Real Anomalies

• DOI: http://doi.org/10.1111/jofi.12771
• Author: JULES H. van BINSBERGEN,CHRISTIAN C. OPP
• Date: 01 August 2019
• Published date: 01 August 2019

THE JOURNAL OF FINANCE •VOL. LXXIV, NO. 4 •AUGUST 2019

Real Anomalies

JULES H. van BINSBERGEN and CHRISTIAN C. OPP∗

ABSTRACT

We examine the importance of cross-sectional asset pricing anomalies (alphas) for the

real economy.To this end, we develop a novel quantitative model of the cross-section of

firms that features lumpy investment and informational inefficiencies, while yielding

distributions in closed form. Our findings indicate that anomalies can cause material

real inefficiencies, which raises the possibility that agents who help eliminate them

add significant value to the economy. The model shows that the magnitude of alphas

alone is a poor indicator of real outcomes, and highlights the importance of the alpha

persistence, the amount of mispriced capital, and the Tobin’sqof firms affected.

OVER THE PAST FEW DECADES, a vast literature has developed that attempts to

document and explain the behavior of asset prices both in the cross-section and

in the time series. The seminal paper on excess volatility—Shiller (1981)—has

spurred a literature that attempts to explain why stock markets are so volatile

and whether such volatility is excessive. The literature has further uncovered

a large number of “anomalies” in the cross-section of asset prices that suggest

the presence of mispricings.1

These empirical findings raise an important question: if the documented pat-

terns in financial market returns do indeed reflect informational inefficiencies

in the economy, how large are the resulting real efficiency losses? In this paper,

∗Jules H. van Binsbergen is with the Wharton School of the University of Pennsylvania and

NBER. Christian C. Opp is with the Wharton School of the University of Pennsylvania. We thank

the Editor Stefan Nagel and an anonymous referee for their helpful and constructive feedback.

Moreover, we thank Andy Abel, Malcolm Baker (discussant), Markus Brunnermeier,John Camp-

bell, Andrea Eisfeldt, Nicolae Gˆ

arleanu, Vincent Glode, Itay Goldstein, Jo˜

ao Gomes, Urban Jer-

mann, Tom Sargent, Sydney Ludvigson, Hanno Lustig, Stijn van Nieuwerburgh, Dimitris Pa-

panikolaou (discussant), Thomas Philippon (discussant), Krishna Ramaswamy, Rob Stambaugh,

Jeremy Stein, Jessica Wachter, Tony Whited, Amir Yaron, and seminar participants at the 7th

Advances in Macro-Finance Tepper-LAEF Conference, Berkeley Haas, BI-SHoF Conference on

Financial Econometrics, London Business School, Michigan Ross, NBER Summer Institute Asset

Pricing, University of Oxford, Princeton University,and the Wharton Conference on Liquidity and

Financial Crises for their helpful comments. We have read the Journal of Finance’s disclosure

policy and have no conflict of interests to disclose.

1Examples include anomalies based on Tobin’sq(the value premium), investment, profitability,

and past return performance (momentum). See, for example, Rosenberg, Reid, and Lanstein (1985)

for the value premium and Jegadeesh and Titman (1993) for momentum. For the profitability

anomaly, see Ball and Brown (1968), Bernard and Thomas (1990), and Novy-Marx (2013). For the

investment anomaly, see Fairfield, Whisenant, and Yohn (2003), Titman, Wei, and Xie (2004), and

Cooper, Gulen, and Schill (2008).

DOI: 10.1111/jofi.12771

1659

1660 The Journal of Finance R

we address this question and find that the economy-wide real distortions can

be substantial.

Our focus on real investment distortions connects our study to the literature

in macroeconomics that quantifies efficiency losses due to resource misalloca-

tion (see, for example, Hsieh and Klenow (2009)).2Compared to this literature,

we focus on a novel friction—cross-sectional asset pricing wedges—that is dis-

ciplined by alphas estimated in the finance literature. We introduce a tractable

dynamic framework that maps distortions in agents’ subjective beliefs to these

alphas, and we use the model to evaluate the aggregate implications for the

real economy.

Any study aiming to quantify misallocations can only do so conditional on

the model that is imposed by the econometrician. In particular, the stance on

what constitutes the efficient benchmark ultimately determines what is iden-

tified as a wedge. In our context, informational inefficiencies encoded in prices

can only be measured with some postulated asset pricing model. This issue is

known in the empirical asset pricing literature as the joint-hypothesis problem

(Fama (1970,1991)). Rather than solve the important problem of determining

which model is the correct efficient benchmark, we develop a flexible method-

ology that allows for the assessment of the real distortions associated with

alphas computed under a variety of standard asset pricing models. While we

use the CAPM to identify alpha wedges, other asset pricing models can be

accommodated.

Moreover, our methodology can be used by economists who believe that mar-

ket prices are always informationally efficient. Conditional on this premise, our

framework identifies the quantitative importance of various risk factors for real

economic activity. That is, it reveals which types of asset pricing facts are im-

portant for macroeconomics. The unifying insight of our paper, independent of

one’s view on the informational efficiency of prices, is that not all anomalies

or risk factors are created equal—some are significantly more important than

others from the perspective of their effects on the real economy. For exposi-

tional clarity, we will interpret our results below conditional on the premise

that alphas do measure informational inefficiencies, but we also provide an in-

depth analysis of how our methodology can be used under the premise that they

capture omitted risk factors. Ultimately, the purpose of this paper is to initiate

discussion on which documented return patterns are relevant for aggregate

economic activity.

While alphas indicate mispricing conditional on the postulated asset pricing

model, we find that the magnitude of alphas alone is generally a poor measure

of the potential real economic distortions. First, alphas merely represent ex-

pected changes in the level of mispricing over a given period. Mispricing is an

inherently dynamic phenomenon—as alphas are realized over time, mispricing

can build up and/or resolve, leaving firms only temporarily affected by price

distortions. It is therefore essential to also account for the persistence of alphas,

2See also Eisfeldt and Rampini (2006) for evidence on the amount of capital reallocation between

firms and the cost of reallocation.

Real Anomalies 1661

a factor that is captured by our dynamic model. For the aggregate economy, a

relatively small but persistent alpha is more distortive than a very short-lived

but large alpha. The investment management literature cares primarily about

the magnitude of alpha (generally after trading costs) regardless of its per-

sistence, yet real corporate investment decisions are not driven by short-lived

mispricing. Second, since alphas measure the percentage change in mispricing,

they do not provide an accurate representation of the economy-wide value of

firms affected. Just as the internal rate of return cannot be used to measure

the value of an investment opportunity (it is the net present value [NPV] that

matters), the magnitude of alpha alone cannot be used to measure the eco-

nomic importance of an anomaly—information on the market capitalization of

affected firms should also be taken into account. Third, and most importantly,

the extent to which mispricing translates into real investment distortions and

surplus losses depends on firms’ real investment opportunities. We show, for

example, that firms with high Tobin’s qrespond more to mispricing than firms

with low Tobin’sq, implying that the mispricing of growth firms tends to lead to

larger efficiency losses. Because firms with low Tobin’s qwish to disinvest but

face significant frictions in doing so (e.g., due to partly irreversible investment),

these firms’ real responses to mispricing are also dampened significantly.

An initial observation indicating the potential real impact of mispricing is

the fact that cross-sectional variation in investment (asset growth) is related

to future abnormal stock returns.3One channel that could explain a relation

between asset growth and mispricing is overvalued firms’ opportunity to raise

cash cheaply without investing in new capital. However, given that the relation

between asset growth and alpha is almost identical after excluding cash hold-

ings from assets (see the Internet Appendix4), this channel does not appear to

have a first-order effect. Instead, the investment-alpha relation suggests that

firms with inflated (deflated) prices—which will be corrected in the future—

overinvest (underinvest) in capital today.

To assess real effects quantitatively, we estimate the joint dynamic distribu-

tion of firm characteristics that have been linked to mispricing and other firm

variables, such as investment and capital. We develop a novel quantitative

model that features stochastic lumpy adjustment (SLA) to capital. This SLA

technology yields closed-form solutions for the distribution of firm dynamics

for any given policy function, thereby facilitating the estimation of the model

based on panel data.5In the model, decision makers use (dis)information en-

coded in market prices when making investment decisions (Hayek (1945)). We

target and successfully match more than 40 moments describing the cross-

sectional distributions of firm size, Tobin’s q, the relation between Tobin’s q

3As discussed below,investment alphas arise under a variety of commonly used empirical asset

pricing models.

4The Internet Appendix is available in the online version of the article on the Journal of

Finance website.

5In Binsbergen and Opp (2019), we propose a modeling approach that features this SLA tech-

nology and yields exact solutions to general equilibrium economies with persistent heterogeneity

(such as the economy considered in Krusell and Smith (1998)).