Of leaders and followers—An econometric analysis of equity analysts and stock market investors

• Author: Rainer Baule,Hannes Wilke
• Published date: 01 January 2019
• Date: 01 January 2019
• DOI: http://doi.org/10.1002/ijfe.1676

Received: 1 December 2017 Revised: 4 September 2018 Accepted: 9 September 2018

DOI: 10.1002/ijfe.1676

RESEARCH ARTICLE

Of leaders and followers—An econometric analysis of

equity analysts and stock market investors

Rainer Baule1Hannes Wilke2

1Faculty of Business and Economics,

University of Hagen, Hagen, Germany

2Financial Stability Department Deutsche

Bundesbank, Frankfurt, Germany

Correspondence

Rainer Baule, University of Hagen,

Universitätsstraße 41, 58097 Hagen,

Germany.

Email: rainer.baule@fernuni-hagen.de

JEL Classification: G12; G14; G17; G2;

C32; C53; D82

Abstract

We measure the information content of publicly available analyst consensus

forecasts for 1-year-forward earnings per share based on two well-established

price discovery measures drawn from the area of market microstructure

research. Employing a 36-year sample of large U.S. companies listed in the S&P

100 Index, we compute (a) Hasbrouck's information shares and (b) Gonzalo and

Granger's common factor component shares to measure the relative contribu-

tion that market participants and financial analysts working on the sell side of

the market have in the process of value discovery. We find that although these

analysts are far from leading the market, they have a small but significant share

in the process of value discovery, amounting to 4.6% (Hasbrouck) and 18.0%

(Gonzalo and Granger), on average. This share is persistent over time but varies

significantly in the cross section. We identify the level of analyst coverage and

company age as drivers of this heterogeneity.

KEYWORDS

analyst herding, earnings-per-share consensus forecasts, information share, marketefficiency, price

discovery, sell-side analysts

1INTRODUCTION

Financial analysts act as information intermediaries for

capital markets. Their key tasks are the estimation of

the companies' prospects and the transformation thereof

into earnings forecasts, target prices, buy–sell–hold rec-

ommendations, and so forth, ideally combining strong

economic knowledge with a close connection to the man-

agement of the companies on their coverage list. However,

the simple existence of analysts who improve the process

of price discovery conflicts with the theory of information-

ally efficient markets. According to the efficient market

hypothesis (EMH), every single piece of information rel-

evant to investors is already incorporated into the stock

price. Only for markets that are not alwaysinformationally

efficient, analysts could provide some informational lead-

ership if they were able to incorporate information into

their forecasts that has not yet been priced (correctly) by

the market participants. The degree of such informational

leadership depends upon the relative amount of informa-

tion that is reflected faster by analyst forecasts, compared

with stock market prices.

The empirical question as to the degree to which analysts

are able to fulfil such a role, and thus justify their existence

(i.e., whether analyst output contains information that is

at all relevant to the market), remains open and is still

vigorously debated in the empirical literature. Most publi-

cations in this area apply event studies in order to identify

and measure abnormal stock returns coinciding with the

release of new analyst content. Another extensive strand of

literature formulatestrading strategies that aim to generate

508 © 2018 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/ijfe IntJ Fin Econ. 2019;24:508–526.

BAULEAND WILKE 509

abnormal returns by exploiting earnings-forecast momen-

tum effects.

In this paper, we propose a new and completely differ-

ent method for analysing the informational contribution of

financial analysts: Adapting econometric approaches from

market microstructure theory, we measure analysts' con-

tribution to the discovery process of the price, or value,

of a stock.1By implementing this method, we can not

only answer the question of whether analyst forecasts lead

stock prices or prices instead lead forecasts but also quan-

tify the extent to which such a lead–lag relationship exists.

To the best of our knowledge, this is the first paper that

directly determines the percentage share of value discov-

ery between stock prices and analyst forecasts, thus pro-

viding a quantitative measure for the information content

offered by analysts.

We focus on financial analysts working on the sell side

of the market. These analysts are typically employed in

research departments of banks and provide a broad range

of publicly available forecasts.2We measure the infor-

mation content of monthly analyst consensus forecasts

of earnings per share (EPS) as the central analyst out-

put. Assuming a simple relationship between future EPS

and stock value (in the spirit of Gordon, 1959, model),

changes in EPS estimates induce changes in the analysts'

view of the stock value. Stock market prices and such

forecast-based implied stock prices should cointegrate and

therefore share a common stochastic component. On the

basis of a vector error-correction model (VECM) and a

36-year sample of large U.S. companies listed in the S&P

100 Index, we compute Hasbrouck (1995) information

shares (ISs) and Gonzalo and Granger (1995) common

factor component (CFC) shares as measures for the contri-

bution to value discovery.

As we use monthly data, we look at stock-price move-

ments at a low frequency. It is therefore not the simple

speed at which information is reflected in stock prices,

but more importantly the interpretation of information in

terms of the long-term perspectives of the company. To

avoid confusion, we refer to this low-frequency process

as value discovery in contrast to the high-frequency price

discovery.

Our findings show that financial analysts working on

thesellsideofthemarketaremostlypureinformation

followers. However, they nonetheless participate in the

process of value discovery to a small extent. On aver-

age, sell-side analysts aggregate a percentage share of

4.6% (Hasbrouck measure) or 18.0% (Gonzalo and Granger

measure), respectively. These figures vary considerably in

the cross section, ranging from 0% to 33.3% (Hasbrouck) or

0 to 59.3% (Gonzalo and Granger). With the exception of

some periods lasting between 2 to 3 years, the small contri-

bution of analysts can be observed over the whole 36-year

time span.

We find that a company's mean level of analyst coverage

has a significant negative effect on the analysts' IS. This

finding can be explained by a herding argument: Individ-

ual analysts may shrink from opposing a strong consensus

forecast based on a large number of individual estimates.

Consequently, herding behaviour causes EPS forecasts to

be more biassed, and thus less informative in the case of

high coverage. Compared with consensus EPS forecasts

that are made up of only a small number of estimates,

high-coverage forecasts are thus less informative. Another

explanation lies in the covered company's incentives to

provide analysts with high-quality information: As ana-

lyst coverage is basically advantageous for a company, its

management has incentive to attract analysts by offer-

ing them high quality personalized company information

when coverage is low. These incentives vanishas the level

of coverage rises, whereas contrary incentives of limiting

the company's information costs increase. Therefore, com-

pared with analysts of low-coverage companies, analysts

of high-coverage companies will receive less personalized,

and more standardized company information.

Moreover, we observe a weak but (mostly) significant

negative effect of company age on analysts' IS, implying

that sell-side analysts' contribution to the process of value

discovery decreases with increasing company age. This

is consistent with sell-side analysts having advantages in

analysing firms that newly emerged—either from their

own or based on a merger. Here, analysts' informational

edges might result from younger firms and their funda-

mental prospects being more dynamic and less foreseeable

to the public.

The remainder of this article is organized as follows. In

Section 2, we present the theoretical foundations and a

review of the related literature. The econometrical frame-

work and the methodological approach are developed in

Section 3. In Section 4, we first describe our data set and

present some descriptive statistics. We then show our key

results and discuss their implications. Section 5 concludes.

2THE INFORMATION CONTENT

OF ANALYST FORECASTS

According to the EMH proposed by Fama (1970), stock

prices always fully reflect all information available to the

capital market. In his seminal article, Fama formulates

three forms of market efficiency that can be distinguished

by the information set that is available to investors: In the

EMH's weak form, stock prices incorporate only histori-

cal information. Therefore, in a weakly efficient market,

analyst output, which is fundamentally based on current