Dividend reinvestment plans : do the companies using DRIPs benefit from greater funding flexibility during market distress?

(1)

1

Dividend Reinvestment Plans – do the companies

using DRIPs benefit from greater funding flexibility

during market distress?

By

Lukasz Szalasnik

Submitted in partial fulfillment for the degree of

MSc Finance

University of Amsterdam

Amsterdam Business School

June 2015

Thesis Supervisor:

(2)

2

ABSTRACT

This study sought to extend the work of Boehm and DeGennaro (2007) with respect to dividend reinvestment plans. These authors employed a set of independent variables in a logistic regression to predict whether a firm would institute a DRIP in the future. In addition to the characteristics employed by Boehm and DeGennaro, this study sought to use three indices whose purpose is to measure the level of financial constraint under which a firm is operating--the Kaplan-Zingales index (KZ), operating--the Whited-Wu index (WW), and operating--the Hadlock-Pierce index (HP). Data for the years 2004-2010 were collected, and the presence of a DRIP determined. Not only did this study fit a logistic regression model to the data and conduct an exploration of a

difference in means but it also fitted a mixed effects regression with interaction to the data. The interactions between the KZ Index and a firm’s employing dividend reinvestment plan and between these two and period (i.e., pre-crisis vs. crisis) were statistically significant.

Keywords: dividend reinvestment plan, Kaplan-Zingales index, Whited-Wu index, Hadlock-Pierce index, logistic regression

(3)

3

DEDICATION

I dedicate this work to my girlfriend, Aga, who has been a constant source of support and motivation throughout the process. I never could have made it without you.

I would also like to dedicate this thesis to my parents, who never let me gave up and taught me how to fight to achieve my goals. I will always appreciate your determination to

(4)

4

LIST OF FIGURES

Figure 1. Regression Model with Interaction for Years 2004-2008 (Pre-Crisis) ... 37 Figure 2. Regression Model with Interaction for Years 2008-2010 (Crisis) ... 37 Figure 3. Group Means (NN, NY, YN, and YY) with respect to Crisis/DRIP Grouping ... 38

(8)

8

LIST OF TABLES

Table 1 Sample Statistics ... 46

Table 2 T-Tests, Firm Status with Respect to DRIPs during the 2003-2007 Period and the 2008-2010 Period ... 47

Table 3 Logit Results, 2003-2010 Data ... 48

Table 4 Correlations, 2004-2010 Data ... 50

(9)

9

CHAPTER 1. INTRODUCTION

Dividend reinvestment plans (DRIPs) have been introduced for a long time and remain a viable alternative for U.S. companies. At first, these DRIPs were market plans in which a

company bought its outstanding shares in the market and then sold them at its shareholders at a discount. In time the procedure was modified, allowing companies not to use the market as an intermediary. In the early 1980s, utilities introduced unissued shares to provide stocks to the investors that wanted to reinvest their dividends. These issuing plans allowed the company a way to procure additional equity in the form of reinvested dividends.

There are benefits that result from managers being allowed to convert the dividends they owe to shareholders into additional shares at a discount. In this way, both costly underwriting procedures and brokerage fees are circumvented. Also firms can fulfill their payment obligations easier, since a significant amount of cash remains in the company. The investors will also have increased confidence in the company.

The present paper addresses two questions. The first one is in what way do economic variables predict whether a company will start to use DRIPs or not. If there is a reliable way to estimate if a company will proceed on this path, then there are benefits to be obtained with this information. The second question is how well have companies that used DRIPs fared during the financial crisis of 2007-2008. In that period a lot of companies were faced with problems of liquidity and it remains to be established if the companies that adopted DRIPs were affected in

(10)

10

a lesser extent by negative factors during the crisis and if they outperformed the other companies during periods of economic calm.

These questions contain several areas that were not examined in previous studies and are investigated here, such as in what extent companies that implemented DRIPs were affected by negative economic factors during the financial crisis. The intuitive argument is that investors have more confidence in the company if the DRIPs alternative is offered and they are less inclined to sell the company’s stock. However, they can also have less confidence in the company and be inclined to sell earlier in financial crisis periods, because more of their funds are invested in the company as they have not received their dividends that were converted into stocks. What argument is true in what periods remains to be confirmed by the data. Another area is in what way can investors that know how likely a company to implement DRIPs is, can benefit from the information. The financial characteristics of firms that implement DRIPs are different from companies that do not. In this way, an investor can estimate what are the firms that are the most likely to initiate DRIPs in the immediate future and with that knowledge he can trade the company’s stocks. The answers to these questions are of importance to

managers, investors and researchers.

The remainder of the thesis is organized as follows. The following chapter is a brief review of the literature concerning dividend reinvestment plans. The third chapter presents a discussion of the methodology and the data that were used in the study. The results and their interpretation are presented in the fourth chapter, while the fifth chapter concludes.

(11)

11

CHAPTER 2. LITERATURE REVIEW

The dividend reinvestment plans (DRIP) allow shareholders to reinvest partially or totally their entitled dividends in new shares. Reilly and Nantell (1979)and Finnerty (1989) argue that there is no new wealth created when a discount DRIP is realized, just a transfer of wealth from the investors that do not participate to the investors who participate by

purchasing the new shares with a discount. Finnerty (1989) elaborates that the wealth

transferred is directly proportional to the discount. He further states that if a company has to perform equity financing while unable to retain all its earnings, the DRP solution is less costly, because a small discount can be less than the costs associated with issuing new shares. The amount of equity capital that is optimal to be raised depends on the elasticity between shareholder participation and the size of the discount. Willis (1989) states that this relation is positive. This is also confirmed by Todd (1992) based by on an examination of US firms that offered the discount DRIP.

Direct stock ownership is not viewed as a viable solution on the long term, because it entails high fees and not enough diversification. Dividend reinvestment plans and direct

investment plans have reduced the problems of direct stock ownership by allowing investors to circumvent the need for brokers. Dividend reinvestment plans allow shareholders to easily convert their dividends into new shares. If this feature is not just for current shareholders and new shareholders can benefit from it too, then it is called a direct investment plan. DRIPs are not a class of security, such as forward and futures contracts. They represent a new method of

(12)

12

selling securities. DRIPs have significantly reduced transaction costs and made it possible for investors to diversify easier.

In the past, mutual funds did something similar, by catering to the needs of the small investors by reducing transaction costs. The Scudder funds opened the first mutual fund in 1920s (Carlson, 1997). The mutual fund operates on the principle that small investors put their money in a pool that allows fund managers to make large purchases of stock. In this way, the transaction costs are shared between all the investors. This solves two of the problems small investors face, because the transaction costs are lower and there is a significant diversification when considering the whole asset structure of the mutual fund.

There are also some drawbacks related to mutual funds, such as management fees that can reach up to 7% and the fact that an investor may not want to have the exact portfolio structure of another investor and take some decisions personally.

Dividend reinvestment plans first started to appear in the 1960s when AT&T offered this facility to the shareholders. The plans have been further expanded by other companies in order to unsure shareholder participation.

Mutual funds as well as DRIPs are a suitable choice for investors who believe that dollar-cost averaging is viable. This involves investing approximately the same amount in the same security at periodic time intervals. In this way, the average purchase price is less than the arithmetic average of the share’s prices on purchase dates. Participating in DRIPs is considered a good idea for these investors, because reinvesting dividend payments is the equivalent of dollar-cost averaging.

(13)

13

DRIPs also have some advantages, because it eliminates the costs related to investment bankers and other administrative and accounting fees when issuing new shares. New-issue DRIPs also allow for large sums to be raised. Scholes and Wolfson (1989) report that firms with no discounts on reinvestments raise 12 percent of the total common and preferred dividends they pay. When a 5 percent discount is offered, this percentage rises to 98. This shows that there is a great elasticity between stock purchase and discounts. Another advantage is that DRIPs provide goodwill to the investors. Even if not all investors use it, it is good to have a feature such as DRIPs in order to give them confidence that the company is professional. The third advantage applies especially to large companies. These companies, if they have a great number of shareholders, can easily determine them to purchase its products. In this way, the firm has extended its market share.

It is important to determine if there is a way to anticipate whether a company will use a discount DRIP and what are the factors that can give an accurate prediction of that fact. That is the purpose of this article. It builds upon the work of Boehm and DeGennaro (2007) who examined the likelihood of firms without DRIP to adopt one. Given the performance registered by Scholes and Wolfson (1989), the ability to estimate when such a prediction will be made consists in a very profitable trading opportunity. Also companies that administer direct investment plans can find more customers more easily. Because their fees are lower than the underwriting costs of banks, they represent a serious competition. This in conjunction with information regarding what firms are going to adopt DRIP confers them an edge over finding clients.

(14)

14

There are many ways on which firms that offer DRIP are different from firms that do not offer them. That is because companies that offer DRIP usually do so in order to obtain cheap capital. That can also mean that there is a problem in the way the assets are structured and the firm may be experiencing difficulties. That is why in a firm must prove it has a stable financial situation when making a DRIP announcement, in order to benefit from credibility from the public.

Once the factors have been established, it remains to be seen if the announcement effects have a positive influence on the share price. Hansen et al. (1985) stated that the

announcements should have a null or positive effect on the stock price. Also shareholders who do not participate are not necessarily in a bad position. Generally it is agreed in the literature that the announcement has a positive influence. Such is the opinion of Dubofsky and Bierman (1988), while Perumpral et al. (1991) found no correlation between these two variables. The announcements often mean that the DRIP features will have discounts and in this situation it is reasonable to assume that the demand for the stock will increase, as more and more

shareholders see the benefits of purchasing more stock.

By benefiting from the discount DRP offers of J. P. Morgan, Scholes and Wolfson (1989)claimed to have tripled their investment in about two years. However, they recognize that even if the market is inefficient, arbitrage opportunities are available only in the short term and eliminated in the long term. Their substantial profits make a convincing case for purchasing shares with discounted DRIP, but companies have modified their policies significantly since their experiment has been made. Dubofsky and Bierman (1988) argue that DRIP resolve market imperfections, such as transaction costs, taxation and information asymmetry.

(15)

15

Peterson et al. (1987) examine stock price reaction to the adoption of dividend reinvestment plans for utilities and industrials and presume that a DRIP action is perceived in the same way as issuing new equity and can cast a shadow of doubt about the future prospects of the firm. Their results are not conclusive since they focus only on utilities companies.

Noe (1988)shows that there will be cases when big companies prefer equity to debt financing, but that on average the quality of firms issuing equity will be lower than the quality of firms financing the debt. Good quality firms that need to issue equity will try to manage the adverse selection problem of new equity issues by offering a discount DRIP. This is opposed as what was stated by Finnerty (1989), because in this signaling equilibrium, the discount

represents the cost of the signal rather than a transfer of wealth from stockholders who do not participate to ones that do participate. In this way the adoption of a DRIP without discount is similar to issuing equity, but the reaction to the adoption of a discount DRIP should be less negative or even positive.

When capital markets are perfect and the information is symmetrical, financial policy will not affect the value of the company. Financial policy also contains dividend policy and dividend reinvestment plans, so the assumption is that implementing a DRIP will not affect firm value. However, there are also time periods and markets where the information is

asymmetrical.

Disadvantages of DRIPs

It is true that DRIPs contain some limitations. Scholes and Wolfson (1989) state that there is an array of implicit costs related to DRIPs. DRIP investors must be vigilant and look over the details of the plans and know when there will be changes in terms. The DRIP purchases

(16)

16

don’t have a cost, but there are always costs to be considered when selling shares. Selling stocks can even mean to request stock certificates in order to deliver them to a broker for sale.

Another disadvantage is that DRIP investors must keep their securities in their own names. This makes it so that lending the securities is unfeasible and the investors forfeit the interest and fees they might gain. By contrast, many brokers keep their customers; securities in street name or their own name, so that the customers have the right to lend the securities for short sales and collect any resulting fees.

There are both positive and negative influences on stock price and shareholder wealth when DRIPs are established. There are the following arguments for a negative influence. Establishing a plan reduces the scrutiny of a firm because the firm avoids entering the financial markets to raise cash. The reduced level of scrutiny determines the investors to be unsure about whether or not to participate and a smaller stock price. Barnea et al. (1980) provide evidence that when companies enter the financial market, the agency costs are reduced and there are more benefits for the investors. Also when a reliable underwriter certifies a

transaction, then the offering is considered to be of quality and credible to the investors. Booth and Smith (1986)state that when the insiders of a company are unable to convey credibility and when outsiders are unable to easily obtain information about a company, there is a reason to use an investment banker. Tinic (1988) argues that when the underwriting certification is lost then the offering will determine a reduced stock price and shareholder wealth. Among the studies that have found a negative relation between DRIPs and stock price, we mention Peterson, Peterson and Moore (1987) as well as Dhillon et al. (1992).

(17)

17

Advantages of DRIPs

The DRIPs have positive effects, such as improving shareholder relations. Companies are convinced that DRIP will make their shares more appealing to investors and create goodwill. If demand is on the rise, it is easy to anticipate a price increase. Hansen et al. (1985) examined the extent to which a DRIP announcement influences stock valuation. Their argument is that the share price will increase if the current value of additional expected earnings is higher than the value of the dividends that have been reinvested. Cherin and Hanson (1995) state that the most important benefit is being able to purchase shares cheaply without brokerage fees. From the company’s point of view, not having to pay underwriting costs is also a welcomed addition. Perumpral (1983) found that the US market had a favorable impression when a DRIP

announcement was made, but this attitude was not universal.

The arguments that support a positive influence on the stock price are the following. Asquith and Mullins (1986) present the argument that when a new offering of common stock is realized there are negative signals sent to the investors that result in a decreased shareholder wealth. DRIPs are an alternative method of raising new equity financing and offers benefits to the company and its shareholders by avoiding the downsides of the negative signals. By avoiding the need of a new offering, the earning retention provides the company that issued DRIPs a way to not pay underwriting fees and other costs related to the issuing. Scholes and Wolfson (1989) argue that DRIPs mitigate the negative influence of new equity offerings. There are two reasons that make DRIPs able to achieve this. The first is that the firm can issue capital over a period of time rather than all at once, so that the investors have more time to be informed about the practices of the firm and more likely to participate. In this way the

(18)

18

information asymmetry will be reduced, because the slow down process would permit some information management has to be made public by accounting disclosures, performance methods and other outlets. The second is that discount DRIPs can be used as a signal to investors to send them the message that all is well with the company as the firm can afford starting this transaction. Overall, the positive aspects of DRIPs, such as avoiding paying

underwriting fees by implementing them, counterbalance other negative aspects, such as the lack of scrutiny and the certification from a prestigious underwriter, resulting in a net benefit for the company.

Healy and Palepu (1988)show that there is a positive relation between divided changes and stock price. The dividend changes are a signal about cash flows and future investments in a firm. As a consequence, managers don’t want to reduce or eliminate dividend payments. It is reasonable to assume that managers are better informed about the true value of the stock price. It is also possible for the stock to be undervalued if there is a certain level of information asymmetry. Managers can also decide to obtain benefits from the information asymmetry, by purchasing the undervalued stock. The degree of information asymmetry can be estimated with the market-to-book ratio. Repurchase DRIPs allow companies to keep paying dividends to shareholders in the form of market shares.

Another advantage is given by Davey (1976), who states that DRIPs is useful for

companies that want to diminish institutional ownership, because it can attract small individual investors. In this way, management has the control over the company and the possibility of a hostile takeover is diminished. Dittmar (2000)states that a repurchase facility can be used as a

(19)

19

defense against a takeover, because he has the opinion that repurchases increase the acquisition price.

Jensen (1986) states that when the level of cash flow exceeds the worthwhile

investment opportunities with positive Net Present Value, there can be disagreement between shareholders and managers regarding the optimum dividend policy. This has the upside of increasing firm value, because the investments that would result in losses are avoided. Bajaj and Vijh (1990) state that repurchasing stock is a more beneficial way of distributing capital, because if the dividends are reduced, there is a significant decrease in confidence from the part of the shareholders. Repurchase DRIPs are a good way to distribute excess capital, because the dividend payments and the stock repurchase alternative are provided to investors at the same time and they can choose what suits them best while having a good impression about the company.

Among the further studies that support a positive relation between DRIPs and stock price, it is worthwhile to mention the studies of Perumpral et al. (1991), Ogden (1994) and Roden and Stripling (1996). All of the above studies found positive stock evolutions to the announcements of DRIPs introduction. Another branch of DRIPs studies has focused more on the investors’ perspective and the potential wealth effect due to discounted stock prices. Of these, a noteworthy study is done by Dammon and Spatt (1992), who tried to derive the value of an option to convert dividends into extra shares.

(20)

20

CHAPTER 3. METHODOLOGY

The subject of interest in this research was the likelihood of firms that are financially constrained of employing dividend reinvestment plans (DRIPs). Moreover, since the financial upheaval of 2007-2008 increased the likelihood of a firm’s experiencing financial constraint, an increased number of firms would be more likely to employ DRIPs in the aftermath of these years. Thus, the hypotheses this study considered were twofold:

H1: Financially constrained firms are more likely to employ DRIPs.

H2: Financially constrained firms benefited from employing them during the aftermath of the 2007-2008 financial crisis.

The two hypotheses required differing approaches, which are discussed below. Also presented below is a description of the data used in the process and the means whereby it was obtained.

Hypothesis 1

Since the variable of interest was implementation of a dividend reinvestment plan, a logistic regression model was a logical approach with which to perform this analysis. The dependent variable, a firm’s use of a DRIP, took on a value of 0 if no DRIP was used and 1 if a DRIP was used. Moreover, the author employed previous research in this area as a stepping off point prior to the study’s breaking new ground.

Initial Analysis Based on Previous Research

The work of Boehm and DeGennaro (2007) was taken as an initial basis for this research. These authors attempted to identify firm characteristics which increased the firm’s likelihood of

(21)

21

adopting a dividend reinvestment plan (DRIP). Among the factors selected for inclusion in the model were total assets, payout ratio, dividend yield, total sales, after-tax return on assets (ROA), after-tax return on common equity (ROE), earnings per share (EPS), and total income. (Please see Appendix 1 for definitions of these measures.) The authors also assigned firms to eight sectors of the economy: mining oil production and consumption; materials and food processing; manufacturing/construction; transportation, utilities and waste disposal; wholesale and retail activity; financial services; other miscellaneous services (Boehm & DeGennaro 2007, 9). The author fitted the data to a logistic regression model, which identified the following independent variables significant at least at a p-value of 0.15: dividend yield, ROA, EPS, common shares outstanding, common shares traded, number of employees. Among sectors, the following were found statistically significant: mining, oil, and construction; materials processing; manufacturing; transportation, utilities, and waste management; and wholesale and retail. The authors report a pseudo of 35.49 with a total of 1149 firms included in the sample. Upon running a model incorporating only those independent variables listed above as significant, they obtained a new pseudo of 35.11 (Boehm & DeGennaro 2007, 26).

The author of the current study emulated this work with a sample of firms drawn from the years 2004 through 2010, which included the financial turbulence of 2007-2008. In addition to the financial characteristics found significant by Boehm and DeGennaro, the current study sought to examine the efficacy of three commonly used measures of financial constraint in predicting whether a firm employed a DRIP. The reasoning behind selecting these was that a firm suffering from greater financial constraint would be more likely to turn to a DRIP to help ease its financial situation. The three indices which measure financial constraint are the

(22)

Kaplan-22

Zingales index (KP), the Whited-Wu index (WW), and the Hadlock-Pierce index (HP). These three indices are discussed briefly below.

Study Indices Intended to Measure Financial Constraint

The three indices to be employed in the study are described as follows.

The Kaplan-Zingales Index (KZ-index) measures a firm’s reliance on external financing,

and higher relative values indicate higher levels of financial constraint. It is calculated as follows (YCharts n.d.):

( ⁄ ) ( )

( ) ( ⁄ ⁄ ) ( ⁄ )

where

Cash Flows for time t = income before extraordinaryt + total depreciation and amortizationt

K = PP&Et-1

Q = (Market Capitalizationt + Total Shareholder's Equityt - Book Value of Common Equityt -

Deferred Tax Assetst) / Total Shareholder's Equityt

Debt = Total Long Term Debtt + Notes Payablet + Current Portion of Long Term Debtt

Dividends = Total Cash Dividends Paidt (common and preferred)

Cash = Cash and Short-Term Investmentst

Higher values of the Whited and Wu Index indicate an increased need for external capital. It is calculated as follows (Lykov 2015):

WW = - 0.091CF - 0.062*DIVPOS + 0.021*TLTD - 0.044*LNTA + 0.102*ISG - 0.035*SG where

TLTD = the ratio of the long-term debt to total assets

DIVPOS = an indicator that takes the value of one if the firm pays cash dividends SG = firm sales growth

LNTA = the natural log of total assets

ISG = the firm’s three-digit industry sales growth CASH = the ratio of liquid assets to total assets

(23)

23 CF = the ratio of cash flow to total assets

The Hadlock-Pierce index employs the total assets of a firm in its calculation. Specifically, it is as follows:

HP =-0.737*Total Assets + 0.043*(Total Assets)2

Study Methodology

A twofold analysis was performed in order to explore Hypothesis 1. First, the study firms were classified in each year according to whether they had a DRIP in (1) the period 2004

through 2007 and (2) the period 2008-2010. A comparison of means was then performed to determine whether firms that had a DRIP during the period 2008-2010 performed better than those that did not with respect to the study variables. The second analysis consisted of a

logistical regression to detect patterns between the study variables and a firm’s having a DRIP.

Hypothesis 2

The purpose of Hypothesis 2 was to explore whether financially constrained firms benefited from employing them during the aftermath of the 2007-2008 financial crisis (i.e., in 2008 and 2009). Requirements for this analysis were threefold: a measure of firm performance, an indicator specifying whether the current year was in the aftermath of a crisis, and an

indicator as to whether a firm employed a DRIP during that time. The latter two were binary in nature, and so simple 0-1 variables were employed for them. For the first measure, Tobin’s Q, calculated for each year, was employed. This measure is calculated as follows:

The model employed in the study is as follows:

(24)

24

( ) ( ) ( ) ( )

where

t = the fiscal year Yt = Tobin’s Q

At = the KZ index, included to measure financial constraint

Bt = a 0-1 variable indicating whether Year t was within the aftermath years of the financial

crisis (i.e., 1 if t = 2008 or 2009 and 0 otherwise)

DRIPt = a 0-1 variable indicating whether the firm employed a DRIP program in Year t (i.e., DRIPt

= 1 if it employed a DRIP and 0 otherwise)

Note the expressions involving the 0-1 dummy variables established control for the effects of firms’ employing DRIPs and for inclusion in the financial crisis aftermath years.

Data Used for Statistical Analysis

Altogether, 29,687 firm-year combinations were identified. Data for these firms was obtained from 10-K, 10-Q, and DEF 14A statements downloaded from the EDGAR database. Analysis of these statements, described below, revealed whether a firm employed a DRIP or not.

In order to download the statements, a VBA script created in MS Excel looped through a list of internet file paths pointing to EDGAR’s FTP server. Thus, for each record, VBA launched a VBS script that downloaded a single file. Several downloads occurred in parallel, however, and a user-defined parameter determined the number of VBS scripts running at any one time. Text analysis was handled similarly. Employing an Excel file containing file paths to downloaded reports, another VBS script analyzed file contents with respect to the term “dividend reinvestment.” If this key phrase were found, the script extracted three lines of text

(25)

25

surrounding it and placed them in a report text file. Although many false positives were created, all output reports of hits were read for reclassification.

(26)

26

CHAPTER 4. RESULTS

This chapter describes in more detail the statistical models fit to the study data and presents the results of the resulting analyses.

Hypothesis 1

As stated earlier, Hypothesis 1 sought to explore whether financially constrained firms are more likely to employ DRIPs. Two separate analyses were performed in order to accomplish this goal.

Comparison of Means

For the variables Assets, Market Capitalization, Income, Sales, Dividend Yield (DY), Return on Assets (ROA), Return on Equity (ROE), and Earnings per Share (EPS), firms were classified as to whether they had a DRIP in the years (1) 2004-2007 and (2) 2008-2010. Thus, firms belonged to one of four groups, where the first component of the classification

represents DRIP status with respect to 2004-2007 and the second 2008-2010: No/No (NN), No/Yes (NY), Yes/No (YN), and Yes/Yes (YY). Then means for the study variables were calculated and a t-test was performed on these differences. Table 2 shows the results.

Based on the assumption that firms having DRIPs in the 2008-2010 period would

outperform firms that did not, we would expect the NN – NY and the NN – YY differences to be negative. All are with the exception of ROE for both NN – NY and NN – YY and EPS for NN – YY. These latter differences were not statistically significant however. In contrast to firms that never employed a DRIP during either period (NN), firms that had a DRIP during both periods (YY) performed better with respect to Assets, Market Capitalization, Income, Sales, DY, and

(27)

27

ROA. These differences were statistically significant for the NN – YY differences, but only Market Capitalization and ROA are for NN – NY differences. (Income would, however, have been statistically significant at the .10 value.) Since the distinction between NN – YY and NN – NY differences is that the firms that adopted DRIPs in 2008-2010 already also had employed them in the former case but not in the latter, this comparison would indicate some advantage associated with having DRIPs in the 2004-2008 period and a smaller one in 2008-2010.

Next, we explored possible effects on firms that had DRIPs in the 2004-2007 period but then did not have them in 2008-2010; these are represented by the YN – YY differences. Of these, Assets, Market Capitalization, Income, Sales, DY, and ROA are negative and statistically significant at the .01 level, indicating an advantage for firms having DRIPs in 2008-2010 that firms that had DRIPs in 2004-2007 but chose not retain them in 2008-2010. This relative advantage was not as great as for firms which did not employ DRIPs through the entire period of the study; absolute NN – YY differences are much greater than YN – YY differences, although, again, both are statistically significant.

However, in the YN – NY analyses, i.e., the difference in means of firms that had DRIPs in 2004-2007 but chose not to have them in 2008-2010 and those firms that did not employ them in 2004-2007 but chose to have them in 2008-2010, differences except ROE and EPS were positive, indicating no relative advantage for firms that adopted DRIPs post crisis compared to those that had in the past employed DRIPs but dropped them post crisis. Of these, Assets, Market Capitalization, Sales, Dividend Yield, and ROA are statistically significant.

Also positive and statistically significant were differences in means between firms that employed DRIPs throughout the entire study period (i.e., 2004-2010) and those that adopted

(28)

28

them only in the post-crisis period of 2008-2010, indicating that firms that had DRIPs

throughout the entire seven years performed better with respect to the study variables than did firms that instituted them during the post-crisis period. The means of those firms were also statistically significant.

So far, the analysis has looked at differences in the post-crisis period, i.e., the second in the two-letter firm classification, with respect to DRIP status. Examination of the first letter, representing the pre-crisis period and 2007, is also revealing. Comparison of NN – YY and YN – YY differences reveals that, while both are negative, the NN – YY differences, without

exception, have greater absolute values than the YN – YY differences, indicating an apparent relationship between having a DRIP prior to the post-crisis period.

Comparison of NN – NY and YN – NY also reveals interesting differences. Except for ROE, the first differences are all negative, although not with as high absolute values as in the two differences just discussed. The latter differences, on the other hand, are positive with the exception of ROA and EPS, indicating that firms that had DRIPs prior to 2008 performed better in terms of Assets, Income, Sales, and Dividend Yield than firms that did not have them but added them in the post-crisis period. However, compared to NY firms, those that had DRIPs throughout the study period performed best of all; the differences for YY – NY firms yielded the highest positives in the table.

Comparing all three firms that did not have DRIPs in 2004-2007 but added them in 2008-2010 indicated best performance with respect to the study variables as follows: firms that had them in both periods (YY), firms that had them in 2004-2007, and firms that had never had

(29)

29

them. Thus, again, having a DRIP in the pre-crisis period appears to be associated with an advantage.

Although this analysis has looked at these measures en masse to this point, examining ROA and ROE in isolation could prove fruitful, simply because these are percentages and so are comparable across firms of different sizes. Also, for all but one of the differences (NN – YN), ROA differences were statistically significant. ROA was positive for YY – NY and NN – YN. The remaining differences were negative with ROA values decreasing in the following order: NN – NY, YN – NY, YN – YY, and NN – YY. These again lend credence that firms that had DRIPs in 2004-2008 derived the most benefits associated with DRIPs (YN – YY), that firms that added them in 2008 when not having had them previously derived some benefits from doing so (NN – NY), and that firms that had them throughout the study period fared best of all (YY – NY, YN – YY).

To summarize, almost exclusively, the largest absolute differences were between firms that did not have DRIPs and those that had them during the entire seven-year period (YY – NN). The next largest absolute differences (YY – NY) were between means of firms that had DRIPs for the entire study period and those that added them in the 2008-2010 period. This would

indicate that adding a DRIP during this period provided only a slight advantage. The third largest absolute differences (YN – YY) were between firms that had DRIPs during the entire study period and those that had them only during the 2004-2007 period, again pointing to an advantage for having a DRIP during the post-crisis period. The fourth largest absolute

differences (NN – YN) pointed to the advantages apparently associated with having DRIPs during the first period. The fifth absolute differences (YN – NY) would combine the effects of having DRIPs during the first period and not during the second, and vice versa; that these

(30)

30

differences were positive again points to the greater relative advantage having DRIPs during the first period appeared to have over having them during the second period. Last were the NN – NY differences, of which only ROA and Market Capitalization were statistically significant, pointed to almost no real advantage for firms that added DRIPs during the second period over firms that never had them. Thus, employing a DRIP during this second period seems to have had the most beneficial effect for firms that had them during the first period.

Logistical Regression Model with DRIP as the Dependent Variable

As presented in the preceding chapter, the study’s first hypothesis was that financially constrained firms were more likely to employ DRIPS. As an initial step, a modified version of the research performed by Boehm and DeGennaro (2007) was taken in order to address Hypothesis 1. Like these authors, this research employed a logistical regression where the dependent variable was a binary categorical variable indicating the presence or absence of a dividend reinvestment program for a firm. Retracing the work of Boehm and DeGennaro was only an initial step, however. Whereas these authors incorporated several fundamental

financial measures as independent variables in their model, this research sought to examine the efficacy of three indices that measure a firm’s financial constraints on likelihood of a firm

employing a DRIP. These indices were the Kaplan-Zingales index (KZ), the Whited-Wu index (WW), and the Hadlock-Pierce index (HP), and the previous chapter described how they were calculated.

Model

(31)

31 ( ) ( ) ( ) ∑ ∑

where the independent variables are largely self-explanatory. The coefficients are marginal effects at the mean to facilitate interpretation of the variables’ effects on the likelihood of a firm’s having a DRIP.

This model incorporated year t as an independent variable with years, designated by t, taking on the values 2004, 2005, 2006, 2007, 2008, 2009, and 2010. The years 2008 through 2010 were considered to be those most affected by the financial turbulence of the 2007-2008 financial upheaval, and so these years were employed in classifying firms as “financially

constrained.” Following Boehm and DeGennaro, industry was comprised of the following: mining oil production and consumption; materials and food processing; manufacturing/construction; transportation, utilities and waste disposal; wholesale and retail activity; financial services; other miscellaneous services. The model was fitted with the three measures of financial constraint. Table 1 presents statistical characteristics of the data.

Data Characteristics

As an initial step, correlations between these variables were measured in order to eliminate, or at least minimize, multicollinearity through selection of independent variables. Note that highly skewed variables, Sales, Income, and Assets, were transformed by taking the log of them into log(Sales), log(Income), and log(Assets). The highest positive correlations were between log(Sales) and log(Assets) (0.5431) and between Dividend Yield and Log(Sales)

(32)

32

log(Assets) was -0.4278 and between it and log(Sales) -0.4118. The fact that assets comprise the independent variable in the computation of this index explains this correlation. None of the other variables tested included a correlation exceeding either -0.12 or +0.12, and all but one were exceeded plus or minus 0.10.

These correlations dictated this study’s selection of variables to be included in one version of the model, whose results are shown in Table 3. Three models were fitted to the data, one for each of the financial constraint indices. Models 1, 4, and 7 included only the financial constraint variable. As can be seen, only the coefficients of the HP Index were statistically significant and at the 0.01 level for not only Model 7 but Models 8 and 9 also. The coefficient is negative and uniform throughout the three models, indicating that the higher the financial constraint experienced by the firm, as measured by the HP Index, the less likely the firm is to have a DRIP.

Discussion

Models 3, 6, and 9 incorporated all of the initial independent variables. In Model 3, the only statistically significant coefficients were log(Assets) and log(Sales). Log(Assets) influence on likelihood of having a DRIP is negative according to the model; its coefficient is negative.

Log(Sales), on the other hand, had a larger and a positive coefficient, indicating that greater sales translate to increased likelihood of a firm employing a DRIP. In Model 3, none of the other variables were statistically significant.

Model 6 also incorporated all of the study variables in addition to the WW Index. Interestingly, this model’s results with respect to statistically significant coefficients were almost identical to Model 3’s. Log(Sales) had a relatively larger and positive coefficient, and

(33)

33

log(Assets) a negative and relatively smaller one. The coefficients themselves were almost identical to those shown in Model 3.

Model 9’s coefficients behaved in a different manner from those in Models 3 and 6, demonstrating the influence including the HP Index had. Not only were log(Assets) and log(Sales), apparently two variables which influence a firm’s employing a DRIP, highly statistically significant, but Fiscal Year, Dividend Yield and ROA were also. Log(Assets) and log(Sales) demonstrated the same pattern as shown in the two previous models discussed—a relatively small and negative log(Assets) and a larger and positive log(Sales). Fiscal Year exerted a small, negative influence on likelihood of a firm’s having a DRIP. This data’s Fiscal Year ranged from 2003 through 2010, and the years exhibiting financial constraint were in the latter part of this range. Therefore, the model indicates that greater financial constraint would decrease likelihood of a firm’s having a DRIP. Dividend Yield, small and significant at only the 0.05 level, exerts a small but positive influence on likelihood of a DRIP, and ROA a somewhat larger and negative influence.

Since these models were constructed based on the eliminating correlation between independent variables, Models 2, 5, and 8 would be expect to provide the most accurate representation of their relationship to likelihood of a firm’s having a DRIP. Model, which incorporated constraint measure KZ Index, excluded log(Assets), whose coefficient was highly significant in Model 3. Log(Sales), which was significant in Model 3, was also significant here. Since log(Assets) and log(Sales) were highly and positively correlated there, log(Sales) would be expected to exert a greater influence with log(Assets) excluded, and this proved to be the case, with this coefficient larger than Model 3’s corresponding one.

(34)

34

Model 5 also excluded log(Assets), but the effect was to significant lower log(Sales) rather than strengthen it, although this coefficient remained highly significant. Moreover, with log(Assets), log(Sales) was the only other statistically significant variable in Model 3.

Models 7, 8, and 9 were the only ones in which a measure of financial constraint, HP Index, proved statistically significant. Moreover, the HP variant with the full complement of initial variables exhibited several other variables with a strong influence on likelihood of a DRIP. Therefore, both log(Assets) and log(Sales) were excluded from Model 8. The effect of excluding both variables was to bring other variables to the foreground. For the first time, Industry became significant with a relatively strong negative coefficient, suggesting that differences in industry could influence likelihood of a DRIP. As in Model 9, Dividend Yield and ROA were significant with the former positive and the latter negative. Excluding both log(Assets) and log(Sales) served to strengthen the effect of these variables. The Dividend Yield coefficient, in particular, increased from 0.005 to 0.013. The ROA coefficient’s change, from -0.16 to -0.182. By far the most interesting seeming results were obtained from the models containing the HP Index. However, this index is based solely on asset quantity. The HP equation is a quadratic polynomial with point of inflection at around -8.57, and its independent variable, asset quantity, will not typically be negative. Asset quantity for the data included in this study ranged from 0.04 to 85,963 with a median of 140.67. Thus, over the range of asset quantity present in the study data, HP would increase at an increasing rate of change ( ) ⁄ . Although taking the log of asset quantity would mute its impact

somewhat, the effect of asset quantity, which has proved statistically significant in every model discussed so far, would nonetheless be magnified in the HP Index. This could account for HP

(35)

35

Index’s significance as a measure of financial constraint in the study models when the other two measures, HZ and WW, were not significant. Thus, asset quantity appears to have a strong negative impact on likelihood of a firm to employ a DRIP. Because of asset quantity so greatly dominates this model, that ROA is significant is also not surprising, and Dividend Yield is also indirectly related to asset quantity.

With respect to the model Wald Chi Square, of the models incorporating only the financial constraint (i.e., KZ, HP, and WW), only that with the HP Index as its single independent variable was statistically significant, due to the fact that it is so strongly representative of the asset quantity. All other models, even those in which log(Asset) had been excluded, were statistically significant at the 0.01 level of significance.

Hypothesis 2

As will be recalled, Hypothesis 2 sought to explore whether firms having DRIPs during the aftermath of the 2007-2008 financial crisis benefited from having them in its immediate aftermath 2008 and 2009. Rather than employing all three measures of financial constraint, only the KZ Index was chosen for further analysis. The following model was fit and run:

( ) ( ) ( ) ( ) where

t = the fiscal year Yt = Tobin’s Q

KZt = the KZ index for year t, included to measure financial constraint

Crisist = a 0-1 variable indicating whether Year t was within the aftermath years of the financial

crisis (i.e., 1 if t = 2008, 2009, or 2010 and 0 otherwise)

DRIPt = a 0-1 variable indicating whether the firm employed a DRIP program in Year t (i.e., DRIPt

= 1 if it employed a DRIP and 0 otherwise in year t)

(36)

36

Graphical Look

The model was first run with no controls and with industry included as a fixed effect. None of the levels within this factor was statistically significant however, and so the model was rerun with just the KZ Index as an independent variable. The resulting model, whose

coefficients are shown in Figure 5, had a statistically significant F statistic and KZ coefficient, although the latter was extremely small and positive. Within this model, also significant were the coefficients for Crisis (a dummy variable equaling 1 if year was in 2004-2008 and 0

otherwise), the interaction term for KZ and DRIP, and the interaction term for KZ, Crisis, and DRIP.

First, 100 KZ values between -98000 and 2000 were generated and input in the model. Figures 1 and 2 below show the results.

-200 0 200 400 600 800 1000 1200 -120000 -100000 -80000 -60000 -40000 -20000 0 20000 Predicted T o b in 's Q KZ-Index

Regression Model with Interaction

Predicted Tobin's Q as a Function of KZ Index During Pre-Crisis Period

(37)

37

Figure 1. Regression Model with Interaction for Years 2004-2008 (Pre-Crisis)

Figure 2. Regression Model with Interaction for Years 2008-2010 (Crisis)

Figure 1 depicts the relationship between KZ and Tobin’s Q during the non-crisis years 2004-2008. The orange line is Tobin’s Q for firms that had DRIP during this period and the blue for one that did not. Figure 2 employs the same colors for DRIP status. Both appear to support the hypothesis that a KZ index is a good predictor of financial well-being.

Results of ANOVA

Next an ANOVA was constructed, adding each variable and then the interactions. For main effects, the model found the KZ Index, DRIP, and Crisis period to be significant; however, since higher orders supersede main effects, this information is inconclusive. Among the second order effects, only KZ x DRIP approach statistical significance; with a p-value of .0578, this is at a

-160 -140 -120 -100 -80 -60 -40 -20 0 20 -120000 -100000 -80000 -60000 -40000 -20000 0 20000 Predicted T o b in 's Q KZ-Index

Regression Model with Interaction

Predicted Tobin's Q as a Function of KZ Index During a Crisis

(38)

38

.10 level of significance. The third order interaction exhibits a p-value of .0001, and so was highly statistically significant.

That DRIP and time (i.e., in Crisis period or not) are not related do not display a significant interaction is borne out by Figure 3, which displays the group means of Crisis=No versus DRIP=No (blue) and DRIP=Yes (orange). The two lines to not cross and are, in fact, almost parallel, indicating no significant interaction.

Figure 3. Group Means (NN, NY, YN, and YY) with respect to Crisis/DRIP Grouping When Crisis=NO, the means of Tobin’s Q for DRIP=N and DRIP=Y were 4.48 and 4.66,

respectively. When Crisis=Yes, the Tobin’s Q means for DRIP=N and DRIP=Y were 3.43 and 3.49, respectively.

The three-way interaction term, KZ-Index x DRIP x Crisis, was statistically significant with a p-value = 0.001. The KZ x Crisis interaction was not statistically significant, but the KZ x DRIP

0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 4,00 4,50 5,00 0 0,5 1 1,5 2 2,5 To b in 's Q

Crisis Period (N=No, Y=Yes)

Group Means by Crisis and DRIP Factors

(39)

39

interaction was, with a p-value of 0.0578. Therefore, the significance of the three-way interaction term appears to be based on the KZ x DRIP interaction.

Conclusion

In this chapter, we have performed statistical tests to explore the two hypothesis governing this thesis:

H1: Financially constrained firms are more likely to employ DRIPs.

H2: Financially constrained firms benefited from employing DRIPs during the aftermath of the 2007-2008 financial crisis.

To explore the first, after first determining correlations between study variables, we analyzed and compared differences of means for the various study variables with respect to whether a firm was employing a DRIP in a year in the pre-crisis period (2004-2007) and in the crisis period (2008-2010) or not. Most variables had statistically significant differences, and comparison indicated a possible effect of having a DRIP during the pre-crisis period on the study variables. Of particular note was ROA, which, as a percentage, was comparable across firms of differing sizes. It was statistically significant across all differences except NN-YN, although this value was positive. The greatest differences occurred between firms that had DRIPs during both period; the least difference was for NN – NY firms, those that had DRIPs only in the crisis

period. Firms seemed to benefit more from having DRIPs in the crisis period when they had employed them in the pre-crisis period also.

Next, we fitted a logistical regression model to subsets and the entire set of study variables to determine which increased the likelihood of a firm employing a DRIP. Differing measures of financial constraint, the KZ, WW, and HP Indices were included, but none tested as

(40)

40

statistically significant except for the HP index, due to its close relationship with a firm’s total assets. This held true for KZ and WW even when they were the sole independent variables in the model. A firm’s log(Assets) and log(Sales) tested as statistically significant here, and, among industries, Finance and Wholesale/Retail.

Thus, with respect to Hypothesis 1, evidence was inconclusive. Even though the

coefficients of the measures of financial constraint were not statistically significant, the models themselves were.

The analyses undertaken to explore Hypothesis 2 provided more evidence that DRIPs aided firms during the 2008-2010 period. Of the higher order interaction terms in the mixed effects regression we fitted to the data, the KZ Index x DRIP and the KZ Index x DRIP x Crisis were statistically significant.

(41)

41

CHAPTER 5. CONCLUSION

The study sought to test whether firms having higher levels of financial constraint would be more likely to maintain DRIPs. Moreover, it extended the work of Boehm and DeGennaro (2007) by incorporating three indices intended to represent level of financial constraint. The three indices were the Kaplan-Zingales index, the Whited-Wu index, and the Hadlock-Pierce index. Logistic regression models were fit to the study data using a subset of the Boehm and DeGennaro independent variables and then with the three indices. Overall, all approaches did not perform as well as those reported in Boehm and DeGennaro (2007), although Chi Square values indicated that they were superior to random selection. Thus, these study results were somewhat inconclusive.

The influence of asset quantity was extremely pronounced, both by itself and as it worked through the HP Index and the ROA. Moreover, its coefficients tended to have the largest absolute magnitude.

Contribution

This research attempted a novel approach to predicting which firms would be most likely to adopt DRIPs. Based on the assumption that firms experiencing constrained financing would turn to dividend reinvestment to conserve cash, it employed the KZ, WW, and HP indices to see if the levels of these contributed to the likelihood that a firm would adopt a DRIP. The statistical models run did lend credence to use of these in deriving greater benefit from adopting a DRIP during financially turbulent times.

(42)

42

Further Study

In the HP Index models (Models 7, 8, and 9 of Table 3), Industry for the first time was significant. This could merit further study. It could be due to significant differences in DRIP across industry or it could simply reflect the large influence asset quantity appears to have on the presence of DRIPs. Some firms (e.g., manufacturing, financial) would tend to have a greater total asset value than do others.

Limitations

To determine whether a firm employed a DRIP, 10-Ks, 10-Qs, and DEF 14As were scanned, and if mention of a DRIP was found, the firm was considered to have a DRIP. This means of gathering data may have introduced some error into the study data. First, the timing of the DRIP’s implementation might not correspond exactly to the year of the document which mentioned it. Moreover, firms with DRIPs might have not mentioned them in any of these documents or firms might refer to DRIPs by some other name. In the latter case, the search method employing the term “dividend reinvestment” would overlook that firm as having a DRIP. In either case, false negatives would result.

Another limitation was the ability of the three indices to reflect financial constraint. For instance, the HP-index emphasizes a firm’s size (as measured by assets), whereas the KZ-index incorporates market capitalization, debt, PP&E, and dividends. The KZ Index would seem more suitable for the purposes of this study.

Also most likely hampering the robustness of the model are multicollinearity between variables and endogeneity. As noted earlier, sales and assets were highly correlated, as were dividend yield and sales. Because of its formulation, the HP Index was highly correlated with

(43)

43

assets and so indirectly with sales. In addition, the possibility of hidden variables underlying these and influencing both, but not included in the models, was very real. The consequences of multicollinearity would be higher standard errors than otherwise and less accurate marginal values. Also a result of confounding variables could be endogeneity, where a variable is

correlated with the model’s error term and always introduces bias. The size of the sample data used in this analysis could, to a certain extent, compensate for any endogeneity in the model but most likely could not eliminate it entirely.

Still another possible distortion could show in the t-value analyses of Table 2. In a test of difference of means, the assumption is that the variances of the underlying distributions are equal for each of the study variables. Compounding this was the fact that sample sizes for the populations were unequal, sometimes very unequal. The effect would be an increase in Type I error.

Conclusion

In spite of its limitations, the study extended the work of Boehm and DeGennaro in dividend reinvestment plans and succeeded in showing that at least one of the financial indices presented herein have value in predicting a firm’s use of a DRIP.

(44)

44

REFERENCES

Asquith, P. & Mullings, D.W., 1986. Equity Issues and Offering Dilution. Journal of Financial Economics, 15, pp.61–90.

Bajaj, M. & Vijh, A., 1990. Dividend clienteles and the information content of dividend changes. Journal of Financial Economics, 26, pp.193–219.

Barnea, A., Haugen, R. & Senbet, L., 1980. A rationale for debt maturing structure and call provisions in the agency theoretic framework. Journal of Finance, 35, pp.1223–1234.

Boehm, T.P. & DeGennaro, R.P., 2007. A Discrete Choice Model of Dividend Reinvestment Plans: Classification and Prediction,

Booth, J. & Smith, R., 1986. Capital raising, underwriting and the certification hypothesis. Journal of Financial Economics, 15, pp.261–281.

Cherin, A.C. & Hanson, R.C., 1995. Dividend reinvestment plans: A review of the literature. Financial Markets, Institutiions and Investments, 4, pp.59–73.

Dammon, R.M. & Spatt, C.S., 1992. An option-theoretic approach to the valuation of dividend reinvestment and voluntary purchase plans. The Journal of Finance, 47, pp.331–347.

Davey, P.J., 1976. Dividend Reinvestment Programs,

Dhillon, U.S., Lasser, D.J. & Ramirez, G.G., 1992. Dividend reinvestment plans: An empirical analysis. Review Quantitative Finance Accounting, 2, pp.205–213.

Dittmar, A.K., 2000. Why Do Firms Repurchase Stock? The Journal of Business, 73, pp.331–355. Dubofsky, D.A. & Bierman, L., 1988. The effect of discount dividend reinvestment plan

announcements on equity value. Akron Business and Economic Review, 19, pp.58–68.

Finnerty, J.D., 1989. New issue dividend reinvestment plans and the cost of capital. Journal of Business Research, 18, pp.127–39.

Hansen, R.S., Pinkerton, J.M. & Keown, A.J., 1985. On dividend reinvestment plans: the adoption decision and stockholder wealth effects. Review of Business and Economic Research, 20, pp.1–10. Healy, P. & Palepu, K., 1988. Earning information conveyed by dividend initiations and omissions. Journal of Financial Economics, 21, pp.149–175.

(45)

45

Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance and takeovers. American Economic Review, 76, pp.323–329.

Lykov, K., 2015. How to interpret Whited Wu Index (WW-index). Economics beta. Available at: http://economics.stackexchange.com/questions/1877/how-to-interpret-whited-wu-index-ww-index [Accessed May 7, 2015].

Noe, T., 1988. Capital structure and signaling game equilibria. Review of Financial Studies, 1, pp.331–356.

Ogden, J.P., 1994. A dividend payment effect in stock returns. Financial Review, 29, pp.345–369. Perumpral, S., Keown, A.J. & Pinkerton, J.M., 1991. Market reaction to the formulation of

automatic dividend reinvestment plans. Review of Business and Economic Research, 26(2), pp.48– 58.

Perumpral, S.E., 1983. No Title. Virginia Polytechnic Institute.

Peterson, P., Peterson, R. & Moore, N., 1987. The adoption of new-issue dividend reinvestment plans and shareholder wealth. Financial Review, 22(221-32).

Reilly, R.R. & Nantell, T.J., 1979. Dividend reinvestment. Journal of Midwest Finance Association, 3, pp.101–13.

Roden, F. & Stripling, T., 1996. Dividend reinvestment plans as efficient methods of raising equity financing. Review of Financial Economics, 5, pp.91–100.

Scholes, M.S. & Wolfson, M.A., 1989. Decentralized investment banking: the case of discount dividend-reinvestment and stock purchase plans. Journal of Financial Eonomics, 24, pp.7–35. Tinic, S., 1988. Anatomy of Initial Public Offerings of Common Stock. The Journal of Finance, 43, pp.789–821.

YCharts, KZ Index. YCharts. Available at: http://ycharts.com/glossary/terms/kz_index [Accessed May 7, 2015].

(46)

46

Tables

Table 1 Sample Statistics

Variable Obs Mean Std. Dev. Min Max Median

Total Assets 25616 867.49 3208.72 0.04 85963.00 140.67 Total Sales 29310 2541.15 11655.79 -1671.00 425071.00 272.26 Total Income 29309 152.80 1129.63 -29580.00 45220.00 7.68 Market Capitalization 29310 3267.24 15450.75 0.35 504239.60 354.66 Dividend Yield 29309 1.78 13.68 -0.01 780.80 0.00 Return on Assets 29309 -0.06 0.79 -100.01 6.45 0.02 Return on Equity 29309 -0.05 19.28 -819.79 2911.01 0.08

(47)

47

Table 2

T-Tests, Firm Status with Respect to DRIPs during the 2003-2007 Period and the 2008-2010 Period

NN - YN NN - YY NN - NY YN - YY YN - NY YY - NY

Assets -8.6*** -19.16*** -0.86 -10.64*** 7.31*** 17.71***

Market Capitalization -4.2*** -13.15*** -2.22** -9.38*** 2.03** 10.93***

Income -3.47*** -10.76*** -1.8 -7.17*** 1.8 9.2***

Sales -7.1*** -16.45*** -1.62 -9.23*** 5.17*** 14.15***

Dividend Yield (DY) -6.11*** -14.37*** -1.56 -7.53*** 4.39*** 12.36***

Return on Assets

(ROA) 0.64 -4.33*** -2.49** -3.57*** -2.51** 4.23***

Return on Equity

(ROE) 0.02 0.34 0.14 0.22 0.07 -0.6

Earnings per Share

(EPS) 0.83 0.66 -0.29 0.21 -1.01 -0.77 9609, 25219 3027, 25219 16007, 25219 3027, 11305 9609, 18004 3027, 18004

** - Significant at the .05 level *** - Significant at the .01 level NN – Firms that did not have DRIPs in 2004-2007 and did not have DRIPs in 2008-2010

NY – Firms that did not have DRIPs in 2004-2007 and did have them in 2008-2010 YN - Firms that had DRIPs in 2004-2007 and did not have them in 2008-2010 YY – Firms that had DRIPs in 2004-2007 and in 2008-2010

(48)

48

Table 3

Logit Results, 2003-2010 Data

( ) ( ) ( )

∑

Dependant variable Firm has a dividend reinvestment plan

Model Logit

Constraint measure KZ WW HP

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Kaplan Zingales Index 0.000 0.000 0.000

(0.66) (0.41) (1.08)

Whited Wu Index -0.001 0.000 0.000

(-1.18) (-1.22) (-1.11)

Hadlock Pierce Index -0.158 -0.160 -0.148

(-15.48)*** (-15.75)*** (-8.80)*** logAssets 0.053 0.053 -0.038 (7.01)*** (7.04)*** (-3.00)*** logIncome -0.090 -0.059 -0.137 -0.059 0.388 -0.015 (-0.40) (-0.38) (-0.32) (-0.38) (1.22) (-0.40) logSales 0.142 0.056 0.159 0.056 0.139 (10.53)*** (3.10)*** (10.66)*** (3.10)*** (6.91)*** Dividend Yield 0.000 0.000 0.000 0.000 0.001 0.000 (0.44) (0.54) (0.16) (0.53) (2.25) (2.10)** ROA 0.008 -0.002 0.024 -0.003 -0.009 -0.007 (0.53) (-1.27) (1.20) (-1.34) (-0.97) (-1.54) ROE 0.000 0.000 0.000 0.000 0.000 0.000 (-0.78) (-0.61) (-0.66) (-0.62) (-0.81) (-0.66) EPS 0.000 0.000 0.000 0.000 0.000 0.000 (-3.15)*** (0.50) (-3.34)*** (0.50) (-2.10) (0.34)

(49)

49

Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Industry Finance 0.505 0.498 0.50476005 0.425 0.418 (3.46) (3.29)*** (3.46)*** (2.68)*** (2.63) Manufacturing 0.206 0.194 0.2058722 0.195 0.188 (2.19) (2.07)** (2.18)** (2.13)** (2.08) Mining 0.329 0.309 0.32807156 0.279 0.267 (2.06) (1.91)* (2.06)** (1.75)* (1.70) Other Svc. 0.175 0.180 0.17476094 0.172 0.168 (1.42) (1.43) (1.43) (1.40) (1.39) Transportation 0.290 0.238 0.28961292 0.228 0.212 (1.82) (1.54) (1.82)* (1.50) (1.43) Wholesale/Retail 0.501 0.437 0.50044041 0.432 0.422 (3.34) (2.74)*** (3.35)*** (2.65)*** (2.58) Years 2004 0.027 0.024 0.008 0.026739 0.032 0.008 0.017 0.016 0.007 (4.84) (4.45)*** (1.78)* (4.84) (5.43)*** (1.78)* (3.44)*** (3.34) (1.68)* 2005 0.009 0.005 -0.008 0.00930714 0.012 -0.008 -0.001 -0.002 -0.008 (1.61) (0.85) (-1.70)* (1.61) (2.00)** (-1.71)* (-0.20) (-0.40) (-1.82)* 2006 0.016 0.009 -0.007 0.01636354 0.017 -0.007 0.002 0.000 -0.009 (2.74) (1.57) (-1.55) (2.74)*** (2.75)*** (-1.55) (0.38) (0.08) (-1.93)* 2007 -0.017 -0.025 -0.035 -0.01769152 -0.019 -0.035 -0.028 -0.030 -0.035 (-2.99) (-4.58)*** (-7.45)*** (-3.03)*** (-3.05)*** (-7.48)*** (-5.95)*** (-6.45) (-8.07)*** 2008 0.009 -0.001 -0.016 0.00932942 0.009 -0.016 -0.007 -0.010 -0.020 (1.44) (-0.12) (-3.12)*** (1.44) (1.25) (-3.13)*** (-1.23) (-1.92) (-4.24)*** 2009 0.021 0.012 -0.005 0.02126801 0.023 -0.005 0.001 -0.001 -0.007 (3.044) (1.84) (-0.84) (3.04)*** (3.17)*** (-0.85) (0.20) (-0.19) (-1.29) 2010 0.027 0.014 -0.007 0.02725041 0.027 -0.007 0.002 -0.001 -0.009 (3.62) (2.03)** (-1.09) (3.61)*** (3.40)*** (-1.11) (0.38) (-0.14) (-1.56) Observations 15,297 15,297 15,297 15,297 15,297 15,297 15,297 15,297 15,297 Chi Square 770.9*** 1351.651*** 863.201*** 1048.67*** 870.98*** 860.677*** 1383.039*** 1610.048*** 1086.313***

(50)

50

Table 4

Correlations, 2004-2010 Data

Industry Year logAsset logIncome logSales

Dividend

Yield ROA ROE EPS

KZ-Index WW HP Industry 1 Year -0.0353 1 logAssets 0.3075 0.0255 1 logIncome -0.0077 0.0082 0.099 1 logSales 0.1276 0.0682 0.5431 0.2686 1 Dividend Yield 0.0095 0.0386 0.1171 0.2592 0.3869 1 ROA 0.0329 -0.0146 0.1078 0.0277 0.0852 0.0182 1 ROE 0.0094 0.0063 0.0097 0.008 0.0133 0.007 0.0072 1 EPS 0.0012 -0.003 -0.023 0.0797 0.087 -0.0006 0.0064 0.0007 1 KZ-Index 0.0054 0.0001 0.0292 0.002 0.0187 -0.002 -0.0194 -0.0002 0.0005 1 WW 0.0004 -0.0097 -0.0159 -0.0066 -0.0282 -0.0103 -0.0363 -0.0078 -0.0018 -0.0018 1 HP -0.0504 -0.1149 -0.4278 -0.0304 -0.4118 -0.0342 -0.2735 -0.0144 -0.0017 -0.0568 0.0315 1

Dividend reinvestment plans : do the companies using DRIPs benefit from greater funding flexibility during market distress?