Forthcoming, Journal of Financial and Quantitative Analysis, 2016

DRIPs and the Dividend Pay Date Effect

Henk Berkman
University of Auckland Business School
Auckland, New Zealand
h.berkman@auckland.ac.nz
64 9 923 4861

Paul D. Koch*
University of Kansas School of Business
Lawrence, KS 66045
pkoch@ku.edu
785-864-7503

*Corresponding author. This version is preliminary. We acknowledge the helpful comments of

Ferhat Akbas, Bob DeYoung, David Emanuel, Kathleen Fuller, Brad Goldie, Ted Juhl, Michal
Kowalik, Joe Lan, Dimitris Margaritis, Alastair Marsden, Felix Meschke, Nada Mora, Peter
Phillips, David Solomon, Ken Spong, Jide Wintoki, and seminar participants at the Annual
Conferences of the Society of Financial Studies Finance Cavalcade, the Financial Management
Association, and the Southern Finance Association, as well as the University of Auckland, the
University of Canterbury, the University of Kansas, and the Federal Reserve Bank of Kansas
City. We also acknowledge the excellent research assistance of Aaron Andra, Suzanna Emelio,
Greg Freix, and Evan Richardson. Please do not quote without permission.

DRIPs and the Dividend Pay Date Effect

Abstract
On the day that dividends are paid we find a significant positive mean abnormal return that is
completely reversed over the following days. This dividend pay date effect has strengthened
since the 1970s, and is consistent with the temporary price pressure hypothesis. The pay date
effect is concentrated among stocks with dividend reinvestment plans (DRIPs), and is larger for
stocks with a higher dividend yield, greater DRIP participation, and greater limits to arbitrage.
Over time, profits from a trading strategy that exploits this behavior are positively related to the
dividend yield and spread, and negatively associated with aggregate liquidity.

JEL Classification: D82, G12, G14.

Key Words: market efficiency, price pressure, dividend reinvestment plan, retail investors,
liquidity, limits to arbitrage, institutional ownership, short sales.

I. Introduction
Many studies examine stock prices around the dividend announcement day or the exdividend date. These events are of interest because they might contain value-relevant news
associated with a dividend surprise, or evoke trading to capture dividends.1 In contrast, when the
dividend pay date arrives, there is no tax-motivated trading and no new information about the
amount or timing of this distribution. Nevertheless, we find striking evidence of a predictable
price increase on the pay date that is completely reversed over the following days. This
temporary inflation is concentrated among firms with dividend reinvestment plans (DRIPs).
A company-sponsored DRIP allows shareholders to automatically reinvest their dividend
income into more shares of the firm, without paying brokerage costs and sometimes at a
discount. Firms with DRIPs tend to be larger, with higher institutional ownership and lower
spreads. On average, 43% of all dividend-paying firms have company-sponsored DRIPs, but
these firms represent 75% of the total market capitalization of all dividend-paying firms.
The fact that there is no new information on the dividend pay date, combined with the
prevalence of DRIPs among U.S. stocks, creates an ideal setting to test the temporary price
pressure hypothesis.2 In this setting, we predict that liquidity suppliers should require a price
increase on the dividend pay date to accommodate the increased demand for shares by
uninformed (re-)investors. Ogden (1994) is the first to test this prediction. During the period,
1962 - 1989, he finds a small but significant mean abnormal return of 7 basis points (bp) on the
pay date, which accumulates to 20 bp over the following three days. However, he finds no

1 DeAngelo, DeAngelo, and Skinner (2009) review this literature.

2 Other events that have been studied to test price pressure are more likely to have information content. For
example, Mitchell, Pulvino, and Stafford (2004) examine price pressure around mergers. Studies of block sales and
secondary distributions include Scholes (1972), Holthausen, Leftwich, and Mayers (1990), and Mikkelson and
Partch (1986). Studies of changes in the S&P 500 Index include Harris and Gurel (1986), Schleifer (1986), Beneish
and Whaley (1996), Kaul, Mehrotra, and Morck (2000), and Chen, Noronha, and Singal (2004). Hartzmark and
Solomon (2013) analyze price pressure in the month that dividends are predicted.

significant price reversal after the pay date, and thus concludes that his evidence does not support
the temporary price pressure hypothesis.
Figure 1 provides analysis similar to Ogdens (1994) over ten-year subperiods spanning
the years, 1975 - 2012. Panel A plots the mean daily abnormal returns over the two weeks before
and after dividend pay dates, for the tercile of stocks with the highest dividend yield, for every
ten-year period since 1975. Panel B plots the analogous patterns in abnormal trading volume.
This figure reveals a significant mean abnormal return on the dividend pay date (AR(0)),
accompanied by a similar spike in abnormal trading volume (RVol(0)). Both spikes have grown
in magnitude during the two most recent decades, since the mid-1990s. For the tercile of high
yield stocks, the mean AR(0) has increased from 12 bp in the mid-1970s to nearly 40 bp in the
most recent decade. In addition, for the two most recent decades, there are several significant
negative mean abnormal returns after day +1, indicating a reversal that accumulates to offset the
temporary inflation around the pay date. We also find a significant abnormal return on day -3.3
The main goal of this paper is to test the temporary price pressure hypothesis by
exploring how the magnitude of the pay date effect (AR(0)) varies across stocks and over time,
with an emphasis on the role of company-sponsored DRIPs. For the period, 2008 - 2012, we
extend the analysis of Ogden by examining the divergent behavior of DRIP firms versus nonDRIP firms. We focus on this recent period because it reveals the greatest price pressure in
Figure 1, and because we have accurate lists of DRIP firms for this time frame.4

3 Yadav (2010) focuses his study on similar evidence of a price spike on day -3. He notes that the 3-day settlement
period became effective in July 1995 with SEC Rule 15c6-1, and he suggests that this AR(-3) is due to shareholders
who buy additional shares on day -3 and settle these trades with the dividend income received on day 0.
4 We obtained annual lists of DRIP firms since 1996 from the American Association of Individual Investors (AAII).
However, while we validate that these lists cover all DRIP firms for the most recent period, before 2008 they appear
to omit many DRIP firms, so that the implied lists of non-DRIP firms are corrupted by the inclusion of many DRIP
firms before 2008. We have made an Internet Appendix available, which shows that our results are robust over the
extended period, 1996 - 2012.

Figure 2 provides a first glance at our main results. Here we examine the price patterns
around the dividend pay date for the subsets of DRIP stocks or non-DRIP stocks in three
portfolios: (I) all dividend-paying stocks, (II) high dividend yield stocks, and (III) a subset of
high yield stocks that are hard to arbitrage.5 Panel A of Figure 2 plots the mean cumulative
abnormal returns (CARs) for the subsets of DRIP stocks in these three portfolios, while Panel B
plots the mean CARs for the analogous subsets of non-DRIP stocks.
Panel A of Figure 2 reveals that the CAR for each portfolio of DRIP stocks is highest on
day +1, before reversing toward zero on subsequent days. For the portfolio of high yield DRIP
stocks that are hard to arbitrage, the mean AR on day 0 is 1.4 percent and the mean CAR reaches
a peak that exceeds 1.5 percent on day +1. Then, for each portfolio, the series of negative ARs
beginning with day +2 accumulate to offset the temporary price spike on day 0, so that the mean
CAR is reduced to zero by day +20. This evidence confirms a reversal that completely offsets the
temporary inflation around the pay date for these portfolios of DRIP stocks.
Panel B of Figure 2 indicates that the analogous three portfolios of non-DRIP stocks also
display temporary inflation around the pay date, although it is much smaller in magnitude. One
potential reason for the temporary (albeit smaller) inflation for non-DRIP stocks is that some
brokerage houses allow retail clients to reinvest dividends automatically, even for stocks with no
DRIP. Also, some shareholders of non-DRIP stocks may reinvest dividends on their own.6
These results imply that on the pay date the marginal investor is trying to undo the
dividend payment by reinvesting the cash back into shares of the same company. This motivation
behind the pay date effect contrasts with prior work which shows that, in the period leading up to

5 The second portfolio (II: High_DY) includes the tercile of all dividend-paying stocks each quarter with the highest
dividend yield, while the third portfolio (III: Hard_Arb) contains the top tercile of high yield stocks that are also in
the bottom tercile by institutional ownership and the top tercile by bid ask spread.
6 In Figure B.1 of Internet Appendix B we show that the results in Figure 2 are robust over the period, 1996 - 2012.

the ex-dividend date, the marginal investor is a dividend-seeking investor as evidenced by price
pressure in that period (e.g., see Hartzmark and Solomon, 2013). Thus, the source of the pay date
effect can be viewed as the exact opposite of the impetus behind the price increase before the exdividend date, which reflects a preference by the marginal investor to receive the dividend.7
Figures 1 and 2 provide support for the temporary price pressure hypothesis, and indicate
an important role for DRIPs behind the pay date effect. This evidence motivates us to pursue
further insights into the nature of this price pressure, by conducting four additional sets of tests.
Our first set of tests relates the magnitude of the pay date effect, AR(0), to cross-sectional
variation in the demand for shares on the pay date. We begin by using a simple DRIP indicator
variable as a proxy for the elevated demand on the pay date for the shares of firms with a
company-sponsored DRIP. We find that AR(0) is significantly higher for the subset of DRIP
firms, after controlling for the influence of other firm attributes. Next we develop a proxy for
firm-specific DRIP participation, based on the observation that a higher level of participation
should result in a higher proportion of shares outstanding held by registered retail investors.8 We
conjecture that firms with greater DRIP participation (i.e., proxied by more registered retail
investors) should have a larger pay date effect. Consistent with this conjecture, we show that: i)
the proportion of shares outstanding held by registered retail shareholders is significantly higher
for DRIP stocks than for non-DRIP stocks, and ii) higher values for this proxy are associated
with a significantly larger pay date effect (AR(0)) for DRIP stocks, but not for non-DRIP stocks.
Our second set of tests investigates whether there is less price pressure for the subset of
DRIP firms that increases shares outstanding around the pay date. Firms that do not issue new
7 This evidence also raises the question why investors wait to reinvest the dividend on the pay date at a high price,
instead of selling before the ex date and buying back after the pay date. We thank the referee for these insights.
8 As we discuss below, DRIP participation is effectively limited to retail investors. In addition, participating retail
investors need to be registered shareholder of record, as opposed to the more common situation where the shares of
retail investors are held in street name.

shares are obliged to meet their DRIP commitments by buying back shares on the open market.
Alternatively, DRIP firms may issue new shares as another way to meet their DRIP obligations,
without exerting price pressure on day 0. As expected, we find that the subset of DRIP firms that
issues new shares around the event has a significantly lower pay date effect, AR(0).
Our third set of tests examines whether short sellers try to exploit the temporary price
increase around the pay date. For our sample of DRIP stocks, we find a significant spike in the
abnormal volume of short selling at the time of the largest positive price spike, on day 0. In
contrast, for non-DRIP stocks there is no such evidence of abnormal short volume around the
pay date, consistent with the lower price pressure that we observe for these stocks.
Finally, we examine whether variation over time in the pay date effect for DRIP stocks is
related to variation in the average cost of trading in the market. We begin by analyzing three
alternative trading strategies that attempt to profit from the price spike on day 0. On any given
day, these strategies prescribe holding the subset of DRIP stocks in each of our three portfolios,
I - III, on their respective pay dates (i.e., buy at the close on day -1 and sell at the close on day 0).
For every day (t) in the extended sample period, 1996 - 2012, we first calculate the daily trading
profits from each strategy as the cross sectional mean abnormal return on the pay date, AR(0)_kt,
for the subset of DRIP stocks in each portfolio (k = I - III) that pays a dividend on that date.
Within each quarter (n) in the sample period, we then aggregate the series of daily profits,
AR(0)_kt, to obtain the implied stream of quarterly profits measured by the cumulative abnormal
return for each strategy during quarter n, CAR(0)_kn, k = I - III.
Our three trading strategies yield streams of quarterly profits that are surprisingly large
and stable over time. The cumulative abnormal returns for each strategy are positive in at least
60 of the 68 quarters in the sample. They average 17.4% per quarter for portfolio I (All_Stocks),

21.3% for portfolio II (High_DY), and 19.3% for portfolio III (Hard_Arb). In addition, they tend
to be higher during recessions, grow over time, are positively related to movements in the firms
dividend yield and spread, and are negatively associated with aggregate liquidity. This evidence
corroborates the view that the pay date effect constitutes a liquidity premium for these portfolios
of DRIP firms, in response to temporary price pressure on the pay date.
This paper contributes to several strands of literature. First we build upon the body of
work that explores the price pressure hypothesis, by investigating an ideal setting where
temporary buying pressure stems from a perfectly predictable non-information event. In doing
so, we exploit a widely used tool to implement a popular investing strategy, DRIPs, which
creates predictable retail demand on the dividend pay date. Second, we contribute to the
anomalies literature by providing an example of post-publication return-predictability that has
become stronger over time, in contrast to the evidence presented in Schwert (2003) and McLean
and Pontiff (2015). Furthermore, we show that the pay date effect is not limited to small stocks
with low institutional ownership, which is also in contrast to most other anomalies (see, for
example, Boehmer and Kelly, 2009, and Chordia et al., 2011). Finally, this paper adds to the
literature on limits to arbitrage by showing that, while the temporary inflation around the pay
date is actively exploited by short sellers, their activity is insufficient to eliminate this price
pressure (see Mitchell, Pulvino, and Stafford, 2002, and Stambaugh, Yu, and Yuan, 2012).

II. DRIPs: The Literature, Implementation, and Participation

II.A. Review of the Literature on Dividend Reinvestment Plans
The use of DRIPs expanded greatly in the 1970s (Pettway and Malone, 1973), but these
plans have attracted little research in the academic literature. Ogden (1994) is the first to examine
price pressure around the dividend pay date. He finds evidence of a small price impact that is

somewhat larger for stocks with DRIPs, but he finds no evidence of a subsequent reversal.
Moreover, he relies on published lists of firms with DRIPs for just two years, 1984 and 1990,
forcing him to make assumptions about which firms had DRIPs during the decade of the 1980s.
Two other working papers also explore price pressure around the dividend pay date.
Blouin and Cloyd (2005) investigate price changes around the pay dates for closed-end funds
during the years, 1988 to 2003. They find a significant price increase around the pay date, but no
significant reversal. Yadav (2010) examines price changes around dividend pay dates over the
years, 1997 to 2008. Using an incomplete list of 300 DRIP stocks, he finds that the mean
abnormal return on the pay date is larger for his sample of DRIP stocks, compared to all stocks.
In addition, similar to our result in Figure 1, he documents a significant abnormal return three
days before the pay date. He attributes this price spike to shareholders who buy more shares on
day -3 and then use their dividend income to settle the trades three days later. He focuses the
remainder of his paper on potential microstructure determinants of this price spike on day -3.9
II.B. Transfer Agents and the Implementation of Company-Sponsored DRIPs
Firms commonly enlist a transfer agent to manage the ownership record for all investors
who trade the companys shares. The transfer agent ensures that all ownership rights are properly
allocated to the registered shareholders of record, including voting rights, the right to new shares
issued from stock splits, stock dividends or rights offerings, and the right to cash dividends.
Firms also typically rely on their transfer agent to administer company-sponsored DRIPs.

9 Another body of work addresses other implications of DRIPs for both asset pricing and corporate finance. For
example, Hansen, Pinkerton, and Keown (1985), Keown, Pinkerton, and Shalini (1991), and Peterson, Peterson, and
Moore (1987) find only a modest market reaction to the firms decision to adopt DRIPs. Scholes and Wolfson
(1989) show how investors could benefit from certain discount features associated with DRIPs in the late 1980s.
Dhillon, Lasser, and Ramirez (1992), Finnerty (1989), and Scholes and Wolfson (1989) examine firms use of
DRIPs to raise capital, and conclude that DRIPs can help to mitigate the adverse price effects of new equity issues.

Details regarding the implementation of each DRIP vary somewhat across firms, and they
are communicated to investors through a prospectus filed with the SEC or a document distributed
by the firm or the transfer agent. Two transfer agents that manage a substantial portion of all
DRIPs sponsored by U.S. companies are Wells Fargo Shareowner Services and Computershare
Trust Company. These two transfer agents have made DRIP documents available to the public.10
This DRIP documentation typically describes three important features about the purchase
of shares involved in the DRIP: (i) how the shares are to be purchased, (ii) when the shares are to
be purchased, and (iii) what purchase price is to be charged to DRIP participants. First, at each
dividend distribution the company will direct the transfer agent to either purchase newly issued
shares from the company, or purchase existing shares in the open market. Second, if the transfer
agent is told to purchase shares in the open market, it is typically directed to purchase the shares
as soon as possible after receiving the funds. This common wording implies a fairly strong
incentive for transfer agents to purchase shares on the dividend pay date (as soon as the dividend
funds are received), in order to avoid litigation regarding any potential breach of fiduciary duty.
Third, the price applied to every DRIP participant is typically the trade-weighted average price
that applies to all shares bought to satisfy the DRIP, if the shares are purchased in the open
market, or the closing price on the pay date if newly issued shares are purchased from the
company. Internet Appendix A provides excerpts from the DRIP documents for two firms, which
illustrate the responsibilities of the transfer agent and the relevant details common in these plans.
II.C. DRIP Participation and Broker Non-Votes in Shareholder Proxy Contests
We conjecture that the existence and implementation of company-sponsored DRIPs is a
major force behind the pay date effect documented in this study. DRIP firms with a higher
10 The web site for Computershare is https://www-us.computershare.com/investor/plans/planslist.asp?stype=drip,
and the web site for Wells Fargo is https://www.shareowneronline.com/UserManagement/DisplayCompany.aspx.

dividend yield and/or greater shareholder participation in their DRIP should have greater demand
for their shares on the pay date, since the transfer agent must buy more shares to implement the
DRIP. However, in designing tests of this conjecture, we are limited by the fact that no firmspecific data are available on the shareholder participation rates in company-sponsored DRIPs.11
Given this limitation, we pursue two approaches to examine how the temporary demand
induced by DRIPs affects the pay date effect. Our first approach uses a simple DRIP indicator
variable as a proxy for the elevated demand on the pay date for DRIP firms versus non-DRIP
firms. Our second approach relies on two common features of company-sponsored DRIPs to
develop a firm-specific proxy for cross-sectional variation in DRIP participation rates.
One feature is that roughly 95 percent of all DRIP firms limit the maximum dollar
dividend amount that any individual shareholder can apply to purchase more shares through the
DRIP. The typical limit ranges from $1,000 to $25,000 per quarter, which effectively excludes
financial institutions, so that only retail investors can participate. Another feature is that firms
require investors to become registered shareholder of record, in order to participate in a DRIP.
This requirement helps firms to manage their retail ownership, and results in all communication
being made directly between the firm and its participating retail shareholders. It also enables the
firms transfer agent to administer the DRIP directly to participating retail shareholders.
It is noteworthy that the shares of retail investors who do not participate in a firms DRIP,
or who purchase stocks with no DRIP, are normally held in street name in retail brokerage
accounts. This label means that the shares are registered in the name of the brokerage firm
through which the stock is bought, rather than the investor who purchased the stock. In this case,
all communication between the company and the investor is routed through the broker. This
11 We have had many conversations with companies, transfer agents, and brokerage houses to request data on firmspecific DRIP participation rates, the shareholdings of DRIP participants, and the timing and pricing of purchases
made in the implementation of DRIPs. None of these entities are willing to share any data or discuss their DRIPs.

practice gives the brokerage house control over details involving shareholder rights for their
retail customers, and reduces the cost of providing brokerage services. The typical brokerage
house charges a fee of several hundred dollars to retail clients who ask to become shareholders of
record, in order to discourage such requests. As a result of this practice, a small number of retail
brokerage houses operate as the shareholders of record on behalf of most retail investors who do
not participate in DRIPs, as well as retail investors who purchase stocks with no DRIP.
Given the above features of DRIPs, we know that an increase in DRIP participation will,
ceteris paribus, result in an increased proportion of shares held by retail investors who are
shareholders of record. This discussion motivates our proxy for the firms DRIP participation
rate as the proportion of shares held by registered retail shareholders each quarter. Figure 3
illustrates our two-step process to identify this proportion of registered retail shareholders. The
first step is to obtain the proportion of shares held by all retail investors. We accomplish this first
task by subtracting (from one) the proportion of shares held by financial institutions that file
13Fs each quarter. The second step requires that we identify the smaller proportion of shares
held by retail investors who are registered shareholders of record. This second task is achieved
by estimating its complement: the proportion of shares held by retail investors who are not
registered shareholders of record (i.e., who hold their shares in street name).
For this second step, we use panel data on shareholder votes in proxy contests to elect
members of the firms board of directors. Importantly, these voting data include the category of
broker non-votes, which reflects the proportion of a firms shares that are held in street name by
brokerage firms on behalf of non-voting shareholders. This proportion of shares classified as
broker non-votes serves to measure the fraction of a firms shares held by non-registered retail

10

shareholders, because: (a) these shares are held in street name (i.e., they are not registered), and
(b) typical retail shareholders do not vote their shares, while most institutions vote their shares.12
Given these data on broker non-votes, we can estimate the proportion of shares
outstanding that are held by registered retail shareholders each quarter by subtracting from one:
(i) the proportion of shares held by financial institutions, and (ii) the proportion of remaining
shares held by non-registered retail shareholders (measured by broker non-votes), as follows:
Partin = 1 - Pct_INSTin - Broker_Nonin;

(1)

where Pct_INSTin = percent of shares outstanding for stock i held by financial institutions
during quarter n, obtained from the firms quarterly 13F filings;
Broker_Nonin = percent of shares outstanding classified as broker non-votes
for stock i during quarter n.
The remaining shares held by registered retail investors include the shares of DRIP participants.
We use this proxy to test two predictions behind the price pressure hypothesis. First,
because Partin includes the proportion of shares held by DRIP participants, it should be higher
for DRIP firms than for non-DRIP firms, ceteris paribus. Second, if Partin is a meaningful proxy
for the additional demand for shares on the dividend pay date due to higher DRIP participation,
then Partin should be related to abnormal returns on the pay date, AR(0), for the subsample of
DRIP firms. In contrast, for firms with no DRIP, this proxy (Partin) has no bearing on DRIP
participation or the demand for shares on the pay date, and should thus be unrelated to AR(0).13

12 The company, Broadridge Financial Solutions, processes the proxy votes for over 90 percent of public

companies and mutual funds in North America. In a recent study in collaboration with Pricewaterhouse
Coopers (PwC), they report that roughly 70 percent of the shares held by retail investors are not voted, while
only 10 percent of institutional shares are not voted (see Broadridge and PwC, 2013).
13 Our proxy for DRIP participation is subject to some measurement error. For example, any institutional shares
that are not registered and not voted (i.e., that appear in both Pct_INST and Broker_Non) are double counted in
Equation (1), resulting in a downward bias for our proxy. We have also adjusted for this potential bias by adding
10% of Pct_INSTin back into Partin (see footnote 12). Our results are not affected by this adjustment.

11

III. Data and Variables

III.A. Data
We analyze daily returns and trading volume for all NYSE, AMEX, and NASDAQ
common stocks (CRSP share code 10 or 11) during the period, July 1975 through December
2012. We examine benchmark-adjusted abnormal returns to measure stock price performance
(see Daniel, Grinblatt, Titman, and Wermers, 1997). The daily abnormal return is defined as the
difference between the actual return on a stock and the return on an equally weighted portfolio of
all firms in the same size and book-to-market quintiles, obtained from Russ Wermers website.14
We obtain the pay dates of all cash dividends from CRSP (distcd = 1200 - 1299). We
apply the screens in Ogden (1994) to keep all dividend events for which: (a) the number of days
between the ex-dividend day and the pay date is at least 10 and no more than 45; (b) the number
of days between the ex-dividend day and the record date is at least 2 and no more than 7; and (c)
there are at least 20 days between a firms consecutive dividend pay dates.15
We have obtained annual lists of DRIP firms since 1996 from the American Association
of Individual Investors (AAII). For 2012, the AAII list is almost identical to an analogous list of
DRIP firms from Morningstar (97% of these DRIP firms appear on both lists). This cross-check
validates the accuracy of the AAII lists over the recent period. However, from 2007 to 2008 the
AAII lists indicate an unrealistic increase of 230 new DRIP firms, from a total of 392 to 622.16
We failed to find any reports in Factiva that corroborate this increase in 2008, suggesting
that the AAII lists omit a large number of firms with DRIPs prior to 2008. Thus, we restrict our
sample to the years 2008 - 2012 for our main analysis, where we compare the behavior of DRIP
14 Portfolio assignments for NYSE, AMEX, and NASDAQ stocks from CRSP are available since July 1975, based
on the approach in Wermers (2003), at http://www.smith.umd.edu/faculty/rwermers/ftpsite/Dgtw/coverpage.htm.
15 These screens help to ensure that we ignore events with coding errors, and reduce the impact of ex-dividend day
price effects on the pay date price effect. They eliminate 15,068 of the 232,322 events in the sample since 1975.
16 The total number of company-sponsored DRIPs on the AAII list ranges from 600 - 1,300 in the years since 1996.
However, roughly half of all DRIPs each year are sponsored by mutual funds, which are excluded in our analysis.

12

firms versus non-DRIP firms and thus require accurate lists of both types of firms. In Internet
Appendix B, we show that our main results are robust over the extended period since 1996.
Furthermore, in our analyses that focus on only DRIP stocks, we examine the extended period,
1996 - 2012, for which we have the accurate lists of DRIP stocks from AAII.
III.B. Variables
We begin by examining the following (cumulative) abnormal returns measured over five
different portions of the 21-day event window, (-10,+10), as well as the ex-dividend date:
Return Measures around the Dividend Pay Date:
1) AR(-3)in

= abnormal return for firm i on day -3, before the pay date in quarter n;

2) AR(0)in

= abnormal return for firm i on day 0, the pay date in quarter n;

3) CAR(0,+1)in

= cumulative abnormal return over days 0 and +1;

4) CAR(+2,+10)in = cumulative abnormal return over days +2 through +10;

5) CAR(0,+10)in

= cumulative abnormal return over days 0 through +10;

6) AR(ex-div)in

= abnormal return on the ex-dividend date for firm i in quarter n;

7) AR(0)in - AR(ex-div)in = difference between AR(0) and AR(ex-div);

where AR(t)in = abnormal benchmark-adjusted return for stock i on day t relative to the
dividend pay date in quarter n.
The timing of the first five measures is dictated by the evidence in Figure 1, which indicates
significant positive abnormal returns on day -3 and day 0, and a significant negative abnormal
return on day +2 followed by a series of smaller but significant negative abnormal returns in the
ensuing days over the following two weeks (days +2 through +10).
Explanatory variables:
1) DRIPin

= 1 if firm i has a company-sponsored DRIP in quarter n, or 0 otherwise;

2) Sizein

= the market capitalization for firm i on day -10, two weeks prior to the
pay date in quarter n, where market capitalization is taken from CRSP;

13

3) Div_Yieldin = the percentage dividend yield for firm i, computed as the cash dividend
amount from CRSP divided by the firms closing stock price on day -10,
two weeks before the pay date in quarter n, and multiplied by 100;
4) Pct_INSTin-1 = the percentage of total shares outstanding for firm i owned by institutional
investors during the previous quarter (n-1), taken from 13F filings;
5) Spreadin

= the closing bid-ask spread for firm i, taken from CRSP, as a percentage of
the closing price on day -10, two weeks prior to the pay date in quarter n;

6) Log_Hiloin

= the intraday stock return volatility for firm (i), measured as the natural log
of the ratio of the daily high and low prices on day -10, multiplied by 100.

These explanatory variables are known at least two weeks prior to the pay date. Thus, the results
in this paper lead to predictive trading strategies that can easily be implemented.
We examine three portfolios of stocks that vary in terms of dividend yield and the limits
to arbitrage that they face. These portfolios are generated by first independently sorting all
dividend-paying stocks each quarter by: (i) dividend yield, (ii) institutional ownership in quarter
n-1, and (iii) the percent spread on day -10. Then we construct portfolios I - III as follows:
I (All_Stocks):

All dividend-paying stocks each quarter;

II (High_DY):

Top 33% of all dividend-paying stocks each quarter by dividend yield;

III (Hard_Arb): Top 33% of all dividend-paying stocks each quarter by dividend yield,
bottom 33% by institutional ownership, and top 33% by spread.17

IV. Main Results: The Pay Date Effect

IV.A. Abnormal Returns for the Subsets of DRIP Stocks and Non-DRIP Stocks
Table 1 provides the average firm event returns over the five time frames defined above,
for the subsets of DRIP stocks and non-DRIP stocks in the first portfolio (I) that includes all
dividend-paying firms each quarter. Table B.1. of Internet Appendix B provides analogous
results for portfolios II (High_DY) and III (Hard_Arb). We present seven sets of means for: (1)
all stocks in the portfolio, (2) DRIP stocks, (3) non-DRIP stocks, (4) the difference of means
17 Other studies that use institutional ownership or transaction costs to examine the influence of limits to arbitrage
on stock return performance include Asquith et al. (2005), Berkman et al. (2009), Boehme et al. (2006), Diether et
al. (2002), Nagel (2005), and Sadka and Scherbina (2007).

14

across DRIP stocks versus non-DRIP stocks, and (5) - (7) the analogous results for a subset of
matched pairs of DRIP firms versus non-DRIP firms within the portfolio.
The subset of matched pairs is generated every quarter, by first requiring matched firms
to be in the same Fama-French five-industry classification, and then using logistic propensity
scoring based on the following four firm characteristics: Log_Sizein, Div_Yieldin, Pct_Instin-1,
and Spreadin. These four attributes regularly enter the quarterly logistic regressions with
significant coefficients, while Log_Hiloin is rarely significant and is thus omitted from the logit
model. We compute the average returns in Table 1, using a panel regression of each return
measure on a constant term, with standard errors clustered by firm (i) and quarter (n).
First consider the mean abnormal returns around the pay date for all dividend-paying
stocks, in the left column of Table 1. Results indicate that an average of 1,169 U.S. stocks pay
dividends each quarter. These stocks have a mean AR(-3) of 13 bp, a mean AR(0) of 24 bp, and
a mean CAR(0,+1) of 31 bp. This price increase is completely reversed over the following two
weeks, with a mean CAR(+2,+10) of -39 bp. All of these mean ARs and CARs are significantly
different from zero. Aggregating the CAR(0,+1) and the subsequent reversal, CAR(+2,+10), we
obtain a mean CAR(0,+10) of -9 bp, which is not significantly different from zero (t-ratio =
-0.4). This evidence supports the temporary price pressure hypothesis.
The next three sets of results in of Table 1 compare the analogous price patterns for the
subsets of all DRIP stocks versus all non-DRIP stocks. This comparison indicates that the mean
AR(0) and CAR(0,+1) are both significantly larger for DRIP firms versus non-DRIP firms.
Importantly, for both DRIP stocks and non-DRIP stocks, the negative return reversal following
the pay date effect, CAR(+2,+10), completely offsets the positive price spike around the pay
date, so that the mean CAR(0,+10) is insignificant for both sets of firms. This evidence again

15

indicates a price spike around the pay date for each set of firms that is completely neutralized
over the following two weeks, further supporting the temporary price pressure hypothesis.
Finally, the analysis of matched pairs reported on the right side of Table 1 controls for
differences in firm characteristics, and yields similar results. For example, the subset of matched
DRIP stocks has a mean AR(0) of 53 bp, which is significantly larger than the mean AR(0) of 13
bp for the matched set of non-DRIP stocks. The mean CAR(0,+1) for the matched pairs of DRIP
stocks versus non-DRIP stocks displays comparable behavior. Once again, the price spike for
both subsets of matched stocks is completely reversed over the following two weeks.
At the bottom of Table 1, we also present the mean abnormal return on the ex-dividend
date, along with the mean difference between the abnormal returns on the pay date versus the exdate.18 For the subset of DRIP stocks, the mean AR(0) on the pay date is significantly larger than
the mean AR on the ex-date. In contrast, for the subset of non-DRIP stocks, the mean AR(0) on
the pay date is smaller than that on the ex-date, although this difference is not significant.
Table B.1 of Internet Appendix B provides similar results for the successively smaller
portfolios with a higher dividend yield (II) and greater limits to arbitrage (III). As we proceed to
portfolios II and III, the price run-up and reversal around the pay date (AR(0), CAR(0,+1), and
CAR(+2,+10)) grow larger in magnitude, and the associated mean differences across DRIP
stocks versus non-DRIP stocks become larger and more significant. These results corroborate the
evidence in Table 1 and Figure 2.
IV.B. Correlations across Abnormal Returns around the Pay Date
Table 2 presents the correlations across the abnormal return measures that reveal upward
price pressure around the pay date (AR(0) and CAR(0,+1)) and the subsequent price reversal
18 Hartzmark and Solomon (2013) provide evidence of temporary price pressure during the month that dividends
are expected, which is followed by a reversal in the following month. They focus on the abnormal returns around the
ex-dividend date, and do not consider the pay date effect or the impact of DRIPs on their results.

16

following the pay date (CAR(+2,+10)). Each quarter we first compute the pairwise cross
sectional correlations across these return measures for all dividend-paying stocks in portfolio I.
We then calculate the time series mean of every pairwise correlation across all quarters, and
present the results in Table 2. Mean Pearson correlations are provided above the diagonal and
mean Spearman correlations are below the diagonal. Table B.2. of Internet Appendix B presents
the analogous results for portfolios II and III.
The correlations of interest are highlighted in the shaded areas of Table 2. These results
indicate that the cumulative abnormal return around the pay date (CAR(0,+1)) is negatively
correlated with the cumulative abnormal return over the following two weeks (CAR(+2,+10)).
Furthermore, the magnitudes of these negative correlations increase as we proceed from portfolio
I in Table 2, to consider the finer subsets of stocks in Portfolios II and III with a higher dividend
yield and greater limits to arbitrage, in Table B.2. These results reinforce the evidence in Table 1
and Table B.1., to further indicate a systematic tendency for stock prices to increase around the
pay date and reverse over the following two weeks, consistent with the price pressure hypothesis.
IV.C. Firm Characteristics for the Subsets of DRIP Stocks and Non-DRIP Stocks
Table 3 presents the average firm characteristics for the subsets of all DRIP stocks and
non-DRIP stocks in portfolio I. Columns 2 - 4 show that the average DRIP firm is significantly
larger in size, and has smaller spreads and lower return volatility, compared with the average
non-DRIP firm.19 These relative attributes of the DRIP firms versus non-DRIP firms apparent in
Table 3 remain fairly stable across the successively smaller subsets of DRIP firms and non-DRIP
firms in portfolios II and III, provided in Table B.3. of Internet Appendix B. Importantly, this
analysis shows that the DRIP firms responsible for the pay date effect are different from the
19 The last three columns of Table 3 present the results for the matched pairs analysis. As expected, three of the four
matching variables (firm size, institutional ownership, and spread) are not significantly different across the matched
pairs. On the other hand, the mean difference for the dividend yield is significant, but small in magnitude (at 5 bp).

17

typical firms involved in many other anomalies. Relative to non-DRIP firms, DRIP firms tend to
be larger, and have higher institutional ownership, smaller spreads, and lower volatility.

V. DRIPs, Demand or Supply, and the Pay Date Effect

V.A. Demand for Shares: DRIPs, DRIP Participation, and the Pay Date Effect
V.A.1. DRIPs and the Pay Date Effect
This section examines how the pay date effect depends upon cross sectional variation in
demand for shares around the pay date. We begin by using a DRIP indicator variable as a proxy
for the elevated demand for DRIP firms versus non-DRIP firms, to explore the relation between
AR(0) and firm attributes for all dividend-paying stocks over the period, 2008 - 2012, as follows:
AR(0)in = 0 + 1 DRIPin + 2 Div_Yieldin + 3 Log_Sizein + 4 Pct_INSTn-1 + 5 Log_Hiloin + in. (2)

In Table 4, we present the results from estimating this panel regression model, with standard
errors clustered by firm (i) and quarter (n). The main variable of interest is the DRIP dummy,
which reveals a significant coefficient of 0.326 (t-ratio = 5.2). This result indicates that the
abnormal return on the pay date (AR(0)) is approximately 33 bp higher for DRIP stocks relative
to non-DRIP stocks, after controlling for other firm characteristics. The coefficients of the firm
characteristics also indicate a significantly higher AR(0) for firms that face greater demand for
shares on the pay date (due to a higher dividend yield), as well as for firms that face more severe
limits to arbitrage (with smaller firm size, lower institutional ownership, and higher volatility).20
V.A.2. DRIP Participation and the Pay Date Effect
We next compare these firm characteristics, as well as our proxy for DRIP participation
and its components, across the subsets of DRIP stocks versus non-DRIP stocks. Our proxy for

20 We have also estimated Equation (2) for the extended sample period covering 1996 - 2012. Results are provided
in Table B.4. of Internet Appendix B, and are robust with respect to those in Table 4. In addition, we have estimated
this panel using the Fama-MacBeth approach. Results are in Table B.5. of Internet Appendix B, and remain robust.

18

DRIP participation relies on broker non-votes rendered in proxy contests to elect board members.
These data are available from Equilar, Inc. for the years, 2010 - 2012. For this sample period, we
have 4,747 dividend events for firms with DRIPs and 4,987 events for firms without DRIPs.
In Panel A of Table 5, we show that this sample of events for DRIP firms has a mean
AR(0) that is 15 bp higher than the analogous sample for non-DRIP firms. In addition, the
average DRIP firm has lower institutional ownership, larger size, smaller spreads, and lower
volatility. The average dividend yield is not significantly different across these groups of firms.
The main result of interest in Panel A of Table 5 pertains to our proxy for DRIP
participation. As expected, the average proportion of shares held by registered retail investors
(Partin) is significantly higher for DRIP firms than for non-DRIP firms. However, the average
level of Partin remains large in magnitude for non-DRIP firms, at almost 21%. This result
indicates that, for the typical firm, the group of registered retail shareholders includes many other
investors besides those who participate in a DRIP (which must be zero for non-DRIP firms).
This evidence suggests that there are other reasons for retail investors to become
registered shareholders, besides DRIP participation. For example, Partin may also be associated
with other firm attributes such as the dividend yield, size, spread, and stock return volatility. We
explore this issue by estimating the following regression model that measures the average
difference in our proxy for DRIP participation across the subsets of DRIP stocks versus nonDRIP stocks, after accounting for the influence of these other firm characteristics:
Partin = 0 + 1 DRIPin + 2 Broker_Nonin + 3 Div_Yieldin + 4 Log_Sizein
+ 5 Spreadin + 6 Log_Hiloin + in.

(3)

The left side of Panel B in Table 5 reports the results for the sample of all DRIP stocks
and non-DRIP stocks over the period, 2010 - 2012. The t-ratios are based on standard errors

19

clustered by firm (i) and quarter (n). The coefficient of the DRIP indicator variable (1) indicates
that, after controlling for other firm characteristics, our proxy for DRIP participation is 3.8
percent greater for DRIP firms versus non-DRIP firms (t-ratio = 3.5). In addition, this proxy is
significantly greater for firms with a higher dividend yield, smaller size, a larger spread, and
lower stock return volatility. We also note that, when we control for these other firm attributes,
there is no significant relation between the proportion of shares held by registered retail
shareholders (Partin) and the portion of non-voting shares held in street name (Broker_Nonin).21
Next we test our prediction that Partin should be related to the pay date effect, AR(0)in,
for DRIP firms, but not for non-DRIP firms. We test this prediction using the following model:
AR(0)in = 0 + 1 DRIPin + 2 Partin + 3 DRIPin*Partin + 4 Broker_Nonin + 5 DRIPin*Broker_Nonin
+ 6 Div_Yieldin + 7 Log_Sizein + 8 Spreadin + 9 Log_HiLoin + in .

(4)

This specification allows the coefficients of two key variables, Partin and Broker_Nonin, to differ
across the subsets of DRIP stocks versus non-DRIP stocks. We argue that higher values of DRIP
participation (Partin) should not be informative for the subset of non-DRIP stocks (i.e., 2 = 0),
while higher DRIP participation should imply greater demand on the pay date and a higher
AR(0) for the subset of DRIP stocks (i.e., 3 > 0 and 2 + 3 > 0). Furthermore, after we control
for the influence of Partin in this model, there is no reason to expect the proportion shareholders
who do not participate in DRIPs (i.e., non-registered retail shareholders, Broker_Nonin) to have
an independent impact on AR(0)in, for either DRIP firms or non-DRIP firms (i.e., 4 = 5 = 0).22
We estimate Equation (4) for the panel including all DRIP stocks and non-DRIP stocks
over the period, 2010 - 2012, with standard errors clustered by firm (i) and quarter (n). The

21 We do not include institutional ownership in Equation (3), since inclusion of both variables, Pct_INST in and
Broker_Nonin, would completely explain the dependent variable (i.e., Part in = 1 - Pct_Instin - Broker_Nonin).
22 We do not include institutional ownership in Equation (4), since inclusion of all three variables, Part in,
Pct_INSTin, and Broker_Nonin, would result in perfect collinearity (i.e., Part in = 1 - Pct_INSTin - Broker_Nonin).

20

results are presented on the right side of Panel B in Table 5. As expected, for non-DRIP stocks
there is no significant relation between our proxy for DRIP participation and the pay date effect
(i.e., 2 is insignificant). In contrast, there is a significant positive relation for DRIP stocks (i.e.,
3 > 0 and 2 + 3 > 0). In addition, after accounting for this association between Partin and
AR(0)in, we find no significant relation between Broker_Nonin and AR(0)in, for either DRIP
stocks or non-DRIP stocks (i.e., 4 = 5 = 0). We also find a significant positive relation between
the dividend yield and AR(0) (i.e., 6 > 0). Together, this evidence indicates that greater demand
for DRIP stocks on the pay date, due to either a higher dividend yield or greater DRIP
participation, is associated with a significantly larger pay date effect, AR(0).23
V.B. Demand for Shares: Changes in Shares Outstanding and the Pay Date Effect
We would like to investigate the conjecture that there should be less price pressure for the
subset of pay date events where firms issue more shares to meet the demand from their DRIP
commitment, rather than buying shares on the open market. However, firms do not publicly
disclose whether they issue new shares for the specific purpose of meeting their DRIP obligation.
In this section we focus on the 43,635 pay date events by all DRIP firms over the period,
1996 - 2012.24 We then use CRSP daily data on shares outstanding (SHROUT), to identify the
subset of DRIP firms that issues new shares over two time frames: on the pay date itself (i.e., on
day 0), and over a longer event window that begins on the pay date and extends through the rest

23 We have also estimated both Equations (3) and (4) using the Fama-MacBeth approach. Results are provided in
Table B.6. of Internet Appendix B and are robust. In addition, in untabulated results we find similar evidence when
we estimate Equations (3) and (4) for the subsets of all DRIP stocks and all non-DRIP stocks, separately.
24 Since we focus on only DRIP firms in this analysis, it is appropriate to rely on our accurate lists of DRIP stocks
from AAII for the extended period since 1996. This sample of 43,635 pay dates excludes events when the DRIP firm
changes shares outstanding during the event window, (-10,+10), due to a stock split, stock dividend, or merger. This
screen ensures that our analysis of the pay date effect is not confused by a dilution of shares from these other events.

21

of the pay month. We consider these two alternative time frames, because of potential delays in
reporting changes in shares outstanding that may be updated by CRSP only at the months end.25
By measuring the change in SHROUT over the longer window, we are assured of
capturing changes that may occur on day 0 to help meet the firms DRIP obligations, but that do
not show up until the end of the month. Of course, this longer window may also weaken the
power of our tests, since it is likely to include many instances where new shares are issued for
purposes other than meeting the DRIP commitment, as well as instances where the number of
shares outstanding is reduced (e.g., because of share repurchases).
First consider the relative frequencies of events when DRIP firms increase SHROUT, for
each of these two time frames. For the short window (i.e., on day 0), there is an increase in
SHROUT on 2.5% of the pay dates. For the longer window, DRIP firms experience an increase
in SHROUT between the pay date and the end of the pay month for 19% of the pay date events.
Next we formally analyze whether the magnitude of the pay date effect, AR(0), is smaller
for the subset of events for which DRIP firms increase shares outstanding, while controlling for
other firm attributes that are associated with AR(0), as follows:
AR(0)in = 0 + 1 DSHROUTin + 2 Div_Yieldin + 3 Log_Sizein
+ 4 Lpct_INSTin + 5 Spreadin + 6 Log_HiLoin + in .

(5)

where DSHROUTin = is assigned a value of one for the subset of DRIP firms with an increase
in shares outstanding, either on the pay date, or during the longer event window that
extends from the pay date to the end of the pay month.
This panel regression is estimated with standard errors clustered by firm (i) and quarter (n).

25 According to CRSP documentation, major changes in SHROUT should be recorded on the exact day of the event
in CRSP. However, firms may make smaller changes in SHROUT that CRSP is not aware of, perhaps to satisfy
DRIP obligations, or for stock grants and option exercises, or share repurchases. Such smaller changes in SHROUT
are often only recorded on the last day of the month, when CRSP receives monthly updates from its data provider.

22

Two sets of regression results are provided in Table 6, for the analysis that considers an
increase in shares outstanding over the two different time frames. Our main focus is on the
coefficient of DSHROUT in Table 6, which is significantly negative in both sets of regression
results. When we consider increases in SHROUT on the pay date, the subset of DRIP firms that
issues new shares has a significantly smaller average pay date effect, by roughly 19 basis points
(i.e., 1 = -.193, t-ratio = -2.7). We also find a negative coefficient for the analogous subset of
DRIP firms that increases SHROUT sometime over the longer window, between the pay date
and the end of the pay month. This coefficient is smaller in magnitude and only marginally
significant. Together, this evidence suggests that at least some firms manage their DRIP
obligations by issuing new shares, and as a result experience significantly less price pressure on
the pay date.
V.C. Supply of Shares: Short Selling around the Pay Date
If sophisticated suppliers of liquidity try to exploit the temporary price increase around
the pay date, then we would expect the volume of short selling to increase at the time of the
largest positive price spike, on day 0. We investigate this possibility by examining daily
movements in abnormal short volume (ASV(t)in) over all 21 days in the event window, t =
(-10,+10). This variable is constructed from Reg SHO data on daily short volume, which is
available for the ten quarters covering the period, January 2005 through June 2007, as follows:
ASV(t)in = ( SV(t) / TV(t) )in - Normal ( SV(t) / TV(t) )in ,

(5)

where ( SV(t) / TV(t) )in = ratio of short volume to total volume on day t, for stock i in quarter n;
and
Normal ( SV(t) / TV(t) )in = mean of ( SV(t) / TV(t) )in over days, t = +11 through +30.26
26 Our daily short-sales data are obtained from the self-regulatory organizations (SROs) that made tick data on short
sales publicly available starting on January 2, 2005, as a result of Regulation SHO. Short sales data for the NYSE
are available through the TAQ database, and all other SROs make short sales data available on their websites. The
end date for the regulation SHO data in our sample is July 1, 2007. Thus we rely on the earlier lists of DRIP firms
from AAII for this period, which precedes the sample period for our main analysis, 2008 - 2012. Note that the subset
of non-DRIP firms used in this section is classified as non-DRIP firms in 2008.

23

For every day in the event window, t = (-10,+10), we examine the mean values of
ASV(t)in for the subsets of DRIP stocks versus non-DRIP stocks in two portfolios: I. All_Stocks
and II. High_DY. For a firm event to be included in this analysis, we require at least one day
with non-zero shorting volume in the post-announcement window, (+11, +30). This requirement
reduces the sample to 6,451 events for portfolio I, and 2,261 events for portfolio II.27 As before,
for every quarter (n) we first compute the cross-sectional average of ASV(t)n for each day (t) in
the event window, and then we calculate the time series mean of these cross-sectional averages
over the 10 quarters in the sample period for which we have short sales data.
Results are plotted in Panels A and B of Figure 4 for the subsets of DRIP stocks and nonDRIP stocks, respectively, in portfolios I and II. The confidence intervals for the mean ASV(t)n
are obtained from the standard errors of the time series means, for the subset of DRIP stocks or
non-DRIP stocks in portfolio II.28 For the two portfolios of DRIP stocks in Panel A, average
abnormal short volume is positive on all but one day in the event window, and it is significantly
greater than zero on days -3 and 0. In addition, the magnitude of the spikes in abnormal short
volume on these two days increases somewhat as we move from the portfolio of all DRIP stocks
(I) to the subset of high yield DRIP stocks (II). For portfolio II, the average abnormal short
volume is 1% of total volume on day 0.
For the analogous subsets of non-DRIP stocks in Panel B of Figure 4, we find no
evidence of abnormal short selling around the pay date. The average abnormal short volume is
small in magnitude for each day in the event window, and it is never significantly greater than
zero. This result is consistent with the lower temporary inflation for these subsets of non-DRIP

27 We do not present the results for the third portfolio of high yield stocks that are hard to arbitrage (III), because of
small sample sizes (there are less than 10 events per quarter for the DRIP and non-DRIP subsets of this portfolio).
28 Similar to our other Figures, in Figure 4 the confidence interval for portfolio II is conservative for portfolio I.

24

stocks that is documented in Table 1 and Figure 2. Together, this evidence supports the view that
short sellers are attracted by the predictable price spikes around the pay date for DRIP stocks, but
their attempt to exploit this price spike is insufficient to eliminate this temporary inflation.
V.D. The Question of Identification
Since DRIPs are not randomly allocated among firms, there might be a concern about
endogeneity regarding the association between DRIPs and the pay date effect documented in this
study. We argue that it is unlikely that the pay date effect is caused by anything other than price
pressure, which is expected to be substantially greater for stocks with DRIPs. Importantly, there
is no systematic new information on the pay date, and no tax-induced trading. Still, it is possible
that another firm attribute may be a causal factor behind both the presence of DRIPs and the pay
date effect. In this light, perhaps the most relevant difference in firm characteristics between
DRIP firms and non-DRIP firms relates to liquidity. However, Table 3 documents that DRIP
firms tend to be larger and more liquid than non-DRIP firms. Thus, to the extent that liquidity
may create a spurious correlation between the presence of DRIPs and the pay date effect, we
would expect the pay date effect (AR(0)) to be smaller for DRIP firms, not larger. In addition,
we explicitly control for differences in liquidity in our analysis and we conduct several tests
which show that, within the sample of DRIP firms, our proxies for increased demand or supply
of shares on the pay date display behavior consistent with the price pressure explanation.

VI. Strategies that Trade on this Price Pattern

VI.A. Measuring the Quarterly Performance from Three Trading Strategies
If the pay date effect reflects a liquidity premium for the temporary increase in demand
on the pay date by uninformed traders, then the magnitude of the average pay day effect should
be larger during periods when the average cost of liquidity is higher. We measure time-variation

25

in the average pay date effect by analyzing the performance over time of three alternative trading
strategies that attempt to profit from the price spike on day 0. These strategies prescribe holding
the subsets of DRIP stocks in each of our three portfolios, I - III, on their respective pay dates.
The three trading strategies are implemented as follows. First, for every day in our
sample period from 1996 - 2012, we identify every DRIP stock in each portfolio that pays a
dividend on the next day (t).29 Then we prescribe buying the subset of all such DRIP stocks in
each portfolio that pays dividends on the next day, and holding for 24 hours (i.e., use market-onclose orders to buy at the close on day t-1 and sell at the close on day t). In addition, we assume a
short position in an ETF that mimics the S&P 500 index. This strategy earns the market-adjusted
abnormal return, AR(0)it, for each DRIP stock (i) that pays a dividend on any given day (t).
We measure the performance of these three trading strategies over time, as follows. First,
for every day (t) in the sample period, we compute the average AR(0)it across the subset of DRIP
stocks (i) in each portfolio (I - III) that pays a dividend on that given date (t). The resulting mean
values, AR(0)_kt, reflect a daily time series of one-day equally weighted abnormal profits for
each trading strategy, k = I - III, for all days (t) where at least one DRIP stock in each portfolio
pays a dividend. There are some days when no DRIP stocks in a given portfolio (k) pay a
dividend. For these days we assume no trading profits (i.e., AR(0)_kt = 0).
Figure 5 plots the resulting stream of daily abnormal profits, AR(0)_It, from applying the
first strategy that holds the DRIP stocks in portfolio I on their respective pay dates. On 3,814
(89%) of the 4,284 trading days in the sample period, 1996 - 2012, at least one DRIP stock pays
a dividend. This strategy produces a mean daily profit (AR(0)_It) that is positive on 2,273 (60%)
of the 3,814 days that produce a trade, and averages 0.31% per day across all of these trade days.

29 Since our trading strategies prescribe buying only the DRIP stocks in portfolios I - III, it is again appropriate to
rely on our accurate lists of DRIP stocks from AAII for the extended period since 1996.

26

The second and third trading strategies yield similar results. For example, on 2,836 (66%)
of all 4,284 trading days, at least one DRIP stock in portfolio II pays a dividend, producing a
mean daily profit (AR(0)_IIt) that is positive on 1,780 (62%) of these trading days, and averages
0.49% per day. Likewise, on 1,363 (32%) of all 4,284 trading days, at least one DRIP stock in
portfolio III pays a dividend, yielding a mean daily profit (AR(0)_IIIt) that is positive on 899
(66%) of these days, and averages 0.86% per day.30
Next, for every quarter (n) in the sample period, we aggregate the series of daily profits,
AR(0)_kt, to obtain the implied stream of cumulative quarterly abnormal profits for each trading
strategy, CAR(0)_kn, k = I - III. Figure 6 plots the results for all quarters (n), from applying these
three trading strategies. For the first portfolio of all DRIP stocks (I), the values of CAR(0)_In are
positive for 60 of the 68 quarters in the sample period, and the average CAR(0)_In across all
quarters is 17.4%. Likewise, the second portfolio of high yield stocks (II) generates a similar
stream of cumulative abnormal profits that are positive for 63 of 68 quarters, and result in a
somewhat larger mean quarterly CAR(0)_IIn of 21.3%. Finally, the third portfolio of stocks that
are hard to arbitrage (III) has a quarterly profit stream that is somewhat more volatile, yet these
cumulative abnormal profits are still positive in 62 quarters, and average 19.3% per quarter. It is
noteworthy that, although each respective trading strategy triggers fewer days each quarter where
a trade is made, each successive strategy produces a higher average abnormal return per day
traded, resulting in similar cumulative abnormal returns over the typical quarter.31

30 In Table D.1. of Internet Appendix D, we estimate the daily alphas from Fama-French three-factor and fourfactor models, and we obtain risk-adjusted returns that are similar to these mean daily profits, AR(0)_kt, k = I - III.
31 In Figure C.1. of Internet Appendix C we show that, for portfolios II and III, the price increase tends to occur
gradually during the trading hours on day 0. Thus, our prescribed trading strategies will capture the mean AR(0). In
addition, we find results similar to Figure 6 when we plot the mean actual returns from these trading strategies when
we do not subtract the daily market return (see CRet(0)_kn, plotted in Figure D.1. of Internet Appendix D).

27

VI.B. Determinants of Time Series Movements in Cumulative Abnormal Profits

Next we investigate the price pressure hypothesis by examining several potential
economic determinants of these three quarterly time series of cumulative abnormal profits.
According to the price pressure hypothesis, the pay date effect (AR(0)) represents a liquidity
premium that is driven by a temporary increase in uninformed demand for the shares of DRIP
stocks on the pay date. This conjecture implies that, over time, the quarterly cumulative profits
(CAR(0)_kn, k = I - III) from these three trading strategies should be larger during periods when
there is: (i) greater demand on the pay date due to higher dividend yields, or (ii) less liquidity as
indicated by higher spreads for these stocks, or (iii) lower aggregate market liquidity. In addition,
price pressure on the dividend pay date may be associated with other measures of market stress
that might capture dividend demand, such as a recession dummy and the VIX (see Hartzmark
and Solomon, 2013), or measures of dividend-related sentiment (see Baker and Wurgler, 2004).
Finally, because of the observed increase in the pay date effect through time (see Figure 1), we
also include a time trend. This discussion motivates the following time series regression model:
CAR(0)_kn = 0 + 1 Recessionn + 2 Trendn + 3 Div_Yield_kn + 4 Spread_kn
+ 5 Agg_LIQn + 6 VIXn + 7 Div_Premn + n ,

(6)

where CAR(0)_kn
= cumulative mean abnormal profits, CAR(0)_kn, aggregated across all
days during quarter n where any DRIP stock in portfolio k (k = I III) pays a dividend;
Recessionn

= 1 if there is a recession (defined by NBER) in quarter n, or 0 otherwise;

Trendn

= time trend that counts the 68 quarters (n) in the sample, 1996 - 2012;

Div_Yield_kn = mean dividend yield across the DRIP stocks in portfolio k (k = I - III),
during quarter n;
Spread_kn

= mean closing percent spread on day -10, two weeks before the pay date,
for the DRIP stocks in portfolio k (k = I - III), during quarter n;

Agg_LIQn

= quarterly average of the monthly measures of aggregate market liquidity

during quarter n, from Pastor and Stambaugh (2003);

VIXn

= quarterly average of the monthly values of the CBOE VIX Index;

28

Div_Premn

= quarterly average of the monthly aggregate market dividend premium

(pdnd), from Baker and Wurgler (2006).

Table 7 provides the regression results for the first trading strategy that involves all DRIP
stocks, CAR(0)_In, for different permutations that include subsets of the variables in Equation
(6). Consider the coefficients for each independent variable, in turn. First, the coefficient of the
recession dummy (1) is significantly positive in most permutations of this model. This evidence
indicates that the pay date effect tends to be larger during recessions, when there is often a
premium placed on dividends. Second, the coefficient of the time trend (2) is significantly
positive in all permutations, indicating that the magnitude of this price pressure has trended
upward over time, as demonstrated in Figure 1. Third, the stream of quarterly profits is positively
related to the average dividend yield for the portfolio of all DRIP stocks (i.e., 3 > 0). This
outcome indicates that a higher average dividend yield is associated with significantly greater
demand for shares on the pay date for DRIP stocks, consistent with the price pressure hypothesis.
Fourth, the stream of cumulative profits is also positively related to movements in the average
spread across all DRIP stocks (i.e. 4 > 0), for most permutations. This result indicates that the
pay date effect tends to be greater during periods of lower liquidity (i.e., higher average spreads)
for the portfolio of DRIP stocks. Fifth, we find similar evidence of a larger pay date effect during
periods of lower aggregate liquidity (i.e., 5 < 0), which is significant for some permutations of
the model. Finally, the VIX and the dividend premium are not significantly related to the stream
of quarterly profits for all DRIP stocks (i.e., 6 and 7 are not significantly different from 0).
Table D.2. of Internet Appendix B presents analogous results for the successively smaller
subsets of DRIP stocks in portfolios II and III. The results are generally consistent with the
evidence in Table 7. We conclude that the pay date effect tends to be larger during periods when

29

there is greater temporary demand for these stocks, or when there is less liquidity in these stocks
or the overall market, providing further support for the price pressure hypothesis.

VII. Summary and Conclusions

This study analyzes the behavior of stock prices around the dividend pay date. Consistent
with the temporary price pressure hypothesis, we find a significant price increase on the pay date
which is followed by a complete reversal. This temporary inflation is concentrated among stocks
with DRIPs. It is also exacerbated for finer subsets of DRIP stocks with a higher dividend yield,
and for high yield DRIP stocks that face greater limits to arbitrage.
These results are corroborated by further cross-sectional and time series analyses. First,
we show that the temporary inflation is larger for DRIP stocks that are subject to greater demand
on the pay date, for firms with greater DRIP participation and a higher dividend yield. Second,
we find that the pay date effect is significantly smaller for the subset of events where the DRIP
firm issues new shares as an alternative potential means to meet their DRIP obligations. Third,
we document that sophisticated investors act on this predictable price pressure by increasing
their short selling activity on the pay date for DRIP firms, although the resulting supply response
is insufficient to completely attenuate the temporary price pressure.
We propose three trading strategies that are designed to take advantage of this predictable
price spike on the pay date. These strategies generate a reliable stream of profits over time that is
both economically and statistically significant. Time series variation in this stream of profits is
positively related to the average dividend yield and bid-ask spread of the DRIP firms traded, and
negatively related to aggregate market-wide liquidity. These profits are surprisingly large and
consistent over time, given the nature of our proposed trading strategy, which simply exploits a
predictable price increase around a recurring non-information event.

30

Finally, we suggest that DRIP firms and transfer agents should consider establishing a
policy whereby they routinely implement their DRIP by purchasing shares over a period of
several days following the pay date, in order to attenuate the price pressure on day 0. Such a
policy is likely to be welfare improving from the perspective of the firms DRIP participants.

31

References
Asquith, Paul, Parag Pathak, and Jay Ritter, 2005, Short interest, institutional ownership, and
stock returns, Journal of Financial Economics 78 (Issue 2), 243-276.
Baker, Malcolm, and Jeffrey Wurgler, 2004, A catering theory of dividends, Journal of Finance
59, 1125-1165.
__________, 2006, Investor sentiment and the cross section of stock returns, Journal of Finance
61, 1645-1680.
Beneish, M., and R. Whaley, 1996, Anatomy of the S&P Game: The effect of changing the
rules, Journal of Finance 51, 1909-1930.
Berkman, Henk., Valentin Dimitrov, Prem C. Jain, Paul D. Koch, and Sheri Tice, 2009,Sell on
the news: Differences of opinion, short-sales constraints, and returns around earnings
announcements. Journal of Financial Economics 92, 376-399.
Blouin, Jennifer, and C. Bryan Cloyd, 2005, Price pressure from dividend reinvestment activity:
Evidence from closed end funds, Wharton School Working Paper.
Boehme, R., B. Danielsen, and S. Sorescu, 2006, Short sale constraints, differences of opinion,
and overvaluation, Journal of Financial and Quantitative Analysis 41 (Issue 2), 455-477.
Boehmer, E., and E. Kelley, 2009, Institutional investors and the informational efficiency of
stock prices, Review of Financial Studies 22, 3563-3594.
Broadridge and PwC, 2013, How well do you know your shareholders?, available at
http://proxypulse.broadridge.com/reports/How-well-do-you-know-your-shareholders.html .
Carnival Corporation DRIP Document, Prospectus Supplement, Automatic Dividend
Reinvestment Plan of Carnival Corporation, Nov. 27, 2007, Carnival Corporation, Filed
Pursuant to Rule 424(b)(3), Registration No. 333-106553, available at the following web site:
https//www-us.computershare.com/Content/Download.asp?docID={4285723E-0BD1-4F6A9FA0-9EA34623625B}&cc=US&lang=en&bhjs=1&fla=1&theme=cpu.
Chen, H., G. Noronha, and V. Singal, 2004, The price response to S&P 500 Index additions and
deletions: Evidence of asymmetry and a new explanation, Journal of Finance 59, 1901-1930.
Chordia, T., R. Roll, and A. Subrahmanyam, 2008, Liquidity and market efficiency, Journal of
Financial Economics 87, 249-268.
Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers, 1997, Measuring mutual
fund performance with characteristicbased benchmarks. The Journal of finance 52, 1035-1058.
DeAngelo, Harry, Linda DeAngelo, and Douglas Skinner, 2009, Corporate payout policy,

32

Foundations and Trends in Finance, 3, 2-3, 95-287.

Dhillon, Upinder, Dennis Lasser, and Gabriel Ramirez, 1992, Dividend reinvestment plans: An
empirical analysis, Review of Quantitative Finance and Accounting 2, 205-213.
Diether, Karl, Christopher Malloy, and Anna Sherbina, 2002, Differences of Opinion and the
Cross-Section of Stock Returns, Journal of Finance 57 (No. 5), 2113-2141.
Finnerty, John, 1989, New issue dividend reinvestment plans and the cost of capital, Journal of
Business Research 18, 127-139.
H.B. Fuller Company DRIP Document, 2011, A Dividend Reinvestment Plan for H.B. Fuller
Company Common Stock, available at the following web site:
https://www.shareowneronline.com/UserManagement/DisplayCompany.aspx.
Hansen, R., J. Pinkerton, and A. Keown, 1985, On dividend reinvestment plans: The adoption
decision and stockholder wealth effects, Review of Business and Economic Research 20, 1-10.
Hartzmark, Samuel M., and David H. Solomon. 2013, The dividend month premium.
Forthcoming, Journal of Financial Economics 109 (No. 3, September), 640-660.
Harris, L., and E. Gurel, 1986, Price and volume effects associated with changes in the S&P 500
list: Evidence for the existence of price pressures, Journal of Finance 41, 815-829,
Holthausen, R., R. Leftwich, and D. Mayers, 1990, Large-block transactions, the speed of
response, and temporary and permanent stock-price effects, Journal of Financial Economics 26,
71-95.
Kaul, A., V. Mehrotra, and R. Morck, 2000, Demand curves for stocks do slope down: New
evidence from an Index Weights Adjustment, Journal of Finance 55, 893-912.
Keown, A., J. Pinkerton, and P. Shalini, 1991, Market reaction to the formulation of automatic
dividend reinvestment plans, Review of Economics and Finance 26 (Spring), 48-58.
McLean, R. David, and Jeffrey Pontiff, 2012, Does academic research destroy stock return
predictability? Journal of Finance, forthcoming.
Mikkelson, W., and M. Partch, 1986, Valuation effects of security offerings and the issuance
process, Journal of Financial Economics 15, 31-60.
Mitchell, Mark, T. Pulvino, and Erik Stafford, 2002, Limited arbitrage in equity markets, Journal
of Finance 62, (No. 2, April), 551-584.
__________, 2004, Price pressure around mergers, Journal of Finance 59, (No. 1, February),
3163.

33

Nagel, Stefan, 2005, Short Sales, Institutional Investors and the cross-section of stock returns,
Journal of Financial Economics 78 (Issue 2), 277-309.
Ogden, Joseph P., 1994, A dividend payment effect in stock returns, Financial Review 29, 3,
345-369.
Pettway, Richard, and R. Phil Malone, 1973, Automatic dividend reinvestment plans of
nonfinancial corporations, Financial Management 2, 4, 11-18.
Peterson, Pamela P., D. R. Peterson, and N. H. Moore, 1987, The adoption of new-issue dividend
reinvestment plans and shareholder wealth, Financial Review 22, 221-232.
Sadka Ronnie, and Anna Scherbina, 2007, Analyst disagreement, mispricing, and liquidity,
Journal of Finance 62, 2367-2403.
Schleifer, A., 1986, Do demand curves for stock slope down? Journal of Finance 41, 579-590.
Scholes, Myron, 1972, The market for corporate securities: Substitution versus price pressure
and the effects of information on share price, Journal of Business 45, 179-211.
Scholes, Myron, and Mark Wolfson, 1989, Decentralized investment banking: The case of
discount dividend-reinvestment and stock-purchase plans, Journal of Financial Economics 23, 735.
Schwert, G. William, 2003, Anomalies and market efficiency, Handbook of the Economics of
Finance, Chapter 15, Ed., G. M. Constantinides, M. Harris, and R. Stulz, Elsevier Science B. V.
Stambaugh, Robert, Jianfeng Yu, and Yu Yuan, 2012, The short of it: Investor sentiment and
anomalies, Journal of Financial Economics (104), 288-302.
Wermers, Russ, 2003, Is money really smart: New evidence on the relation between mutual fund
flows, manager behavior, and performance persistance, University of Maryland Working paper.
Yadav, Vijay, 2010, The settlement period effect in stock returns around the dividend payment
days, INSEAD Working paper.

34

Figure 1. Mean Abnormal Returns and Trading Volume around the Dividend Pay Date
Panel A plots the mean abnormal returns for all 21 days in the event window, t = (-10,+10), around the dividend pay
date (on day 0) for each 10-year period since 1975. Daily abnormal returns (AR) are obtained by subtracting the
return on a benchmark portfolio matched to each stock by size and book-to-market. Panel B plots the analogous
patterns in the adjusted rank of volume (RVol), obtained by ranking the 21 days in each window by trading volume,
and adjusting these ranks to range from -0.5 to +0.5 (i.e., RVol(t) = Rank(t) / 21 - 0.5). First, every quarter we select
the top tercile of stocks by dividend yield. Second, every quarter we compute the cross-sectional mean AR(t) and
RVol(t) for each day (t) in the event window. Third, for each day in the window, we compute the time series mean
of the cross-sectional means across all quarters in every ten-year period since 1975. The 95% confidence interval is
plotted for the mean AR(t) and RVol(t) from the most recent decade, since this period has the widest interval.

Panel A. Mean Abnormal Returns around Dividend Pay Dates for Last Four Decades
0.4
0.3
1975-1984

Mean AR(t), in %

0.2

1985-1994
1995-2004

0.1

2005-2012
L95: 05-12

0
-10

-8

-6

-4

-2

10

U95: 05-12

-0.1
-0.2
10 Days Before and After Dividend Pay Date

Panel B. Mean Abnormal Volume around Dividend Pay Dates for Last Four Decades
0.1
0.08
1975-1984

0.06

1985-1994
Mean Rvol(t)

0.04

1995-2004
2005-2012

0.02

L95: 05-12
0
-10

-8

-6

-4

-2

U95: 05-12
0

-0.02
-0.04
10 Days Before and After Dividend Pay Date

35

10

1975
1975
1975
1975
1975
1975
1975
1975
1975
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985
1985

Figure 2. Mean CARs around Dividend Pay Date for DRIP Stocks or Non-DRIP Stocks
This Figure plots mean cumulative abnormal returns (CARs) across all 31 days in the event window, (-10,+20),
around the dividend pay date (on day 0), for the DRIP stocks or non-DRIP stocks in three portfolios. We first
independently sort all dividend-paying stocks each quarter by dividend yield, institutional ownershp in the
prior quarter, and the spread as a percent of the closing price on day -10. Then we select three portfolios:
I. All_Stocks = all dividend-paying stocks each quarter;
II. High_DY = the top 33% of all dividend-paying stocks each quarter by dividend yield;
III. Hard_Arb = top 33% by dividend yield, bottom 33% by institutional ownership, and top 33% by spread.
Next, daily abnormal returns are computed by subtracting the return on a benchmark portfolio matched to
each stock by size and book-to-market ratio. Then, for the DRIP stocks or non-DRIP stocks in each portfolio,
we compute the cross-sectional average CARs for all 31 days in the event window, during every quarter in the
period, 2008 - 2012. Finally, for each portfolio we compute the time series mean of these cross-sectional
averages across all quarters. Panels A and B plot the resulting mean CARs for the DRIP stocks and non-DRIP
stocks, respectively, in each portfolio.

Panel A. Mean CARs for Subsets of DRIP Stocks in Three Portfolios

1.5

Percent

1
All_Stocks
0.5

High_DY
Hard_Arb

0
-10

-5

10

15

20

-0.5
10 Days Before and 20 Days After Dividend Pay Date

Panel B. Mean CARs for Subsets of Non-DRIP Stocks in Three Portfolios

1.5

Percent

1
All_Stocks
0.5

High_DY
Hard_Arb

0
-10

-5

10

-0.5
10 Days Before and 20 Days After Dividend Pay Date

36

15

20

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

Figure 3. Proxy for Firm-Specific DRIP Participation Rates

This Figure illustrates the construction of our quarterly proxy for firm-specific DRIP participation rates,
as the proportion of shares outstanding held by registered retail investors:
Partin = 1 - Pct_INSTin - Broker_Nonin ,
where

(1)

Pct_INSTin = percent of shares outstanding for stock (i) held by financial institutions
during quarter (n), obtained from the firms quarterly 13F filings;
Broker_Nonin = percent of shares outstanding classified as broker non-votes for stock (i)
during quarter (n).

We proxy a firms DRIP participation rate each quarter, by estimating the proportion of the firms shares
held by retail investors who are registered shareholders of record, since a basic requirement for investors
to participate in a DRIP is to register their shares. The flow chart below shows that we construct this
measure as the proportion of shares that remain after subtracting: (i) shares held by financial institutions,
and (ii) shares classified as broker non-votes in proxy contests to elect the firms board members (i.e.,
non-voting shares held in street name). It is appropriate to begin by subtracting institutional ownership
and thereby focus on retail investors, since institutions are excluded from participating in most DRIPs. It is
also appropriate to further subtract broker non-votes as an estimate of the proportion of non-registered
retail shareholders, since typical non-participating retail shareholders hold their shares in street name and
do not vote their shares. After subtracting this proportion of non-registered retail investors, the remaining
shares held by registered retail investors include the shares of DRIP participants.

Broker Non-Votes

Institutional
Shareholders
Shares

Non-Registered

Outstanding

Retail
Shareholders
Retail

DRIP

Shareholders

Registered

Participants

Retail
Shareholders

Proxy for DRIP Participation (Parti)

Others

37

Figure 4. Mean Abnormal Short Volume around the Dividend Pay Date
This Figure plots the mean daily movements in abnormal short volume (ASVitn) for the subsets of DRIP stocks
versus non-DRIP stocks in two portfolios each quarter: I (All stocks) and II (High_DY). There are too few firms
in Portfolio III (Hard_Arb) with nonzero short volume to obtain reliable results. Abnormal short volume is
defined as: ASVitn = (SV / TV)itn - Normal(SV / TV)in , where (SV / TV)itn = short volume as a proportion of total
volume for stock i on day t during quarter n, and Normal(SV / TV)in is the mean (SV / TV)it over days +11 - +30
after the dividend pay date (on day 0). First, every quarter, for each day in the window, (-10,+10), we compute
the cross-sectional mean of ASVitn across the DRIP stocks or non-DRIP stocks in each portfolio. Second, we
compute the time series mean of these cross-sectional means over the ten quarters for which we have short
sales data, January 2005 - June 2007. The standard error of each time series mean is then used to construct
the 95% confidence interval for the DRIP or Non-DRIP stocks in Portfolio II, for all 21 days in the event window.
Panel A. Abnormal Short Selling for the Subsets of DRIP Stocks in Two Portfolios
1

0.5
Percent

I. All Stocks
II. High_DY

0
-10

-8

-6

-4

-2

10

L95 - II
U95 - II

-0.5

-1
10 Days Before and After Dividend Pay Date

Panel B. Abnormal Short Selling for the Subsets of Non-DRIP Stocks in Two Portfolios
1

0.5

Percent

I. All Stocks
II. High_DY
0
-10

-8

-6

-4

-2

10

L95 - II
U95 - II

-0.5

-1
10 Days Before and After Dividend Pay Date

38

Figure 5. Time Series of Daily Profits, the Mean Market-Adjusted AR(0)_It,

for the Subset of DRIP Stocks in Portfolio I that Pay Dividends on Any Date, t
This Figure plots the daily time series of equally weighted abnormal profits, AR(0)_It, obtained by averaging the daily
abnormal returns on the dividend pay date, AR(0)it, across all stocks in Portfolio I that pay a dividend on day t. First,
daily returns on the dividend pay date for each stock are measured from the close on day -1 to the close on day 0.
Second, market-adjusted abnormal returns, AR(0)it, are obtained by subtracting the daily return on the S&P 500 index
from the daily return for each stock. Third, every day (t) in the sample period, 1996 - 2012, we compute the mean of
the cross sectional abnormal returns on the pay date, AR(0)it , across the subset of DRIP stocks in Portfolio I that pays
a dividend on that date. We then plot this time series of daily mean abnormal returns, AR(0)_It, for all days (t) when
at least one DRIP stock in Portfolio I pays a dividend. The results reflect the stream of daily abnormal profits for the
first trading strategy that prescribes holding the DRIP stocks in Portfolio I on their respective dividend pay dates.
30

25

Mean AR(0)_It, in %

20

15

89% of all trading days in the sample period have

1 DRIP stock in Portfolio I that pays a dividend.

60% of these days have a mean daily AR(0)t > 0.

The average of all mean daily AR(0)t = 0.31%.

10
5

0
1996
-5

1998

2000

2002

2004

2006

2008

2010

-10
All Days (t) When At Least One DRIP Stock in Portfolio I Pays a Dividend, January 1996 - December 2012

39

2012

Figure 6. Time Series of Quarterly Profits, the Cumulative Abnormal Returns, CAR(0)_k n,
for the Subsets of DRIP Stocks in Portfolios, k = I - III, that Pay Dividends on Any Date, t, During Quarter n
This Figure plots the quarterly time series of the CAR(0)_kn obtained by aggregating the daily cross-sectional mean
abnormal returns on the dividend pay date, AR(0)_kt, across all days during every quarter for which at least one DRIP
stock in each Portfolio (k = I - III) pays a dividend. First, daily returns on the pay date are measured from the close on
day -1 to the close on day 0. Second, daily market-adjusted abnormal returns are obtained by subtracting the daily
return on the S&P 500 index from the daily return for each stock. Third, every day (t), we compute the mean cross
sectional abnormal return on the pay date, AR(0)it , for the subset of DRIP stocks in each portfolio that pays dividends
on that date. Fourth, we compute the quarterly sum of this series of mean daily abnormal returns, AR(0)t, over all
days (t) during the quarter (n) for which at least one DRIP stock in each portfolio pays a dividend. The results reflect
the quarterly aggregate cumulative abnormal return, CAR(0)_kn, from the three trading strategies that prescribe
holding the DRIP stocks in each portfolio (k = I - III) on their respective dividend pay dates during a given quarter (n).
140
I. All_Stocks

120

II. High_DY

CAR(0_kn , in %

100

III. Hard_Arb

80

Avg CAR(0)_kn /qtr:

I: 17.4%
II: 21.3%
III: 19.3%

60
40

Avg #Firms/qtr:
I: 507
II: 235
III: 74

20
0
1996

1998

2000

2002

2004

2006

-20
40

2008

2010

2012

Table 1. Average Behavior of Stock Prices around Dividend Pay Dates

for the Subsets of DRIP Firms and Non-DRIP Firms in Portfolio I (All Dividend-Paying Stocks)
This Table presents the average abnormal returns over different time frames around the dividend pay dates for all dividend-paying stocks
during the sample period, 2008 - 2012. The variables, AR(-3)in, AR(0)in, CAR(0,+1)in , CAR(+2,+10)in , and CAR(0,+10)in are the percent
(cumulative) abnormal returns measured over different portions of the 21-day event window (-10,+10) around the pay date (on day 0) for
the ith firm in the nth quarter. These abnormal returns are computed by subtracting the daily return on a benchmark portfolio matched to
each stock by size and book-to-market. "AR(ex-div)in" is the analogous abnormal return on the ex-dividend date, and "AR(0) - AR(ex-div)"
is the mean of the difference between AR(0)in and AR(ex-div)in. We provide seven sets of results for: (1) all stocks in Portfolio I, (2) all
DRIP stocks, (3) all Non-DRIP stocks, (4) the difference of means across all DRIP stocks and all Non-DRIP stocks, and (5)-(7) the
analogous results for a subset of matched pairs of all DRIP stocks and non-DRIP stocks. We describe the matching scheme in the text.
We compute the average abnormal returns around the pay date for every portfolio, using a panel regression of each return measure on a
constant term, with standard errors clustered by firm (i ) and quarter (n). Analogous mean abnormal returns for the subsets of DRIP firms
and non-DRIP firms in each Portfolio are computed similarly.

Portfolio I. All Stocks

All DRIP Stocks

All Non-DRIP Stocks

Difference of Means
(2) - (3)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Mean

Mean

Mean

DRIP - Non-DRIP

DRIP

Non-DRIP

Mean Diff

1,169 ***

507 ***

656 ***

-149 ***

130

130

AR(-3)in %

.13 ***

.12 ***

.14 ***

-.02

.04

.18**

AR(0)in %

.24 ***

.38 ***

.13 **

.25 ***

.53***

.13**

.40 ***

CAR(0,+1)in %

.31 ***

.44 ***

.21 **

.24 ***

.64***

.25**

.39 ***

-.65**

-.26*

-.38

-.02

-.01

.00

.18**

.16***

.02

.35***

-.02

# firms per qtr

CAR(+2,+10)in %

-.39 **

CAR(0,+10)in %

-.09

AR(ex-div)in %
AR(0)in - AR(ex-div)in

-.40 **

-.39 ***

.04

.15 ***

.13 ***

.09 **

.26 ***

-.18
.17 ***
-.04

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

41

-.01
.23

**

-.04
.29

***

Matched Pairs

-.14

.37 **

Table 2. Correlations across Return Measures over Different Time Frames around the Dividend Pay Date
This Table provides correlations across the return measures taken over three different time frames around the dividend pay da te: AR(0),
CAR(0,+1), and CAR(+2,+10). We compute these correlations across all stocks, DRIP stocks, and non -DRIP stocks within Portfolio I (all dividendpaying stocks) each quarter. The mean correlations are calculated in two stages. First, every quarter we compute each pair wise cross-sectional
Pearson or Spearman correlation across the dividend events for every group of stocks. Second, we compute the time series mea n for each
pairwise cross-sectional correlation across all quarters in the sample period, 2008 - 2012. The standard deviation of every time series mean
correlation is then used to construct the t-test of the null hypothesis that every mean correlation equals zero. The mean Pears on correlations are
above the diagonal, and the mean Spearman correlations appear below the diagonal. Correlations in BOLD are significant at the .05 level.
Portfolio I: All Dividend-Paying Stocks
AR(0)

CAR(0,1)

CAR(2,10)

All DRIP Stocks

AR(0)

CAR(0,1)

All Non-DRIP Stocks

CAR(2,10)

AR(0)

CAR(0,1)

CAR(2,10)

AR(0)

1.00

.69

-.02

1.00

.69

-.03

1.00

.69

-.01

CAR(0,1)

.66

1.00

-.06

.67

1.00

-.08

.65

1.00

-.06

CAR(2,10)

-.03

-.05

1.00

-.03

-.05

1.00

-.02

-.06

1.00

42

Table 3. Firm Characteristics for the Subsets of All DRIP Firms and All Non-DRIP Firms
This Table summarizes the descriptive statistics for the main firm characteristics analyzed in this study, for the subsets of DRIP stocks
and non-DRIP stocks in Portfolio I (All_Stocks), which includes all dividend-paying stocks each quarter over the period, 2008 - 2012.
Sizein is market capitalization of the ith firm on day -10, two weeks prior to the nth quarterly dividend pay date (on day 0). Div_Yieldin is the
quarterly dividend amount, as a percent of the closing price on day -10. Pct_Instin-1 is the percent of total shares outstanding for stock i
held by financial institutions in the previous quarter (n-1). Spreadin is the daily closing spread, as a percent of the closing share price on
day -10. Finally, Log_Hiloin is the natural log of the ratio of the daily high to the daily low taken on day -10, presented as a percentage
(i.e., multiplied by 100). We present seven sets of results for: (1) all stocks in Portfolio I, (2) all DRIP stocks, (3) all Non-DRIP stocks, (4)
the difference of means across all DRIP stocks and all Non-DRIP stocks, and (5)-(7) the analogous results for a subset of matched pairs
of DRIP stocks and non-DRIP stocks within Portfolio I. The matching scheme is described in the text. We compute the average firm
characteristics for every portfolio, using a panel regression of each variable on a constant term, with standard errors clustered by firm (i )
and quarter (n). Analogous mean firm characteristics for the subsets of DRIP firms and non-DRIP firms in each Portfolio, and their mean
differences, are computed similarly.

Portfolio I. All Stocks

All DRIP Stocks

All Non-DRIP Stocks

Difference of Means
(2) - (3)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Mean

Mean

Mean

DRIP - Non-DRIP

DRIP

Non-DRIP

Mean Diff

130

130

Matched Pairs

# firms per qtr

1,169 ***

507 ***

656 ***

-149 ***

Sizein (millions)

$7,316 ***

$12,588 ***

$3,241 ***

$9,347 ***

$3,602***

$3,515***

$87

Div_Yieldin %

.77 ***

.77 ***

.77 ***

.00

.66***

.71***

-.05

Pct_Instin-1 %

61.73 ***

62.71 ***

60.97 ***

1.74

60.15***

60.77***

-.62

Spreadin %

.80 ***

.58 ***

.97 ***

-.39 ***

.82***

.86***

-.03

Log_Hiloin %

3.76 ***

3.48 ***

3.98 ***

-.50 ***

3.76***

3.83***

-.07

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

43

***

Table 4. DRIPs, Firm Characteristics, and the Dividend Pay Date Effect
This Table analyzes the following relation between firm characteristics and the abnormal return on the dividend
pay date, AR(0), for all dividend-paying stocks (both with and without DRIPs) over the period, 2008 - 2012:
AR(0)in = 0 + 1 DRIPin + 2 Div_Yieldin + 3 Log_Sizein + 4 Pct_INSTin-1 + 5 Spreadin + 6 Log_Hiloin + in . (2)
All variables are defined in Table 1 and Table 3. We present the results from estimating this panel regression
with standard errors clustered by firm (i) and quarter (n).

Intercept

.694
3.5 ***

DRIPin

.326
5.2 ***

Div_Yieldin

13.98
2.1 *

Log_Sizein

-.042
-2.9 ***

Pct_INSTin-1

-.407
-2.2 **

Spreadin

4.54
1.3

Log_Hiloin

2.61
2.8 **

Avg # Firms / qtr

1,212

Panel R2

.0068

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

44

Table 5. DRIP Participation, Firm Characteristics, and the Pay Date Effect, AR(0)
This Table analyzes the following proxy for the participation rate in a firm's dividend reinvestment plan (DRIP):
Partin = 1 - Pct_INSTin - Broker_Nonin ;
where Pct_INSTin = percent of shares outstanding for stock (i) held by financial institutions during quarter (n);
and Broker_Nonin = percent of shares outstanding with proxy votes classified as broker non-votes.
Data on broker non-votes, and thus our proxy for DRIP participation, are available for the period, 2010 - 2012.
In Panel A, we present summary statistics for various firm characteristics, including our proxy for DRIP
participation, across the subsamples of DRIP firms and non-DRIP firms over the period, 2010 - 2012.
In Panel B, we estimate the following two panel regression models that describe: (i) the relation between our
proxy for DRIP participation and firm characteristics, and (ii) the relation between the pay date effect, AR(0),
and our proxy for DRIP participation, as well as other firm characteristics:
Partin = 0 + 1 DRIPin + 2 Broker_Nonin + 3 Div_Yieldin + 4 Log_Sizein + 5 Spreadin + 6 Log_Hiloin + in. (3)
AR(0)in = 0 + 1 DRIPin + 2 Partin + 3 DRIPin * Partin + 4 Broker_Nonin + 5 DRIPin * Broker_Nonin
+ 6 Div_Yieldin + 7 Log_Sizein + 8 Spreadin + 9 Log_Hiloin + in .

(4)

We estimate the panel regression models in Equations (3) and (4) with standard errors clustered by firm (i) and
quarter (n). All variables are defined in Tables 1 and 3.a

Panel A. Summary Statistics for Our DRIP Participation Proxy and Firm Characteristics
Across the Subsets of DRIP Firms versus Non-DRIP Firms, 2010 - 2012
Variable

DRIP Firms

Non-DRIP Firms

Mean Difference

AR(0)in (%)

0.17

0.02

0.15***

Pct_Instin-1 (%)

67.4

71.0

-3.6***

Div_Yieldin (%)

0.707

0.713

-0.006

Sizein (millions)

$3,590

$1,481

$2,109***

Spreadin (%)

0.12

0.17

-0.05***

Log_Hiloin (%)

2.4

2.8

-0.4***

Part in (%)

22.0

20.9

1.1***

Broker_Nonin (%)

11.2

8.4

2.8***

*** indicates statistical significance at the .10 level; ** at the .05 level; and *** at the .01 level.

45

Table 5, continued
Panel B. Estimation of Two Panel Regression Models:
Equation (3): DRIP Participation and Firm Characteristics
Equation (4): The Pay Date Effect, DRIP Participation, and Firm Characteristics
Dependent Variable for (3) = Partin
Intercept

Dependent Variable for (4) = AR(0)in

.596

Intercept

9.8 ***

DRIPin

0.3

DRIPin

.038

3.5 ***

.050
-.022
-0.3

Part in

-.122
-1.1

DRIPin*Part in

.618
2.7 **

Broker_Nonin

Broker_Nonin

.010

0.1

.060
0.1

DRIPin*Broker_Nonin

.324
0.5

Div_Yieldin

Div_Yieldin

1.46

2.2 **

Log_Sizein

3.5 ***

Log_Sizein

-.028

-7.3 ***

Spreadin

5
6

Spreadin

10.52

Log_hiloin

-.613

Avg # Firms/qtr

-.676
-0.3

2 + 3

t-statistic

Panel R2

3.37
0.5

-4.0 ***

-.002
-0.3

6.8 ***

Log_Hiloin

4.61

Panel R2

0.148
804

Avg # Firms/qtr

.496
3.2 ***

0.0045
801

*** indicates statistical significance at the .10 level; ** at the .05 level; and *** at the .01 level.
a

In Table B.3 of Internet Appendix B, we document similar results when we estimate Equations (3) and (4) using the
Fama-MacBeth approach.

46

Table 6. DRIPs, Changes in Shares Outstanding, and the Dividend Pay Date Effect
This Table examines the average difference in the abnormal return on the pay date, AR(0), for the subset of
DRIP firms that increase shares outstanding (SHROUT) around the pay date, while controlling for other firm
characteristics, as follows:
AR(0)in = 0 + 1 DSHROUTin + 2 Div_Yieldin + 3 Log_Sizein + 4 Pct_INSTin-1 + 5 Spreadin + 6 Log_Hiloin + in, (5)

where DSHROUTin = 1 if the DRIP firm increases shares outstanding around the pay date, and zero otherwise.
All other variables are defined in Table 1 and Table 3. We measure the change in shares outstanding over two
time frames: (i) on the pay date (day 0), and (ii) over the event window that extends from the pay date to the
end of the pay month. The latter time frame represents a conservative approach which ensures that we
include changes in SHROUT that may occur on the pay date (day 0), but that may not be recorded by CRSP
until the end of the month. This regression model is estimated for the sample of pay date events for all DRIP
stocks each quarter over the period 1996 - 2012, for which CRSP shares outstanding is not missing for either
time frame. The panel is estimated with standard errors clustered by firm (i) and quarter (n).

Time Frame for Measuring Change in Shares Outstanding

Intercept
DSHROUTin

0
1

On Pay Date (day 0)

From Pay Date (day 0)

until End of Pay Month

.666

.674

1.5

1.5

-.193

-.063

-2.7 ***

Div_Yieldin

29.20
4.2 ***

Log_Sizein

-.026
-1.9 *

Pct_INSTin-1

-.711
-3.6 ***

Spreadin
Log_Hiloin

5
6

-1.7 *

29.26
4.3 ***

-.026
-1.9 *

-.711
-3.6 ***

3.66

3.62

0.5

0.5

12.97

12.99

4.3 ***

4.3 ***

Number of Events

43,635

43,635

Panel R2

.041

.041

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

47

Table 7. Determinants of Quarterly Profits from the First Trading Strategy

This Table presents estimates for the time series regression model that analyzes determinants
of the quarterly cumulative abnormal profits from our first trading strategy, as follows:
CAR(0)_In = 0 + 1 Recessionn + 2 Trendn + 3 Div_Yield_In + 4 Spread_In
+ 5 Agg_LIQn + 6 VIXn + 7 Div_Premn + n .
(6)
The dependent variable, CAR(0)_In, is computed in two steps. First, for every day (t) in our
sample period, 1996 - 2012, we compute the stream of average daily profits as the crosssectional mean market-adjusted AR(0)_It across all DRIP stocks in Portfolio I (all dividend-paying
stocks) that pay dividends on that day. Second, for each quarter (n), we aggregate this stream
of daily profits, AR(0)_It, across all days in which at least one DRIP stock in Portfolio I pays a
dividend, to obtain the cumulative abnormal profit, CAR(0)_In, for Portfolio I. Recessionn is a
dummy variable that equals one for all quarters during recessions, and zero otherwise. Trendn is
a deterministic trend that counts the quarters in our sample. Div_Yield_In is the mean dividend
yield in quarter n for the DRIP stocks from Portfolio I. Spread_In is the mean daily closing
percent spread on day -10 prior to the dividend pay dates in quarter n, for the DRIP stocks in
Portfolio I. Agg_LIQn is the aggregate market liquidity measure of Pastor and Stambaugh
(2003), averaged across the three months during quarter n. VIXn is the average of the monthly
CBOE VIX index values during quarter n. Div_Premn is the quarterly average of the monthly
aggregate market dividend premium (pdnd), from Baker and Wurgler (2006). The Newey-West
t-ratios appear beneath the parameter estimates.

Recessionn

t-ratio

Trendn

t-ratio

Div_Yield_kn

t-ratio

Spread_kn

8.14

1.9 *

1.2

.24

.29

.23

.41

1.9 *

2.5 **

1.8 *

2.1 **

85.36

74.18
4.0 ***

10.48

11.09

2.9 ***

3.2 ***

70.30
3.7 ***

8.66
1.9 *

3.3 ***

10.43
1.5

-.75

-.68

-2.6 ***

-1.9 *

-1.5

.35

.41

1.1

0.8

6
7

-.07

t-ratio
Adj R2
Overall F

62.69

-.97

t-ratio

Div_Premn a

4.91

2.2 **

t-ratio

VIXn

6.99

2.8 ***

4.4 ***

t-ratio

Agg_LIQn

11.24

-0.3
.54
20.5 ***

.59
20.5 ***

.60
17.4 ***

.63
15.1 ***

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.
The models in the first three columns are estimated over the entire 68-quarter sample period, 1996-2012. The
last column is estimated over the period, 1996-2010, due to data limitations on pdnd from Baker and Wurgler (2006).

48

Internet Appendix
Appendix A. Excerpts from Selected Dividend Reinvestment Plans
The following excerpts exemplify the relevant details common in DRIP documentation.
A.1. H.B. Fuller Company DRIP Document (2011), selected excerpts:
As the Plan Administrator, Wells Fargo Shareowner Services, a division of Wells Fargo
Bank, N.A., (the Plan Administrator) offers investors a simple and convenient method of
investing in H.B. Fuller Company common stock. The Plan Administrator will apply all of the
participants designated dividends to purchase whole and fractional shares acquired under the
Dividend Reinvestment Plan. Such purchases may be made on any securities exchange where
such shares are traded, in the over-the-counter market or in negotiated transactions, and may be
on such terms as to price, delivery and otherwise as the Plan Administrator may determine.
Dividends are invested as soon as administratively possible on or following the dividend
payable date, generally within five (5) trading days. In the case of each purchase, the price at
which the Plan Administrator shall be deemed to have acquired H.B. Fuller common stock for
the participants account shall be the weighted average price of all shares purchased plus any per
share fees. Depending on the number of shares being purchased and current trading volumes in
the shares, purchases may be executed in multiple transactions that may occur on more than one
day.
A.2. Carnival Corporation DRIP Document (2007), selected excerpts:
The shares of Carnival Corporation common stock purchased under the (Dividend
Reinvestment) Plan may be newly issued shares or shares purchased for participants in the open
market, at Carnival Corporations option. The Plan currently provides that shares purchased for

49

participants with reinvested dividends will be purchased at fair market value, as determined in
the Plan.
WHO ADMINISTERS THE PLAN?
Computershare Trust Company, N.A. (the Agent), a bank unaffiliated with Carnival
Corporation, administers the Plan. The Agent arranges for the custody of share certificates, keeps
records, sends statements of account to participants, and makes purchases of shares of Carnival
Corporation common stock under the Plan for the accounts of participants. The Agent will send
each participant a statement of his or her account under the Plan as soon as practicable following
each purchase of shares of Carnival Corporation common stock. Each statement will show (a)
any dividends credited; (b) plan shares purchased and fractional shares allocated; (c) the cost per
share of the purchased shares and fractional shares; (d) the number of whole shares for which
certificates have been issued, if any; and the beginning and ending balances of whole shares and
fractional shares The Agent will also serve as custodian of shares purchased under the Plan to
protect participants from loss, theft or destruction of stock certificates.
WHAT IS THE SOURCE OF SHARES PURCHASED UNDER THE PLAN?
Shares purchased under the Plan will come from the authorized and unissued shares of
the Carnival Corporation common stock or from shares purchased on the open market by the
Agent, as determined by Carnival Corporation.
With respect to any open market purchases made under the Plan, the Agent will have full
discretion as to all matters relating to purchases, including determination of the number of
shares, if any, to be purchased on any day, the time of day, the price paid for such shares, the
markets in which such shares are to be purchased
WHEN WILL FUNDS BE INVESTED UNDER THE PLAN?

50

If shares are purchased from Carnival Corporation, the purchases will be made on the
dividend payment date and such shares will be credited to participants accounts on the dividend
payment date. If shares are to be purchased in the open market, the Agent is to use its best efforts
to apply all funds received by it to the purchase of shares within 30 days of the receipt of such
funds from Carnival Corporation
WHAT IS THE PURCHASE PRICE OF THE SHARES?
If the Carnival Corporation common stock is purchased from Carnival Corporation, the
price per share will be the closing price for the Carnival Corporation common stock on the
NYSE Composite Tape on the dividend payment date, as reported in THE WALL STREET
JOURNAL or other authoritative source. The price per share for open market purchases will be
the weighted average price paid by the Agent for all shares of Carnival Corporation common
stock purchased by it for participants in the Plan through negotiation with the seller. No share of
Carnival Corporation common stock will be purchased at a price in excess of current market
prices at the time of purchase.

Appendix B. Robustness Tests

B.1. Descriptive Statistics for DRIP Stocks and Non-DRIP Stocks in Portfolios II and III
In Tables 1 - 3 of the main text, we provide the summary statistics for the main variables
in our analysis for the first portfolio (I: All Stocks), as well as for the subsets of all DRIP stocks
and all non-DRIP stocks, over the sample period, 2008 - 2012. In Tables B.1 - B.3 of this
Appendix, we present the analogous summary statistics for the second and third portfolios (II:
High_DY and III: Hard_Arb), over the same sample period. In Table B.1., as we proceed to
consider each successive portfolio, II and III, the price run-up and reversal around the pay date
(AR(0), CAR(0,+1), and CAR(+2,+10)) grow larger in magnitude, and the associated mean

51

differences across DRIP stocks versus non-DRIP stocks become larger and more significant. In
Table B.2, the negative correlations between AR(0) or CAR(0,+1) and CAR(+2,+10) become
larger in magnitude as we examine the finer subsets of stocks in portfolios II and III with a
higher dividend yield and greater limits to arbitrage. Finally, in Table B.3, the subsets of DRIP
stocks and non-DRIP stocks in portfolios II and III reveal average firm attributes that remain
fairly stable. For example, these respective subsets of DRIP stocks tend to be larger in size, and
have higher institutional ownership, smaller spreads, and lower return volatility, in comparison
with non-DRIP stocks.
B.2. Re-Estimating the Main Analysis over the Extended Period, 1996 - 2012
In this Appendix we duplicate the comparative analysis of DRIP stocks versus non-DRIP
stocks provided in Figure 2 and Table 4, but we analyze the extended period covering the years,
1996 - 2012, for which we have lists of DRIP firms from AAII. As noted in the text, the AAII
lists indicate an unrealistic increase of 230 new DRIP firms from 2007 to 2008. This increase
strongly suggests that these lists omit a substantial number of firms with DRIPs prior to 2008. As
a result, before 2008 the implied lists of non-DRIP firms are likely to be corrupted by the
inclusion of many firms with DRIPs. Because of this problem with our list of non-DRIP firms, in
our main analysis of Figure 2 and Table 4 we restrict the sample period to the years, 2008 - 2012.
Here we analyze all dividend-paying stocks for the years, 1996 - 2012, using the AAII
lists to distinguish DRIP stocks from non-DRIP stocks every year. However, for the earlier years
before 2008, we pursue the conservative approach of excluding the 230 dividend-paying firms
that the AAI lists indicate began their DRIPs in 2008. This exclusion ensures that we will not
incorrectly include these likely DRIP firms in our list of non-DRIP firms for the earlier years.

52

The evidence for the extended sample period is presented in Figure B.1. and Table B.4.,
and is similar to the results provided in Figure 2 and Table 4, respectively, in the main body of
the paper. This evidence indicates that our major results also pertain to the longer sample period
beginning in 1996, for which we have lists of DRIP firms from AAII.32
B.3. Using the Fama-MacBeth Approach to Estimate the Panel Regression Models
Tables B.5. and B.6. present the results from applying the Fama-MacBeth approach to
estimate the panel regression models in Equations (2), (3), and (4). Equation (2) pertains to the
relation between AR(0) and firm characteristics, which is analyzed in Table 4 over the period,
2008 - 2012, using clustered standard errors. Equations (3) and (4) involve our proxy for DRIP
participation, which is analyzed in Table 5 over the period, 2010 - 2012, using clustered standard
errors. The results in Tables B.5. and B.6. are similar to the evidence in Tables 4 and 5,
indicating that this support for the price pressure hypothesis does not depend on the technique
used to estimate these panel regression models.

Appendix C. The Intraday Price Pattern on the Dividend Pay Date

In this appendix we examine the intraday pattern in price movements on the dividend pay
date, for the subsets of DRIP stocks in Portfolios I - III. This analysis reveals whether investors
(or their transfer agents) tend to exert price pressure at certain times during the day that dividend
funds are distributed. We analyze intraday midquotes at five-minute intervals for the first and

32 The AAII annual data on DRIP firms for the earlier years, 1996 - 2002, also provide additional information about
certain features of these DRIPs. For example, for this sub-period, 25% of all DRIPs charged a fee for participation,
while 7% offered the opportunity to reinvest dividends at a discount of 1% to 5% below the market price. In analysis
not reported here, we find no evidence that the existence or magnitude of a fee or a discount affects the pay date
effect, AR(0), for the 25% of DRIP firms that charged a fee, or the 7% of DRIP firms that offered a discount.

53

last 30 minutes of trading on day 0, and at thirty-minute intervals during the rest of this trading
day. We also analyze the last three hours of the previous trading day, on day -1.33
For each stock analyzed, we begin by computing the ratio of the midquote at every
intraday interval (T) to the closing midquote on day 0. For each portfolio, the average intraday
price pattern is then calculated in two stages. First, for every quarter in our sample period, 1996
through 2009, we calculate the cross-sectional average price ratio across dividend events at every
intraday time interval (T) during days -1 and 0. Second, we compute the time series means of
these quarterly cross-sectional average intraday price ratios, across all quarters in the sample.
Results are plotted in Figure C.1. for the DRIP stocks in Portfolios I - III. For each
successive subset of DRIP stocks, the mean intraday pattern begins at a lower price point, and
thereby reflects a larger increase on day 0. The magnitude of the price increase from the close on
day -1 to the close on day 0 ranges from 5 basis points (bp) for the DRIP stocks in Portfolio I
(i.e., 1 - .9995), to 40 bp for the DRIP stocks in Portfolio II (i.e., 1 - .9960), to 85 bp for those in
Portfolio III (i.e., 1 - .9915). For portfolio I, this evidence suggests a smaller close-to-close return
than is documented in Table 1 and Figure 2. For portfolios II and III, this evidence closely
corresponds to the results in Table 1 and Figure 2. These results indicate that each successive
portfolio of DRIP stocks has a larger mean abnormal close-to-close return on day 0.
In Figure C.1., these intraday patterns of price movements reveal how the price increase
transpires gradually throughout the trading hours on the dividend pay date, for these portfolios of
DRIP stocks. First, prices are roughly flat during the last three hours of trading on day -1, before
rising 10-15 bp in the last five minutes of trading. Then, for portfolios II and III, the mean
opening midquote on day 0 is within a few basis points of the closing price on day -1, indicating

33 Midquotes are analyzed rather than trade prices, because trade prices may tend to occur at the bid or the ask at
certain times of the day (e.g., at the open or the close). See Berkman et al. (2012) for issues regarding this approach.

54

that the mean overnight return before day 0 is also flat for these portfolios of DRIP stocks. This
result suggests that transfer agents do not focus their buying at the open on the dividend pay date.
After the open on day 0, the average price for each portfolio increases gradually
throughout the trading day, and then accelerates during the last five minutes of trading. It is
noteworthy that a large portion of the variation in the mean close-to-close return on day 0, across
the DRIP stocks in Portfolios I - III, appears during the last five minutes of trading. Figure C.1.
reveals an average 5-minute price increase at the close on day 0 that ranges from roughly 10 bp
for Portfolio I to 25 bp for Portfolio III. Together, this evidence suggests that transfer agents
gradually buy these stocks throughout day 0, and then perhaps accelerate their purchase orders
just before the close in an apparent attempt to complete their DRIP purchases on the pay date.

Appendix D. Extended Analysis of the Time Series of Quarterly Profits

This Appendix extends our analysis of the daily and quarterly time series of abnormal
market-adjusted profits, AR(0)_kt and CAR(0)_kn, k = I - III, from our three proposed trading
strategies that are plotted in Figures 5 and 6, respectively. First, we compute the average riskadjusted daily return for each strategy by estimating the Fama-French alpha for the daily stream
of excess profits. Second, we analyze the quarterly stream of time series movements in actual
profits for each strategy, without subtracting the market return. Third, we explore the robustness
of the results from our time series regression analysis of determinants of the quarterly
performance of the second and third trading strategies, CAR(0)_kn, k = II and III.
D.1. Fama-French Regression on the Daily Stream of Excess Actual Profits, Ret(0)_kt - Rft
Table D.1. provides further analysis of the risk and reward characteristics of the daily
stream of excess profits from our three trading strategies over the extended period, 1996 - 2012.
This Table presents the results from regressing the daily mean excess return from each strategy,

55

(Ret(0)_kt - Rft), k = I - III, against the three daily Fama-French factors, along with the daily
momentum factor. This Table shows that the Fama-French daily alphas from this analysis are
very large and highly significant (alpha = 0.29% for portfolio I, 0.50% for portfolio II, and
0.97% for portfolio III). This evidence indicates that, even after controlling for common sources
of risk, our proposed trading strategies yield average risk-adjusted returns that range from 30 to
100 basis points per day.
D.2. Time Series Movements in Quarterly Actual Profits
One conspicuous feature of Figures 5 and 6 is the presence of large spikes in the mean
daily abnormal profits AR(0)_It and quarterly cumulative abnormal profits, CAR(0)_kn, during
the liquidity crisis of 2008-2009. We conjecture that these spikes reflect a higher liquidity
premium during the crisis. Here we explore the alternative possibility that these spikes in marketadjusted abnormal returns could be due to the large decline in the market during the crisis.
In Figure D.1., we abstract from market movements by plotting the analogous time series
of quarterly cumulative actual profits from these three strategies (CRet(0)_kn, k = I - III),
without subtracting the market return. The use of market-on-close orders to implement these
strategies would assure that a trader realizes these average quarterly actual returns (CRet(0)_kn)
plotted in Figure D.1. This Figure reveals spikes in CRet(0)_kn, k = I - III, during 2008 and 2009
that are similar to the analogous spikes in CAR(0)_kn plotted in Figure 6. Furthermore, when we
do not subtract the market return (which is positive in most quarters), the cumulative actual
returns in Figure D.1. tend to be higher than the analogous cumulative abnormal returns in
Figure 6 for most quarters throughout the sample period. As a result, the cumulative actual return
(CRet(0)_kn) averaged across all quarters is somewhat higher than the average quarterly

56

cumulative abnormal return (CAR(0)_kn) in Figure 6, for all three strategies (i.e., 18.9% for
portfolio I, 23.7% for II, and 20.5% for III).
D.3. Determinants of Time Series Movements in Quarterly Profits for Portfolios II and III
Panels A and B of Table D.2. present the results from estimating Equation (6) for the
successively smaller subsets of DRIP stocks in portfolios II and III, respectively, over the
extended sample period, 1996 - 2012. The four columns in each Panel provide the coefficient
estimates for different permutations that include different combinations of the independent
variables in Equation (6). The results are generally consistent with the evidence in Table 7, for
the portfolio of all DRIP stocks. Consider the results for each independent variable, in turn.
First, the coefficient of the recession dummy (1) is positive for all permutations of the
model estimated in Panels A and B of Table D.2., and significantly so in Panel B. This evidence
suggests a tendency for a larger pay date effect during recessions, and thus greater cumulative
profits for our second and third trading strategies in such times of market stress, when there is
often a premium placed on dividend-paying stocks. Second, the coefficient of the time trend (2)
is positive and significant for all but one permutation of the model in Panels A and B, indicating
a tendency for the pay date effect to grow in magnitude over time, consistent with the evidence
in Figure 1. Third, in both Panels A and B, the average dividend yield is positively related to the
magnitude of AR(0) for all permutations of the model (i.e., 3 > 0), and significantly so in Panel
A, for the DRIP stocks in portfolio II. This outcome supports the view that a higher dividend
yield for these portfolios of DRIP stocks tends to be associated with greater profits from trading
on the pay date effect. Fourth, the stream of cumulative profits is also significantly positively
related to time series movements in the average spread (i.e., 4 > 0), for the DRIP stocks in both
portfolios II and III. This evidence indicates that the pay date effect is larger during periods of

57

lower liquidity (i.e., higher spreads) for these portfolios. Fifth, the coefficient of aggregate
liquidity (5) is negative for all permutations of the model in Panels A and B, and marginally
significant for one permutation in Panel A. This outcome suggests a weak tendency for the DRIP
stocks in portfolios II and III to have a larger pay date effect during periods of lower aggregate
market liquidity. Sixth, there is no evidence in Table D.2. of a significant relation between the
VIX and the stream of cumulative profits from these two trading strategies. Finally, the
coefficient of the dividend premium is positive in both Panels A and B, and significantly so in
Panel B, indicating that the stream of profits from holding the DRIP stocks in portfolio III are
higher during periods of greater demand for dividend-paying stocks.34

34 In untabulated results, we have also estimated the quarterly time series regression model in Equation (6), using
the quarterly cumulative actual return, CRet(0)_kn, as the dependent variable. In addition, we have also estimated
the daily Fama-French 3-factor and 4-factor models using the daily actual excess return, Ret(0)_kt - Rft, as the
dependent variable. Together, these results reinforce the evidence in Figure 6 and Tables 6, D.1., and D.2.

58

Figure B.1. Mean ARs and CARs for DRIP Stocks or Non-DRIP Stocks: Extended Period
This Figure analyzes dividend pay dates over the extended period covering 1996 - 2012. We plot the mean
abnormal returns (ARs) and cumulative abnormal returns (CARs) across all 21 days in the event window,
(-10,+10), around dividend pay dates (on day 0), for the DRIP or non-DRIP stocks in three portfolios:
I. All_Stocks = all dividend-paying stocks each quarter;
II. High_DY = the top 33% of all dividend-paying stocks each quarter by dividend yield;
III. Hard_Arb = top 33% by dividend yield, bottom 33% by institutional ownership, and top 33% by spread.
First, daily abnormal returns are computed by subtracting the return on a benchmark portfolio matched to
each stock by size and book-to-market ratio. Second, for the DRIP or non-DRIP stocks in each portfolio, we
compute the cross-sectional average ARs and CARs for all 21 days, during every quarter in the period, 1996 2012. Third, for each portfolio we compute the time series mean of the cross-sectional average ARs and CARs
across all quarters. Panels A and B plot the mean ARs and CARs, respectively, for the DRIP stocks in each
portfolio. Panels C and D plot analogous results for the non-DRIP stocks in each portfolio. The 95% confidence
band for the ARs in the third portfolio is provided in Panels A and C, since this portfolio has the widest band.

Panel A. Mean ARs for Subsets of DRIP Stocks in Three Portfolios

1

All_Stocks

Percent

0.5

High_DY
Hard_Arb
L95 - Hard_Arb

0
-10

-8

-6

-4

-2

10

U95 - Hard_Arb

-0.5
10 Days Before and After Dividend Pay Date

Panel B. Mean CARs for Subsets of DRIP Stocks in Three Portfolios

1

0.5
Percent

All_Stocks
High_DY
Hard_Arb

0
-10

-8

-6

-4

-2

-0.5
10 Days Before and After Dividend Pay Date

59

10

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Figure B.1., continued

Panel C. Mean ARs for Subsets of Non-DRIP Stocks in Three Portfolios
1

All_Stocks

Percent

0.5

High_DY
Hard_Arb
L95 - Hard_Arb

0
-10

-8

-6

-4

-2

10

U95 - Hard_Arb

-0.5
10 Days Before and After Dividend Pay Date

Panel D. Mean CARs for Subsets of Non-DRIP Stocks in Three Portfolios

1

0.5
Percent

All_Stocks
High_DY
Hard_Arb

0
-10

-8

-6

-4

-2

-0.5
10 Days Before and After Dividend Pay Date

60

10

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

Figure C.1. Intraday Price Pattern on the Dividend Pay Date

This Figure plots the average pattern of price movements over the last three hours of trading on the day
before the pay date (day -1), and all trading hours on the pay date (day 0), for the subsets of DRIP stocks in
portfolios I - III over the period, 1996 - 2009. We analyze intraday midquotes at five-minute intervals for
the first and last 30 minutes of trading, and at thirty-minute intervals during the rest of the trading day.
First, for every stock we compute the ratio of the midquote at every intraday time interval (T) to the
closing midquote on day 0. Second, for every quarter, we calculate the cross-sectional average price ratio
across the firms in every portfolio, at every intraday interval (T). Third, for each portfolio we compute the
time series means of these quarterly cross-sectional means, across all quarters.
1
0.998

0.996
I. All Stocks
II. High_DY

0.994

III. Hard_Arb

0.992
0.99
0.988

1:00
2:00
3:00
4:00
- (Day -1) - - - - - - Close

10:00 11:00 12:00 1:00 2:00 3:00 4:00

Open - - - - - - - - - (Day 0) - - - - - - - - - Close

Time of Day (T)

61

Figure D.1. Time Series of Quarterly Profits, the Mean Cumulative Actual Return, CRet(0)_k n,
for the Subsets of DRIP Stocks in Portfolios, k = I - III, that Pay Dividends on Any Date, t, During Quarter n
This Figure plots the quarterly time series of the cumulative actual return, CRet(0)_kn, obtained by aggregating the
daily cross-sectional mean actual returns on the dividend pay date, Ret(0)_kt, across all days in every quarter for
which at least one DRIP stock in each Portfolio (k = I - III) pays a dividend. Daily returns are measured from the
close on day -1 to the close on day 0. Every day (t), we first compute the mean cross sectional actual return on the
dividend pay date, Ret(0)_kt , for the subset of DRIP stocks in each Portfolio (k = I - III) that pays dividends on that
date. We then compute the quarterly sum of this series of daily mean actual returns, Ret(0)_kt, over all days (t)
during the quarter (n) for which at least one DRIP stock in each Portfolio pays a dividend. The results reflect the
quarterly aggregate cumulative actual return, CRet(0)_kn, from three separate trading strategies that prescribe
holding the DRIP stocks in each portfolio (k = I - III) on their respective dividend pay dates during a given quarter (n).
140
120

I. All_Stocks

II. High_DY

100

CRet(0)_kn, in %

III. Hard_Arb
80
Avg CRet(0)_kn /qtr:
I: 18.9%
II: 23.7%
III: 20.5%

60
40

Avg #Firms/qtr:
I: 507
II: 235
III: 74

20
0
1996

1998

2000

2002

2004

2006

-20
62

2008

2010

2012

Table B.1. Average Behavior of Stock Prices around Dividend Pay Dates
for the Subsets of DRIP Firms and Non-DRIP Firms in Portfolios II and III
This Table presents the average abnormal returns for Portfolios II and III over different time frames around dividend pay dates for the
sample period, 2008 - 2012. The variables, AR(-3)in, AR(0)in, CAR(0,+1)in , CAR(+2,+10)in , and CAR(0,+10)in are the percent (cumulative)
abnormal returns measured over different portions of the 21-day event window (-10,+10) around the pay date (on day 0) for the ith firm in
the nth quarter. These abnormal returns are computed by subtracting the daily return on a benchmark portfolio matched to each stock by
size and book-to-market. "AR(ex-div)in" is the analogous abnormal return on the ex-dividend date, and "AR(0) - AR(ex-div)" is the mean of
the difference between AR(0)in and AR(ex-div)in. Panel A provides the results for Portfolio II (High_DY), which includes the tercile of all
dividend-paying stocks each quarter with the highest dividend yield. Panel B gives the analogous results for Portfolio III (Hard_Arb), which
includes hard-to-arbitrage stocks (i.e., the subset of all dividend-paying stocks each quarter in the: (i) top tercile by dividend yield, (ii)
bottom tercile by institutional ownership in quarter n-1, and (iii) top tercile by the percent spread on day -10). In each Panel we provide
seven sets of results for: (1) all stocks in that portfolio, (2) DRIP stocks, (3) Non-DRIP stocks, (4) the difference of means across DRIP
stocks and Non-DRIP stocks, and (5)-(7) the analogous results for a subset of matched pairs of DRIP stocks and non-DRIP stocks in that
portfolio. We describe the matching scheme in the text. We compute the average abnormal returns around the pay date for every
portfolio, using a panel regression of each return measure on a constant term, with standard errors clustered by firm (i ) and quarter (n).
Analogous mean firm event returns for the subsets of DRIP and non-DRIP firms in each Portfolio are computed similarly.
Portfolio II. High_DY

DRIP Stocks in II.

Non-DRIP Stocks in II.

II. Difference of Means

(2) - (3)

(1)

(2)

(3)

(4)

Mean

Mean

Mean

# firms per qtr

468 ***

235 ***

AR(-3)in %

.24 ***

AR(0)in %
CAR(0,+1)in %

Panel A.
Portfolio II

CAR(+2,+10)in %

II. Matched Pairs

DRIP - Non-DRIP

(5)
DRIP

(6)
Non-DRIP

(7)
Mean Diff

230 ***

75

75

.18 ***

.30 ***

-.12

.26***

.19*

.06

.46 ***

.66 ***

.26 **

.40

***

1.02***

.34**

.69 ***

.57 ***

.74 ***

.40 **

.34

***

1.20***

.36*

.84 ***

-.56 **

.05

-.31

-.27

-.16

.38

**

.88

.08

.80 *

-.18
.58

***

.20**

.34***

***

.83***

.00

-.15
.83 ***

-.54 **

-.51 *

CAR(0,+10)in %

.03

.22

ex-div AR(0)in %

.23 ***

.14 **

AR(0)in - AR(ex-div)in

.23 ***

.52 ***

.32 ***
-.07

63

-.04

Table B.1., continued

Panel B.
Portfolio III

Portfolio III. Hard_Arb

DRIP Stocks in III.

Non-DRIP Stocks in III.

III. Difference of Means

(2) - (3)

(1)

(2)

(3)

(4)

Mean

Mean

Mean

DRIP - Non-DRIP

III. Matched Pairs

(5)
DRIP

(6)
Non-DRIP

(7)
Mean Diff

24

24

.47***

.48***

-.01

# firms per qtr

195 ***

74 ***

120 ***

-47 ***

AR(-3)in %

.43 ***

.43 ***

.42 ***

.00

AR(0)in %

.73 ***

1.40 ***

.32 **

1.08

***

2.08***

.56**

1.52 ***

CAR(0,+1)in %

.92 ***

1.63 ***

.49 **

1.14

***

2.23***

.64**

1.57 ***

CAR(+2,+10)in %

-.87 ***

-.75 ***

-.30

-.90

-.81**

-.10

1.32

-.17

1.47 *

.38*

.46**

1.71***

.10

-.09
1.61 ***

-1.05 **

CAR(0,+10)in %

.04

.55

ex-div AR(0)in %

.48 ***

.48 ***

.25 *

.93 ***

AR(0)in - AR(ex-div)in

-.27
.49 ***
-.16

64

.82
-.01
1.09

**

***

Table B.2. Correlations across Return Measures over Different Time Frames around the Dividend Pay Date
for Portfolio II (High_DY) and Portfolio III (Hard_Arb)
This Table provides correlations across the return measures taken over three different time frames around the dividend pay da te: AR(0),
CAR(0,+1), and CAR(+2,+10). We compute these correlations across all stocks, DRIP stocks, and non -DRIP stocks within two portfolios selected
each quarter. Panel A presents the results for Portfolio II (High_DY), which includes the tercile of all dividend -paying stocks each quarter with the
highest dividend yield. Panel B provides the analogous results for Portfolio III (Hard_Arb), which includes hard -to-arbitrage stocks (i.e., the subset
of all dividend-paying stocks each quarter in the: (i) top tercile by dividend yield, (ii) bottom tercile by institutional owne rship in quarter n-1, and (iii)
top tercile by the percent spread on day -10). The mean correlations are calculated in two stages. First, every quarter we co mpute each pairwise
cross-sectional Pearson or Spearman correlation across the dividend events for each portfolio. Second, we compute the time seri es mean for each
pairwise cross-sectional correlation across all quarters in the sample period covering 2008 - 2012. The standard deviation of every time series
mean correlation is then used to construct the t-test of the null hypothesis that every mean correlation equals zero. The mean Pearson correlations
are above the diagonal, and the mean Spearman correlations appear below the diagonal. Correlations in BOLD are significant at the .05 level.
Panel A.

Portfolio II: High_DY

AR(0)

CAR(0,1)

CAR(2,10)

DRIP Stocks in Portfolio II

AR(0)

CAR(0,1)

CAR(2,10)

Non-DRIP Stocks in Portfolio II

AR(0)

CAR(0,1)

CAR(2,10)

AR(0)

1.00

.69

-.03

1.00

.70

-.05

1.00

.68

-.02

CAR(0,1)

.65

1.00

-.10

.67

1.00

-.12

.63

1.00

-.10

CAR(2,10)

-.04

-.07

1.00

-.04

-.07

1.00

-.04

-.08

1.00

Panel B.

Portfolio III: Hard_Arb

AR(0)

CAR(0,1)

CAR(2,10)

DRIP Stocks in Portfolio III

AR(0)

CAR(0,1)

CAR(2,10)

Non-DRIP Stocks in Portfolio III

AR(0)

CAR(0,1)

CAR(2,10)

AR(0)

1.00

.66

-.07

1.00

.68

-.08

1.00

.64

-.07

CAR(0,1)

.62

1.00

-.17

.64

1.00

-.20

.60

1.00

-.14

CAR(2,10)

-.07

-.14

1.00

-.07

-.16

1.00

-.06

-.12

1.00

65

Table B.3. Firm Characteristics for the Subsets of DRIP Firms and Non-DRIP Firms in Portfolios II and III
This Table summarizes the descriptive statistics for the firm characteristics analyzed in this study over the period, 2008 - 2012. The
characteristics are defined in Table 3. Panel A provides the results for Portfolio II (High_DY), and Panel B gives the results for Portfolio III
(Hard_Arb), which are described in Table B.1. In each Panel we present seven sets of results for: (1) all stocks in that portfolio, (2) DRIP
stocks, (3) Non-DRIP stocks, (4) the difference of means across DRIP stocks and Non-DRIP stocks, and (5)-(7) the analogous results for
a subset of matched pairs of DRIP stocks and non-DRIP stocks within that portfolio. The matching scheme is described in the text. We
compute the average firm characteristics for every portfolio, using a panel regression of each variable on a constant term, with standard
errors clustered by firm (i ) and quarter (n).

Panel A.

Portfolio II. High_DY

DRIP Stocks in II.

Non-DRIP Stocks in II.

II. Difference of Means

Portfolio II

Mean

Mean

Mean

DRIP - Non-DRIP

468 ***

235 ***

230 ***

$6,854 ***

$11,624 ***

$1,969 ***

Div_Yieldin %

1.33 ***

1.15 ***

Pct_Instin-1 %

51.56 ***

Spreadin %

II. Matched Pairs

Mean Diff

DRIP

Non-DRIP

75

75

$9,655 ***

$2,037***

$2,288***

-$251

1.50 ***

-.35

***

1.10***

1.17***

-.06

54.89 ***

48.15 ***

6.73

***

47.86***

49.05***

-1.20

1.14 ***

.78 ***

1.50 ***

-.72

***

1.34***

1.31***

.04

Log_Hiloin %

3.74 ***

3.35 ***

4.14 ***

-.80

***

3.71***

4.01***

-.30

Panel B.

Portfolio III. Hard_Arb

DRIP Stocks in III.

Non-DRIP Stocks in III.

III. Difference of Means

Portfolio III

Mean

Mean

Mean

DRIP - Non-DRIP

# firms per qtr

Sizein (millions)

***

***

III. Matched Pairs

DRIP

Non-DRIP

24

24

Mean Diff

# firms per qtr

195 ***

74 ***

120 ***

Sizein (millions)

$234 ***

$281 ***

$206 ***

$75 *

$180***

$208***

-$29 *

Div_Yieldin %

1.40 ***

1.17 ***

1.54 ***

-.37

1.14***

1.17***

-.03

Pct_Instin-1 %

24.96 ***

26.10 ***

24.26 ***

1.84

21.80***

21.87***

-.06

Spreadin %

2.54 ***

2.28 ***

2.71 ***

-.43

2.81***

2.84***

-.03

Log_Hiloin %

4.19 ***

3.91 ***

4.36 ***

-.45

3.93***

4.48***

-.55

66

-47 ***

***

***

***

Table B.4. Firm Characteristics and the Dividend Pay Date Effect: Extended Sample
This Table analyzes the following relation between firm characteristics and the abnormal return on the dividend
pay date, AR(0), over the extended period, 1996 - 2012, for which we have annual lists of DRIP firms from AAII:
AR(0)in = 0 + 1 DRIPin + 2 Div_Yieldin + 3 Log_Sizein + 4 Pct_INSTin-1 + 5 Spreadin + 6 Log_Hiloin + in . (2)
All variables are defined in Table 1 and Table 3. The annual lists from AAII contain an unusually large increase
of 230 new DRIP firms in 2008, suggesting that the earlier AAII lists omit many of these 230 DRIP firms prior to
2008. While we have validated the accuracy of the AAII lists for the most recent period, we cannot be sure
whether these 230 firms had DRIPs prior to 2008. We address this concern by taking the conservative
approach of excluding these 230 firms from the extended sample for the earlier years prior to 2008.
We present the results from applying two alternative methodologies to estimate this panel regression model,
including the Fama-MacBeth approach and a panel regression with standard errors clustered by firm (i) and
quarter (n). For the Fama-MacBeth approach, Newey-West robust standard errors of the mean quarterly FamaMacBeth coefficients are used to construct the t-ratios, which appear beneath the mean parameter estimates.

Variable
Intercept

Fama-MacBeth

Clustered Std Errors

by firm and quarter

.196

.252

1.6

DRIPin

.211
7.1 ***

Div_Yieldin

12.97
4.4 ***

Log_Sizein
Pct_INSTin-1

3
4

Log_Hiloin

5
6

.252
7.1 ***

13.85
3.5 ***

-.004

-.012

-0.5

-1.5

-.324

-.318

-4.9 ***

Spreadin

2.1 **

-4.7 ***

1.65

.346

1.4

0.2

1.04

2.25

1.6

2.9 ***

Avg # Firms / qtr

1,169

1,176

(Avg) R2

.019

.0065

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

67

Table B.5. Firm Characteristics and the Pay Date Effect: Fama-MacBeth Estimates
This Table uses the Fama-MacBeth approach to estimate the following panel regression model that describes
the relation between firm characteristics and the abnormal return on the dividend pay date, AR(0):
AR(0)in = 0 + 1 DRIPin + 2 Div_Yieldin + 3 Log_Sizein + 4 Pct_INSTin-1 + 5 Spreadin + 6 Log_Hiloin + in . (2)
This model is applied to the sample including all dividend-paying stocks (both with and without DRIPs) over the
period, 2008 - 2012. All variables are defined in Tables 1 and 3. Newey-West robust standard errors of the
mean quarterly Fama-MacBeth coefficients are used to construct the t-ratios, which appear beneath the mean
parameter estimates.

Intercept

.682
3.2 ***

DRIPin

.299
6.3 ***

Div_Yieldin

12.10
2.0 *

Log_Sizein

-.038
-2.4 **

Pct_INSTin-1

-.399
-2.7 **

Spreadin

4.70
2.2 **

Log_Hiloin

1.41
1.2

Avg # Firms / qtr

1,205

Avg R2 / qtr

.026

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.

68

Table B.6. Firm Characteristics, DRIP Participation, and the Pay Date Effect:
Fama-MacBeth Estimates
This Table uses the Fama-MacBeth approach to estimate the following two panel regression models that
describe: (i) the relation between our proxy for DRIP participation and firm characteristics, and (ii) the relation
between the pay date effect, AR(0), and our proxy for DRIP participation, as well as other firm characteristics:
Partin = 0 + 1 DRIPin + 2 Broker_Nonin + 3 Div_Yieldin + 4 Log_Sizein + 5 Spreadin + 6 Log_Hiloin + in. (3)
AR(0)in = 0 + 1 DRIPin + 2 Partin + 3 DRIPin* Partin + 4 Broker_Nonin + 5 DRIPin* Broker_Nonin
+ 6 Div_Yieldin + 7 Log_Sizein + 8 Spreadin + 9 Log_Hiloin + in .

(4)

The sample period covers the years for which we have data on broker non-votes, 2010 - 2012. All variables
are defined in Tables 1 and 3. Newey-West robust standard errors of the mean quarterly Fama-MacBeth
coefficients are used to construct the t-statistics, which appear beneath the parameter estimates.

Dependent Variable for (3) = Partin

Intercept

Dependent Variable for (4) = AR(0)in

.546

Intercept

9.6 ***

DRIPin

-0.2

DRIPin

.032

6.4 ***

-.027
-.030
-0.5

Part1in

-.163
-1.5

DRIPin*Part1in

.630
2.5 **

Broker_Nonin

Broker_Nonin

-.054

-1.9 *

-.012
-0.1

DRIPin*Broker_Nonin

.328
0.6

Div_Yieldin

Div_Yieldin

3.86

5.2 ***

Log_Sizein

2.0 *

Log_Sizein

-.025

-7.8 ***

Spreadin

5
6

Spreadin

15.64

Log_hiloin

-.568

Avg # Firms/qtr

1.04
0.6

2 + 3

t-statistic

Avg R2 / qtr

5.97
1.4

-5.6 ***

.001
0.1

5.5 ***

Log_Hiloin

8.06

Avg R2 / qtr

0.178
804

Avg # Firms/qtr

.467

3.2 ***

0.021
801

*** indicates statistical significance at the .10 level; ** at the .05 level; and *** at the .01 level.
a

We test H0: (2 + 3) = 0, by constructing the t-statistic for the time series mean of the sum of these two coefficients.

69

Table D.1. Fama-French Regressions on Daily Excess Returns, (Ret(0)_kt - Rft), from Three Trading Strategies
This Table presents the results from estimating a Fama-French 3 or 4-factor model to analyze the mean daily excess actual returns on the
dividend pay date, (Ret(0)_kt - Rft), from the three trading strategies that prescribe holding the subsets of DRIP stocks in the three portfolios
(k = I - III) that pay dividends on any given date (t). First, we construct Portfolios I - III each quarter over the period, 1996 - 2012, as described in
the text. Second, we compute the daily actual return on every dividend pay date (t) for each stock (i), Ret(0)it, over this period. Third, we
compute the mean daily actual return on the pay date, Ret(0)_kt, across all DRIP stocks in each portfolio (k = I - III) that pay dividends on any
given date (t). The resulting time series of daily mean returns, Ret(0)_kt, represent the daily actual returns to each of our three trading
strategies. Finally, we subtract the daily riskfree rate from each series of daily returns, to obtain the time series of daily mean excess returns,
(Ret(0)_kt - Rft), that represents the dependent variable analyzed in the Fama-French regression. Newey-West robust standard errors are used
to construct the t-statistics.
DRIP Stocks in Portfolio I (All_Stocks)

Intercept

t-stat

(Rm - Rf)

t-stat

HML

t-stat

SMB
UMD

3-factors

4-factors

3-factors

4-factors

.290

.295

.496

.505

.971

.985

12.0 ***

11.9 ***

14.2 ***

14.3 ***

12.0 ***

12.0 ***

.920

.889

.812

.759

.570

.518

23.7 ***

27.3 ***

15.6 ***

16.3 ***

.631

.576

.667

.563

.251

t-stat

8.1 ***

.261

4.7 ***

6.3 ***

.190

5.0 ***

t-stat

.398

.401

842.8 ***

640.1 ***

.686

.274

.281

357.8 ***

277.9 ***

N = 2,836 days

5.5 ***

.544

3.2 ***

3.1 ***

.209

.240

1.5

1.6

-.239

-3.1 ***

N = 3,814 days

# of days

2.9 ***

5.6 ***

-.200

-2.3 **

F-Stat

6.3 ***

.205

2.6 ***

-.117

Adj R2

DRIP Stocks in Portfolio III (Hard_Arb)

4-factors

7.7 ***

DRIP Stocks in Portfolio II (High_DY)

3-factors

-1.8 *

.100

.104

51.5 ***

40.7 ***

N = 1,363 days

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.
a

We obtain similar results when we analyze profits in terms of the market-adjusted or benchmark-adjusted abnormal
returns on the pay date, AR(0), rather than actual returns, Ret(0), and when we analyze the more recent period
covering the years, 2008 - 2012.

70

Table D.2. Determinants of Quarterly Profits from Trading Strategies

that Prescribe Holding the DRIP Stocks in Portfolios II and III
Panels A and B of this Table present results for the time series regression model that analyzes
determinants of the quarterly cumulative abnormal profits from our last two strategies, as follows:
CAR(0)_kn = 0 + 1 Recessionn + 2 Trendn + 3 Div_Yield_kn + 4 Spread_kn
+ 5 Agg_LIQn + 6 VIXn + 7 Div_Premn + n , where k = Portfolio II or III.
(6)
The dependent variable, CAR(0)_kn, is computed in two steps. First, for every day (t) in our
sample period, 1996 - 2012, we compute the stream of average daily profits as the crosssectional mean market-adjusted AR(0)_kt across all DRIP stocks in each Portfolio (k = II or III)
that pay dividends on that day. Second, for each quarter (n), we aggregate this stream of daily
profits, AR(0)_kt, across all days in which at least one DRIP stock in each portfolio pays a
dividend, to obtain the cumulative abnormal profit, CAR(0)_kn, for each portfolio (k = II or III).
Recessionn is a dummy variable that equals one for all quarters during recessions, and zero
otherwise. Trendn is a deterministic trend that counts the quarters in our sample. Div_Yield_kn
is the mean dividend yield in quarter n for the DRIP stocks from each Portfolio k (k = II or III).
Spread_kn is the mean daily closing percent spread on day -10 prior to the dividend pay dates in
quarter n, for the DRIP stocks in each Portfolio (k = II or III). Agg_LIQn is the aggregate liquidity
measure of Pastor and Stambaugh (2003), averaged across the three months during quarter n.
VIXn is the average of the monthly CBOE VIX index values during quarter n. Div_Premn is the
quarterly average of the monthly aggregate market dividend premium (pdnd), from Baker and
Wurgler (2006). The Newey-West t-ratios appear beneath the parameter estimates.

Panel A. Portfolio II: High Dividend Yield Stocks

Recessionn

t-ratio

Trendn
t-ratio

Div_Yield_kn

t-ratio

10.23

7.56

6.37

.15

1.3

1.0

0.8

0.0

.32

.41

.31

.71

1.9 *

2.4 **

1.5

2.7 ***

70.73

59.47

3.5 ***

Spread_kn

t-ratio

3.1 ***

14.91

16.33

3.2 ***

3.4 ***

Agg_LIQn

-1.01

t-ratio

-1.8 *

VIXn
t-ratio

Div_Premn

57.28
3.1 ***

12.63
2.1 **

2.5 **

19.93
2.7 ***

-.64

-.74

-1.0

-1.2

.56

.27

1.1

0.4

.20

t-ratio
Adj R2
Overall F

50.04

0.7
.60
26.0 ***

.63
23.8 ***

.64
20.6 ***

.69
19.7 ***

* indicates significance at the .10 level; ** at the .05 level; and *** at the .01 level.
The models in the first three columns are estimated over the entire 68-quarter sample period, 1996-2012. The
last column is estimated over the period, 1996-2010, due to data limitations on pdnd from Baker and Wurgler (2006).

71

Table D.2., continued

Panel B. Portfolio III: Hard-to-Arbitrage Stocks
Recessionn

t-ratio

Trendn

t-ratio

Div_Yield_kn

t-ratio

Spread_kn

t-ratio

Agg_LIQn

20.88

19.13
2.3 **

2.3 **

.88

.89

.82

5.0 ***

5.4 ***

5.3 ***

1.17
6.5 ***

8.98

6.47

5.57

0.8

0.5

0.5

13.89

13.62

12.00

16.31

4.8 ***

5.3 ***

4.7 ***

5.6 ***

-.53

-.27

-.36

-1.0

-0.5

-0.7

.46

-.14

0.9

-0.2

6
7

.51

t-ratio
Adj R2
Overall F

1.4

0.8

t-ratio

Div_Premn a

9.93

9.83

t-ratio

VIXn

18.76

2.4 **

2.4 **
.68
37.2 ***

.69
30.4 ***

72

.69
25.6 ***

.75
26.6 ***