Printer Friendly

Explaining the NYSE listing choices of NASDAQ firms.

Arnold R. Cowan and Ajai K. Singh are Assistant Professors, and Richard B. Carter and Frederick H. Dark are Associate Professors, all with the Department of Finance, College of Business, Iowa State University, Ames, Iowa.

* Recent studies report increases in shareholder wealth associated with announcements of New York Stock Exchange listings by corporations previously traded over-the-counter. In light of this evidence, it becomes important to understand why some firms, which apparently could list their stocks on the New York Stock Exchange, fail to do so. Either the managers of the firms are acting suboptimally, from their shareholders' point of view, or the net benefits of an exchange listing are not uniformly positive for all eligible companies.

This paper examines the characteristics of NASDAQ firms that are eligible to list on the New York Stock Exchange. We document the characteristics of firms that remain in the NASDAQ system in contrast to those that list on the NYSE. Market microstructure theory implies that firms list their stocks on major exchanges to reduce transaction costs to their investors. However, the theory also suggests that the potential cost-reducing benefits of an exchange listing are not necessarily uniform across all eligible firms. We discuss these points in more depth in Section I.

The empirical results show that firms that leave NASDAQ to list on the New York Stock Exchange tend to have smaller stock market capitalization, fewer shareholders, fewer market makers and smaller prices per share than NASDAQ firms that could list but do not. Listing firms have larger volume, on average, than qualified nonlisting firms, but trade in smaller dollar amounts per transaction. Firms with larger unexpected bid-ask spreads are more likely to move. The results are consistent with the idea that firms are more likely to list on the NYSE when the perceived benefits, including increased liquidity, are greater.

Firms that list on the NYSE tend to do so after a period of strong earnings growth, relative to eligible nonlisting firms. This suggests that the pre-listing stock price run-ups reported elsewhere are not due solely to leakage of private information about the impending listing announcement. Consistent with the growth result, firms that move to the exchange tend to have been qualified to move for a shorter time than other firms in the same industry that could move but do not. Firms appear to decide soon after they qualify for the NYSE whether they will move to the exchange, on average. This is also consistent with the reduction of estimation risk as a motive for listing, which we discuss in Section III.B.

Listing firms are less likely than other qualified firms to have dual or multiple classes of traded common stock. Voting rights typically differ across classes of common stock, and the NYSE discourages unequal voting rights as a matter of policy. Firms with dual classes may choose not to list on the exchange to avoid being forced to equalize voting rights across classes. This is consistent with the idea that the costs of an exchange listing are unequally distributed among qualified firms.

I. Background

A. Research on Exchange Listings

Research on stock exchange listings dates back more than 50 years. Early studies offer mixed conclusions as to whether or not exchange listing is beneficial. Some studies suggest that listing reduces systematic risk and the cost of equity capital, but they do not specify in detail the mechanism that causes the cost or risk reduction.(1) Using modern event study methods, however, Sanger and McConnell |35~ report that over-the-counter firms listing on the New York Stock Exchange experience positive abnormal stock returns when the intended listing is announced. They also find that the average stock price reaction is significantly smaller in the period following the introduction of the NASDAQ system in 1972. The advent of NASDAQ is widely perceived to have narrowed the advantage that the NYSE had, relative to the over-the-counter market, in the operational efficiency of trading. Sanger and McConnell |35~ infer from their results that the benefits of an exchange listing derive from improvements in the way that a firm's stock is traded.(2)

One recent study takes advantage of the theory of market microstructure to specify how exchange listing increases shareholder wealth. Grammatikos and Papaioannou |21~ suggest that listing on the New York Stock Exchange reduces the bid-ask spread because of advantages inherent in the exchange's specialist system. Amihud and Mendelson |1~ show analytically that the required rate of return on a security increases with the bid-ask spread. Listing may reduce the cost of equity by reducing the bid-ask spread. Assuming that the firm's expected future cash flows are not reduced, a reduction in the required rate of return implies that the market value of equity will increase. Listing thus may increase the wealth of existing shareholders. Grammatikos and Papaioannou report empirical evidence that supports this argument. They find that the abnormal return at the announcement of listing is more positive, the greater the percentage bid-ask spread in the pre-listing market for the stock. Edelman and Baker |17~ report a similar result for firms listing on the American Stock Exchange.

Like other recent studies, Grammatikos and Papaioannou |21~ use an ex post sample of firms that actually list on the NYSE. The use of an ex post sample limits the interpretation of their results. Among firms that do list on the New York Stock Exchange, those with larger bid-ask spreads appear to benefit more from the decision. However, we cannot infer that stocks with higher bid-ask spreads than others traded over the counter are, or should be, most likely to list on an exchange. Evidence that can support or refute such a conclusion must be based on an ex ante sample. That is, we need a sample of all firms qualifying to list on an exchange, rather than just those firms that decide to list. It is also important to consider other variables, in addition to the bid-ask spread, that may differ across actual and potential listing firms.

This paper constructs an ex ante sample by using publicly available data to identify NASDAQ firms eligible for a New York Stock Exchange listing. To the best of our knowledge, ours is the first extensive examination of the differences between qualified nonlisting firms and firms that do list.

B. Market Microstructure, Stock Characteristics, and Spreads

A major component of the cost of trading common stocks is the bid-ask spread. Several differences exist between the trading arrangements, or microstructure, of the New York Stock Exchange and NASDAQ that may contribute to differences in bid-ask spread for a given stock depending on where it is traded.

At first glance, the NYSE is a monopoly dealer market and the NASDAQ is a competitive dealer market. On the NASDAQ system, any dealer meeting net capital requirements may make a market in any number of NASDAQ securities. Making a market implies that the dealer stands ready to sell securities from and buy securities for the dealer's own portfolio. Multiple firms may act as market makers in a given security at any one time. On the New York Stock Exchange, only the exchange-appointed specialist may make a market in a stock; each stock has only one specialist. Ho and Stoll |24~ show that competition among market makers reduces equilibrium market spreads. Thus, one might expect spreads to be lower on NASDAQ than on the NYSE. However, the NYSE specialist faces competition from members of the public. Brokers representing the public can bypass the specialist and trade directly with each other at prices between the specialist bid and ask quotes (see Bernstein |6~ and Mann and Seijas |27~). Also, members of the public may place limit orders that supersede the specialist's quotes, thereby tightening the spread (see Ho and Stoll |24~ and Stoll |38~).

The quotes of all NASDAQ market makers are available to all subscribing security brokers. Until January 1989, however, NASDAQ had no provision for a public limit order to be executed other than by the dealer with whom it was placed, or for limit order quotes to be available to other traders.(3) Overall, the degree of effective competition in market making may be as great on the NYSE as it is on NASDAQ.

Stock exchange regulations require specialists to stabilize prices more than unregulated dealers would. A specialist meets this requirement by providing quotes and participating in transactions in order to prevent trade-to-trade price changes that, by the exchange's definition, would be unacceptably large. The specialist may have to buy when prices are falling and sell when prices are rising. These obligatory transactions may cause undesired fluctuations in the specialist's inventory. The NASDAQ market maker has minimal obligations. A dealer making a market in a particular stock must continuously post both bid and ask quotes, and honor those quotes for a limited number of shares. A dealer may begin, or cease, making a market in a stock at the start of any trading day. (There is a two-day waiting period after withdrawing from a stock before the dealer can resume making a market in the stock |31~.) Price stabilization by NYSE specialists, and the monitoring by the exchange of specialists and trading, may induce firms to list.

Recent empirical studies offer conflicting evidence of the ability of the exchange and NASDAQ trading mechanisms to minimize bid-ask spreads and other trading costs. Hasbrouck and Schwartz |22~ find that average quoted bid-ask spreads of stocks traded on the NYSE are smaller than spreads of NASDAQ stocks. Reinganum |33~ reports that for portfolios of small firms that are at least as large as the smallest NYSE firms, the returns of NASDAQ portfolios contain a smaller liquidity premium than the returns of NYSE portfolios. The results suggest that the NASDAQ seems to provide more liquidity to small stocks than does the NYSE. Reinganum finds little difference in liquidity for large firm portfolios.

The explicit costs of listing on the NYSE are comparatively small. Sanger and McConnell |35~ estimate the total present value of all initial and continuing annual listing fees to be 0.29% of total equity value for an average size firm. However, there may be implicit costs of listing as well, such as a prohibition against dual classes of stock with different voting rights and increased filing requirements.

Some NASDAQ firms may be able to benefit more than others from listing on the NYSE. Small, thinly traded NASDAQ stocks with relatively few market makers should be most likely to benefit from the increased public interaction and specialist stabilization on the New York Stock Exchange. Large, actively traded stocks with relatively many market makers are likely to have enough competitive quotes on NASDAQ to minimize spreads.(4) On the other hand, Reinganum |33~ suggests that small stocks may receive little benefit from listing because specialists have incentives to concentrate their attention on the stocks that generate the most trading volume. The characteristics of stocks that list on the NYSE thus is ultimately an empirical question.

C. Empirical Implications of Market Microstructure Theory

If the benefit of listing is greatest for smaller or less actively traded issues, then we should observe that listing firms are smaller and less actively traded in the pre-listing period than qualified nonlisting firms. We use two measures of size, the market value of equity and the book value of assets. To measure trading activity, we use trading volume, the size of the average trade, the number of shares outstanding, and the number of shareholders.(5) Because competition among market makers reduces equilibrium bid-ask spreads (see Ho and Stoll |24~), we also test the hypothesis that listing firms have fewer market makers than nonlisting firms.

Microstructure theory does not predict that bid-ask spreads necessarily will differ, on average, between listing and nonlisting firms. Rather, it simply predicts that the firms that list on the exchange will be those whose characteristics offer the greatest probability that they can reduce their spreads by doing so. Amihud and Mendelson |1~ show that a given change in the spread reduces the cost of capital more for firms with already low spreads than for firms with high spreads. Thus, firms with high spreads may not be any more likely to seek a listing than firms with low spreads. On the other hand, Amihud and Mendelson |2~ suggest that a given cash investment in a spread-reducing project may produce a greater reduction in the cost of capital when the spread is high. Thus, an NYSE listing, at a given cost, may have a greater net present value for firms with higher spreads. Empirically, Grammatikos and Papaioannou |21~ find that there is a positive relationship between the stock market reaction to listing and the bid-ask spread in NASDAQ trading. We test whether there is a difference in spreads between listing and nonlisting firms.

D. Information Leakage Versus Performance

Sanger and McConnell |35~ report that NASDAQ firms listing on the New York Stock Exchange experience positive abnormal stock returns during the year before listing announcements. The authors propose two alternative, though not mutually exclusive, explanations. First, firms may decide to list following a period of especially strong performance. We refer to this as the performance hypothesis. Second, the positive abnormal returns may result from trading in advance of the public announcement by investors who have inside information that the firm plans to apply for a New York Stock Exchange listing. The distinction is important because if the large pre-listing stock returns are driven by leakage of information, a proper estimate of the net benefits of a listing must incorporate the returns.

We test the performance hypothesis by comparing the pre-tax profit growth rates of listing firms just prior to listing to those of nonlisting control firms. If the performance hypothesis is correct, the growth rate of earnings should be greater for the listing firms than for the nonlisting control firms.

II. Data Sources and Sample Selection Procedures

To identify NASDAQ firms that listed on the New York Stock Exchange, we examine the delisting data structure of every stock in the 1990 edition of the Center for Research in Security Prices (CRSP) NASDAQ file. We find 488 companies that list on the New York Stock Exchange. Our final sample consists of 277 nonfinancial firms that have data on the Standard & Poor's COMPUSTAT files.

We also generate a pool of control firms that are eligible, as of the end of a fiscal year, to list on the New York Stock Exchange, but do not do so. Exhibit 1 displays the minimum requirements for listing a common stock during the period covered by this study. We compare data for all nonfinancial NASDAQ firms on the COMPUSTAT files with the New York Stock Exchange minimum listing requirements. We search both the active and research COMPUSTAT files. Searching only the active file would result in a control sample consisting only of firms that survived as independent, publicly traded companies through 1990. The use of survivors could bias the control sample toward larger, more active firms. This problem is averted by using the research file, which contains companies that cease to exist as independent firms or cease to be publicly traded. The result is a sample of firms that apparently qualify to list on the New York Stock Exchange during at least one year.

Comparing listing firms only with those that are known never to have listed on an exchange could introduce an ex post selection bias into the tests. Therefore, we also include in the control sample any firms that eventually list on an exchange. Firms that list on the American Stock Exchange are included because they have not made their listing choices at the time they are in the control sample. Excluding these firms would require the use of ex post information. No firm that moved from the NASDAQ to regional exchanges ever qualified before that time for a New York Stock Exchange listing. Listing firms serve as part of the control sample for the second and earlier fiscal years prior TABULAR DATA OMITTED to their year of listing, provided they meet the New York Stock Exchange listing requirements at the time. For example, a firm listing in 1985 that was eligible since 1982 would serve as a control firm for 1982 and 1983.

Our tests compare NASDAQ firms that list on the New York Stock Exchange against others that, on the basis of publicly available data, appear to qualify for listing but do not list. To control for the effects of unobservable confounding variables, it is important to match listing firms with nonlisting firms that are as similar as possible. We match firms with nonlisting controls on the basis of two-digit SIC code industry classification, and compare listing firm data with control firm data for the same year.(6) Because listing eligibility is based in large part on annual financial statement data, we use the fiscal year rather than the calendar year. Financial and market data for the fiscal year of listing (fiscal year 0) may be distorted by changes that occur after the listing date. We therefore restrict market data for fiscal year 0 to that portion of the year ending two months before listing. Only annual financial statement data are available on the research file, so we conduct our comparisons on data from both fiscal year 0 and the preceding fiscal year (fiscal year -1). Twenty listing firms do not meet New York Stock Exchange criteria in fiscal year -1 (in other words, they listed during the first fiscal year they were eligible). We exclude the 20 firms from the tests of fiscal year -1.

Exhibit 2 lists the number of firms with COMPUSTAT and CRSP data in each year. No eligible, nonlisting firms are identified for 1990. The reason is that to be a control firm for a given fiscal year, a firm must not list in the following year. We do not know whether firms eligible to list in 1990 may have listed in 1991; therefore, we cannot properly classify them as control firms. However, firms that list in 1990 are included in the tests of fiscal year -1 (1989 in the case of these firms). On average, there are 313 nonlisting firms in the control sample per year. Exhibit 2 shows only ten firms listing on the American Stock Exchange. A total of 633 NASDAQ firms actually list on the American Stock Exchange between 1972 and 1990. Of the 633, only the ten counted in the exhibit meet the New York Stock Exchange requirements at the time they list on the American Stock Exchange.
Exibit 2. Number of NASDAQ Firms Eligible for Listing and
Actually in the Period 1973-1990 by Year
 NASDAQ Firms Listing NYSE-Eligible
 Firms Eligible on Firms
 for NYSE NYSE Listing on
 Listing at During ASE
 Fiscal Year the During
Fiscal Year End(a) Year the Year
1973 84 11 0
1974 100 6 0
1975 245 7 0
1976 286 10 1
1977 312 18 0
1978 327 12 2
1979 334 15 1
1980 341 21 1
1981 344 19 1
1982 343 18 0
1983 360 11 0
1984 375 12 0
1985 389 13 1
1986 354 18 0
1987 357 24 0
1988 376 26 2
1989 375 18 1
1990 -- 18 0
Total 277 10
a Listing eligiblity is determined from published NYSE criteria
and annual firm financial data from COMPUSTAT. Only firms with
both COMPUSTAT and CRSP NASDAQ file data are included in this

III. Empirical Results

A. Difference Tests

We used a paired-difference t-test to compare listing firms to the matched control portfolios. Since paired difference tests are potentially sensitive to skewness, we also use the nonparametric Wilcoxon-Mann-Whitney test, which makes less stringent distributional assumptions (see Sprent |36~).

1. Tests of Microstructure-Related Characteristics

Listing firms are smaller than nonlisting firms in the same industry, on average. In Exhibit 3, Panel A reports that the mean book value of assets of all listing firms at the end of the fiscal year of listing is $714.6 million, while the mean book value of all control firms is $391.8 million. However, the medians for the listing and control firms are $130.2 million and $213.4 million, respectively. Also, the group means do not reflect the pairing of listing firms with control firms by year and industry. The average paired difference (listing minus control) is significantly negative at 0.1%. The Wilcoxon-Mann-Whitney test also is consistent with a smaller mean book value for the listing sample appears to depend on a few very large firms. The mean market value of equity of listing firms is $437.9 million while that of control firms is $366.6 million. Again, the ranking is reversed for the group medians, and the paired-t and Wilcoxon-Mann-Whitney statistics have the opposite sign to the difference of group means. The tests show the listing firms to have a smaller market value than the control firms; the difference is significant at the five percent level. The pre-listing year comparisons in Panel B are similar.

The ratio of market value to book value of equity can serve as a proxy for how much of firms' market value is due to growth opportunities. There are no statistically significant differences in the ratio between listing and control firms.

The number of shareholders and the number of shares outstanding both represent potential trading volume. Exhibit 3 shows that on a paired basis, the mean of each of the variables is smaller in the listing than in the control sample. Panel A indicates that listing firms have a mean of 6,400 shareholders at the end of the fiscal year of listing, while control firms have 6,800 shareholders, on average. The paired difference is significantly negative at 0.1%. At the same time, listing firms have 18.3 million shares outstanding while control firms have an average of 14.4 million shares outstanding. The average paired difference (using the log of shares) is negative but statistically insignificant. Panel B shows that the listing and control firms have, on average, 5,400 and 6,600 shareholders, respectively, at the end of the pre-listing year. The paired difference is significant at 0.1%. On a paired basis, the listing firms also have significantly fewer shares outstanding during the year before they list than control firms.

Listing firms tend to experience greater actual trading volume in the NASDAQ market than control firms.(7) From the beginning of fiscal year 0 until six weeks prior to listing, the mean volume of listing firms' stocks traded is 84,365 shares (shown on Panel A). The corresponding figure for controls is 68,889. The paired comparison has the same sign as the difference of group means; the log of average volume of listing firms is significantly greater at the 0.1% level. Over the full fiscal year -1, Panel B shows that listing firms have a mean volume of 81,724 shares and control firms have a mean volume of 65,983 shares. The paired difference is significant at one percent.(8) Relative volume, defined as shares traded divided by shares outstanding, is also greater for listing firms. The greater volume may mean that listing firm stocks are, in one respect, more liquid than control firm stocks. Another possibility is that there is less information being generated about the firm, resulting in greater disagreement among traders and thus, greater volume (see Karpoff |25~ and Kim and Verrecchia |26~).

Another measure of trading activity is the dollar value of the average trade. The listing firms have smaller mean and especially median average trade sizes than the nonlisting firms. Thus, although listing firms have higher volume, their shares tend to be traded in smaller dollar amounts. This suggests that there may be less trading by institutional TABULAR DATA OMITTED investors, which in turn may imply lower potential liquidity. Alternatively, the smaller trade size may reflect a smaller share price.(9)

Because prices are set in increments of eighths of a dollar, relative bid-ask spreads may reflect, in part, the price level of a firm's shares (Blume and Stambaugh |8~). Panels A and B of Exhibit 3 report that the average share price of listing firms in both years is significantly smaller than that of the control firms. Listing firms also tend to have fewer market makers than nonlisting firms.

Listing firms have somewhat smaller average bid-ask spreads than their matched control portfolios. Panel A shows that, in the year of listing (up to six weeks before the listing date), listing firms have a mean 2.5% (median 2.1%) bid-ask spread and control firms have a mean 2.7% (median 2.3%) spread. The paired difference is marginally significant (p |approximately~ 0.06) by the t-test and the difference between the groups is significant by the Wilcoxon-Mann-Whitney test. In the year before the year of listing, the average listing firm has a 2.7% spread and the average control firm has a 2.8% spread; both group medians are 2.4%. There is no significant difference in the average bid-ask spread between listing and control firms in the year before listing. Although listing firms do not have larger spreads than control firms, reduction of the bid-ask spread still could be a motive for listing. However, listing firms are not simply those with larger spreads than other eligible firms in the same industry.

Listing firms, on average, have a greater variance of return and a greater beta than control firms. The difference in return variance is statistically significant only by the paired t-test in both the listing and pre-listing years, while the difference in beta is also significant according to the nonparametric test. Bhandari, Grammatikos, Makhija, and Papaioannou |7~, and Clarkson and Thompson |11~ report that the betas of common stocks newly listed on the NYSE decline during the first year of trading on the exchange. Clarkson and Thompson argue that the reduction in beta is the result of decreasing estimation risk as more information is generated about listed firms. The tendency for listing firms in our sample to have greater betas is consistent with this argument and suggests that the reduction of estimation risk is a motive for listing.

Firms may wait until they have surpassed the minimum listing requirements for a number of years before applying for a listing. One reason to delay applying may be a belief that the New York Stock Exchange actually applies more stringent standards in evaluating a listing application than the published minima. Alternatively, managers may not be confident that they will be able to meet the NYSE's requirements for remaining listed until they have surpassed the initial listing requirements for several years. If such a delay is common, then it may not be appropriate to include in the control sample firms that have qualified to list only for a single year. We test whether listing firms have met the minimum criteria for a greater number of consecutive years than nonlisting control firms. Panel A of Exhibit 3 shows that this is not the case. Listing firms, as of the end of fiscal year 0, have met the published minimum listing requirements for an average of 1.6 years, while control firms have met the requirements for an average of 4.3 years. The difference between listing and control firms is highly statistically significant according to the paired-t and nonparametric tests.

Moving from the NASDAQ to the NYSE may be more costly if firms must restructure their ownership and voting structure to gain approval. In both the listing year and the year before, roughly one percent of firms that move to the exchange have two or more classes of common stock traded. Ten to 11 percent of control firms have more than one class of common. Different classes of common stock often have unequal voting rights, a practice discouraged or prohibited by the NYSE throughout the sample period. Thus, a firm with unequal voting classes of common would probably have to recapitalize with a single voting class before being considered for a listing. The difference between the listing and control samples means that whether a NASDAQ firm moves to the exchange appears to depend on the costs of listing.

Firms may have less incentive to move to the NYSE if they are in the NASDAQ/NMS system, which provides centralized trade reporting like the exchange. However, there is no statistically significant difference in the proportion of listing and control firms that are in NASDAQ/NMS.

In summary, NASDAQ companies that choose to list their common stock on the New York Stock Exchange are smaller, have fewer shareholders, fewer shares outstanding, smaller average trades, lower prices per share, fewer market makers and greater betas than the control firms. The control firms are firms in the same industry that qualify for the New York Stock Exchange in the same year but choose -- at least for the succeeding two years -- to remain with NASDAQ. At the same time, listing and nonlisting control firms' stock quotes on NASDAQ exhibit very similar bid-ask spreads; spreads of listing firms are, if anything, smaller. Listing firms also have greater trading volume than control firms. The results suggest that listing choices depend on stock liquidity characteristics and thus, the potential benefits of a listing.

2. Evidence on the Performance Hypothesis

Panel A of Exhibit 3 reports that listing firms' pre-tax profit grew 29.1% during the year of listing, while control firms' pre-tax profit grew 16.4% over the same period. The difference is significant according to both the parametric and Wilcoxon-Mann-Whitney tests. Panel B shows that the growth rates in the pre-listing year average 33.0% for listing firms and 18.7% for nonlisting firms. The difference is statistically significant at least at five percent in the pre-listing year by the parametric and nonparametric tests. The results support the performance hypothesis. In addition, the fact that listing firms have qualified for the New York Stock Exchange for a shorter time than the control firms is consistent with the performance hypothesis. Listing firms tend to be those that have recently achieved the minimum listing criteria through a burst of growth in earnings and market value. If the positive abnormal stock returns prior to the public announcement of listing that Sanger and McConnell |35~ report were due solely to trading on private information about the future New York Stock Exchange listing, we would not observe the superior growth in pre-tax earnings.(10) However, part of the pre-listing stock performance still could be due to advance trading on nonpublic information about listing.

B. Logistic Regression Tests

In this section, we use logistic regression to further assess the contributions of liquidity-related trading characteristics and performance to the listing decision.

1. Expected Bid-Ask Spreads

As Grammatikos and Papaioannou |21~ argue, reduction of the bid-ask spread is a likely motive for exchange listing. It is desirable to determine whether the bid-ask spread has any bearing on the decision to list apart from the firm's other relevant market microstructure characteristics. To improve the chance of detecting such an effect, we use the unexpected spread in place of the raw spread in the logistic regressions. The expected spread is calculated on the basis of OLS parameter estimates from a broader sample of NASDAQ stocks. We sort all the stocks on the NASDAQ CRSP file that have at least 260 trading days of data into quintiles depending on the number of days available. We select a random sample of 2,000 stocks with replacement; a stock in the quintile with the most days has five times the chance of selection as a stock in the quintile with the fewest days. For each stock, we randomly select an analysis date within the stock's period of trading on NASDAQ.

The dependent variable is the average closing bid-ask spread over a 200-trading-day period starting with the randomly selected analysis date. The independent variables for prediction of the spread are the logarithms of mean trading volume in shares, mean bid price per share, mean equity market value (the bid price times the number of shares outstanding), return variance and mean number of market makers, and an indicator variable for whether the stock is listed on the NASDAQ/NMS system. We calculate the means and return variance from daily observations over the same 200-day period used for the spread. For NMS stocks, the bid prices are not available on the main NASDAQ CRSP file; we obtain them from the CRSP/NMS supplemental file. Of the 2,000 stocks, 1,169 have data on all the variables.

Exhibit 4 shows the results of the OLS estimation. The volume, bid price, and the number of market makers are significantly negatively related to the spread. The market value, return variance, and NASDAQ/NMS indicator are not significantly related to the spread, which could be due to correlation with the other variables. Jointly, the regressors explain 63.6% of the variability in spread across the sample. The results are consistent with other studies of spread determination (for example, see Chiang and Venkatesh |10~).
Exhibit 4. Estimates From Ordinary Least Squares Regression of
the Relative Bid-Ask Spread on Natural Logarithms of Market
Value, Volume, Price, Return Variance, and Number of Market
Makers, and an Indicator Variable for NMS Listing, for 1,169
Randomly Selected NASDAQ Stocks in the Period 1973-1990
Variable Coefficient t-statistic
Intercept 0.376 11.46
Log of market value 0.001 0.25
Log of volume -0.017 -9.37
Log of bid price -0.060 -20.20
Log of return variance 0.003 0.79
Log of number of market makers -0.023 -3.61
NASDAQ/NMS indicator -0.002 -0.32
F for regression: 340.75
Adjusted |R.sup.2~: 63.6%

2. Simple Logistic Regressions

To examine the impact of trading characteristics and performance on exchange listing, we run several regressions. Exhibit 5 reports simple logistic regressions on fourteen explanatory variables. Even though many of the variables are correlated with each other, we report these simple logistic regressions in order to further describe TABULAR DATA OMITTED the difference between listing and nonlisting firms. In the logistic regressions, we pool all the listing and qualified nonlisting firms for which we have data. The unit of observation is a single firm, rather than a matched pair of a listing firm and a portfolio of control firms as in Exhibit 3. This increases the sample size for some variables, because there are listing firms for which we could not find controls and potential control firms without a matching listing firm.

The simple logistic regressions show that firms are more likely to move from NASDAQ to the NYSE if they have larger unexpected spreads, smaller prices per share, higher volume, smaller market value of equity, larger return variances or betas, only one class of common stock or larger growth rates of pre-tax profit. The probability of listing also increases with smaller average trade values and fewer market makers, but these effects are only marginally statistically significant (0.10 |is greater than~ p |is greater than~ 0.05). The most statistically significant influence on listing in the simple logistic regressions is the number of years qualified; the longer a firm is qualified, the less likely it is to move to the exchange. This is contrary to our expectation that firms find it easier to move after a longer period of qualification. A possible explanation for the result is that the period of qualification is a proxy for information availability. (Barry and Brown |4~ use the period of listing on the NYSE as a proxy for information.) Firms qualify for listing by growing in size, earnings, and public ownership. These factors may also encourage the generation of outside estimates of firm cash flows and value. If the reduction of estimation risk is a motive for listing, as Clarkson and Thompson |11~ suggest, the incentive to list would be weaker the longer a firm is qualified.

3. Multiple Logistic Regressions

Exhibit 6 reports multiple logistic regressions. The twelve variables in Regression (1) collectively explain 29.2% of the variation in the listing decision. (There is a penalty for additional explanatory variables built into the adjusted |R.sup.2~ calculation, so the maximum possible is less than 100%.) Somers' |D.sub.yx~, an index of rank correlation between the predicted and actual listing outcomes |20~, is 76.6% (of a possible 100%). Thus, the regression has, at best, moderate power to predict which firms actually list, but fairly good power to explain which firms are more likely than others to list.

Regressions (2) through (5) report multiple logistic regressions on subsets of explanatory variables. The unexpected spread, bid price per share, shares outstanding, years qualified for listing and dual-class common stock indicator and number of market makers contribute significantly to explaining the listing decision. The return variance, beta and NASDAQ/NMS indicator are marginally or insignificantly related to the listing decision in the presence of other explanatory variables. The insignificance of the return variance and beta in the multiple regressions probably is due to correlation with other variables.

Coupled with the simple logistic regression results, the multiple logistic regression evidence supports the view that firms moving to the exchange have lower costs and greater benefits of moving than other firms that could move. Firms are more likely to move when their stocks have greater bid-ask spreads on NASDAQ, after adjusting for their market microstructure characteristics, than other qualified firms. Thus, NASDAQ does not appear to provide as liquid a market for these stocks as would be expected. The relative lack of liquidity on the NASDAQ implies that the potential benefit of listing on the exchange is greater than it is for other qualified firms. Also, consistent with the liquidity argument, firms are more likely to list the lower their share price, number of shares outstanding, and number of market makers. The probability of listing also is inversely related to the number of years qualified for listing. As we discuss in Section III.B.2 above, this suggests that the reduction of estimation risk motivates the listing decision. Finally, firms are less likely to list when they have dual classes of common stock, a feature that may be costly to eliminate.

IV. Summary

This paper explores the characteristics of NASDAQ firms that list their stocks on the New York Stock Exchange as compared to control firms in the same industry that could have listed on the exchange in the same year but did not. We find that, on average, listing firms are smaller, have fewer shareholders, smaller share prices, and fewer market makers, are traded in smaller dollar amounts and have larger return variances and betas than the control firms. These factors suggest that listing firms are less liquid than control firms. Consistent with the predictions of Grammatikos and Papaioannou |21~, the unexpected bid-ask spread, adjusted for the relationship in the broader NASDAQ market between spread and market microstructure variables, is positively related to the probability of listing in logistic regressions. Although listing firm spreads are no larger than those of nonlisting firms in the same industry, firms are more likely to list when their spreads are larger than other market microstructure characteristics would predict. The results support the idea that firms eligible to list on the New York Stock Exchange actually TABULAR DATA OMITTED list when their liquidity in the NASDAQ market is lower than would be expected on the basis of firm characteristics.

Listing firms tend to have met the published listing requirements for fewer years than control firms. The longer the firm is qualified, the less likely it is to move to the NYSE. We argue that the period of qualification can be a proxy for information generated about the firm. The inverse relationship between the period of qualification and the decision to move to the exchange supports the reduction of estimation risk as a motive for listing.

Listing firms experience exceptionally strong average growth in pre-tax accounting profit, relative to the control firms, in the fiscal year of and fiscal year preceding the switch from NASDAQ to the New York Stock Exchange. The exceptional growth supports the hypothesis that the pre-listing announcement positive abnormal stock returns that Sanger and McConnell |35~ report reflect a tendency to list on the New York Stock Exchange after a period of abnormally good firm performance. Thus, trading on inside information about the decision to apply for listing probably is not the sole cause of the large returns. However, the growth rate of pretax profit has no incremental predictive power after the other variables are included in a multiple logistic regression model.

In summary, the evidence supports the idea that firms list on the New York Stock Exchange in search of a more liquid market for their stock. However, the potential liquidity-enhancing benefits, and the costs, of listing are unequally distributed across NYSE-qualified firms.

1 Ule |40~ and Merjos |28~, |29~, |30~ try to determine whether listing affects the value of the stock, but do not identify the source of the impact. Van Horne |41~, Reints and Vandenberg |34~, Ying, Lewellen, Schlarbaum, and Lase |42~, and Fabozzi |18~ find no value to exchange listing. Boardman, Dark, and Lease |9~ find no stock price effects of corporate bond listings. Ferri, Moore, and Schirm |19~ report that listing equity warrants increases their value, especially for smaller issues. Dhaliwal |15~ finds that exchange-listed stocks, in 1971, have lower systematic risk than over-the-counter stocks. Phillips and Zecher |32~ are unable to replicate Dhaliwal's result using more recent data.

2 See Baker and Meeks |3~ for a more detailed and extensive review of empirical research on exchange listing.

3 In January 1989, NASDAQ instituted a limit order file in its small order execution system |37~. The SOES provides for automatic execution of market orders, and now limit orders, to buy or sell up to 1,000 shares of a security.

4 Conroy and Winkler |13~ model the role of the specialist's limit order book in reducing bid-ask spreads of thinly traded securities. Cohen, Maier, Schwartz, and Whitcomb |12~ and Ho and Macris |23~ also analyze the relationship between trading thinness and the role of the specialist.

5 Demsetz |14~ and Benston and Hagerman |5~ use the number of shareholders as a proxy for long-run average volume.

6 We also conducted our tests on a control sample formed on the basis of three-digit SIC codes. The results were qualitatively similar to the results obtained with the two-digit sample. When we attempted to select controls on the basis of four-digit SIC codes, we were unable to find enough firms with matching codes to permit meaningful tests. Also, the fourth digit may not convey much information. For example, it is not obvious whether SIC 3494, "valves and pipe fittings" should be considered a different industry from SIC 3498, "fabricated pipe and fittings."

7 The reduced sample size for the volume statistics is due to the limited availability of volume data on the CRSP NASDAQ and COMPUSTAT files.

8 In a previous draft, we reported results for firms listing during 1973-1987. We found that listing firms had lower volume on both a paired and a group basis than control firms. There are two major differences in the samples in the present version. First, the use of the latest available CRSP file enables us to approximately triple the number of listing firms with volume data. Second, the control sample in the previous version came from the limited-subscription COMPUSTAT file, which contained only selected NASDAQ firms. In the present version, we use the full-coverage COMPUSTAT file, which includes all NASDAQ firms.

9 Easley and O'Hara |16~ model an equilibrium in which larger trade sizes are associated with informed traders and larger bid-ask spreads. However, their model refers to a sequence of trades for a given stock. It does not try to explain how the long-run average trade size differs across stocks.

10 A referee pointed out that the superior earnings growth would have to be unanticipated to cause abnormal stock returns. It is unlikely that all of the superior earnings growth in the fiscal year of listing -- by the end of which the firm was trading on the NYSE -- was anticipated with certainty so far ahead of the listing that it would not have contributed to the pre-listing run-up. Even if consensus estimates of earnings predicted the growth exactly in every case, the resolution of uncertainty about actual earnings should still have resulted in positive stock returns.


1. Y. Amihud and H. Mendelson, "Asset Pricing and The Bid-Ask Spread," Journal of Financial Economics (December 1986), pp. 223-250.

2. Y. Amihud and H. Mendelson, "Liquidity and Asset Prices: Financial Management Implications," Financial Management (Spring 1988), pp. 5-15.

3. H.K. Baker and S.E. Meeks, "Research on Exchange Listings and Delistings: A Review and Synthesis," Financial Practice and Education (Spring 1991), pp. 57-71.

4. C.B. Barry and S.J. Brown, "Differential Information and the Small Firm Effect," Journal of Financial Economics (June 1984), pp. 283-294.

5. G.J. Benston and R.L. Hagerman, "Determinants of Bid-Asked Spreads in the Over-the-Counter Market," Journal of Financial Economics (December 1974), pp. 353-364.

6. P.L. Bernstein, "Liquidity, Stock Markets, and Market Makers," Financial Management (Summer 1987), pp. 54-62.

7. A. Bhandari, T. Grammatikos, A.K. Makhija, and G. Papaioannou, "Risk and Return on Newly Listed Stocks: The Post-Listing Experience," Journal of Financial Research (Summer 1989), pp. 93-102.

8. M. Blume and R. Stambaugh, "Biases in Computed Returns: An Application to the Size Effect," Journal of Financial Economics (November 1983), pp. 387-404.

9. C.M. Boardman, F.H. Dark, and R.C. Lease, "On the Listing of Corporate Debt: A Note," Journal of Financial and Quantitative Analysis (March 1986), pp. 107-114.

10. R. Chiang and P.C. Venkatesh, "Insider Holdings and Perceptions of Information Asymmetry: A Note," Journal of Finance (September 1988), pp. 1041-1048.

11. P.M. Clarkson and R. Thompson, "Empirical Estimates of Beta When Investors Face Estimation Risk," Journal of Finance (June 1990), pp. 431-454.

12. K.J. Cohen, S.F. Maier, R.A. Schwartz, and D.K. Whitcomb, "The Returns Generation Process, Returns Variance, and the Effect of Thinness in Securities Markets," Journal of Finance (March 1978), pp. 149-167.

13. R.M. Conroy and R.L. Winkler, "Informational Differences Between Limit and Market Orders for a Market Maker," Journal of Financial and Quantitative Analysis (December 1981), pp. 703-724.

14. H. Demsetz, "The Cost of Transacting," Quarterly Journal of Economics (February 1968), pp. 33-53.

15. D.S. Dhaliwal, "Exchange-Listing Effects on a Firm's Cost of Equity Capital," Journal of Business Research (June 1983), pp. 139-151.

16. D. Easley and M. O'Hara, "Price, Trade Size, and Information in Securities Markets," Journal of Financial Economics (September 1987), pp. 69-90.

17. R.B. Edelman and H.K. Baker, "Liquidity and Stock Exchange Listing," Financial Review (May 1990), pp. 231-249.

18. F.J. Fabozzi, "Does Listing on the Amex Increase the Value of Equity?," Financial Management (Spring 1981), pp. 43-50.

19. M.G. Ferri, S.B. Moore, and D.C. Schirm, "The Listing, Size, and Value of Equity Warrants," Financial Review (February 1989), pp. 135-146.

20. L.A. Goodman and W.H. Kruskal, Measures of Association for Cross-Classifications, New York, Springer-Verlag, 1979.

21. T. Grammatikos and G. Papaioannou, "Market Reaction to NYSE Listings: Tests of the Marketability Gains Hypothesis," Journal of Financial Research (Fall 1986), pp. 215-227.

22. J. Hasbrouck and R.A. Schwartz, "Liquidity and Execution Costs in Equity Markets," Journal of Portfolio Management (Summer 1988), pp. 10-17.

23. T.S.Y. Ho and R.G. Macris, "Dealer Market Structure and Performance," in Market Making and the Changing Structure of the Securities Industry, Y. Amihud, T.S.Y. Ho and R.A. Schwartz (eds.), Lexington, MA, Lexington Books, 1985.

24. T.S.Y. Ho and H.R. Stoll, "The Dynamics of Dealer Markets Under Competition," Journal of Finance (September 1983), pp. 1053-1074.

25. J.M. Karpoff, "The Relation Between Price Changes and Trading Volume: A Survey," Journal of Financial and Quantitative Analysis (March 1987), pp. 109-126.

26. O. Kim and R.E. Verrecchia, "Trading Volume and Price Reactions to Public Announcements," Journal of Accounting Research (Autumn 1991), pp. 302-321.

27. S.V. Mann and R.W. Seijas, "Bid-Ask Spreads, NYSE Specialists, and NASD Dealers," Journal of Portfolio Management (Fall 1991), pp. 54-58.

28. A. Merjos, "Going on the Big Board: Stocks Act Better Before Listing Than Right Afterward," Barron's (January 29, 1962), p. 5ff.

29. A. Merjos, "Like Money in the Bank: Big Board Listing, the Record Suggests, is a Valuable Asset," Barron's (July 8, 1963), p. 9ff.

30. A. Merjos, "Up on the Curb," Barron's (May 1, 1967), p. 9ff.

31. Reference Guide for NASDAQ Companies, Washington, D.C., National Association of Securities Dealers, Inc., 1987.

32. S.M. Phillips and J.R. Zecher, "Exchange Listing and the Cost of Equity Capital," Capital Market Working Papers, No. 8. Washington, D.C., U.S. Securities and Exchange Commission, 1982.

33. M.R. Reinganum, "Market Microstructure and Asset Pricing: An Empirical Investigation of NYSE and NASDAQ Securities," Journal of Financial Economics (November/December 1990), pp. 127-147.

34. W. Reints and P. Vandenberg, "The Impact of Changes in Trading Location on a Security's Systematic Risk," Journal of Financial and Quantitative Analysis (December 1975), pp. 881-890.

35. G.C. Sanger and J.J. McConnell, "Stock Exchange Listings, Firm Value, and Security Market Efficiency: The Impact of NASDAQ," Journal of Financial and Quantitative Analysis (March 1986), pp. 1-25.

36. P. Sprent, Applied Nonparametric Statistical Methods, London, Chapman and Hall, 1989.

37. S. Steptoe, "Stocks Advance on Vigorous Volume as Several Technology Issues Rebound" (OTC Focus column), Wall Street Journal (January 19, 1989).

38. H.R. Stoll, "The New York Stock Exchange Specialist System," Working Paper No. 83-129, Owen Graduate School of Management, Vanderbilt University, 1984.

39. H.R. Stoll, "Alternative Views of Market Making," in Market Making and the Changing Structure of the Securities Industry, Y. Amihud, T.S.Y. Ho and R.A. Schwartz (eds.), Lexington, MA, Lexington Books, 1985.

40. M.G. Ule, "Price Movements of Newly-Listed Common Stocks," Journal of Business (October 1937), pp. 346-369.

41. J. Van Horne, "New Listings and Their Price Behavior," Journal of Finance (September 1970), pp. 783-794.

42. L.K.W. Ying, W.G. Lewellen, G.G. Schlarbaum, and R.C. Lease, "Stock Exchange Listings and Securities Returns," Journal of Financial and Quantitative Analysis (September 1977), pp. 415-432.
COPYRIGHT 1992 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Market Microstructure and Corporate Finance Special Issue; New York Stock Exchange, National Association of Securities Dealers Automated Quotations Systems
Author:Cowan, Arnold R.; Carter, Richard B.; Dark, Frederick H.; Singh, Ajai K.
Publication:Financial Management
Date:Dec 22, 1992
Previous Article:AMEX-to-NYSE transfers, market microstructure, and shareholder wealth.
Next Article:On the management of financial guarantees.

Related Articles
Big gains for small stocks.
Defining the playing field: here's the skinny on the three major stock exchanges.
Market microstructure.
Japanese editorial excerpts -3-+.
Japanese editorial excerpts -3-.
Metal Management applies to list on NYSE.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters