Nasdaq ex‐day behavior: An out‐of‐sample test

• Date: 01 April 2020
• DOI: http://doi.org/10.1002/rfe.1083
• Author: Mark Wu,Sabatino (Dino) Silveri,Shishir Paudel
• Published date: 01 April 2020

Rev Financ Econ. 2020;38:405–420. wileyonlinelibrary.com/journal/rfe

|

405

© 2019 University of New Orleans

1

|

INTRODUCTION

We test various explanations of the ex‐day stock price behavior using Nasdaq‐listed firms. For over half a century, academics

have noted that, on average, stock prices fall by less than the dividend amount on the ex‐day.1

Much research has been under-

taken to explain this stylized fact as it has important implications for both corporate finance and asset pricing. The predominant

and most enduring explanation of this ex‐day behavior is taxes. In their seminal paper on the issue, Elton and Gruber (1970)

point out a dollar of capital gains is worth more than a dollar of dividends whenever capital gains are taxed more favorably than

dividends. As this has often been the case in the U.S., the equilibrium price drop will be less than the dividend. Measuring the

price‐drop ratio (hereafter PDR) as the ratio of the change in price to the dividend amount, Elton and Gruber (1970) document

that the PDR is on average around 80%, i.e., prices fall on average by 80% of the dividend amount on the ex‐day. As the im-

puted marginal tax rate from the PDR is consistent with the highest marginal tax rate at the time, Elton and Gruber interpret

the results as consistent with their tax hypothesis. Many papers have since found support for this tax explanation (a partial list

includes Barclay, 1987; Elton, Gruber, & Blake, 2005; Graham & Kumar, 2006; Graham, Michaely, & Roberts, 2003; Naranjo,

Nimalendran, & Ryngaert, 2000; Whitworth & Rao, 2010).

The literature investigating the U.S. evidence almost exclusively focuses on NYSE‐listed firms.2

This is understandable

for papers written before Nasdaq came into being or before Nasdaq data were readily accessible but it is a little surprising for

papers written since then. Perhaps it reflects a view that the technology‐laden exchange does not have many dividend‐paying

firms. As surprising as it may appear to some, Nasdaq has a non‐negligible number of dividend payers. After imposing various

restrictions, we have a sample of 107,200 dividend observations for the 1973–2015 period, compared to a sample of 160,477

NYSE observations with corresponding data filters. Moreover, an underlying assumption of tax‐based explanations is that retail

investor characteristics such as marginal tax rates are impounded into ex‐day stock prices. To the extent that retail investors have

Received: 29 April 2019

|

Revised: 20 August 2019

|

Accepted: 9 September 2019

DOI: 10.1002/rfe.1083

ORIGINAL ARTICLE

Nasdaq ex‐day behavior: An out‐of‐sample test

ShishirPaudel1

|

Sabatino (Dino)Silveri2

|

MarkWu3

1College of Business Administration,

University of Wisconsin‐La Crosse,

La Crosse, WI, USA

2Fogelman College of Business and

Economics,University of Memphis,

Memphis, TN, USA

3Mario J. Gabelli School of Business,Roger

Williams University, Bristol, RI, USA

Correspondence

Shishir Paudel, College of Business

Administration, University of Wisconsin‐La

Crosse; 1725 State Street, La Crosse,

WI 54601, USA.

Email: spaudel@uwlax.edu

Funding information

Roger Williams University

Abstract

We test various explanations of the ex‐dividend day price anomaly using Nasdaq‐

listed firms. Similar to NYSE‐listed firms, on average the prices of Nasdaq‐listed

firms drop by less than the dividend amount. However, the average Nasdaq price‐

drop is substantially smaller than what existing theories would predict and translates

to an imputed dividend tax rate that is double the maximum tax rate. We thus find

little support for the tax hypothesis. We also find little support for the short‐term

trading hypothesis and various other explanations. The significant disconnect we

document between Nasdaq dividends and price changes seems to support the “free

dividends fallacy.”

KEYWORDS

asset pricing, dividends, Nasdaq, taxes, transaction costs

JEL CLASSIFICATION

G10; G12; G35; H20; H24

406

|

PAUDEL Et AL.

a greater ownership concentration in Nasdaq firms relative to NYSE firms, the tax‐based explanations may be more relevant

in the Nasdaq setting.

We find that the average Nasdaq PDR is only 39% (the median is 27%), and it is substantially below a lower theoretical

bound implied by the tax hypothesis. Moreover, the PDR translates to an imputed marginal dividend tax rate of 71%, almost

double the average top marginal dividend tax rate over the sample period. Both the dividend tax rate and the capital gains tax

rate change a number of times during our sample period. To deal with this, we partition our sample into various tax regimes.

We find that within each partition, the Nasdaq PDR is significantly below the lower theoretical bound implied by the tax hy-

pothesis. The average PDRs translate to imputed dividend tax rates up to two and half times the actual maximum rate. Again,

the evidence is inconsistent with the tax hypothesis.

An implication of the tax hypothesis is the formation of tax clienteles, i.e., investors facing relatively low marginal tax rates

will favor high dividend‐yielding stocks and investors facing relatively high marginal tax rates will favor low dividend‐yielding

stocks. If tax clienteles exist then PDRs will increase in dividend yield, as Elton and Gruber (1970), Barclay (1987), Koski

(1996), Graham et al. (2003) and Whitworth and Rao (2010), amongst others, find for NYSE firms. However, we find that the

Nasdaq PDR significantly decreases in dividend yield, opposite the prediction from tax clientele models.3

A reason for our findings may be transaction costs as Nasdaq firms potentially have larger transaction costs relative to NYSE

firms.4

We note, however, that significant transaction costs make arbitrage more difficult and it is in such cases that PDRs are

most likely to reflect the marginal tax rate of retail investors (see, e.g., Boyd & Jagannathan, 1994; Elton, Gruber, & Rentzler,

1984; Karpoff & Walkling, 1990). Nonetheless, we partition our sample into quartiles based on various proxies for transaction

costs including liquidity, bid‐ask spread, stock price, volatility, and firm size. For each transaction cost proxy and for each quar-

tile, the average PDRs remain significantly below the lower theoretical bound implied by the tax hypothesis.

As an alternative test of the tax hypothesis, we follow Eades, Hess, and Kim (1984) and explore the ex‐day price behavior

around non‐taxable distributions (i.e. stock dividends and stock splits). If taxes are the driving force behind the ex‐day results,

then we should not observe similar ex‐day price behavior for non‐taxable distributions. However, we find significant abnormal

ex‐day returns for the non‐taxable Nasdaq distributions, providing further evidence against the tax hypothesis.

Not all investors, however, have a tax preference for dividends over capital gains. For example, short‐term traders and tax‐

exempt institutions do not face a tax differential between dividends and capital gains. Moreover, corporations have a preference

for dividends due to the 70% exclusion on corporation dividend income. Thus, in the absence of transaction costs, an expected

PDR of less than one presents profit opportunities to some market participants.5

Consequently, another major strand of the lit-

erature argues one cannot infer the marginal tax rate of retail investors from the PDR. Rather, the PDR reflects the transaction

costs faced by tax‐neutral or tax‐advantaged entities (see, e.g., Boyd & Jagannathan, 1994; Kalay, 1982; Karpoff & Walkling,

1988, 1990; Koski, 1996; Michaely & Vila, 1996). This view is called the “short‐term trading hypothesis.”6

We employ the framework of Boyd and Jagannathan (1994) to test the short‐term trading hypothesis. Boyd and Jagannathan

(1994) develop a model that incorporates trading among individual investors, tax‐neutral arbitrageurs and tax‐advantaged

corporate traders in the presence of both transaction costs and risk. In regressions of the ex‐day percentage price drop on divi-

dend yield, Boyd and Jagannathan (1994) find a slope coefficient of one. Within their framework, this suggests that short‐term

traders are the marginal price‐setters around ex‐days and provides evidence consistent with the short‐term trading hypothesis.

We find, however, that for Nasdaq‐listed firms the slope coefficients are significantly less than one, even after controlling for

taxes and transaction costs. This is consistently the case in both the overall sample and the more recent period. Thus, we do

not find support for the short‐term trading hypothesis. We note that the control variable regression coefficients are statistically

significant and have signs consistent with what theory predicts. However, the combined explanatory power is not sufficient to

explain the ex‐day behavior of Nasdaq‐listed firms.

Another strand of the literature focuses on market microstructure‐based explanations. Bali and Hite (1998) argue price

discreteness imposed by minimum tick‐sizes can induce PDRs of less than one. However, using NYSE firms Graham et al.

(2003) and Jakob and Ma (2004) do not find support for this argument. For example, Graham et al. (2003) find PDRs signifi-

cantly decrease around tick‐size reductions for NYSE firms, opposite the tick‐size prediction. Following a similar procedure,

we investigate whether Nasdaq PDRs change around mandated tick‐size reductions. We do not find any significant change in

Nasdaq PDRs. We also investigate other potential microstructure arguments. Prior to the mid‐1990s, Nasdaq dealers avoided

odd‐eighths price quotes (Christie & Schultz, 1994). In addition, prior to 1993 Nasdaq dealers did not automatically adjust

prices for outstanding limit orders on ex‐dividend days (Dubofsky, 1992; Jakob & Ma, 2004). Both of these factors could po-

tentially produce PDRs significantly less than one. When we exclude these periods, our results and inferences are unaffected.7

Thus, these market microstructure‐based arguments do not explain our results.

Finally, a recent study by Hartzmark and Solomon (2019) finds evidence that investors trade as if they track dividends and

price changes as separate and independent accounts, without fully appreciating the fact that dividends come at the expense of