Analysts’ Dynamic Decisions: Timeliness versus Accuracy

• Author: Byungjin Kwak,Changhee Lee,Steven J. Jordan
• DOI: http://doi.org/10.1002/for.2438
• Date: 01 July 2017
• Published date: 01 July 2017

Analysts’Dynamic Decisions: Timeliness versus Accuracy

STEVEN J. JORDAN,

1

BYUNGJIN KWAK

2

*AND CHANGHEE LEE

3

1

College of Business, Alfaisal University, Riyadh, Kingdom of Saudi Arabia

2

KAIST Graduate School of Finance and Accounting, Korea Advanced Institute of Science and

Technology, Seoul, Korea

3

Anisfield School of Business, Ramapo College of New Jersey, Mahwah, NJ, USA

ABSTRACT

For analysts there is a tradeoff between the accuracy and the timeliness of their forecasts. Prior literature heavily

investigates analyst forecast accuracy. Few papers investigate the importance of timeliness. To our best knowledge,

there are no empirical papers to date to investigate the dynamic interplay between these key characteristics. We show

that if analysts experience a period of high accuracy relative to their peers, they subsequently focus more on the

timeliness of their forecasts in the subsequent period and thus issue their forecasts earlier than they did in the prior

period. Copyright © 2016 John Wiley & Sons, Ltd.

key words timeliness of analyst forecast; analyst characteristics; accuracy of analyst forecast

INTRODUCTION

For analysts making a forecast there is a tradeoff between the accuracy and the timeliness of their forecasts (S chipper,

1991). Compensation is based on the contribution an analyst makes to their brokerage house. An important element

used to evaluate contribution is the volume generated by the analyst’s forecast.

1

All else being equal, a timelier

forecast will generate higher volume (Schipper, 1991).

2,3

From an investor’s perspective, timeliness is not the only

objective. Accuracy of the forecast matters too, as inaccurate forecast can lead to investment losses. If an analyst

develops a reputation for inaccurate forecasts, investors will learn to ignore this analyst’s recommendations. Thus,

it is clear that analysts care not only for timeliness, but also for accuracy.

Prior literature heavily investigates analyst forecast accuracy. There are, however, fewer papers that investigate the

importance of timeliness.

4

Guttman (2010) states: ‘All else equal, analysts with a higher precision of initial private

information tend to forecast earlier, and analysts with higher learning ability tend to forecast later.’Since learning

ability of an analyst is a fixed personal characteristic (or at least changes slowly compared to our time step), it will have

little variation from one period to the next and thus little to no explanatory power.

5

However, the precision of an

analyst’s information about a firm can vary from period to period. Thus, in our data, prior period’s accuracy compared

to current period’s accuracy should be correlated to the change in an analyst’s quality of information. Theoretically,

Guttman shows that analysts with higher quality of private information issue their forecasts earlier compared to those

with lower-quality private information. If analysts believe that the current period’s precision of their private informa-

tion is positively correlated with the realized accuracy of forecasts in the prior period, they will issue their forecasts

earlier than others. One potential explanation is that, since analysts do not perfectly know the precision of their initial

private information, the realized accuracy of forecasts in the prior period is likely to affect their beliefs about the

precision of their private information.

6

Another potential explanation is that, when analysts have more resources, better

forecast models, or better relationships with managers, the accuracy of prior period forecast is likely to reinforce their

belief that they have better quality of information. Analysts with better resources, models, or relationships, who issued

accurate forecasts in the prior period, may believe that they have better quality of information in the current period

*Correspondence to: Byungjin Kwak, KAIST Graduate School of Finance and Accounting, Korea Advanced Institute of Science and

Technology, Seoul, Korea. E-mail: bjkwak@business.kaist.ac.kr

1

Cowen et al. (2006) documents that ‘brokerage firms usually reward their research analysts using a single measure of performance: trading vol-

ume in the stocks they cover’.

2

Beyer and Guttman (2011) provide a model that incorporates this behavior. In their model, a timely forecast is issued earlier, i.e. at a time the

precision of public information is relatively low. In their model, investors are risk averse; thus their demand for the firm’s shares is a function of

the residual uncertainty they face. Since an analyst’s forecast decreases the uncertainty, a timely forecast with a given precision generates higher

expected trading volume (since it occurs in a higher-uncertainty environment) than an equally precise but less timely forecast.

3

Jordan et al. (2014) add to Jackson’s work by investigating whether analyst forecasts generate trade volume. They find a positive and significant

relationship indicating that volume is at least in part driven by analysts’forecasts. First forecasts generate 1.7 times the volume than later fore-

casts; thus the timelier the forecast the more volume generated.

4

See, for example, the survey article by Ramnath et al. (2008).

5

An analyst fixed effect will capture all highly persistent explanatory variables, such as learning ability. Thus our regressions with analyst fixed

effects control for learning ability, allowing us to capture a purer measure of information quality.

6

We appreciate a reviewer for helping us to improve the argument.

Journal of Forecasting,J. Forecast. 36, 368–381 (2017)

Published online 1 September 2016 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/for.2438

Copyright © 2016 John Wiley & Sons, Ltd.