Past,Present and Future Compensation Research Perspectives

Question 1. What do you see as underresearched, yet important areas for work in compensation going forward?

Peter Bamberger: Altogether, compensation is underresearched. There were only 30 people at the most recent EIASM rewards/compensation conference. Four particular areas where more research is needed are the following: (1) the sharing economy and how this has changed the nature of pay; (2) benefits—an increasingly large part of total compensation, yet receiving little research attention; (3) the nature and structure of compensation and benefits in small to medium enterprises and startup firms and (4) the relative efficacy of merit pay versus bonus pay. Research in all four domains could be practically quite valuable.

Barry Gerhart: I’d like to see more work that considers intrinsic motivation in the context of compensation. We’ve been pushing back on work that discounts the importance of pay because it isn’t based on the evidence and has inappropriately applied the findings by Deci and colleagues. ⁴¹ So more proactively looking at the harmful outcomes that have been suggested with solid, empirical research would be important to this conversation.

Jason Shaw: For the most part, I think you can choose any area of compensation research and conclude that it is underresearched. Beyond what the others have suggested, I would like to see more research on issues related to timing and form of pay, as well as studies over time that can assess how long reactions tend to last and under what conditions they are more stable.

Question 2. Wage inequality is currently receiving a lot of attention in the United States; how do you see compensation research contributing to the conversation on wage inequality in the years to come?

Peter Bamberger: I think the structural and organizational aspects of pay that may exacerbate inequality should be considered. I’m also interested in seeing more work on transparency and its role in inequality and understanding the mechanisms through which pay practices effect inequality.

Barry Gerhart: We need to think about how we define inequality. Inequity, or inequality without good reason, is the real problem, not inequality. This appears to be an important conversation, and we need to think about our role in it.

Jason Shaw: This question is geared to the United States, but I see it as a global issue. Disparity and inequality issues are being discussed and debated around the world. The literature is fairly clear that the issue isn’t inequality, per se, but rather unexplained or indefensible inequality that is the problem. People are accepting of, and can even motivated by, inequalities that are linked to acceptable reasons for differences. We need a better understanding of how organizations get in the position of having a great deal of unexplainable variation in pay. Beyond this, we are frequently tied into pay dispersion types and issues, but we can also look at other forms of inequality.

Question 3. What other lines of work could or should inform our approaches? What theories could be promising to transfer to and apply in compensation research?

Peter Bamberger: I think both signaling theory and affective events theory would be valuable in the compensation research domain.

Barry Gerhart: I question if we need new theories in this domain—do they really help? We have a lot of good theories already.

Jason Shaw: I like the idea of being open to new ways of seeing our current issues with different theories. But there are certain things that existing theory explains rather well and we can always use these principles as a foundation for moving forward.

Question 1. Given the breadth of compensation as a research topic, what do you consider the most important compensation topics in micro research? In macro research?

Nina Gupta: Compensation research faces increasing challenges as a result of growing international, informational, technological and similar developments around the world. It is very tempting to pursue these challenges—they are there, they are interesting, few people have addressed them before and so on. But, in my opinion, the pursuit of these challenges should be accompanied by an equally fervent search for answers to issues that provide broad answers to fundamental problems that compensation researchers must also address. Dealing with nuances also means that we must first deal with basics. One of the most significant incentive issues that has continued to challenge compensation researchers for decades (if not centuries) is that of performance-based pay. Both at the macro and micro levels, it is imperative that we get a cleaner understanding of how performance-based pay works, the outcomes it affects and the complexities inherent in its design.

Micro researchers have addressed the challenges of performance-based pay for a long time, focusing both on the performance measures used and the ways these measures are tied to pay. We already know a lot about performance-based pay at the micro level. We know that objective measures of performance often suffer from criterion deficiency, so that large swaths of critical behaviors essential for good performance are ignored because good objective measures don’t exist. We know that subjective measures of performance, particularly supervisory ratings, suffer from criterion contamination, often reflecting more about supervisory biases than about differences in employee performance. We know that making minuscule performance-based differentiations in pay is likely to be demotivating. We know that the absence of widely shared pay information, that is, pay secrecy, can lead to widespread misperceptions about the actual pay system. The lessons from micro research are many and varied. But at the micro level, we still need to know a lot more about the dysfunctional consequences of poorly designed and poorly communicated performance-based pay systems. These deficiencies in micro-level performance-based pay systems allow consultants and motivational speakers to propagate the myth that financial incentives don’t work. We have much more work to do to demonstrate convincingly that they do work—that most of the problems we see with performance-based systems are problems of bad design and bad communication.

At the macro level, many of the lessons from the micro level are often ignored. Partly because of the need for summary measures, research tends to combine any and all forms of performance-based pay into one measure of whether an organization uses performance-based pay. This despite the fact that we know, from the micro research, that objective and subjective measures act differently, that individual-, group-, and organizational-level measures act differently or that different incentives (whether at the same level or at different levels) can work at cross-purposes with one another. Even when a single type of performance-based pay system is considered (e.g., merit pay), we are likely to treat a merit system that dispenses an average of 20% of base pay in merit raises as being equivalent to one that dispenses 2% of base pay in merit raises. In essence, we ignore many of the lessons from micro research when we look at performance-based pay at the macro level. A macro-level understanding of performance-based pay that incorporates these micro lessons is essential. We need to be more diligent in our approaches. Only then can we truly understand compensation phenomena at the macro level as well.

I do not mean to imply in the foregoing that all micro and all macro research on performance-based pay suffers from these problems. There is some great work being done. What I do hope is that we attend more carefully to the basics, both at the micro level and at the macro level.

Anthony Nyberg: I agree with Nina, and I would also add that there is a need to better incorporate time and sort out theory when bridging micro and macro compensation research. From a micro perspective, we need to do more to tease apart how individual incentives function within group settings, and we need to think more in terms of understanding how those individual incentives influence the aggregation of individual behaviors over time to influence higher order outcomes. From a macro perspective, we need more consideration of the mechanisms that drive individual behaviors over time (often referred to as microfoundations). Together, these suggest a deep need for developing meso-level compensation theory that incorporates time. For instance, we have solid theories about how incentives motivate individual behaviors, but know less about how incentives motivate group or team behaviors. Simply stated, to fully understand how compensation functions in an organization, we need to understand the causal mechanisms within and across levels, and the level that is currently most neglected is the meso-level.

Tae-Youn Park: I think we need to explore the role of social psychological factors more when studying the effects of compensation practices. Recently, my collaborators and I conducted a review of the pay dispersion literature, and we were able to find 382 correlations from 149 studies that examined the relationship between pay dispersion and its outcomes. It was an impressive number. We found that most of those studies examined the direct effect of pay dispersion. Several studies investigated the role of contingent factors, which were mostly human resource (HR) practice-related factors such as average pay level, the use of pay-for-performance, the degree of individual productivity differences and task interdependence. We found almost no studies that considered social psychological factors. However, I would be surprised if social psychological factors such as relationships among team members, organizational culture and manager’s managerial style do not have any influence on pay dispersion effects. Although I base this on the status of the pay dispersion literature, I think the status of other compensation literature is similar in that they often focus too much on HR practice-related factors and overlook the role of social psychological factors. This may partly be due to the difficulty in obtaining data, but I think it is still important for us to make efforts to theorize and empirically examine the role of social psychological factors when studying compensation issues.

Question 2. What are the biggest challenges of bridging micro and macro research on compensation?

Tae-Youn Park: We have good theories at the individual level—such as expectancy theory, equity theory and agency theory—but I often see that these theories are used to explain macro-level phenomena such as organizational performance. Perhaps this is because we do not have many good theories at the collective level. Although finding good macro-level compensation theories can be challenging, we still need to be very careful when applying micro theories because, as we know, a practice that is good for micro-level outcomes can be bad for macro-level outcomes, especially when employees are working highly interdependently. Thus, it is important to understand the similarities and differences between micro-level compensation phenomena and macro-level compensation phenomena and use the right theory that fits the level of analysis.

Anthony Nyberg: I see three major challenges to bridging micro and macro compensation research. First, we lack strong theory to help explain individual and organizational responses to incentives. Second, we lack clear understanding about what it means to combine different incentives within a person, people with different incentives in a group and groups with different incentives in an organization. For years, methods challenges complicated rigorous cross-level testing, but the increased use of tools for aggregating and the increased appreciation for examining these issues over time has led us to a time where we are now more advanced methodologically than theoretically. Third, if we are going to make progress in these areas, then we need reviewers and editors to work with authors to help bridge divergent areas rather than working against them by holding unrealistic standards. For instance, we would all do well when reviewing to help thoughtful authors to see additional paths and literatures while holding them accountable to their own literature’s criterion rather than enforcing our norms or railing against errors of omission.

Nina Gupta: Again, theoretical and empirical work in the micro area tends to have advanced much more than work in the macro area, partly because it has a longer history. Compensation research at the macro level often extrapolates individual-level theories and applies these theories at the organizational level. This practice raises issues of the validity and appropriateness of cross-level generalizations. This is what Tae-Youn and Anthony have been highlighting. I agree with them—theories need to be developed at different levels, and cross-level theories need to be developed.

At the empirical level, one of the most vexing problems is that of aggregation. We often take individual-level data and aggregate these data to the organizational level. Statistical techniques enable this work relatively easily, but they also obfuscate concerns about what specifically needs to be aggregated. They omit answers to many questions. For example, should data be aggregated across all employees? Should production employees be treated differently from managerial employees, since they are likely to have different attitudes and different reactions? Should data be aggregated at the group or meso level, or should data be aggregated at the organizational or macro level? Should data be aggregated across good and bad performers? Organizations often have different human research management approaches to different segments of the work force. Should data be aggregated across these segments or within each segment? Answers to questions like these are unlikely to be absolute. They probably depend on the exigencies of the particular research question, the particular situation, the particular employee groups, and so on, under consideration. But it is essential that researchers think about these answers explicitly before undertaking their research.

Question 3. What are some of the biggest mistakes you see people make when trying to using micro theories to inform macro research (and vice versa)? What do you recommend people think about early in their study design to prevent such problems?

Anthony Nyberg: We are often too narrowly focused on our own literatures, which allows us to remain ignorant of others’ views. Too often, authors from one discipline fail to spend the time or make the effort to examine alternative paradigms. One obvious answer is to expand ones reading list. Perhaps a more effective means for incorporating a broader set of perspectives is to work with more varied authors who have different backgrounds and perspectives. Admittedly, working with authors who have very different perspectives can be extremely challenging, and often extremely frustrating, but the end product can be extremely rewarding, particularly in terms of bridging conversations, and hence making meaningful changes to our science.

Tae-Youn Park: One of the mistakes I often see is that people overly focus on data analysis and the use of sophisticated statistical techniques and pay less attention to where and how the data are generated. The implications of compensation practices can be vastly different depending on the characteristics of tasks and jobs employees are working on. In particular, the consideration of job characteristics, such as interdependence and complexity, become very important when bridging micro and macro theories in the compensation literature. Thus, I think it is important to have a full understanding of the task and job characteristics of the sample when we analyze data, and also when possible we need to make efforts to explicitly measure those characteristics using a valid scale and consider their roles when analyzing data.

Nina Gupta: Obviously, as Anthony said, we need to be careful—we need to make sure that micro researchers have a good understanding of the relevant macro theories and perspectives, and that macro researchers have a good understanding of the relevant micro theories and perspectives. Without such understanding, we cannot develop unified paradigms that address compensation issues within and across levels. Such an understanding would also help address the issues I raised above—it would allow us to delineate the employees and groups that must be aggregated to address a specific research question.

In addition, I am troubled by the heavy reliance on existing data to answer compensation questions at the macro level. Micro research has typically used a variety of empirical approaches—surveys, laboratory experiments, simulations and field experiments, to name a few. Within the micro area, it is much more likely that the research design actually dovetails with the research question of interest. By contrast, a sizeable portion of macro compensation research relies on existing data sets obtained from settings where pay data are publicly available. The most common targets for study in macro compensation studies are sports teams and top management teams. This research is valuable. But our perspectives, particularly in macro research, must be broadened to include other samples, other settings and especially other empirical approaches. A significant constraint in using existing data is that key etiological dynamics may not have been measured properly (because the data were collected for purposes other than to answer a specific research question). If key variables are omitted from the research design, clear and comprehensive understanding of the relevant phenomena is impossible. We must have substantial macro research that asks key research questions first, and obtains the data to answer the questions with a targeted view to answering the questions.

Question 1. Compensation bridges many disciplines, including management, economics and psychology—how is publishing for outlets in these varied fields different?

Barry Gerhart: I think the main focus of what is studied differs by field. For example, psychology journals tend to focus on publishing studies that examine psychological mechanisms (“black box”) between compensation practices and individual/team outcomes. On the other hand, economists tend to focus more at the policy level and economics journals tend to focus on publishing studies that examine the outcomes of policies.

Michael Sturman: Each field has different priorities, and this makes the conversation between fields challenging. For example, in explaining a phenomenon, economists tend to prioritize establishing mathematical models while psychologists tend to prioritize examining perceptions. In terms of methods, economists tend to focus on precision in parameter estimation, while psychologists pay much more attention to issues of measurement. Management research seems to be stuck in the middle. It is very difficult bridging these two fields and thereby convincing reviewers in different fields.

Sanghee Park: The study topic that is frequently examined also differs by field. Psychologists frequently examine motivational issues such as subordinates’ intrinsic and extrinsic motivation, economists frequently study the topic of executive compensation and government policies and management scholars tend to study a wide range, including the topics from these two fields. However, other fields such as marketing, accounting/finance and sociology also study compensation. In the field of marketing, there are many studies that examine compensation issues of sales force. Accounting and finance scholars are more similar to economists in that they frequently use theories in economics and archival data. Last, sociologists are more macro-focused and frequently examine the issue of wage inequality and discrimination such as the gender pay gap.

Question 2. As a compensation reviewer, what are the biggest mistakes/problems you see in work submitted on the topic of compensation? What recommendations do you have for researchers to avoid these mistakes?

Michael Sturman: Fundamentally, a study needs to be a good study. There is no exception in the field of compensation. As in other fields, the research question should be interesting and contributing to the theory, research design should be well aligned with the research question, implementation should be sound, alternative explanations should be ruled out, measures should be good and so on. One problem I often do encounter specific to compensation research is the generalizability of the research when it is based on a micro experimental study. Some studies are often too context-specific to the experiment and cannot be applied to real-world compensation settings. Studies need to think more about generalizability and applicability in designing experiments. On the other hand, research with “real-world” compensation data can be quite messy. Pay gets adjusted in companies for many ad hoc reasons, and it often proves challenging to get good compensation data along with sufficient contextual information to rule out alternative explanations. Thus, for both experimental and field studies, researchers need to consider the implications of these difficulties, but editors and reviewers need to understand that compensation research has these particular kind of problems.

Sanghee Park: First, from a theory perspective, I have seen many articles that are unclear about conceptualizing constructs and use of theories. For example, compensation practices at individual, team and executive levels obviously differ and different theories should be utilized in developing models. Studies need to be clearer about the construct that they are examining and why a certain theory is applied in building a model that the study is testing. Second, from an empirical perspective, reviewers are often concerned with the lack of information regarding the study context. Although it is difficult to be too rich in context due to the confidentiality agreement often made with the compensation data provider, it is true that thorough presentation of the empirical data you studied should be presented to help us understand a complete context and to fully assess the validity of the research.

Michael Sturman: I strongly agree with Park’s comments on context. Study context needs to be described very specifically. Compensation systems are typically complex, and employees are often covered by multiple pay plans. Even sports pay data are highly complex, but I often see research articles that gloss over this complexity and make inappropriate assumptions about how well sports compensation data actually reflects performance or an individual’s market value. Basically, for all compensation studies using field data, we really need to understand the contexts in these studies.

Barry Gerhart: I have to agree with Sturman’s comments on fundamentals. Focus on the basics of what is a good study. Have many people read your article. Address common issues raised in friendly reviews. In terms of methodology, try to use more advanced methodologies (e.g., longitudinal or multi-level methods) that address common statistical issues. Although this is not a requirement (and theoretical issues are certainly more important), use of these advanced methods can increase the possibility of your paper being published.

Question 3. One of the most widely acknowledged challenges in our field is obtaining data. What have you learned about obtaining data? Any recommendations that could help new researchers?

Sanghee Park: Most firms are very concerned about maintaining confidentiality and compensation data are often confidential. Companies are very reluctant to provide these data. Another hurdle is a company’s own employee surveys. In particular, with a big corporation, they tend to have their own analytics team and conduct their own employee surveys periodically. They are concerned because they don’t want their employees to be overwhelmed with too many surveys.

Building a trustworthy relationship with practitioners is very important. I would suggest attending practitioner conferences. It can be a great starting point in building relationships with practitioners. Moreover, it is a great opportunity to know more about the concerns of companies. Companies are often restricted in budgets in getting professional consultants. You can also think about suggesting free consulting projects to these companies in exchange for data.

Michael Sturman: I would add the use of an alumni network. Building a trustworthy relationship is very gradual. The fact that you went to the same school with the alumni or that you are a faculty of the alumni’s school can speed-up the relationship-building process. In my experience, this is a highly underutilized networking opportunity for researchers.

Barry Gerhart: Other possible methods can be (1) sending out letters to companies, (2) working with consulting companies, (3) reaching out to former students and (4) conducting an experiment.