Methodology and Data Directory UMM :Data Elmu:jurnal:J-a:Journal of Economics and Business:Vol52.Issue4.July2000:

investment banks were negative and insignificant. They also find that risk did not decline significantly for any of the groups around the initial increase in underwriting power for commercial banks. Bhargava and Fraser 1998 use a multivariate regression model to study prior effects of four Federal Reserve Board decisions that allowed commercial banks to participate in investment banking through Section 20 subsidiaries. When the less expansive powers to underwrite mortgage securities and municipal revenue bonds were initially proposed in April 1987, banks that participated experienced positive and significant abnormal returns. When these powers were expanded in January 1989 to include corporate debt and equity, as well as the subsequent decision to double the amount allowed in September 1989, banks experienced negative abnormal returns and increases in risk. They also study the 1996 proposal by the Federal Reserve Board to increase the allowable Section 20 revenue to increase to 25 and find no wealth effects for commercial banks or investment banks. Bhargava and Fraser do not study the wealth effects of large banks expanding underwrit- ing power on smaller banks and only look at the initial proposal to increase the section 20 loophole. 3

IV. Methodology and Data

Collectively, prior research indicates that commercial bank underwritten securities per- form better, or at least no worse than, investment bank underwritten securities. Thus, the role of informed monitor for banks, as well as the possible synergies with investment banks, apparently outweighs the potential conflicts of interest. Banks act as an “inside investor” because banks not only audit a firm but invest in the firm in the form of a loan, or as an equity underwriter. Other research based on simulations of mergers between investment and commercial banks indicates investment banking is more profitable, but higher risk and a lack of diversification offset the profitability. This study is concerned with the market impact increased investment banking powers has on the banking industry, as well as groups of banks. That is, will banks benefit from the repeal of Glass–Steagall or from the widening of Section 20 loopholes? In addition, which group of commercial banks has potential competitive gains from increased invest- ment banking powers? This paper differs from other research in that the most recent approval of the increase in the loophole in 1996 is studied and includes interim announce- ments, and results are compared across groups of commercial banks. The first hypothesis to test is whether or not the banking industry overall will benefit from investment banking activity. If commercial banks expand into investment banking via takeovers of existing investment banks, the takeover premium could be high, perhaps due to hubris as in Roll 1986 or an overestimation of scale and scope economies as in Trifts and Scanlon 1987. 4 If takeover premiums are sufficiently high on average, the market could have a negative reaction to the increased investment banking takeover or merger announcements. On the other hand, if private information, diversification, synergy, economies of scale, and other benefits outweigh the takeover and merger costs, stock price reaction should be positive on average to the announcement of increased investment banking powers. The results could also be insignificant if these effects offset. 3 Bhargava and Fraser also study the wealth effect on investment banks. In this study, there are no significant effects for any of the events for investment banks, thus the results are omitted in this paper. 4 Note that many other studies find negative, insignificant, and positive CARs around bank takeover announcements. See Rhodes 1994 for a summary of bank takeover event studies. 348 K. B. Cyree As a corollary, it is hypothesized that banks have no changes in risk when entering underwriting activities. Kwast 1989 shows underwriting has higher returns and risk, and Gande, Puri, and Saunders 1999 indicate banks underwrite riskier lower rated and small debt as compared to investment banks. These prior results indicate a potential for a risk increase for commercial banks, which would be important for regulators and shareholders. However, if diversification benefits offset some of these risks, there could be insignificant or reductions in risk. Insignificant changes in risk were found by Apilado et al. 1993 around initial increased investment banking powers by commercial banks. Table 1 shows relevant announcements concerning congressional discussion of the repeal of Glass–Steagall and of the subsequent Federal Reserve increase in Section 20 affiliate activity levels. If the repeal of Glass–Steagall is positive for commercial banks, then announcements favorable to the repeal of the Act or that indicate support for an increase in the Section 20 loophole will have positive stock price reactions on average. 5 The expected signs are shown in Table 1, assuming that the ability of commercial banks to underwrite previously ineligible securities is valued by the market. The Goldman Sachs’ December 3, 1996 announcement on whether or not to buy a commercial bank has an uncertain stock price reaction. This event is included because it illustrates the com- petitive effects on commercial banks, absent any regulatory effects, of investment banks entering traditional banking activities. The announcement by Goldman Sachs’ is ambig- uous because it could mean more competition for traditional commercial banks negative or large takeover premiums for the target bank positive. If either of these announcement effects offset, the result will be an insignificant stock price reaction. The second hypothesis is that no group had a differential reaction to the increase in Section 20 activity. A differential reaction could indicate the market’s assessment of the 5 The repeal of Glass–Steagall can be viewed as an increase in the Section 20 loophole to 100. Table 1. Chronology of Announcements Concerning Increased Investment Banking Powers for Commercial Banks Date Event exp. sign Announcement Source May 9, 1995 Event 1 1 House Banking Committee approves a bill that would repeal Glass–Steagall. NY Times September 14, 1995 Event 2 1 Reps. Leach and Salomon reach a tentative agreement to remove Glass–Steagall. NY Times October 26, 1995 Event 3 2 Prospects dimmed for repeal of Glass–Steagall. NY Times April 17, 1996 Event 4 1 Speculation that the Fed will raise the “ineligible security” Section 20 underwriting activity to 25. WSJ May 3, 1996 Event 5 1 House Banking Committee Chairman Jim Leach tries to bring the repeal of Glass–Steagall to a floor vote. WSJ June 12, 1996 Event 6 2 House Chairman Leach pulls legislation and asks Fed Chairman Greenspan to enlarge the Section 20 loophole. WSJ July 28, 1996 Event 7 1 Federal Reserve proposes change to 25 of revenue for Section 20 affiliates. WSJ December 3, 1996 Event 8 ? Goldman Sachs studies whether or not to buy a commercial bank. WSJ December 20, 1996 Event 9 1 Rule change adopted to allow increase of affiliate ineligible revenue to 25 WSJ Notes: NY Times is the New York Times and WSJ is the Wall Street Journal. The Erosion of the Glass Steagall Act 349 competitive effects of the ability to further expand investment banking activities. To test this hypothesis, the publicly traded commercial banks are split into four groups: 1 Money Center banks; 2 banks that already have Section 20 affiliates but are not Money Center banks; 3 Large Regional banks not in the prior groups; and 4 Small Regional banks not in the prior groups. Money center banks, with the exception of Wells Fargo, all have Section 20 subsidiaries before the time of study. However, Money Center banks perform currency trading, derivative trading, hedging activities, and other services at a much larger scale than most regional banks with Section 20 affiliates and would likely benefit more from scope economies in these areas. Apilado, Gallo, and Lockwood 1993 find that highly significant abnormal returns for Money Center banks for the 1987 Fed decision to allow Citicorp, J. P. Morgan, and Banker’s Trust to underwrite securities. The Money Center bank group is the same as Carow and Larsen 1997 with the Bank of New York replacing Chemical Bank after the ChaseChemical merger. The Bank of New York is chosen because it is included in Salomon Brothers Money Center bank group and is often considered a Money Center bank [see Saunders 1994]. Because of the differences in services and products, along with the more national scope of Money Center banks, banks with Section 20 subsidiaries that are not Money Centers are analyzed separately. 6 Likewise, size is an important factor as indicated by Krosner and Rajan 1997 who found that prior to Glass–Steagall, banks actively engaged in underwriting through separate securities affiliates were larger than other commercial banks. The definitions of large and Small Regional banks are those banks with more or less than 10 billion in assets. Ten billion dollars is chosen because it is a relatively common demarcation point [e.g., see Carow and Larsen 1997] and is used by the United States Banker. Also, it is likely that banks below 10 billion in asset size would not benefit from increasing investment banking. 7 Money Center banks and those banks already with Section 20 subsidiaries are expected to have the highest probability of benefiting from increased investment banking powers because of prior access to the underwriting market. It is also hypothesized that Large Regional banks are most likely to benefit vis-a`-vis smaller regional banks because Large Regional banks can acquire larger investment banks and offer greater service to corporate customers. To test the effects of the increase in investment banking powers for commercial banks, stock returns for all publicly traded banks on the Center for Research in Securities Prices CRSP tape for 1995 through 1996 are used. Returns are from 30 days before the first event to 10 days after the last event. Because the events surrounding the increasing of investment banking power are spread over 19 months, there are over 425 returns. Commercial banks with missing returns are deleted. Banks with contaminating events, such as becoming a takeover target, are removed from the sample. There are 98 publicly traded banks in the final sample. The appendix lists the banks, by group, as defined above. 6 Banks with Section 20 subsidiaries are obtained from the Federal Reserve. My list differs from Gande, Puri, and Saunders 1999 because many of the foreign bank ADRs have missing returns. I also include some banks with only Section 20 Tier I authority that cannot underwrite corporate equity since these banks are likely to be able to gain equity powers more easily. Results are virtually unchanged when Tier I banks are included or not as having Section 20 subsidiaries so I present the more general case here. 7 Results for all tables that use the size categorization are qualitatively the same when the definition is changed to 5 or 15 billion with stronger results for 15 billion. The tables are available from the author upon request. 350 K. B. Cyree Because regulatory events are clustered in time, the appropriate methodology to test for abnormal performance is Seemingly Unrelated Regression SUR as shown in Schipper and Thompson 1994, Binder 1985, Saunders and Smirlock 1987, Allen and Wilhelm 1988, Millon–Cornett and Tehranian 1990, or more recently Madura and Bartunek 1994, Wagster 1996, or Bhargava and Fraser 1998. SUR allows for heteroskedas- ticity across equations and contemporaneous dependence of the disturbances. The model used is as in Saunders and Smirlock 1987 and Bhargava and Fraser 1998 to study increased investment banking powers whereas accounting for possible non-synchronous trading. The reaction to announcements affecting the increase in investment banking activity is measured through a series of multi-variable equations. Each equally weighted portfolio responds to the announcements through share price: R p 5 a p 1 b 1p R m 1 b 2p R m~t21 1 O t51 9 A pt D t 1 e p 1 where R p is the return on the equally weighted portfolio of bank stocks in portfolio p, from one to four, R m is the return on the equally weighted CRSP index, and D t is a dummy variable that equals one on the event day and the day before days 21 and 0 to account for information leakage and zero otherwise. The lagged market term is to account for non-synchronous trading as in Saunders and Smirlock 1987 and Bhargava and Fraser 1998. Binder 1985 shows that SUR is equivalent to ordinary least squares OLS in testing for abnormal returns across all firms using Equation 1, except the dependent variable is an equally weighted portfolio of all commercial banks in the sample. In this context, the equation is discussed as if it is SUR to be consistent with the other results. An industry-wide event such as increased investment banking power for commercial banks will likely change the risk characteristics of the commercial banks in anticipation of acquiring investment banks. As shown by Wall 1987 and Kwast 1989, investment banking activities are riskier than commercial banking where risk is measured as the SD of return on equity. Further, it is likely that banks with the highest probability of increasing investment bank activity will have the largest change in risk. To capture the potential increase in risk, Eq. 1 is modified to include risk-shift variables in the intercept, market return parameter, and non-synchronous trading parameter: R p 5 a p 1 a9 p D 1 1 b 1p R m 1 b 2p R m~t21 1 b9 1p D 1 R m 1 b9 2p D 1 R m~t21 1 O t51 9 A pt D t 1 e p 2 where D 1 is a dummy variable that is one after the dimming of prospects for repeal of Glass–Steagall in October 1995. 8 The other variables are the same as in Equation 1 and the prime denotes a shift in the intercept, systematic risk, and non-synchronous trading sensitivities. This equation follows Saunders and Smirlock 1987 and Bhargava and Fraser 1998. 8 A risk-shift after the speculation of the Fed Event 4 was also evaluated with similar results, thus the results are not presented here. The Erosion of the Glass Steagall Act 351 Abnormal share price reactions or risk shifts for the four portfolios individually around announcements on event period t is tested under the null hypothesis of no abnormal returns or risk shifts. Also, the abnormal returns and risk shifts are tested for different reactions between groups under the null hypothesis of no differences between groups. SUR is a more powerful test of the hypothesis of no significant share price reactions across groups of banks and for differences among groups. The reason for using SUR is that events are clustered in time and that the change in regulation impacts the whole industry.

V. Empirical Results