Pushing the future back: The impact of policy uncertainty on the market pricing of future earnings

• Author: Michael A. Mayberry,Michael S. Drake,Jaron H. Wilde
• DOI: http://doi.org/10.1111/jbfa.12316
• Published date: 01 July 2018
• Date: 01 July 2018

DOI: 10.1111/jbfa.12316

Pushing the future back: The impact of policy

uncertainty on the market pricing of future

earnings

Michael S. Drake1Michael A. Mayberry2Jaron H. Wilde3

1MarriottSchool of Management, Brigham Young

University,Provo, Utah, USA

2WarringtonCollege of Business Administration,

Universityof Florida, Gainesville, Florida, USA

3HenryB. Tippie College of Business, University

ofIowa, Iowa City, Iowa, USA

Correspondence

MichaelS. Drake, Brigham YoungUniversity, 521

TNRB,Provo, Utah, 84602.

Email:mikedrake@byu.edu

Abstract

We examine whether policy uncertainty triggered by presidential

elections pushes the future back byreducing the extent to which cur-

rent prices reflect information about future earnings. We estimate

future earnings response coefficients (FERCs) for the years 1975–

2013, a period that covers ten presidential elections. We find that

FERCs are significantly lower( by11.8%) during presidential election

years compared to other years. Additional analyses using pseudo

election years, ex-ante polls, contract prices from the Iowa Elec-

tronic Political Market, and cross-sectional firm characteristics pro-

vide corroborating evidence that the lower FERCs during election

years are related to policy uncertainty. We also investigate poten-

tial explanations for the lower FERCs during election years. Wefind

that the lower FERCs relate to forecasting difficulty rather than to

changes in the discount rateor in the amount of noise trading. Finally,

we find that market prices movetoward future earnings to a greater

degree during presidential election years compared with other years

once the policy uncertainty is resolved. A trading strategy based on

this drift yields significant abnormal returns. Overall, we contribute

to the literature byproviding the first empirical evidence that shocks

to policy uncertainty influence the pricing of earnings information.

KEYWORDS

future earnings response coefficient, policy uncertainty

1INTRODUCTION

Since Ball and Brown (1968) capital markets researchers in accounting havebeen interested in understanding the fac-

tors that help explain the extent to which stock prices reflect earnings information. Lundholm and Myers(2002) initi-

ated an important branch of this literatureby examining factors (e.g., disclosure quality) that ‘bring the future forward’,

increasing the extent to which current prices reflect future earnings information. This branch of literatureidentifies a

variety of factors that bring the future forward (see, for example, Ayers& Freeman, 2003; Ettredge, Soo Young, Smith,

& Zarowin, 2005; and Choi, Myers, Zang, & Ziebart, 2011). Broadly speaking, we are interested in the opposite notion.

J Bus Fin Acc. 2018;45:895–927. wileyonlinelibrary.com/journal/jbfa c

2018 John Wiley & Sons Ltd 895

896 DRAKE ET AL.

That is, we examine whether a specific factor – namely, government policy uncertainty – will push the future back,

decreasing the extent to which current prices reflect future earnings information.

Stockprices reflect a value-weighted consensus of investor opinions about future performance (Lee, 2001). We con-

jecture that uncertainty about political policies will inject an additional constraint into the process through which the

market impounds these opinions into prices. Our rationale is that policy uncertainty yields a broader arrayof possible

future (post-election) earnings outcomes that makes it more difficult for market participants to forecast future eco-

nomic benefits and capitalize these expectations into prices. While there is likely to be some amount of policy uncer-

tainty present in marketsat any given point in time, national elections inject significant policy uncertainty into this sys-

tem because they often lead to changes in regulations, taxes, economic philosophy,and fiscal policy that are difficult

to predict and potentially affect firms’ future performance (Białkowski, Gottschalk, & Wisniewski, 2008; Boutchkova,

Doshi, Durnev,& Molchanov, 2012). We examinewhether policy uncertainty triggered by important national elections

impacts the amount of future earnings information that is reflected in current prices.

We rely on exogenousshocks to policy uncertainty surrounding the presidential election season in the US, as pres-

idential election years havethe most potential for dramatic shifts in policy resulting from changes in elected officials.1

In a presidential election year, the presidency,all of the House of Representatives seats, and one-third of the Senate

seats are up for re-election. We study US presidential election years because, unlikeother countries, the timing of the

elections is not at the discretion of the sitting government. The US Constitution requires a presidential election every

four years.Our assumption is that election years introduce incremental policy uncertainty into the market place, above

and beyond that observed duringany other time period, providing exogenous variation in policy uncertainty.

Following Lundholm and Myers (2002), we test our predictions by examiningthe future earnings response coeffi-

cient(FERC), which explains current returns using prior, current and future earnings. The coefficient on future earnings,

the FERC,estimates the amount of future earnings information embedded in current prices. This model is well suited to

examineour research question because the current period policy uncertainty relates to how the newly elected regimes

will impact future earnings. We estimate our models using a large sample of firms over the 1975 to 2013 time period,

which covers 10 presidential election seasons, beginning with Carter versusFord in 1976 and ending with Obama ver-

sus Romney in 2012.

In our primary test we find that FERCs are significantly lower (less positive)for election years than for non-election

years. This result obtains after controlling for firm-specific characteristics(i.e., firm size, profitability, growth and earn-

ings volatility), macro-economic variables (consumer price index, risk-free rate, unemployment and gross domestic

product), and economic policy uncertainty (as measured in Baker, Bloom, & Davis,2013). In addition, a Monte-Carlo

simulation provides no evidence that FERCs are lower during pseudo election years, which mitigates concerns that

our main findings are due to random chance. We conduct a series of robustness tests and confirm that our primary

result is not attributable to election year earnings management, election year earnings quality, or to anysingle pres-

idential election season. We also verify that our results hold in samples of firms with strong and weak information

environments.

Next, we conduct a series of additional tests to provide corroborating evidence that the attenuated earnings

responses during election years are related to policy uncertainty.In the first test we examine whether the attenuating

effect of elections on FERCs is stronger for closer presidential races. Importantly,for this test we restrict the sample

to election years and thus, exploit variation in political uncertainty within election years only.We find that the FERCs

are significantly lower for ‘tighter’ elections when the Gallup poll projected margin of victory is lower.We also use con-

tract prices from the Iowa Electronic Market (IEM) as an alternativeway to measure close presidential races. We find

that FERCs are lower when the IEM predicts a closer race. Together, these results provide additional support that the

informativeness of stock prices with respect to future earnings information is lower in the presence of higher policy

uncertainty.

1Manystudies in accounting and finance are based on a single shock to firms. For example, Ramanna and Roychowdhury (2010) exclusivelyfocus on the 2004

electionand Jayachandran (2006) examine a single legislator's change in political affiliation. We have nine such shocks across multiple decades, strengthening

ourability to draw inferences across different time periods and in different political climates.

DRAKE ET AL.897

In the second test we examinewhether the attenuating effect of elections on FERCs is stronger in firms with income

streams that are potentially more sensitive to political outcomes. We find that the attenuating effect of policy uncer-

tainty on FERCs during election years is relatively more pronounced for firms for which the US Government is a key

customer and for firms with a high probability of engaging in tax sheltering. We also decompose future earnings in

the FERC model into future US-based earnings and future foreign earnings. Consistent with US presidential elections

inducing uncertainty regarding future domestic earnings in particular, we find that the dampening effect of policy

uncertainty on the FERC is observed for future US-based earnings, but not for future foreign earnings.

In our next set of analyses, we investigatethree potential channels through which policy uncertainty could result in

lower earnings informativeness. Given that the FERC estimates the relationship between current returns and future

earnings, it follows that policy uncertainty could affect the FERC by: (1) inducing less informative forecasts of future

earnings, (2) increasing the cost of equity capital, and/or (3) inducing less informativecurrent returns. We investigate

these possibilities and find evidence that supports the forecasting explanationonly. More specifically, we find that ana-

lyst forecast accuracy improves to a significantly greater degree in the window following the election period than it

does during corresponding periods in non-election years. This result is consistent with future earnings being more dif-

ficult to forecast during election years when it is more challenging to assess how future policies and actions are likely

to affect firms’ future earnings.

Weconclude with a number of tests that examine post-election price movements. Our objective with this analysis is

toinvestigate whether markets impound information about future earnings to a greater degree once the election result

is known and the policy uncertainty is reduced. We first examineshort-window returns (two trading days) around elec-

tions and corresponding days around non-election years and find that the association between short-window returns

and future earnings is significantly more positive around national elections than for the corresponding days in non-

election years. This result suggests that with the news of the election results, and thus the exogenousreduction in pol-

icy uncertainty,market prices move towards future earnings to a greater degree during election years relative to other

years. We also investigate the relationship between future earnings and longer-window returns after the election by

estimating returns over the following 60 trading days. We find that the relationship between these longer window

returns and future earnings is also more positive for election years than for non-election years.

Our finding that prices move to a greater extent in the direction of future earnings during the post- election time

period suggests that it may be possible to implement a trading rule that generates significant abnormal returns. To

investigate this possibility,we implement the following trading strategy: Each year, on the first Tuesday following the

first Monday of November (the election date for presidential election years and the pseudo election date for other

years), firms are sorted into five portfolios based on current earnings, which is used as the forecast for future earnings

(Bradshaw,Drake, Myers, & Myers, 2012). Three days after the election date (or the pseudo election date in the case of

non-election years), the strategy goes long (short) in the top (bottom) quintile based on this earnings forecast and the

positions are held for 60 trading days. The strategy is dormant until the subsequent year.Using calendar-time, four-

factor models, we find that this strategy yields an alpha that is 7.0 basis points higher (or4.20% over the 60-day period)

during election yearsthan during non-election years. It is important to note however that investors cannot implement a

strategyto earn this 7.0 basis point spread because the election and non-election year portfolios cannot be constructed

at the same time. Any given year is either an election or non-election year,but never both. However, during election

years, we find that the strategyyields, on average, 2 basis points per day over the 60-trading day window, which is 7.2%

annualized.

In summary, our analyses contribute a new stylized fact to the literature about how policy uncertainty impacts

markets: prices are less informative with respect to earnings information when policy uncertainty is high, due in part

to the challenge of forecasting earnings. In effect, policy uncertainty pushes the future back from a price formation

perspective. This study also contributes to several specific streams of literature.First, we contribute to the literature

that investigatesfactors influencing the relationship between current returns and future earnings. Prior research finds

that disclosure quality (Choi et al., 2011; Lundholm & Myers, 2002), reporting standards (Ettredge et al., 2005; Hsu &

Pourjalali, 2014), analysts and institutional investors (Ayers & Freeman, 2003), and short sellers (Drake, Myers,

Myers, & Stuart, 2015) all bring the future forward by influencing the extent to which prices reflect future earnings