al., and several other studies provide evidence that analyst coverage is negatively associated with information asymmetry (Hong, Lim, and Stein, 2000; Gleason and Lee, 2003).

Other researchers provide evidence that, especially when confronted with complex information, analysts make important recommendation and forecasting errors that do not reduce information asymmetry (Gilson, 2000; Louis, 2004; Feng, 2005; and Shane and Stock, 2006). Even worse, analysts have come under heavy criticism in recent years for allegedly issuing intentionally biased recommendations or biased earnings forecasts in order to gain lucrative brokerage or underwriting fees for their firms. Evidence to support the claim that conflicts of interest lead analysts to intentionally biased recommendations or forecasts is provided by Lin and McNichols (1998), Michaely and Womack (1999), and others. Also, evidence that analysts tend to disproportionately cover firms that they view favorably is provided by McNichols and O'Brien (1997), Rajan and Servaes (1997), Bradley, Jordan, and Ritter (2003), and Cliff and Denis (2004). Building on studies that highlight analysts' economic incentives for providing firm coverage, Doukas, Kim, and Pantzalis (2005) state that analysts increase coverage for certain firms in anticipation of greater underwriting and brokerage business. In turn, firms receiving high analyst coverage experience high investor demand for their stocks, resulting in overvaluation. Doukas, et al., find that firms receiving high analyst coverage have overvalued stocks that subsequently experience low future returns. They also find that firms receiving weak analyst coverage have undervalued stocks that subsequently experience high future returns.

We add to the literature on analyst behavior and security prices by examining shareholder reactions to severe losses in analyst coverage. There are two main reasons for analysts to drop existing coverage of a firm. First, analysts may conclude that the firm is no longer a good prospect for generating future income (through brokerage and underwriting fees) for the analyst's firm. Second, analysts may become pessimistic about the firm's future share performance and would rather drop coverage than issue a sell recommendation. These motivations are not mutually exclusive and brokerage firms rarely give public explanations for dropping coverage of a firm's stock. Therefore, shareholders are left alone to infer the information content of dropped analyst coverage. If shareholders believe that analysts generally drop coverage because they have private, negative information that they choose not to reveal through a sell recommendation, then shareholders would interpret dropped coverage as "bad news." This bad news would likely motivate many shareholders to sell their shares in the firm. A severe decrease in analyst coverage of a firm might lead to shareholder overreaction and security undervaluation because shareholders fear that analysts have chosen to drop coverage rather than to issue sell recommendations. If shareholders initially overestimate the role of private, negative information in analysts' decisions to drop coverage, then an initial mispricing caused by shareholders' overreaction to dropped coverage would only be corrected over time as the feared bad news fails to materialize. Under this scenario, positive abnormal returns would be earned in a period following dropped analyst coverage.

Our primary research question is whether firms suffering severe losses in analyst coverage subsequently earn abnormal returns consistent with investor overreaction and security mispricing. We gather a sample of firms from the period 1988-2002 that experienced more than a 50% loss in analyst coverage during a single calendar year. We then calculate abnormal returns for the first 60 trading days in the year following coverage loss. Abnormal returns are calculated using the Fama-French (1993) three factor model, plus an adjustment for momentum. On average, firms suffering severe losses in coverage during the prior calendar year earn positive abnormal returns of 11.6% during the first 60 trading days of the current calendar year. This evidence of abnormal performance supports the view that shareholders initially mispriced stocks in reaction to analysts' decision to drop coverage. After controlling for prior share performance, price-to-book, market capitalization, risk, and trading volume, further evidence suggests that the initial mispricing is more extreme for firms suffering a greater percentage loss in analyst coverage. Abnormal returns in the first 60 trading days of the current year are negatively related to the percentage of coverage loss in the prior year. We conclude that shareholders initially overreact to coverage loss and their overreaction is greater when the coverage loss is greater.

RELATED LITERATURE

Jensen and Meckling (1976) argued that security analysts provide a valuable function by monitoring managers and thereby decreasing the costs of agency conflict between shareholders and managers. Jensen and Meckling (1976) also suggest that analysts cause security prices to trade closer to fundamental values, by reducing information asymmetries between shareholders and managers. Some more recent researchers, such as Chang, Dasgupta, and Hilary (2006), assume that firms followed by more analysts have a lower level of information asymmetry, although Chag, et al., acknowledge that analysts may simply be attracted to more transparent firms.

Beginning at least with Bhushan (1989), researchers began examining the economic incentives for analysts to cover firms. Given that brokerage firm resources are limited, and not all firms can be covered, analysts must decide which firms to cover. Over time researchers became more focused on analysts' incentives to provide coverage, and perhaps optimistically biased coverage, for those firms more likely to generate investment banking fees and trading fees. For example, Cheng, Liu, and Qian (2006) discuss the incentives that (sell-side) analysts have to issue overly optimistic research, because this serves the interests of their firms' underwriting and trading business. Chung and Cho (2005) find that analysts are more likely to provide coverage for firms that are handled by their affiliated market makers. Cliff and Denis (2004) find that firms conducting IPOs compensate their lead underwriting firms for providing analyst coverage by underpricing their IPOs. Hong and Kubik (2003) find evidence that brokerage firms reward overly optimistic analysts who endorse stocks. Bradley, Jordan and Ritter (2003) find that analysts initiate coverage for about three fourths of IPOs at the expiration of the quiet period and that the initial ratings are almost always favorable. Barth, Kasznik, and McNichols (2001) find that analyst coverage is significantly greater for firms with higher trading volume and equity issuance, i.e., sources of income for brokers and underwriters. Barth, et al., conclude that analysts weigh the private benefits and the private costs to their own firms when deciding which stocks to cover.

Other researchers have emphasized analysts' incentives to selectively cover firms that they view favorably and to drop coverage of firms that they view unfavorably. McNichols and O'Brien (1997) find evidence that analysts are more likely to drop coverage of a firm when they have private, negative information about the firm. Specifically, they find that analysts' ratings changes are mostly unfavorable immediately prior to dropping coverage. McNichols and O'Brien also provide evidence that analysts' earnings forecast errors are more negative for stocks that they recently dropped than for those firms that analysts continue to cover. This finding suggests that analysts often prefer to discontinue coverage, rather than revise their earnings forecasts downward or issue sell recommendations. Das, Guo, and Zhang (2006) also support the idea that analysts provide coverage for firms that they view favorably.

Another strand of the analyst literature focuses on the effect analyst coverage has on stock values and some researchers even challenge the notion that greater analyst coverage forces security prices towards their fundamental values. Merton (1987) shows that firm value is a positive function of investors' awareness of the firm. To the extent that analysts increase awareness of a firm by providing coverage, analyst coverage can increase share values. After controlling for various factors, Chung and Jo (1996) find that Tobin's q is positively related to the number of analysts covering the firm. Of course, an increase in share value driven by analyst coverage does not necessarily mean that analyst coverage moves share prices closer to their fundamental values. Jegadeesh, Kim, Krische, and Lee (2004) find that sell-side analysts disproportionately recommend expensive stocks. They report that, among stocks with unfavorable characteristics (regarding momentum, growth, volume, and valuation), stocks recommended by analysts experience lower subsequent returns. Jensen (2004) suggests that excessive analyst coverage can cause stock prices to trade above fundamental values and that this leads to agency costs of overvalued stock. Finally, Doukas, Kim, and Pantzalis (2005) argue that excessively high analyst coverage (caused by investment banking and brokerage trading interests) drives stock prices above fundamental values, because analysts cause investors to be overly optimistic about such firms. Doukas, et al., find that stocks with weak analyst coverage trade below their fundamentally values.

DATA AND METHODOLOGY

Our primary research objective is to test the hypothesis that a severe loss of analyst coverage will cause a firm's stock to trade below its fundamental value. Analysts may drop coverage of a firm because the firm is no longer a good prospect for generating future investment banking or brokerage income. Alternatively, analysts may drop coverage because they become pessimistic about the firm's future share performance. Investors generally must infer the reason for dropped coverage. If investors typically emphasize the latter explanation when they initially interpret the coverage drop decision, they may overreact by selling shares and driving stock prices to below fundamental values. We test this hypothesis by examining abnormal share returns in the first 60 trading days of the calendar year following the year of dropped coverage. We would interpret positive abnormal share performance following the year of lost coverage as evidence that dropped coverage is associated with undervaluation.

Using I/B/E/S data covering the years 1988-2002, we gather a sample of firms experiencing greater than a 50% decrease in analyst coverage during a single calendar year. Analyst coverage is defined as the number of analysts providing at least one annual earnings forecast for the firm during the year. We require that a firm be included in the I/B/E/S database both in the year of lost coverage and in the prior year. That is, we do not assume that a firm has lost 100% of its coverage if it appears in the database one year and fails to appear in the database the next year. This ensures that I/B/E/S is reporting each sample firm's data for both years, but it also effectively excludes firms that lose all analyst coverage. So that abnormal share returns can be calculated, firms included in the final sample must be included in the Center for Research in Security Prices (CRSP) database. Our final sample includes 1249 firm years for which we have sufficient data to calculate abnormal returns. For additional tests, including regressions of abnormal returns on firm-specific variables, we require that firms are included in the Compustat database. Thus, the sample size varies and is reduced in some tests because of Compustat data limitations. In summary, all data concerning analyst coverage are drawn from I/B/E/S, all data used to calculate abnormal returns are drawn from CRSP, and all other firm-specific data are drawn from Compustat.

We calculate abnormal share performance over a 60 trading day period using the Fama-French (1993) three-factor model with the momentum factor adjustment recommended by Carhart (1997). The estimation period is the 255 trading days ending 46 trading days before the first trading day of the year immediately following the year of severe change in analyst coverage. Daily abnormal returns are cumulated over the first 60 trading days in the year following the change in coverage year.

RESULTS

Descriptive statistics and share returns for the sample of coverage losing firms appear in Table 1. Because we draw a sample of firms that experience an extreme (greater than 50%) loss in analyst coverage, this selection requirement results in a sample of mostly small cap firms with relatively modest initial analyst coverage. The mean (median) market value of equity for sampled firms at the end of the year of lost coverage is $1.4 billion ($92 million). The mean (median) number of analysts covering sampled firms in the year prior to coverage loss is six (five). The mean and median percentage decrease in analyst coverage is about 67%.

To gain some perspective on the overall share performance of firms suffering extreme coverage losses we report in Table 1 the raw returns and the market-adjusted returns calculated the year before, the year of, and the year following coverage losses. The market-adjusted return for an individual firm is calculated as the sample firm's total annual return minus the total return on a small stock index for the same year. Annual returns for the small stock index are obtained from Kenneth French's web site at Dartmouth University. In particular, we use the returns on the smallest quintile of U.S. firms. The mean raw return in the year before coverage loss is--12.88%. The mean market-adjusted return in the year before coverage loss is -27.34%. Both of these returns are significant at the 0.01 level and they suggest that analysts often drop firms that have performed poorly in the prior year. The results concerning annual returns in the year of coverage loss are less conclusive. The mean raw return in the year of coverage loss is 9.16% (p = 0.054), however the mean market-adjusted return in the year of coverage loss is -4.08% (p = 0.386). Finally, the mean returns for the year following coverage loss are strongly positive. The raw return in the year following coverage loss is 33.57% and the market-adjusted return in the year following coverage loss is 14.59%. Both of these results are significant at the 0.01 level.

Although the positive mean annual returns following the year of coverage loss could suggest that firms suffering coverage losses were oversold and undervalued at the end of the lost coverage year, these calculations do not well control for the effects of firm size, risk, price-book, or momentum. To directly test whether abnormal share returns are positive in the year following extreme coverage loss, we use the Fama-French (1993) three-factor model with the momentum adjustment mentioned earlier. Using this model with the sample of 1249 extreme decreases in analyst coverage, we find a mean cumulative abnormal return of 11.64% (significant at the 0.01 level) calculated over the first 60 trading days following the year of coverage loss. The most likely explanation for this positive abnormal return is that, during a year in which firms suffer a severe loss in analyst coverage, their stocks are heavily sold and become undervalued. If investors initially fear that analysts drop coverage because of private, negative information relating to the firm's future share performance, then investors would rationally choose to sell their shares before the "bad news" becomes publicly revealed. As the feared bad news often fails to materialize over time, because many analysts drop coverage for other reasons, share prices should return to their fundamental values, thus producing positive abnormal returns, on average.

To further investigate the influence of dropped analyst coverage on share prices and future returns, we estimate several models by regressing the 60-day abnormal returns on the degree of coverage loss. A finding that abnormal returns are greater following more severe coverage losses would support the hypothesis that dropped analyst coverage causes investors to sell shares until stock prices fall below fundamental values. In these regressions we include several control variables so that we can isolate the effects of the lost coverage. Specifically, we regress abnormal returns on the following explanatory variables: Coverage loss; Market-adjusted return; Volatility; Market cap; Price-book; and Volume. Coverage loss is the percentage change in analyst coverage. Market-adjusted return is the firm's total annual stock return minus the return on a small stock index. Volatility is the standard deviation of the monthly stock returns. Market cap is the total market value of equity. Price-book is the firm's stock price divided by the book value of equity per share. Volume is the number of shares of the firm's stock traded. Coverage loss, Market-adjusted return, Volatility, and Volume are calculated for the coverage loss year. Market cap and Price-book are calculated at the end of the coverage loss year.

A negative relation between Coverage loss and abnormal returns would indicate that abnormal returns are higher when the prior year's coverage losses are more severe. Therefore, a negative sign on Coverage loss suggests a positive relationship between dropped coverage and undervaluation in the coverage loss year. We include Market-adjusted return as a control variable, because stocks performing poorly in the prior calendar year may experience a turnaround earlier in the current calendar year for several reasons suggested in the literature, such as tax-loss selling effects. Of course it is also true that, if investors overreact more severely to coverage losses, returns in the coverage loss year will be lower and subsequent price recover may be greater. We include Volatility as a control variable for several reasons including: (a) riskier stocks are likely to produce higher returns; (b) evidence suggests that analysts prefer to cover riskier stocks; and (c) investors may value analyst coverage more for volatile stocks and thus react more severely to a loss in coverage for these firms. Bhushan (1989) argues that investor demand for analyst coverage will be greater for more volatile stocks, because the potential gains and losses from firm-specific information is greater for these stocks. We include Market cap as a control variable because small firms: (a) typically produce greater returns; (b) may be more susceptible to calendar year effects; and (c) may be subject to greater information asymmetries so that investors react more severely to losses in analyst coverage for these firms. Investors may value analyst coverage more highly for high Price-book firms because these firms generally have greater growth opportunities that are more difficult to value absent analyst coverage. We include Price-book as a control variable for this reason and also because prior evidence suggests analysts are more likely to cover high Price-book firms (see, for example, Jegadeesh, et al., 2004). Finally, we include Volume as a control variable because prior evidence suggests analysts prefer to cover high volume stocks (see, for example, Barth, et al., 2001, and Jegadeesh, et al., 2004) and because price reactions to coverage losses may be greater for more thinly traded stocks.

The regression results appear in Table 2. Using simple ordinary least squares regression, multiple tests of the null hypothesis of homoskedasticity are rejected at the 0.001 level. Although our results are extremely similar using simple OLS, and none of our major conclusions change depending on the method used, for brevity we only report regression results using White's (1980) heteroskedasticiy-consistent standard errors.

Model 1 of Table 2 shows that when Coverage loss is the only explanatory variable considered, it is negatively related to abnormal returns (p = 0.004). Models 2 through 5 show that, as the control variables are introduced to the specifications, Coverage loss is consistently, negatively related to abnormal returns at a significance level of 0.025 or better. Thus, after controlling for the effects of prior stock performance, stock return volatility, firm size, price-book ratio, and trading volume, the more severe the loss in analyst coverage during a particular year, the greater are the abnormal returns in the early months of the following year. The most plausible interpretation of this finding is that shareholders overreact to news of lost analyst coverage and they drive stock prices to below their fundamental values. Additional evidence suggests that Market cap and Market-adjusted return are negatively related to abnormal returns, whereas Price-book is positively related to abnormal returns. These findings indicate that smaller firms, firms that suffered the worst relative performance in the prior year, and firms with higher price-book ratios tend to perform better in the early months following the coverage loss year. We conduct a number of tests (not shown) to check on the robustness of our regression results concerning Coverage loss. Using the specification shown as Model 5 in Table 2 as a base model, we tried several alternatives to the firm performance variable Market-adjusted return. Specifically, we substituted, one variable at a time: (a) the raw stock return from the coverage loss year; (b) the market-adjusted return calculated for the coverage loss year using returns on the Wilshire 5000 index as the benchmark index; and (3) the accounting return on assets calculated for the coverage loss year. Results using these alternative firm performance measures in the coverage loss year are very similar. In each case the performance variable is significantly, negatively related to abnormal returns. More importantly, in each case Coverage loss is significantly, negatively related to abnormal returns at the p = 0.019 level or better. As an additional robustness check, we altered the Model 5 specification so that the dependent variable is the market-adjusted annual return for the year following the coverage loss. When an annual return is substituted for a 60-day return, obviously many events unrelated to analyst coverage loss intercede to affect the dependent variable. As expected, the model's R-squared and the significance levels of explanatory variables deteriorate dramatically. Nevertheless, the coefficient on Coverage loss remains negative and is significant at the p = 0.069 level.

SUMMARY AND CONCLUSIONS

This study investigates the share price effects of extreme losses in security analyst coverage. Using a sample of firms that lose more than 50% of their analyst coverage in a single calendar year, we find that abnormal returns in the early months of the subsequent year are strongly positive. The mean abnormal return calculated over the first 60 trading days following the year of coverage loss is 11.6%. Furthermore, the returns in the year following coverage loss are negatively related to the percentage change in analyst coverage during the year of coverage loss. These results are obtained after controlling for the effects of firm size, price-book, prior share performance, risk, and trading volume.

The most plausible interpretation of this evidence is that investors respond to extreme losses in analyst coverage by selling shares in the coverage loss year and driving stock prices to below their fundamental values. As stock prices recover and move closer to their fundamental values, shares of coverage losing firms experience positive abnormal returns. Analysts' bias in favor of covering stocks that they can recommend is well-documented in the finance literature and has been widely reported in the financial press. Therefore, investors are likely to view dropped coverage as an indication that analysts have private, negative information regarding the firm's future prospects. Under this scenario, investors would rationally choose to sell their shares at the time of dropped coverage, before the feared "bad news" becomes publicly revealed. However, analysts have other incentives to add or drop coverage of firms, which also have been documented in the literature. These incentives relate to analysts' desire to generate brokerage and investment banking fees for their own firms. If analysts drop coverage of firms because of brokerage and investment banking concerns, rather than because of private information about the firm's prospects, then investors would overreact by selling their shares during periods of severe coverage loss.

Although we conclude that investor overreaction to extreme losses in analyst coverage is the best explanation for our findings, we cannot completely rule out an alternative interpretation. After firms suffer a loss in analyst coverage, the problems of information asymmetry between managers and investors are likely to become more severe. Therefore, it is possible that the relationship we observe between coverage loss and abnormal returns shortly following coverage loss is evidence of a permanent asymmetric information risk premium. We note this alternative explanation for completeness, but we surmise that the magnitude of the abnormal returns is better explained by an initial shareholder overreaction to coverage loss resulting in temporary mispricing.

REFERENCES

Barth, M., R. Kasznik, & M. McNichols (2001). Analyst Coverage and Intangible Assets. Journal of Accounting Research, 39, 1-34.

Bhushan, R. (1989). Firm Characteristics and Analyst Following. Journal of Accounting and Economics, 11, 255-274.

Bradley, D., B. Jordan, & J. Ritter (2003). The Quite Period Goes Out with a Bang. Journal of Finance, 58, 1-36.

Carhart, M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance, 52, 57-82.

Chang, X., S. Dasgupta, & G. Hilary (2006). Analyst Coverage and Financing Decisions. Journal of Finance, 61, 3009-3048.

Cheng, Y., M. Liu, & J. Qian (2006). Buy-Side Analysts, Sell-Side Analysts, and Investment Decisions of Money Managers. Journal of Financial and Quantitative Analysis, 41, 51-83.

Chung, K. & H. Jo (1996). The Impact of Security Analysts' Monitoring and Marketing Functions on the Market Value of Firms. Journal of Financial and Quantitative Analysis, 31, 493-512.

Chung, K. & S. Cho (2005). Security Analysis and Market Making. Journal of Financial Intermediation, 14, 114-141.

Cliff, M. & D. Denis (2004). Do Initial Public Offering Firms Purchase Analyst Coverage with Underpricing? Journal of Finance, 59, 2871-2901.

Das, S., R. Guo, & H. Zhang (2006). Analysts' Selective Coverage and Subsequent Performance of Newly Public Firms. Journal of Finance, 61, 1159-1185.

Doukas, J., C. Kim, & C. Pantzalis (2005). The Two Faces of Analyst Coverage. Financial Management, 34, 99-125.

Fama, E. & K. French (1993). Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics, 33, 3-56.

Feng, G. (2005). Innovation, Future Earnings, and Market Efficiency. Journal of Accounting, Auditing and Finance, 20, 385-418.

Gilson, S. (2000). Analysts and Information Gaps: Lessons from the UAL Buyout. Financial Analysts Journal, 56, 82-110.

Gleason, C. & Lee, C. (2003). Analyst Forecast Revisions and Market Price Discovery. Accounting Review, 78, 193-225.

Hong, H. & J. Kubik (2003). Analyzing the Analysts: Career Concerns and Biased Earnings Forecasts. Journal of Finance, 58, 313-351.

Hong, H., T. Lim, & J. Stein (2000). Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies. Journal of Finance, 55, 265-295.

Jegadeesh, N., J. Kim, S. Krische, & C. Lee (2004). Analyzing the Analysts: When Do Recommendations Add Value? Journal of Finance, 59, 1083-1124.

Jensen, M. (2004). The Agency Costs of Overvalued Equity and the Current State of Corporate Finance. European Financial Management, 4, 549-565.

Jensen, M. & W. Meckling (1976). Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure. Journal of Financial Economics, 3, 305-360.

Lin, H. & M. McNichols (1998). Underwriting Relationships, Analysts' Earnings Forecasts and Investment Recommendations. Journal of Accounting and Economics, 25, 101-127.

Louis, H. (2004). Earnings Management and the Market Performance of Acquiring Firms. Journal of Financial Economics, 74, 121-148.

McNichols, M. & P. O'Brien (1997). Self-selection and Analyst Coverage. Journal of Accounting Research, 35, 167-199.

Merton, R. (1987). A Simple Model of Capital Market Equilibrium with Incomplete Information. Journal of Finance, 42, 483-510.

Michaely, R. & K. Womack (1999). Conflict of Interest and the Credibility of Underwriter Analyst Recommendations. Review of Financial Studies, 12, 653-686.

Rajan, R. & H. Servaes (1997). Analyst Following of Initial Public Offerings. Journal of Finance, 52, 507-529.

Shane, P. & T. Stock (2006). Security Analyst and Stock Market Efficiency in Anticipating Tax-Motivated Income Shifting. Accounting Review, 81, 227-250.

White, H. (1980). A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica, 48, 817-838.

Rich Fortin, New Mexico State University

Greg Roth, New Mexico State University

Table 1: Descriptive Statistics and Share Returns
For Firms Suffering Severe Losses in Analyst Coverage

Variable N Mean Median

Market cap (in $millions) 1249 1401.2 92.13
Analyst Coveraget-1 1249 6.02 5
Analyst Coveraget 1249 2.03 1
Coverage loss 1249 -0.671 -0.667
Annual Returnt-1 1126 -0.129 -0.177
Annual Returnt 1206 0.092 * -0.06
Annual Returnt+1 1124 0.336 *** -0.07
Market-Adjusted Annual Returnt-1 1127 -0.273 -0.311
Market-Adjusted Annual Returnt 1206 -0.041 -0.175
Market-Adjusted Annual Returnt+1 1124 0.146 *** -0.078
Abnormal Return 1249 0.116 *** 0.038

Variable Standard Min Max
 Deviation

Market cap (in $millions) 7548.68 1.32 155440.1
Analyst Coveraget-1 4.05 3 41
Analyst Coveraget 1.73 1 16
Coverage loss 0.079 -0.938 -0.524
Annual Returnt-1 0.561 -0.991 4.941
Annual Returnt 1.647 -0.992 47.932
Annual Returnt+1 1.35 -0.998 23.929
Market-Adjusted Annual Returnt-1 0.55 -1.456 4.54
Market-Adjusted Annual Returnt 1.634 -1.366 47.841
Market-Adjusted Annual Returnt+1 1.287 -1.741 23.183
Abnormal Return 0.423 -1.736 3.597

Shown are descriptive statistics for firms suffering severe losses
in security analyst coverage.

Each firm was selected from the I/B/E/S database and experienced
greater than a 50% loss in analyst coverage during a single
calendar year.

The sample period includes analyst coverage losses from 1988-2002.
Market cap is the total market value of firm equity at the end
of the year of severe coverage loss.

Analyst Coveraget-1 is the number of analysts covering the firm
before the year of coverage loss. Analyst Coveraget is the number
of analysts covering the firm at the end of the coverage loss year.

Coverage loss is the percentage change in the number of analysts
covering the firm's stock during the year of coverage loss.

Annual Returnt-1, Annual Returnt, and Annual Returnt+1, refer to the
raw share returns in the calendar year before, during, and after the
coverage loss, respectively.

Market-Adjusted Annual Returnt-1, Market-Adjusted Annual Returnt,
and Market-Adjusted Annual Returnt+1, refer to the market-adjusted
share returns in the calendar year before, during, and after the
coverage loss, respectively.

Abnormal Return is the cumulative mean abnormal return calculated
for the first 60 trading days following the year of coverage loss.

For the various mean return measures, ***, **, and *, indicates
statistical significance at the 1%, 5%, and 10% level, respectively.

Table 2: Regressions of Abnormal Returns Following Severe Losses
in Analyst Coverage

 (1) (2) (3)

Intercept -0.172 -0.156 -0.138
 (0.072) (0.091) (0.137)
Coverage loss -0.428 -0.358 -0.339
 (0.004) (0.011) (0.016)
Market-adjusted return -0.001 -0.001
 (0.000) (0.000)
Volatility 0.001 0.001
 (0.201) (0.230)
Market cap -2.8e-06
 (0.000)
Price-book

Volume

[R.sup.2] 0.007 0.075 0.078
N 1249 1097 1097

 (4) (5)

Intercept -0.134 -0.134
 (0.143) (0.152)
Coverage loss -0.317 -0.316
 (0.022) (0.025)
Market-adjusted return -0.002 -0.002
 (0.000) (0.000)
Volatility 0.000 0.000
 -0.866 -0.865
Market cap -3.4e-06 -3.4e-06
 (0.000) (0.000)
Price-book 0.008 0.008
 (0.000) (0.000)
Volume -5.4e-06
 (0.969)
[R.sup.2] 0.115 0.115
N 1023 1023

Shown are the results of regressing abnormal returns on
several variables.

The dependent variable is the cumulative abnormal return calculated
for the first 60 trading days following the calendar year in which
the sample firm lost more than 50% of its security analyst coverage.

The sample period includes analyst coverage losses from 1988 to 2002.
Coverage loss is the percentage change in analyst coverage.

Market-adjusted return is the firm's total annual stock return minus
the return on a small stock index. Volatility is the standard
deviation of the monthly stock returns.

Market cap is the total market value of equity. Price-book is the
firm's stock price divided by the book value of equity per share.

Volume is the number of shares of the firm's stock traded.

Coverage loss, Market-adjusted return, Volatility, and Volume are
calculated for the calendar year in which the coverage loss occurred.

Market cap and Price-book are calculated at the end of the coverage
loss year. Coefficient estimates are shown on the top row for each
variable.

P-values are shown in parentheses and are calculated using White's
(1980) corrected standard errors.