# The effects of option introduction on analyst coverage and earnings estimates.

I. Introduction

We examine the informational environment of the market after options are introduced on NYSE/ Amex and Nasdaq stocks between 1983 and 2004, with a primary focus on the introductions since 1990. In particular, we test several aspects of analyst coverage and earnings estimates around stock option listings empirically and compare the effects for more recently-optioned stocks to those optioned earlier. Mayhew & Mihov (2004) show that stocks optioned in the 1990s tended to be of younger, smaller firms, with lower prices and less investor recognition than those optioned earlier. Many more were Nasdaq stocks, with substantial recent increases in volatility and volume, and recent records of impressive operating and earnings growth. A key consideration in our analysis is to recognize that stocks most likely to be optioned are also stocks that should naturally draw more interest from investors and analysts, even in the absence of the option introduction. Thus there is a question of endogeneity: are improvements in the informational environment the result of factors that led to the option introduction, or are they due to the introduction itself? To distinguish between these factors, we design a control sample of stocks that are very similar to the optioned stocks on the basis of pre-optioning turnover and volatility, but which were not selected for option introduction, as a benchmark.

We find that option listing results in the production of important additional information about the firm. Empirically we find larger increases in analyst coverage of Nasdaq than NYSE/Amex stocks following option introduction. Analysts' forecasts of Nasdaq stock earnings are significantly improved in the two to six months after option introduction, while those of NYSE/Amex stocks are not. The change in responsiveness of optioned stock prices to earnings surprises is more significant for optioned than non-optioned stocks, and optioned Nasdaq stocks display additional response to positive earnings surprises.

Finally, revisions of earnings estimates are generally negative after option introduction--a phenomenon optioned stocks have in common with non-optioned stocks in general. The revisions for optioned Nasdaq stocks are less significant than those of optioned NYSE/Amex stocks. One possible explanation is that even after informational benefits following option introduction are realized, the prospects and outlooks of Nasdaq stocks remain more difficult to analyze than of NYSE/Amex stocks.

Our evidence indicates that option introduction continues to have significant effects on the underlying stocks and firms. It also supports the hypothesis that option effects are still very significant for Nasdaq stocks, which are on average smaller, have lower stock prices, receive less analyst attention, and have more uncertain prospects. Moreover, these effects still take place even in markets that can be characterized as much more developed than earlier in the 1970s when trading on organized options exchanges first took place. Finally, it suggests that the potential role of put and call options in terms of enhancing the informational environment is not limited to the earliest option introductions, and that even the most recent options have significant beneficial effects on the markets.

The remainder of the paper is organized as follows: We review related literature and develop our hypotheses in Section II. Section III describes the data, while Section IV presents our empirical tests and results. Section V concludes.

II. Literature Review and Development of Hypotheses

1. Effects of Option Introduction and Trading

Much theoretical analysis of contingent claims, especially options, examines their ability to improve market completeness. An incomplete market is one in which there are not enough independent, "primitive" securities to "span" the state space, or to allow investors to select securities that provide claims to wealth in all desired states. Ross (1976) argues that options written on existing financial assets can enhance market efficiency by increasing the number of states, or outcomes, that are spanned, or covered, by claims on their underlying securities. (i,ii) In their general equilibrium model of an incomplete market, Detemple and Selden (1991) show that the valuations of primary and derivative securities are determined jointly rather than independently of each other.

The role of options in helping complete financial markets stands in contrast to the view that they are redundant securities, as assumed in option pricing models like Black/Scholes. If options are primarily redundant, then their introduction and subsequent trading should have little or no effect on the underlying stocks. A very large body of empirical evidence establishes that option introduction is associated with significant effects on return and volatility of the underlying stocks. One of the first is Conrad (1989), who finds a significant increase in prices of the underlying stocks after option listings in her sample of call option introductions in the years 1974-1980. She develops the "additional information" hypothesis, and suggests that option introduction brings greater exposure and interest in the stock by financial analysts, (iii) Detemple & Jorion (1990) conclude that the major effects of option introduction on stock returns were in the earliest introductions, those in the 1970s and 1980s. The inference is that these earlier options contributed in a more significant degree to improved market quality than later ones.

Closely related is research on how option trading affects a stock's information environment. To the extent that options are cheaper and easier to trade than the underlying stock, one might expect to see informed traders turning to options. Manaster & Rendleman (1982) hypothesize that because options provide a preferred outlet for informed investors, they may affect the manner in which stock prices adjust to the release of information. Easley, O'Hara & Srinivas (1998) analyze the linkages of price, volume, and information between the stock and options markets. They argue that option volume often has information content for future stock price movements, and find that "positive news" and "negative news" option volumes have significant predictive power. (iv)

Several studies focus on the quality of the information environment as related to various measures of analyst coverage. Based on a sample of 88 stocks that were optioned in the years 1973-1986, Skinner (1990) finds an increase in the average number of analysts covering a stock from 10.14 to 16.70 in the two years before and after the option listing. Similarly, Damodaran & Lim (1991) find significant increases in both the numbers of analysts who follow the stock and media coverage as measured by the number of articles that mention the stock published in The Wall Street Journal after option introduction. Jennings & Starks (1986) find that prices of optioned stocks adjust more quickly to earnings announcements than prices of non-optioned stocks. (v) Ho, Hassel, & Swidler (1995) examine a sample of 371 option introductions in the years 1977-1988 and find that analyst forecasts are more accurate afterward.

Improved analyst coverage and accuracy have long been associated with informational benefits. Merton (1987) models a capital market with incomplete information, and shows analytically how expected stock returns vary with the degree of investor awareness. His view is supported in Arbel & Strebel's (1982) hypothesis of the neglected firm effect. Moyer, Chatfield, & Sisneros (1989) find that analyst monitoring helps make security markets more informationally efficient, (vi) These studies support the hypotheses of increased interest in the underlying stocks after option introduction, and an improved information environment for them.

Finally, Hong, Lim, & Stein (2000) test whether "momentum reflects the gradual diffusion of firm-specific information." They account for the positive relation between firm size and analyst coverage, and find that profits from a momentum strategy are largest for stocks with the least coverage. They find that the "marginal importance of analyst coverage is greatest among small stocks." (vii)

These empirical studies suggest that while option introduction is generally followed by greater analyst coverage and accuracy, the impact should tend to be larger for the smaller, less well-known stocks with strong growth prospects. Formally, we state these (positive) hypotheses:

H1: Increase in analyst coverage

More analysts cover stocks after they are optioned, and analyst coverage increases more for the smaller Nasdaq stocks than the larger ones on the NYSE/Amex.

H2: Improved accuracy of analysts' forecasts

Analyst forecasts become more accurate after stocks are optioned, and the improvement is larger for the more recently-optioned Nasdaq stocks.

2. Reaction to Earnings Surprises

Skinner (1990) finds that price reactions to earnings announcements are smaller for optioned than non-optioned stocks. His results are confirmed by Ho (1993), who examines the volatility of earnings surprises for a sample of 255 optioned stocks and another 176 non-optioned stocks in the years 1980-1983. She finds that residual variance is lower for the optioned than the non-optioned stocks, after controlling for firm size, trading volume, the number of analysts, and the number of institutional investors. She also finds that prices of optioned stocks anticipate earnings growth earlier than of non-optioned stocks.

Mendenhall & Fehrs (1999) extend the analysis to cover option introductions through the year 1993. They find that the responses are time-varying, and are more likely to be positive in the later years of their sample period. They suggest that the larger positive responses are consistent with an improved information environment in which the response of optioned stocks to earnings announcements is faster and more concentrated around the announcement than otherwise. Their results are somewhat different from those of Skinner and Ho in that they find announcement period returns higher--rather than lower--for optioned stocks. However, their essential conclusion is consistent with the hypothesis that options improve the information environment. (viii)

Dimitrov, Jain, & Tice (2006) examine the magnitude of stock excess returns around earnings announcements in relation to six measures of information uncertainty. Two of these measures are the number of analysts with quarterly EPS forecasts two days before the announcement date, and the dispersion of analysts' forecasts. They find that information uncertainty is significantly negatively related to the magnitude of announcement period excess returns, and that the effect of uncertainty is particularly important for firms that are not covered by analysts.

This leads to our hypothesis:

H3: Greater stock price responsiveness to earnings announcements

Stock price responsiveness to earnings surprises will be larger, and adjustments will be completed in a shorter frame of time for all optioned stocks. The effects are larger for Nasdaq stocks.

3. Over-optimism and Revisions of Earnings Estimates

Analysts' tendency to display a consistent upward bias in earnings estimates reflects optimism. For example, Ali, Klein, & Rosenfeld (1992) find that analysts tend to set optimistic estimates of current year, as well as of longer-term, EPS figures.

Dugar & Nathan (1995) hypothesize that analysts from firms that offer investment banking services to a particular firm tend to make more optimistic earnings forecasts of that firm than other analysts. Consistent with this, Ackert & Athanassakos (2003) show that analyst coverage is simultaneously determined with institutional ownership and forecast bias. McNichols & O'Brien (1997) find evidence consistent with the hypothesis that analysts exercise a degree of self selection, in that they are more likely to cover stocks for which they have favorable expectations.

This suggests that we should expect to see overoptimism for the stocks in our optioned sample, and downward earnings revisions during the fiscal year. The effects should tend to be more significant for stocks with the most optimism. In this case, it is not clear a priori whether, or how, the revisions might be different for Nasdaq stocks. If their estimates are revised downward by larger amounts, it is consistent with more optimism for them, coupled with more rapid incorporation of new negative information. If their estimates are not revised downward as much, then other factors predominate. This leads us to our fourth hypothesis:

H4: Revisions of earnings estimates

Forecast revisions are made more quickly after a stock is optioned. Revisions for Nasdaq stocks could be more significant if increased information and coverage leads to greater consensus about their prospects, or less significant if there is still considerable uncertainty surrounding their prospects.

In summary, these studies on analyst coverage and estimates generally conclude that options improve the information environment of the underlying stocks. However, some of the empirical findings appear quite different, possibly due to the varying samples and time periods covered. By focusing on the differences between Nasdaq and NYSE/Amex stocks, one of our objectives is to identify one possible source of these differences.

III. The Data, Sample, and Control Stocks

We obtain the initial listing dates for all options traded on NYSE/Amex and Nasdaq between January 1983 and December 2004 from the Chicago Board Options Exchange. We exclude subsequent option listings of the same stock on another exchange. We also exclude ADRs because of the difficulty of finding a matching control stock.

Daily stock return and trading volume data are from the Center for Research in Security Prices database. Average earnings per share (EPS) forecasts, the number of analysts that issue earnings forecasts, and actual quarterly earnings are from the Institutional Brokers Estimate Service (IBES). Data on the market-to-book ratios are obtained from Compustat.

Our initial sample of optioned stocks includes 827 NYSE/Amex and 1,527 Nasdaq stocks. Descriptive data are presented in Table 1, which shows the number of option introductions by year. There are clear differences between the NYSE/ Amex and Nasdaq stocks. The median market capitalization of the Nasdaq stocks is $353 million, roughly one-half of that of NYSE/Amex stocks, $704 million (See Table 1, last row). Similarly, the median price of the optioned Nasdaq stocks is lower than of the NYSE/Amex stocks: $21.88 versus $28.25. The Nasdaq stocks are more volatile, as shown by the standard deviation of returns in the pre-introduction period: 3.72% versus 2.20% for NYSE/Amex stocks. They have less analyst coverage, and have higher growth prospects as measured by the market-to-book ratio. (ix)

We form a control sample of non-optioned stocks to account for general increases in analyst coverage, trading volume, and stock volatility over time. To qualify for the control sample, a stock must not be optioned, and must remain unoptioned for at least two years after the option listing date. It must have the same 2-digit SIC code as its underlying optioned stock. (x) The control stocks are then selected on the basis of stock volatility and turnover, defined as trading volume divided by the number of shares outstanding. These criteria reflect the fact that, according to Mayhew and Mihov (2004), volatility and turnover are key factors considered by options exchanges in listing new options. The control stock must have a 250-trading day average turnover between 80% and 120% of those of the optioned stock. After locating up to five control-stock candidates, we then select the one whose standard deviation of daily returns in the same 250-trading-day period is closest to that of the optioned firm as that stock's control stock. Of the 827 NYSE/Amex stocks in our basic sample, we find control stocks for 580 of them (70%). Of the 1,527 Nasdaq stocks in the initial sample, we find control stocks for 1,137, or 74%, of them.

Table 2 shows the mean and median values of turnover and return volatility for the optioned and control samples. The control stocks are very closely matched to their underlying optioned stocks. The mean overall turnover of the optioned stock sample is 0.66, very close to that for the control stocks of 0.61. The volatility of the optioned stocks has a mean value of 3.56, very close to that of 3.74 for the control stocks. The respective medians for volume and volatility are also very similar.

IV. Empirical Tests and Results

1. Degree of Analyst Coverage

For each optioned stock, the numbers of analysts covering the stock for two years before and after the option-listing month (excluding the month of the option listing) are obtained from IBES. We include analysts who submit EPS forecasts of five types: for the current and next fiscal years, the current and next fiscal quarters, and the growth rate over the next 3 to 5 years. For each type, we report the number of analysts submitting forecasts in the periods of 6, 12, and 24 months before and after option introduction. (xi,xii)

The results are summarized in Table 3. In the case of EPS forecasts of NYSE/Amex stocks for the next fiscal year, 8.4 analysts on average submitted estimates in the 6 months before optioning. After option introduction, there were 8.8 such analysts. (See Table 3, Panel A.) The average increase is 0.4 analysts, or about 5%. All of the mean changes for the other forecasts are also positive (Panels B, C, D, and E).

To test whether the increase in the number of analysts following optioned stocks is significant, we compare each firm's change to that of its control stock over the same period. The mean control-stock-adjusted results of analysts following the NYSE/Amex stocks around option introduction are presented in Table 3, Panel A, Column (5). In this case, the mean control-adjusted change is 0.4 analysts, the same as for the uncontrolled change. Importantly, all of the changes shown in this column are positive.

We hypothesize that the increase is larger for the typically smaller Nasdaq stocks because of their lower coverage before the option introduction. Stocks that are already widely known and traded--those with substantial investor recognition--may gain some analyst coverage after option introduction, but those that are the least known and traded potentially have the most to gain by option introduction.

These expectations are strongly supported in the sample data. For an example, refer again to Table 3, Panel A. For Nasdaq stocks, the average number of analysts submitting EPS estimates for the current fiscal year in the six months before optioning is 5.0, compared to 8.4 for NYSE/Amex stocks. The number after optioning is 5.7, an increase of 0.7, or 14%. This compares to the increase of 0.4 (5%) for the NYSE/Amex stocks. The increase for Nasdaq stocks is larger than that of the NYSE/Amex stocks in all cases shown in the table, both in terms of number and percent.

Finally, we test for these effects jointly in a multiple regression of the percentage change in analyst coverage as a function of whether the stock is traded on Nasdaq, and the number of analysts that follow the stock before option introduction. A variable for the market capitalization at time of option introduction is included to control for the size factor, known to be related to the degree of analyst coverage. We also include a dummy variable to divide the time period 1983-2004 into periods before and after 1990 because of the general increase in analyst coverage over time. We estimate this regression equation separately for percentage changes in 6, 12, and 24 month periods before and after option introduction:

[[DELTA]%No_Analysts.sub.j] = [alpha] + [[beta].sub.1] [D_Nasdaq.sub.j] + [[beta].sub.2] [Mkt_Cap.sub.j] + [[beta].sub.3] [D_Post_1990.sub.j] + [[beta].sub.4] [No_nalysts_Pre.sub.j] + [[epsilon].sub.j] (1)

Where:

[[DELTA]%No_Analysts.sub.j] = Percentage change in the number of analysts following stock j in the 6, 12, or 24 months around option introduction.

[D_Nasdaq.sub.j] = Dummy variable for Nasdaq stocks. Value = 1 if the stock is traded in Nasdaq, and 0 otherwise.

[Mkt_Cap.sub.j] = Market capitalization of stock j.

[D_Post_1990.sub.j] = Dummy variable for the time period. Value = 1 if the option introduction date is alter 1990, and 0 otherwise.

[No_Analysts_Pre.sub.j] = Number of analysts in 6, 12, or 24 months before option introduction for stock j.

[[epsilon].sub.j] = Disturbance term, assumed to have distribution N(0, [[sigma].sup.2]).

The regression results are summarized in Table 4. All of the variables are significant and have the expected signs. The increase in analyst coverage 6 months after option introduction for Nasdaq stocks is 6.2306 percentage points, consistent with our expectation that these stocks have more to gain. This coefficient is also positive and significant in the regressions for 12 and 24 months around option introduction. The coefficient on the number of analysts following the stock before option introduction is negative, consistent with the expectation that stocks with less analyst attention potentially have the most to gain. The coefficient on market value is positive, consistent with the general tendency for analysts to cover large stocks. The coefficient on the post-1990 dummy variable is negative. As analyst coverage increased in the 1990s, the potential percentage increase due to option introduction is accordingly smaller.

We conclude that this empirical evidence is strongly consistent with our expectations in hypothesis H1.

2. Accuracy of Earnings Forecasts

If trading in options brings more information to the market, then we would expect analysts' earnings forecasts to improve after option introduction. Accordingly, forecast errors should be reduced. We further expect that they should be reduced more for Nasdaq stocks that potentially have more to gain in terms of informational efficiency.

To test this, we collect monthly EPS estimates for the current fiscal year, denoted FY1, from IBES and compare them to actual earnings at the end of each year. Since several monthly estimates are usually submitted before the actual earnings are reported, each monthly estimate's ex post error is standardized by its own value to reduce any size bias. Since over- and under-estimates both represent errors, we measure forecast error (FE) in absolute value terms:

[FE.sub.j,t] = 100 * [absolute value of ([Act_EPS.sub.j,t] - [FY1.sub.j,t])/[FY1.sub.j,t]] (2)

Where:

[Act_EPS.sub.j,t] = actual earnings announced by firm j for fiscal year t

[FY1.sub.j,t] = monthly consensus estimate for firm j's fiscal year t

We standardize each month's error by the control firm's FE to adjust for trends in forecast errors across years. The standardized FE for firm j, denoted [SFE.sub.j], is defined for each month as follows:

[SFE.sub.j] = [FE.sub.j] - [FE.sub.j,s control]

Table 5, Column (3), shows a clear decrease in standardized forecast errors (SFEs) in the months after optioning. All of the errors are negative, and those for the Nasdaq stocks are much larger (in absolute value terms) than those for the NYSE/ Amex stocks.

We also compare each of the 12 monthly SFEs after option listing to its corresponding SFE a year earlier to adjust for seasonality as well as for control stocks. The null hypothesis is that option listings do not help improve analysts' forecast accuracy. Significant declines in the SFEs after option introduction from before would enable us to reject this null hypothesis.

When compared to the SFEs one year earlier (Table 5, Column (4)), the median difference in SFEs remains negative in most cases, but more so for Nasdaq stocks (Panel B) than exchange-listed stocks (Panel A). One of the most interesting observations is that in the fourth month after option introduction, the median SFE for Nasdaq stocks is -4.9, compared to +2.0 for NYSE/Amex stocks. This suggests that the early effects are for large improvements in forecast accuracy for Nasdaq stocks, even while those for NYSE/Amex stocks worsen. Both Wilcoxon (Column (5)) and sign tests (Column (7)) confirm that the improvement in forecast accuracy in the post-option period relative to that in the pre-option period is statistically significant.

We conclude that, as a whole, analysts do a better job after options listings, and their performance improves more for Nasdaq than NYSE/ Amex stocks, especially two to six months after option introduction. This is consistent with our expectations in Hypothesis H2.

3. Reactions to Earnings Surprises upon Announcement

We next investigate whether the market reaction to firms' quarterly earnings announcements changes after option introduction. We analyze the abnormal returns upon announcement, event days (0, +2), as well as the drift in a longer event period, (-9, +30).

We include up to eight quarterly earnings announcements of each optioned stock: four quarterly announcements before and four after the option listing date. Quarterly earnings announcements made within ten days of option introduction are excluded to avoid any possible complications from option listing effects. We include only firms with option introductions in the years 1985-2004 because of limits on the availability of earnings announcement data from IBES.

We define the earnings surprise variable, E_SURP, as the difference between the actual EPS value and the consensus forecast at that time:

[E_SURP.sub.j,n] = 100 * ([Act_EPS.sub.j,n] - [FQ1.sub.j,n])/[absolute value of [FQ1.sub.j,n]] (3)

Where:

[Act_EPS.sub.j,n] = actual earnings announced by firm j, fiscal quarter n

[FQ1.sub.j,n] = consensus earnings estimate for firm j, fiscal quarter n

In order to focus on the most significant surprises, we include only positive earnings surprises of 5% or more, and negative earnings surprises of 5% or more. We form two sub-samples, one of positive earnings surprises, E_Surp_Pos, and the other for negative surprises, E_Surp_Neg.

To test for the effect of earnings surprises on stock returns, we define market-adjusted abnormal returns as follows: (xiii)

[AR.sub.j,t] = [RET.sub.j,t] - [S&P500.sub.t] (4)

Where:

[AR.sub.j,t] = Estimated abnormal return on stock j, day t

[RET.sub.j,t] = Return on stock j, day t

[S&P500.sub.t] = Market return, based on the S&P Index of 500 stocks.

We cumulate ARs in the announcement period window (0, +2) to form the cumulative abnormal returns, CARs.

To test for differences in reaction to earnings surprises due to option listing and exchange location, cumulative abnormal returns in the announcement period are regressed cross-sectionally on the magnitude of earnings surprise (positive or negative separately), along with several dummy variables to control for additional factors. The equation for positive earnings surprises is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

Where:

[CAR.sub.j,t] = Cumulative abnormal return in the 3-day announcement period (0, +2). Stock returns are adjusted for the market, represented by the S&P 500 Index. CARs are defined for stock j, announcement day t.

[E_Surp_Pos.sub.j,t] = Earnings surprise, in percent, for stock j, day t.

[D_Option_Post.sub.j,t] = Dummy variable for stocks in the optioned sample after they are optioned. Value = 1 if options exist on stock j and 0 otherwise. This variable has values of 0 for optioned stocks before their option introduction, and for all control stocks.

[E_Surp_Pos.sub.j,t] * [D_Option_Pos.sub.j,t] = Interaction term

[D_Nasdaq.sub.j,t] = Dummy variable for Nasdaq stocks.

Value = 1 if the stock is traded in Nasdaq, and 0 otherwise.

[D_Post_1990.sub.j.t] = Dummy variable for time period.

Value = 1 if the earnings announcement date is after 1990, and 0 otherwise.

[D_Option_Stock.sub.j,t] = Dummy variable for stocks in the optioned sample, whether before or after they are optioned. Value = 1 if the stock is in the optioned sample, and 0 if the stock is in the control sample.

[D_Control_Post.sub.j,t] = Dummy variable for control stocks in the post-introduction period. This variable has a value of 1 if the announcement date falls after the control's underlying optioned stock has been optioned, and 0 otherwise.

[[epsilon].sub.j,t] = Disturbance term, assumed to have distribution N(0, [[sigma].sup.2] ).

The equation for negative earnings surprises is the same except that the negative earnings surprises variable, E_Surp_Neg, is used instead of the positive one E_Surp_Pos.

Table 6 reports the results of the regression model of the three-day cumulative excess returns against positive (Panel A) and negative (Panel B) quarterly earnings surprises. We estimate two versions of Equation (5). The first includes only the optioned stocks, and the second also includes the control stocks, for a robustness test.

We focus first on the positive earnings surprise variable, E_Surp_Pos, whose coefficient is sometimes called the "earnings response coefficient." This estimate is positive and significant, as expected (Panel A). The intercept dummy variable for optioned stocks after option introduction, D_Option_Post, has a negative coefficient of -2.3826, also statistically significant. The coefficient of the interaction variable, the product of the earnings surprise and the post-optioning dummy variables, has a value of 0.9983, also significant.

Thus, we conclude that the earnings response coefficient is higher for optioned than non-optioned stocks. This is consistent with our hypothesis H3 that the response of optioned stocks to earnings news is stronger than otherwise. (xiv)

The coefficient on the Nasdaq dummy variable, D_Nasdaq, is positive, 0.4439, with a significant 0.0408 level of confidence. This is consistent with our hypothesis that the response of optioned Nasdaq stocks is larger than for optioned NYSE/Amex stocks. The dummy variable for announcements after 1990 is also significantly positive, indicating a change in responsiveness over time not clearly attributable to option introduction.

For robustness, we test an additional version of Equation (5) which includes the control stocks as well as the optioned stocks. This equation includes the optioned stock dummy variable (D_Option_Stock). The results of this regression are shown in the right-hand columns of Table 6. Interestingly, the coefficient estimates on the three option introduction variables, E_Surp_Pos, D_Option_Post, and the interaction between them, all have nearly the same values and levels of significance as when the control stocks are excluded. Also notable is that the general fit of the relation, as measured by adjusted [R.sup.2] and F-Ratio, is much lower. The coefficient on the Nasdaq dummy variable is not significant in this regression.

The coefficient on the optioned stock dummy variable, D_Option_Stock, is positive and significant, suggesting that while the control stocks are well matched on the basis of trading volume and volatility, there are, as one would expect, other differences between them. The positive value is consistent with the hypothesis that the earnings announcement conveys more information for non-optioned stocks. We also include a dummy variable that assigns a value of 1 to control stocks if the announcement lies in the post-option period of their underlying stocks. As expected, this variable, D_Control_Post, is not statistically significant.

The results for negative earnings surprises are reported in Table 6, Panel B. Most of the results for negative earnings surprises are similar to the positive ones. The option intercept dummy variable, D_Option_Post, has a significant positive value. Since in this sample, the earnings surprises and CARs are mostly negative, a positive value means that the CARs are lower in absolute value terms (closer to zero). Again, the coefficient on the interaction variable is significant and positive, indicating that the earnings response coefficient is larger after stocks are optioned than before. In this case, the coefficient on the Nasdaq dummy variable has the expected sign, but is not significant. Thus we conclude that when it comes to negative information, option introduction matters more than trading location.

4. Reactions to Earnings Surprises in a Larger Event Window

Next we consider a broader event period, from nine days before the announcement to 30 days afterward, event period (-9, +30). For each market, we again separate out positive and negative earnings surprises. The average daily cumulative excess returns are shown graphically in Chart 1 for the optioned sample and Chart 2 for the control sample.

The CARs of the optioned stocks in Chart 1 show a very distinctive difference before and after optioning. The positive and negative earnings surprises before option introduction both show marked upward drifts in the 30 days after the earnings announcements. Since these announcements are in the year before options are introduced, these positive drifts may reflect unusually good performance of these stocks before they are selected for optioning. After they are optioned, this post-announcement tendency for continued positive abnormal returns disappears. The initial response to the earnings announcements is then completed in the event period (-1, +2). This is consistent with an efficient market.

[GRAPHIC 1 OMITTED]

[GRAPHIC 2 OMITTED]

The CARs in Chart 2 represent the effects of earnings announcements for the control stocks. These are largely as expected: the effects are apparent in the days immediately around the announcement, and there is no significant drift after day +2.

These differences are further detailed in Table 7, which shows the CARs for the whole window, (-9, +30), for the optioned vs. control stocks (Panel A), and within the optioned sample, for Nasdaq vs. NYSE/Amex stocks (Panel B). The data in Panel A confirm the conclusions illustrated in Charts 1 and 2. Before option introduction, there was large upward drift in excess returns after earnings announcements, but that drift disappears afterward. The data in Panel B show that the upward drift of Nasdaq stocks was larger than for NYSE/ Amex stocks before option introduction, but nearly the same after. This provides additional evidence that option introduction has larger effects for Nasdaq than NYSE/Amex stocks.

5. Revisions of Earnings Estimates

Any evidence of downward revisions implies that analysts have previously over-estimated the target firm's future earnings. Indeed, analyst earnings forecasts tend to be over-optimistic, and downward revisions are the norm.

In this section, we test whether option listings have an influence on earnings revisions. How might this happen? As we find above, on average, more analysts follow stocks after they are optioned, and their forecast accuracy is improved. More generally, trading in options should enhance the information environment in which the stock is traded. If the revisions are made more quickly, or if the basic level of over-optimism is reduced, then this can be interpreted as evidence of a positive informational effect of options listing.

We focus on analysts' forecasts of current fiscal year's EPS (FY1) both because there is a longer historical record of data and because more analysts issue this estimate. Our measure of estimate revision is the difference between two consecutive monthly estimates, scaled either by the corresponding stock price or the absolute value of previous month's estimate:

[ER_P.sub.j,t] = 100 * ([FY1.sub.j,t] - [FY1.sub.j,t-1])/[Price.sub.j,t-1] (6a)

[ER_FY.sub.j,t] = 100 * ([FY1.sub.j,t] - [FY1.sub.j,t-1])/[absolute value of [FY1.sub.j,t-1]] (6b)

Where:

[ER_P.sub.j,t] = Estimate revision as percent of base month stock price

[ER_FY.sub.j.t] = Estimate revision as percentage change in forecast value

[FY1.sub.j,t] = Consensus estimate in month t of firm j's earnings for the current fiscal year, FY1

[FY1.sub.j,t] = Consensus estimate in month t-1 of firm j's earnings for the current fiscal year, FY1

[Price.sub.j.t-1] = Stock price in prior month t-1

Cumulative three-, six-, and twelve-month revisions of current fiscal year earnings estimates after option introduction are compared to the corresponding revisions a year earlier to control for possible seasonal factors. (xv) Each optioned firm's difference in estimate revision before and after option introduction is also adjusted by the control stock's difference during the same period.

The median estimate revisions adjusted for seasonal trends are shown in Table 8, Column (3), and the percentage of newly-optioned firms with positive adjusted estimate revisions are in Column (4). The median revisions adjusted for control stocks and the percentage of positive changes are presented in Columns (5) and (7), respectively.

The results for NYSE/Amex stocks are in Panel A, Table 8. The revisions based on both Equations (6a) and (6b) are overwhelmingly negative (see Columns (3) and (5)). This reflects the fact that the percent of positive observations in Columns (4) and (7) in Table 8 are less than 50%. This indicates that analysts as a whole lower their FYI estimates after options listings. The statistical significance is confirmed by the low p-values in Columns (6) and (8), for 6- and 12-month periods. The results for Nasdaq stocks are in Panel B, Table 8. While these revisions are also negative, they are less significant than for the NYSE/Amex stocks.

Analysts generally are overly optimistic at the beginning of the fiscal year and systematically revise their forecasts downward as the year proceeds. In our sample, EPS estimates of both groups of stocks are revised downwards, but the NYSE/ Amex revisions are more significant. Above we find that more new analysts initiate coverage of Nasdaq stocks after option introduction than of NYSE/Amex stocks. Since Nasdaq's listing requirements are less restrictive than those on the NYSE, it is plausible that the earnings outlook for most Nasdaq stocks are, and remain, more uncertain than those of the larger NYSE/Amex stocks, even after option introduction. It also may be that the new analysts tend to issue slightly less optimistic estimates to avoid possible legal repercussions of overly-optimistic estimates.

V. Conclusions

This study examines the informational environment of the market after options are introduced on NYSE/Amex and Nasdaq stocks between 1983 and 2004. We compare the optioned sample to a similar non-optioned control sample. We find that option listing results in the production of important additional information about the firm. The forecast accuracy of analyst earnings improves significantly and the improvement appears to be more significant for Nasdaq than for NYSE/Amex stocks.

Empirically we find larger increases in analyst coverage of Nasdaq stocks following option introduction. Analysts' forecasts of Nasdaq stock earnings are significantly improved in the two to six months after option introduction, while those of NYSE/Amex stocks are not, all relative to similar non-optioned control stocks.

The change in responsiveness of optioned stock prices to earnings surprises is more significant for optioned than non-optioned stocks, and optioned Nasdaq stocks display additional response to positive earnings surprises.

Finally, revisions of earnings estimates are generally negative after option introduction--a phenomenon optioned stocks have in common with non-optioned stocks in general. The revisions for optioned Nasdaq stocks are less significant than those of optioned NYSE/Amex stocks. One possible explanation is that even after informational benefits following option introduction are realized, the prospects and outlooks of Nasdaq stocks remain more difficult to analyze than of NYSE/Amex stocks.

We conclude that options enhance the informational environment of all stocks on which they are traded, and that they matter more for smaller, lower-priced stocks with less analyst attention before they were optioned. Our findings extend the literature on the informational benefits of option introduction to show that they are not confined to, or even largely concentrated on, the larger, better-known NYSE/ Amex stocks that were the first to be optioned.

Our findings extend previous research on market completeness in a critical way because they show a new type of evidence about the beneficial effects of options in terms of enhancing market completeness. The fact that most of the large, heavily-traded, and followed stocks have already been optioned suggests that the financial markets--as a whole as well as for those particular stocks--are now significantly more developed than before. Our evidence suggests that the benefits gained to date are not exhaustive, and that option introductions on smaller, less well-known stocks, continue making additional contributions to the market environment.

References

Ackert, Lucy F., and George Athanassakos. 2003 "A simultaneous equations analysis of analysts' forecast bias, analyst following, and institutional ownership," .Journal of Business Finance & Accounting, 30(7) & (8), 1017-1041.

Ali, Ashiq, April Klein, and James Rosenfeld. 1992 "Analysts' use of information about permanent and transitory earnings components in forecasting annual EPS," The Accounting Review, 67(1), 183-198.

Arbel, Avner, and Paul Strebel. 1982 "The neglected and small firm effects," The Financial Review, 17(4), 201-218.

Arditti, Fred D., and Kose John. 1980 "Spanning the State Space with Options," Journal of Financial and Quantitative Analysis, 15, 1-9.

Bansal, Vipul K., Pruitt, Stephen W., and Wei, K. C. John. 1989 "An empirical reexamination of the impact of CBOE option initiation on the volatility and trading volume of the underlying equities: 1973-1986." Financial Review, 24(1), 19-30.

Barth, Mary E., Ron Kasznik, and Maureen F. McNichols. 2001 "Analyst coverage and intangible assets," Journal of Accounting Research, 39(1), 1-34.

Bollen, Nicolas P. B. 1998 "A note on the impact of options on stock return volatility." Journal of Banking and Finance, 22(9) 1181-1191.

Brous, Peter A., 1992 "Common Stock Offerings and Earnings Expectations: A Test of the Release of Unfavorable Information," The Journal of Finance, 47(4), 1517-1536.

Chakravarty, Sugato, Huseyin Gulen, and Stewart Mayhew, 2004 "Informed trading in stock and option markets," The Journal of Finance, 59(3), 1235-1257.

Chang, Kee H., and Hoje Jo. 1996 "The impact of security analysts' monitoring and marketing functions on the market value of firms," Journal of Financial and Quantitative Analysis, 31 (4), 493-512.

Conrad, Jennifer. 1989 "The price effect of option introduction," The Journal of Finance, 44(2), 487-498.

Damodaran, Aswath, and Joseph Lira. 1991 "The effect of options listing on the underlying stocks' return processes," Journal of Banking and Finance, 15(3), 647-664.

Detemple, Jerome, and Philippe Jorion. 1990 "Option listing and stock returns: an empirical analysis," Journal of Banking and Finance, 14(4), 781-801.

Detemple, Jerome, and Larry Seldon. 1991 "A general equilibrium analysis of option and stock market interactions," International Economic Review, 32(2), 279-303.

Dimitrov, Valentin, Prem Jain, and Sheri Tice. 2006 "Sell on the news: information uncertainty and returns around earnings announcements." Working paper, SSRN-id = 938749.

Dugar, Amitabh, and Siva Nathan. 1995 "The effects of investment banking relationships on financial analysts' earnings forecasts and investment recommendations," Contemporary Accounting Research, 12(1), 131-160.

Easley, David, Maureen O'Hara, and P. S. Srinivas. 1998 "Option volume and stock prices: Evidence on where informed traders trade," The Journal of Finance, 53(2), 431-465.

Faff, Robert, and David Hilliel. 2005 "Complete markets, informed trading and equity option introductions," Journal of Banking and Finance, 29(6), 1359-1384.

Figlewski, Stephen, and Gwendolyn Webb. 1993 "Options, short sales, and market completeness," The Journal of Finance, 48(2), 761-777.

Freund, Steven, P. McCann, and Gwendolyn Webb. 1994 "A regression analysis of the effects of option introduction on stock variances." The Journal of Derivatives, 1 (3), 25-38.

Ho, Li-Chin Jennifer. 1993 "Option trading and the relation between price and earnings: a crosssectional analysis." The Accounting Review, 68(2), 368-384.

Ho, Li-Chin Jennifer, John M. Hassel, and Steve Swidler. 1995 "An empirical examination of the dispersion and accuracy of analyst forecasts surrounding option listing," Review of Financial Economics, 4(2), 171-185.

Hong, Harrison, Terrence Lim, and Jeremy C. Stein. 2000 "Bad news travels slowly: size, analyst coverage, and the profitability of momentum strategies," The Journal of Finance, 55(1), 265-295.

Jennings, Robert H., and Laura T. Starks. 1986 "Earnings announcements, stock price adjustments, and the existence of option markets," The Journal of Finance, 41(1), 107-125.

Jubinski, Daniel, and Marc Tomljanovich. 2007 "Options listings and individual equity volatility." The Journal of Futures Markets, 27(1), 1-27.

Jubinski, Daniel, and Marc Tomljanovich. 2003 "An examination of the introduction of options on the volatility of the underlying individual equity securities." Working paper, SSRN-id = 479841.

Kecskes, Ambrus, and Kent L. Womack. 2007 "Adds and drops of analyst coverage: does the stock market overreact?" Working paper, January 28, SSRN-id = 960501.

Kim, Sunyoung, and Jason Lee. 2006 "Effect of option listing on price reactions to earnings announcements," Working paper, SSRN-id = 874857.

Kumar, Raman, Atulya Sarin, and Kuldeep Shastri. 1998 "The impact of options trading on the market quality of the underlying security: an empirical analysis," The Journal of Finance, 53(2), 717-732.

Manaster, Steven, and Richard J. Jr. Rendleman. 1982 "Option prices as predictors of equilibrium stock prices," The Journal of Finance, 37(4), 1043-1057.

Mayhew, Stewart. 2002 "Competition, market structure, and bid-ask spreads in stock option markets" The Journal of Finance, 57(2), 931-958.

Mayhew, Stewart, and Vassil Mihov. 2004 "How do exchanges select stocks for option listing?" The Journal of Finance, 59(1), 447-471.

Mazouz, Khelifa, 2004 "The effect of CBOE option listing on the volatility of NYSE traded stocks: a time-varying variance approach." The Journal of Empirical Finance, 11(5), 695-708.

McNichols, Maureen, and Patricia C. O'Brien. 1997 "Self-selection and analyst coverage," Journal of Accounting Research, 35(Supplement), 167-199.

Mendenhall, Richard R., and Donald H. Fehrs. 1999 "Option listing and the stock-price response to earnings announcements," Journal of Accounting and Economics, 27, 57-87.

Merton, Robert. 1987 "A simple model of capital market equilibrium with incomplete information," Journal of Finance, 42(3), 483-510.

Moyer, R. Charles, Robert E. Chatfield, and Phillip M. Sisneros. 1989 "Security analyst monitoring activity: agency costs and information demands," Journal of Financial and Quantitative Analysis, 24(4), 503-512.

O'Brien, Patricia C., and Ravi Bhushan. 1990 "Analyst following and institutional ownership," Journal of Accounting Research, 28(Supplement), 55-76.

Roll, Richard, Eduardo Schwartz, and Avanidhar Subrahmanyam. 2007 "Options trading activity and firm valuation," Working paper, SSRNid = 1087112.

Ross, Stephen A. 1976 "Options and Efficiency," Quarterly Journal of Economics, 90, 75-89.

Skinner, Douglas J. 1989 "Options markets and stock return volatility," Journal of Financial Economics, 23(1), 61-78.

Skinner, Douglas J. 1990 "Options markets and the information content of accounting earnings releases," Journal of Accounting and Economic's, 13(3), 191-211.

Sorescu, Sorin. M. 2000 "The effect of options on stock prices: 1973-1995." The Journal of Finance, 55(1), 487-514.

St. Pierre, Ellen F., 1998 "The impact of option introduction on the conditional return distribution of underlying securities." The Financial Review, 33(1), 105-119.

Notes

(i.) More specifically, Ross (1976) models a market in which there are n primitive securities and m possible states. The number of possible states, m, is higher than the number of primitive securities, n. This market is incomplete in the sense that the available supply of n securities is not sufficient to allow investors to design an investment portfolio that has desired coverage across all possible outcome states. Technically, the rank of the state-space tableau is not full. Ross shows how options can be written on the underlying primitive securities so that investors can achieve their desired coverage across the states. In his model, the market is complete when options on the primitive securities augment the rank of the state-space tableau to the point that it is full.

(ii.) Although Ross's analysis does not fully cover cases when portfolio returns are required to be non-negative, Arditti and John (1980) resolve this problem by identifying conditions under which the full state space can be spanned by put options.

(iii.) Conrad also finds that the excess return volatility declines after introduction and that systematic risk is unchanged. Similarly, Skinner (1989), Damodaran & Lim (1991), and Bansal, Pruitt, & Wei (1989) also find decreases in volatility after introduction. Bollen (1998) examines a larger sample and compares post-introduction volatility to that of a control sample and concludes that options introduction has no effect on volatility. This is consistent with Freund, McCann, & Webb (1993), as well as in studies that test for changes in conditional volatility, including Jubinsky & Tomljanovich (2003, 2007), St. Pierre (1998), and Mazouz (2004).

(iv.) Additional aspects of options on the informational environment are addressed by Roll, Schwartz, and Subrahmanyam (2007), Faff and Hilliel (2005), Chakravarty, Gulen, & Mayhew (2004), and Kumar, Sarin & Shastri (1998), among others.

(v.) A different aspect of the informational environment is examined by Figlewski & Webb (1993), who hypothesize that options offer new, less restrictive ways for investors to act on negative information, and find evidence consistent with their hypothesis. This implies that options help reduce the amount of information that was previously excluded from the marketplace. They find evidence consistent with their hypothesis. This is significant because it implies that options help reduce the amount of information that was previously excluded from the marketplace. Their findings are supported by Sorescu (2000).

(vi.) Additional supporting findings in those of Chung and Jo (1996), who find that analysts help increase the extent of investor awareness of covered firms. O'Brien & Bhushan (1990) model the simultaneous determination of analyst coverage and institutional ownership. They find that analyst coverage increases more in firms (1) with smaller prior analyst following; (2) whose return volatility has declined; and (3) with firms in growing industries.

(vii.) Barth, Kasznik, & McNichols (2001) show that firms with significant intangible assets tend to have larger analyst following. They measure intangible assets by research and development (R&D) and advertising expenses, and by depreciation and recognized intangible assets, including goodwill. Firms with high R&D expenditures are often considered to be firms with substantial growth prospects. Additionally, Kecskes & Womack (2007) find that analysts tend to add coverage of stocks that are in greater demand by investors and with better operating performance. They tend to drop firms in less demand and with poorer performance.

(viii.) Kim & Lee (2006) examine reactions to earnings announcements of optioned stocks in 1985-2003. They find that the reactions of optioned stocks are more intense, and that their post-earnings announcement drift is smaller. They compare optioned firms to all other firms, and therefore do not form a well-defined control sample. Overall, their findings are consistent with those of Mendenhall & Fehrs.

(ix.) The trading volumes of the Nasdaq and NYSE/Amex groups cannot be compared directly because the trading volume data are generated differently in the two markets.

(x.) Control stocks are also required to have coverage in the Institutional Brokers Estimates System (IBES) at least six months before the option listing date of their underlying sample stock.

(xi.) We exclude the number of analysts in the option listing month, in part because IBES defines its statistical periods from midmonth to mid-month rather than on the basis of month-ends.

(xii.) As an example, consider Microsoft. Options were first listed on Microsoft stock on March 12, 1987. The average number of analysts issuing current fiscal year EPS estimates for Microsoft increased from 12.33 six months before the option listing date (9/86 to 2/87) to 21.67 six months afterward (4/87 to 9/87), resulting in an increase of 9.34 analysts. In percentage terms, this is an increase of 76%. Our control stock for Microsoft is IMS International Inc. During the six months before Microsoft was optioned, the average number of analysts following IMS was 11.33. The number increased to 11.83 during the six-month period after Microsoft was optioned, for an increase of just 0.50 analysts. The control-stock-adjusted change in the average number of analysts following Microsoft is thus 8.84 (= 9.34-0.50).

(xiii.) Because we define abnormal returns based on a market adjustment, we do not employ a separate estimation window outside of our event window.

(xiv.) The combination of a lower intercept and a higher slope means that the linear relation between news and stock price response has rotated counter-clockwise in space defined with earnings surprises on the x-axis, and stock abnormal returns on the y-axis.

(xv.) Brous (1992) shows that the monthly forecast revisions are serially correlated because not all analysts update their forecasts on a monthly basis.

Susana Yu, Corresponding Author, Montclair State University, yus@mail.montclair.edu, (917) 834-5238

Kishore Tandon, Baruch College/The City University of New York, Kishore.Tandon@baruch.cuny.edu, (646) 312-3450

Gwendolyn Webb, Baruch College/The City University of New York, Gwendolyn.Webb@baruch.cuny.edu, (646) 312-3485

We examine the informational environment of the market after options are introduced on NYSE/ Amex and Nasdaq stocks between 1983 and 2004, with a primary focus on the introductions since 1990. In particular, we test several aspects of analyst coverage and earnings estimates around stock option listings empirically and compare the effects for more recently-optioned stocks to those optioned earlier. Mayhew & Mihov (2004) show that stocks optioned in the 1990s tended to be of younger, smaller firms, with lower prices and less investor recognition than those optioned earlier. Many more were Nasdaq stocks, with substantial recent increases in volatility and volume, and recent records of impressive operating and earnings growth. A key consideration in our analysis is to recognize that stocks most likely to be optioned are also stocks that should naturally draw more interest from investors and analysts, even in the absence of the option introduction. Thus there is a question of endogeneity: are improvements in the informational environment the result of factors that led to the option introduction, or are they due to the introduction itself? To distinguish between these factors, we design a control sample of stocks that are very similar to the optioned stocks on the basis of pre-optioning turnover and volatility, but which were not selected for option introduction, as a benchmark.

We find that option listing results in the production of important additional information about the firm. Empirically we find larger increases in analyst coverage of Nasdaq than NYSE/Amex stocks following option introduction. Analysts' forecasts of Nasdaq stock earnings are significantly improved in the two to six months after option introduction, while those of NYSE/Amex stocks are not. The change in responsiveness of optioned stock prices to earnings surprises is more significant for optioned than non-optioned stocks, and optioned Nasdaq stocks display additional response to positive earnings surprises.

Finally, revisions of earnings estimates are generally negative after option introduction--a phenomenon optioned stocks have in common with non-optioned stocks in general. The revisions for optioned Nasdaq stocks are less significant than those of optioned NYSE/Amex stocks. One possible explanation is that even after informational benefits following option introduction are realized, the prospects and outlooks of Nasdaq stocks remain more difficult to analyze than of NYSE/Amex stocks.

Our evidence indicates that option introduction continues to have significant effects on the underlying stocks and firms. It also supports the hypothesis that option effects are still very significant for Nasdaq stocks, which are on average smaller, have lower stock prices, receive less analyst attention, and have more uncertain prospects. Moreover, these effects still take place even in markets that can be characterized as much more developed than earlier in the 1970s when trading on organized options exchanges first took place. Finally, it suggests that the potential role of put and call options in terms of enhancing the informational environment is not limited to the earliest option introductions, and that even the most recent options have significant beneficial effects on the markets.

The remainder of the paper is organized as follows: We review related literature and develop our hypotheses in Section II. Section III describes the data, while Section IV presents our empirical tests and results. Section V concludes.

II. Literature Review and Development of Hypotheses

1. Effects of Option Introduction and Trading

Much theoretical analysis of contingent claims, especially options, examines their ability to improve market completeness. An incomplete market is one in which there are not enough independent, "primitive" securities to "span" the state space, or to allow investors to select securities that provide claims to wealth in all desired states. Ross (1976) argues that options written on existing financial assets can enhance market efficiency by increasing the number of states, or outcomes, that are spanned, or covered, by claims on their underlying securities. (i,ii) In their general equilibrium model of an incomplete market, Detemple and Selden (1991) show that the valuations of primary and derivative securities are determined jointly rather than independently of each other.

The role of options in helping complete financial markets stands in contrast to the view that they are redundant securities, as assumed in option pricing models like Black/Scholes. If options are primarily redundant, then their introduction and subsequent trading should have little or no effect on the underlying stocks. A very large body of empirical evidence establishes that option introduction is associated with significant effects on return and volatility of the underlying stocks. One of the first is Conrad (1989), who finds a significant increase in prices of the underlying stocks after option listings in her sample of call option introductions in the years 1974-1980. She develops the "additional information" hypothesis, and suggests that option introduction brings greater exposure and interest in the stock by financial analysts, (iii) Detemple & Jorion (1990) conclude that the major effects of option introduction on stock returns were in the earliest introductions, those in the 1970s and 1980s. The inference is that these earlier options contributed in a more significant degree to improved market quality than later ones.

Closely related is research on how option trading affects a stock's information environment. To the extent that options are cheaper and easier to trade than the underlying stock, one might expect to see informed traders turning to options. Manaster & Rendleman (1982) hypothesize that because options provide a preferred outlet for informed investors, they may affect the manner in which stock prices adjust to the release of information. Easley, O'Hara & Srinivas (1998) analyze the linkages of price, volume, and information between the stock and options markets. They argue that option volume often has information content for future stock price movements, and find that "positive news" and "negative news" option volumes have significant predictive power. (iv)

Several studies focus on the quality of the information environment as related to various measures of analyst coverage. Based on a sample of 88 stocks that were optioned in the years 1973-1986, Skinner (1990) finds an increase in the average number of analysts covering a stock from 10.14 to 16.70 in the two years before and after the option listing. Similarly, Damodaran & Lim (1991) find significant increases in both the numbers of analysts who follow the stock and media coverage as measured by the number of articles that mention the stock published in The Wall Street Journal after option introduction. Jennings & Starks (1986) find that prices of optioned stocks adjust more quickly to earnings announcements than prices of non-optioned stocks. (v) Ho, Hassel, & Swidler (1995) examine a sample of 371 option introductions in the years 1977-1988 and find that analyst forecasts are more accurate afterward.

Improved analyst coverage and accuracy have long been associated with informational benefits. Merton (1987) models a capital market with incomplete information, and shows analytically how expected stock returns vary with the degree of investor awareness. His view is supported in Arbel & Strebel's (1982) hypothesis of the neglected firm effect. Moyer, Chatfield, & Sisneros (1989) find that analyst monitoring helps make security markets more informationally efficient, (vi) These studies support the hypotheses of increased interest in the underlying stocks after option introduction, and an improved information environment for them.

Finally, Hong, Lim, & Stein (2000) test whether "momentum reflects the gradual diffusion of firm-specific information." They account for the positive relation between firm size and analyst coverage, and find that profits from a momentum strategy are largest for stocks with the least coverage. They find that the "marginal importance of analyst coverage is greatest among small stocks." (vii)

These empirical studies suggest that while option introduction is generally followed by greater analyst coverage and accuracy, the impact should tend to be larger for the smaller, less well-known stocks with strong growth prospects. Formally, we state these (positive) hypotheses:

H1: Increase in analyst coverage

More analysts cover stocks after they are optioned, and analyst coverage increases more for the smaller Nasdaq stocks than the larger ones on the NYSE/Amex.

H2: Improved accuracy of analysts' forecasts

Analyst forecasts become more accurate after stocks are optioned, and the improvement is larger for the more recently-optioned Nasdaq stocks.

2. Reaction to Earnings Surprises

Skinner (1990) finds that price reactions to earnings announcements are smaller for optioned than non-optioned stocks. His results are confirmed by Ho (1993), who examines the volatility of earnings surprises for a sample of 255 optioned stocks and another 176 non-optioned stocks in the years 1980-1983. She finds that residual variance is lower for the optioned than the non-optioned stocks, after controlling for firm size, trading volume, the number of analysts, and the number of institutional investors. She also finds that prices of optioned stocks anticipate earnings growth earlier than of non-optioned stocks.

Mendenhall & Fehrs (1999) extend the analysis to cover option introductions through the year 1993. They find that the responses are time-varying, and are more likely to be positive in the later years of their sample period. They suggest that the larger positive responses are consistent with an improved information environment in which the response of optioned stocks to earnings announcements is faster and more concentrated around the announcement than otherwise. Their results are somewhat different from those of Skinner and Ho in that they find announcement period returns higher--rather than lower--for optioned stocks. However, their essential conclusion is consistent with the hypothesis that options improve the information environment. (viii)

Dimitrov, Jain, & Tice (2006) examine the magnitude of stock excess returns around earnings announcements in relation to six measures of information uncertainty. Two of these measures are the number of analysts with quarterly EPS forecasts two days before the announcement date, and the dispersion of analysts' forecasts. They find that information uncertainty is significantly negatively related to the magnitude of announcement period excess returns, and that the effect of uncertainty is particularly important for firms that are not covered by analysts.

This leads to our hypothesis:

H3: Greater stock price responsiveness to earnings announcements

Stock price responsiveness to earnings surprises will be larger, and adjustments will be completed in a shorter frame of time for all optioned stocks. The effects are larger for Nasdaq stocks.

3. Over-optimism and Revisions of Earnings Estimates

Analysts' tendency to display a consistent upward bias in earnings estimates reflects optimism. For example, Ali, Klein, & Rosenfeld (1992) find that analysts tend to set optimistic estimates of current year, as well as of longer-term, EPS figures.

Dugar & Nathan (1995) hypothesize that analysts from firms that offer investment banking services to a particular firm tend to make more optimistic earnings forecasts of that firm than other analysts. Consistent with this, Ackert & Athanassakos (2003) show that analyst coverage is simultaneously determined with institutional ownership and forecast bias. McNichols & O'Brien (1997) find evidence consistent with the hypothesis that analysts exercise a degree of self selection, in that they are more likely to cover stocks for which they have favorable expectations.

This suggests that we should expect to see overoptimism for the stocks in our optioned sample, and downward earnings revisions during the fiscal year. The effects should tend to be more significant for stocks with the most optimism. In this case, it is not clear a priori whether, or how, the revisions might be different for Nasdaq stocks. If their estimates are revised downward by larger amounts, it is consistent with more optimism for them, coupled with more rapid incorporation of new negative information. If their estimates are not revised downward as much, then other factors predominate. This leads us to our fourth hypothesis:

H4: Revisions of earnings estimates

Forecast revisions are made more quickly after a stock is optioned. Revisions for Nasdaq stocks could be more significant if increased information and coverage leads to greater consensus about their prospects, or less significant if there is still considerable uncertainty surrounding their prospects.

In summary, these studies on analyst coverage and estimates generally conclude that options improve the information environment of the underlying stocks. However, some of the empirical findings appear quite different, possibly due to the varying samples and time periods covered. By focusing on the differences between Nasdaq and NYSE/Amex stocks, one of our objectives is to identify one possible source of these differences.

III. The Data, Sample, and Control Stocks

We obtain the initial listing dates for all options traded on NYSE/Amex and Nasdaq between January 1983 and December 2004 from the Chicago Board Options Exchange. We exclude subsequent option listings of the same stock on another exchange. We also exclude ADRs because of the difficulty of finding a matching control stock.

Daily stock return and trading volume data are from the Center for Research in Security Prices database. Average earnings per share (EPS) forecasts, the number of analysts that issue earnings forecasts, and actual quarterly earnings are from the Institutional Brokers Estimate Service (IBES). Data on the market-to-book ratios are obtained from Compustat.

Our initial sample of optioned stocks includes 827 NYSE/Amex and 1,527 Nasdaq stocks. Descriptive data are presented in Table 1, which shows the number of option introductions by year. There are clear differences between the NYSE/ Amex and Nasdaq stocks. The median market capitalization of the Nasdaq stocks is $353 million, roughly one-half of that of NYSE/Amex stocks, $704 million (See Table 1, last row). Similarly, the median price of the optioned Nasdaq stocks is lower than of the NYSE/Amex stocks: $21.88 versus $28.25. The Nasdaq stocks are more volatile, as shown by the standard deviation of returns in the pre-introduction period: 3.72% versus 2.20% for NYSE/Amex stocks. They have less analyst coverage, and have higher growth prospects as measured by the market-to-book ratio. (ix)

We form a control sample of non-optioned stocks to account for general increases in analyst coverage, trading volume, and stock volatility over time. To qualify for the control sample, a stock must not be optioned, and must remain unoptioned for at least two years after the option listing date. It must have the same 2-digit SIC code as its underlying optioned stock. (x) The control stocks are then selected on the basis of stock volatility and turnover, defined as trading volume divided by the number of shares outstanding. These criteria reflect the fact that, according to Mayhew and Mihov (2004), volatility and turnover are key factors considered by options exchanges in listing new options. The control stock must have a 250-trading day average turnover between 80% and 120% of those of the optioned stock. After locating up to five control-stock candidates, we then select the one whose standard deviation of daily returns in the same 250-trading-day period is closest to that of the optioned firm as that stock's control stock. Of the 827 NYSE/Amex stocks in our basic sample, we find control stocks for 580 of them (70%). Of the 1,527 Nasdaq stocks in the initial sample, we find control stocks for 1,137, or 74%, of them.

Table 2 shows the mean and median values of turnover and return volatility for the optioned and control samples. The control stocks are very closely matched to their underlying optioned stocks. The mean overall turnover of the optioned stock sample is 0.66, very close to that for the control stocks of 0.61. The volatility of the optioned stocks has a mean value of 3.56, very close to that of 3.74 for the control stocks. The respective medians for volume and volatility are also very similar.

IV. Empirical Tests and Results

1. Degree of Analyst Coverage

For each optioned stock, the numbers of analysts covering the stock for two years before and after the option-listing month (excluding the month of the option listing) are obtained from IBES. We include analysts who submit EPS forecasts of five types: for the current and next fiscal years, the current and next fiscal quarters, and the growth rate over the next 3 to 5 years. For each type, we report the number of analysts submitting forecasts in the periods of 6, 12, and 24 months before and after option introduction. (xi,xii)

The results are summarized in Table 3. In the case of EPS forecasts of NYSE/Amex stocks for the next fiscal year, 8.4 analysts on average submitted estimates in the 6 months before optioning. After option introduction, there were 8.8 such analysts. (See Table 3, Panel A.) The average increase is 0.4 analysts, or about 5%. All of the mean changes for the other forecasts are also positive (Panels B, C, D, and E).

To test whether the increase in the number of analysts following optioned stocks is significant, we compare each firm's change to that of its control stock over the same period. The mean control-stock-adjusted results of analysts following the NYSE/Amex stocks around option introduction are presented in Table 3, Panel A, Column (5). In this case, the mean control-adjusted change is 0.4 analysts, the same as for the uncontrolled change. Importantly, all of the changes shown in this column are positive.

We hypothesize that the increase is larger for the typically smaller Nasdaq stocks because of their lower coverage before the option introduction. Stocks that are already widely known and traded--those with substantial investor recognition--may gain some analyst coverage after option introduction, but those that are the least known and traded potentially have the most to gain by option introduction.

These expectations are strongly supported in the sample data. For an example, refer again to Table 3, Panel A. For Nasdaq stocks, the average number of analysts submitting EPS estimates for the current fiscal year in the six months before optioning is 5.0, compared to 8.4 for NYSE/Amex stocks. The number after optioning is 5.7, an increase of 0.7, or 14%. This compares to the increase of 0.4 (5%) for the NYSE/Amex stocks. The increase for Nasdaq stocks is larger than that of the NYSE/Amex stocks in all cases shown in the table, both in terms of number and percent.

Finally, we test for these effects jointly in a multiple regression of the percentage change in analyst coverage as a function of whether the stock is traded on Nasdaq, and the number of analysts that follow the stock before option introduction. A variable for the market capitalization at time of option introduction is included to control for the size factor, known to be related to the degree of analyst coverage. We also include a dummy variable to divide the time period 1983-2004 into periods before and after 1990 because of the general increase in analyst coverage over time. We estimate this regression equation separately for percentage changes in 6, 12, and 24 month periods before and after option introduction:

[[DELTA]%No_Analysts.sub.j] = [alpha] + [[beta].sub.1] [D_Nasdaq.sub.j] + [[beta].sub.2] [Mkt_Cap.sub.j] + [[beta].sub.3] [D_Post_1990.sub.j] + [[beta].sub.4] [No_nalysts_Pre.sub.j] + [[epsilon].sub.j] (1)

Where:

[[DELTA]%No_Analysts.sub.j] = Percentage change in the number of analysts following stock j in the 6, 12, or 24 months around option introduction.

[D_Nasdaq.sub.j] = Dummy variable for Nasdaq stocks. Value = 1 if the stock is traded in Nasdaq, and 0 otherwise.

[Mkt_Cap.sub.j] = Market capitalization of stock j.

[D_Post_1990.sub.j] = Dummy variable for the time period. Value = 1 if the option introduction date is alter 1990, and 0 otherwise.

[No_Analysts_Pre.sub.j] = Number of analysts in 6, 12, or 24 months before option introduction for stock j.

[[epsilon].sub.j] = Disturbance term, assumed to have distribution N(0, [[sigma].sup.2]).

The regression results are summarized in Table 4. All of the variables are significant and have the expected signs. The increase in analyst coverage 6 months after option introduction for Nasdaq stocks is 6.2306 percentage points, consistent with our expectation that these stocks have more to gain. This coefficient is also positive and significant in the regressions for 12 and 24 months around option introduction. The coefficient on the number of analysts following the stock before option introduction is negative, consistent with the expectation that stocks with less analyst attention potentially have the most to gain. The coefficient on market value is positive, consistent with the general tendency for analysts to cover large stocks. The coefficient on the post-1990 dummy variable is negative. As analyst coverage increased in the 1990s, the potential percentage increase due to option introduction is accordingly smaller.

We conclude that this empirical evidence is strongly consistent with our expectations in hypothesis H1.

2. Accuracy of Earnings Forecasts

If trading in options brings more information to the market, then we would expect analysts' earnings forecasts to improve after option introduction. Accordingly, forecast errors should be reduced. We further expect that they should be reduced more for Nasdaq stocks that potentially have more to gain in terms of informational efficiency.

To test this, we collect monthly EPS estimates for the current fiscal year, denoted FY1, from IBES and compare them to actual earnings at the end of each year. Since several monthly estimates are usually submitted before the actual earnings are reported, each monthly estimate's ex post error is standardized by its own value to reduce any size bias. Since over- and under-estimates both represent errors, we measure forecast error (FE) in absolute value terms:

[FE.sub.j,t] = 100 * [absolute value of ([Act_EPS.sub.j,t] - [FY1.sub.j,t])/[FY1.sub.j,t]] (2)

Where:

[Act_EPS.sub.j,t] = actual earnings announced by firm j for fiscal year t

[FY1.sub.j,t] = monthly consensus estimate for firm j's fiscal year t

We standardize each month's error by the control firm's FE to adjust for trends in forecast errors across years. The standardized FE for firm j, denoted [SFE.sub.j], is defined for each month as follows:

[SFE.sub.j] = [FE.sub.j] - [FE.sub.j,s control]

Table 5, Column (3), shows a clear decrease in standardized forecast errors (SFEs) in the months after optioning. All of the errors are negative, and those for the Nasdaq stocks are much larger (in absolute value terms) than those for the NYSE/ Amex stocks.

We also compare each of the 12 monthly SFEs after option listing to its corresponding SFE a year earlier to adjust for seasonality as well as for control stocks. The null hypothesis is that option listings do not help improve analysts' forecast accuracy. Significant declines in the SFEs after option introduction from before would enable us to reject this null hypothesis.

When compared to the SFEs one year earlier (Table 5, Column (4)), the median difference in SFEs remains negative in most cases, but more so for Nasdaq stocks (Panel B) than exchange-listed stocks (Panel A). One of the most interesting observations is that in the fourth month after option introduction, the median SFE for Nasdaq stocks is -4.9, compared to +2.0 for NYSE/Amex stocks. This suggests that the early effects are for large improvements in forecast accuracy for Nasdaq stocks, even while those for NYSE/Amex stocks worsen. Both Wilcoxon (Column (5)) and sign tests (Column (7)) confirm that the improvement in forecast accuracy in the post-option period relative to that in the pre-option period is statistically significant.

We conclude that, as a whole, analysts do a better job after options listings, and their performance improves more for Nasdaq than NYSE/ Amex stocks, especially two to six months after option introduction. This is consistent with our expectations in Hypothesis H2.

3. Reactions to Earnings Surprises upon Announcement

We next investigate whether the market reaction to firms' quarterly earnings announcements changes after option introduction. We analyze the abnormal returns upon announcement, event days (0, +2), as well as the drift in a longer event period, (-9, +30).

We include up to eight quarterly earnings announcements of each optioned stock: four quarterly announcements before and four after the option listing date. Quarterly earnings announcements made within ten days of option introduction are excluded to avoid any possible complications from option listing effects. We include only firms with option introductions in the years 1985-2004 because of limits on the availability of earnings announcement data from IBES.

We define the earnings surprise variable, E_SURP, as the difference between the actual EPS value and the consensus forecast at that time:

[E_SURP.sub.j,n] = 100 * ([Act_EPS.sub.j,n] - [FQ1.sub.j,n])/[absolute value of [FQ1.sub.j,n]] (3)

Where:

[Act_EPS.sub.j,n] = actual earnings announced by firm j, fiscal quarter n

[FQ1.sub.j,n] = consensus earnings estimate for firm j, fiscal quarter n

In order to focus on the most significant surprises, we include only positive earnings surprises of 5% or more, and negative earnings surprises of 5% or more. We form two sub-samples, one of positive earnings surprises, E_Surp_Pos, and the other for negative surprises, E_Surp_Neg.

To test for the effect of earnings surprises on stock returns, we define market-adjusted abnormal returns as follows: (xiii)

[AR.sub.j,t] = [RET.sub.j,t] - [S&P500.sub.t] (4)

Where:

[AR.sub.j,t] = Estimated abnormal return on stock j, day t

[RET.sub.j,t] = Return on stock j, day t

[S&P500.sub.t] = Market return, based on the S&P Index of 500 stocks.

We cumulate ARs in the announcement period window (0, +2) to form the cumulative abnormal returns, CARs.

To test for differences in reaction to earnings surprises due to option listing and exchange location, cumulative abnormal returns in the announcement period are regressed cross-sectionally on the magnitude of earnings surprise (positive or negative separately), along with several dummy variables to control for additional factors. The equation for positive earnings surprises is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

Where:

[CAR.sub.j,t] = Cumulative abnormal return in the 3-day announcement period (0, +2). Stock returns are adjusted for the market, represented by the S&P 500 Index. CARs are defined for stock j, announcement day t.

[E_Surp_Pos.sub.j,t] = Earnings surprise, in percent, for stock j, day t.

[D_Option_Post.sub.j,t] = Dummy variable for stocks in the optioned sample after they are optioned. Value = 1 if options exist on stock j and 0 otherwise. This variable has values of 0 for optioned stocks before their option introduction, and for all control stocks.

[E_Surp_Pos.sub.j,t] * [D_Option_Pos.sub.j,t] = Interaction term

[D_Nasdaq.sub.j,t] = Dummy variable for Nasdaq stocks.

Value = 1 if the stock is traded in Nasdaq, and 0 otherwise.

[D_Post_1990.sub.j.t] = Dummy variable for time period.

Value = 1 if the earnings announcement date is after 1990, and 0 otherwise.

[D_Option_Stock.sub.j,t] = Dummy variable for stocks in the optioned sample, whether before or after they are optioned. Value = 1 if the stock is in the optioned sample, and 0 if the stock is in the control sample.

[D_Control_Post.sub.j,t] = Dummy variable for control stocks in the post-introduction period. This variable has a value of 1 if the announcement date falls after the control's underlying optioned stock has been optioned, and 0 otherwise.

[[epsilon].sub.j,t] = Disturbance term, assumed to have distribution N(0, [[sigma].sup.2] ).

The equation for negative earnings surprises is the same except that the negative earnings surprises variable, E_Surp_Neg, is used instead of the positive one E_Surp_Pos.

Table 6 reports the results of the regression model of the three-day cumulative excess returns against positive (Panel A) and negative (Panel B) quarterly earnings surprises. We estimate two versions of Equation (5). The first includes only the optioned stocks, and the second also includes the control stocks, for a robustness test.

We focus first on the positive earnings surprise variable, E_Surp_Pos, whose coefficient is sometimes called the "earnings response coefficient." This estimate is positive and significant, as expected (Panel A). The intercept dummy variable for optioned stocks after option introduction, D_Option_Post, has a negative coefficient of -2.3826, also statistically significant. The coefficient of the interaction variable, the product of the earnings surprise and the post-optioning dummy variables, has a value of 0.9983, also significant.

Thus, we conclude that the earnings response coefficient is higher for optioned than non-optioned stocks. This is consistent with our hypothesis H3 that the response of optioned stocks to earnings news is stronger than otherwise. (xiv)

The coefficient on the Nasdaq dummy variable, D_Nasdaq, is positive, 0.4439, with a significant 0.0408 level of confidence. This is consistent with our hypothesis that the response of optioned Nasdaq stocks is larger than for optioned NYSE/Amex stocks. The dummy variable for announcements after 1990 is also significantly positive, indicating a change in responsiveness over time not clearly attributable to option introduction.

For robustness, we test an additional version of Equation (5) which includes the control stocks as well as the optioned stocks. This equation includes the optioned stock dummy variable (D_Option_Stock). The results of this regression are shown in the right-hand columns of Table 6. Interestingly, the coefficient estimates on the three option introduction variables, E_Surp_Pos, D_Option_Post, and the interaction between them, all have nearly the same values and levels of significance as when the control stocks are excluded. Also notable is that the general fit of the relation, as measured by adjusted [R.sup.2] and F-Ratio, is much lower. The coefficient on the Nasdaq dummy variable is not significant in this regression.

The coefficient on the optioned stock dummy variable, D_Option_Stock, is positive and significant, suggesting that while the control stocks are well matched on the basis of trading volume and volatility, there are, as one would expect, other differences between them. The positive value is consistent with the hypothesis that the earnings announcement conveys more information for non-optioned stocks. We also include a dummy variable that assigns a value of 1 to control stocks if the announcement lies in the post-option period of their underlying stocks. As expected, this variable, D_Control_Post, is not statistically significant.

The results for negative earnings surprises are reported in Table 6, Panel B. Most of the results for negative earnings surprises are similar to the positive ones. The option intercept dummy variable, D_Option_Post, has a significant positive value. Since in this sample, the earnings surprises and CARs are mostly negative, a positive value means that the CARs are lower in absolute value terms (closer to zero). Again, the coefficient on the interaction variable is significant and positive, indicating that the earnings response coefficient is larger after stocks are optioned than before. In this case, the coefficient on the Nasdaq dummy variable has the expected sign, but is not significant. Thus we conclude that when it comes to negative information, option introduction matters more than trading location.

4. Reactions to Earnings Surprises in a Larger Event Window

Next we consider a broader event period, from nine days before the announcement to 30 days afterward, event period (-9, +30). For each market, we again separate out positive and negative earnings surprises. The average daily cumulative excess returns are shown graphically in Chart 1 for the optioned sample and Chart 2 for the control sample.

The CARs of the optioned stocks in Chart 1 show a very distinctive difference before and after optioning. The positive and negative earnings surprises before option introduction both show marked upward drifts in the 30 days after the earnings announcements. Since these announcements are in the year before options are introduced, these positive drifts may reflect unusually good performance of these stocks before they are selected for optioning. After they are optioned, this post-announcement tendency for continued positive abnormal returns disappears. The initial response to the earnings announcements is then completed in the event period (-1, +2). This is consistent with an efficient market.

[GRAPHIC 1 OMITTED]

[GRAPHIC 2 OMITTED]

The CARs in Chart 2 represent the effects of earnings announcements for the control stocks. These are largely as expected: the effects are apparent in the days immediately around the announcement, and there is no significant drift after day +2.

These differences are further detailed in Table 7, which shows the CARs for the whole window, (-9, +30), for the optioned vs. control stocks (Panel A), and within the optioned sample, for Nasdaq vs. NYSE/Amex stocks (Panel B). The data in Panel A confirm the conclusions illustrated in Charts 1 and 2. Before option introduction, there was large upward drift in excess returns after earnings announcements, but that drift disappears afterward. The data in Panel B show that the upward drift of Nasdaq stocks was larger than for NYSE/ Amex stocks before option introduction, but nearly the same after. This provides additional evidence that option introduction has larger effects for Nasdaq than NYSE/Amex stocks.

5. Revisions of Earnings Estimates

Any evidence of downward revisions implies that analysts have previously over-estimated the target firm's future earnings. Indeed, analyst earnings forecasts tend to be over-optimistic, and downward revisions are the norm.

In this section, we test whether option listings have an influence on earnings revisions. How might this happen? As we find above, on average, more analysts follow stocks after they are optioned, and their forecast accuracy is improved. More generally, trading in options should enhance the information environment in which the stock is traded. If the revisions are made more quickly, or if the basic level of over-optimism is reduced, then this can be interpreted as evidence of a positive informational effect of options listing.

We focus on analysts' forecasts of current fiscal year's EPS (FY1) both because there is a longer historical record of data and because more analysts issue this estimate. Our measure of estimate revision is the difference between two consecutive monthly estimates, scaled either by the corresponding stock price or the absolute value of previous month's estimate:

[ER_P.sub.j,t] = 100 * ([FY1.sub.j,t] - [FY1.sub.j,t-1])/[Price.sub.j,t-1] (6a)

[ER_FY.sub.j,t] = 100 * ([FY1.sub.j,t] - [FY1.sub.j,t-1])/[absolute value of [FY1.sub.j,t-1]] (6b)

Where:

[ER_P.sub.j,t] = Estimate revision as percent of base month stock price

[ER_FY.sub.j.t] = Estimate revision as percentage change in forecast value

[FY1.sub.j,t] = Consensus estimate in month t of firm j's earnings for the current fiscal year, FY1

[FY1.sub.j,t] = Consensus estimate in month t-1 of firm j's earnings for the current fiscal year, FY1

[Price.sub.j.t-1] = Stock price in prior month t-1

Cumulative three-, six-, and twelve-month revisions of current fiscal year earnings estimates after option introduction are compared to the corresponding revisions a year earlier to control for possible seasonal factors. (xv) Each optioned firm's difference in estimate revision before and after option introduction is also adjusted by the control stock's difference during the same period.

The median estimate revisions adjusted for seasonal trends are shown in Table 8, Column (3), and the percentage of newly-optioned firms with positive adjusted estimate revisions are in Column (4). The median revisions adjusted for control stocks and the percentage of positive changes are presented in Columns (5) and (7), respectively.

The results for NYSE/Amex stocks are in Panel A, Table 8. The revisions based on both Equations (6a) and (6b) are overwhelmingly negative (see Columns (3) and (5)). This reflects the fact that the percent of positive observations in Columns (4) and (7) in Table 8 are less than 50%. This indicates that analysts as a whole lower their FYI estimates after options listings. The statistical significance is confirmed by the low p-values in Columns (6) and (8), for 6- and 12-month periods. The results for Nasdaq stocks are in Panel B, Table 8. While these revisions are also negative, they are less significant than for the NYSE/Amex stocks.

Analysts generally are overly optimistic at the beginning of the fiscal year and systematically revise their forecasts downward as the year proceeds. In our sample, EPS estimates of both groups of stocks are revised downwards, but the NYSE/ Amex revisions are more significant. Above we find that more new analysts initiate coverage of Nasdaq stocks after option introduction than of NYSE/Amex stocks. Since Nasdaq's listing requirements are less restrictive than those on the NYSE, it is plausible that the earnings outlook for most Nasdaq stocks are, and remain, more uncertain than those of the larger NYSE/Amex stocks, even after option introduction. It also may be that the new analysts tend to issue slightly less optimistic estimates to avoid possible legal repercussions of overly-optimistic estimates.

V. Conclusions

This study examines the informational environment of the market after options are introduced on NYSE/Amex and Nasdaq stocks between 1983 and 2004. We compare the optioned sample to a similar non-optioned control sample. We find that option listing results in the production of important additional information about the firm. The forecast accuracy of analyst earnings improves significantly and the improvement appears to be more significant for Nasdaq than for NYSE/Amex stocks.

Empirically we find larger increases in analyst coverage of Nasdaq stocks following option introduction. Analysts' forecasts of Nasdaq stock earnings are significantly improved in the two to six months after option introduction, while those of NYSE/Amex stocks are not, all relative to similar non-optioned control stocks.

The change in responsiveness of optioned stock prices to earnings surprises is more significant for optioned than non-optioned stocks, and optioned Nasdaq stocks display additional response to positive earnings surprises.

Finally, revisions of earnings estimates are generally negative after option introduction--a phenomenon optioned stocks have in common with non-optioned stocks in general. The revisions for optioned Nasdaq stocks are less significant than those of optioned NYSE/Amex stocks. One possible explanation is that even after informational benefits following option introduction are realized, the prospects and outlooks of Nasdaq stocks remain more difficult to analyze than of NYSE/Amex stocks.

We conclude that options enhance the informational environment of all stocks on which they are traded, and that they matter more for smaller, lower-priced stocks with less analyst attention before they were optioned. Our findings extend the literature on the informational benefits of option introduction to show that they are not confined to, or even largely concentrated on, the larger, better-known NYSE/ Amex stocks that were the first to be optioned.

Our findings extend previous research on market completeness in a critical way because they show a new type of evidence about the beneficial effects of options in terms of enhancing market completeness. The fact that most of the large, heavily-traded, and followed stocks have already been optioned suggests that the financial markets--as a whole as well as for those particular stocks--are now significantly more developed than before. Our evidence suggests that the benefits gained to date are not exhaustive, and that option introductions on smaller, less well-known stocks, continue making additional contributions to the market environment.

References

Ackert, Lucy F., and George Athanassakos. 2003 "A simultaneous equations analysis of analysts' forecast bias, analyst following, and institutional ownership," .Journal of Business Finance & Accounting, 30(7) & (8), 1017-1041.

Ali, Ashiq, April Klein, and James Rosenfeld. 1992 "Analysts' use of information about permanent and transitory earnings components in forecasting annual EPS," The Accounting Review, 67(1), 183-198.

Arbel, Avner, and Paul Strebel. 1982 "The neglected and small firm effects," The Financial Review, 17(4), 201-218.

Arditti, Fred D., and Kose John. 1980 "Spanning the State Space with Options," Journal of Financial and Quantitative Analysis, 15, 1-9.

Bansal, Vipul K., Pruitt, Stephen W., and Wei, K. C. John. 1989 "An empirical reexamination of the impact of CBOE option initiation on the volatility and trading volume of the underlying equities: 1973-1986." Financial Review, 24(1), 19-30.

Barth, Mary E., Ron Kasznik, and Maureen F. McNichols. 2001 "Analyst coverage and intangible assets," Journal of Accounting Research, 39(1), 1-34.

Bollen, Nicolas P. B. 1998 "A note on the impact of options on stock return volatility." Journal of Banking and Finance, 22(9) 1181-1191.

Brous, Peter A., 1992 "Common Stock Offerings and Earnings Expectations: A Test of the Release of Unfavorable Information," The Journal of Finance, 47(4), 1517-1536.

Chakravarty, Sugato, Huseyin Gulen, and Stewart Mayhew, 2004 "Informed trading in stock and option markets," The Journal of Finance, 59(3), 1235-1257.

Chang, Kee H., and Hoje Jo. 1996 "The impact of security analysts' monitoring and marketing functions on the market value of firms," Journal of Financial and Quantitative Analysis, 31 (4), 493-512.

Conrad, Jennifer. 1989 "The price effect of option introduction," The Journal of Finance, 44(2), 487-498.

Damodaran, Aswath, and Joseph Lira. 1991 "The effect of options listing on the underlying stocks' return processes," Journal of Banking and Finance, 15(3), 647-664.

Detemple, Jerome, and Philippe Jorion. 1990 "Option listing and stock returns: an empirical analysis," Journal of Banking and Finance, 14(4), 781-801.

Detemple, Jerome, and Larry Seldon. 1991 "A general equilibrium analysis of option and stock market interactions," International Economic Review, 32(2), 279-303.

Dimitrov, Valentin, Prem Jain, and Sheri Tice. 2006 "Sell on the news: information uncertainty and returns around earnings announcements." Working paper, SSRN-id = 938749.

Dugar, Amitabh, and Siva Nathan. 1995 "The effects of investment banking relationships on financial analysts' earnings forecasts and investment recommendations," Contemporary Accounting Research, 12(1), 131-160.

Easley, David, Maureen O'Hara, and P. S. Srinivas. 1998 "Option volume and stock prices: Evidence on where informed traders trade," The Journal of Finance, 53(2), 431-465.

Faff, Robert, and David Hilliel. 2005 "Complete markets, informed trading and equity option introductions," Journal of Banking and Finance, 29(6), 1359-1384.

Figlewski, Stephen, and Gwendolyn Webb. 1993 "Options, short sales, and market completeness," The Journal of Finance, 48(2), 761-777.

Freund, Steven, P. McCann, and Gwendolyn Webb. 1994 "A regression analysis of the effects of option introduction on stock variances." The Journal of Derivatives, 1 (3), 25-38.

Ho, Li-Chin Jennifer. 1993 "Option trading and the relation between price and earnings: a crosssectional analysis." The Accounting Review, 68(2), 368-384.

Ho, Li-Chin Jennifer, John M. Hassel, and Steve Swidler. 1995 "An empirical examination of the dispersion and accuracy of analyst forecasts surrounding option listing," Review of Financial Economics, 4(2), 171-185.

Hong, Harrison, Terrence Lim, and Jeremy C. Stein. 2000 "Bad news travels slowly: size, analyst coverage, and the profitability of momentum strategies," The Journal of Finance, 55(1), 265-295.

Jennings, Robert H., and Laura T. Starks. 1986 "Earnings announcements, stock price adjustments, and the existence of option markets," The Journal of Finance, 41(1), 107-125.

Jubinski, Daniel, and Marc Tomljanovich. 2007 "Options listings and individual equity volatility." The Journal of Futures Markets, 27(1), 1-27.

Jubinski, Daniel, and Marc Tomljanovich. 2003 "An examination of the introduction of options on the volatility of the underlying individual equity securities." Working paper, SSRN-id = 479841.

Kecskes, Ambrus, and Kent L. Womack. 2007 "Adds and drops of analyst coverage: does the stock market overreact?" Working paper, January 28, SSRN-id = 960501.

Kim, Sunyoung, and Jason Lee. 2006 "Effect of option listing on price reactions to earnings announcements," Working paper, SSRN-id = 874857.

Kumar, Raman, Atulya Sarin, and Kuldeep Shastri. 1998 "The impact of options trading on the market quality of the underlying security: an empirical analysis," The Journal of Finance, 53(2), 717-732.

Manaster, Steven, and Richard J. Jr. Rendleman. 1982 "Option prices as predictors of equilibrium stock prices," The Journal of Finance, 37(4), 1043-1057.

Mayhew, Stewart. 2002 "Competition, market structure, and bid-ask spreads in stock option markets" The Journal of Finance, 57(2), 931-958.

Mayhew, Stewart, and Vassil Mihov. 2004 "How do exchanges select stocks for option listing?" The Journal of Finance, 59(1), 447-471.

Mazouz, Khelifa, 2004 "The effect of CBOE option listing on the volatility of NYSE traded stocks: a time-varying variance approach." The Journal of Empirical Finance, 11(5), 695-708.

McNichols, Maureen, and Patricia C. O'Brien. 1997 "Self-selection and analyst coverage," Journal of Accounting Research, 35(Supplement), 167-199.

Mendenhall, Richard R., and Donald H. Fehrs. 1999 "Option listing and the stock-price response to earnings announcements," Journal of Accounting and Economics, 27, 57-87.

Merton, Robert. 1987 "A simple model of capital market equilibrium with incomplete information," Journal of Finance, 42(3), 483-510.

Moyer, R. Charles, Robert E. Chatfield, and Phillip M. Sisneros. 1989 "Security analyst monitoring activity: agency costs and information demands," Journal of Financial and Quantitative Analysis, 24(4), 503-512.

O'Brien, Patricia C., and Ravi Bhushan. 1990 "Analyst following and institutional ownership," Journal of Accounting Research, 28(Supplement), 55-76.

Roll, Richard, Eduardo Schwartz, and Avanidhar Subrahmanyam. 2007 "Options trading activity and firm valuation," Working paper, SSRNid = 1087112.

Ross, Stephen A. 1976 "Options and Efficiency," Quarterly Journal of Economics, 90, 75-89.

Skinner, Douglas J. 1989 "Options markets and stock return volatility," Journal of Financial Economics, 23(1), 61-78.

Skinner, Douglas J. 1990 "Options markets and the information content of accounting earnings releases," Journal of Accounting and Economic's, 13(3), 191-211.

Sorescu, Sorin. M. 2000 "The effect of options on stock prices: 1973-1995." The Journal of Finance, 55(1), 487-514.

St. Pierre, Ellen F., 1998 "The impact of option introduction on the conditional return distribution of underlying securities." The Financial Review, 33(1), 105-119.

Notes

(i.) More specifically, Ross (1976) models a market in which there are n primitive securities and m possible states. The number of possible states, m, is higher than the number of primitive securities, n. This market is incomplete in the sense that the available supply of n securities is not sufficient to allow investors to design an investment portfolio that has desired coverage across all possible outcome states. Technically, the rank of the state-space tableau is not full. Ross shows how options can be written on the underlying primitive securities so that investors can achieve their desired coverage across the states. In his model, the market is complete when options on the primitive securities augment the rank of the state-space tableau to the point that it is full.

(ii.) Although Ross's analysis does not fully cover cases when portfolio returns are required to be non-negative, Arditti and John (1980) resolve this problem by identifying conditions under which the full state space can be spanned by put options.

(iii.) Conrad also finds that the excess return volatility declines after introduction and that systematic risk is unchanged. Similarly, Skinner (1989), Damodaran & Lim (1991), and Bansal, Pruitt, & Wei (1989) also find decreases in volatility after introduction. Bollen (1998) examines a larger sample and compares post-introduction volatility to that of a control sample and concludes that options introduction has no effect on volatility. This is consistent with Freund, McCann, & Webb (1993), as well as in studies that test for changes in conditional volatility, including Jubinsky & Tomljanovich (2003, 2007), St. Pierre (1998), and Mazouz (2004).

(iv.) Additional aspects of options on the informational environment are addressed by Roll, Schwartz, and Subrahmanyam (2007), Faff and Hilliel (2005), Chakravarty, Gulen, & Mayhew (2004), and Kumar, Sarin & Shastri (1998), among others.

(v.) A different aspect of the informational environment is examined by Figlewski & Webb (1993), who hypothesize that options offer new, less restrictive ways for investors to act on negative information, and find evidence consistent with their hypothesis. This implies that options help reduce the amount of information that was previously excluded from the marketplace. They find evidence consistent with their hypothesis. This is significant because it implies that options help reduce the amount of information that was previously excluded from the marketplace. Their findings are supported by Sorescu (2000).

(vi.) Additional supporting findings in those of Chung and Jo (1996), who find that analysts help increase the extent of investor awareness of covered firms. O'Brien & Bhushan (1990) model the simultaneous determination of analyst coverage and institutional ownership. They find that analyst coverage increases more in firms (1) with smaller prior analyst following; (2) whose return volatility has declined; and (3) with firms in growing industries.

(vii.) Barth, Kasznik, & McNichols (2001) show that firms with significant intangible assets tend to have larger analyst following. They measure intangible assets by research and development (R&D) and advertising expenses, and by depreciation and recognized intangible assets, including goodwill. Firms with high R&D expenditures are often considered to be firms with substantial growth prospects. Additionally, Kecskes & Womack (2007) find that analysts tend to add coverage of stocks that are in greater demand by investors and with better operating performance. They tend to drop firms in less demand and with poorer performance.

(viii.) Kim & Lee (2006) examine reactions to earnings announcements of optioned stocks in 1985-2003. They find that the reactions of optioned stocks are more intense, and that their post-earnings announcement drift is smaller. They compare optioned firms to all other firms, and therefore do not form a well-defined control sample. Overall, their findings are consistent with those of Mendenhall & Fehrs.

(ix.) The trading volumes of the Nasdaq and NYSE/Amex groups cannot be compared directly because the trading volume data are generated differently in the two markets.

(x.) Control stocks are also required to have coverage in the Institutional Brokers Estimates System (IBES) at least six months before the option listing date of their underlying sample stock.

(xi.) We exclude the number of analysts in the option listing month, in part because IBES defines its statistical periods from midmonth to mid-month rather than on the basis of month-ends.

(xii.) As an example, consider Microsoft. Options were first listed on Microsoft stock on March 12, 1987. The average number of analysts issuing current fiscal year EPS estimates for Microsoft increased from 12.33 six months before the option listing date (9/86 to 2/87) to 21.67 six months afterward (4/87 to 9/87), resulting in an increase of 9.34 analysts. In percentage terms, this is an increase of 76%. Our control stock for Microsoft is IMS International Inc. During the six months before Microsoft was optioned, the average number of analysts following IMS was 11.33. The number increased to 11.83 during the six-month period after Microsoft was optioned, for an increase of just 0.50 analysts. The control-stock-adjusted change in the average number of analysts following Microsoft is thus 8.84 (= 9.34-0.50).

(xiii.) Because we define abnormal returns based on a market adjustment, we do not employ a separate estimation window outside of our event window.

(xiv.) The combination of a lower intercept and a higher slope means that the linear relation between news and stock price response has rotated counter-clockwise in space defined with earnings surprises on the x-axis, and stock abnormal returns on the y-axis.

(xv.) Brous (1992) shows that the monthly forecast revisions are serially correlated because not all analysts update their forecasts on a monthly basis.

Susana Yu, Corresponding Author, Montclair State University, yus@mail.montclair.edu, (917) 834-5238

Kishore Tandon, Baruch College/The City University of New York, Kishore.Tandon@baruch.cuny.edu, (646) 312-3450

Gwendolyn Webb, Baruch College/The City University of New York, Gwendolyn.Webb@baruch.cuny.edu, (646) 312-3485

TABLE 1. Descriptive Data on Our Sample of Optioned Stocks NYSE/Amex Stocks Market Capital- Trading ization Stock Volume Year Number (mil) Price (% of S.O.) 1983 20 $533 $24.19 0.39 1984 15 741 26.38 0.35 1985 23 2,003 40.88 0.23 1986 24 1,503 47.13 0.26 1987 50 1,707 36.94 0.26 1988 42 1,003 31.06 0.25 1989 38 756 33.81 0.28 1990 49 834 27.75 0.20 1991 37 565 25.13 0.30 1992 21 387 21.63 0.37 1993 39 478 23.25 0.31 1994 35 622 25.13 0.25 1995 53 463 25.88 0.29 1996 45 588 28.25 0.27 1997 101 621 26.13 0.26 1998 77 673 25.44 0.22 1999 34 525 18.94 0.31 2000 16 583 31.25 0.55 2001 28 938 29.93 0.45 2002 29 860 32.78 0.42 2003 12 992 29.05 0.43 2004 39 902 28.67 0.49 Total or 827 704 28.25 0.29 Median NYSE/Amex Stocks # of Standard Analysts Market/ Deviation Issuing Book Year of Return FYI Ratio 1983 3.09 10 2.56 1984 2.67 17 1.92 1985 1.54 21 1.71 1986 1.84 17 2.16 1987 2.05 15 2.56 1988 2.50 11 1.95 1989 1.81 14 2.14 1990 1.77 13 1.73 1991 2.50 7 2.82 1992 2.49 5 2.55 1993 2.39 7 2.32 1994 2.10 7 2.93 1995 2.18 6 2.94 1996 2.11 7 2.44 1997 2.05 6 2.31 1998 2.21 5 2.70 1999 3.57 5 2.26 2000 4.00 4 3.92 2001 3.56 5 2.94 2002 2.77 6 2.15 2003 2.43 5 2.15 2004 2.24 4 2.50 Total or 2.20 7 2.39 Median Nasdaq Stocks Market Capital- Trading ization Stock Volume Year Number (mil) Price (% of S.O.) 1983 0 -- -- -- 1984 1 $4,672 $64.25 0.27 1985 27 487 19.00 0.67 1986 10 1,198 32.50 0.42 1987 17 521 28.75 0.94 1988 33 510 21.25 0.40 1989 17 370 17.25 0.58 1990 39 434 22.50 0.56 1991 56 333 20.13 0.96 1992 47 255 20.25 0.93 1993 52 290 21.63 0.83 1994 87 280 19.13 0.80 1995 133 297 22.63 0.66 1996 178 342 24.56 0.81 1997 204 285 20.31 0.65 1998 162 336 22.63 0.64 1999 93 317 21.50 0.99 2000 112 631 32.66 1.10 2001 90 379 20.53 0.77 2002 42 341 18.27 0.56 2003 40 319 16.24 0.62 2004 87 367 18.89 0.74 Total or 1,527 353 21.88 0.75 Median Nasdaq Stocks Standard # of Devia- Analysts Market/ tion of Issuing Book Year Return FYI Ratio 1983 -- -- -- 1984 1.72 23 1.69 1985 2.67 16 2.63 1986 2.14 16 4.33 1987 3.05 14 4.44 1988 3.07 10 3.00 1989 2.43 11 2.79 1990 2.43 12 2.74 1991 3.75 8 3.79 1992 3.60 5 5.17 1993 3.35 5 4.10 1994 3.45 5 3.11 1995 3.41 5 3.81 1996 3.61 5 5.21 1997 3.62 4 3.73 1998 3.59 5 3.88 1999 5.14 4 5.68 2000 6.52 4 8.11 2001 5.29 4 5.40 2002 4.29 4 4.01 2003 4.11 3 5.56 2004 3.47 4 4.07 Total or 3.72 5 4.18 Median This table summarizes our sample of optioned stocks. Initial listing dates for all options traded on NYSE-Amex and Nasdaq from 1983-2004 are from the CBOE. We exclude subsequent option listings of the same stock on another exchange. Daily stock return and trading volume data are from CRSP. The trading volumes and standard deviations shown are based on daily data in the year before option introduction. Average EPS forecasts, the number of analysts that issue earnings forecasts, and actual quarterly earnings are from IBES. Data for the market-to-book ratios are from Compustat. TABLE 2. Results of Matching Control Stocks on the Basis of Share Turnover and Return Volatility Daily Trading Volume as Percent of Shares Outstanding Optioned Stocks Control Stocks Number of Year Pairs Mean Median Mean Median 1983 15 0.42 0.40 0.37 0.33 1984 10 0.35 0.37 0.30 0.30 1985 27 0.42 0.35 0.34 0.31 1986 21 0.41 0.26 0.43 0.26 1987 40 0.38 0.30 0.34 0.30 1988 41 0.41 0.30 0.36 0.26 1989 43 0.40 0.31 0.39 0.31 1990 59 0.44 0.31 0.40 0.31 1991 66 0.64 0.52 0.59 0.53 1992 46 0.80 0.72 0.66 0.55 1993 53 0.65 0.56 0.57 0.58 1994 94 0.67 0.62 0.62 0.52 1995 141 0.68 0.47 0.60 0.44 1996 175 0.81 0.74 0.78 0.69 1997 228 0.59 0.49 0.57 0.51 1998 193 0.56 0.46 0.53 0.45 1999 88 0.88 0.74 0.80 0.66 2000 110 1.11 1.03 1.05 0.98 2001 86 0.71 0.67 0.65 0.63 2002 51 0.62 0.56 0.51 0.49 2003 42 0.60 0.57 0.54 0.47 2004 88 0.72 0.62 0.62 0.51 Total or 1,717 0.66 0.55 0.61 0.51 Median Standard Deviation of Daily Returns Optioned Stocks Control Stocks Year Mean Median Mean Median 1983 2.91 3.04 3.01 3.03 1984 2.43 2.30 2.35 2.20 1985 2.15 1.85 2.10 1.78 1986 2.15 1.94 2.43 2.17 1987 2.13 2.07 2.20 2.13 1988 2.93 2.64 2.98 2.78 1989 2.01 1.89 2.06 2.10 1990 2.03 1.83 2.11 1.82 1991 3.34 3.20 3.35 3.22 1992 3.37 3.41 3.46 3.46 1993 3.11 2.99 3.46 3.28 1994 3.07 3.10 3.17 3.18 1995 3.12 3.19 3.23 3.21 1996 3.60 3.36 3.73 3.47 1997 3.22 3.20 3.40 3.35 1998 3.19 3.09 3.52 3.38 1999 4.97 4.79 5.19 5.35 2000 6.69 6.19 6.55 6.42 2001 5.19 5.04 6.04 5.87 2002 3.84 3.69 4.15 3.95 2003 4.32 3.80 4.69 3.92 2004 3.27 3.10 3.40 3.03 Total or 3.56 3.27 3.74 3.40 Median This table shows the mean and median trading volumes and standard deviations of daily returns for the stocks in our option sample and their control stocks, by year. Turnover is the daily trading volume as a percent of the number of shares outstanding. Return volatility is measured by the standard deviation of stock return. Both measures are calculated based on data for the 250 trading days before option listing. The daily data are from CRSP. In this table, we report the results of the control stock sample for the optioned NYSE/Amex and Nasdaq stocks combined. TABLE 3. The Number of Analysts Reporting EPS Forecasts around Option Introduction NYSE/Amex Stocks (Sample Size = 580) Average Number of Time Interval Analysts Reporting around Estimate Average Difference, Option Intro- Before After After-- duction Optioning Optioning Before (1) (2) (3) (4) Panel A: EPS Forecasts for the Current Fiscal Year 6 months 8.4 8.8 0.4 12 months 8.2 8.8 0.7 24 months 7.8 8.9 1.1 Panel B: EPS Forecasts for the Next Fiscal Year 6 months 6.5 6.8 0.3 12 months 6.2 6.9 0.6 24 months 5.9 7.0 1.1 Panel C: EPS Forecasts for the Current Fiscal Quarter 6 months 4.0 4.4 0.4 12 months 3.8 4.6 0.9 24 months 3.5 4.8 1.3 Panel D: EPS Forecasts for the Next Fiscal Quarter 6 months 2.4 2.7 0.3 12 months 2.3 2.9 0.6 24 months 2.2 3.1 0.9 Panel E: Forecast of EPS Growth Rate in Next 3-5 Years 6 months 4.5 4.7 0.3 12 months 4.3 4.8 0.5 24 months 4.1 4.8 0.7 NYSE/Amex Stocks (Sample Size = 580) Time Interval around Average Difference, After--Before, Option Intro- % Adjusted for duction Difference Control Stocks (1) (5) Panel A: EPS Forecasts for the Current Fiscal Year 6 months 4.8% 0.4 12 months 8.5% 0.6 24 months 14.1% 0.8 Panel B: EPS Forecasts for the Next Fiscal Year 6 months 4.6% 0.2 12 months 9.7% 0.6 24 months 18.6% 0.8 Panel C: EPS Forecasts for the Current Fiscal Quarter 6 months 10.0% 0.4 12 months 23.7% 0.7 24 months 37.1% 0.9 Panel D: EPS Forecasts for the Next Fiscal Quarter 6 months 12.5% 0.2 12 months 26.1% 0.4 24 months 40.9% 0.6 Panel E: Forecast of EPS Growth Rate in Next 3-5 Years 6 months 6.7% 0.2 12 months 11.6% 0.3 24 months 17.1% 0.5 Nasdaq Stocks (Sample Size = 1,137) Average Number of Time Interval Analysts Reporting around Estimate Average Difference, Option Intro- Before After After-- duction Optioning Optioning Before (1) (6) (7) (8) Panel A: EPS Forecasts for the Current Fiscal Year 6 months 5.0 5.7 0.7 12 months 4.7 5.9 1.2 24 months 4.3 6.1 1.8 Panel B: EPS Forecasts for the Next Fiscal Year 6 months 4.1 4.8 0.7 12 months 3.9 4.9 1.0 24 months 3.6 5.1 1.5 Panel C: EPS Forecasts for the Current Fiscal Quarter 6 months 3.5 4.1 0.7 12 months 3.2 4.3 1.1 24 months 3.0 4.6 1.6 Panel D: EPS Forecasts for the Next Fiscal Quarter 6 months 2.7 3.2 0.6 12 months 2.5 3.4 0.9 24 months 2.3 3.6 1.3 Panel E: Forecast of EPS Growth Rate in Next 3-5 Years 6 months 2.5 2.9 0.4 12 months 2.4 3.1 0.7 24 months 2.2 3.3 1.1 Nasdaq Stocks (Sample Size = 1,137) Time Interval around Average Difference, After--Before, Option Intro- % Adjusted for duction Difference Control Stocks (1) (9) Panel A: EPS Forecasts for the Current Fiscal Year 6 months 14.0% 0.7 12 months 25.5% 1.1 24 months 41.9% 1.6 Panel B: EPS Forecasts for the Next Fiscal Year 6 months 17.1% 0.7 12 months 25.6% 1.0 24 months 41.7% 1.4 Panel C: EPS Forecasts for the Current Fiscal Quarter 6 months 20.0% 0.7 12 months 34.4% 1.0 24 months 53.3% 1.4 Panel D: EPS Forecasts for the Next Fiscal Quarter 6 months 22.2% 0.5 12 months 36.0% 0.8 24 months 56.5% 1.1 Panel E: Forecast of EPS Growth Rate in Next 3-5 Years 6 months 16.0% 0.4 12 months 29.2% 0.6 24 months 50.0% 0.8 This table shows the number of analysts submitting EPS forecasts for the current fiscal year on the optioned and control stocks in our sample. The data are from IBES. We include the numbers of analysts covering the stock for two years before and two years after the options-listing month. We exclude the options-listing month itself. TABLE 4. Regression Analysis of Percentage Change in Analyst Coverage after Option Introduction Dependent Variables: Percentage Change in Average Number of Analysts After Option Introduction 6 Months After Independent Variable Coefficient p-value Intercept 40.8860 0.0000 Nasdaq Dummy Variable (Value = 1 if 6.2306 0.0099 Nasdaq stock, = 0 if NYSE/Amex stock) Market Capitalization ($Millions) 0.0095 0.0000 Dummy Variable for Option Introductions -9.9084 0.0090 after 1990 (Value = 1 if after 1990, = 0 if 1990 or before) Number of Analysts 6 Months before -3.5635 0.0000 Option Introduction Number of Analysts 12 Months before -- -- Option Introduction Number of Analysts 24 Months before -- -- Option Introduction Number of Observations 1,717 Adjusted R Square 0.0961 Standard Error 43.3682 F Ratio 46.6 p-value 0.0000 Dependent Variables: Percentage Change in Average Number of Analysts After Option Introduction 12 Months After Independent Variable Coefficient p-value Intercept 70.8998 0.0000 Nasdaq Dummy Variable (Value = 1 if 8.0939 0.0124 Nasdaq stock, = 0 if NYSE/Amex stock) Market Capitalization ($Millions) 0.0167 0.0000 Dummy Variable for Option Introductions -17.5682 0.0006 after 1990 (Value = 1 if after 1990, = 0 if 1990 or before) Number of Analysts 6 Months before -- -- Option Introduction Number of Analysts 12 Months before -6.0860 0.0000 Option Introduction Number of Analysts 24 Months before -- -- Option Introduction Number of Observations 1,717 Adjusted R Square 0.1396 Standard Error 58.0327 F Ratio 70.6 p-value 0.0000 Dependent Variables: Percentage Change in Average Number of Analysts After Option Introduction 24 Months After Independent Variable Coefficient p-value Intercept 107.2266 0.0000 Nasdaq Dummy Variable (Value = 1 if 11.4081 0.0068 Nasdaq stock, = 0 if NYSE/Amex stock) Market Capitalization ($Millions) 0.0263 0.0000 Dummy Variable for Option Introductions -30.9927 0.0000 after 1990 (Value = 1 if after 1990, = 0 if 1990 or before) Number of Analysts 6 Months before -- -- Option Introduction Number of Analysts 12 Months before -- -- Option Introduction Number of Analysts 24 Months before -9.1876 0.0000 Option Introduction Number of Observations 1,717 Adjusted R Square 0.1610 Standard Error 75.4258 F Ratio 83.3 p-value 0.0000 This table shows the result of regressions of the percentage change in analysts submitting EPS estimates for the current fiscal year in each of three periods around option introduction. The independent variables include: (1) a Nasdaq dummy variable; (2) the firm's market capitalization; (3) a post-1990 dummy variable; and (4) the number of analysts that followed the stock before it was optioned. The regression is estimated using OLS. TABLE 5. Standardized Forecast Errors of Current Fiscal Year EPS Month Number (Relative to of Median month of Observa- SFE after optioning) tions Optioning (1) (2) (3) Panel A: NYSE/Amex Stocks 0 315 -1.8 1 329 -0.5 2 333 -1.5 3 340 -0.4 4 345 -1.6 5 371 -2.0 6 355 -1.9 7 356 -2.6 8 359 -2.4 9 353 -2.5 10 359 -0.3 11 355 0.0 Panel B: Nasdaq Stocks 0 499 -2.3 1 526 -3.0 2 549 -2.8 3 578 -3.4 4 610 -3.8 5 655 -2.5 6 659 -3.3 7 680 -2.8 8 643 -2.2 9 638 -2.6 10 621 -3.3 11 595 -2.5 Month Median Difference: p-value of (Relative to SFE Post Option Wilcoxon Sign month of Listing Minus SFE 1 Rank Test: optioning) Year Ago Difference = 0 (1) (4) (5) Panel A: NYSE/Amex Stocks 0 -0.9 0.316 1 -2.2 0.492 2 1.6 0.888 3 -0.1 0.352 4 2.0 0.796 5 -0.1 0.285 6 -1.1 0.119 7 -3.3 0.033 8 -4.6 0.001 9 -5.3 0.027 10 -2.0 0.164 11 -1.3 0.212 Panel B: Nasdaq Stocks 0 -0.8 0.164 1 -1.4 0.284 2 -3.1 0.063 3 -3.8 0.007 4 -4.9 0.000 5 -2.0 0.099 6 -2.6 0.049 7 0.3 0.860 8 -1.5 0.348 9 -1.4 0.230 10 -0.2 0.464 11 -0.9 0.161 Month Percent of Firms p-value of (Relative to with Increase in Sign Test: month of SFE after Option Percent = optioning) Listing 50 (1) (6) (7) Panel A: NYSE/Amex Stocks 0 48.9 0.368 1 47.7 0.220 2 51.7 0.744 3 49.7 0.478 4 53.0 0.881 5 49.6 0.459 6 47.9 0.229 7 44.7 0.025 8 42.1 0.002 9 42.5 0.003 10 46.5 0.103 11 46.5 0.102 Panel B: Nasdaq Stocks 0 49.3 0.394 1 47.5 0.138 2 46.3 0.044 3 44.1 0.003 4 42.1 0.000 5 46.6 0.043 6 45.5 0.012 7 50.1 0.546 8 49.0 0.318 9 48.4 0.226 10 49.8 0.468 11 49.1 0.341 This table shows the standardized forecast errors after option introduction for our sample of NYSE-Amex and Nasdaq optioned stocks. The standardized forecast error (SFE) in Column (3) shows forecast errors of the optioned stocks minus the corresponding forecast error of their control stocks. The differences are also shown relative to their values a year earlier to correct for any seasonal effects (Column (4)). The sample covers 1985-2004 because of limitations of the IBES data. TABLE 6. Regression Analysis of Earnings Announcement Surprises Sample Stocks Only Independent Variable Description Coefficient p-Value Panel A: Positive Earnings Surprises, +5% or More Intercept 1.2345 0.0007 E_Surp_Pos Value of Earnings 0.0033 0.0024 Surprises greater than or equal to 5% D_Option_Post Dummy Variable for -2.3826 0.0000 optioned stock post introduction Value = 1 for optioned stock, 0 otherwise E_Surp_Pos * Interaction Term for 0.9983 0.0000 D_Option_Post Earnings Surprises and Post Option Introduction for Sample Stocks D_Nasdaq Dummy Variable for 0.4439 0.0408 Nasdaq Stocks Value = 1 if stock is traded in Nasdaq, 0 otherwise D_Post_1990 Dummy Variable for 0.7748 0.0341 Time Period Value = 1 if the announcement was made after 1990, and 0 otherwise D_Option_Stock Dummy Variable for -- -- Optioned Stock Value = 1 if the stock is in the optioned sample, 0 otherwise D_Control_Post Dummy Variable for -- -- for Control Stock after Matching Optioned Stock's Option Introduction Value = 1 if the announcement is for a control stock after its matching optioned stock has been optioned. Number of 5,516 Observations Adj R Squared 0.4962 Standard Error 7.0560 F Ratio 1,087.18 p-value 0.0000 Independent Sample Stocks Only Variable Description Coefficient p-Value Panel B: Negative Earnings Surprises, -5% or Lower Intercept -0.2667 0.3666 E_Surp_Neg Value of Earnings 0.0002 0.5137 Surprise less than or equal to -5% D_Option_Post Dummy Variable for 1.2546 0.0000 optioned stock post introduction Value = 1 for optioned stock, 0 otherwise E_Surp_Neg * Interaction Term for 0.9970 0.0000 D_Option_Post Earnings Surprises and Post Option Introduction for Sample Stocks D_Nasdaq Dummy Variable for -0.1217 0.6046 Nasdaq Stocks Value = 1 if stock is traded in Nasdaq, 0 otherwise D_Post_1990 Dummy Variable for -1.0234 0.0010 Time Period Value = 1 if the announcement was made after 1990, and 0 otherwise D_Option_Stock Dummy Variable for -- -- Optioned Stock Value = 1 if the stock is in the optioned sample, 0 otherwise D_Control_Post Dummy Variable for -- -- for Control Stock after Matching Optioned Stock's Option Introduction Value = 1 if the announcement is for 2,944 a control stock after its matching optioned stock has been optioned. Number of Observations Adj R Squared 0.6607 Standard Error 5.7907 F Ratio 1,147.37 p-value 0.0000 All Sample and Control Stocks Independent Variable Description Coefficient p-Value Panel A: Positive Earnings Surprises, +5% or More Intercept 0.3363 0.3053 E_Surp_Pos Value of Earnings 0.0043 0.0000 Surprises greater than or equal to 5% D_Option_Post Dummy Variable for -2.3781 0.0000 optioned stock post introduction Value = 1 for optioned stock, 0 otherwise E_Surp_Pos * Interaction Term for 0.9979 0.0000 D_Option_Post Earnings Surprises and Post Option Introduction for Sample Stocks D_Nasdaq Dummy Variable for 0.2686 0.1653 Nasdaq Stocks Value = 1 if stock is traded in Nasdaq, 0 otherwise D_Post_1990 Dummy Variable for 0.7835 0.0092 Time Period Value = 1 if the announcement was made after 1990, and 0 otherwise D_Option_Stock Dummy Variable for 0.9772 0.0000 Optioned Stock Value = 1 if the stock is in the optioned sample, 0 otherwise D_Control_Post Dummy Variable for -0.0005 0.9983 for Control Stock after Matching Optioned Stock's Option Introduction Value = 1 if the announcement is for a control stock after its matching optioned stock has been optioned. Number of 9,798 Observations Adj R Squared 0.2905 Standard Error 8.2327 F Ratio 574.18 p-value 0.0000 All Sample and Control Independent Stocks Variable Description Coefficient p-Value Panel B: Negative Earnings Surprises, -5% or Lower Intercept -0.3727 0.3150 E_Surp_Neg Value of Earnings 0.0000 0.8545 Surprise less than or equal to -5% D_Option_Post Dummy Variable for 1.2676 0.0003 optioned stock post introduction Value = 1 for optioned stock, 0 otherwise E_Surp_Neg * Interaction Term for 0.9950 0.0000 D_Option_Post Earnings Surprises and Post Option Introduction for Sample Stocks D_Nasdaq Dummy Variable for -0.4304 0.0962 Nasdaq Stocks Value = 1 if stock is traded in Nasdaq, 0 otherwise D_Post_1990 Dummy Variable for -1.3046 0.0001 Time Period Value = 1 if the announcement was made after 1990, and 0 otherwise D_Option_Stock Dummy Variable for 0.5120 0.1257 Optioned Stock Value = 1 if the stock is in the optioned sample, 0 otherwise D_Control_Post Dummy Variable for -0.9631 0.0016 for Control Stock after Matching Optioned Stock's Option Introduction Value = 1 if the announcement is for 6,601 a control stock after its matching optioned stock has been optioned. Number of Observations Adj R Squared 0.2569 Standard Error 9.2257 F Ratio 326.96 p-value 0.0000 This table shows the results of estimating a regression equation of stock abnormal returns around earnings announcements. The dependent variable is the cumulative abnormal return in the event window (0, +2). The abnormal returns are market-adjusted, based on the contemporaneous return on the S&P 500 Index. The dependent variables include: (1) degree of earnings surprise, measured by the difference between the actual EPS figure and the consensus estimate in IBES; (2) a post option-introduction variable; (3) an interaction term between earnings surprises and option introduction; (4) a Nasdaq dummy variable; (5) a dummy variable for introductions in 1990 or later; (6) a dummy variable for the control stocks; and (7) a dummy variable for the control stock for the period after its underlying sample stock was optioned. The regression is run separately for positive earnings surprises that are 5% or more (Panel A) and negative earnings surprises that are -5% or more (Panel B). TABLE 7. Cumulative Average Returns around Earnings Surprises before and after Option Introduction Positive Earnings Surprises Pre Option Post Option Introduction Introduction Panel A: Comparison of Optioned vs. Control Stocks Optioned Stocks CAR 11.56% 2.95% No. 2,861 2,654 Control Stocks CAR 4.61% 2.77% No. 2,289 2,012 Panel B: Comparison of Optioned Nasdaq vs. Optioned NYSE/Amex Stocks Nasdaq CAR 12.58% 3.13% No. 2,088 1,882 NYSE CAR 8.82% 2.49% No. 773 772 Negative Earnings Surprises Pre Option Yost Option Introduction Introduction Panel A: Comparison of Optioned vs. Control Stocks Optioned Stocks CAR 4.44% -4.86% No. 1,273 1,697 Control Stocks CAR -2.61% -4.45% No. 1,727 1,899 Panel B: Comparison of Optioned Nasdaq vs. Optioned NYSE/Amex Stocks Nasdaq CAR 5.98% -5.63% No. 803 1,098 NYSE CAR 1.95% -3.44% No. 470 599 This table shows cumulative average returns (CARS) in the event window (-9, +30) around earnings announcements for optioned stocks in our sample and their matching control stocks before and after option introduction. The event day is the announcement of earnings. The excess returns are adjusted for the contemporaneous returns of the S&P 500 Index. Earnings surprises are the difference between the actual EPS figure and the consensus forecast in IBES. The positive earnings surprises are those +5% or more, and the negative ones are -5% or lower. TABLE 8. Revisions of Earnings Estimates Median revision of earnings of optioned stocks relative to revision 1 year ago Month Number % of Optioned (Relative to of Percent Stocks with Month of Observa- Revision Positive Optioning) (ions in EPS Revision (1) (2) (3) (4) Panel A: NYSE Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 260 -0.1 41.4 6 months 413 -0.3 37.5 12 months 507 -0.5 36.3 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 276 -1.3 42.2 6 months 436 -3.8 36.6 12 months 524 -8.2 34.5 Panel B: Nasdaq Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 378 -0.1 41.6 6 months 712 -0.1 41.4 12 months 951 -0.5 32.9 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 389 -1.2 43.9 6 months 734 -2.7 41.9 12 months 973 -10.4 33.8 Median revision of earnings of optioned stocks relative to revision 1 year ago and adjusted for earnings revisions of control stocks Month p-value of (Relative to Percent Wilcoxon Month of Revision Sign-Rank Test: Optioning) in EPS Difference = 0 (1) (5) (6) Panel A: NYSE Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 0.0 0.639 6 months -0.2 0.074 12 months -0.5 0.023 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 0.1 0.834 6 months -2.7 0.083 12 months -7.5 0.084 Panel B: Nasdaq Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 0.0 0.484 6 months 0.2 0.337 12 months -0.2 0.697 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 1.5 0.787 6 months 0.1 0.184 12 months -5.8 0.156 Median revision of earnings of optioned stocks relative to revision 1 year ago and adjusted for earnings revisions of control stocks Month % of Optioned p-value of (Relative to Stocks with Sign Test: Month of Positive Percent = Optioning) Revision 50 (1) (7) (8) Panel A: NYSE Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 50.0 0.951 6 months 45.0 0.050 12 months 43.2 0.003 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 50.7 0.764 6 months 45.2 0.050 12 months 43.7 0.005 Panel B: Nasdaq Stocks Estimate revision for current fiscal year EPS standardized by previous stock price 3 months 50.0 0.959 6 months 53.7 0.047 12 months 48.6 0.399 Estimate revision for current fiscal year EPS standardized by previous estimate of the same fiscal year's EPS 3 months 52.2 0.362 6 months 50.1 0.912 12 months 46.6 0.035 This table shows percent changes in earnings estimates for the current fiscal year at different intervals in the years 1983-2004. Revisions shown in Column (3) are adjusted for comparable revisions a year earlier. Revisions shown in Column (5) are also adjusted for control stocks.

Printer friendly Cite/link Email Feedback | |

Author: | Yu, Susana; Tandon, Kishore; Webb, Gwendolyn |
---|---|

Publication: | American Economist |

Date: | Sep 22, 2010 |

Words: | 13421 |

Previous Article: | The effects of excise taxes on non-homogeneous populations. |

Next Article: | Memorializing Paul A. Samuelson: a review of his major works, 1915-2009. |

Topics: |