Past, Present and Future Compensation Research: Economist Perspectives

At the 2016 Academy of Management conference, a group of leading human resource scholars held a panel discussion assessing the state of compensation research, which was subsequently published in Compensation and Benefits Review [CBR]. Afterwards, CBR editor Charles Fay asked me to organize a panel discussion of economists on the same topic. This seems a good time to assess where we now stand in this field. Over the last several decades economists have produced an enormous literature on compensation, especially incentives. The 2016 Nobel Prize in Economics was shared by Bengt Holmstrom (with Oliver Hart). Bengt’s work provided a rigorous theoretical foundation for understanding incentives, and inspired much of the empirical literature on the topic in economics, accounting, and governance. Meanwhile, “Workforce Analytics” has emerged as a new tool that firms use to manage their employees. It uses statistical analysis of data on employees, policies, and outcomes—precisely what personnel economists have done for two decades in studying compensation and related issues. Given recognition by the Nobel Committee, and adoption of similar methods by practitioners, where does compensation research go from here?

The panel discussion was just published in CBR. I had the pleasure of having conversations about the state of compensation research with Kevin Hallock, Edward Lazear, Kevin J. Murphy, and Canice Prendergast. Collectively the panelists have made significant contributions to compensation research by extending the theory, collecting new and novel datasets, using new research methods, and expanding the range of topics studied. As tends to be the case with economists, much of the focus in our conversations was on incentives, but other topics were also raised.

There was general agreement that the theory of compensation and incentives is a significant success. This body of work has provided a rigorous, complex, and rich structure for thinking about pay, evaluation and incentives. Some of it is abstract, but much has been more practical, targeted at better understanding observed practices. The theory has been confronted with data for many years, and has been very successful in building our understanding of practice. That said, one or two noted that while the theory allows for great complexity in firm compensation practices, researchers sometimes use too-simplistic versions when analyzing data. As Kevin M. put it, empirical researchers need to “understand and embrace the richness of pay.” Canice also notes that a fundamental question—how strongly do people respond to incentives—has eluded a clear answer.

To the extent that theory should be refocused, there were several suggestions. Subjective performance evaluation is widespread (e.g., it is one of the most important functions of independent directors in monitoring a CEO), yet there is not very much theory on this practice. Risk is an issue that has been treated too simplistically. It is often treated as exogenous. However, employees can often manage risks, because performing the job provides them with knowledge about their uncertain environment, so they can plan for and react to uncertain events. (Note that accounting scholars sometimes refer to “controllable risks.”) Another neglected topic is intrinsic motivation. Some panelists emphasized that employees are motivated by more than compensation, and incentives might interact with other forms of motivation (such as research creativity). This has long been an area of interest in the field of organizational behavior, of course, but not in economics.

Even though some important theoretical issues have been neglected, a primary theme that emerged is that compensation research should focus less on theory, and more on pushing the boundaries of existing empirical work. As Canice put it, “we are very limited by the absence of good empirical facts.” Examples of topics for which we have inadequate evidence include total pay (analyzing the entire employment package, not just salary, bonus and equity pay); the role of political pressures in constraining CEO compensation; why compensation practices are very simple compared to the complexity suggested by theory; and how pay relates to recruitment and career mobility.

What types of data might be collected to enrich compensation research? First, outside of executive compensation, we rarely have access to financial or accounting data. In fact, information on all types of employee outputs (financial, productivity, quality, service, etc.) would be extremely interesting. Even less common is information on employee “effort” (hours worked, intensity, diligence, collaboration, etc.). A third broad category is detailed information on compensation policies and practices. How important are subjective evaluations compared to numeric metrics? How are evaluations conducted? What bonus formulas are used? How are options granted? Along the same line, information on other firm characteristics and policies would improve compensation research, including recruiting and selection methods, training, organizational structures, group composition, and social networks. In the area of governance, information on CEO wealth portfolios would be extremely interesting. Finally, greater use of qualitative information (possibly scanned from text), such as the feedback on employee evaluation forms, would expand the range of empirical work.

Finally, there was some discussion of research methods. Over time, social science research has moved away from cross-sectional and descriptive analyses, towards emphasizing causal inference. Several panelists expressed concern that “causality police” make it difficult to publish interesting new evidence, by giving too much emphasis to econometric methods aimed at causality. Field experiments are a positive development as causality is more clearly established without obscure econometrics. Unfortunately, few firms are willing to run experiments with their workforces. Eddie suggests that the field would benefit from more “common sense and not cute strategies or overly complex models.”

Summing up, compensation research in economics has been a great success story. However, there are many areas for further expansion and improvement in the field. Much of this promise comes from the explosion of new types of data. Firms have realized that they have rich data which can be used to analyze how they manage their workforces. Personnel economists have been doing this for a long time, but the new interest by firms is likely to make it easier for academic researchers to expand our “data entrepreneurship.” That will allow us to study more interesting and relevant questions, and deepen our understanding of more familiar questions. Two final notes. First, compensation research spans many academic disciplines, but is not very interdisciplinary. Perhaps more cross-field research would also be valuable. Second, how well are we doing in making our research accessible to and useable by practitioners?

The complete piece is available for download here.