Does Earnings Management Explain the Performance of Canadian Private. Placements of Equity?

Transcription

1 Does Earnings Management Explain the Performance of Canadian Private Placements of Equity? MAHER KOOLI Maher Kooli is a associate professor of finance in the School of Business and Management at University of Quebec in Montreal (UQAM), CP 6192, succursale Centre-Ville, Montreal (Quebec) CANADA, H3C 4R2, Tel : , 2082#, Fax : , kooli.maher@uqam.ca.

2 Does Earnings Management Explain the Performance of Canadian Private Placements of Equity? Abstract Using a sample of 434 Canadian private placements of equity (PPEs) that occurred from 1996 to 2005, we first examine the long-run performance following PPEs, and secondly, we analyze the earnings management hypothesis as an explanation to the aftermarket performance. We find that Canadian PPEs do underperform on a calendar-time basis as well as on event-time basis. We also find that the most aggressive earnings management firms issues larger offerings than the most conservative ones but post the worst long term performance. The result for the most aggressive quartile is consistent with the over-optimism hypothesis. However, we find that private placements issuers unlike public issuers are less inclined to manage earnings around the time of the offering. 1

3 An overview of the initial public offerings (IPOs) and seasoned equity offerings (SEOs) studies reveals the existence of severe aftermarket underperformance for issuers. This phenomenon has been reported in the U.S. and in other countries 1. It is also observed by Hertzel, Lemmon, Link and Ress [2002] with U.S. private placements of equity (PPEs). Hertzel et al., indeed, observe a positive market reaction to the PPE announcement followed by a long-run underperformance. They retain the investor optimism hypothesis as a possible explanation for this stock price behaviour. This result seems surprising giving the particularity of PPE, typically negotiated with a group of informed investors without SEC supervision. If the aftermarket underperformance phenomenon also exists among PPEs, then it raises questions concerning aftermarket efficiency. Therefore, it seems important, before accepting or rejecting the efficient market hypothesis, to further examine the robustness of the U.S. findings using non U.S data. The PPEs market in Canada is of special interest, given the fact that companies issuing equity privately in Canada tend to be much smaller than their U.S. counterparts but the number of Canadian PPEs is much higher than that in the U.S. Canadian Exchanges have also different listing requirements from those in the U.S 2. Hence, we may expect dissimilar results. This paper presents two distinctive features. First, we use both event-time as well as a calendar-time framework to measure PPE aftermarket performance. Secondly, we examine the cross-sectional variance of long-run performance. We focus specifically on the earnings management hypothesis. Teoh, Welch and Wong [1998a and 1998b], Teoh, Wong and Rao [1998], and Ducharme, Malatesta and Sefick [2004] provide evidence that earnings management can explain, at least in part, the aftermarket performance of public equity issues. Recently, Chou, Gombola and Liu [2006] investigate whether earnings management explanation for long-run stock performance of public issues also holds for equity private issues. Their results suggest that the observed underperformance following private placements 2

4 is not attributable to earnings management. Indeed, due to limited information asymmetry between informed private investors and managers of firms issuing private equity, we could expect less earning manipulation in PPE than in IPO or SEO. However, Managers of PPE could still manage earnings to compensate private investors through high discount 3. In doing so, they mislead outside investors that could not see any earnings manipulation. These investors could be disappointed when earnings are lower than expected, which could explain the aftermarket underperformance of PPE. In other words, after a period of over-optimism boosted by earnings manipulation, outside investors are disappointed and private equity issuers underperform in the long run. We reexamine this hypothesis for our PPE Canadian sample. This paper is one of the few studies available that examine PPEs in Canada. The rest of the paper is organized as follows: In the next section, we present data and methods used to measure the aftermarket performance of PPEs and to detect earnings manipulation. Long-run performance and earnings management test results are presented in section 3. The last section concludes the paper. DATA AND METHODS We first describe the data and then the methods. Data The database on PPEs in Canada is built from the Record of New Issues (RNI) held by The Financial Post Data Group, and covers the period going from 1996 to The following criteria are used in selecting the final sample: (1) We retain only common-share offerings, and exclude units, 4 closed-end funds, and real estate investment trust offerings, (2) Issuing firms are listed on the Toronto Stock Exchange, (3) financial statements, stock price/return data for issuers, market capitalisation and book-to-market ratio are available on the Research Insight Compustat and Financial Post Corporate Retriever databases, and (4) to address the crosssectional dependence problem induced by overlapping observations [Lyon, Barber, and Tsai, 3

5 1999], we retain a purged sample. If a PPE occurs within three years of a previously included offering by the same firm, we remove the later observation. No look-ahead bias is present in this restriction [Loughran and Vijh, 1997]. Our final sample consists of 434 Canadian PPEs. Exhibit 1 reports the time distribution of the final sample. Similar to IPOs, Canadian PPEs are clustered in time. The most active years are 2003, 2004 and These years represent 60.83% of the sample. The largest number of PPEs is observed for 2005, while the largest PPE gross proceeds is observed for *** Insert Exhibit 1 about here*** Exhibit 2 reports the distribution of the PPE sample across two-digit SIC code, and Exhibit 3 presents the distribution of the sample by size. The most active industries in PPEs are the mining industry, which represents (in number) 65.21% of the sample. Inspection of exhibit 3 shows that small PPEs (gross proceeds < CAN$ 5 million) represent 62.44%, of the sample, while large PPEs (gross proceeds > CAN$ 50 million) represent 2.76% of the sample. Thus, Canadian PPEs in the sample are skewed toward small private offerings and mining issuers. *** Insert Exhibit 2 and 3 about here*** Methods We first present the different methods used to examine the long-run performance of Canadian PPEs and then the model used to detect earnings management. Long-run performance of PPEs: To guarantee the robustness of the results, we build our conclusions on different methods. We first examine the cumulative abnormal returns (CARs) as an event-time method. We then consider the alphas from the Fama-French [1993] three factors (FF-TFPM) and the fourth factors (FFPM) models as calendar-time methods. 4

6 Event-Time Analysis : The analysis of CARs is warranted if a researcher is interested in answering the following question: do sample firms persistently earn abnormal monthly returns? A key feature of the analysis is the careful construction of reference portfolios, which alleviates the new listing and rebalancing biases [Barber and Lyon, 1997 and Kothari and Warner, 1997]. The reference portfolios are purged from event firms and are formed continually on the basis of firm size and book-to-market ratios. To construct the size control portfolio, all Canadian stocks are ranked each month according to their market capitalisations, and four quartile portfolios are formed (with equal numbers of firms in each portfolio). Independently, all Canadian stocks are also ranked according to their book-to-market ratios, and four portfolios are formed (with equal numbers of firms in each portfolio). The returns of the 16 monthly rebalanced [Rau and Vermaelen, 1998] portfolios are calculated as the valueweighted average of the individual-firm monthly returns in each of the size-be/me quartile intersections. Each PPE is then assigned a control portfolio based on its market capitalisation and book-to-market ratio over the performance test period examined. Thus, the average cumulative abnormal return CAR 1 to q month q (12, 24, 36) is calculated as: for the PPE portfolio from the offering month to the event q N CAR1 to q ARs with s 1 ARs N s 1 i 1 s r i, s (1) AR s : the average abnormal return of the PPE portfolio in event time s N s : number of firms for which returns are available in event time s (s=1 to 36) r i,s: abnormal return of firm i (the difference between the return from the issuing firm, R i,s, and the return from the control portfolio, R cpi,s ) Calendar-Time Analysis: In addition to the event-time analysis, we employ the calendar-time method, which allows the simulation of an investment strategy that could be implemented by a portfolio manager. Fama [1998] recommends the construction of monthly portfolios in 5

7 calendar time to be used in measuring the average abnormal long-run performance, for the following reasons: first, monthly returns are less subject to the bad model problem; secondly, monthly portfolios allow the cross-correlation between the firms in the sample to be taken into consideration, and thirdly, the portfolio returns allow better statistical inferences. First, we consider the alphas from the Fama-French [1993] three factors model (FF-TFPM) as a calendar-time portfolio method. For the alpha coefficient, we calculate, for each calendar month, the return on a portfolio composed of firms that issued equity within the following T years (T=1 to 3) of the calendar month. Then, the calendar-time return on this portfolio is used to estimate the following regression: TFPM : R s SMB h HML e (4) p, t R f, t p p ( Rm, t R f, t ) p t p t p, t The dependent variable of the regression is the monthly excess return of the portfolios (R p,t - R f,t ), which corresponds for a given month, t, to the returns of the portfolio of IPOs (R p,t ) less the risk-free rate (the monthly rate of 91-day Canadian Government Treasury bills, R f,t ). The independent variables are the excess market return and 3 zero-investment portfolios constructed such as to mimic the risk factors common to all securities. 5 p, s p, h p stand for the loadings of the portfolio on each risk factor: the market, SMB (size) and HML (book-tomarket ratio). The parameter ( ) in equation (4) indicates the monthly average abnormal return of the PPE sample. Second, we consider the alphas from the four factors model (FFPM), which includes the FF- TFPM and a momentum factor suggested by Carhart [1997]. The regression model is: FFPM : R PR YR e (5) p, t R f, t p p ( Rm, t R f, t ) s p SMBt hp HMLt p p 1 t p, t The dependent variable of the regression is the monthly excess return of the portfolios (R p,t - R f,t ), which corresponds for a given month, t, to the returns of the portfolio of IPOs (R p,t ) less the risk-free rate (the monthly rate of 91-day Canadian Government Treasury bills, R f,t ). The independent variables are the excess market return and 3 zero-investment portfolios 6

8 constructed such as to mimic the risk factors common to all securities. p, s p, h p and p p stand for the loadings of the portfolio on each risk factor: the market, SMB (size), HML (book-tomarket ratio), and PR1YR (momentum) 6. The parameter ( ) in equation (5) indicates the monthly average abnormal return of the PPE sample. Measuring earnings management Chou, Gombola and Liu [2006] note that earnings management can be accomplished through recognition of revenues or delaying recognition of expenses relative to cash flows. It can also be accomplished through changes in accounting methods and in capital structure. However, not all accruals items are subject to manipulation. Long-term accruals are less susceptible to manage with respect to long-term accounting choices already taken. Thus, we focus in this paper, as Teoh, Welch and Wong [1998a and 1998b] on short-term accruals which are easier to manage. However, we have to decompose current accruals into two components, one that is nondiscretionary, dictated by firm and industry conditions and one that is discretionary that reflects managers desire for a higher short-term share price. Following Teoh, Welch and Wong [1998a and 1998b], we use the modified Jones [1991] model to estimate expected accruals. Current accruals (CA) are the change in noncash current assets minus the change in operating current liabilities: CA = [current assets (4)-cash (1)] [current liabilities (5)-current maturity of long-term debt (44)] (6) Note that numbers in parentheses are Compustat item numbers. To estimate non discretionary accruals, we run a cross-sectional regression of current accruals in a given year on the change in sales. We use an estimation sample that includes all Canadian industry matched firms excluding PPE issuers. We scale by total assets to control for 7

9 heteroskedasticity. We run the following cross-sectional regression using the estimation sample: CA j, t / TA j, t-1 = α 0 (1/TA j, t-1 ) + α 1 (Δ Sales j, t / TA j, t-1 ) + ε j, t, (7) where j firms belong in the same two-digit SIC code as the issuing firm but excluding the issuer, TA j, t-1 is total assets in year t-1, and Δ Sales j, t is the change in sales in year t for firm j. The nondiscretionary current accruals (NDCA), is estimated as: NDCA i,t = (1/TA i, t-1 ) + ((Δ Sales i, t - ΔTR i, t )/ TA j, t-1 ), (8) where ΔTR i, t is the change in trade receivables in year t for PPE I, is the estimated intercept, and is the slope coefficient for PPE i. Discretionary current accruals (DCA i,t ) for PPE i for year t is then estimated as: DCA i,t = CA i,t / TA i,t-1 NDCA i,t (9) RESULTS Long term performance of Canadian PPEs Event-Time Returns: In this section, we present analyses of the returns of Canadian PPEs in event time. Exhibit 4 presents the CARs for the three years following the private issue. *** Insert Exhibit 4 about here*** The data shows that CARs are negative and statistically significant in the three years following the private issue. For example, they reach -5.94% (t-statistic= 1.87) over 1 year, % (t-statistic= -1.86) over 2 years, and -6.66% (t-statistic= -5.22) over 3 years. These results confirm the U.S. evidence of underperformance following a private placement. For example, Hertzel et al. [2002] calculate a mean three-year buy-and-hold abnormal returns for the size-and-book-to market-matched control portfolios of % (t-statistic= -4.68), for a 8

10 sample of 952 U.S. PPEs. Exhibit 5 presents CARs by issue year. It shows that CAR varies from year to year. For example, the highest performance is for 1997 PPEs (3-year CAR = %, (t-statistic = -1.70)), while the lowest performance is for 2001 PPEs (3-year CAR = 9.35 %, (t-statistic = -1.15)). *** Insert Exhibit 5 about here*** Exhibit 6 presents CARs distribution by industry. It shows that 3-year PPE performance varies across industries. For example, the transport, communication, electric, gas and sanitary services PPEs post the weakest performance (3-year CAR = %), but statistically non significant. The finance, insurance and real estate industry 3-year CARs is % (tstatistics = -2.43), while wholesale trade PPE posts a positive 3-year CAR of 5.69 % but not statistically significant. *** Insert Exhibit 6 about here*** Exhibit 7 presents CARs distribution by size quartile. 4.4 représente le RAC (0, +36 mois) par quartile de produit brut. It shows that 3-year CAR for the first quartile (PPEs less than 2.1 million dollars) is % significant at the 10 % level (t-statistic = ). The fourth quartile, (PPEs over than 8.01 million dollars) posts even a more pronounced underperformance (3-year CAR = %, t-statistic= ). These observations suggest that the underperformance of PPE is not exclusive to small private offerings. However, we should mention that Canadian PPEs are on average much smaller than the U.S. ones, which could explain our results. *** Insert Exhibit 7 about here*** 9

11 Calendar-Time Returns: In this section, we examine calendar-time returns for the PPEs Canadian sample. As mentioned previously, calendar-time portfolios represent an important improvement over the traditional event methodology, which assumes independence of individual-firm abnormal returns. Exhibit 8 reports the three-factor time series regression results for the three years following the issue. The intercept from the FF-TFPM regression is negative (-1.23%) and significant (tstatistic=-2.09). The significantly positive HML and SMB loadings means that the Canadian PPEs examined are small and value stocks, which confirms our previous observations. *** Insert Exhibit 8 about here*** Exhibit 9 reports the four-factor time series regression results for the three years following the issue. The intercept from the FFPM regression is -1.15% (t-statistic=-1.71), -1.21% (tstatistic=-1.99) and -1.14% (t-statistic=-1.84), for 1, 2 and 3-year CAR, respectively. Overall, Canadian PPEs do underperform on a calendar-time basis as on event-time basis. ***Insert Exhibit 9 about here*** Earnings management hypothesis Exhibit 10 reports the median and the mean of discretionary current accruals (DCA) estimated by the modified Jones [1991] model, for the 3 years before and after the private issue 7. The median DCA for the issue year is 0.2% but statistically non significant. None of the other years surrounding the PPE show a significant positive mean or median value for DCA. Even if Exhibit 10 results do not show any significant earnings management evidence, further investigation is warranted giving the DCA positive value at the issue year. Hence, following, Teoh, Wong and Rao [1998] and Chou, Gombola and Liu [2006], we classify private issuers in quartiles according to issue-year DCA and then examine the relation between the long term 10

12 performances and DCA quartiles. We expect that aggressive earnings management issuers post the worst long term performance. ***Insert Exhibits 10 about here*** Exhibit 11 reports the distribution of our PPE sample by issue-year DCA quartiles. The mean (median) DCA of the first quartile (the most aggressive earnings management issuers) is (0.2462), while the mean (median) DCA of the fourth quartile (the most conservative earnings management issuers) is ( ). In terms of size, we find that the most aggressive quartile issues larger offerings than the most conservative quartile (17.56 million $CAN vs. 4.55). This result is different from the U.S. evidence. Chou, Gombola and Liu [2006] find, indeed, that the most PPE aggressive DCA quartile are smallest in firm size. Teoh, Welch and Rao [1998] also confirm this observation for their IPO sample. ***Insert Exhibits 11 about here*** Exhibit 12 reports the 3-year performance 8 of our PPE sample across DCA quartiles. Exhibit 12 shows that issues in the most aggressive DCA quartile have the worst mean 3-year CARs (-9.89% (t-statistic= -2.37)). For the three other quartiles, the results are not statistically significant. These observations are confirmed by Chou, Gombola and Liu [2006]. They find that (p.28) there is not much of a pattern of abnormal returns across DCA quartiles, except for significant underperformance shown by the most aggressive DCA quartile. These results also confirm the fact that private placements issuers unlike public issuers are less inclined to manage earnings around the time of the offering. Our results for the most aggressive quartile confirm Teoh, Welch and Wong [1998a and 1998b] hypothesis, which stipulates that earnings manipulation could be used to bump up investor optimism and could explain, at least in part, the aftermarket underperformance. Consistent with this view, Hertzel et al. [2002, p.1] suggest that investors are overoptimistic about the prospects of firms issuing equity, regardless of the method of issuance (public or private issues). 11

13 ***Insert Exhibit 12 about here*** CONCLUSION This study attempts to fulfill the need for out-of-sample evidence on the long-run performance of PPEs. Using a sample of 434 Canadian PPEs that occurred from 1996 to 2005, we first examine the long-run performance following PPEs, and secondly, we analyze the earnings management hypothesis as an explanation to the aftermarket behaviour. The main results are the following: First, Canadian PPEs do underperform on a calendar-time basis as well as on event-time basis. For example, 3-year CARs is -6.66% and is statistically significant. Compared to public offerings, PPE underperformance is much smaller than that reported by Kooli, L Her and Suret [2006] for a sample of 141 Canadian public offerings during Second, the distribution of our PPE sample by issue-year DCA quartiles shows that the most aggressive quartile earnings management firms issues larger offerings than the most conservative quartile. Third, when we examine the relation between 3-year CARs and DCA quartiles, we find that the most aggressive earnings management issuers post the worst long term performance. This result is consistent with the over-optimism hypothesis as suggested by Hertzel et al. [2002]. However, we should note that, except for the most aggressive private issuers, private placements issuers unlike public issuers are less inclined to manage earnings around the time of the offering. Although, we shed some light about the stock market reaction to PPE in Canada, the issues raised and the complex nature of these transactions calls for further analysis. 12

14 REFERENCES Barber, B., and J. Lyon. Detecting long-run abnormal stock returns: the empirical power and specification of test statistics. Journal of Financial Economics, 43(1997), pp Carhart, M. On persistence in mutual fund performance. Journal of Finance, 52(1997), p Carpentier, C., J.F. L'Her, and J.M. Suret. PIPEs: A Canadian perspective. The Journal of Private Equity, 8(2005), pp Chou, De-Wai., Michael. Gombola and Feng-Ying. Liu. Earnings management and long-run stock performance following private equity placements. Drexel University working paper, DuCharme, Larry L., Paul H. Malatesta and Stephan E. Sefcik. Earnings management, stock issues, and shareholder lawsuits. Journal of Financial Economics, 71(2004), pp Fama, E. F., and K. R. French. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33(1993), pp Fama, E.F. Market efficiency, long term returns, and behavioral finance. Journal of Financial Economics, 49(1998), pp Hertzel, M., and R. L. Smith, Market discounts and shareholder gains for placing equity privately. Journal of Finance 48 (1993), pp Hertzel, M., M., Lemmon, J., Linck and L. Rees, Long-run performance following private placements of equity. Journal of Finance, 57(2002), pp Jones, J. Earnings management during import relief investigation. Journal of Accounting Research, 29 (1991), pp Kooli, M. J.F. L Her and J.M. Suret. Do IPOs really underperform in the long run? New 13

15 evidence from Canadian market. The Journal of Private Equity, (9)2006, pp Kothari, S.P. and J.B. Warner. Measuring long-horizon security price performance. Journal of Financial Economics, 43(1997), pp Loughran, T. and A.M. Vijh. Do long-term shareholders benefit from corporate acquisitions? Journal of Finance, 52(1997), pp Rau, P.R. and T. Vermaelen. Glamour, value and the post-acquisition performance of acquiring firms. Journal of Financial Economics, 49(1998), pp Ritter, J.R., and I. Welch. A review of ipo activity, pricing, and allocations. Journal of Finance, 57(2002), pp Teoh, S. H., T.J. Wong., and G.R. Rao., Are accruals during initial public offerings opportunistic? Review of Accounting Studies, 3(1998), pp Teoh, S.H., I. Welch, and T.J. Wong, Earnings management and the underperformance of seasoned equity offerings. Journal of Financial Economics, 50 (1998a), pp Teoh, S.H., I. Welch, and T.J. Wong, Earnings management and the long-run market performance of initial public offerings. Journal of Finance, 53 (1998b), pp

16 EXHIBIT 1 Canadian PPE time distribution Year Number Total gross proceeds (million CAN$) Average gross proceeds (million CAN$) EXHIBIT 2 Canadian PPE distribution by primary SIC code Industry SIC Number % Total gross proceed (million CAN$) % Average gross proceed (million CAN$) Mining % % 7.66 Construction % % 8.26 Manufacturing % % Transportation, communications, % % 8.36 electric, gas and sanitary services Wholesale trade % % 6.78 Retail trade % % Finance, insurance and real estate % % Services % % 8.55 Total % % EXHIBIT 3 Canadian PPE distribution by size Gross proceeds (CAN$ millions) Number % Gross proceeds < CAN$ 5 million % CAN$ 5 million < Gross proceeds CAN$ 10 million % CAN$ 5 million < Gross proceeds CAN$ 50 million % Gross proceeds > CAN$ 50 million % Total % 15

17 EXHIBIT 4 Cumulative abnormal returns (CARs) The sample consists of 434 Canadian PPEs, from 1996 through Month CAR Bootstrapped skewness-adjusted t-statistic (0,+12) %* (0,+24) %* (0,+36) %*** *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively EXHIBIT 5 Cumulative abnormal returns distribution (CAR) by year The sample consists of 434 Canadian PPEs, from 1996 through Year 3-year CAR t-statistic % % %** % % %* % %*** %*** %*** *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively EXHIBIT 6 Cumulative abnormal returns distribution (CAR) by industry The sample consists of 434 Canadian PPEs, from 1996 through Industry CAR (0, +36 month) t-statistic Mining %*** Construction % -1 Manufacturing %** Transportation, communications, electric, gas and sanitary services % Wholesale trade 5.69 % Retail trade % Finance, insurance and real estate %** Services 1.14 % Total %*** *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. 16

18 EXHIBIT 7 Cumulative abnormal returns distribution (CAR) by size The sample consists of 434 Canadian PPEs, from 1996 through Gross proceeds (CAN$ millions) CAR (0, +36 month) t-statistic 1 st quartile < %* <= 2 nd quartile < % <= 3 rd quartile < % th quartile >= %*** *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. EXHIBIT 8 Calendar-time Fama-French (1993) three-factor model. The sample consists of 434 Canadian PPEs, from 1996 through (0, +12 month) t-statistic (0, +24 month) t-statistic (0, +36 month) t-statistic Abnormal return %** %** %** R m -R ft ** ** ** SMB *** *** ** 2.38 HML *** *** *** Adjusted R *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. EXHIBIT 9 Calendar-time four-factor model The sample consists of 434 Canadian PPEs, from 1996 through (0, +12 month) t-statistic (0, +24 month) t-statistic (0, +36 month) t-statistic Abnormal return %* %** %* R m -R ft *** ** ** SMB *** ** ** 2.30 HML *** *** *** 3.41 PRY1R Adjusted R *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. 17

19 EXHIBIT 10 Discretionary current accruals (DCA) for Canadian PPEs from Year (issue year) Median Mean EXHIBIT 11 Discretionary current accruals (DCA) quartiles for Canadian PPEs from Quartiles 3-year CAR t-statistic 1 st quartile (most aggressive, DCA >=0.072) %** nd quartile (0.001 <= DCA < 0.072) % rd quartile ( <=DCA < 0.001) % th quartile (most conservative, DCA < ) % *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. EXHIBIT 12 3-year Cumulative abnormal returns for the Discretionary current accruals (DCA) quartiles of 443 Canadian PPEs from Mean DCA Median Average gross proceed (million CAN$) 1 st quartile (most aggressive, DCA >=0.072) nd quartile (0.001 <= DCA < 0.072)) rd quartile ( <=DCA < 0.001) th quartile (most conservative, DCA < ) *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. 18

20 ENDNOTES 1 See Ritter and Welch [2002] for an excellent review of IPO studies. 2 See Carpentier, L Her and Suret [2005] for a detailed comparison between Canadian and U.S. private placements of equity. 3 Hertzel and Smith [1993] suggest that discounts reflect compensation to private placement investors for information production. 4 Unit offerings are excluded because we were was not able to separate the values of the components of offerings (usually common stock with warrants). 5 All data comes from Compustat, and the market return is a value-weighted return computed within the sample. We have constructed SMB and HML in keeping with Fama and French (1993). 6 See Carhart [1997] for the momentum factor construction. 7 We should note in this section that our period of study is , because the data from Compustat to calculate 2005 DCA was not available. 8 Using the alphas from the FF-TFPM or the FFPM rather than CARs does not alter the results. 19