In a forthcoming article, Chen and Ritter (2000) suggest that gross spreads (fees) received by U.S. underwriters for common equity initial public offerings (IPOs) are considerably higher than those received by underwriters in other countries. They claim that fees charged to underwrite new issues of common equity in Japan, Hong Kong and Europe are about half the level of U.S. fees. In addition, the authors cite Lee, Lochhead, Ritter and Zhao (1996), who report that U.S. fees for bonds, convertible bonds and seasoned equity offerings do not show significant clustering at one particular fee level, whereas the proportion of medium-sized common equity U.S. IPOs ($20 million to $80 million) earning a fee of exactly 7% has approximately tripled over the last 10 years.

Hot on the heels of a recent $1 billion (U.S.) settlement for price-fixing by market makers on the NASDAQ,1 the above revelations have prompted another class action lawsuit undertaken by a retired Florida real estate businessman who names 27 of the top U.S. brokerages. The suit claims that investment firms have colluded for years on the prices they charge to underwrite IPOs. If the suit is successful, damages awarded may run into the billions of dollars.2

This study examines a sample of 330 common equity IPOs that listed on the Toronto Stock Exchange (TSE) in the period 1984-1997. The aim here is to analyse various aspects of fees charged by Canadian underwriters, and to compare and contrast the findings with those of the U.S. study.

This research will be of particular interest to Canadian firms who have recently issued, or are considering the issuing of, common equity for the first time. It will also be of interest to Canadian brokerage firms who can compare their practices to others in the industry, and to regulators who are charged with the responsibility of maintaining fair competition among firms servicing the new issues market. Investors themselves may be interested in facts surrounding one of the significant costs that are embedded in the new shares that they subscribe to. Finally, lawyers on the other side of the aforementioned lawsuit may want to know that fees charged by Canadian underwriters are only slightly less on average than those charged by their U.S. counterparts.

The Data
The data used in this study come from several sources. For the period 1984-93 we used the TSE Annual New Listings Report to identify TSE IPOs. To maintain sample homogeneity, unit offerings, warrants, preferreds and other hybrids were filtered out. Prospectuses for the remaining IPOs were then examined to capture information on syndicate composition and fees charged.

For the period 1994-97, we used the Financial Post's Record of New Issues. This annual source lists all new issues in Canada. It is split up into sections on equity, debt and preferred shares. Common equity IPOs were picked out, and the resulting list was cross-checked against the TSE Review to highlight only those common equity IPOs that listed on the TSE.

The final sample consists of 330 common equity IPOs. Most of these are for industrial companies, but 27 firms out of the total are in either oil and gas or mining.

The Distribution of IPO Fees
To control for inflation, IPO offer size is adjusted to 1997 prices using the Canadian annual Consumer Price Index. Over-allotment options are excluded from this study, and would be expected to increase revenue (but not fee percentages charged) for underwriters since they allow additional shares to be purchased from the issuing firm at the same discount as the rest of the issue.3

Table 1 (see "IPO Fee Frequency Distribution and Descriptive Statistics," below) shows the frequency distribution and descriptive statistics for IPO fees in the sample. There is some apparent clustering of fee frequency at the three integer levels shown (5%, 6% and 7%) with particularly strong clustering at exactly 6% (28.18% of the sample).4

The fee range containing the highest proportion of IPOs is 6%-6.99% (57.57% of the sample). The mean fee charged in the sample is 6.09%. While this fee is high relative to those charged in countries outside the U.S., it is still less than 1% below the mean fee in the U.S. study (6.82%).

It is interesting to note that the distribution of Canadian fees is close to normally distributed as the mean, median and mode virtually coincide, skewness is positive but close to zero and kurtosis is only slightly larger than three.

Chen and Ritter show evidence that fee clustering at exactly 7% is particularly high and increasing over time for medium-sized IPOs ($20 million-$79.99 million--57% of the sample). To facilitate a comparison with Chen and Ritter, we split our sample into small (under $10 million--15% of the sample), medium ($10 million-$50 million--61% of the sample) and large (over $50 million--25% of the sample) IPOs.

Table 2 (see "IPOs by Year, Offer Size and Underwriter Fee, 1984-1997," p. 30) gives a frequency breakdown by fee range, offer size and year.5 Medium sized IPOs do exhibit higher fee clustering at 6% (36% of all medium IPOs). But the proportion is considerably less than the 73% at 7% in Chen and Ritter.

Chen and Ritter also note that clustering at 7% in the most recent three years (1995-1997) is a staggering 90%. Our data indicate that clustering at 6% in the three years 1995-1997 for medium-sized IPOs is actually lower at 33 1/3%. We find that the percentage of medium-sized IPOs, with fees at exactly 6%, is not increasing over time. In fact, these fees tend to follow a cyclical pattern.

The underpricing of IPOs is a well-documented phenomenon. If underwriters underprice to compensate uninformed investors as Rock (1986) suggests, or because they benefit from asymmetric information in terms of knowledge of true value compared to issuing firms as in Baron (1982), then underwriters may be predisposed to reduce fees for underpriced (relative to overpriced) issues since selling efforts are reduced.

We split the sample additionally into underpriced and overpriced issues. It does not appear that fees are generally lower for underpriced offerings. Indeed, the opposite may be true as 23.7% (32.5%) of underpriced (non-underpriced) IPOs have fees below 6%, while 46.8% (40.8%) of underpriced (overpriced) IPOs have fees above 6%.

Next, we look at equally- and value-weighted averages (weighted by issue size) over time. These results appear in Table 3 (see "Average IPO Fees by Year and Offer Size, 1984-1997," p. 30). On average, fees have not changed dramatically over the 14 years of this study. Although there was a bit of a decline in the lean years following the 1987 crash, that was followed by a complete recovery for medium-and large-sized IPOs in particular.

A chi square test of independence between fees and time for medium-sized IPOs is conducted next. Observed cell frequencies and results indicate that we cannot reject the null of independence between fees and time at the 5% level. It further supports the notion that fees for medium-sized IPOs are not increasing over time, in contrast with the Chen and Ritter study.

The IPO Fee Structure of Canadian Underwriters
In the United States there are literally hundreds of firms that are in the business of underwriting new security issues. The lawsuit currently underway points a finger at 27 of the largest firms and is undoubtedly motivated by the fact that these firms are sufficiently solvent to make the necessary payment(s) should the verdict go against them.

In Canada, there are far fewer investment bankers, and only about a dozen or so prominent ones. The majority of these prominent underwriters are affiliated with the country's largest banks as legislation was enacted more than a decade ago to permit it.6 For the years 1993-1997, syndicate participation by 11 prominent Canadian underwriters appears in Table 4 (see "TSE IPO Syndicate Participation by Leading Canadian Underwriters, 1993-1997") for all IPOs in the sample, as well as for only medium-sized IPOs.

We concentrated our analysis on the most recent period of time to limit the complications arising from prior changes in the identity of several key players in the industry, and to try to conform to the time period that Chen and Ritter focus on (1995-1997).7 RBC Dominion Securities leads in terms of syndicate participation frequency for the full sub-sample, while Midland Walwyn was involved in the largest number of medium-sized IPOs.8

Table 5 (see "Mean Fee for Medium TSE IPOs by Brokerage Firm, 1993-1997," below) contains mean fees charged by the 11 brokerages for the syndicates that they were involved in. The only mean fees that are significantly different at conventional levels are those of Toronto Dominion Securities with Nesbitt Burns, and Toronto Dominion Securities with ScotiaMcleod. All other paired differences have p-values that are greater than 10%.

The Collusion Question
The study by Chen and Ritter suggests that there may be collusion among leading U.S. investment bankers, since the proportion of fees that are exactly 7% is extremely high and has been steadily increasing. An examination of fees charged by 12 brokerages is also shown in the Chen and Ritter study, and the concentration at 7% is also unusually high.

These facts do not, by themselves, prove that collusion took place. In reality it is always easier to disprove an assertion than to prove it true. To disprove, all that is usually required is to provide counter-examples or other plausible explanations. To prove that an assertion is true, an exhaustive analysis of all possible, reasonable, alternatives must also be proven true.

In the Canadian context, one cannot as easily assert that investment bankers collude to set prices charged for either all or medium-sized IPOs. The analysis shows that fee frequency distributions for leading Canadian IPO underwriters of common equity, in the period 1993-1997, exhibiting a concentration at exactly 6% is never more than 37.7% (First Marathon). Figure 1 (see "TSE IPO Fee Distributions for Leading Canadian Underwriters 1993-1997, Medium Sized Issues," p. 32 and 34) focuses attention on medium-sized IPOs. Here the fee concentration at exactly 6% runs as high as 57.8% (Nesbitt Burns), with ScotiaMcleod coming in second (48.7%). Is there sufficient evidence to assert that widespread collusion exists among firms servicing the Canadian new issues market?

A stronger case for collusion could be made if the difference in the proportion of fees at exactly 6% was statistically insignificant between an overwhelming majority of pairs of investment bankers. We consider this possibility by performing a series of paired t-tests for the difference between proportions at exactly 6%. We perform this test for each of the 55 pairs of combinations of the 11 leading Canadian brokerage firms.

Eighteen differences (33% of the tests) are significant at the 10% level, while seven differences (13%) are significant at the 5% level and two differences (4%) are significant at the 1% level. These results do not indicate an unusually high similar concentration of fees at 6%.

Conclusion
Fees charged by U.S. investment banks for distributing common equity IPOs are high compared to fees charged outside of North America. But they are only marginally greater on average than fees charged by Canadian firms. Canadian firms seem to concentrate fees charged for distributing medium-sized IPOs ($10-50 million) at the 6% level, compared to U.S. fee clustering at 7%. Further, the clustering of fees at 6% for medium-sized Canadian IPOs follows a cyclical pattern over time in contrast to the steadily increasing proportion of fees at exactly 7% charged by American brokerages.

While there exists some evidence to conclude that American brokerages may collude on fees charged, the Canadian evidence on this point is much weaker. In particular, paired t-tests for the difference in proportion of fees at exactly 6% charged by 11 leading Canadian underwriters showed that there were a substantial number of instances where significant differences existed. It is reasonably safe to say that any lawsuit directed at Canadian underwriters concerning collusion to set prices in the new issue market would be largely unsuccessful.

Endnotes
1. See Christie and Schultz (1994).

2. See Lang (1998).

3. See Chung, Kryzanowski and Rakita (1997) for a discussion of the role of the over-allotment option in Canadian IPOs.

4. The second highest level of fee frequency clustering was at 6.5% (14.55% of the sample).

5. We were forced to aggregate some of the frequency data across several year ranges due to the small number of IPOs issued in particular years. The five-year interval following the 1987 market crash was a period of extremely low IPO activity in Canada.

6. The Royal Bank of Canada owns RBC Dominion Securities while The Bank of Montreal owns Nesbitt Burns. The Canadian Imperial Bank of Commerce purchased Wood Gundy as far back as 1988 but only effected a name change to CIBC Wood Gundy in 1995. The National Bank owns Levesque, Beaubien and Geoffrion, The Bank of Nova Scotia owns ScotiaMcleod and The Toronto Dominion Bank owns TD Securities.

7. We haven't conformed exactly to the Chen and Ritter time period since we thought that 1993-1994 data should be included as they were particularly strong IPO years in Canada and we wanted to maintain a more meaningful sample size.

8. There have been a number of consolidations in the industry. Throughout this section we attribute IPO involvement of firms that no longer exist to the surviving firm. For example, The Bank of Montreal purchased Burns Fry and merged it with Nesbitt Thomson to create Nesbitt Burns on Sept. 1, 1994. IPO involvement by Burns Fry is therefore included in the information for Nesbitt Burns. Similarly for RBC Dominion Securities which completed its acquisition of Richardson Greenshields on Nov. 1, 1996. On the other hand, Merrill Lynch Canada finalized the purchase of Midland Walwyn on Aug. 26, 1998. Prior to this date, Merrill Lynch Canada was not a major force in the new issue market and is therefore grouped into the "Other" category. At the start of 1998, the board of directors of Marleau Lemire decided to shut down operations in the wake of several years of large losses, the exodus of employees and mounting lawsuits.

References
Baron, David, 1982. "A Model of the Demand for Investment Banking Advising and Distribution Services for New Issues." The Journal of Finance, (September), 37, 955-976.

Christie, William G., and Paul H. Schultz, 1994. "Why do NASDAQ Market Makers Avoid Odd-eighth Quotes?" The Journal of Finance 49, 1813-1840.

Chen, Hsuan-Chi, and Jay R. Ritter, 2000. "The Seven Percent Solution." Forthcoming The Journal of Finance.

Chung, Richard, Lawrence Kryzanowski and Ian Rakita, 1997. "Stabilization and the Role of the Over-allotment Option in Canadian IPOs." Unpublished Concordia University working paper.

Lang, Amanda, 1998. "73-year-old behind IPO suit spending his golden years in court." National Post, November 11, C1.

Lee, Inmoo, Scott Lochhead, Jay Ritter, and Quanshui Zhao, 1996. "The Costs of Raising Capital." Journal of Financial Research19, 59-74.

Rock, Kevin, 1986. "Why New Issues Are Underpriced." Journal of Financial Economics 15, 187-212.

Lawrence Kryzanowski is professor of finance and Ian Rakita is a lecturer in the finance department, faculty of commerce and administration, at Concordia University in Montreal.