Table 12. The Effect of the End of Consolidated Trading on the NYSE in the Pre- and Consolidated Period (December 26, 1924 - April 8, 1927)

1 There is a growing literature that shows that potential competition can have an important effect in reducing prices and improving consumer welfare. Whinston and Collins (1992) and Goolsbee and Syverson (2005), for example, find that airlines reduce fares on routes that face potential entry from low cost carriers. Similarly, Neal (1987) compares options that are eligible for listing on multiple exchanges with those restricted to trade on a single exchange and finds a significant reduction in trading costs for those eligible for listing on multiple exchanges even if the actual volume is primarily confined to a single exchange.

2 For background on the formation of the Consolidated Stock Exchange, see Nelson (1907), Garvy (1944), Sobel (1972), and Mulherin, Netter and Overdahl (1991). An appendix of relevant news stories is available from the authors.

3 Ott (2004) shows that odd lot trading accounted for as much as 40 percent of the business of NYSE members by 1921.

4 Although the data collected from The New York Times contains total volume for listed and unlisted securities, the volume data from the NYSE website only reports data for listed securities. As a result, our total volume data for the NYSE slightly undercounts total volume for the period between 1888 and 1910 for the Big Board, when the NYSE closed its unlisted securities department. This discrepancy does not affect our formal analyses which are conducted on individual securities where we have the actual data from both exchanges.

5 Ott (2004) shows that politics played an important role in the collapse of the Consolidated. She argues that the NYSE engaged in a public relations campaign from 1913 until 1929 and captured the New York State Attorney General’s Office which investigated the Consolidated for stock fraud and wash sales.

6 We have also estimated the model while jointly clustering standard errors by company and by time as suggested by Petersen (2007). These results are also similar to those reported in Table 3.

7 Silber (2005) reports a similar non-reporting of data on NYSE competitors by major financial newspapers in his analysis of the closure of stock markets from the end of July to December 1914 following the outbreak of World War I.

8 The robustness checks are available from the authors upon request.

9 The Commercial and Financial Chronicle reported bid-ask spreads for some stocks on the regional exchanges that did not have trading volume for that particular day. For consistency, the average bid-ask spread for the NYSE is the average for all securities and includes some stocks that did not have any trading volume so the data in this figure are not directly comparable to those in later tables. For a couple of weeks when The Commercial and Financial Chronicle did not report bid-ask spreads for the regional markets, we interpolated the average bid-ask spread using the week before and after the missing observations. The Boston Exchange accounts for over 90 percent of the observations in Figure 3. For our formal analysis, we are forced to focus on the Boston Exchange due to the lack of sufficient observations with trading volume on the other regional exchanges.

10 The 11.6 percent decline is the calculated as follows e^-.123-1 where 0.123 is the coefficient on the presence of consolidated trading variable in the regression of the natural log of bid-ask spreads on the presence of consolidated trading and control variables.

11 As a robustness check, we also estimated the following model of the determinants of bid-ask spreads that did not contain market control variables: SPREAD_it = α₀+ β₁VOL_it + β₂CLOSE_it+ β₃STDEV_i + β₄COMP_t + ε_it. In this specification, we found that the Consolidated competition was statistically significant at the one percent level with a coefficient of similar size to the model with the market control variables.

12 Neal (1987) reports a similar reduction in bid-ask spreads in the options markets from competition.

13 We also find evidence that NYSE seat prices declined 20 percent in the first six months that the Consolidated began to trade NYSE listed stocks. We are unable to formally test this hypothesis given that the NYSE did not report any trades in the seat market in the three-year period before the Consolidated began to compete head-to-head with the Big Board. We also do not have Consolidated seat prices from this period.

14 Railroads comprised 97 out of 116 firms in our sample.

15 The findings are similar to some studies of option markets where low volume contracts experience the greatest decline in spreads as the result of potential competition (Neal, 1987).

16 Given that the Consolidated traded the NYSE’s most liquid and active stocks, it is unlikely that our welfare analysis would significantly change if we dropped the assumption that bid-ask spreads for all NYSE stocks declined.

17 In an earlier version of this paper, we use the resignation of Consolidated President William Silkworth as the effective end of the Consolidated and examine the impact of that event on spreads for the securities in the Dow Jones Averages. Those result produce similar negative impacts on bid-ask spreads to those reported here suggesting that the Consolidated’s demise was more gradual than abrupt.

18 Data prior to 1926 come from the New York Times while data after December 31, 1925 come from the CRSP tapes.

19 As with the initiation of trading, we also estimated the impact of the Consolidated using the window of 76 trading days around the end of Consolidated trading. These results using daily data, available on request, produce similar results with the post-Consolidated variable being statistically significant in all specifications and with coefficient estimates being similar in magnitude to those reported for the initiation of Consolidated trading.