Market demand for conservative analysts

Abstract

Sell-side analysts, on balance, have incentives to emphasize good company news and downplay the bad, resulting in inefficient forecasts. We conjecture that this behavior generates a demand for forecasts from conservative analysts who unwind this pattern, at least in part, resulting in more efficient forecasts. To investigate, we introduce a measure of analyst conservatism and assess the market reaction to analysts’ forecast revisions conditioned on their past levels of conservatism. We find a stronger market reaction to forecast revisions by more conservative analysts, and that this result is heightened for companies with greater institutional investor following.

Introduction

As information intermediaries, sell-side analysts gather and process public and private company information for the benefit of client investors who lack the resources to conduct such analyses on their own. The research of analysts, therefore, is geared toward first creating an information asymmetry between analysts and client investors, and then, ideally, reducing this asymmetry through systematic and objective disclosures. However, analysts, on balance, tend to emphasize good and downplay bad company news, resulting in inefficient forecasts (Easterwood and Nutt, 1999). While many incentives documented in prior work are consistent with the norm of aggressive research (Hong and Kubik, 2003; D’Avilio et al., 2002; Francis and Philbrick, 1993, Das et al., 1998; among others), we argue that some analysts will forecast conservatively to garner influence with equity investors, who should value more efficient earnings research.¹ Accordingly, we conjecture that equity investors in general—and institutional investors in particular—will respond more strongly to research from conservative analysts, who generate more informative forecasts by unwinding, at least in part, the aggressive research practices representative of analysts more generally.²

Since it is not clear if analysts act conservatively in an absolute sense, we introduce a conservatism measure that is conditional (based on an individual analyst's reactions to news direction, bad versus good) and relative (based on an individual analyst's reactions relative to those of peer analysts).³ This measure is formed each calendar year by ranking, among the analyst population, the average asymmetry in an individual analyst's forecast revisions to bad versus good news; analysts making stronger revisions in response to bad news versus good news relative to their peers are considered more conservative.⁴ After this modeling, we examine how analyst characteristics vary with the conservatism measure. We find that more conservative analysts are more likely to be Institutional Investor award winners, work for larger investment houses, have more experience, and yield forecasts that are more accurate and more persistent, i.e., more predictive of future earnings revisions. Collectively, these findings suggest that more conservative analysts are more able, have better resources, and provide more efficient forecasts than less conservative analysts.

In order to examine the market response to conservative analysts’ earnings research, we assess the short-window market responses to analysts’ forecast revisions in the testing period (i.e., current year) conditioned on their level of conservatism established during the estimation period (i.e., prior year). We find a stronger market response to forecast revisions of conservative analysts; the response is 22% larger for the revisions of the top versus the bottom conservatism quintile. Extending the analysis to the setting of institutional ownership, we conjecture and find that institutions’ fiduciary duties coupled with their greater ability to discern efficient earnings research result in a relatively stronger market response to conservative analysts’ forecasts.

To further the interpretation of our results, we conduct two additional analyses. The first tests a prediction of Bayesian investor learning that analysts develop a more precise reputation for conservatism as their estimation period lengthens, thereby allowing the market to observe a greater number of forecasts (Chen et al., 2005). We find that the market reaction to conservative analysts’ forecasts monotonically increases as the estimation period is set longer, from 1 to 4 years. The second analysis investigates whether the market response to our measure of analyst conservatism aggregates to the brokerage level; we find stronger market responses to revisions of analysts employed by more conservative brokers.⁵

Several efforts toward enhancing the reliability of our measure and robustness of our findings are made. Among them, we verify that the conservatism measure is not subsumed by alternative constructs such as brokerage size, Institutional Investor award status, past forecast accuracy, or past bias. Also, altering design choices, such as varying the requirements of our conservatism measure, or substituting a stock returns-based news proxy for our analyst-based news proxy, yield inferentially similar results.

Our findings offer several insights to the empirical literature on the economics of financial intermediation. First, we document a significant cross-sectional variation in individual analysts’ asymmetric response to bad versus good news—which by definition forms the degree of analysts’ conservatism—and show a stronger market response to more conservative analysts’ earnings research. This response is rational in the sense that earnings forecasts of more conservative analysts are ex post more accurate and more informative about longer-term earnings revisions. While prior literature has extensively documented the benefits of producing aggressive research, our findings of greater price impact for conservative analysts’ earnings research suggest the existence of an alternative incentive—market influence—to produce conservative research.

Second, we find a stronger market response to conservative analysts’ earnings research in the presence of greater institutional ownership. Institutional investors shape analysts’ careers in the sense that they vote for the best analysts in the annual Institutional Investor survey (Stickel, 1992). In addition, institutions pay for research either directly or indirectly, when they allocate their trading across brokerage firms, thereby generating trading commissions for the analysts’ employers (Ljungqvist et al., 2007). Our empirical results are consistent with more conservative analysts’ research having enhanced credibility in this important setting, as well as institutional investors’ ability to discern efficient earnings research.

We organize our paper as follows. The next section describes the related literature and develops our hypotheses. Section 3 describes our conservatism measure and specifies empirical tests, and Section 4 explains our sample selection and provides results. Section 5 provides concluding remarks.