Types of investor uncertainty and the cost of equity capital

• Author: Robert J. Resutek,Chad R. Larson
• DOI: http://doi.org/10.1111/jbfa.12283
• Published date: 01 October 2017
• Date: 01 October 2017

DOI: 10.1111/jbfa.12283

Types of investor uncertainty and the cost

of equity capital

Chad R. Larson1Robert J. Resutek2

1C.T. Bauer College of Business, University of

Houston,USA

2TullSchool of Accounting, University of Georgia,

USA

Correspondence

ChadR. Larson, C.T.Bauer College of Business,

Universityof Houston, 7680 Calhoun Road

Houston,TX 77004, USA.

Email:crlarson@uh.edu

Abstract

We jointly test the effects of two types of investor uncertainty,

one related to future firm performance and unrelated to accruals

(cash flow uncertainty) and one directly related to accrual estima-

tion errors (accounting quality uncertainty). Distinct from prior stud-

ies, our uncertainty estimates are based on a matched-firm design

that minimizes the mechanical relationship between the two uncer-

tainty variables. Wefind a strong negative relationship between cash

flow uncertainty and multiple estimates of the cost of equity capi-

tal. With respect to accounting quality uncertainty, we find a strong

positive association with both expected stock returns and implied

costs of equity, but only in settings that control for cash flow uncer-

tainty.Collectively, our results suggest the need to consider different

types of investor uncertainty when examining how investor uncer-

tainty affects the cost of equity capital.

KEYWORDS

cost of equity capital, earnings management, earnings quality, earn-

ings uncertainty,information uncertainty

1INTRODUCTION

This study examines the relationship between two different types of investor uncertainty and costs of equity capital

as proxied by expected stock returns and implied costs of equity.1Thefirst type is cash flow uncertainty (CFU). Cash

flow uncertainty is important to consider for at least two reasons: (i) it is a measure of investor uncertainty in future

performance that is unaffected by accrual measurement error,and (ii) it is a form of investor uncertainty associated with

the fundamental parameters of firm growth. The second source is accounting quality uncertainty (AQU). We interpret

accounting quality uncertainty as capturing how uncertain investorsare about firm value because of accrual accounting

and empirically estimate AQU using the conventional cross-sectional residual accrual volatility design of Dechow and

Dichev (2002).

Motivation for our study stems from studies in accounting and finance that document two strikingly divergent

empirical relations. The first relation, noted primarily in the finance literature,centers on a strong negative relationship

1Throughoutthe article when discussing the theoretical construct, the cost of equity capital, we use the terms ‘expected stock returns’ and the ‘cost of equity’

interchangeably.When discussing empirical results and theoretical proxies, we specifically refer to expected stock returns and implied costs of equity.

J Bus Fin Acc. 2017;44:1169–1193. wileyonlinelibrary.com/journal/jbfa c

2017 John Wiley & Sons Ltd 1169

1170 LARSON ANDRESUTEK

between investor uncertainty in future performance and expected stock returns (Ang, Hodrick, Xing, & Zhang, 2006;

Diether, Malloy,& Scherbina, 2002; Johnson, 2004). The second relation, explored primarily in the accounting litera-

ture, centers on a strong positive relationship between implied equity costs and investor uncertainty attributable to

low earnings quality (Francis, Lafond, Olsson, & Schipper,2004, 2005) and a largely insignificant relationship between

expected stock returns and earnings quality (Core,Guay, & Verdi, 2008; McInnis, 2010).

Prior empirical studies have largely overlooked the fact that if different types of investoruncertainty explain the

cost of equity differently, these differences could significantly influence empirical inferences. Specifically, if differ-

ent types of investor uncertainty are not considered jointly, and the different types explain expected stock returns

in offsetting directions, the economic importance of investor uncertainty will be attenuated if the types are posi-

tively correlated; i.e., correlated omitted variable bias. This econometric concern seems reasonable given the incon-

sistent empirical evidence prior studies note on the relationship between different measures of investor uncertainty

and expected stock returns (e.g., Core et al., 2008; Francis et al., 2004; McInnis, 2010; Perotti & Wagenhofer, 2014;

Rountree, Weston, & Allayannis, 2008).

Key to effectively tackling questions involving different sources of investoruncertainty is an empirical design that

can estimate both sources with limited mechanical relation to one another. Dating back to at least McNichols (2002),

researchershave noted a strong mechanical relation between a firm's accounting quality and time-series variation in its

cash flows. Conventionalmeasures of accounting quality uncertainty (AQ U) such as residual accrual volatility (Dechow

& Dichev,2002) or earnings smoothness (Leuz, Nanda, & Wysocki, 2003) are, by construction, mechanically linked to

cash flow volatility.The tight mechanical connection between cash flow volatility and AQU variables makes it difficult

to determine whether inferences drawn on investoruncertainty are due to uncertainty in a firm's cash flow process or

the precision with which the accrual process estimates unrealized cash flows.2

We use a novel research design to sidestep some of the empirical difficulties caused by this tight mechanical link.

Specifically,we use a matched-firm empirical design that produces estimates of uncertainty in future cash flows (CFU)

from the cash flow patterns of firms other than the firm of interest (Blouin, Core, & Guay, 2010;Donelson & Resutek,

2015). By construction, CFU will have minimal mechanical relation to our AQU estimates, which we estimate using

a conventional residual accrual volatility design (Dechow & Dichev, 2002). Our research design allows us to more

definitively test whether different types of investor uncertainty havedifferential explanatory powers for future stock

returns by minimizing the mechanical overlapbetween the two uncertainty variables.

Our central empirical result suggests that these two types of uncertainty,CFU and AQU , have significant, but coun-

tervailing, effects on a firm's cost of equity. Our results suggest that the inconsistent empirical evidence offered in

prior studies on the relationship between investor uncertainty,expected stock returns, and the implied cost of equity

may not be ‘spurious’ or the result of poor research designs. Instead, we find that by carefully constructing differ-

ent measures of these two types of investor uncertainty and allowing them to vary,independently, within a research

design, empirical inferences consistent with theory are observable in the data. We present our empirical results in two

parts.

Our primary empirical analysis offers several new insights on how different types of investor uncertainty relate

to future stock returns and the implied cost of equity. First, in predictive regressions of future monthly stock returns

over the standard twelve-month horizon, slopes on CFU are consistently negative, with the slope precisions ranging

between 4.0 and 6.0 standard errors from zero. We find similar inferences in portfolio-based empirical designs where

the average spread in excessstock returns across low and high CFU firms is between 0.25% and 0.50% per month and

statistically significant above conventionallevels.

Second, consistent with prior studies (Core et al., 2008; McInnis, 2010; Ogneva, 2012), we do not find evidence

of a relationship (or a slightly negative relationship) between AQU and expected stock returns. In settings controlling

for CFU, however,we find a consistently positive relationship between AQ U andexpected stock returns. Across both

regression and portfolio empirical designs, our results suggest the average spread in expectedmonthly returns across

2Liuand Wysocki (2007) and Dichev andTang (2009) make similar points.