Investment Dynamics and Earnings‐Return Properties: A Structural Approach

• Published date: 01 June 2019
• DOI: http://doi.org/10.1111/1475-679X.12253
• Author: MATTHIAS BREUER,DAVID WINDISCH
• Date: 01 June 2019

DOI: 10.1111/1475-679X.12253

Journal of Accounting Research

Vol. 57 No. 3 June 2019

Printed in U.S.A.

Investment Dynamics and

Earnings-Return Properties:

A Structural Approach

MATTHIAS BREUER ∗AND DAVID WINDISCH †

Received 18 January 2017; accepted 12 November 2018

ABSTRACT

We propose the standard neoclassical model of investment under uncertainty

with short-run adjustment frictions as a benchmark for earnings-return pat-

terns absent accounting influences. We show that our proposed benchmark

generates a wide range of earnings-return patterns documented in account-

ing research. Notably, our model generates a concave earnings-return rela-

tion, similar to that of Basu [1997], and predicts that the earnings-return

concavity increases with the volatility of firms’ underlying shock processes

and decreases with the level of firms’ investments. We find strong empir-

ical support for these predictions. Overall, our evidence suggests that our

proposed benchmark is useful for understanding the joint dynamics of vari-

ables of interest to accounting research (e.g., earnings, returns, investment,

∗Columbia University; †University of Graz.

Accepted by Rachel Hayes. We gratefully acknowledge helpful code publicly provided

by Dean Corbae and comments and suggestions from two anonymous referees, Ray Ball,

Philip G. Berger, Jeremy Bertomeu, Jonathan Black (discussant), Pietro Bonetti, Jung Ho

Choi, Holger Daske (discussant), Max G¨

odl, Christian Leuz, Brett Lombardi, Valeri Niko-

laev, Ali Cem Randa, Georg Schneider (discussant), Alfred Wagenhofer, Anastasia Za-

kolyukina, and seminar participants at the University of Chicago Booth School of Busi-

ness, Duke University, SKEMA Business School, University of Bath, VHB Conference 2017,

EAA Annual Meeting 2017, AAA Annual Meeting 2017, and the AS-VHB/IAAER Annual

Accounting Conference 2018. An online appendix to this paper can be downloaded at

http://research.chicagobooth.edu/arc/journal-of-accounting-research/online-supplements.

639

CUniversity of Chicago on behalf of the Accounting Research Center,2018

640 M.BREUER AND D.WINDISCH

market-to-book) absent accounting influences, a necessary precondition for

inferring the effects of accounting from these dynamics.

JEL codes: D25; G10; G31; M40; M41

Keywords: dynamic investment; conservatism; earnings-return relation

1. Introduction

A substantial body of accounting research investigates the properties of ac-

counting earnings, such as timeliness and conservatism, by studying the

contemporaneous relation between earnings and returns.1A key challenge

to this literature is to discern the effects of accounting-related factors,

such as accounting rules and managerial reporting discretion, on earnings-

return patterns from the effects of economic fundamentals. Despite recent

progress (e.g., Dutta and Patatoukas [2017], Hemmer and Labro [2018]),

we still lack a basic understanding of how economic fundamentals jointly

determine earnings and returns absent accounting rules and measurement

issues. Without such an understanding of the economic null, however, in-

ferences that can be drawn from empirically observed earnings-return pat-

terns about accounting effects are limited.

In this paper, we propose a simple model of the firm and investigate

its implications for the properties of earnings and returns absent account-

ing rules and measurement issues, to make progress toward an economic

benchmark for earnings-return patterns. In line with economics and fi-

nance literatures, we model the firm as an infinite-horizon dynamic invest-

ment problem under uncertainty with symmetric (convex) capital adjust-

ment costs.2In each period, a risk-neutral manager, acting on behalf of

shareholders, faces a partially persistent shock to the profitability of capi-

tal, the firm’s only fixed production factor, and chooses next period’s capi-

tal stock (through this period’s investment) to maximize the present value

of future cash flows.

Absent a closed-form solution, we numerically solve the model using dy-

namic programming techniques. Equipped with the optimal investment

behavior (policy function) and firm value (value function) for each pos-

sible state, we simulate a time series of earnings, returns, and capital (in-

vestments) in response to a deliberately symmetric exogenous profitability

shock process. The simulated data allow us to investigate the earnings and

return patterns generated by our model.

We show that our proposed model generates a wide range of well-

documented empirical earnings and return patterns. Compared to pre-

vious static valuation models (e.g., Feltham and Ohlson [1995], Ohlson

[1995], Holthausen and Watts [2001]), our model adds two features con-

tributing to its ability to match empirical earnings and return patterns.

1See Kothari [2001] for a survey of the literature.

2See Strebulaev and Whited [2012] for a survey of the literature.

INVESTMENT DYNAMICS AND EARNINGS-RETURN PROPERTIES 641

First, it features real expansion and adaption options (e.g., Burgstahler

and Dichev [1997], Zhang [2000]). These real options introduce notable

nonlinearities in the predicted earnings and return patterns, allowing our

model to match empirical patterns, such as the differential persistence of

profits and losses (e.g., Hayn [1995]) and the nonlinear shape of earnings-

response coefficients (e.g., Freeman and Tse [1992], Wysocki [1999]).

Second, our model features a short-run fixed production factor (due to

time-to-build and convex capital adjustment costs; e.g., Lucas and Prescott

[1971], Hayashi [1982]). This short-run fixed production factor introduces

real dynamics into investment decisions and the earnings process, allowing

our model to match empirical patterns, such as the autocorrelation of in-

vestments (e.g., Cooper and Haltiwanger [2006], Strebulaev and Whited

[2012]) and the earnings-return concavity (e.g., Basu [1997]).

Our model’s ability to generate a concave earnings-return relation is par-

ticularly intriguing, given that it is based on symmetric primitives (e.g.,

symmetric profitability shocks and adjustment costs).3Two channels jointly

generate the prominent earnings-return concavity, even in the absence

of accounting conservatism. First, current-period earnings respond more

strongly to negative than to positive persistent profitability shocks. The in-

tuition behind this is as follows. Suppose a firm experiences a persistent

shock to the profitability of its short-run fixed capital stock (e.g., property,

plant, and equipment) in the current period. To achieve an optimal level

of capital in the future, given the current-period shock, the firm invests or

divests, depending on the sign of the shock. In changing its capital stock,

however, it incurs adjustment costs (e.g., costs associated with production

disruptions and managerial planning) that reduce its current-period earn-

ings. Thus, adjustment costs exacerbate the earnings-decreasing effects of

negative profitability shocks and reduce the earnings-increasing effects of

positive profitability shocks.

Second, and more importantly, future earnings respond more strongly

to positive than to negative persistent profitability shocks. Capital invest-

ments, in response to a persistent positive shock in the current period, al-

low firms to take advantage of their increased demand or productivity in

the future. Hence, the capital expansion in response to a persistent posi-

tive shock increases future earnings (expansion (call) option; e.g., Pindyck

[1988], Wysocki [1999]). By contrast, capital divestments in response to

a persistent negative shock in the current period allow firms to minimize

the impact of their decreased demand or productivity in the future. Thus,

the capital curtailment in response to a persistent negative shock curbs

3For a precise and consistent nomenclature, we refer to an earnings-return relation with

a decreasing slope as a “concave” relation or “earnings-return concavity.” The Basu [1997]

coefficient or asymmetry, for example, represents a special case of a weakly concave earnings-

return relation. By contrast, we refer to an earnings-return relation with an increasing slope

as a “convex” relation or “earnings-return convexity.”