r. david mclean, jeffrey pontiff and akiko watanabe - Center for ...
r. david mclean, jeffrey pontiff and akiko watanabe - Center for ...
r. david mclean, jeffrey pontiff and akiko watanabe - Center for ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Share Issuance <strong>and</strong> Cross-Sectional Returns:<br />
R. David McLean<br />
University of Alberta<br />
rd<strong>mclean</strong>@ualberta.ca<br />
(780) 492-8005<br />
International Evidence Ψ<br />
Jeffrey Pontiff<br />
Boston College<br />
<strong>pontiff</strong>@bc.edu<br />
(617) 552-6786<br />
April 4, 2008<br />
Abstract<br />
Akiko Watanabe<br />
University of Alberta<br />
<strong>akiko</strong>.<strong>watanabe</strong>@ualberta.ca<br />
(780) 492-0385<br />
Share issuance predicts cross-sectional returns in a non-U.S. sample of stocks from 41<br />
different countries. The issuance effect is more statistically significant than either the size or<br />
momentum effects, <strong>and</strong> is similar to the book-to-market effect. As in the U.S., the<br />
international issuance effect is robust across both small <strong>and</strong> large firms. Compared to the<br />
U.S., the effect is driven more by low returns after share creation rather than positive returns<br />
following share repurchases. The share issuance effect is stronger in countries in which there<br />
is greater share issuance activity, greater stock market development, <strong>and</strong> stronger investor<br />
protection. The results suggest that the return-predictive ability of share issuance is related to<br />
the ease with which firms can issue <strong>and</strong> repurchase their shares.<br />
Ψ We thank Brad Barber, David Chapman, Wayne Ferson, Mark Huson, Masahiro Watanabe, Seminar<br />
Participants at Michigan State University, <strong>and</strong> an anonymous referee <strong>for</strong> helpful discussions. McLean is grateful<br />
to the Southam/Edmonton Journal Fellowship Award <strong>for</strong> financial support.
Recent studies by Daniel <strong>and</strong> Titman (2006), Pontiff <strong>and</strong> Woodgate (2006), <strong>and</strong><br />
Fama <strong>and</strong> French (2007) show that there is a significant negative relationship between share<br />
issuance <strong>and</strong> average returns in the cross section of U.S. firms. In this paper, we take a first<br />
look at the return predictive ability of firm-level share issuance in international markets. Our<br />
analysis is divided into two parts. In the first part we test whether the issuance effect is<br />
present among non-U.S. firms, <strong>and</strong> compare our results to those reported in U.S. studies. In<br />
the second part of our analysis we test whether the issuance effect differs across countries,<br />
<strong>and</strong> if so whether proxies <strong>for</strong> equity market development, investor protection, <strong>and</strong> other<br />
country characteristics can explain any cross-country differences.<br />
With respect to the first part of our analysis, the use of international data enables a<br />
better underst<strong>and</strong>ing of whether sources of return predictability identified in the U.S. pose<br />
challenges to asset pricing, or whether they are statistical artefacts from data-mining.<br />
Previous research in this vein includes Rouwenhorst (1998) who studies momentum effects<br />
in twelve European markets, <strong>and</strong> Fama <strong>and</strong> French (1998) who examine value effects in<br />
twelve developed countries. These papers find that momentum <strong>and</strong> value display similar<br />
predictability in international markets as in the U.S. Such an investigation is particularly<br />
important <strong>for</strong> share issuance, since Pontiff <strong>and</strong> Woodgate (2006) find that the relationship<br />
between issuance <strong>and</strong> returns is insignificant in their pre-1970 sample, suggesting that the<br />
issuance effect may be sample specific. Our analysis based on non-U.S. firms hence provides<br />
a useful out-of-sample test of the issuance effect.<br />
We use a large sample of firms drawn from 41 non-U.S. countries <strong>and</strong> examine the<br />
existence of an international issuance effect over a 25-year period between 1981 <strong>and</strong> 2006.<br />
Using net issuance measures that reflect both share issuance <strong>and</strong> repurchases, we find that<br />
1
there is a significant issuance effect in international markets. Similar to the recent U.S.<br />
evidence, international share issuance has strong return-predictive ability. A one st<strong>and</strong>ard<br />
deviation difference in annual share issuance is associated with a 0.16% difference in<br />
subsequent monthly returns in the non-U.S. markets, which is half of the magnitude in the<br />
post-1970 U.S. findings reported in Pontiff <strong>and</strong> Woodgate (2006). The predictive ability of<br />
annual issuance is more statistically significant than that of size <strong>and</strong> momentum, <strong>and</strong> is of the<br />
same magnitude as book-to-market. We also find that the issuance effect is robust across<br />
both small <strong>and</strong> large firms, which is consistent with the findings in Fama <strong>and</strong> French (2007).<br />
The evidence in Loughran <strong>and</strong> Ritter (1995), Loughran <strong>and</strong> Vijh (1997), <strong>and</strong> Fama<br />
<strong>and</strong> French (2007) shows that in the U.S. abnormal returns are associated with both positive<br />
share issuance <strong>and</strong> share repurchase. Our sample of non-U.S. firms allows <strong>for</strong> out-of-sample<br />
tests of both effects. We find that non-U.S. firms that issue stock tend to experience larger<br />
negative abnormal returns than do U.S. issuers. 1 There is a buyback effect across non-U.S.<br />
firms, but it is weaker than that found in the U.S., <strong>and</strong> unlike the U.S. effect it is limited to<br />
small firms.<br />
In the second part of our analysis we examine cross-country differences in the<br />
issuance effect. The U.S. evidence suggests that the ease with which firms can issue <strong>and</strong><br />
buyback their shares is an important factor in explaining the issuance effect. Pontiff <strong>and</strong><br />
Woodgate (2006) show that pre-1970 only 37.4% of U.S. firms had nonzero values of share<br />
issuance, <strong>and</strong> that during this period there was no share issuance effect. Post-1970, 75.8% of<br />
U.S. firms had nonzero values of share issuance, <strong>and</strong> during this period the return-predictive<br />
ability of share issuance is strong.<br />
1 Ikenberry, Lakonishok, <strong>and</strong> Vermaelen (2000) find evidence of low abnormal returns following both seasoned<br />
equity offerings <strong>and</strong> stock-financed mergers in Canada. Their sample includes 161 share issuance events over<br />
the period 1989-1995.<br />
2
Across the 41 countries in our sample there is a good deal of variation in share<br />
issuance activity. As an example, in Chile only 11% of our observations have nonzero<br />
issuance values; while in the U.K. 55% of the observations have nonzero issuance values,<br />
suggesting that it is less costly <strong>for</strong> firms to issue <strong>and</strong> repurchase shares in the U.K. than in<br />
Chile. Share issuance should also be less costly in more developed stock markets, so we use<br />
stock market development variables as proxies <strong>for</strong> the cost of share issuance activity as well.<br />
We also test whether the issuance effect is related to investor protection laws. Morck,<br />
Yeung, <strong>and</strong> Yeu (2000) contend that less developed stock markets, which offer fewer<br />
investor protections, have less efficient markets. Morck et al. reason that arbitrageurs are<br />
more willing to trade in developed markets which protect their property rights. If this<br />
reasoning is correct, then we might expect the issuance effect to be stronger in less developed<br />
markets, as the firms in these markets should contain larger mispricing.<br />
We find that the issuance effect is stronger in countries with more issuance activity,<br />
in countries with more developed stock markets, <strong>and</strong> in countries with better investor<br />
protection laws. One interpretation of these results is that market timing is more prevalent in<br />
countries in which it is less costly <strong>for</strong> firms to issue <strong>and</strong> buyback shares. La Porta et al.<br />
(2006) show that countries that have better investor protection laws have more developed<br />
stock markets, <strong>and</strong> offer firms greater access to equity capital. This might explain why the<br />
issuance effect is stronger in countries with better investor protection<br />
Our findings may also be consistent with an efficient markets explanation. A Firm<br />
may attempt to maintain its optimal capital structure by issuing <strong>and</strong> repurchasing shares in<br />
response to changes in its investment opportunities. If changes in investment opportunities<br />
are correlated to changes in risk exposure, then share issuance is likely to have a stronger<br />
3
correlation to stock returns in an environment where it is less costly to issue shares, so firms<br />
rebalance their capital structure frequently, than in an environment in which the costs of<br />
share issuance exceed the benefits of maintaining a target capital structure.<br />
The remainder of the paper is organized as follows. Section 1 discusses data <strong>and</strong><br />
estimation procedures. Section 2 presents regression results using continuous measures of<br />
share issuances, <strong>and</strong> Section 3 provides results based on issuance portfolios that allow us to<br />
study positive <strong>and</strong> negative share issuance effects separately. Section 4 studies the issuance<br />
effect across countries. Section 5 concludes.<br />
1. Data, Variables, <strong>and</strong> Estimation<br />
1.1. Data<br />
Our sample consists of 41 non-U.S. countries, which are listed in Table 1. The data<br />
used in this study are obtained from Thomson Datastream. We use data from monthly stock<br />
price indices to calculate returns. We also use market capitalizations, market-to-book values,<br />
shares outst<strong>and</strong>ing, <strong>and</strong> the capital adjustment index to compute our share issuance measures.<br />
Returns <strong>and</strong> market capitalizations are converted to U.S. dollars. We select common stocks<br />
listed on each country’s major stock exchange(s) from both active <strong>and</strong> defunct research files<br />
of Datastream in order to avoid survivorship bias. We screen the data <strong>for</strong> coding errors via<br />
the methods of Ince <strong>and</strong> Porter (2006) <strong>and</strong> winsorize each variable within country at the top<br />
<strong>and</strong> bottom 1% to eliminate the affects of outliers.<br />
Due to the availability of firm book values in Datastream, our regression analysis<br />
begins in July of 1981 <strong>and</strong> ends in June of 2006. The start <strong>and</strong> end dates vary across<br />
countries based on each country’s data availability (see columns 2 <strong>and</strong> 3 of Table 1). We use<br />
4
data prior to the listed start dates to construct some of the variables that are described in the<br />
following subsection. To be included in our sample, a stock must have sufficient in<strong>for</strong>mation<br />
to generate the annual issuance measure (explained below), market value of equity, lagged<br />
six-month holding period return, <strong>and</strong> current month’s return. Each month we limit our<br />
sample to countries that have at least 50 firm observations in that month.<br />
Our final sample consists of 2,819,906 total firm-month observations (see column 4<br />
of Table 1). Japan represents the largest part of our sample, accounting <strong>for</strong> 14.72% of the<br />
total observations (column 5) <strong>and</strong> 29.56% of the total market value per month (column 9).<br />
The United Kingdom is the second largest <strong>and</strong> accounts <strong>for</strong> 13.27% of the total observations<br />
<strong>and</strong> 11.68% of the total market value. The rest of the countries typically accounts <strong>for</strong> less<br />
than 5% of the total observations <strong>and</strong> market value.<br />
1.2. Variables<br />
Share issuance. The main variable of our interest is the real change in shares<br />
outst<strong>and</strong>ing, or the change in the number of shares outst<strong>and</strong>ing adjusted <strong>for</strong> distribution<br />
events such as stock splits <strong>and</strong> stock dividends. We use the capital adjustment index from<br />
Datastream recorded at the end of month t (CAIt) to calculate the number of real shares<br />
outst<strong>and</strong>ing <strong>for</strong> that month (Adjusted Sharest). The CAI is the cumulative product of the<br />
inverse of the individual-period capital adjustment factor (AX) <strong>and</strong> is analogous to the Total<br />
Factor of Pontiff <strong>and</strong> Woodgate (2006):<br />
CAIt = ∏<br />
i=<br />
t<br />
1<br />
1 /AXi<br />
.<br />
Adjusted Sharest is then given by:<br />
Adjusted Sharest = Shares Outst<strong>and</strong>ingt / CAIt.<br />
5
We use Adjusted Shares to construct two continuous measures of share issuance.<br />
First, we compute a one-year issuance measure (ISSUE) used in Pontiff <strong>and</strong> Woodgate<br />
(2006) <strong>and</strong> Fama <strong>and</strong> French (2007) as:<br />
ISSUEt,t-12 = Ln(Adjusted Sharest) – Ln(Adjusted Sharest-12).<br />
Size. We calculate the natural logarithm of the June-end U.S.-dollar converted market<br />
value of equity from Datastream. This variable, ME, is used to <strong>for</strong>ecast returns from July of<br />
this year through June of the following year.<br />
Book-to-market. We use the Fama <strong>and</strong> French (1992) procedure <strong>and</strong> construct the<br />
natural logarithm of the previous fiscal year-end book-to-market ratio, BM. 2 The book-to-<br />
market ratio is the inverse of the market-to-book value provided by Datastream. To ensure<br />
the availability of accounting in<strong>for</strong>mation, BM is used to <strong>for</strong>ecast returns from July of the<br />
current year through June of the following year.<br />
We follow Pontiff <strong>and</strong> Woodgate (2006) <strong>and</strong> create a book-to-market dummy<br />
variable, BM-Dum. If the book value of equity is either missing or negative, we assign both<br />
BM <strong>and</strong> BM-Dum values of zero. Otherwise, BM-Dum is set to one. The use of BM-Dum<br />
allows us to include firms with either missing or negative book values of equity without<br />
influencing inference of the slope coefficient on BM.<br />
Momentum. Our momentum measure, MOM, is given by the U.S.-dollar returns over<br />
the previous six months. A one month lag of this variable, MOMt-7,t-1, is used to <strong>for</strong>ecast<br />
holding period returns from month t+1 in order to avoid the effect of bid-ask bounce.<br />
Holding Period Returns. The dependent variables in our regressions are subsequent<br />
holding period U.S.-dollar returns. We measure returns over the first month <strong>and</strong> first year.<br />
2 The fiscal year end <strong>for</strong> Japan is March, whereas it is December <strong>for</strong> all other countries.<br />
6
1.3. Estimation<br />
We estimate a set of Fama <strong>and</strong> MacBeth (1973) regressions each month to calculate<br />
linear relations between holding period returns <strong>and</strong> our independent variables. We then report<br />
time-series averages of intercepts, slope coefficients, <strong>and</strong> adjusted R 2 s obtained from the<br />
cross-sectional regressions. We use the procedure of Pontiff (1996) to calculate t-statistics <strong>for</strong><br />
the regression coefficients with autocorrelation-consistent st<strong>and</strong>ard errors that correct <strong>for</strong> the<br />
holding period overlap.<br />
Our goal in this paper is to study the pervasiveness of the issuance effect in<br />
international markets. There<strong>for</strong>e, we pool firm observations from the 41 countries into one<br />
sample <strong>and</strong> conduct all of our analyses using this pooled data. We estimate every regression<br />
both with <strong>and</strong> without country dummy variables. When the country dummies are included,<br />
the regression coefficients provide estimates of the within-country effects. When the<br />
dummies are excluded, the coefficients measure the effects across the entire sample.<br />
Henderson, Jegadeesh, <strong>and</strong> Weisbach (2006) show that countries with large positive<br />
share issuances have low subsequent returns. Their results suggest that there is an aggregate<br />
issuance effect across countries, but do not tell us whether there is an issuance effect across<br />
stocks within each country. The cross-stock issuance effect in the U.S. is documented by<br />
Daniel <strong>and</strong> Titman (2006), Pontiff <strong>and</strong> Woodgate (2006), <strong>and</strong> Fama <strong>and</strong> French (2007). The<br />
use of country dummies in our regression analyses allows us to test whether such a cross-<br />
stock effect exists in international markets.<br />
In addition, we also estimate all of our regressions by both equal-weighting <strong>and</strong><br />
value-weighting each observation. If the results from the two regressions are comparable,<br />
then we infer that the issuance effect is similar across both small <strong>and</strong> large firms in our<br />
7
sample. This investigation is motivated by Fama <strong>and</strong> French (2007), who find that the<br />
issuance effect is pervasive across different size groups in the U.S.<br />
Firms are on average larger in more developed markets than in less developed ones,<br />
so value-weighting in our sample might just be measuring a developed market effect. To<br />
address this issue we use a third weighting scheme, in which we scale each value-weight by<br />
the average market value in the firm’s country (scaled-weight). The scaled-weight is a<br />
within-country value-weight; it allows us to test whether the issuance effect is present in<br />
stocks that are large relative to other stocks in the same country.<br />
2. Return Predictive Ability of Continuous Issuance Measures<br />
2.1. Summary Statistics <strong>for</strong> Continuous Issuance Measures<br />
Table 2 presents summary statistics <strong>for</strong> the variables used in our study. We focus on<br />
the continuous issuance measures, ISSUE. We observe that ISSUE has a pronounced right<br />
skew; it has a mean value of 0.048, which is greater than its 75th percentile value of 0.007.<br />
Pontiff <strong>and</strong> Woodgate (2006) report similar findings <strong>for</strong> the U.S.; their ISSUE measure has a<br />
mean of 0.04 <strong>and</strong> a 75th percentile value of 0.03. 3 These numbers also indicate that the<br />
average level of ISSUE is similar between U.S. <strong>and</strong> non-U.S. firms.<br />
2.2. Determinants of International Share Issuance<br />
We study the characteristics of international share issuances further by examining<br />
how they are related to firm characteristics. Table 3 reports the results from Fama-MacBeth<br />
regressions in which ISSUE is regressed on the values of size, book-to-market, <strong>and</strong><br />
3 Unless otherwise noted, our comparisons of summary statistics <strong>and</strong> regression results to those of Pontiff <strong>and</strong><br />
Woodgate (2006) use their post-1970 study since our sample begins in 1981.<br />
8
momentum that are available immediately be<strong>for</strong>e the ISSUE-construction months, as well as<br />
on the twelve-month lag of ISSUE (LAG-ISSUE). We estimate six different regression<br />
specifications; first using equal-weights, value-weights, <strong>and</strong> scaled-weights with country<br />
dummies <strong>and</strong> then without country dummies.<br />
In all six regressions, the coefficients on size are both negative <strong>and</strong> significant,<br />
suggesting that large firms issue fewer shares than do small firms. The coefficients on the<br />
momentum <strong>and</strong> lagged issuance measures are both positive <strong>and</strong> significant in all six<br />
regressions. This implies that firms with high past returns <strong>and</strong> firms that have recently issued<br />
shares are likely to issue more shares. These findings are consistent with Pontiff <strong>and</strong><br />
Woodgate’s (2006) U.S. findings.<br />
The book-to-market coefficients are positive in all six regressions, but are significant<br />
only in the equal-weighted regressions. This suggests that high book-to-market firms issue<br />
more shares <strong>and</strong> low book-to-market firms buyback more shares, <strong>and</strong> that this pattern is more<br />
strongly observed among large firms. The result is in stark contrast to the findings <strong>for</strong> U.S.<br />
firms, in which low book-to-market firms tend to issue more shares (see Loughran <strong>and</strong> Ritter<br />
(1995), Baker <strong>and</strong> Wurgler (2002) <strong>and</strong> Pontiff <strong>and</strong> Woodgate (2006)).<br />
The use of country dummies does not have a strong effect on the regression<br />
coefficients; economic <strong>and</strong> statistical significances of the coefficients are mostly similar in<br />
the regressions with country dummies <strong>and</strong> those without. However, the average R 2 statistics<br />
from the regressions with country dummies are more than twice as large as those from the<br />
regressions without the dummies. Thus, country of origin explains as much of the variation in<br />
share issuances as do the individual firm characteristics that are included in the regressions.<br />
9
2.3. Share Issuance <strong>and</strong> Future Returns<br />
We now turn to one of the main focuses of our paper, <strong>and</strong> test whether international<br />
share issuance can predict the cross section of stock returns. We conduct our analyses using<br />
holding period returns measured over the first month <strong>and</strong> first year, <strong>and</strong> with equal-weighted,<br />
value-weighted, <strong>and</strong> scaled-weighted returns. 4<br />
Panels A <strong>and</strong> B of Table 4 summarize the results of the regressions in which ISSUE is<br />
the only explanatory variable along with country dummies (Panel A) <strong>and</strong> without the<br />
dummies (Panel B). The ISSUE coefficients are negative <strong>and</strong> significant in all twelve of the<br />
regressions reported in Panels A <strong>and</strong> B. This indicates that annual issuance has a significant<br />
<strong>and</strong> persistent ability to predict cross-sectional returns both within country (Panel A) <strong>and</strong><br />
across countries (Panel B).<br />
The issuance effect is robust to all three weighting schemes, suggesting that the effect<br />
is strong in both large <strong>and</strong> small stocks. Looking across the columns in Panel A <strong>and</strong> B we see<br />
that, <strong>for</strong> a given holding period the, coefficients <strong>and</strong> t-statistics are similar <strong>for</strong> each of the<br />
three weighting schemes. This is consistent with Fama <strong>and</strong> French (2007), who find that the<br />
issuance effect is present across both large <strong>and</strong> small stocks in U.S. markets.<br />
A comparison of our results in to those of Pontiff <strong>and</strong> Woodgate (2006) reveals that<br />
when estimated with a continuous issuance measure the economic magnitude of the annual<br />
issuance effect is stronger in the U.S. than in international markets. For example, In Panel A<br />
our equal-weighted one-month holding period regression yields an ISSUE coefficient<br />
estimate of -0.79, implying that a one st<strong>and</strong>ard deviation increase (0.21) in annual issuance<br />
leads to a 0.16% decline in the cross section of monthly international returns. From the same<br />
4 In an earlier version of this paper we also used returns measured over six months, second year, <strong>and</strong> third year.<br />
The results over these horizons are similar to the results <strong>for</strong> the one month <strong>and</strong> one year horizons, so <strong>for</strong> the<br />
sake of brevity we do not report them.<br />
10
egression that includes only the ISSUE variable, Pontiff <strong>and</strong> Woodgate (2006) find its slope<br />
to be -2.23; this indicates that a st<strong>and</strong>ard deviation increase (0.15) in annual issuance is<br />
associated with a 0.33% decline in the subsequent month’s cross-sectional returns in the U.S.<br />
A similar comparison made <strong>for</strong> the 1-year holding period also shows that the continuous<br />
annual issuance measure exhibits a stronger economic effect on future returns in the U.S.<br />
than in non-U.S. countries.<br />
2.4. The Impact of Size, Book-to-Market, <strong>and</strong> Momentum on the Share Issuance Effect<br />
To test the robustness of the international issuance effect, we include size (ME),<br />
book-to-market (BM), <strong>and</strong> momentum (MOM) as control variables in the regressions. We<br />
report these results in Panels C <strong>and</strong> D of Table 4. Controlling <strong>for</strong> these effects is important<br />
since they exist in international markets (see Fama <strong>and</strong> French (1998), Rouwenhorst (1998,<br />
1999), <strong>and</strong> Griffin, Ji, <strong>and</strong> Martin (2003)).<br />
Overall, we observe that the effects of ISSUE are robust to the inclusion of other firm-<br />
specific variables. The issuance measure has significant negative coefficients in all twelve<br />
regressions reported in Panels C <strong>and</strong> D, <strong>and</strong> the effect remains similar in both economic<br />
magnitude <strong>and</strong> statistical significance as compared to the results in Panels A <strong>and</strong> B. These<br />
results are similar to those reported in Pontiff <strong>and</strong> Woodgate (2006) <strong>and</strong> Fama <strong>and</strong> French<br />
(2007), who find that the U.S. issuance effect remains pervasive after controlling <strong>for</strong> other<br />
firm characteristics known to affect future returns.<br />
The results in Panels C <strong>and</strong> D show that the statistical significance of issuance<br />
compares favourably to those of other firm-specific variables. The t-statistics <strong>for</strong> the ISSUE<br />
slopes are considerably larger than those <strong>for</strong> ME <strong>and</strong> comparable to those <strong>for</strong> BM throughout<br />
11
the regressions. They are also greater than those <strong>for</strong> MOM in both the value- <strong>and</strong> scaled-<br />
weighted regressions.<br />
2.6. Discussion<br />
Based on our analysis, we offer several additional observations <strong>and</strong> interpretations of<br />
our results. First, we observe large differences in average R 2 statistics resulting from the same<br />
regression specification estimated with <strong>and</strong> without country dummies. For example, the<br />
average R 2 statistics are mostly 20% <strong>and</strong> above in Panel A of Table 4 (with country<br />
dummies), whereas they are at most 0.30% in Panel B of the same table (without country<br />
dummies). The R 2 statistic tells us how much stock-return variation is associated with the<br />
variation in the explanatory variables. There<strong>for</strong>e, this difference in average R 2 suggests that<br />
country effects alone can explain about 20% of the cross-sectional variation in international<br />
stock returns.<br />
A comparison of the results from the same regression that differs only in terms of<br />
country-dummy usage also shows that the t-statistics on the ISSUE coefficients are larger<br />
when the dummies are used. This finding indicates that the return predictive ability of share<br />
issuance tends to be more robust within a country than across countries.<br />
Annual share issuance alone explains very little of the cross-sectional variation in<br />
international stock returns. Across all holding periods, the highest average R 2 is only 0.30%,<br />
<strong>and</strong> is lower in four out of the five holding periods examined by Pontiff <strong>and</strong> Woodgate<br />
(2006). Pontiff <strong>and</strong> Woodgate show that in the U.S., size, book-to-market, <strong>and</strong> momentum<br />
explain 2-8 times the cross-sectional return variation explained by share issuance. Since the<br />
ability of annual issuance to explain cross-sectional return variation is even smaller in<br />
12
international markets than in the U.S., it is doubtful that share issuance proxies <strong>for</strong> a risk<br />
factor in the same vein as book-to-market <strong>and</strong> size in the U.S. Share issuance may still reflect<br />
rational asset pricing, although its contribution to expected returns is likely to be through a<br />
non-risk based avenue, such as providing compensation <strong>for</strong> taxes or transaction costs.<br />
The relation between book-to-market <strong>and</strong> issuance runs counter to market-timing<br />
literature that uses book-to-market as a proxy <strong>for</strong> mispricing (Lakonishok, Shleifer, <strong>and</strong><br />
Vishny (1994)). Counter to this literature, Table 2 shows that high book-to-market firms<br />
tend to issue more stock.<br />
We find that firms are more likely to issue shares after having had high past returns,<br />
<strong>and</strong> that this result is stronger <strong>for</strong> small firms. This finding is consistent with both the U.S.<br />
evidence <strong>and</strong> with market timing theories. However, high past returns could be the result of<br />
improved investment opportunities, <strong>and</strong> smaller firms may be more dependent on equity<br />
financing than large firms. There<strong>for</strong>e, capital raising <strong>for</strong> legitimate investment opportunities<br />
could also explain the observed relation between past returns <strong>and</strong> share issuance.<br />
3. Effects of Positive <strong>and</strong> Negative Share Issuances<br />
Previous studies have investigated long-run returns following specific events that<br />
affect shares outst<strong>and</strong>ing, such as seasoned-equity offerings (Loughran <strong>and</strong> Ritter (1995))<br />
<strong>and</strong> stock-financed mergers (Loughran <strong>and</strong> Vigh (1997)). Fama <strong>and</strong> French (2007) compare<br />
issuance-return predictability based on the whether the firm is an issuer or repurchaser, <strong>and</strong><br />
the magnitude of the respective activity. Motivated by the results in these studies, we test<br />
whether positive <strong>and</strong> negative issuances can separately predict returns across non-U.S. firms.<br />
Our investigation varies from the repurchase literature in that our issuance variable measures<br />
13
actual issuance activity, whereas the repurchase literature uses announcements [<strong>for</strong> example,<br />
Ikenberry, Lakonishok, <strong>and</strong> Vermaelen (1995), Ikenberry, Lakonishok, <strong>and</strong> Vermaelen<br />
(2000), <strong>and</strong> Rau <strong>and</strong> Vermaelen (2002)].<br />
3.1. Methodology<br />
In order to compare the U.S. <strong>and</strong> non-U.S. evidence, we follow Fama <strong>and</strong> French<br />
(2007) <strong>and</strong> sort stocks every month into eight groups based on the level of issuance. Like<br />
Fama <strong>and</strong> French (2007), we use all but the smallest quintile stocks in our sample to<br />
determine the portfolio breakpoints, although the smallest stocks are also included in the<br />
portfolios. POS1 through POS5 are the quintiles of firms with positive share issuances, with<br />
POS5 consisting of the largest issuers. NEG1 <strong>and</strong> NEG2 are the portfolios of firms with<br />
negative issuance, with NEG1 containing the largest net share repurchasers, with ISSUE<br />
values below that month’s median negative value. Firms with zero issuance are included in<br />
the portfolio, ZERO.<br />
There is some variation in the laws concerning buybacks across the countries in our<br />
sample (see Table 7), <strong>and</strong> many of the countries in our sample had changes in their laws<br />
regarding buybacks during our sample period. The law changes either made buybacks legal<br />
<strong>for</strong> the first time, or made buybacks more feasible from either a tax or regulatory perspective.<br />
To test whether buyback restrictions affect the issuance <strong>and</strong> repurchase effects, we exclude<br />
firm-month observations <strong>for</strong> which buybacks were either illegal or infeasible, <strong>and</strong> repeat our<br />
analysis.<br />
3.2. Summary Statistics <strong>for</strong> Positive <strong>and</strong> Negative Issuance Portfolios<br />
14
Summary statistics <strong>for</strong> the portfolios are presented in Table 8. Panel A reports the<br />
statistics <strong>for</strong> the entire sample. The table shows that in an average month 58.05% of the firms<br />
in our sample have ISSUE values of zero, while about 6% have negative share issuances <strong>and</strong><br />
the remaining 36% have positive share issuances. The smallest three positive-issuance<br />
portfolios have negligible values of ISSUE; they are 0.1%, 0.9%, <strong>and</strong> 3.8%. This is similar to<br />
the findings in Fama <strong>and</strong> French (2007) who report values of 0.14%, 0.57%, <strong>and</strong> 1.48% <strong>for</strong><br />
the same quintiles with U.S. stocks. 5<br />
Compared to the U.S., the biggest issuers <strong>and</strong> the biggest net repurchasers in the<br />
world tend to have more extreme changes in their shares outst<strong>and</strong>ing. For the highest quintile<br />
of positive issuers we find a mean issuance of 46.2%, compared to Fama <strong>and</strong> French’s (2007)<br />
finding in the U.S. of 24.04%. For the lower half of repurchasers we find a mean value of<br />
issuance of -13.3%, compared with -5.73% from Fama <strong>and</strong> French’s (2007) study.<br />
Panel B displays the results <strong>for</strong> the sample that excludes firm-month observations that<br />
have buyback restrictions. We expect the absolute values of the NEG portfolios to increase<br />
when buybacks become either legal or more attractive, <strong>and</strong> the results in Panel B relative to<br />
those in Panel A show this. The results in Panel B also show that the value of POS increases<br />
when buybacks are allowed. The POS4 <strong>and</strong> POS5 portfolios increase from 12.00 <strong>and</strong> 46.20<br />
in Panel A, to 13.98 <strong>and</strong> 50.23 in Panel B. One reason <strong>for</strong> this increase could be market<br />
timing, in that firms are more willing to issue overvalued shares if they can buy the shares<br />
back if their equity becomes undervalued.<br />
5 Buybacks in the U.S. were restricted prior to 1982, so Fama <strong>and</strong> French’s sample, which begins in 1963, also<br />
includes observations <strong>for</strong> which buybacks were restricted.<br />
15
3.3. Positive <strong>and</strong> Negative Share Issuances <strong>and</strong> Subsequent Returns<br />
In Table 9 we again test whether share issuance can predict returns, only now we<br />
replace the ISSUE measure with dummy variables that indicate which of the seven share<br />
issuance portfolios that a firm belongs to. Panel A reports the results <strong>for</strong> the full sample.<br />
NEG1 is significant only with equal-weightings, <strong>and</strong> NEG2 is not significant with any of the<br />
weightings. This shows that the international issuance effect is weak, <strong>and</strong> limited to smaller<br />
firms. This is different than the results that Fama <strong>and</strong> French (2007) obtain with U.S. stocks;<br />
they find that negative issuance portfolios have positive <strong>and</strong> statistically significant abnormal<br />
returns in both large <strong>and</strong> small stocks, showing that the buyback effect is weaker<br />
internationally than it is in the U.S.<br />
At the one-month horizon the POS4, <strong>and</strong> POS5 coefficients are negative <strong>and</strong><br />
statistically significant under each of the three weighting schemes. This result is stronger than<br />
that reported by Fama <strong>and</strong> French (2007) <strong>for</strong> U.S. stocks; they find that only the fifth quintile<br />
of issuers has negative abnormal returns. The POS5 slope is -0.44% <strong>and</strong> its t-statistic is -6.00,<br />
suggesting that if a firm is in POS5 its expected returns over the subsequent month are lower<br />
by 0.44%. Fama <strong>and</strong> French (2007) find that this group of firms in the U.S. underper<strong>for</strong>ms by<br />
0.27% per month. The POS4 slope is -0.28% <strong>and</strong> its t-statistic is -4.57. Fama <strong>and</strong> French<br />
(2007) find that this group of firms in the U.S. has a positive abnormal return of 0.07%.<br />
Taken together, these results show that the impact of positive issuances on future returns is<br />
both more pervasive <strong>and</strong> of greater economic magnitude internationally than in the U.S.<br />
Overall the results in Panel A show that the positive issuance effect found in the U.S.<br />
is not sample specific, <strong>and</strong> exists in a similar fashion across non-U.S. firms. The results also<br />
show that there are both similarities <strong>and</strong> differences between the U.S. <strong>and</strong> non-U.S. buyback<br />
16
effects. In the U.S., the buyback effect exists in both small <strong>and</strong> large firms, whereas<br />
internationally the effect is found primarily among small firms.<br />
Panel B restricts the sample to observations <strong>for</strong> which buybacks were feasible.<br />
Surprisingly, the NEG coefficients are not positive <strong>and</strong> significant in any of the regressions.<br />
However the POS4 <strong>and</strong> POS5 coefficients in Panel B are noticeable larger, as are their t-<br />
statistics, when compared to those in Panel A. This shows that when firms can buy their<br />
shares back, the post-issuance returns are worse following positive issuances. This again<br />
could be consistent with market timing, as firms may be more willing to issue overvalued<br />
shares if they know that they can buy those shares back if their equity becomes undervalued<br />
in the future.<br />
4. Determinants of Cross-Country Differences in the Share Issuance Effect<br />
In this Section we test whether the share issuance effect varies across countries, <strong>and</strong> if<br />
so what type of factors can explain cross-country differences in the effect.<br />
4.1. Hypothesis Development<br />
The issuance effect might be stronger in countries in which it is easier <strong>for</strong> firms to<br />
issue <strong>and</strong> buyback their own shares. To underst<strong>and</strong> why, assume that the benefits from share<br />
issuance <strong>and</strong> repurchase activities include both taking advantage of mispricing <strong>and</strong> target<br />
capital structure adjustments. However there are costs involved with issuing <strong>and</strong><br />
repurchasing shares <strong>and</strong> rational managers will only engage in capital structure changes if the<br />
total benefit is larger than the cost. In this framework if the cost of transacting in one’s own<br />
stock decreases, then we would expect firms to transact in their own stock more frequently,<br />
17
<strong>and</strong> we would also expect these transactions to convey more in<strong>for</strong>mation about cross-<br />
sectional mispricing.<br />
The findings in Pontiff <strong>and</strong> Woodgate (2006) support this hypothesis. Pontiff <strong>and</strong><br />
Woodgate (2006) find that in the U.S. pre-1970, 37.4% of firms have nonzero annual<br />
issuance, <strong>and</strong> that during this time period the relation between issuance <strong>and</strong> future stock<br />
returns is insignificant <strong>and</strong> often positive. Post-1970, Pontiff <strong>and</strong> Woodgate (2006) find a<br />
negative <strong>and</strong> statistically significant relation between issuance <strong>and</strong> future returns. During this<br />
period issuance <strong>and</strong> repurchase activity are more frequent, with 75.8% of firms having<br />
nonzero annual issuance.<br />
Like Pontiff <strong>and</strong> Woodgate, we use the percentage of firm-month observations with<br />
nonzero ISSUE values as a proxy <strong>for</strong> the ease with which firms can issue <strong>and</strong> repurchase their<br />
shares. There is significant variation in the issuance activity among the countries in our<br />
sample (see Table 7), <strong>and</strong> we test whether the issuance effect is greater in countries where<br />
issuance activity is greater.<br />
Share issuance should be less costly in countries with more developed stock markets,<br />
so we also use several stock market development measures as proxies <strong>for</strong> the cost of issuing<br />
shares. La Porta et al. (2006) introduce several measures of stock market development <strong>and</strong><br />
we use their measures.<br />
We also test whether the issuance effect is related to investor protection laws. One<br />
might expect that the issuance effect will be weaker in countries with stronger investor<br />
protection laws. To see why, consider an environment in which some investors are irrational,<br />
so mispricing can arise. However there are arbitragers that can profit from this mispricing,<br />
thereby keeping markets efficient. Further assume that investor protection laws are effective.<br />
18
These laws reduce in<strong>for</strong>mation asymmetries between issuers <strong>and</strong> investors, <strong>and</strong> prevent<br />
managers from expropriating the firm’s resources. In a state with strong protection,<br />
arbitrageurs will have similar in<strong>for</strong>mation as managers. Because they do not fear<br />
expropriation, arbitrageurs will trade on their in<strong>for</strong>mation keep prices close to fundamental<br />
values [see Morck, Yeung, <strong>and</strong> Yeu (2000)]. In states with weak investor protection, the<br />
in<strong>for</strong>mation asymmetries between managers <strong>and</strong> arbitrageurs will be greater, <strong>and</strong> arbitrageurs<br />
are less willing to trade due to a lack of protection. This should result in larger mispricing in<br />
low protection countries, <strong>and</strong> stronger share issuance effects.<br />
4.2. Variable Descriptions<br />
To test these hypotheses we use several different measures of stock market<br />
development <strong>and</strong> investor protection that have been used in previous studies. Each of the<br />
variables is displayed in Table 7. We include U.S. firms in this part of our analysis.<br />
4.2.1. Measures of Firm-Level Issuance, Issuance Activity, <strong>and</strong> Stock Market Development<br />
Country-Level Issuance Coefficients. We estimate monthly <strong>and</strong> yearly issuance<br />
coefficients <strong>for</strong> each country, using monthly returns, equal weights, <strong>and</strong> controls <strong>for</strong> size,<br />
book-to-market, <strong>and</strong> momentum. Share issuance activity is infrequent in many smaller<br />
countries, with many firms having values of zero <strong>for</strong> ISSUE. There<strong>for</strong>e we only estimate the<br />
coefficient in countries which have at least 100 firms in each cross-section. We display the<br />
monthly coefficients in Panel A of Table 7 <strong>and</strong> the yearly coefficient in Panel B of Table 7.<br />
Percentage with Nonzero Issuance. Each month we measure the percentage of firms<br />
in each country with nonzero values of ISSUE. Countries with a higher value of this measure<br />
19
contain firms that more actively issue <strong>and</strong> repurchase their own shares, suggesting that share<br />
issuance activity is not costly.<br />
Liquidity is the total dollar value of stocks traded scaled by GDP, <strong>for</strong> the period 1996-<br />
2000. This measure is used by La Porta et al. (2006) as a proxy <strong>for</strong> the level of stock market<br />
development. 6<br />
Turnover is the total dollar value of stocks traded, scaled by the value of shares<br />
outst<strong>and</strong>ing, <strong>for</strong> the period 1996-2000. This measure is used by La Porta et al. (2006) as a<br />
measure of stock market development.<br />
Log GDP per capita is per capita GDP in U.S. dollars in 2000. La Porta et al. (2006)<br />
show that countries with higher per capita GDP have more developed stock markets.<br />
Buybacks. We create a dummy variable which is equal to 1 if buybacks are feasible,<br />
<strong>and</strong> zero if they are not feasible. We define buybacks as not feasible if they are illegal, or if<br />
tax or other regulatory issues make them highly unattractive. As an example, we define<br />
Australia as feasible post-1995. Buybacks became legal in Australia in 1989, but regulatory<br />
issues made them infeasible (<strong>and</strong> highly infrequent) until rule changes in 1995. We obtained<br />
our buyback data from various sources, with Kim, Schremper, <strong>and</strong> Varaiya (2004) providing<br />
data <strong>for</strong> the ten largest stock markets.<br />
Short Selling. We create a dummy variable which is equal to 1 if short selling is not<br />
allowed <strong>and</strong> zero if it is allowed. We obtained this measure from Bris et al. (2007). If short<br />
selling was legal prior to 1990, then Bris et al. report “Be<strong>for</strong>e 1990” as the effective date. For<br />
these countries, we assume that short selling was allowed in each of the years prior to 1990.<br />
6<br />
The measures from La Porta et al. (1999 <strong>and</strong> 2006) were obtained from Andrei Shleifer’s website. We thank<br />
Andrei Shleifer <strong>for</strong> making the data available.<br />
20
4.2.2. Measures of Investor Protection<br />
Law. We create a dummy variable which is is equal 1 if a country is of common law<br />
origin <strong>and</strong> zero if the country is of civil law origin. This measure is from La Porta et al.<br />
(1997). La Porta et al. show that common law countries tend to have stronger investor<br />
protection laws than civil law countries.<br />
Accounting is an index of accounting st<strong>and</strong>ards; a higher value represents better<br />
accounting st<strong>and</strong>ards. The index is based on the reporting or omission of 90 items from<br />
annual reports. A higher value signals better accounting st<strong>and</strong>ards. This measure is from La<br />
Porta et al. (1997).<br />
Criminal is an index which measures the ease with which accountants, directors, <strong>and</strong><br />
distributors can be pursued in criminal courts <strong>for</strong> not fully disclosing in<strong>for</strong>mation in<br />
prospectuses. A higher value signals greater investor protection. This variable is from La<br />
Porta et al. (2006).<br />
Liability is an index which measures the ease with which accountants, directors, <strong>and</strong><br />
distributors can be pursued in civil courts <strong>for</strong> not fully disclosing in<strong>for</strong>mation in<br />
prospectuses. A higher value signals greater investor protection. This variable is from La<br />
Porta et al. (2006). La Porta et al. show that countries with higher values of Liability have<br />
more developed equity markets.<br />
Protect is the principle component of three different investor protection indices. The<br />
indices are Liability, an index of disclosure requirements, <strong>and</strong> an index of anti-director rights.<br />
A higher value of Protect signals greater investor protection. This variable is from La Porta<br />
21
et al. (2006), who show that countries with higher values of Protect have more developed<br />
equity markets.<br />
Earnings Management Score (EMS) is from Leuz et al. (2003). EMS is the average<br />
rank of four different earnings management measures. A higher value of the EMS suggests<br />
poorer earnings quality.<br />
4.2.3 Correlations Among Variables<br />
Panel C of Table 7 reports the correlations among the variables. We focus our<br />
discussion on the correlations between the 1-month ISSUE coefficient <strong>and</strong> the other<br />
variables; the 1-year ISSUE coefficient’s correlations are similar. The 1-month ISSUE<br />
coefficient is negatively correlated to the frequency of nonzero issuance; hence countries<br />
with more issuance activity have stronger issuance effects. The market development<br />
indicators Liquidity, Turnover, <strong>and</strong> Log GDP are each negatively correlated to the 1-month<br />
ISSUE coefficient as well, suggesting that the issuance effect is stronger in more developed<br />
markets. The remaining correlations suggest that the issuance effect is stronger in countries<br />
with stronger investor protection, <strong>and</strong> better earnings quality. Taken in their entirety, the<br />
results suggest that the issuance effect is stronger in countries in which it is easier <strong>for</strong> firms to<br />
issue <strong>and</strong> buyback their shares. We test this hypothesis more rigorously with regressions<br />
which are reported in Table 8.<br />
4.3. Interaction Regressions<br />
4.3.1. Methodology<br />
22
Our objective is to test whether country characteristics can explain differences in the<br />
issuance effect across countries. To maximize the power of our tests, we return to our pooled<br />
sample <strong>and</strong> regress monthly stock returns on ISSUE along with interaction variables which<br />
are the product of ISSUE <strong>and</strong> each of the country characteristics described in Table 7. The<br />
interaction coefficients measure to what extent the issuance effect differs across countries<br />
with different levels of issuance activity, stock market development, <strong>and</strong> investor protection.<br />
4.2.1. Issuance Activity, Market Development, Buyback, <strong>and</strong> Short Selling<br />
Interactions<br />
Panels A <strong>and</strong> B of Table 8 show that the issuance effect is stronger in countries in<br />
which firms more actively trade in their own shares, in countries with more developed equity<br />
markets, <strong>and</strong> in countries where buybacks are allowed. Panel A reports results <strong>for</strong> one-month<br />
holding period return regressions, while Panel B reports the one-year holding period results.<br />
We first describe the results in Panel A.<br />
In regression 1 ISSUE is interacted with the percentage of observations within the<br />
country that have a non-zero value of ISSUE. The interaction term is negative <strong>and</strong> highly<br />
significant, with a t-statistic of -5.24., showing that the issuance effect is stronger in countries<br />
with more issuance activity.<br />
Regressions two, three, <strong>and</strong> four report the market development interactions. All three<br />
of the interaction terms are negative; the t-statistics range from -3.70 to -5.07, showing that<br />
the issuance effect is stronger in markets that are more developed.<br />
The buyback interaction is also negative <strong>and</strong> significant. Recall that Buyback is equal<br />
to one if buybacks are permitted <strong>and</strong> zero otherwise. The interaction term is -0.58, while the<br />
23
ISSUE coefficient is -0.70. Taken together, the coefficients show that the issuance coefficient<br />
is -0.58 in countries that do not allow buybacks, <strong>and</strong> -1.28 in countries that do. The t-statistic<br />
<strong>for</strong> the buyback coefficient is -1.87, which is much smaller than the development <strong>and</strong><br />
nonzero interactions, suggesting that buyback regulations are not as important as issuance<br />
activity <strong>and</strong> equity market development with respect to explaining the issuance effect.<br />
The final regression in Panel A is an interaction with a short sale prohibition dummy.<br />
If the low returns following share issuances are the result of overvaluation, then the effect<br />
might be stronger in countries that prohibit short selling. However this is not what we find, as<br />
the interaction term is positive <strong>and</strong> not significant. Short selling tends to be allowed in<br />
developed markets more than in non-developed markets; this might explain why the<br />
coefficient has a positive sign.<br />
The results in Panel B are very a similar to those in Panel A. The issuance activity<br />
<strong>and</strong> market development interactions are all highly significant, showing that the issuance<br />
effect is strongest in countries where it is easier to issue shares. The buyback dummy is<br />
marginally significant, showing that the effect is stronger in countries where buybacks are<br />
feasible, while the short selling interaction is not significant.<br />
Taken in their entirety the results show that issuance effect is strongest in countries<br />
with the greatest amount of issuance activity, <strong>and</strong> with the most developed capital markets.<br />
These findings suggest that the effect is greatest in places where it is easier <strong>for</strong> firms to issue<br />
<strong>and</strong> repurchase shares.<br />
4.3.2. Investor Protection <strong>and</strong> Earnings Management Interactions<br />
24
Panels C <strong>and</strong> D of Table 8 show that the issuance effect is stronger in countries with<br />
better investor protection laws, <strong>and</strong> in countries with better earnings quality. We first<br />
describe the results in Panel C, which reports the monthly holding period return regression<br />
results. In the first regression, the interaction term is a dummy variable equal to one if the<br />
country is a common law country. The interaction term -0.96 (t-statistic = -5.58) <strong>and</strong> the<br />
ISSUE coefficient is -0.43 (t-statistic = -3.56). Hence in a civil law country, the average<br />
ISSUE coefficient is -0.43, while in common law countries it is -1.39.<br />
Regressions 2, 3, <strong>and</strong> 6 rein<strong>for</strong>ce the notion that the issuance effect is stronger in<br />
countries which offer greater investor protection. The coefficients are negative <strong>and</strong><br />
significant <strong>for</strong> the Accounting, Liability, <strong>and</strong> Protect interaction terms. The t-statistics range<br />
from -3.96 to -4.64. In regression 2 the Criminal interaction term is positive, but not<br />
significant.<br />
Regression 6 shows that the issuance effect is stronger in countries with better<br />
earnings quality, <strong>and</strong> weaker in countries with poor earnings quality. Leuz et al. (2003) show<br />
that earnings quality is better in countries with that offer better investor protection, so the<br />
results here again show that the issuance effect is stronger in countries with stronger investor<br />
protection laws.<br />
Panel D of Table 8 reports the regression results with the 1-year holding period. The<br />
signs <strong>and</strong> significance <strong>for</strong> each of the interaction terms are the same as in Panel C, showing<br />
that the issuance effect is stronger in countries that offer better investor protection <strong>and</strong> in<br />
countries that have better earnings quality.<br />
The results in Table 8 are not consistent with the hypothesis that markets with greater<br />
investor protection are more efficient than markets that offer less protection. La Porta et al.<br />
25
(2006) show that better governed countries have more developed equity markets, <strong>and</strong> provide<br />
firms with easier access to equity capital. This would suggest that it is easier to <strong>for</strong> firms to<br />
issue <strong>and</strong> repurchase mispriced shares in markets that offer better investor protection, so the<br />
results here could be consistent with market timing.<br />
5. Conclusion<br />
In this paper we take a first look at the return predictive ability of firm-level share<br />
issuance in international markets. We study the issuance effect across a large sample of non-<br />
U.S. firms, drawn from 41 different countries over a 25 year period. We find robust evidence<br />
of a share issuance effect in our sample. The predictive ability of annual issuance is generally<br />
more statistically significant than that of either size or momentum, <strong>and</strong> is similar to that of<br />
book-to-market.<br />
The international issuance effect is especially strong <strong>for</strong> positive share issuances. A<br />
comparison of our results to those in Fama <strong>and</strong> French (2007) shows that positive share<br />
issuance events have a stronger effect on subsequent returns internationally than they do in<br />
the U.S. Like in the U.S., international negative issuances also predict high returns, but the<br />
international buyback effect is weaker than that from positive issuances, <strong>and</strong> weaker than the<br />
buyback effect found in the U.S.<br />
The predictive ability of share issuance provides challenges <strong>for</strong> rational asset pricing<br />
models as well as behavioural models. We find that cross-sectional average R 2 s from<br />
regressions of returns on issuance are very low, casting doubt on a rational risk-based link,<br />
although we can not rule out the possibility that issuance proxies <strong>for</strong> a rational non-risk based<br />
factor, such as taxes or transaction costs.<br />
26
The findings suggest that the share issuance effect is related to the ease with which<br />
firms can issue <strong>and</strong> buyback their shares. The issuance effect is stronger in countries with in<br />
which there is more frequent issuance activity, more developed stock markets, <strong>and</strong> stronger<br />
investor protection laws.<br />
27
References<br />
Baker, Malcolm, <strong>and</strong> Jeffrey Wurgler, 2002, Market timing <strong>and</strong> capital structure, Journal of<br />
Finance 57, 1-32.<br />
Bris, Arturo, William Goetzman, <strong>and</strong> Ning Zhu, 2007, Efficiency <strong>and</strong> the bear: short sales<br />
<strong>and</strong> markets around the world, Journal of Finance 62, 1029-1079.<br />
Daniel, Kent, <strong>and</strong> Sheridan Titman, 2006, Market reaction to tangible <strong>and</strong> intangible<br />
in<strong>for</strong>mation, Journal of Finance 61, 1605-1643.<br />
Fama, Eugene F., <strong>and</strong> Kenneth R. French, 1992, The cross-section of expected stock returns,<br />
Journal of Finance 47, 427-465.<br />
Fama, Eugene F., <strong>and</strong> Kenneth R. French, 1998, Value versus growth: The international<br />
evidence, Journal of Finance 53, 1975-1999.<br />
Fama, Eugene F., <strong>and</strong> Kenneth R. French, 2007, Dissecting anomalies, <strong>for</strong>thcoming in the<br />
Journal of Finance.<br />
Fama, Eugene F., <strong>and</strong> James D. MacBeth, 1973, Risk, return, <strong>and</strong> equilibrium: Empirical<br />
tests, Journal of Political Economy 71, 607-636.<br />
Griffin, John, Xiuqing, Ji, <strong>and</strong> Spencer Martin, 2003, Momentum investing <strong>and</strong> business<br />
cycle risk: Evidence from pole to pole, Journal of Finance 58, 2515-5247.<br />
Henderson, Brian J., Narasimhan Jegadeesh, <strong>and</strong> Michael S. Weisbach, 2006, World markets<br />
<strong>for</strong> raising new capital, Journal of Financial Economics 82, 63-101.<br />
Ikenberry, David, Josef Lakonishok, <strong>and</strong> Theo Vermaelen, 1995, Market underreaction to<br />
open market share repurchases, Journal of Financial Economics 39, 181-208.<br />
Ikenberry, David, Josef Lakonishok, <strong>and</strong> Theo Vermaelen, 2000, Stock repurchases in<br />
Canada: Per<strong>for</strong>mance <strong>and</strong> strategic Liquidity, Journal of Finance 55, 181-208.<br />
Ince, Ozgur S., <strong>and</strong> R. Burt Porter, 2006, Individual equity return data from Thomson<br />
Datastream: h<strong>and</strong>le with care!, Journal of Financial Research 29, 463-479.<br />
Jegadeesh, Narasimhan, <strong>and</strong> Sheridan Titman, 1993, Returns to buying winners <strong>and</strong> selling<br />
losers: Implications <strong>for</strong> stock market efficiency, Journal of Finance 48, 65–91.<br />
Jegadeesh, Narasimhan, <strong>and</strong> Sheridan Titman, 2001, Profitability of momentum strategies:<br />
An evaluation of alternative explanations, Journal of Finance 56, 699-720.<br />
Lakonishok, Josef, Andrei Shleifer, <strong>and</strong> Robert W. Vishny, 1994, Contrarian investment,<br />
extrapolation, <strong>and</strong> risk, Journal of Finance 49, 1541-1578.<br />
28
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, <strong>and</strong> Robert Vishny, 1997,<br />
Law <strong>and</strong> Finance, Journal of Political Economy 103, 1113-1155.<br />
La Porta, Rafael, Florencio Lopez-de-Silanes, <strong>and</strong> Andrei Shleifer, 2006, What works in<br />
securities laws?, Journal of Finance 61, 1-32.<br />
Leuz, Christian, Dhananjay N<strong>and</strong>a, <strong>and</strong> Peter D. Wysocki, Earnings <strong>and</strong> investor protection:<br />
an international comparison, Journal of Financial Economics, 69, 505-527.<br />
Loughran, Tim, <strong>and</strong> Jay Ritter, 1995, The new issues puzzle, Journal of Finance 50, 23-51.<br />
Loughran, Tim, <strong>and</strong> Arn<strong>and</strong> Vijh 1997, Do long-term shareholders benefit from corporate<br />
acquisitions? Journal of Finance 52, 1765-1790.<br />
Morck, R<strong>and</strong>all, Bernard Yeung, <strong>and</strong> Wayne Yu, 2000, The in<strong>for</strong>mation content of stock<br />
markets: Why do emerging markets have synchronous stock price movements? Journal of<br />
Financial Economics 58, 215-260.<br />
Pontiff, Jeffrey, 1996, Costly arbitrage <strong>and</strong> closed-end fund discounts, Quarterly Journal of<br />
Economics 111, 1135-1151.<br />
Pontiff, Jeffrey, <strong>and</strong> Artemiza Woodgate, 2006, Share issuance <strong>and</strong> cross-sectional returns,<br />
<strong>for</strong>thcoming in the Journal of Finance.<br />
Rau, Raghavendra P., <strong>and</strong> Theo Vermaelen, 2002, Regulation, taxes, <strong>and</strong> share repurchases<br />
in the United Kingdom, Journal of Business 75, 245-282.<br />
Rouwenhorst, K. Geert, 1998, International momentum strategies, Journal of Finance 53,<br />
267-284.<br />
Rouwenhorst, K. Geert, 1999, Local return factors <strong>and</strong> turnover in emerging markets,<br />
Journal of Finance 54, 1439-1464.<br />
29
Table 1: Summary Statistics <strong>for</strong> Sample Countries<br />
This table provides summary statistics <strong>for</strong> the 41 countries included in our sample. Columns 2 <strong>and</strong> 3<br />
list the beginning <strong>and</strong> ending dates during which each country is included in our regression analysis.<br />
The table additionally reports the total number of firm-month observations (column 4) <strong>and</strong> the average<br />
number of firm observations per month (column 6). The values of these statistics represented as<br />
percentages of the corresponding total across countries are given in the remaining columns 5, <strong>and</strong> 7.<br />
The average monthly total market capitalization in millions of U.S. dollars is given in column 8. The<br />
percentage of firms with non-zero one-year change in the number of shares outst<strong>and</strong>ing adjusted <strong>for</strong><br />
distribution events such as stock splits <strong>and</strong> stock dividends (ISSUE) is displayed in column 9. ISSUE<br />
is computed over month t-12 to t: ISSUEt,t-12 = Ln(Adjusted Sharest) – Ln(Adjusted Sharest-12).<br />
30
Country Start Date End Date<br />
Total Number<br />
of Firm-<br />
Month<br />
Observations<br />
Percentage<br />
of Total<br />
Sample (%)<br />
31<br />
Average<br />
Number of<br />
Firm<br />
Observations<br />
per Month<br />
Monthly<br />
Average<br />
Percentage of<br />
Sample (%)<br />
Average<br />
Monthly Total<br />
Market Value<br />
Average<br />
Monthly<br />
Percentage of<br />
Total Market<br />
Value (%)<br />
Argentina 03/1995 06/2006 8,511 0.30 63 0.53 24,942 0.45<br />
Australia 07/1981 06/2006 138,645 4.92 462 3.89 96,456 1.75<br />
Austria 01/1991 06/2006 9,613 0.34 75 0.63 22,865 0.42<br />
Belgium 10/1985 06/2006 28,052 0.99 128 1.07 55,312 1.01<br />
Brazil 03/1995 06/2006 58,402 2.07 429 3.62 137,145 2.49<br />
Canada 07/1981 06/2006 218,867 7.76 730 6.14 223,245 4.06<br />
Chile 02/1991 06/2006 24,855 0.88 153 1.29 44,459 0.81<br />
China 08/1994 06/2006 90,180 3.20 631 5.31 212,619 3.87<br />
Czech Republic 07/1996 11/2001 3,768 0.13 77 0.65 6,151 0.11<br />
Denmark 01/1989 06/2006 36,657 1.30 183 1.54 49,618 0.90<br />
Egypt 01/2001 06/2006 4,479 0.16 76 0.64 9,233 0.17<br />
Finl<strong>and</strong> 07/1990 06/2006 16,719 0.59 96 0.80 41,519 0.75<br />
France 07/1981 06/2006 132,542 4.70 443 3.73 331,623 6.03<br />
Germany 07/1981 06/2006 112,621 3.99 377 3.17 300,368 5.46<br />
Greece 08/1989 06/2006 36,381 1.29 179 1.51 30,446 0.55<br />
Hong Kong 09/1981 06/2006 94,629 3.36 325 2.74 123,696 2.25<br />
India 08/1991 06/2006 124,086 4.40 693 5.84 73,522 1.34<br />
Indonesia 04/1992 05/2006 12,551 0.45 113 0.95 22,908 0.42<br />
Irel<strong>and</strong> 10/1996 12/2001 864 0.03 51 0.43 39,212 0.71<br />
Italy 07/1981 06/2006 60,813 2.16 203 1.71 165,867 3.02<br />
Japan 07/1981 06/2006 415,165 14.72 1,384 11.65 1,625,890 29.56<br />
Malaysia 08/1987 06/2006 88,985 3.16 392 3.30 73,995 1.35<br />
Mexico 01/1990 06/2006 22,699 0.80 117 0.99 72,992 1.33<br />
Netherl<strong>and</strong>s 07/1981 06/2006 44,442 1.58 149 1.25 167,715 3.05<br />
New Zeal<strong>and</strong> 08/1989 06/2006 14,765 0.52 73 0.61 11,541 0.21<br />
Norway 08/1983 06/2006 26,847 0.95 104 0.87 25,239 0.46<br />
Pakistan 02/1994 06/2006 27,574 0.98 185 1.56 6,459 0.12<br />
Peru 10/1993 06/2006 16,670 0.59 109 0.92 7,772 0.14<br />
Philippines 07/1991 06/2006 28,023 0.99 156 1.31 20,857 0.38<br />
Pol<strong>and</strong> 02/1999 06/2006 11,528 0.41 133 1.12 18,976 0.35<br />
Portugal 12/1989 06/2006 11,862 0.42 77 0.65 22,344 0.41<br />
Singapore 08/1984 06/2006 40,403 1.43 157 1.32 65,385 1.19<br />
South Africa 05/1982 06/2006 66,391 2.35 253 2.13 88,217 1.60<br />
South Korea 08/1987 06/2006 148,122 5.25 653 5.49 82,878 1.51<br />
Spain 10/1988 06/2006 20,815 0.74 98 0.82 160,075 2.91<br />
Sweden 08/1983 06/2006 54,317 1.93 198 1.66 97,062 1.76<br />
Switzerl<strong>and</strong> 07/1981 06/2006 58,621 2.08 197 1.66 63,230 1.15<br />
Taiwan 02/1991 06/2006 63,985 2.27 346 2.91 173,789 3.16<br />
Thail<strong>and</strong> 08/1988 06/2006 41,520 1.47 209 1.76 35,353 0.64<br />
Turkey 07/1990 06/2006 29,810 1.06 157 1.32 26,860 0.49<br />
U.K. 07/1981 06/2006 374,127 13.27 1,247 10.50 642,183 11.68<br />
Total 2,819,906 100.00 11,878 100.00 5,500,018 100.00
Table 2: Aggregate Summary Statistics<br />
This table reports aggregate summary statistics <strong>for</strong> one-month <strong>and</strong> one-year holding period returns, the<br />
natural logarithm of the June-end market value (ME), the natural logarithm of the previous year’s<br />
fiscal year-end book-to-market ratio (BM), the past six-month stock return (MOM), <strong>and</strong> one-year<br />
(ISSUE) changes in the number of shares outst<strong>and</strong>ing adjusted <strong>for</strong> distribution events such as stock<br />
splits <strong>and</strong> stock dividends. ISSUE is computed over month t-12 to t: ISSUEt,t-12 = Ln(Adjusted Sharest)<br />
– Ln(Adjusted Sharest-12). The sample is drawn from 41 countries, covers a period between July 1981<br />
<strong>and</strong> June 2006, <strong>and</strong> consists of 2,819,906 total firm-month observations.<br />
Total Number of<br />
St<strong>and</strong>ard 25<br />
Variable Observations Mean Deviation<br />
th<br />
75<br />
Percentile Median<br />
th<br />
Percentile<br />
1-Month Return 2,819,906 0.011 0.141 -0.057 0.000 0.063<br />
1 st -Year Return 2,338,097 0.162 0.642 -0.179 0.054 0.353<br />
ME 2,819,906 4.382 2.037 3.004 4.419 5.777<br />
BM 1,901,598 -0.390 0.850 -0.908 -0.372 0.139<br />
MOM 2,819,906 0.068 0.409 -0.146 0.012 0.204<br />
ISSUE 2,819,906 0.048 0.206 0.000 0.000 0.008<br />
32
Table 3: Fama-MacBeth Cross-Sectional Regressions of Annual Share Issuance on Firm<br />
Characteristics<br />
This table reports the results of Fama-MacBeth cross-sectional regressions. The annual share issuance<br />
measure (ISSUE) is regressed on the following firm-specific variables available immediately be<strong>for</strong>e<br />
the ISSUE-construction months: the natural logarithm of the June-end market value (ME), the natural<br />
logarithm of the previous year’s fiscal year-end book-to-market ratio (BM), the past six-month stock<br />
return (MOM), <strong>and</strong> the annual share-issuance measure lagged by 12 months (LAG-ISSUE). If the book<br />
value of equity is either missing or negative, then we assign both BM <strong>and</strong> BM-Dum values of zero.<br />
Otherwise, BM-Dum receives a value of one. ISSUE is the real change in the number of shares<br />
outst<strong>and</strong>ing adjusted, computed over month t-12 to t: ISSUEt,t-12 = Ln(Adjusted Sharest) – Ln(Adjusted<br />
Sharest-12). We estimate six different regression specifications using either equal-weighting, or valueweighting,<br />
or scaled weighting each observation, <strong>and</strong> by either including or excluding country<br />
dummies. The scaled-weights are computed by dividing each value-weight by the average market of<br />
the firm’s home country. The coefficients <strong>and</strong> adjusted R 2 s are in percentages <strong>and</strong> are given by the<br />
time-series averages of the corresponding statistics obtained from the monthly cross-sectional<br />
regressions. T-statistics, corrected <strong>for</strong> overlapping ISSUE-construction periods, are reported in<br />
parentheses. Coefficients with a 10% significance level or higher are bolded. The sample is drawn<br />
from 41 countries, covers a period between January 1983 <strong>and</strong> June 2006, <strong>and</strong> consists of 1,944,531<br />
firm-month observations.<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 6.71 5.42 6.32 6.55 5.15 6.05<br />
(9.67) (10.03) (11.85) (8.49) (10.01) (10.97)<br />
ME -0.50 -0.37 -0.52 -0.68 -0.42 -0.56<br />
(-4.87) (-10.21) (-10.48) (-5.70) (-7.39) (-8.76)<br />
BM 0.25 0.45 0.50 0.41 0.46 0.60<br />
(1.17) (3.89) (3.81) (1.13) (2.04) (3.86)<br />
BM-Dum -0.07 0.35 0.59 0.09 0.13 0.61<br />
(-0.28) (1.07) (1.83) (0.36) (0.39) (1.92)<br />
MOM 1.61 0.96 1.02 1.41 0.95 0.85<br />
(4.45) (2.24) (2.54) (4.23) (2.95) (3.11)<br />
LAG-ISSUE 7.69 3.65 4.13 10.83 6.74 6.71<br />
(3.60) (1.83) (2.37) (5.58) (5.49) (5.48)<br />
Average Adj. R 2 8.48 8.24 8.20 3.23 2.69 2.50<br />
Country Dummies Yes Yes Yes No No No<br />
33
Table 4: Fama-MacBeth Regressions of Holding Period Returns on Share Issuance <strong>and</strong><br />
Share Issuance with Controls<br />
This table reports the results of Fama-MacBeth regressions both with <strong>and</strong> without country dummies.<br />
The regressions are estimated with equal-weights, value-weights, <strong>and</strong> value-weights scaled by the<br />
average market value within the firm’s country. In Panels A <strong>and</strong> B holding period returns, measured<br />
over the first month <strong>and</strong> first year, are regressed on one-year (ISSUE) real changes in the number of<br />
shares outst<strong>and</strong>ing. ISSUE is computed over month t-12 to t: ISSUEt,t-12 = Ln(Adjusted Sharest) –<br />
Ln(Adjusted Sharest-12). In Panels C <strong>and</strong> D holding period returns, measured over the first month <strong>and</strong><br />
first year, are regressed on the natural logarithm of the June-end market value (ME), the natural<br />
logarithm of the previous year’s fiscal year-end book-to-market ratio (BM), the past six-month stock<br />
return (MOM), <strong>and</strong> one-year (ISSUE) real changes in the number of shares outst<strong>and</strong>ing. If book value<br />
of equity is either missing or negative, then we assign both BM <strong>and</strong> BM-Dum values of zero.<br />
Otherwise, BM-Dum receives a value of one. T-statistics, corrected <strong>for</strong> overlapping holding periods,<br />
are reported in parentheses. Coefficients with a 10% significance level or higher are bolded. The<br />
sample is drawn from 41 countries, covers a period between July 1981 <strong>and</strong> June 2006, <strong>and</strong> consists of<br />
2,819,906 total firm-month observations.<br />
Panel A<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1 st -Year 1 st -Year 1 st -Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.15 1.3 1.30 18.24 16.82 16.83<br />
(3.91) (4.19) (4.20) (4.37) (5.22) (5.23)<br />
ISSUE -0.79 -0.59 -0.66 -7.10 -6.58 -6.39<br />
(-7.12) (-3.78) (-5.07) (-6.25) (-3.63) (-3.96)<br />
Average Adj. R 2 19.82 23.57 27.46 22.15 25.34 28.13<br />
Country Dummies Yes Yes Yes Yes Yes Yes<br />
Panel B<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1 st -Year 1 st -Year 1 st -Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.24 1.07 1.23 17.63 13.78 16.38<br />
(5.23) (3.94) (4.90) (0.45) (3.27) (4.83)<br />
ISSUE -0.96 -0.73 -0.82 -7.05 -7.25 -7.00<br />
(-5.28) (-2.60) (-3.93) (-3.55) (-2.33) (-3.39)<br />
Average Adj. R 2 0.18 0.29 0.22 0.26 0.27 0.28<br />
Country Dummies No No No No No No<br />
34
Table 4: Continued<br />
Panel C<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1 st -Year 1 st -Year 1 st -Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.17 0.95 0.84 7.70 15.25 15.47<br />
(4.05) (2.38) (2.52) (3.57) (3.09) (3.43)<br />
ME -0.04 0.02 0.04 -0.21 -0.04 -0.08<br />
(-1.63) (0.54) (1.23) (-1.20) (-0.08) (-0.29)<br />
BM 0.34 0.36 0.36 2.13 4.21 4.24<br />
(7.27) (4.84) (5.99) (4.33) (3.78) (3.60)<br />
BM-Dum 0.33 0.36 0.36 2.15 3.75 4.08<br />
(6.59) (3.77) (4.74) (5.32) (3.44) (3.30)<br />
MOM 0.85 0.36 0.63 6.29 5.18 6.39<br />
(5.19) (1.04) (2.44) (7.58) (1.79) (2.49)<br />
ISSUE -0.71 -0.65 -0.68 -3.30 -6.77 -6.59<br />
(-6.84) (-4.41) (-5.53) (-5.47) (-3.77) (-4.08)<br />
Average Adj. R 2 20.73 26.71 29.11 24.58 28.79 30.26<br />
Country Dummies Yes Yes Yes Yes Yes Yes<br />
Panel D<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1 st -Year 1 st -Year 1 st -Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.58 0.85 1.27 24.17 15.91 21.75<br />
(6.07) (2.37) (3.50) (5.27) (3.87) (5.22)<br />
ME -0.10 0.01 -0.03 -1.55 -0.29 -0.85<br />
(-3.10) (0.26) (-0.64) (-4.67) (-0.74) (-2.28)<br />
BM 0.34 0.34 0.38 4.69 4.25 4.90<br />
(3.90) (3.25) (4.01) (6.77) (4.83) (7.12)<br />
BM-Dum 0.26 0.39 0.32 3.28 4.18 3.85<br />
(2.27) (3.14) (2.76) (2.04) (3.72) (5.16)<br />
MOM 0.94 0.31 0.62 8.04 3.58 5.31<br />
(3.42) (0.71) (1.70) (2.50) (0.78) (1.35)<br />
ISSUE -0.91 -0.78 -0.92 -6.87 -8.66 -8.71<br />
(-5.37) (-3.22) (-4.77) (-4.16) (-2.56) (-3.69)<br />
Average Adj. R 2 3.26 6.64 4.90 3.89 7.03 5.50<br />
Country Dummies No No No No No No<br />
35
Table 5: Summary Statistics <strong>for</strong> Portfolios Sorted on Annual Share Issuance<br />
This table provides summary statistics <strong>for</strong> portfolios sorted on annual share issuance (ISSUE). ISSUE<br />
is the real change in the number of shares outst<strong>and</strong>ing computed over month t-12 to t: ISSUEt,t-12 =<br />
Ln(Adjusted Sharest) – Ln(Adjusted Sharest-12). Each month, we follow Fama <strong>and</strong> French (2007) <strong>and</strong><br />
use all but the smallest quintile stocks to determine our portfolio breakpoints. The smallest stocks are<br />
included in the portfolios. We sort stocks with negative issuance into two portfolios; NEG1 is a<br />
portfolio <strong>for</strong> stocks below the negative issuance median <strong>and</strong> NEG2 is <strong>for</strong> stocks above the negative<br />
issuance median. We sort stocks with positive issuance into five quintiles (POS1-POS5); POS1 is <strong>for</strong><br />
stocks in the lowest positive issuance quintile <strong>and</strong> POS5 is <strong>for</strong> stocks in the highest issuance quintile.<br />
ZERO is a portfolio that includes stocks with zero net annual issuance. The sample is drawn from 41<br />
countries, consists of 2,819,906 total firm-month observations, <strong>and</strong> is from July of 1981 to June of<br />
2006. Panel A reports results <strong>for</strong> the entire sample, while Panel B excludes observations <strong>for</strong> which<br />
buybacks were either illegal or unattractive from either a tax or other regulatory perspective.<br />
Panel A<br />
Average Monthly Average Number<br />
Average Cross-Sectional of Firm Average Monthly<br />
ISSUE Rank Monthly Value St<strong>and</strong>ard Deviation of Observations per Percentage of<br />
Portfolio<br />
of ISSUE<br />
ISSUE<br />
Month<br />
Sample (%)<br />
NEG1 -0.133 0.298 306 3.26<br />
NEG2 -0.002 0.002 288 3.06<br />
ZERO 0.000 0.000 5,457 58.05<br />
POS1 0.001 0.001 623 6.63<br />
POS2 0.009 0.004 628 6.68<br />
POS3 0.038 0.014 651 6.93<br />
POS4 0.120 0.034 681 7.25<br />
POS5 0.462 0.359 766 8.15<br />
Panel B<br />
Average Monthly Average Number<br />
Average Cross-Sectional of Firm Average Monthly<br />
ISSUE Rank Monthly Value St<strong>and</strong>ard Deviation of Observations per Percentage of<br />
Portfolio<br />
of ISSUE<br />
ISSUE<br />
Month<br />
Sample (%)<br />
NEG1 -18.41 34.88 220 3.53<br />
NEG2 -0.27 0.26 208 3.34<br />
ZERO 0.00 0.00 3,450 55.43<br />
POS1 0.15 0.12 432 6.94<br />
POS2 1.01 0.46 435 6.99<br />
POS3 4.43 1.73 458 7.36<br />
POS4 13.98 3.86 483 7.76<br />
POS5 50.23 36.61 538 8.65<br />
36
Table 6: Equal-Weighted Fama-MacBeth Regressions of Holding Period Returns on<br />
Annual Share Issuance Rankings <strong>and</strong> Other Firm Characteristics<br />
This table reports the results of Fama-MacBeth regressions with country dummies. The regressions<br />
are estimated with equal-weights, value-weights, <strong>and</strong> value-weights scaled by the average market<br />
value within the firm’s country. Holding period returns, measured over the first month <strong>and</strong> first year,<br />
are regressed on the natural logarithm of the June-end market value (ME), the natural logarithm of the<br />
previous year’s fiscal year-end book-to-market ratio (BM), the past six-month stock return (MOM),<br />
<strong>and</strong> annual-issuance (ISSUE) rank dummies. If book value of equity is either missing or negative, then<br />
we assign both BM <strong>and</strong> BM-Dum values of zero. Otherwise, BM-Dum receives a value of one. ISSUE<br />
is the real change in the number of shares outst<strong>and</strong>ing adjusted computed over month t-12 to t:<br />
ISSUEt,t-12 = Ln(Adjusted Sharest) – Ln(Adjusted Sharest-12). Each month, we follow Fama <strong>and</strong> French<br />
(2007) <strong>and</strong> use all but the smallest quintile stocks to determine our portfolio breakpoints. The smallest<br />
stocks are included in the portfolios. We sort stocks with negative issuance into two portfolios; NEG1<br />
is a portfolio <strong>for</strong> stocks below the negative issuance median <strong>and</strong> NEG2 is <strong>for</strong> stocks above the<br />
negative issuance median. We sort stocks with positive issuance into five quintiles (POS1-POS5);<br />
POS1 is <strong>for</strong> stocks in the lowest positive issuance quintile <strong>and</strong> POS5 is <strong>for</strong> stocks in the highest<br />
issuance quintile. An ISSUE rank dummy variable is assigned a value of one if the firm is in the<br />
portfolio, <strong>and</strong> zero otherwise. The coefficients <strong>and</strong> adjusted R 2 s are in percentages <strong>and</strong> are given by<br />
the time-series averages of the corresponding statistics obtained from the monthly cross-sectional<br />
regressions. t-statistics, corrected <strong>for</strong> overlapping holding periods, are reported in parentheses.<br />
Coefficients with a 10% significance level or higher are bolded. The sample is drawn from 41<br />
countries, covers a period between July 1981 <strong>and</strong> June 2006, <strong>and</strong> consists of 2,819,906 total firmmonth<br />
observations. Panel A reports results <strong>for</strong> the entire sample, while Panel B excludes<br />
observations <strong>for</strong> which buybacks were either illegal or unattractive from either a tax or other<br />
regulatory perspective.<br />
37
Panel A<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1st-Year 1st-Year 1st-Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.18 0.85 15.45 0.98 18.70 15.42<br />
(4.12) (2.55) (3.15) (2.48) (3.55) (3.36)<br />
ME -0.040 0.03 -0.04 0.02 -0.72 -0.10<br />
(-1.58) (1.16) (-0.09) (0.44) (-2.42) -(-0.37)<br />
BM 0.33 0.35 4.16 0.34 4.42 4.23<br />
(7.26) (5.98) (3.79) (4.76) (3.81) (3.61)<br />
BM-Dum 0.33 0.36 3.88 0.36 4.19 4.14<br />
(6.63) (4.82) (3.47) (3.83) (4.87) (3.38)<br />
MOM 0.85 0.63 5.21 0.36 9.10 6.41<br />
(5.22) (2.46) (1.81) (1.04) (4.04) (2.51)<br />
NEG1 0.12 0.06 -0.21 0.06 1.14 -0.02<br />
(2.08) (0.82) (-0.29) (0.79) (1.94) (-0.02)<br />
NEG2 0.01 -0.07 -0.59 -0.11 0.30 -0.15<br />
(0.18) (-0.87) (-0.78) (-1.10) (0.68) (-0.17)<br />
POS1 0.011 -0.03 -0.20 -0.06 0.76 0.47<br />
(0.28) (-0.51) (-0.29) (-0.90) (2.76) (-0.80)<br />
POS2 0.00 0.02 -0.39 0.00 0.10 0.33<br />
(0.01) (-0.43) -(0.47) (-0.04) (0.22) (0.48)<br />
POS3 -0.08 -0.01 -0.69 -0.01 -0.03 -0.53<br />
(-1.89) (-0.15) (-0.79) (-0.10) (-0.05) (-0.79)<br />
POS4 -0.28 -0.19 -2.39 -0.18 -2.10 -2.19<br />
(-4.57) (-2.71) (-2.38) (-2.28) (-2.73) (-2.36)<br />
POS5 -0.44 -0.33 -3.81 -0.35 -4.08 -3.67<br />
(-6.00) (-4.45) (-6.11) (-4.35) (-5.71) (-4.69)<br />
Average Adj. R 2 20.78 29.40 29.41 27.25 23.84 30.62<br />
Country Dummies Yes Yes Yes Yes Yes Yes<br />
38
Panel B<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Return Horizon 1-Month 1-Month 1-Month 1st-Year 1st-Year 1st-Year<br />
Weighting Equal Value Scaled Equal Value Scaled<br />
Intercept 1.19 0.77 15.05 0.67 19.63 16.14<br />
(4.09) (2.25) (2.61) (1.81) (3.25) (2.97)<br />
ME -0.05 0.04 0.06 0.06 -0.99 -0.10<br />
(-1.84) (-1.36) (-0.14) (-1.66) (-2.10) (-0.28)<br />
BM 0.29 0.35 3.20 0.31 3.56 3.89<br />
(5.63) (5.03) (2.70) (4.34) (2.87) (3.26)<br />
BM-Dum 0.40 0.31 2.55 0.27 4.57 3.49<br />
(6.99) (3.45) (1.76) (2.65) (4.57) (2.61)<br />
MOM 1.06 0.81 8.24 0.84 9.72 6.47<br />
(5.97) (2.98) (2.76) (2.70) (3.31) (2.19)<br />
NEG1 0.09 -0.04 -2.77 -0.10 0.37 -2.31<br />
(1.10) (-0.34) (-1.76) (-0.89) (-0.39) (-1.45)<br />
NEG2 0.03 -0.06 -0.45 -0.12 0.61 0.41<br />
(0.36) (-0.47) (-0.24) (-0.94) (0.91) (0.24)<br />
POS1 0.08 -0.05 -0.31 -0.08 1.10 0.06<br />
(1.53) (-0.81) (-0.55) (-1.18) (2.33) (0.10)<br />
POS2 -0.01 0.08 -0.31 0.05 -0.31 0.28<br />
(-0.14) (1.14) (-0.35) (0.69) (-0.42) (0.37)<br />
POS3 -0.17 -0.01 -0.54 -0.04 -0.56 -0.26<br />
(-2.89) (-0.09) (-0.51) (-0.52) (-0.82) (-0.37)<br />
POS4 -0.35 -0.27 -3.24 -0.30 -2.98 -3.32<br />
(-4.71) (-3.18) (-2.80) (-3.46) (-3.87) (-3.53)<br />
POS5 -0.55 -0.43 -5.52 -0.48 -5.03 -5.11<br />
(-6.30) (-4.48) (-6.45) (-5.00) (-5.84) (-6.13)<br />
Average Adj. R 2 16.02 24.93 25.43 23.26 19.78 27.00<br />
Country Dummies Yes Yes Yes Yes Yes Yes<br />
39
Table 7: Country-Level Issuance, Issuance Activity, Market Development, <strong>and</strong><br />
Governance Variables<br />
This Table displays country level data. The first column in Panel A is the share issuance coefficient<br />
from a within-country equal-weighted regression of monthly returns on ISSUE along with controls <strong>for</strong><br />
size, book-to-market, <strong>and</strong> momentum. The second column reports the percentage of firm-month<br />
observations with nonzero values of ISSUE. The items in columns 3, 4, <strong>and</strong> 5 in were obtained from<br />
La Porta et al. (2006). Liquidity is the total dollar value of stocks traded scaled by GDP, <strong>for</strong> the period<br />
1996-2000. Turnover is the total dollar value of stocks traded, scaled by the value of shares<br />
outst<strong>and</strong>ing, <strong>for</strong> the period 1996-2000. Log GDP is per capita GDP in U.S. dollars in 2000. Buybacks<br />
Allowed? is the year in which buybacks first became feasible; we collected this data from various<br />
sources. Short Selling Allowed? is the year in which a country first allowed short selling; it is from<br />
Bris et a. (2007). The first two columns in Panel B are from La Porta et al (1997). Law reports<br />
whether the country has either a common or civil law legal origin. Accounting is an index of<br />
accounting st<strong>and</strong>ards, a higher value suggests better accounting st<strong>and</strong>ards. The next three columns are<br />
from La Porta et al. (2006). With each measure a higher value signals greater investor protection.<br />
Criminal <strong>and</strong> Liability are indices which measure the ease with which accountants, directors, <strong>and</strong><br />
distributors can be pursued in criminal <strong>and</strong> civil courts <strong>for</strong> not fully disclosing in<strong>for</strong>mation in<br />
prospectuses. Protect is the principle component of Liability <strong>and</strong> two other investor protection indices.<br />
Earnings Management is from Leutz et al. (2003). A higher value of the Earnings Management<br />
suggests poorer earnings quality. Panel C reports the correlations between the variables.<br />
40
Panel A: Issuance Activity <strong>and</strong> Market Development Variables<br />
1-Month<br />
Issue<br />
Coefficient<br />
%<br />
Nonzero<br />
Issuance Liquidity Turnover<br />
41<br />
Short<br />
Selling<br />
Allowed?<br />
Log GDP Buybacks<br />
Country<br />
per capita Allowed?<br />
Argentina 0.24 5.83 14.42 8.95 Full 1999<br />
Australia -1.03 0.61 45.16 48.16 9.89 1996 Be<strong>for</strong>e 1990<br />
Austria 0.14 6.71 49.52 10.05 2000 Be<strong>for</strong>e 1990<br />
Belgium 1.00 0.18 16.83 24.68 10.02 2000 Be<strong>for</strong>e 1990<br />
Brazil 0.48 0.24 18.29 61.14 8.14 Full Be<strong>for</strong>e 1990<br />
Canada -0.60 0.65 57.86 65.64 10.05 Full Be<strong>for</strong>e 1990<br />
Chile 2.26 0.11 9.14 10.00 8.44 Full Be<strong>for</strong>e 1990<br />
China 0.59 0.33 9.14 10.00 8.44 Not Allowed<br />
Czech Republic 0.06 Be<strong>for</strong>e 1990<br />
Denmark -0.76 0.22 36.27 64.23 10.31 2001 Be<strong>for</strong>e 1990<br />
Egypt 0.08 7.76 32.81 7.28<br />
Finl<strong>and</strong> -0.95 0.36 70.97 51.34 10.06 1998 1998<br />
France -0.81 0.40 44.90 63.82 9.99 1999 Be<strong>for</strong>e 1990<br />
Germany -1.23 0.20 37.80 118.46 10.03 1999 Be<strong>for</strong>e 1990<br />
Greece -0.42 0.29 60.84 78.79 9.27 Full Not Allowed<br />
Hong Kong -0.84 0.53 179.05 64.94 10.10 1992 1996<br />
India 0.22 0.26 54.65 63.72 6.16 2000<br />
Indonesia 0.28 0.26 13.78 65.20 6.59 Not Allowed<br />
Irel<strong>and</strong> 0.70 30.79 59.96 10.14 Full Be<strong>for</strong>e 1990<br />
Italy -0.57 0.38 36.58 80.08 9.84 Full Be<strong>for</strong>e 1990<br />
Japan 0.51 0.42 35.50 49.40 10.54 1996 Be<strong>for</strong>e 1990<br />
Malaysia 0.40 0.35 98.54 50.77 8.25 1998 1995 to 1997<br />
Mexico -0.53 0.45 9.89 34.42 8.67 Be<strong>for</strong>e 1990<br />
Netherl<strong>and</strong>s -1.43 0.45 113.49 95.36 10.06 2002 Be<strong>for</strong>e 1990<br />
New Zeal<strong>and</strong> 0.53 17.82 42.68 9.48 2000 1992<br />
Norway -0.57 0.44 30.15 79.64 10.49 2000 1992<br />
Pakistan -1.01 0.22 26.50 220.97 6.10 Full Not Allowed<br />
Peru -3.83 0.38 5.19 21.65 7.64 Not Allowed<br />
Philippines -0.23 0.28 21.45 39.52 6.83 1998<br />
Pol<strong>and</strong> 0.01 0.17 1998 2000<br />
Portugal 0.16 30.98 68.18 9.27 Be<strong>for</strong>e 1990<br />
Singapore -0.78 0.42 79.15 49.62 10.05 1999 Not Allowed<br />
South Africa -1.59 0.40 41.77 27.74 7.98 2001 Be<strong>for</strong>e 1990<br />
South Korea -0.87 0.38 110.16 204.92 9.18 Full Not Allowed<br />
Spain 0.01 0.37 107.98 174.82 9.56 Full 1992<br />
Sweden -0.78 0.40 92.22 78.08 10.15 2001 1991<br />
Switzerl<strong>and</strong> -0.31 0.25 206.27 91.22 10.41 1998 Be<strong>for</strong>e 1990<br />
Taiwan 0.06 0.72 320.69 314.74 9.54 2001 Be<strong>for</strong>e 1990<br />
Thail<strong>and</strong> 0.50 0.45 22.55 62.51 7.58 2002 1997<br />
Turkey -2.31 0.49 43.68 148.00 8.02 Full Be<strong>for</strong>e 1990<br />
UK -1.19 0.55 83.02 50.62 10.08 1981 Be<strong>for</strong>e 1990<br />
US -2.06 0.87 178.88 125.30 10.46 1982 Be<strong>for</strong>e 1990
Country<br />
Panel B: Governance Variables<br />
1-Year<br />
Issue<br />
Coefficient Law Accounting Criminal Liability Protect<br />
42<br />
Earnings<br />
Management<br />
Score<br />
Argentina Civil 45 0.17 0.22 0.479<br />
Australia -10.20 Common 75 0.83 0.66 0.784 4.8<br />
Austria Civil 54 0.50 0.11 0.104 28.3<br />
Belgium 10.48 Civil 61 0.50 0.44 0.068 19.5<br />
Brazil 1.55 Civil 54 0.33 0.33 0.442<br />
Canada -3.22 Common 74 0.83 1.00 0.959 5.3<br />
Chile 21.31 Civil 52 0.50 0.33 0.610<br />
China 34.68 Civil<br />
Czech Republic Civil<br />
Denmark -15.66 Civil 62 0.00 0.55 0.363 16.0<br />
Egypt Civil 24 0.42 0.22 0.202<br />
Finl<strong>and</strong> 28.63 Civil 77 0.50 0.66 0.465 12.0<br />
France -7.76 Civil 69 0.33 0.22 0.473 13.5<br />
Germany -6.88 Civil 62 0.50 0.00 0.000 21.5<br />
Greece -9.23 Civil 55 0.50 0.50 0.319 28.3<br />
Hong Kong -11.13 Common 69 1.00 0.66 0.851 19.5<br />
India 0.26 Common 57 0.83 0.66 0.769 19.1<br />
Indonesia 4.63 Civil 0.50 0.66 0.507 18.3<br />
Irel<strong>and</strong> Common 0.83 0.44 0.478 5.1<br />
Italy -14.37 Civil 62 0.50 0.22 0.197 24.8<br />
Japan 7.82 Civil 65 0.00 0.66 0.420 20.5<br />
Malaysia 3.31 Common 76 1.00 0.66 0.730 14.8<br />
Mexico -6.62 Civil 60 0.50 0.11 0.100<br />
Netherl<strong>and</strong>s -18.99 Civil 64 0.50 0.89 0.540 16.5<br />
New Zeal<strong>and</strong> 0.00 Common 70 0.33 0.44 0.460<br />
Norway -3.69 Civil 74 1.00 0.39 0.440 5.8<br />
Pakistan -18.58 Common 0.08 0.39 0.630 17.8<br />
Peru -72.05 Civil 38 0.50 0.66 0.660<br />
Philippines -10.16 Civil 65 0.50 1.00 0.810 8.8<br />
Pol<strong>and</strong> -2.07 Civil<br />
Portugal Civil 36 0.00 0.66 0.570 25.1<br />
Singapore -14.42 Common 78 1.00 0.66 0.770 21.6<br />
South Africa -7.98 Common 70 0.42 0.66 0.600 5.6<br />
South Korea -7.93 Civil 62 0.33 0.66 0.360 26.8<br />
Spain 5.64 Civil 64 0.50 0.66 0.550 18.6<br />
Sweden -5.08 Civil 83 0.58 0.28 0.390 6.8<br />
Switzerl<strong>and</strong> -1.09 Civil 38 0.33 0.44 0.300 22.0<br />
Taiwan -7.31 Civil 65 0.83 0.66 0.550 22.5<br />
Thail<strong>and</strong> 11.18 Common 64 0.58 0.22 0.370 18.3<br />
Turkey -20.16 Civil 51 0.50 0.22 0.338<br />
UK -10.40 Common 78 0.42 0.78 0.770 7.0<br />
US -16.40 Common 71 0.50 1.00 1.000 2.0
1-Month 1-Year %<br />
Panel C: Correlation Matrix<br />
Log<br />
Erngs.<br />
Issue Issue Nonzero Liquidity Turnover GDP Law Accntg. Criminal Liability Protect Mng.<br />
1-Month Issue 1.00<br />
1-Year Issue 0.79 1.00<br />
% Non-Zero -0.42 -0.24 1.00<br />
Liquidity -0.09 -0.07 0.49 1.00<br />
Turnover -0.14 -0.17 0.29 0.64 1.00<br />
Log GDP -0.09 0.05 0.37 0.35 0.02 1.00<br />
Law -0.13 -0.09 0.51 0.11 -0.05 -0.08 1.00<br />
Accntg. 0.04 0.22 0.58 0.21 0.10 0.40 0.47 1.00<br />
Criminal 0.04 0.03 0.43 0.30 0.01 0.08 0.41 0.45 1.00<br />
Liability -0.16 -0.13 0.43 0.38 0.09 0.04 0.34 0.32 0.17 1.00<br />
Protect -0.16 -0.18 0.48 0.29 0.01 -0.12 0.61 0.36 0.33 0.78 1.00<br />
Erngs. Mng. 0.42 0.05 -0.57 0.07 0.27 -0.08 -0.45 -0.65 -0.26 -0.37 -0.54 1.00<br />
43
Table 8: Equal-Weighted Fama-MacBeth Regressions of Monthly Returns on Annual Share<br />
Issuance with Issuance Activity, Market Development, <strong>and</strong> Governance Interactions <strong>and</strong><br />
Controls<br />
This table reports the results of equal-weighted Fama-MacBeth regressions. Holding period returns measured<br />
over one month <strong>and</strong> one year are regressed on the natural logarithm of June-end market value (ME), the<br />
natural logarithm of the previous year’s fiscal year-end book-to-market ratio (BM), the past six-month stock<br />
return (MOM), <strong>and</strong> one-year (ISSUE) real changes in the number of shares outst<strong>and</strong>ing. If book value of<br />
equity is either missing or negative, then we assign both BM <strong>and</strong> BM-Dum values of zero. Otherwise, BM-<br />
Dum receives a value of one. ISSUE is interacted with each of the variables described in Table 7. T-statistics,<br />
corrected <strong>for</strong> overlapping holding periods, are reported in parentheses. Coefficients with a 10% significance<br />
level or higher are bolded. The sample is drawn from 41 countries, covers a period between July 1981 <strong>and</strong><br />
June 2006, <strong>and</strong> consists of 2,819,906 total firm-month observations.<br />
44
Panel A: Issuance Activity <strong>and</strong> Market Development Interactions: 1-Month Returns<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Interactive % Nonzero<br />
Log GDP Buybacks No Short<br />
Variable Issuance Liquidity Turnover per capita Allowed Selling<br />
Intercept 1.15 1.03 1.00 5.06 0.58 0.74<br />
(1.65) (2.16) (2.33) (0.72) (1.99) (2.52)<br />
ME 0.03 0.04 0.04 0.04 0.04 0.04<br />
(1.16) (1.20) (1.22) (1.19) (1.18) (1.25)<br />
BM 0.38 0.39 0.39 0.39 0.39 0.40<br />
(7.21) (7.23) (7.25) (7.23) (7.07) (7.35)<br />
BM-Dum 0.26 0.26 0.26 0.26 0.27 0.26<br />
(6.08) (6.04) (6.09) (6.08) (6.09) (6.08)<br />
MOM 0.99 0.99 0.99 0.99 1.02 1.00<br />
(5.67) (5.64) (5.63) (5.63) (5.73) (5.51)<br />
ISSUE 0.58 -0.27 -0.44 8.34 -0.70 -1.17<br />
(2.36) (-1.85) (-2.17) (4.57) (-2.90) (-8.56)<br />
Variable -0.42 0.00 0.00 -0.42 0.25 0.27<br />
(-0.47) (0.48) (0.51) (-0.61) (0.81) (0.57)<br />
Interaction -2.94 -0.01 -0.01 -0.95 -0.58 0.51<br />
(-5.24) (-5.01) (-3.70) (-5.07) (-1.87) (1.59)<br />
Average<br />
Adj. R 2 15.83 15.81 15.80 15.80 15.27 15.28<br />
45
Panel B: Issuance Activity <strong>and</strong> Market Development Interactions: 1-Year Returns<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Interactive % Nonzero<br />
Log GDP Buybacks No Short<br />
Variable Issuance Liquidity Turnover per capita Allowed Selling<br />
Intercept 20.15 18.23 17.92 60.52 13.80 14.94<br />
(1.83) (2.38) (2.59) (0.54) (3.30) (5.37)<br />
ME -0.14 -0.12 -0.11 -0.12 -0.15 -0.10<br />
(-0.52) (-0.45) (-0.44) (-0.46) (-0.56) (-0.36)<br />
BM 4.35 4.37 4.41 4.40 4.39 4.44<br />
(3.89) (3.94) (3.98) (3.96) (3.85) (4.14)<br />
BM-Dum 2.78 2.89 2.92 2.91 2.77 2.91<br />
(3.24) (3.33) (3.37) (3.36) (3.17) (3.35)<br />
MOM 9.65 9.66 9.67 9.66 9.75 9.01<br />
(4.79) (4.76) (4.76) (4.76) (4.80) (4.22)<br />
ISSUE 4.30 -1.70 -5.06 53.09 -6.90 -9.95<br />
(1.12) (-1.06) (-1.92) (1.88) (-3.22) (-7.69)<br />
Variable -5.80 -0.02 -0.02 -4.17 1.84 1.91<br />
(-0.40) (-0.32) (-0.32) (-0.33) (0.39) (0.30)<br />
Interaction -23.25 -0.08 -0.05 -6.11 -4.24 -1.32<br />
(-3.39) (-4.61) (-1.83) (-2.15) (-1.68) (-0.28)<br />
Average<br />
Adj. R 2 20.03 19.96 19.95 19.96 19.35 19.67<br />
46
Panel C: Governance Interactions: 1-Month Returns<br />
Regression (1) (2) (3) (4)<br />
Interactive<br />
Variable Common Account Criminal Liability Protect<br />
47<br />
(5)<br />
(6)<br />
Erngs.<br />
Mgmt.<br />
Intercept 0.72 -1.17 1.81 1.20 1.46 0.71<br />
(2.67) (-0.38) (1.12) (1.58) (1.24) (1.97)<br />
ME 0.04 0.04 0.04 0.04 0.04 0.03<br />
(1.19) (1.20) (1.23) (1.21) (1.20) (1.09)<br />
BM 0.39 0.39 0.40 0.39 0.39 0.39<br />
(7.24) (7.15) (7.23) (7.20) (7.19) (6.85)<br />
BM-Dum 0.26 0.27 0.27 0.27 0.26 0.26<br />
(6.14) (6.06) (6.19) (6.12) (6.10) (5.69)<br />
MOM 0.99 1.00 1.00 1.00 0.99 1.03<br />
(5.65) (5.61) (5.60) (5.61) (5.61) (5.70)<br />
ISSUE -0.43 1.41 -1.34 -0.02 0.19 -1.66<br />
(-3.56) (2.42) (-5.88) (-0.11) (0.88) (-7.44)<br />
Variable 0.05 0.03 -2.19 -0.45 -0.72 0.03<br />
(0.16) (0.66) (-0.62) (-0.52) (-0.55) (0.56)<br />
Interaction -0.96 -0.04 0.53 -1.46 -1.72 0.06<br />
(-5.58) (-3.99) (1.36) (-3.96) (-4.64) (4.89)<br />
Average<br />
Adj. R 2 15.80 15.27 15.26 15.28 15.28 14.66
Panel D: Governance Interactions: 1-Year Returns<br />
Regression (1) (2) (3) (4) (5) (6)<br />
Interactive<br />
Variable Common Account Criminal Liability Protect<br />
48<br />
Erngs.<br />
Mgmt.<br />
Intercept 14.04 -5.92 27.40 20.66 23.39 14.61<br />
(3.38) (-0.14) (1.13) (1.73) (1.30) (4.14)<br />
ME -0.13 -0.13 -0.11 -0.11 -0.12 -0.16<br />
(-0.51) (-0.49) (-0.42) (-0.44) (-0.44) (-0.61)<br />
BM 4.40 4.39 4.43 4.38 4.37 4.34<br />
(3.95) (3.94) (3.98) (3.92) (3.91) (3.75)<br />
BM-Dum 2.79 2.85 2.93 2.90 2.89 2.72<br />
(3.27) (3.31) (3.38) (3.32) (3.31) (3.03)<br />
MOM 9.65 9.74 9.74 9.71 9.71 10.07<br />
(4.80) (4.73) (4.75) (4.74) (4.75) (5.11)<br />
ISSUE -2.18 15.09 -11.11 -0.68 2.49 -13.28<br />
(-1.28) (-1.75) (-3.51) (-0.18) (-0.65) (-5.98)<br />
Variable 1.26 0.29 -25.83 -6.05 -8.57 0.39<br />
(0.31) (0.41) (-0.46) (-0.46) (-0.40) (0.40)<br />
Interaction -9.64 -0.35 4.33 -11.44 -15.29 0.46<br />
(-3.94) (-2.63) (0.73) (-2.07) (-2.77) (2.46)<br />
Average<br />
Adj. R 2 20.02 19.18 19.29 19.30 19.31 18.81