You are on page 1of 21

Journal of Financial Economics 109 (2013) 661–681

Contents lists available at SciVerse ScienceDirect

Journal of Financial Economics
journal homepage: www.elsevier.com/locate/jfec

Controlling shareholders and market timing
in share issuance$
Borja Larrain a,n, Francisco Urzúa I. b
a
b

Pontificia Universidad Católica de Chile, Escuela de Administración and Finance UC, Avenida Vicuña Mackenna 4860, Macul, Santiago, Chile
Rotterdam School of Management, Erasmus University, The Netherlands

a r t i c l e i n f o

abstract

Article history:
Received 29 February 2012
Received in revised form
30 October 2012
Accepted 20 November 2012
Available online 23 April 2013

We examine market timing in the equity issuance of firms controlled by large shareholders using a hand-collected data set of controlling shareholders' ownership stakes in
Chile between 1990 and 2009. When a firm issues shares, the controlling shareholder can
either maintain or change his ownership stake depending on how many of the new shares
he subscribes. Issuance predicts poor future returns and is preceded by high returns, but
only when the controlling shareholder's stake is significantly reduced. Consistent with
market timing, the results are stronger in the absence of institutional investors and in hot
issuance markets.
& 2013 Elsevier B.V. All rights reserved.

JEL classification:
G14
G32
G34
Keywords:
Controlling shareholders
Issuance
Market timing
Ownership

1. Introduction
Most corporations in continental Europe, Asia, and
Latin America have large controlling shareholders (Barca


We are especially grateful to the referee, C. Fritz Foley, whose
excellent comments substantially improved the paper. We are also
indebted to seminar participants at the Pontificia Universidad Católica
de Chile, the first Finance UC International Conference, Malcolm Baker,
Ruth Bradley, Fabio Braggion, Marco Da Rin, Stefano Rossi (discussant),
Eduardo Walker, and, particularly, Andrei Shleifer for insightful comments and suggestions. We would also like to thank Fernando Lefort,
Eduardo Walker, and Moqi Xu for providing some of the data used in this
paper. Francisco Muñoz and Carla Castillo provided excellent research
assistance. This paper corresponds to the second chapter of Francisco
Urzúa I.'s dissertation at Tilburg University. Larrain acknowledges partial
financial support provided by the Programa Bicentenario de Ciencia y
Tecnología through the Concurso de Anillos de Investigación en Ciencias
Sociales (code SOC-04) and by Grupo Security through Finance UC.
n
Corresponding author. Tel.: +562 2354 4025.
E-mail addresses: borja.larrain@uc.cl (B. Larrain),
urzuainfante@rsm.nl (F. Urzúa I.).

0304-405X/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jfineco.2013.03.013

and Becht, 2001; Claessens, Djankov, and Lang, 2000;
Faccio and Lang, 2002; La Porta, López-de-Silanes, and
Shleifer, 1999). Large shareholders can mitigate the agency
conflict between managers and shareholders, but they
can also pursue interests that are at odds with those
of minority shareholders (Burkart, Gromb, and Panunzi,
1997; Grossman and Hart, 1980; Shleifer and Vishny, 1986).
Expropriation of minority shareholders or tunneling can
take many forms, from the most obvious ones such as
outright fraud or theft to less obvious (and harder to
detect) forms such as transactions with related parties at
inflated prices (Johnson, La Porta, López-de-Silanes, and
Shleifer, 2000).
In this paper, we study another form of opportunistic
behavior by controlling shareholders: market timing in
equity issuance or the sale of overpriced shares to outside
investors. The market timing hypothesis rests on three
assumptions. First, the controlling shareholder is better
informed than outside investors. Second, some outside
investors are naive in the sense that, faced with an

662

B. Larrain, F. Urzúa I. / Journal of Financial Economics 109 (2013) 661–681

issuance, they do not perceive themselves as being at a
disadvantage. Third, those outside investors who do interpret the controlling shareholder's intentions correctly face
limits to arbitrage.
The controlling shareholder has incentives for the firm
to issue overpriced shares because, although his proportional ownership falls with issuance, the overall value of
his stake increases. Simply put, the result for the controlling shareholder is a smaller fraction of future dividends,
but these dividends are of higher value. The main prediction of the market timing hypothesis is that returns
following issuance are poor because outside shareholders
are not immediately able to perceive the overvaluation or
act against it. As information is gradually incorporated into
prices or as investor optimism fades, the overvaluation
disappears and returns are poor. The critical implication of
this hypothesis is, however, that future returns are poor
conditional on issuance with dilution of the controlling
shareholder and not simply conditional on any issuance.
Other types of issuance as, for example, when the controlling shareholder subscribes the new shares at pro rata
indicate that the company is not overvalued and, therefore,
do not predict poor returns.
In this paper, we study post-issuance return predictability according to the stake of the controlling shareholder. The quality of the data available for Chile allows us
to determine the ownership stake of the controlling shareholder of all listed firms over a period of 20 years (1990–
2009). Our data are unique not only because of the long
period covered but also because they allow us to identify
the controlling shareholder by name and the size of his
stake in a precise way. This process requires intimate
knowledge of many firms intertwined through pyramidal
structures and other control mechanisms (Morck,
Wolfenzon, and Yeung, 2005). Moreover, under Chilean
law, all shareholders possess preemptive rights, allowing
them to subscribe new issues on a pro rata basis. This
implies that, contrary to the typical assumption of the
market timing literature, the size of the equity issuance
per se is not a proper measure of dilution. To measure
dilution we need to know how many of the new shares are
subscribed by the controlling shareholder.
We find that share issuance in general predicts low
future returns, as previously shown by Pontiff and
Woodgate (2008) and McLean, Pontiff, and Watanabe
(2009). However, consistent with the market timing
hypothesis, we find that all of this predictive power comes
from equity issues that imply substantial dilution of the
controlling shareholder. Monthly dollar returns are on
average 0.81% for diluting-issuers as compared with
2.46% for nonissuers. This implies that minority shareholders who buy shares of diluting-issuers, instead of
investing in nonissuers, lose on average 20% in a year.
Other issuances have a negligible impact on future returns.
For instance, monthly returns are on average 2.31% after
equity issues when the controlling shareholder's stake
does not change (i.e., when the controlling shareholder
subscribes the issue at pro rata).
The alternative to the market timing hypothesis is that
shares are issued at fair price and low post-issuance
returns reflect the relatively low risk of these companies.

We address the risk-based explanation in two ways. First,
all of our tests control for the standard risk factors
identified in the asset pricing literature such as size, value,
and momentum (Fama and French, 1992, 2008). Second,
we explore changes in risk around issuance. For example,
Carlson, Fisher, and Giammarino (2010) find that market
betas decrease after US seasoned equity offerings (SEOs),
which they interpret as a sign of issuance going hand-inhand with a decrease in risk. In our sample, we instead
find that, contrary to the risk-based explanation, the
market betas of poor-performing issuers increase after
issuance.
Consistent with the second assumption of the market
timing hypothesis, we find that the under-performance of
diluting issuers is more pronounced among firms that do
not have institutional investors (e.g., private pension
funds) in their shareholder base. Institutional investors
are arguably more sophisticated than retail investors and
less prone to irrational optimism. Similarly, the underperformance is stronger if the firm issues equity in a hot
issuance market. According to the behavioral literature,
hot markets are dominated by naive, optimistic investors
(Baker and Stein, 2004), which explains the differential
impact of issuance in these periods. Finally, we show
that no return under-performance is evident following
instances of dilution when the controlling shareholder
reduces its stake by selling old shares (a block sale) instead
of issuing new shares. In block sales, the opportunity for
overvaluation is limited not only by the fact that outside
investors are likely to be wealthier and more sophisticated
but also because the controlling shareholder's intentions
are more apparent. In a block sale, the controlling shareholder receives the proceeds directly, and, in an equity
issuance, they go to the firm, probably to finance new
projects. It is easier to disguise overpricing with share
issuance rather than block sales precisely because share
issuance involves investment. If investors are optimistic
about the firm's prospects, they like the firm to issue
shares for investment, but no reason exists for block sales
except overpricing.
In terms of pre-issuance characteristics, we find that
the dilution of the controlling shareholder is preceded by
high returns and high stock liquidity, which are both
typical features of overvaluation (Helwege, Pirinsky, and
Stulz, 2007). Dilution is followed by more capital expenditures, although profitability (return on equity, ROE) is
lower than after other equity issues and, if investors are
disappointed by the company's poor cash flows, this could
explain why overvaluation eventually fades away.
Our results contribute to the literature on large shareholders. First, we highlight that, in most firms around the
world, it is essential to focus on the controlling shareholder to understand financing policy. Our study of equity
financing complements other dimensions of corporate
policy in relation to large shareholders including dividend
policy (Chetty and Saez, 2005; Faccio, Lang, and Young,
2001; La Porta, López-de-Silanes, Shleifer, and Vishny,
2000; Shleifer and Vishny, 1986), the cost of borrowing
(Lin, Ma, Malatesta, and Xuan, 2011), chief executive officer
(CEO) compensation (Bertrand and Mullainathan, 2001;
Burkart, Gromb, and Panunzi, 1997), board compensation

Second. The behavioral literature has explored several different explanations for this slow reversion of investor sentiment. and Scherbina. The market timing hypothesis The market timing hypothesis rests on three assumptions. Ciccotello. Lewellen. 2002. 2012. which can explain the predictability of returns (Chen. Lee. predicts low returns in a cross section of stocks. First. If managers take advantage of windows of opportunity for issuing equity. Investors end up overpaying for the shares to the benefit of the controlling shareholder. 2005. 2010). and Gyoshev. and Loughran and Ritter. Pontiff. 1 See La Porta. and Vishny (1998) for the importance of legal protection in other cases. 1997). those outside investors who do understand the controlling shareholder's intentions are constrained in their buying and selling of securities (i. The rest of the paper is organized as follows. unlike the US. Suppose also that the market value of assets-in-place is above the valuation of the controlling shareholder (i. our results show that legal protection of minority investors cannot prevent abuses when investors fall prey to their own naivety. The rational explanation is that issuance coincides with a decrease in risk (e. therefore. From our second assumption. 1977. lower expected returns (Carlson. 2005). the decision to issue equity resides with the controlling shareholder and not with management. Besides short-sale . 2003). Shleifer. Frazzini and Lamont. López-de-Silanes. 663 2. and Warner. The controlling shareholder can take advantage of a cheap source of funding for the new project if the firm issues equity at the current market price but does not want to subscribe the new issuance because. his assessment of the discounted value of future dividends). It is well known that SEOs under-perform on average (Loughran and Ritter. we study the before and after of share issuances in terms of firm characteristics (such as ROE. Only optimistic valuations are reflected in the market price if there are short–sale constraints. 1995) and. 2004. Intintoli and Kahle. Urzúa I. outside investors do not infer the overvaluation from the fact that the controlling shareholder does not want to subscribe the new issuance (something that they can observe). López-de-Silanes. Black. Suppose that a firm controlled by a large shareholder is contemplating an equity issue to fund a new (observable) project. are unconstrained.e. 2000. and Watanabe (2009) on international evidence. and others) that can predict issuance or that are affected by issuance and. The rational response to overvaluation is short selling. Diether. and Kahle. or when they do not pay attention to all information. although they are drawn from an institutional environment different from that studied in most of these other papers. and Vishny. Jenter.. 2009.. Third. 2010. Hong and Stein (2007).g.3 Our results tend to side with the behavioral explanation. The main prediction of the market timing hypothesis is that the returns following issuance are poor as investors slowly come to realize that the firm is overvalued. The controlling shareholder has full control over the decision. then it is reasonable to expect they would do the same when trading stock of their own companies. as a result. capital expenditures. F. and not just SEOs. 2001. 1998). Li. and Giammarino. as a method of issuing equity. large controlling shareholders are prevalent and. but short selling is costly and prohibited in many markets. the market timing hypothesis needs to explain why rational investors (or arbitrageurs) fail to take advantage of overvaluation. Fisher. mention three broad classes of theories: (1) gradual dissemination of information. Shleifer. some outside investors do not infer their position of disadvantage from the behavior of the controlling shareholder. Larrain. 2000. 2010. 2009). 2009. 3 Insider trading around SEOs is related to the market timing hypothesis and has been studied in US corporations in which managers heavily influence corporate decisions. Second. who understand the controlling shareholder's intentions. First. a fall in the firm's beta) and. In Section 2. In Section 5. rights are common in many countries (La Porta. In theory. In addition to the slow reversion of sentiment. We also contribute to the literature on issuance and returns. Jenter. Section 3 describes our data in detail. Miller. the shares are overpriced.e. Malloy. 2002. Scheinkman and Xiong. overvaluation disappears immediately if rational investors. 2009). Livdan. or when they are overconfident about their own priors. and Pástor and Veronesi. Second. they do not rule out market timing. we study a market in which. this is not necessarily so. in rights offerings. in Section 6. instead of SEOs. this literature finds support for the market timing hypothesis (Clarke. 2 See Pontiff and Woodgate (2008) on US evidence and McLean. One possibility is that outside investors get particularly excited about new projects and their optimism tails off slowly only as the new company's true cash flows are gradually revealed after issuance. (2) limited investor attention. This is where the assumption of limits to arbitrage comes into play.1 Although preemptive rights are usually considered a remedy for the type of equity tunneling under which controlling shareholders dilute minority shareholders (Atanasov. 2008). The controlling shareholder's stake falls with the issuance but its overall value increases because the firm receives a cash infusion. they face limits on arbitrage). it has been shown that equity issuance broadly speaking. In Section 4. Hong. Broadly speaking. the controlling shareholder is better informed than outside investors. Dunbar. we focus on rights offerings. SEOs automatically imply dilution of the controlling shareholder while. Simply put. and Spiess and Affleck-Graves. and Zhang. 1995. although all shareholders have preemptive rights. market prices come to reflect true values in a slow and predictable way when agents incorporate information only gradually. we present our conclusions. we review the main assumptions and prediction of the market timing hypothesis.. 2008. and (3) heterogeneous priors.B. more recently.. 1995). in his opinion. Kahle. and investment (Cronqvist and Fahlenbrach. Although they have largely disappeared from publicly traded US companies (Eckbo. The behavioral explanation is that smart managers take advantage of irrational investors by issuing overpriced shares (Baker and Wurgler. we present the main return regressions. This still begs the question of why investors are fooled for such long periods of time. Greenwood and Hanson. / Journal of Financial Economics 109 (2013) 661–681 (Urzúa I. and Stein. This variation in dilution across different rights offerings allows us to develop sharper tests of the market timing hypothesis. 2006.2 The reasons for this correlation between issuance and future returns are not yet clear. for example.

According to the market timing hypothesis. Larrain. due mostly to the extreme observations. Summers. For these and other reasons. we get basically the same results as those reported throughout the paper. and Shleifer. for example. in practice. the level of control premium. 3. 3. Issuing too much equity in comparison with the firm's investment prospects can make the controlling shareholder's intentions apparent to investors and eliminate the overvaluation. it can produce a much lower price and eliminate the advantage of the issuance. 1997). Pontiff. as in the asymmetric information model. Issuances that do not imply the controlling shareholder's dilution (for example. if the demand curve is steep. Control is clearly an endogenous variable and needs to be taken into account in our analysis. Most large firms in Chile are listed. and Waldmann. 1995). a private negotiation between the controlling shareholder and another large shareholder is likely (Zingales. we collect stock prices and financial statements from readily available data sets. If we winsorize returns. / Journal of Financial Economics 109 (2013) 661–681 constraints. La Porta. Volpin.4 Our sample consists of approximately 21 thousand firm-month observations. even when fully convinced of their overvaluation.1. The price at the moment of the transaction reflects a control premium to compensate for the private benefits associated with control but. Wurgler and Zhuravskaya. 1984). Chile is. 2002). Second. 4 The reason for trimming returns is that some extreme observations are most likely the result of coding errors in the database. volatility (Pontiff. Table 1 provides summary statistics for the main variables used in return regressions. First. harder to get away with overvaluation when the controlling shareholder is selling old shares and receives the proceeds directly as compared with cases of equity issuance when the proceeds go to the company. Mayer. and Wagner. the lack of close substitutes for a stock (Shleifer. but no reason exists for block sales except overpricing. where many large firms are privately held (Franks. By the same token. less than 1% of the equity issues in our sample entailed a transfer of control so. Although the main trade-off between the controlling shareholder's dilution and investment is present in both models. when the controlling shareholder's intentions are more transparent. the standard deviation of raw (untrimmed) monthly returns is 755%. and the fact that most capital is managed in delegated portfolios (Shleifer and Vishny. long-run return predictability implies under reaction to new information instead of an instantaneous adjustment (Loughran and Ritter. 2006). Similarly. when the controlling shareholder subscribes new shares at pro rata) are a sign that the company is not overvalued and should not predict poor returns. they like the firm to issue shares for investment. investors are not able to infer the overvaluation from the decision to issue equity. For example.5% and also winsorize other variables at the top and bottom 1% to eliminate the effect of outliers. We have assumed so far that. The sample covers 85% of Chilean stock market capitalization in the average year. and the overall level of ownership concentration (Djankov. the controlling shareholder must decide how much to sell to maximize personal benefit. a more precise prediction of the market timing hypothesis is that future returns are poor only in cases of issuance with dilution. F. Chile's aggressive privatization program in the 1980s and early 1990s explains companies' ample stock market representation. arbitrageurs can fail to act against overvalued stocks. If investors are optimistic about the firm's prospects. Still. It is. in the former. Long-run return predictability distinguishes the market timing hypothesis from the asymmetric information model (Myers and Majluf. poor subsequent returns should not be expected. other limits to arbitrage include the same slow reversion of sentiment (DeLong. 1990). similar to other emerging and developed economies in terms of legal protection of investors. instead of trimming. 1986). this does not imply long-run predictability after control transfers. the controlling shareholder is indifferent between diluting or not if the market value coincides with his own valuation and does not dilute if the market value is below his own valuation. The amount that can be sold is limited by the controlling shareholder's desire to avoid adverse inference about the firm's prospects (Myers and Majluf. nonetheless. the assumption that the controlling shareholder maintains control is realistic. Data There are two data sources for our tests. The latter model predicts an instantaneous decline in stock prices when a firm announces an equity issue because the market immediately understands that the assets-in-place are worth less than previously expected. and Watanabe (2009) for a similar treatment. Similarly. The data on stock prices and financial statements used are obtained from Economatica. López-de-Silanes. Urzúa I. perhaps. 2008). the level of due diligence and bargaining power on each side of the transaction make it less likely that shares are overvalued. we hand-collect information about ownership structure from several sources.664 B. in contrast to other emerging markets and even some developed markets such as Germany. However. 1995). France or Italy. In those few cases in which control is at stake.45% (1%) with a standard deviation of 11%. However. poor post-issuance returns should not be expected when investors are sophisticated because they understand the controlling shareholder's incentives and adjust prices immediately. We trim returns at the top and bottom 0. 2012). The mean (median) monthly return in dollar terms is 2. after issuance. Shleifer. In this case. the key difference between market timing and asymmetric information is that. Stock prices and financial statements Our sample covers almost all non financial Chilean companies listed on the Santiago Stock Exchange between 1990 and 2009. We exclude only highly illiquid and small listed companies such as country clubs and schools. 1984) and. Thus. See McLean. the controlling shareholder retains control and the large private benefits associated with it. 1986. . by a downward-sloping demand curve for the stock (Shleifer. with financial companies accounting for most of the remaining 15%. the frequency of initial public offerings.

plant. 2005). Urzúa I. blockholdings are measured for insiders (officers and directors) as an anonymous group.21 12. and Stulz (2007).40 10.02 0. be checked by hand to understand their ownership structure. and Watanabe. These include ROE.228 1. López-de-Silanes. In our sample.svs. and Yeung. is controlled by the Claro family through a pyramid containing two listed companies (Elecmetal and Cristalerías) and several .). Size (ME) is the natural logarithm of total market equity (in dollars) at the end of June of each year.65 1. Firms' annual reports must.357 20. and equipment between two consecutive years) over assets.65 2. 2008.19 0. when available). therefore. Issuance (ISSUE) is defined as the log-change in the number of shares outstanding comparing the end of December of two consecutive years. 0. 1999. 2008). including the US. indicating the number of shares held by each.133 0. Fecus Plus. we can also identify the presence of multiple classes of shares with different voting rights. use a shorter sample (16 years) in their study of ownership dynamics in US firms. Pirinsky. Wolfenzon.g.80 0. Similarly to McLean. The sample covers nonfinancial Chilean firms from 1990 to 2009.26 0. however. From 1990 to 2003. Because the 12 largest shareholders are almost always other companies (some of them listed. allowing us to determine when the stake is diluted. the privatization of social security. the website of the Superintendencia de Valores y Seguros (the Chilean stock market regulator. a standard mechanism for achieving control in many emerging and developed countries (La Porta. the natural logarithm of the previous year-end book-to-market ratio (BM).g.35 0. and Shleifer. and Watanabe (2009) report a mean value of 5. 2009. the past six-month stock return (MOM).8%. Morck. In an example that serves to illustrate our methodology. Listed companies in Chile are required by law to disclose their 12 largest shareholders in their annual reports. this information is in itself little help in identifying a company's ultimate controlling shareholder. McLean. / Journal of Financial Economics 109 (2013) 661–681 665 Table 1 Summary statistics for variables in return regressions.22 0. Pontiff. Approximately one-third of the firms in our sample are controlled through pyramids.01 0. To the best of our knowledge.04 −0.05 0.82 11. we obtain the 12 largest shareholders from two private databases: Fecus Plus and Economatica. In their sample of 41 countries.10 0. Annual reports from 2004 onward are publicly available at www. we find that issuance is highly skewed to the right. From the annual reports. Returns are trimmed at the 1% level. Data are taken from Economatica. Ownership data The real challenge for our tests is to obtain data on ownership structures. The market beta is defined as the regression coefficient of a firm's stock returns on the Chilean market return over the previous 24 months (from month t−24 to month t−1. The mean value of 4% is above the 75th percentile. and the log change in the number of shares outstanding adjusted for stock splits in the previous calendar year (ISSUE). and Stulz (2007) or Foley and Greenwood (2010).11 0. Larrain. annual reports provide additional information (e. gaps exist in the price series. we compute the fraction of shares held by the controlling shareholder in each firm between 1990 and 2009. but Chile experienced a unique transition in these two decades (1990–2009) when per capita GDP almost doubled in dollar terms and tripled in purchasing power parity terms. etc.94 0. the book-to-market ratio (BM) is the natural logarithm of the book value of equity divided by its market value in December of each year.63 −0. F.246 21. Variable Monthly returns Annual returns Beta ME BM MOM ISSUE Number of observations Mean Standard deviation 25th percentile Median 75th percentile 21.71 0.137 20..00 0. The definition and timing of variables follows the recent literature on issuance and returns (Fama and French.8% corresponds approximately to the 83rd percentile. the regression coefficient of stock returns on the Chilean market return over the previous 24 months (beta). and Watanabe (2009).32 0. such an extensive database would be difficult to assemble in other countries.098 19. private). not common in Chile (fewer than ten firms in our sample). which is zero.13 −0.11 0. and Superintendencia de Valores y Seguros (SVS). Annual reports explain whether control is exercised through one holding company that owns all the controller's shares or alternatively through several firms related to the controlling shareholder. We have slightly fewer observations for this last variable because it requires continuous data over the previous six months and. With all this information. the natural logarithm of end of June market value (ME). dividends over book equity.3% and a 75th percentile of 0. These are.77 −0. macro stability. Helwege.2. total assets. management and board composition) that also helps to identify the ultimate controlling shareholder.06 0. others. one of Chile's largest winemakers. trade openness. Pontiff. The underlying reasons for this impressive economic boom were the structural reforms implemented by the government of General Augusto Pinochet in the 1970s and 1980s (e.53 −0.25 0. This table reports aggregate summary statistics for one-month and one-year holding period returns. McLean. Viña Santa Rita.35 −0.12 0.B..81 11. Pontiff. and capital expenditures (the change in property.04 0. for some small firms. Pirinsky. for example.07 0. Pontiff and Woodgate. Table 2 provides summary statistics for other variables derived from companies' annual balance sheets and income statements.950 20.00 Average returns are high. and momentum (MOM) is the buy-and-hold return over the previous six months (from month t−7 to month t−1). 3. Finally. Shares outstanding are adjusted for splits.00 0.cl. We are also able to identify the controlling shareholder by name and size of stake in a precise way. hereafter SVS) and a few companies also post older reports online.39 0. All other variables are winsorized at the 1% level. In other work on ownership structures such as Helwege.

09 0.00 0. See the Appendix for further information about cash flow rights in the Chilean market. the natural logarithm of book assets.09 0. holds 55% of Santa Rita. cash flow rights.946 2.38 0. and Wagner (2012) show evidence of the persistence of control in continental Europe. at least a six-month window between issuance and returns.15 0.04 0.04 10. have lower subsequent returns (footnote continued) a sample of Israeli SEOs.01 0. Future returns consider those from July of year t through June of year t+1.133 0.10 −0.03 −0.10 12..25 0.e. Larrain. 4. in other words.33 0.1. dilute his stake by not exercising or selling these rights or increase it by buying rights from other shareholders. and a dummy variable that identifies firm-year observations with a change in the SCS. the private benefits of control are large in Chile. almost 99% of companies in our sample are controlled by a large shareholder. we report summary statistics for the frequency of large changes (more than 75%) in the controlling shareholder's stake as a result of either the issue of new shares or the sale or purchase of blocks of shares. which holds 34% of Cristalerías. Data are taken from Economatica.00 0.68 0.00 0.18 2. .23 −0. Pontiff and Woodgate (2008). / Journal of Financial Economics 109 (2013) 661–681 Table 2 Summary statistics of annual firm-level characteristics and ownership variables. that control is achieved with a stake larger than 20% (Adams and Ferreira.07 0.32 0. 1999).79 0.05 0. the difference between the SCS and cash flow rights.08 0.00 Panel B: Ownership variables Stake of controlling shareholder (SCS) Cash flow rights SCS—cash flow rights Dummy for change in SCS intertwined privately held companies.11 0. leverage.07 3.11 0. asset growth.01 0. However.12 0. Cash flow rights are slightly below the stake of the controlling shareholder. This timing convention follows the work of Fama and French (2008). 2008.03 0. Variable Panel A: Firm-level characteristics ROE Log book assets Leverage Debt growth Asset growth CAPEX/assets Dividends/book equity Number of observations Mean Standard deviation 25th percentile Median 75th percentile 2. can be determined either by multiplying all ownership stakes in the pyramidal chain or by multiplying the stake that the controlling shareholder holds in each share class by the dividend rights of each class. Table 3 reports average future returns for different equity issues according to what happens to the stake of the controlling shareholder.078 3.02 0. All variables are winsorized at the 1% level. The sample covers nonfinancial Chilean firms from 1990 to 2009. when the year of issuance is year t−1 and there is.24 0. as is standard in the literature on control. Panel B shows summary statistics for the stake of the controlling shareholder (SCS). implying that the effect of issuance we find is a medium or long-run effect. which. In the last row of Table 2. The controlling shareholder's mean and median stake is about two-thirds. Kraizberg.97 0.03 0.04 0. controls 55% of the shares of Santa Rita. Considering only links through listed companies. capital expenditure (CAPEX) as a fraction of total assets.83 0. Average returns and portfolios Chilean law gives shareholders in publicly listed companies preemptive rights on a pro rata basis.00 0.072 3.030 3.889 0. with only two out of the 325 issuances in our sample implying a transfer of control.54 0. i. 4. Ciccotello. Volpin.822 1.819 2. We find that issues that result in substantial dilution.5 This is probably related to the 5 Our evidence on the persistence of control throughout share issuances fits well with the results of Hauser. ranging between 16% and 23% of the market price. the fraction of dividends received by the controlling shareholder. According to Dyck and Zingales (2004) and Nenova (2003). 2010). 12% of the observations in our sample correspond to large changes and are almost evenly split between positive and negative changes.01 0.17 13.666 B. López-de-Silanes. in turn. La Porta.42 0. Control is very persistent despite changes in the controlling shareholder's stake. Panel B of Table 2 shows summary statistics for ownership variables.61 0. and dividends as a fraction of book equity. and Dahan (2003). the Claro family.20 1. debt growth.59 0.51 0. F. and others in predictive regressions.33 0.071 2.16 0.00 0.27 0. and Shleifer.68 0. Urzúa I. This assumes. who show that controlling shareholders tend to stay in control in sizable private benefits associated with control. The Claro family directly owns 50% of Elecmetal. Franks. The controlling shareholder can retain his stake by simply exercising his rights. therefore. These are intended to protect minority shareholders against dilution as shown by recent evidence (Atanasov. Mayer. the Claro family's stake increases to 78% once holdings through privately held companies are taken into account. Low returns after issuance with dilution In this section we explore the behavior of returns after issuance by forming portfolios and then at the firm level with cross-sectional regressions.50 0. defined as a decrease in the controlling shareholder's stake that is equal to or larger than 5%. and Superintendencia de Valores y Seguros (SVS). Fecus Plus.002 2.990 3.10 12.20 0. with an average difference between the two of 9%.13 0. Cash flow rights. and Gyoshev. Black. Panel A shows annual summary statistics for return on equity (ROE). According to this definition.37 0. Following the 20% rule for assigning control.

37% 11. Panel A: Issuance statistics considering the full sample Stake of controlling shareholder (SCS) Standard deviation of monthly returns Average annual returns Standard deviation of annual returns Average ISSUE Standard deviation of ISSUE Number of monthly observations Percentage of full sample 0.56% 1.06% 138.94% 2.28% 13.63% 2.Table 3 Summary statistics for firms issuing equity.82% 70. Urzúa I.41% 16.32% 11.06% 16.06% 11.43% 37.85% 11.06% 11.92 3.20% 2. and the percentage of the full sample represented by firms issuing equity.91% 41.99% 38.02% 9.82% 12. Fecus Plus. These firms are split into five groups according to changes in the stake of the controlling shareholder (SCS) caused by the equity issue.58% 12.00% 1.82% 15.10% 52.33% 2.43% 42. Larrain.54% 1.75% 64. / Journal of Financial Economics 109 (2013) 661–681 Decreases by more than 5% Decreases between 0% and 5% Does not change Increases between 0% and 5% Increases by more than 5% Average monthly returns 667 .43% 11.23% 10.91% 46.10% 24.47% 0.83% 36.39% 36.15% 12.09% 2.60% 14.11% 10. Panel A shows the mean and standard deviation of monthly and annual returns.45% 1.23% 44.39% −5.83% 52.91% 40.04% 32. Data are taken from Economatica.18% 2.50% 11. Panel B shows the same statistics but splitting the sample between small and large issues (using the median of shares issued) and between firms with a low and a high stake of the controlling shareholder (using the median SCS).03% B.13% 38.65% 9.56% 12.57% 60.95% 26.10% 12.10% 5.68% 33.13% 5.15% 76.23% 12.89% 11.35% 62. and Superintendencia de Valores y Seguros (SVS).64% 21.31% 12.42% 41.72% 76.54% 2.77% −6.63% 10.90 6.71% 6.58% 22.08% 37.75% 7.58% 2. the number of monthly observations.25% 11. the log change in split-adjusted shares outstanding (or ISSUE).20% 11.61% 2.40% 3.51% 7.25% 589 615 1468 713 510 2.89% 15.85% 3.19% 36.40% 12. F.46% 9.99% 11.68% 27.36 2.43% 17.03% 1.92% 98.13% 12.90% 42.72% −3.95% 67.29% 2.84% 46.16% −3.40 Panel B: Issuance statistics splitting the sample by size of issue and SCS Small issues Stake of controlling shareholder (SCS) Decreases by more than 5% Decreases between 0% and 5% Does not change Increases between 0% and 5% Increases by more than 5% Large issues Average monthly returns Standard deviation of monthly returns Average annual returns Standard deviation of annual returns Average monthly returns Standard deviation of monthly returns Average annual returns Standard deviation of annual returns 0.89% 2.76% 2.16% 19. The sample covers nonfinancial Chilean firms from 1990 to 2009.44% 11.17% 38.96% 8.30% 143.77 2.19% 8.91% 2.43% Low SCS Stake of controlling shareholder (SCS) Decreases by more than 5% Decreases between 0% and 5% Does not change Increases between 0% and 5% Increases by more than 5% High SCS Average monthly returns Standard deviation of monthly returns Average annual returns Standard deviation of annual returns Average monthly returns Standard deviation of monthly returns Average annual returns Standard deviation of annual returns 0.42% 11.49% 9.15% 12.00% 11.36% 1.

the effect of issuance on returns is non-monotonic: issues in which the stake of the controlling shareholder decreases perform abnormally badly. A share issuance in 1992 implied a drop in the Claro family's stake from 95% to 73%. . from July of year t+1 through June of year t+2. so there is a single observation for each year. a spread of about 25% emerges between issuers that dilute and other issuers. in 2000. because we can identify an exact date of issuance (day and month within the year) for only 56% of the events in our sample. The spread. Issuance occurs in year −1.46% in equity issues with substantial dilution and 7.4 0. A repurchase from minority shareholders would be different because. 2 uses fewer observations than Fig. In the year after issuance (year 1). and other issues have average returns of about 2–2. Issuers that dilute strongly have very high returns the year before issuance (year −2) compared with other issuers. from July of year t through June of year t+1. In Fig. in that case. The average return in the following 12 months (July 2001 through June 2002) was 3. but share issuance with a strong dilution of the controlling shareholder does predict low returns. again without diluting the Claro family. distinguishing between those with and without a strong dilution of the controlling shareholder's stake (with dilution and without dilution). In 1996.6 After approximately 15 months from issuance.8 0. 2 shows similar patterns as the previous figure but in months from the effective date of issuance. For instance. the average size of the issue is 6. A similar pattern can be seen in annual returns (July of year t through June of year t+1). Overperformance would suggest that the controlling shareholder is able to increase his stake by buying undervalued shares. Fig. we split the sample by the size of the issuance and by the prior stake of the controlling shareholder (splitting by the sample median in each case). it is more likely that issues in which the controlling shareholder's stake increases are priced fairly.10% in those without a change in the stake of the controlling shareholder. The figure shows the accumulated return difference (in percentage points) between issuers with and without a strong dilution of the controlling shareholder. new shares are not issued. Issuance occurs in year t−1. is the first year after portfolio formation. year two is the return from July of year t+1 through June of year t+2. As shown in Table 3. than other issues. 2. in line with the full-sample average. however.7% of shares outstanding. due to legal restrictions. Year one corresponds to the average annual return between July of year t through June of year t+1. or than the return regressions. given that these gains have to be shared with minority investors. monthly return of Santa Rita in the subsequent 12 months (July 1993 through June 1994) was a paltry −5%. the option always exists of not carrying it out and seeking alternative sources of financing. Portfolio formation is in June of year t. / Journal of Financial Economics 109 (2013) 661–681 1. The x-axis shows months from the effective month of issuance (month 0). who can also subscribe the issuance because of their preemptive rights.54%. a second issuance of 23% of shares outstanding took place without diluting the Claro family. 1. Finally. The average 6 Fig.5%. 3 we illustrate the main point with the example of Santa Rita. Urzúa I. The returns of issuers that do not dilute are fairly stable throughout the event window while. The average monthly return in this category is 0. In other words. and so on. Accumulated return difference between diluting issuers and other issuers in event time. there was another equity issue of 6. the accumulated return difference between diluting issuers and other issuers stabilizes at around 15% and there is no noticeable reversion in this pattern.5%. However. if the issue is severely underpriced. Larrain. Fig. The x-axis shows years from portfolio formation (June of year t). For the sake of visual clarity. Empirical evidence in several markets suggests that repurchases predict future return overperformance (Peyer and Vermaelen. for issuers that dilute. the results do not vary significantly with the controlling shareholder's stake before the issuance. 1 illustrates the return patterns in event time. Average annual returns of issuers in event time. The figure shows average annual returns for firms that issue shares. 2009) but. Poor returns are concentrated among issuers in which the controlling shareholder is strongly diluted in small and large issues alike.2 Years from issuance Fig. share issuance per se does not predict low returns. subsequently seems to disappear. The average return in the following 12 months (July 1997 through June 1998) was basically 0%. Year one. In Panel B of Table 3. In other words. is the second year after portfolio formation.2 10 1 Issuance without dilution Annual returns 0. they increase enormously before the issuance and decrease afterward. which had three different equity issues in the sample period. we present annual returns averaged across issuers. but issues in which the stake of the controlling shareholder increases do not perform abnormally well. Similarly. Fig.668 B. Table 3 shows that the average size of the issue does not have a clear correlation with the change in the controlling shareholder's stake. Year two.2 0 -3 -2 -1 1 2 3 Accumulated return (%) Issuance with dilution Accumulated return difference between issuers with and without dilution 5 0 -3 0 3 6 9 12 15 18 21 24 27 30 33 -5 -10 -15 -20 -25 Months from issuance -0. they are rare in Chile. 1. F. and so on.6 0.

t−j þ eMOMi. momentum (MOM). the spread between issuers with and without dilution is 1. The mean is the average return over the 12-month window.t−j þ cMEi.t−j Þ þ εi.t−j þ eMOMi.74 (¼−0. Average returns at the portfolio level should be robust to distortions caused by isolated market events that could. (2) captures the extra marginal effect of issuance when the SCS is significantly reduced. Residuals in this regression are allowed to be heteroskedastic and clustered by month (or by year in the case of annual returns).2 -0. In the third column of Table 5. size (ME). 8 The value effect associated with BM has been shown in international stock markets by Fama and French (1998). affect the summary statistics presented in the previous table. and Watanabe (2009) in their sample of 41 markets.t ð1Þ ð2Þ where Ri.t−j þ cMEi. and momentum. Monthly returns of Santa Rita (1992–2009).14% in future average returns.62% in future returns in the group of issues with strong dilution of the controlling shareholder. Portfolios have the advantage of collapsing all observations at a particular date into a single return.47). In the US.8 Adding share issuance does not significantly affect the magnitude of the four previous variables.t−j þ f ISSUEi. The stake of the controlling shareholder did not decrease in the other two share issuances.3 0.1 0. Beta is computed with a rolling window of the 24 months prior to the return.35% (t-statistic of 2.59). We also interact ISSUE with a dummy for those issues in which the stake of the controlling shareholder (SCS) decreases by 5% or more.t−j þ f ISSUEi. while the Fama-Macbeth estimator weighs all cross sections equally. but not of beta. which implies that the total effect of ISSUE in cases with a large decrease in the SCS is −4. Other issuers and nonissuers have returns of about 2% on average. .1 -0. The coefficient at is a time fixed effect. This is important given the relatively small number of stocks we have when compared with applications in the US market. Pontiff and Woodgate (2008) find a post-issuance return decline of 0. For example.2. 2008). 3.t−j Ri. Return regressions The basic panel regression is: Regression results are reported in Table 5. and issuance from December of year t−1.27.t−j þ εi.27–4. the coefficient of ISSUE alone (f) falls in magnitude to −0.t ¼ at þ bβi. This dummy focuses on changes in the SCS that occur through issuance and does not cover those that occur through the sale of preexisting shares. A 1 standard deviation increase in issuance leads to a decline of 0.13) predicts a decline of 0. The panel estimator allows us to (footnote continued) give more weight to richer cross sections. The results for value-weighted portfolios and annual (non-overlapping) returns are similar. This means that returns from July of year t through to June of year t+1 are regressed on size computed in June of year t. and share issuance (ISSUE) are as defined previously with the timing conventions that follow Fama and French (1992.33%.2 0. As expected from Fama and French (1992. book-tomarket (BM). ISSUE has a coefficient of −1. Table 4 presents average returns for portfolios of different issuers. there is a significant effect of size.t−j þ dBMi. The figure shows Santa Rita's monthly stock return between 1992 and 2009. Our sample has 74 stocks with full data in 1990 and 134 stocks with full data in 2009. momentum from the previous six months.4 Monthly return 0.B. The highlighted windows represent periods of 12 months (from July of year t through June of year t+1) after a share issuance in year t−1.5 0. theoretically. the book-to-market ratio of December of year t−1.t ¼ at þ bβi. value (BM). however.t þgðISSUEi. 2008) and Pontiff and Woodgate (2008). The spreads between portfolios are generally large and significant. Larrain.t is the dollar return on stock i in month t.23).47 (t-statistic¼ −3. Pontiff. The regression is: Ri.81%. This return decline is exactly the same as that found by McLean. Beta.45). The first window corresponds to the period following a share issuance when the controlling shareholder was diluted.t−j þ dBMi. Coefficient g is −4. which implies that a 1 standard deviation increase in issuance (0. Urzúa I. / Journal of Financial Economics 109 (2013) 661–681 669 0. F.t−j  Dummy Iss w=Decrease SCS Larger 5%i.7 Coefficient g in Eq. These results imply that the 7 Fama-Macbeth regressions give similar results to the panel regressions reported throughout the paper.10 (t-statistic ¼−1. We find that the equally weighted portfolio of issuers with a large decrease in the stake of the controlling shareholder has an average return of only 0.0 -0.3 Fig. 4.

37% (1.44).91% (4.38) 2.83) 1.46% (6.69% (4.31% (4.04) 30.71) 16. The sample covers nonfinancial Chilean firms from 1990 to 2009.91% (4.61% (1.36) 1.66) 1.25) 24.62) 20.05) 24.18% (2.38% (2.24% (2. either as a result of issuance or a block sale of pre existing shares [“Dummy Decrease SCS Larger 5% (Iss or Bs)”]. A squared term for ISSUE in Column 4 gives results different from the interaction of ISSUE and the change in the SCS.36) 0. Fecus Plus.. although with fewer observations to avoid overlap. In Column 6.15) 3.22% (2.30) – – 15. in Column 7.81% (1.10% (5.28% (2. Results show that an equity issue with a substantial decrease in the SCS predicts a decline in future returns of 1.56% (0.77) 19.49% (3.21) 2. These portfolios are split into six groups: five according to changes in the stake of the controlling shareholder caused by the equity issue and a sixth corresponding to the no-issuance portfolio.59) 1.90% (2. to indicate the five groups of issuance from decreases to increases in the SCS.68) – – 1.96) 35.14) 38.77) – – 11.03% (1. the stake itself does not affect the impact of issuance. we study the robustness of the previous result to several changes in the regression specification.70) 7. Data are taken from Economatica.03% (2. implying that each percentage point of reduction in the SCS adds −0.82% (1. We next interact ISSUE with the SCS prior to issuance. We first interact ISSUE with the change in the SCS (i. Finally. The variable ISSUE. Urzúa I.13% (4. As suggested by Table 3 and confirmed in this regression. we include a dummy for all the reductions in the SCS larger than 5%.98% (2.e.06% (3.91) 19.18% (0.01% (3. This table shows average returns of equal.49) 1. Columns 8–14 show the same robustness for annual returns and the results are similar. instead.75) 1.10% (4.12) 1. we control for the change in the SCS while also including our main issuance variables. and Superintendencia de Valores y Seguros (SVS).47) 2. with the continuous variable and not the dummy variable). 4. which does not modify the main conclusion (Column 5 shows the regression with the change in the SCS alone for comparison).67).09% (2.39) 14.18% (3. This shows that the effect we find is not merely a non-monotonic effect of issuance.94) 15.21% (1. Because these are dummy variables.52) 25. The table also shows t-tests comparing all portfolios with the portfolio with the largest dilution of the SCS. Robustness In Table 6.06% (0.63% (2. .46) 10.67% (2.06) – – 1.62% (0. Equal-weighted portfolios Portfolios Monthly returns Portfolio (1)—SCS decreases by more than 5% Portfolio (2)—SCS decreases between 0% and 5% Portfolio (3)—SCS does not change Portfolio (4)—SCS increases between 0% and 5% Portfolio (5)—SCS increases by more than 5% Portfolio (6)—No Issuance Annual returns Portfolio (1)—SCS decreases by more than 5% Portfolio (2)—SCS decreases between 0% and 5% Portfolio (3)—SCS does not change Portfolio (4)—SCS increases between 0% and 5% Portfolio (5)—SCS increases by more than 5% Portfolio (6)—No Issuance Value-weighted portfolios Average return (t-statistic) Return spread with portfolio (1) (t-statistic) Average return (t-statistic) Return spread with portfolio (1) (t-statistic) 0. instead of with the change in the SCS. the key variable that modifies the effect of issuance is the change in the SCS (or the change relative to the previous level as shown in Column 3 of Table 6). paint a similar picture in terms of both the magnitude and statistical significance of the coefficients.11% (0. In the fourth column of Table 5.37) 2.73) 1.53) 32.3. the coefficient attached to them is the average effect of each type of issuance on future returns (once we control for the other variables in the regression).55) 0. F.08% (1.43) 2.01) 1.670 B.06) 0. Larrain.16% (3.76) 23. Instead.31% (2.24% (1.80% (4. instead of the continuous variable ISSUE. Our main variables of interest remain significant and the effects are even larger. we use a set of dummy variables. which is equal to −1.10% (2.and value-weighted portfolios.88) 24.18% (1.35% (2.81% (0. Regressions with annual returns.32 (t-statistic ¼−1.72% (t-statistic¼ −3. As noted in Table 3.36) 2. none of them is statistically significant or comparable in magnitude. captures the marginal effect of issuance. / Journal of Financial Economics 109 (2013) 661–681 Table 4 Average returns of portfolios formed according to share issuance and change in the stake of the controlling shareholder (SCS). The coefficient on the interaction is significant at the 5% level.09) 13. Although some of the other types of issuance also have negative coefficients.60% (0.15) 2.089 to the coefficient of ISSUE.38) predictive power of ISSUE comes almost exclusively from the observations with dilution.41) 27. the effect of the SCS is non-monotonic: Issues with an increase in the SCS do not predict abnormally positive returns as opposed to the negative returns predicted by decreases in the SCS.

507) Dummy Iss w/Decrease SCS larger 5% Dummy Iss w/Decrease SCS between 0% and 5% Dummy Iss w/o Change in SCS Dummy Iss w/Increase in SCS between 0% and 5% Dummy Iss w/Increase in SCS larger 5% Time fixed effects Number of observations R-squared Annual returns Yes 19.183nnn (0. we perform some of the tests we did for issuance for these block sales.859) 6. Panel regressions of monthly and annual returns (both multiplied by one hundred) on the regression coefficient of stock returns on the market return over the previous 24 months (beta).552 0.586 (3.234 (0.710) 3.125) 0.689) −9.017nnn (0. particularly if it occurs in a hot issuance market.053 (5.581 (2.826) 2. holding approximately 10% of all shares.229 (0. All regressions include month or year fixed effects. In block sales.522 0.788 (0.055) 0. the natural logarithm of the previous year-end book-to-market ratio (BM).149) 0. they like the firm to issue shares for investment.812) −4.309 Yes 1.348) 2. the interaction between ISSUE and a dummy variable that identifies observations of ISSUE with a decrease in the stake of the controlling shareholder larger than 5%. Consistent with this idea.522 0. stock returns are not poor after block sales.114) 0.954 (4.238nn (0.603) Yes 1. Data are taken from Economatica. and Superintendencia de Valores y Seguros (SVS).326nnn (0.404) −0. If investors are optimistic about the firm's prospects. Panel A in Table 8 is in the style of Table 3.312 Yes 1.768) 0.371 (2.908) 6.057) 2. our results in Columns 3 and 4 of Table 7 are stronger for the subset of firms that do not have pension funds in their shareholder base. and Panel B is in the style of Table 5.599) 6. The regression with annual returns uses non-overlapping observations. The regressions show that the interaction of ISSUE and the dummy for large decreases in the SCS is larger in magnitude and more statistically significant in the hot market subsample.200 (0.949) −64.204) 0.708) ISSUE ISSUE  Dummy Iss w/Decrease SCS larger 5% −1.207 (0. Significance at the 10%.124) 0. Urzúa I.015nnn (5.004) −0.393) 0.399) −0.968) 0. nn.056) −1.552 0.608 0. institutional investors in the Chilean market.522 0. Finally. In Table 8. domestic pension funds have become the largest.056) −0.335nnn (0.202nnn (0.004) −0.B. optimistic investors predominate in hot markets (Baker and Stein. and nnn.109 (0. Due to disclosure Yes 18.269 4.405 (19.322nnn (0.086) −3.054 (4.549) 2.103nnn (0.523 (2.456 0. we split the sample into subgroups according to several measures of investor sophistication at the time of issuance.849) −4.723nnn (0. and a set of dummy variables that identifies observations of ISSUE with other changes in the SCS. and arguably most sophisticated.087) −3. 5% and 1% is indicated by n. The sample covers nonfinancial Chilean firms from 1990 to 2009. We find that.245nnn (0.196nnn (0. naive. respectively. Investor sophistication In Table 7.614nn (24. We first divide it into hot and cold markets. The differential impact of hot and cold markets is in line with Mclean.279nnn (0. but no reason exists for block sales except .017nnn (0.988) 7. Pontiff.473nnn (1. This implies that issuance with dilution has a negative effect on future returns.321 (0.124) 0. the natural logarithm of end of June market value (ME).087) −3. the incentives are more apparent to outside investors. with the former defined as those months when the fraction of firms with positive issuance is above the sample median (16%).205) 0. the log change in the number of shares outstanding adjusted for stock splits in the previous calendar year (ISSUE).311nnn (0. from large block sales of more than 5% to large block purchases of more than 5%.270 Yes 18.273 (0.385) 0.691 (3.204) 0.239nn (0. We split changes in the stake of the controlling shareholder that are not associated with issuance into five groups.205nnn (0. contrary to the case in which the controlling shareholder reduces his stake through issuance. we have access to details of their stock portfolios in each period.202) 0.314 requirements. Fecus Plus. Larrain. 2004).321) −10. According to the behavioral literature.082) −3.004) −0. / Journal of Financial Economics 109 (2013) 661–681 671 Table 5 Return regressions: the effect of share issuance according to changes in the stake of the controlling shareholder (SCS).004) −0.911 (8. the past six-month stock return (MOM).328nnn (0. F.552 0.018nnn (0.898 (4.308) −0. we study the cases in which the controlling shareholder reduces its stake by selling old shares.272 Yes 18. and Watanabe (2009).029 (16.108) 0.468 (2.048 (0. Dependent variable Monthly returns Independent variables Beta BM MOM ME (1) (2) (3) (4) (5) (6) (7) (8) 0.225 (0.017nnn (0. Our second proxy for investor sophistication is related to institutional investors.122) 0.500) −0.623) 0.4. Since the privatization of social security in the early 1980s. We expect firms without pension funds in their shareholder base to have an easier time engaging in market timing because there is more room for sentiment and overvaluation in these firms.270 −24.514 (10.239nn (0.309 Yes 1.176nnn (0. Standard errors are clustered by time period.

721) 3.540 (2.90nn (46.64) −9.83 (21.12) 0.78) 0.297nnn (0.631nn (1.229 (0.81 (18. the ratio of change in the SCS over the SCS before the issuance.019 (19.06) −0. Fecus Plus.0172nnn (0.530 0. and ISSUE are as in Table 5.270 Yes 1.522 0.31nn (0.09) −3.229 0.72 (4.98nn (3.64 (3.198nnn (0.28nnn (28.27 Yes 18.27) 0.00) −0.306 Yes 1. Data are taken from Economatica.204nnn (0.12) 0.47) 2.303nn (0.33nnn (0.204) 0.149) 0. Standard errors are clustered by time period.60 (3.137) 0.33nnn (0. nn.293 (0.02nnn (0.270 Yes 18.69) −34.047nn (1.70) 23.530 0.124) 0.530 0.87 (41.20nnn (0.22 (0.17) 0.116nnn (1.02) 6. and nnn.31 Yes 1.31 Yes 1.02nnn (0.73) 2.05 (3.245nn (0.488) 1.315) Yes 18.0869) −3.238nn (0.69) 23.06) −1.522 0.077) 6.21) 0.36 (45.27 Yes 18.552 0.207 (0.229 0.09) −50.730 (4.229 0.230 (0.21 (2.25 (53.19nnn (0.629) −9.18nnn (0.145nnn (0.00) −0.24nn (0.31 Yes 1.339 0.64) 36.0169nnn (0.02nnn (0.11nnn (0. The regression with annual returns uses non-overlapping observations.02) 7.843) 0.60) −59.06) −1.504 (0. The sample covers nonfinancial Chilean firms from 1990 to 2009.48 (2.530 0.02nnn (0.16) 0.201nnn (0.811) 2.19) 117. and Superintendencia de Valores y Seguros (SVS).181) 0.11 (4.25nn (0.27 Yes 18.124) 0.13) 0.312 B.21) 0.204) 0.05) 6.308nn (0.12) 0.115 (19. F.24 (0.174 (0. Urzúa I.552 0.775) 0. Significance at the 10%.25nn (0.229 0.672 Table 6 Robustness checks: the effect of share issuance and changes in the stake of the controlling shareholder (SCS). 5% and 1% is indicated by n. and a dummy for decreases in the SCS larger than 5% and that occurred through issuance (Iss) or block sales (Bs).023) 6.09) −3.04 (2.648 (3. respectively.124) 0.98) ISSUE2 0.249nn (0.0168nnn (0.0560) −0.23 (0. the square of ISSUE.40 (3.0856) −3.20nnn (0.0867) −3.486) −0.09) −3.06) −1.494 (6.31) ISSUE  Dummy Iss w/Decrease SCS larger 5% 3.205) 0.47 (3.98) 2. The table also considers the interaction between ISSUE and four variables: the change in the stake of the controlling shareholder.90 (4. Larrain.444 (17.320nn (0.00433) −0.21 (0.75 (4.643 (3.0567) −0.12) 0.310) Change in SCS −3. All regressions include month or year fixed effects.23 (0.270 −4.27 Yes 18.84) 2. / Journal of Financial Economics 109 (2013) 661–681 Beta Annual returns .25 (2.64) −10.21) 0.32n (0.550) Yes 1. ME.20nnn (0.55 (17.70) nn ISSUE  (Change in SCS/SCS) 84.80nn (28. MOM.07nnn (0.68) ISSUE  SCS −1.17nnn (0.778 (4.00433) −0. the SCS before the issuance.920 (1.56) 2.16 (18.18) 0.309 −2.08) 6.43) −62.836 (22.858 (4.47) 2.24nn (0.21) 0.31nn (0.31 Yes 1.48nn (21.82) 0.643) 2. Dependent variable Monthly returns Independent variables BM MOM ME ISSUE ISSUE  Change in SCS (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) 0.79) 8.18) 0.095nnn (0.00) −0. BM.236 (0.538 0.78) 7.898) 6.0560) −1.09) −3.00430) −0. Panel regressions of monthly and annual returns (both multiplied by one hundred) where beta.385) Dummy Decrease SCS larger 5% (Iss or Bs) Time fixed effects Number of observations R−squared Yes 18.18) 2.00) −0.

and have more bargaining power than in issuances. we compare issuance and block sales in companies that had pension funds before and after the dilution.007) −0. the controlling Calderón family reduced its stake from 81% to 61% and the 20% block. we examine other firm characteristics.006) −0. respectively.032) 0.259 (0. Fecus Plus. the average decrease in the controlling shareholder's stake is 14.573 (0. which gives the right both to a board seat and to enter a shareholders' agreement. a potential confirmation of the market timing hypothesis. such as high past returns and . but only after 15% of issuances with strong dilution. and the behavior of pension funds to support the idea that outside investors in block sales are typically wealthier.308n (0. one of Chile's main retail chains. Hot and cold markets are defined with respect to the sample median of firms with positive ISSUE (16%). We provide three pieces of evidence related to boards of directors. We find that changes in its size (most likely the addition of one director) occur after 20% of block sales. The dynamics of these characteristics around issuance can help paint a fuller picture of market timing or. we do not expect to see new blocks of ownership if only retail investors acquire the shares sold by the controlling shareholder but rather when large investors are involved.660 (3. shift the balance toward other explanations.054 (0. Changes in board composition. Third.247) 0. Larrain. was acquired by the Saieh family. we find that a firm goes from having an average of 2. we look at boards of directors in the year of the issuance or block sale in comparison with the previous year. / Journal of Financial Economics 109 (2013) 661–681 673 Table 7 Subsamples: the effect of share issuance and changes in the stake of the controlling shareholder (SCS). the natural logarithm of the previous year-end book-to-market ratio (BM). in this section.186nn (0. We find that pension funds buy on average (median) 34% (21%) of the shares in issuances and 43% (37%) in block sales. but we believe it points in the same direction as the previous literature on block sales (Barclay and Holderness. In other words. are more sophisticated. Firms before an issuance with dilution have approximately the same average number of blocks (2.200 (1. 1991). the number of ownership blocks. Hot markets have more firms with positive ISSUE than the median. Firm characteristics before and after equity issuance The market timing hypothesis focuses on long-run returns but. the past six-month stock return (MOM).B. This evidence is only suggestive.019nnn (0.018 (0. The absence of poor returns after block sales is. which is based on investors not understanding the intentions of the controlling shareholder.005) −0.299 0. In the case of block sales. Second. .321) 0.261) 0. as in Barclay and Holderness (1989.473nnn (1.774) −2.325 overpricing.290 6. Dependent variable: monthly returns Subsample of observations with Independent variables Beta BM MOM ME ISSUE ISSUE  Dummy Iss w/Decrease SCS larger 5% Number of observations R-squared Cold markets (1) Hot markets (2) Zero institutional ownership (3) Positive institutional ownership (4) −0.172 (0.006) −0.246 9.025nnn (0. pension funds—the largest institutional investors—buy more shares in block sales than in equity issues.353) −6. is a clear example of a block sale that involved another large shareholder.213 0. Data are taken from Economatica. Significance at the 10%.161) 9. The sample covers nonfinancial Chilean firms from 1990 to 2009. Panel regressions of monthly returns (multiplied by one hundred) on the regression coefficient of stock returns on the market return over the previous 24 months (beta).191nn (0. Standard errors are clustered by time period.032 (1. are also more likely after block sales than after issuances with dilution: 64% versus 47%. the natural logarithm of end of June market value (ME).294nnn (0. and Superintendencia de Valores y Seguros (SVS). Investors in block sales are also more likely to be sophisticated.185) 0. the log change in the number of shares outstanding adjusted for stock splits in the previous calendar year (ISSUE).152) 0.142) −4.528nnn (0. 1989.081) −1.015nnn (0. when the controlling shareholder reduces his stake by selling old shares.685) 0. 5. and nnn. Ripley. if they fail to match the predictions of the market timing hypothesis. All regressions include month fixed effects. the market timing literature suggests that proxies for overvaluation and investor sentiment. In terms of pre-issuance characteristics.617nn (0.067 0. and the interaction between ISSUE and a dummy variable that identifies observations of ISSUE with a decrease in the stake of the controlling shareholder larger than 5% (Dummy Iss w/Decrease SCS larger 5%). therefore.309 0. Urzúa I. First.498 (0. 5% and 1% is indicated by n nn . When considering blocks larger than 5%.007 (0. In 2009.047nnn (1.079) 1. Institutional ownership refers to the ownership stake of domestic private pension funds in a company.955) −3.249) 0.25 blocks afterward.87 blocks before the block sale to 3.05). F.137 (0.159) 0. This suggests that.766 (2.618) 0.3% or precisely the amount required for a seat on a typical seven-member board.092) 0. 1991). without a change in its size.180) 0.215 12.080) −0. they are on average acquired by larger investors.85) but not as many blocks after issuance (3. Changes in the board are likely to indicate the presence of a new large shareholder.398nnn (0.

95 1.089 (0. Fisher. the natural logarithm of end of June market value (ME).058 (0.752 (7.323nnn (0. In this model.726) Yes 18.709) −64. Pirinsky. 2007).810) −4. Fecus Plus. the number of monthly observations. a lower than normal future ROE is consistent with a reversal Annual returns 0.557 (19.270 Yes 1. and nnn. Panel A shows the mean and standard deviation of monthly and annual returns.522 0.124) 0. and a set of dummy variables that identifies observations of block sale or purchase with other changes in the SCS.213 (0.067 (3. Panel B shows panel regressions of monthly and annual returns (both multiplied by one hundred) on the regression coefficient of stock returns on the market return over the previous 24 months (beta).312 of investor optimism.17% 25.18% 74. Urzúa I.366) 1.652 (9.102) 3.084 (0. the company's . the model does not make clear predictions regarding post-issuance outcomes such as profitability (ROE) or capital expenditures.096 (4.334) 0.91% 11240 52.89% 10.44% 25.92% 12. Equity issuance is zero in all of these groups.93% 1199 5. respectively.004) −0.017nnn (0. the log change in the number of shares outstanding adjusted for stock splits in the previous calendar year (ISSUE).82% 2780 13.10 2. 5% and 1% is indicated by n.204) 0. the interaction between ISSUE and a dummy variable that identifies observations of ISSUE with a decrease in the SCS larger than 5%. F.386) 2.921 (4.056) ISSUE ISSUE  Dummy Iss w/Decrease SCS larger 5% Block sale of more than 5% Block sale between 0% and 5% No block sale and no issuance Block purchase between 0% and 5% Block purchase more than 5% Time fixed effects Number of observations R-squared 0.041 (5.087) −3.252 (0.233) 0.63% 12. capital expenditures are the nexus between issuance and future returns.269 liquidity. When a firm issues equity.60 Block sale of more than 5% Block sale between 0% and 5% No block sale and no issuance Block purchase between 0% and 5% Block purchase more than 5% Panel B: Return regressions Dependent variable Independent variables Beta BM MOM ME Monthly returns 0.204) 0.63 2.218) 0.098 (0.552 0.65 2.06% 60.237nn (0.674 B. and Giammarino (2006).385) 0.552 0.233 (0.326nnn (0.20% 11.308) 0. An alternative theory is the real options model of Carlson.602nn (24.385 (5.24% 770 3. the natural logarithm of the previous year-end book-to-market ratio (BM).28% 27.435) 0.013 (4.73% 11. and Stulz.719) 3.271nnn (0. Standard errors are clustered by time period.60% 79. and the percentage of the full sample represenetd by firms in each of five groups split according to changes in the stake of the controlling shareholder caused by block sales or purchases.473nnn (1.509 (2. Significance at the 10%. although high capital expenditures are consistent with growth opportunities as a source of irrational optimism and.140) 0.427) Yes 1.125) 0.017nnn (0. Data are taken from Economatica. Although these are not necessary conditions for market timing. and Superintendencia de Valores y Seguros (SVS).522 0.266 (0.291 (0. However.356) 0. The sample covers nonfinancial Chilean firms from 1990 to 2009.309 3.29% 11.236nn (0. should predict dilution (Helwege.957) 7. they can complement the previous evidence on stock returns. nn.895) 6.760) −5. / Journal of Financial Economics 109 (2013) 661–681 Table 8 Block sales or purchases and changes in the stake of the controlling shareholder (SCS).97% 41.52% 6.22% 976 4.004) −0. All regressions include month or year fixed effects.444) Yes 18. Larrain.193nnn (0. Panel A: Descriptive statistics for block sales Stake of controlling shareholder (SCS) Average monthly returns Standard deviation of monthly returns Average annual returns Standard deviation of annual returns Number of monthly observations Percentage of full sample 2.161nnn (0.085) −3.52% 46.056) −0. the past six-month stock return (MOM).628) 2.301 (2. similarly.295) 1.202nnn (0.

ð3Þ where Φ is the cumulative standard normal distribution.128nn (0. Fecus Plus.272) 0.472) 0.321nnn (0.511) −1. and Stulz (2007).617) 0.435 (0.495) 0.360) 0. the book value of assets (in logs).011 (0.t−1 Þ. the natural logarithm of book assets.209) −0.096 (0.117) −1.090n (0.113 (0. The sample covers nonfinancial Chilean firms from 1990 to 2009. In this model. This probability is modeled as a function of three sets of variables: pit ¼ Φðα Firm Characteristicsi.526nnn (0.330 (0. and a dummy variable that identifies if there was a change in the SCS.538) −0.326nnn (0.674 (0.398) 0.041) 0.623 (0.211 (0.355 (0. therefore. The independent variables are all lagged by one year.083) 0.234) 0.B. share .202nn (0.367 (0. This change in risk is the defining feature of the real options model vis-à-vis the market timing hypothesis and it is.086) 0.044 (0. the stock's idiosyncratic volatility.039) 0.543) −0. Stock market variables include the book-to-market ratio.389) −0.139) 0.358) 0. respectively. Contrary to the market timing hypothesis. F. All variables are measured one year prior to issuance.925n (0.181n (0.338) 0. Fisher.598) −0.678n (0. annual stock turnover.091 (0.t−1 þ β Stock Marketi.426) 0. Larrain. where pit is the probability that firm i issues equity in year t. Dependent variable Independent variables Firm characteristics at t−1: ROE Log book assets Leverage Stock market variables at t−1: Stock return Turnover Idiosyncratic volatility BM Market return Market turnover Issuance Issuance with decrease in the SCS >5% (1) (2) Issuance with increase in the SCS >5% (4) −0. the real options model does not give explicit predictions regarding the difference between issuance with and without dilution of the controlling shareholder.170n (0. annual market return. and leverage.t−1 þγ Ownershipi.397 (0. stock returns.031nn (0. Typical exante features of issuance would be good investment opportunities as represented.018 (0. nn. Firm characteristics include variables taken from the balance sheet or income statement such as ROE. the difference between the SCS and cash flow rights. Urzúa I.908nn (0.042 (0. Standard errors were calculated using cluster by time period. Significance at the 10%. 2010).179) −0.520) −0. are transformed into safe real assets.312nn (0.552nn (1. the under-performance of issuers is simply a reflection of the lower risk of these stocks. annual stock returns.079) 0.278 (0. and Superintendencia de Valores y Seguros (SVS). 5.459 (0.420) Dummy change in SCS 0.233) −2.569 (0. for example. in a lower market beta.502) −0.175 (0. which is the metric of risk in the capital asset pricing model (Carlson. and leverage.429 (0.159 (0.374) 3.689) 0. and nnn.1.033 (0. and annual market turnover.851) −0.265 0.023 (0.569 (0.054) 0. which behave like options.716nn (SCS) (0.269) 0.129n (1.964nnn (0. Data are taken from Economatica. The fall in risk should be seen.010nn (0.178 (0. 5% and 1% is indicated by n.184) −1. and Giammarino. Issuance with dilution after high returns and high liquidity For the ex ante determinants of issuance.246n (0.096) 0.339) 0. for example.178) −0.101) 0.170 −0.187) 0.353) −2. Independent variables include firm characteristics: return on equity (ROE).143n (0.270 (0.381) −0.019 (0. Pirinsky. This table shows probit regressions for general equity issuance and equity issuance with different changes in the stake of the controlling shareholder. crucial to explore it.492) Number of observations Issuance without decrease in the SCS >5% (3) 1569 risk falls because risky investment opportunities.378 (0. by the company's market-to-book ratio (Tobin's q) and firms should invest heavily after issuance.341) SCS—cash flow rights 0.520) −0.276) −0.648) 1569 1569 1569 Ownership variables at t−1: Stake of controlling shareholder −0.374) −0. stock market variables: the natural logarithm of the previous year-end book-to-market ratio (BM).166) 0.494nn (0. / Journal of Financial Economics 109 (2013) 661–681 675 Table 9 The decision to issue equity and changes in the stake of the controlling shareholder (SCS). and ownership variables: the SCS.106) Constant −1.455n (0.090) 0.237 (0. we conduct a multivariate probit analysis similar to Helwege.042) 1.

the SCS.126 1.022) 0.028) 0.5% Yes 2.503 0.055nnn (0.209 (0.016) −0.117) −0.016) 0. and equipment).036 (0. return on equity (ROE). plant.437nn (0. and nnn.343 0.006 (0.172) −0. Independent variables include a dummy identifying share issuances that imply a large decrease (more than 5% in absolute value) in the stake of the controlling shareholder (SCS) (Dummy Iss w/Decrease SCS larger 5%).131 4.011 (0.038) −0.020) 0.091nnn (0. nn.045) 7.262nnn (0.0% Yes 2.099nnn (0.018) 0.062 (0.007) −0.015) −0.676 Dependent variable Capital expenditures Independent variables Issuance variables in t: Dummy Iss w/Decrease SCS larger 5% (1) Dummy Iss w/o Decrease SCS larger 5% (2) Firm characteristics in t−1: Log book assets Leverage BM Ownership variables in t−1: Stake of controlling shareholder (SCS) SCS—cash flow rights p-value test (1)¼ (2) Year and firm fixed effects Number of observations R-squared Asset growth ROE Same year (1) Three years (2) Five years (3) Same year (4) Three years (5) Five years (6) Same year (7) Three years (8) Five years (9) 0.044 (0.082) 0.042 (0.224) −0.361nnn (0.213 10.072) −0.090) −0.013) −0.027) −0.034) −0. and turnover.150 (0.003 (0.009) −0.016 (0. respectively.009) 0. Fecus Plus.011 (0. leverage. asset growth. the log book-to-market ratio (BM).086n (0. Larrain.144) −0.064nnn (0.036 0.063nn (0. the natural logarithm of book assets.182) 0.006) −0.198 B.027) 0.036) 0.718 0.058) 0. a dummy for the rest of the equity issues (Dummy Iss w/o Decrease SCS larger 5%).015) −0.001 (0.020 (0.160n (0.063) −0. and the difference between the SCS and cash flow rights.049) −0.643nnn (0.030 (0.542nnn (0.061) −0.097nnn (0.031nnn (0.111nnn (0.041 (0.010) −0. Standard errors are clustered by firm.0% Yes 2.259nnn (0.054 (0.072nnn (0. The sample covers nonfinancial Chilean firms from 1990 to 2009. debt growth.364 0.289nnn (0.142) −0.350 96. / Journal of Financial Economics 109 (2013) 661–681 Table 10 Post-issuance firm performance.362 0.045) −0.019 (0. Urzúa I.022n (0.058) 0. F.219) −0.088nnn (0.198 4.234 0.217 85.099) 0.013) −0.049nnn (0. All dependent variables are measured over an interval between one and five years following the measurement of independent variables.013 (0.058 (0.000 (0.165) −0.362nnn (0.056nnn (0. We follow the definitions of Kim and Weisbach (2008) as specified in the main text.032) 0.706 0.054nn (0. and Superintendencia de Valores y Seguros (SVS).044nnn (0. All regressions include year and firm fixed effects.015) −0.078 (0.1% Yes 1.089) −0.4% Yes 1.047) 0.056) −0.799 0.031) −0.247nnn (0. dividends. .194) −0.020nnn (0.118nnn (0.117) −0. 5% and 1% is indicated by n.055) −0.028 0.026) 0.359nn (0.037nnn (0.522nnn (0. financing patterns.1% Yes 1.046) −0.517 15.023) −0.6% Yes 2.017) −0.044) −0. Panel regressions with the following dependent variables: capital expenditures (annual differences in property.054 (0. Significance at the 10%.031nn (0. leverage.003 (0.037) 0.8% Yes 1.011) −0.170 (0.013) −0.024) −0.1% Yes 1.125) −0.010 (0. and investment.224nnn (0.042nnn (0.162nnn (0.033) 0.022) −0. Data are taken from Economatica.031) −0.043 (0.028n (0.171 (0.042) −0.051) −0.

050nnn (0.718 0.069) −0.014) 0. Urzúa I.044) −0.008) 0.008) 0.350 0.022) 0.035) 0.009) 0.695nnn (0.063nnn (0.9% Yes 2.032n (0.039) 0.001 (0.085) −0.168 1.008 (0.022 (0.014 (0.030) −0.127nn (0.042) 0.003 (0.016) −0.046 (0.9% Yes 2.038 (0.004 (0.716 0.239n (0.379nnn (0.014 (0.056 (0.010 (0.002 (0.011 (0.018n (0.6% Yes 1.134 B.033 (0.036 0.014) 0.008) 0.136nnn (0.225nnn (0.364 0.4% Yes 2.081n (0.034 (0.139 52.009 (0.010) 0.040) −0.213 (0.054) −0.485 5.046) −0.032) −0.050) −0.022nn (0.037 0.051) 0.014) 0.015 (0.010) −0.020nnn (0.155) 0.009) 0.144 68.019) −0.002 (0.190nnn (0. F.025) −0.019 (0.057) −0.0% Yes 1.008) 0.040) −0.004) 0.011) −0.001 (0.026 0.048) −0.018) −0.026 (0.017) −0.031nn (0.138) −0.023) −0.033) 0.053nnn (0.033) 0.073 (0.014 (0.020) −0.007 (0.040) −1.048 12.046) 0.016n (0.013) 0.071) −0.139) −0.132) −0.091 (0.256nnn (0.010) −0.719 0.3% Yes 2.481 5.004 (0. Larrain.017 (0.040) −0.322 70.011) −0.026) −0.361 0.4% Yes 2.002 (0.Dependent variable Debt growth Independent variables Issuance variables in t: Dummy Iss w/Decrease SCS larger 5% (1) Dummy Iss w/o Decrease SCS larger 5% (2) Leverage BM Ownership variables in t−1: Stake of controlling shareholder (SCS) SCS—cash flow rights p-value test (1) ¼(2) Year and firm fixed effects Number of observations R-squared Turnover Same year (1) Three years (2) Five years (3) Same year (4) Three years (5) Five years (6) Same year (7) Three years (8) Five years (9) 0.646nnn (0.8% Yes 2.041n (0.117 19.508nnn (0.045 (0.182nnn (0.250nnn (0.023) 0.030 (0.013 (0.010) −0.5% Yes 1.024 (0.032) −0.051) 0. / Journal of Financial Economics 109 (2013) 661–681 Firm characteristics in t−1: Log book assets Leverage 677 .016 (0.162nnn (0.022nnn (0.024nn (0.062) −0.014) 0.021) 0.046) −0.101) −0.344nnn (0.014) 0.017) 0.066 (0.079nn (0.046) 9.

43) 0. The regression includes firm-level controls measured at the end of the year prior issuance. Σk ¼ j vi.14 (4.39) −0.t−1 # ð6Þ The stock variables we study are total assets and debt. and idiosyncratic return volatility.t−1 ð5Þ and low yfi.90) −0. Post-issuance firm performance þ β Dummy Iss w=o Decrease SCS Larger 5%i. high share turnover) are strong predictors of this type of issuance.04 (1. a change in the controlling shareholder's stake during the previous year increases the chance of an equity issue in the current year. except for the one with decreases in the SCS larger than 5%.55) −0. Market betas are computed using a 24-month rolling window of previous returns.01 (−0. the change β[t]−β[t−24] refers to the difference between the current beta and the beta computed in month t−24 (using data from t−25 through to t−49).07 (−2.22) 0. and Zingales (1998): 0.89) 0. j¼0.19)   vi.tþj ¼ ln yi.68% (the unconditional probability of issuance with large decreases in the controlling shareholder's stake is 3. a higher controlling shareholder's stake predicts a lower chance of an equity issue. plus market returns and market turnover. leverage. Column 3 shows all issues except for those with large decreases in the controlling shareholder's stake.2. we define stock and flow outcome variables as " 5.t þ γ Controlsi. Larrain. The table shows average changes in market betas and their respective t-statistics. implies that the likelihood of this type of equity issue by 0. This second dummy encompasses all the issuance dummies in Table 5.08 (−2. for example. and firm-level fixed effects. For example. Out of the ownership variables. and share turnover averaged over the corresponding horizon (j¼0.4): yaverage ¼ i. We also explore a horizon of up to five years after issuance. We include changes in the previous year to control for instances of quick rebalancing or situations in which transactions in two consecutive years are part of a single large transaction. we explore the determinants of equity issues in general. Month t refers to the month of the return used in previous regressions. chance of issuance in these cases. Column 4 shows only issues with large increases in the controlling shareholder's stake. Data are taken from Economatica.11 (−3. In addition.4.4%). and financing patterns. Higher leverage predicts a higher chance of equity issuance because it is more likely that the firm has reached the limit of its borrowing capacity. Dummy Iss w=o Decrease SCS Larger 5%i. A 1 standard deviation increase in past returns.04 (1.t is a dummy −0. capital expenditures.tþk ð7Þ . Urzúa I. A block sale to a strategic partner is.78) 0.17 (5. to save space and to focus on the most relevant comparisons.t is a dummy variable equal to one for other equity issues. irrespective of the effect on the controlling shareholder's stake. These are different from those for general issuance.06 (−1. not higher. Following Kim and Weisbach (2008).10 (2.59) −0. We also examine the effect of issuance on future ROE.t+j is the outcome of interest for firm i measured with information up to the end of year t+j.tþj .14) 0. year fixed effects. Panetta. for example. For simplicity. we look at whether equity issues predict changes in accounting performance. The results in Table 9 suggest that only equity issues with substantial dilution are more likely to be accompanied by signs of overvaluation such as high liquidity and high returns Average change in beta (t-statistics) Stake of controlling shareholder (SCS) β[t]−β[t−24] β[t+12]−β[t−12] β[t+24]−β[t] Decreases by more than 5% Decreases between 0% and 5% Does not change Increases between 0% and 5% Increases by more than 5% ystock i.tþj −vi.678 B.02 (−0. and a dummy to indicate whether there was a change in the controlling shareholder's stake in the previous year. Betas are grouped in five categories according to changes in the stake of the controlling shareholder (SCS) caused by the equity issue. Table 11 Changes in market betas for issuers.tþj 1 k¼j ∑ v 1 þ j k ¼ 0 i.t−1 þ1 assetsi. Only leverage is marginally significant among firm characteristics and stock market variables.t−1 þ Fixed Effects þ εi. The second column shows results for equity issues with large decreases in the controlling shareholder's stake.tþj Next.tþk ¼ ln k ¼ 0 þ1 assetsi. Ownership variables include the stake of the controlling shareholder.16 (−5. High past returns and high market turnover predict a lower. the difference between this and cash flow rights. In the first column. The flow variable is capital expenditures.82) −0. although the effects are relatively weak. / Journal of Financial Economics 109 (2013) 661–681 turnover (a proxy for liquidity). and Superintendencia de Valores y Seguros (SVS). we refer to asset growth and debt growth in each case. sometimes followed in the subsequent year by an equity issuance. Dummy Iss w=o Decrease SCS Larger 5%i. The regression for j¼ 0 represents the effect during the same year of the issuance. Our main regression follows Pagano.05 (1.69) −0.24) variable equal to one if there is an equity issue with a decrease in the stake of the controlling shareholder (SCS) larger than 5% in year t.19 (4. Table 9 shows results for the probit regressions. Stock market variables lose their predictive power or the predictive relationship is reversed. F.tþj ¼ α Dummy Iss w=Decrease SCS Larger 5%i. The sample covers nonfinancial Chilean firms from 1990 to 2009.2. Fecus Plus.98) 0.02 (−0.18) −0.t ð4Þ where yi. Good stock market conditions (high returns.2.

2009. where month t is the month of the return reported in previous tables. Columns 7–9). 9 Consistent with Lin. its cash flow rights increase to 20% while its control rights reach 78%.1%. therefore. Larrain. in the first column of Table 10. and Giammarino (2010). Besalco. after three years. We find that profitability. Minority shareholders lose on average 20% in a year by buying shares in firms in which the controlling shareholder is diluting his stake as compared with investing in other firms. however.6% and 4. The wedge between control and cash flow rights in Chile is comparable to the 10% and 6% . The median stake of a controlling shareholder is 57% in Germany and 50% in Italy. We find that firms that engage in market timing have higher stock returns before issuance and higher capital expenditures after issuance.1% (from an average ROE of 10% in the full sample). Columns 1–3). Appendix A. the increase in investment is more pronounced following issues with a large decrease in the SCS. The only betas that show a statistically significant decline (for different horizons) are those of nondiluting issuers. Using a hand-collected data set of the ownership stakes of controlling shareholders of Chilean companies between 1990 and 2009. Urzúa I. but also lower ROE after issuance. and Giammarino (2006) for the fall in risk. This is inconsistent with leverage as the explanation for return underperformance. In this case. instead of decrease. and Pagano. No evidence exists that the betas of the worst performing issuers fall after issuance as would be required by the risk-based explanation. and Lang. we find that the difference between the controlling shareholder's stake and cash flow rights is an obstacle for debt growth. We do not find evidence of a decrease in the market betas for these stocks. We find that betas increase. separation of control and cash flow rights is common in our sample. we find that debt growth is particularly high after issues with a large decrease in the controlling shareholder's stake (Table 10. Similarly to Carlson. Columns 4–6). the separation between control and cash flow rights is. Our sample is similar to their sample of European and emerging markets in terms of ownership concentration and pyramidal structures. but not so much as to make a significant difference. and Faccio and Lang. who conclude that financing investment is an important motive for equity issues in a sample of 38 countries. was explicit about its intention to increase the liquidity of the firm's stock by not subscribing its equity issue in 2009 at pro rata. See the 2009 annual report of Besalco at www.59). respectively.19 between months t−24 and t (t-statistic ¼4.8% after three and five years.8% higher in firms in which the controlling shareholder is diluted while they are only 3.1%.05 on average. Panel B. leverage falls by less after three to five years in the case of issuance with dilution (Table 10. / Journal of Financial Economics 109 (2013) 661–681 Table 10 reports results for these predictive regressions. we examine issuance decisions and look at whether equity issues are priced fairly or to the advantage of the controlling shareholder. 2002). 679 5. The difference with other issuers is. Considering only links through listed companies. we find that share issuance predicts low future returns only when the controlling shareholder's stake is significantly reduced. falls more strongly after issuance with dilution (Table 10.B. 10 For example. In the short run.7% higher in other issuers. However.3% (¼50%  34%  55%). Fisher. A similar pattern can be seen in total asset growth. Column 2 shows that. Europe (Barca and Becht. Real investment increases after issuance in terms of both capital expenditures and total asset growth. the claim of the Claro family on Santa Rita's dividends would be 9. in issuers when the controlling shareholder's stake falls by more than 5%. Looking forward to 24 months from time t. F.2 shows.10 We find that share turnover is marginally higher after issuance with dilution.8% higher than other issuers (16. Djankov. the beta also increases by 0. The post-issuance drop in profitability has been previously shown (Loughran and Ritter. In this paper. Increasing share turnover is one of the reasons often mentioned by controlling shareholders for not subscribing new issuances. The dynamics of leverage are interesting because low leverage is an alternative explanation for the poor postissuance returns (Eckbo. although only indirect evidence. capital expenditures are 8. 2001. For example.4%) and this difference has a p-value of 4. 1997. Control and cash flow rights in Chile (see Table 2) are higher than in Europe. Cash flow rights in Chile As the Santa Rita example in Subsection 3. 1998) but without reference to the stake of the controlling shareholder. 2000). The p-value of this difference is 7. and Norli. Fisher. Risk dynamics The short-run increase in investment that we find in Table 10 is consistent with the explanation of Carlson.svs. a construction company. The average beta increases by 0. not significant. which is at odds with a risk-based explanation. Conclusions Large controlling shareholders are prevalent in markets around the world. and the US (Villalonga and Amit. 2009). by controlling for this variable. therefore. already capturing part of the positive effect of dilution on share turnover. and Zingales. and Xuan (2011). 6. June 26 and September 9.9 As a consequence. the fall in ROE is 3. although this increase is not statistically significant (t-statistic ¼1. Ma.cl and El Mercurio (a leading Chilean newspaper). Panel B. Masulis.18). Malatesta.3. we report in Table 11 the change in market betas for different issuers. No discernible difference emerges after five years. The differential effect between the two types of issuance has p-values of 1. measured as average ROE. we are. 58%. diluting issuers invest at a rate that is 10.2−5. Columns 7–9 in Panel B of Table 10 present evidence regarding share turnover. The main determinant of share turnover according to Table 10 is the prior stake of the controlling shareholder and. When stakes held through private companies are included. The effect is not noticeable in the first year but. Panetta. after three and five years. The separation between control and cash flow rights is also standard in East Asia (Claessens. 2000). This fits well with the results of Kim and Weisbach (2008).

. 1009–1034. Baker... 2000. his cash flow rights decrease from 59% to 54. Carlson.. Foley. The Control of Corporate Europe.. Dyck. R. E. S. there is also a decline in cash flow rights larger than 5%. Panunzi. Atanasov. Becht. 72–85. Greenwood. and the value of the firm. Value versus growth: the international evidence. S. Bell Journal of Economics 11. Mullainathan. Fama. Journal of Applied Corporate Finance 20.. Journal of Political Economy 98. Dividends and expropriation.. M. This can be illustrated by looking at the average firm in our sample as shown in Table 2. Journal of Finance 59. Norli.. J. Journal of Financial Economics 25. 2000. The evolution of corporate ownership after IPO: the impact of investor protection... Barclay. The cross section of expected stock returns. Dahan.. Black. 1997. Review of Financial Studies 25. The separation of ownership and control in East Asian corporations. Ferreira. Kahle. M. 2010. in press reports and road shows at the time of issuance. 2012. 1675–1712. Holderness. Journal of Finance 67. Review of Financial Studies 22.. 2008.. through another Firm B in which the shareholder owns an 87% stake (87%  68% ¼59%). Burkart. Franks.Y. O. Greenwood. The controlling shareholder owns 68% of the shares of this average firm (Firm A). F.. V. SEO risk dynamics. M.. Journal of Corporate Finance 9. Kahle. J. The structure and formation of business groups: evidence from Korean chaebols.C. Are CEOs rewarded for luck? The ones without principals are. 1975–1999. 251–291. This information is available. Mayer... 2010. R. Market liquidity as a sentiment indicator. 1231–1260. 2008. Journal of Finance 61. R.. Scherbina.. we compute changes in the stake of the controlling shareholder through voting rights. Wurgler. Malloy. for example... French. Differences of opinion and the cross section of stock returns. these are almost equivalent to changes in cash flow rights.. 2009. C. 1990. Hong.. 2002. M. C. Journal of Finance 60. 2004. Journal of Finance 62. French. while. French. 861–878... Private benefits of control: an international comparison... M.. S.. H. H.. 693–728. Journal of Financial Markets 7. B. M.. Other cases with a large decline in control rights correspond to declines that are marginally above the 5% threshold so that cash flow rights fall by something marginally below 5% (as in the example above). If control is not at stake. The Chilean wedge is.8%). 1989. Shleifer.. for example. Barclay.. DeLong.. R.M. H. Urzúa I. 81–112.. however. much lower than the average found by Almeida. 2008. Fisher. 1992. Wolfenzon. Masulis. 1575–1599. 1980. 2005.F. F. Dividend taxes and corporate behavior: evidence from the 2003 dividend tax cut. and Wolfenzon (2011) in Korea. Because we study return predictability.. Review of Financial Studies 23. K. L.. Long-run performance of secondary equity issues: a direct test of the windows of opportunity hypothesis. Lamont.P. K. 171–205.. 995–1028. O. canceled seasoned equity offerings. F.. References Adams. Stulz. 2008. R.. 2003.. Baker. 2113–2141. Journal of Finance 55. 54–78. Seasoned equity offerings: the effect of inside ownership and float. C.. Journal of Financial Economics 58. 42–64. Stein. L. Larrain. Fama. Faccio.. C. Noise trader risk in financial markets. Dunbar. Lang. 430–465. P. American Economic Review 91. All our results are robust to using a measure of the change in the controlling shareholder's stake based on cash flow rights. In short. Journal of Financial Economics 99.. 1653–1678. Giammarino. Financial Management 39.. R. Journal of Financial Economics 65. Journal of Financial Economics 66. E.. C. Claessens. Fisher. respectively. B. 2010.. C. O. D. his stake falls by 5% in terms of control rights and by 4. L. F. J. D. L. 761–798. Shleifer. Large shareholders and corporate policies. 371–395. Waldmann. Park. Holderness.. Quarterly Journal of Economics 120.. Journal of Financial Quantitative Analysis 36. Seasoned public offerings: resolution of the “new issues puzzle”. 2004. Djankov. 2007. S. Wagner. Share issuance and factor timing. C. Disagreement and the stock market. Corporate investment and asset price dynamics: implications for SEO event studies and longrun performance... 2001. to compute changes in cash flow rights. R. Subrahmanyam. Hart.. 537–600... M. in 95% of issuances in which there is a decline of control rights larger than 5%... we find that. 2001. Clarke.. Giammarino. E.. Fahlenbrach. H. 427–465.. Journal of Financial Economics 88. The life cycle of family ownership: international evidence... K.. Oxford University Press. If the controlling shareholder reduces his direct stake in Firm A from 68% to 63% through a share issuance. The relatively high coincidence of control and cash flow rights in our sample implies that the reductions in control rights and cash flow rights are similar when there is dilution. A. 2010. 2002. F.. 183–199. which is more than 40%. M... R. Hong.. Bertrand. J. Journal of Finance 57. Journal of Business 77. Diether... 791–833.. from the point of view of outside investors. J. Review of Financial Studies 23.. How does law affect finance? An examination of equity tunneling in Bulgaria. Young. J. it is preferable to use a measure of the controlling shareholder's stake that is readily available to all market participants at the time of issuance. Journal of Finance 53. 51–91. E... Why do firms become widely held? An analysis of the dynamics of corporate ownership.. A. 901–932.8% (87%  63%¼ 54. Gromb. Given the small wedge in Chilean companies. Equity issues and the disappearing rights offer phenomenon. the free-rider problem. K. Zingales. The ultimate ownership of western European corporations.. D. 703–738. In our sample.. Negotiated block trades and corporate control. implying that Firm A is controlled. Frazzini. Journal of Financial Economics 88. Lang. A. The law and economics of self-dealing. 2011. the controlling shareholder is. Journal of Finance 63. 2007.J. R.. Breadth of ownership and stock returns. Journal of Finance 47. J.C.D. Market timing and managerial portfolio decisions.. Park. Fama. H. 1998. Throughout the paper. / Journal of Financial Economics 109 (2013) 661–681 observed in Italy and Germany. in practice. Dissecting anomalies. L. His cash flow rights are on average 59%. Ciccotello. 365–395. 2000. 2008.. S. Hanson. S. V.. 447–475.. 2005. M. Chen. 2006. E. Review of Finance 12. 271–299. Kraizberg. Barca. A.. Faccio. England. K. 155–173. and the theory of the corporation. R.. Eckbo. 3941–3976. Almeida. Intintoli. J. Long-run performance and insider trading in completed vs. La Porta. it is necessary to understand the entire web of firms (public and private) interconnected in the chain of control. Pirinsky. Oxford. M. One share–one vote: the empirical evidence. 415–430. . 299–322. M.. Dumb money: mutual funds flows and cross-section of stocks returns.. Clarke.. Subrahmanyam.. 2004.. Large shareholders. 4026–4077. Journal of Finance 46. B. S. Helwege. Carlson. Journal of Financial Economics 96. Eckbo. 2012.. 575–603. Kahle.A. Summers.E. Gyoshev. Cronqvist.. A. 1903–1949. R. Jenter. A. selling cash flow rights but. it is more straightforward to compute changes in terms of voting rights. Djankov. S. Quarterly Journal of Economics 112. Journal of Economic Perspectives 21. M. Quarterly Journal of Economics 116. R... López-de-Silanes. Volpin. B. 109–128. Journal of Financial Economics 56.P.. Lang. The equity share in new issues and aggregate stock returns. Private benefits from control of public corporations. 2001. L. Takeover bids. Stein.. Stein.. A. J.680 B. S. Hauser. Dunbar.. Chetty. K. Grossman.. 2219–2257. 2001.E.2% in terms of cash flow rights.. 2002. K. C.. Price behavior and insider trading around seasoned equity offerings: the case of majorityowned firms. Saez. D. monitoring. 1991.

Ownership structure and the cost of corporate borrowing... Journal of Corporate Finance 6. Vishny. T. A. 245–256. 2011.. 583–608. Journal of Financial Economics 13. Johnson. Lewellen. Under-performance in long-run stock returns following seasoned equity offerings. E. Vishny. 23–51.. R. P. 2009. L. Lee. Large shareholders and corporate control... 471–517. 2002. Journal of Finance 54. C. 2009. 461–488. F. B. D.. and divergence of opinion. R. Weisbach. Journal of Finance 60. 281–307 .. B. D. 3047–3091. Ritter. Risk. T. 921–945. Corporate ownership around the world. 1998. Loughran. 655–720. A. F. J. Insider trading and the long-run performance of new security issues. 1–23.. K.. F. Review of Financial Studies 22. A. economic entrenchment. K. Journal of Corporate Finance 15. Journal of Finance 52. family firms controlled? Review of Financial Studies 22. A. Zingales. Journal of Political Economy 111.. Shleifer. A. Pontiff. 1998. D. Journal of Finance 32. / Journal of Financial Economics 109 (2013) 661–681 Jenter. Journal of Financial Economics 100. Anomalies.. 1183–1219. L. Majluf. S. Review of Economic Studies 62.. 1997... Kim. López-de-Silanes. Agency problems and dividend policies around the world. Malatesta. J. 1997. 27–64. M. Villalonga. Journal of Accounting and Economics 42.S. R.. A.. uncertainty. Shleifer. U. L. López-de-Silanes. 2009... 2000... La Porta. Ritter. Overconfidence and speculative bubbles.. Zhuravskaya. . Why do companies go public? An empirical analysis. Vishny. L. 2005. Xuan. Larrain. 35–52. S.. Warner.. J. T. The operating performance of firms conducting seasoned equity offerings.. 1151–1168. 2008.. 1986. 1997... 1113–1155. R. 2009. Motivations for public equity offerings: an international perspective. 1–17.. 25–53... Miller. Journal of Political Economy 94.. Scheinkman. Journal of Economic Literature 43.. Pagano.. D. Costly arbitrage and the myth of idiosyncratic risk. Security issue timing: what do managers know. Journal of Finance 53. 1995. 1–33. Yeung. The limits of arbitrage. 2006. 1713–1757. 1999. Peyer. Insider ownership and the decision to go public. 579–590... Amit. F. Do demand curves for stocks slope down? Journal of Finance 41. Too few dividends? Groups tunneling through chair and board compensation. Loughran. Watanabe. Vishny. J. Pontiff. Kahle. 1995. Rational IPO waves. 1823–1851.R. Shleifer. J. 1984. F.. I. Wolfenzon. Pontiff.. Xiong. J. Tunneling. 681 Myers.. 1439–1466. Corporate financing and investment decisions when firms have information that investors do not have.. Affleck-Graves. Share issuance and cross-sectional returns. 1693–1745.. 4301–4334.. A. La Porta. The new issues puzzles. Pástor. Nenova. Wurgler. How are U.. Share issuance and crosssectional returns: international evidence. American Economic Review 90. F. López-de-Silanes. 187–221. A. J. 1977. R. 1995. Ma. Veronesi. 2003.. 325–351. 2003. Journal of Finance 52.. 2009.W. La Porta. W. F.... R. 35–55. Zhang. McLean. Journal of Financial Economics 94.. Woodgate. W.K.. 1986.. Y. Urzúa I... Y.. 413–443. R. 243–267. Zingales..B. 2005. and when do they know it? Journal of Finance 66. 2008. J. Journal of Financial Economics 38. La Porta. D. Journal of Finance 55. Shleifer.. Morck. R. The nature and persistence of buyback anomalies.W. 22–27. Law and finance. Journal of Finance 50... 2000.. R.. M.. Does arbitrage flatten demand curves for stocks? Journal of Business 75. Shleifer. Journal of Financial Economics 68. Review of Financial Studies 22.. and growth. P. Spiess. E.. R. Livdan.R. Journal of Finance 63.S. J. Journal of Political Economy 106... Vermaelen. A. E.. The value of corporate voting rights and control: a cross-country analysis. López-de-Silanes. Shleifer. 2009. N.. 2000.. T. Li. Panetta. Shleifer. Corporate governance.. Journal of Financial Economics 87. Lin. Do firms knowingly sell overvalued equity? Journal of Finance 52. Urzúa I. 425–448.