Academic pay in the United Kingdom and the United States: the differential returns to productivity and the lifetime earnings gap.

Terrell, Dek

1. Introduction

The product market for academic research is truly international. Research produced by scholars in one country is "consumed" by research scholars in another through networks that afford almost instantaneous transmission of research output. These networks are easily accessed and include formal structures such as the presentation of research output at conferences and department seminars and the publication of research output in peer-reviewed journals. Transmission of research output also takes place through less formal arrangements such as the circulation of working papers by individual faculty members. Research can also be jointly produced by two or more scholars residing at universities located in two or more countries. Technological advances in the field of telecommunications have significantly reduced the costs of engaging in such joint production. Additionally, many scholars spend time visiting at universities in other countries where they can initiate and conduct research projects with their resident colleagues.

However, despite the fact that the "product" market for academic research is international, academic labor markets for research scholars appear to remain separated. These markets are separated not only by distance but also by customs, traditions, and institutional factors that may lead to earnings structures that differ substantially across these separated labor markets. Some of these institutional differences have fairly obvious manifestations, such as, for example, the difference in professorial rank structures in the United States and Canada on the one hand, and the United Kingdom and Australia on the other. But there may be other less tangible differences that separate two academic markets, such as concerns for equity or fairness. While difficult to document, equalitarian considerations may be fairly important in the determination of earnings in one country and less so in another.

In the present paper, we examine differences in the earnings structures between academic economists in the United States and the United Kingdom. It is widely recognized that academic salaries in the United States are higher than in the United Kingdom. Further, since faculty members in the United Kingdom below the rank of professor are paid according to a national pay scale (discussed below), it is presumed that research productivity differentials will not be fully reflected in the observed salary differentials for lecturers and senior lecturers. To the extent that returns to research productivity in the United Kingdom are low, we should observe flatter earnings profiles and less earnings dispersion across experience levels relative to the U.S. labor market. But at least part of the observed earnings differentials between the United States and the United Kingdom may reflect a more pronounced concern for equity in the United Kingdom. One might reasonably infer such equity concerns from an examination of the U.K. national salary schedules, which have no explicit discipline-based pay differentials. So, at least nominally, lecturers in Economics earn no more than their counterparts in English. (1)

The separation in international labor markets is not unique to academia, but the study of compensation issues in academic labor markets has one distinct advantage. Academic labor markets allow us to examine the relationship between experience and earnings in an environment in which measuring individual productivity is less problematic than it is for workers in a cross section of private firms. While salary information is readily available from published public university budgets in the United States, an individual faculty member's salary in the United Kingdom is confidential. We were fortunate, however, to obtain salary information from academic economists through a national survey. Salary data were linked to demographic and productivity data taken directly from each individual's curriculum vitae. Our data set enables us to thoroughly explore salary differentials between the United Kingdom and the United States since it permits more precise measures of past and current relative productivity of individuals engaged in comparable work. Furthermore, by concentrating on researchers in one field--economics--it is possible to compare the relative earnings and relative productivity of individuals in fairly homogeneous jobs, in the same international product market, but who happen to work in separated labor markets.

We employ a standard human capital earnings model to explore differential reward structures in academia between the United States and the United Kingdom. Despite the major institutional differences in compensation schemes, a comparably specified human capital model does a good job explaining earnings variations for academic economist in both countries. Our estimates suggest that rewards for research are more immediate and direct in the United States. Because of the national salary scale, the payoffs to experience and seniority are greater and the payoffs to research are lower in the United Kingdom than in the United States. After adjusting for productivity and demographic factors, we find that U.S. economists are paid approximately 40% more than otherwise equivalent economists in the United Kingdom. Simulating career age-earnings profiles for both markets, we find that the earnings gap widens with experience for relatively productive research economists and may even narrow with age for relatively less productive research economists. Nevertheless, the cumulative lifetime earnings foregone are substantial for both sets of United Kingdom economists.

2. Compensation Schemes in the United Kingdom and United States

Compensation schemes differ between the United Kingdom and the United States in a number of important ways. Academic labor markets in the United States are highly competitive, and the rewards provided to successful researchers are quite substantial. In the United States, universities are funded through tuition, private donations, and state support for public universities. The federal government plays a minor role in the direct financing of higher education. Individual universities differ substantially in terms of their scope, mission, quality, and pay structures. Universities must, however, compete in a national market for research scholars, particularly top scholars.

In the United Kingdom, the central government provides the funding for higher education, and most faculty members are paid under a fixed national salary scale that is negotiated by the Association of University Teachers (AUT) and the University and College's Employer Association. (2) Also, the salary scale sets a minimum and a maximum and annual increments for all ranks below the rank of professor. (3) A minimum is set on professors' salaries at a level above the top pay for reader/senior lecturers, and individual professors' salaries are based on negotiations between each university's personnel department and the individual. Discretionary salary increments may be awarded for exceptional productivity, but in practice these rewards are rare (Ward 1998; Bert 1999).

The national salary scale and uniform automatic pay increases reduce university competition to hire top faculty and lower incentives for individual faculty to produce or to move. (4) Further, Williams, Blackstone, and Metcalf (1974) speculate that central funding in conjunction with the national salary schedules and accompanying regulations may facilitate collusion among the universities acting as agents for the government and that this monopsony situation results in lower salaries for faculty.

Both countries base promotions at least in part on research productivity, and promotions are associated with pay increases. Among major universities in the United States, individuals are promoted and rewarded on the basis of their research productivity (Katz 1973; McDowell, Singell, and Ziliak 1999). Each promotion is accompanied by a one-time increase in permanent pay, but future increments depend on future productivity. In the United Kingdom, promotions allow the individual to move from one pay scale to a higher one with automatic future pay raises. Prior to the 1980s, promotion to senior lecturer was based on longevity and teaching quality, promotion to reader was based on longevity and research quality, and promotion to professor required an outstanding research record, a national reputation, and administrative skills (Williams, Blackstone, and Metcalf 1974). For many years, the rank of professor in U.K. universities was reserved for a very select group. Often there was only one professional chair per department, and that person served as head of the department. To hold costs down, limits were placed on the proportion of faculty in the senior ranks (reader/senior lecturer and professor) at each university. While these practices are no longer followed, the proportion of professors remains much lower in the United Kingdom than in the United States. In our current samples 38% of U.K. economists are professors compared to 54% for the United States.

3. The Samples

Because salary information in the United Kingdom is confidential, there have been relatively few studies of academic pay in the United Kingdom (Bowen 1963; Metcalf 1970; Ward 1998; Blackaby and Frank 2000), and none of the existing studies examine the relationship between pay and research productivity. Our U.K. sample was obtained from email requests sent to approximately 1000 academic economists at 60 universities whose addresses were available on the internet. Individuals were asked to supply a copy of their current curriculum vitae and their academic salary. Complete information was obtained from 126 individuals. Given the low response rate and the relatively small sample, we have two concerns. First, the small sample may cause the regression estimates to be sensitive to specification. We explicitly address this issue by estimating a number of specifications. We find the qualitative results to be insensitive to specification.

Also, with a response rate of 13%, we are concerned with how representative our sample is of the general population of academic economists in the United Kingdom. While not random, the composition of our sample appears to differ only slightly from samples used in a number of other U.K. studies. Table 1 compares the composition of our sample with those collected in recent surveys by Blackaby and Frank (2000); Booth, Burton, and Mumford (2000); Machin and Oswald (2000); and Higher Education Funding Council for England (HEFCE 1997). Our sample has a higher proportion of professors and a lower proportion of lecturers than other samples. We believe the lower proportion of lecturers may be due to their belief that their salaries are determined by the national salary scale and, therefore, are unrelated to productivity. Professors may have responded at a higher rate because their salaries are flexible and affected by productivity and because they have better academic records. Unfortunately, no females at the professorial rank responded to our survey.

Table 1 also shows the distribution of our sample by department quality. As a measure of department quality, we use Research Assessment Exercise (RAE) scores. Since 1986, the United Kingdom has been using peer-review panels to evaluate and rank academic departments in each discipline. Peer-review panels in each of the periodic research assessment exercises make assessments about the quality of research produced within each department. A categorical rank is then assigned to every department, which in 1996 ranged from 1 to [5.sup.+], with [5.sup.+] being the highest rank. Compared to other studies, our sample has a higher proportion of respondents from level 4 programs and a lower proportion from level 1 and 2 departments. The level 4 response may be attributed in part to the fact that one of us (W.J.M.) was a visiting scholar at a level 4 department in the fall 1999 and presented seminars related to this research topic at two level 4 departments. The low response rate for RAE level 1 and 2 departments was caused, in part, by the lack of email addresses for faculty at these schools. But, by in large, departments at the 1 and 2 levels do not have strong research agendas.

Compared to other studies, there appears to be an underrepresentation of faculty at RAE [5.sup.+] schools. However, for purposes of comparative analysis, this is not a major problem because there are no individuals from top 25 departments in our U.S. sample. According to data presented by Kalaitzidakis, Mamuneas, and Stengos (1999), the London School of Economics is the only U.K. school that would rank among the top 25 U.S. economics departments. While our U.K. sample does not appear to be substantially different than those of other studies shown in Table 1, it is not a random sample. Consequently, we are cautious in our attempts to generalize the results presented below. However, to our knowledge, this study represents the first attempt to directly compare life-cycle implications of the salary structures in the United States and the United Kingdom within a consistent econometric framework.

We employ a U.S. sample used in an earlier study. The sample consists of all faculty members in nine economics departments, each with a Ph.D. program and a strong research focus. The nine departments were selected through personal contacts and the willingness of their faculty to participate in the study. These programs ranked from approximately 30th to 70th in the Scott and Mitias (1996) study. Because our sample excludes the top 30 programs, the results may not generalize to the entire U.S. market, but the sample does match the quality level of programs where most of the individuals in our U.K. sample are employed. We would, of course, have preferred a large random sample from both countries drawn from the populations of academic economists in both countries. Like most other empirical studies, we proceed with these limitations in mind.

Both samples include salaries for the academic year. In addition, U.K. wages for 19 professors include a 1500 [pounds sterling] London pay adjustment. The primary results of the paper were not impacted by the exclusion of the London pay premium.

4. The Empirical Model

The dependent variable in all the reported regressions is the log of the 1999-2000 academic year salary. (5) In each sample, we exclude individuals who, in the sample year, occupied an administrative position above the level of department chairperson. With only one exception, the list of independent variables is identical to those used by Moore, Newman, and Turnbull (1998) in an earlier study. Experience measures the years of academic employment subsequent to the date of the highest degree earned. Seniority is defined as the number of years employed at the current institution. Level 1 Articles measures the number of articles published in the top-tier general interest economics journals: American Economic Review, Econometrica, Economic Journal, Economica, International Economic Review, Journal of Economic Theory, Journal of Political Economy, Quarterly Journal of Economics, Review of Economic Studies, and the Review of Economics and Statistics.

Level 2 Articles counts the number of articles published in highly respected second-tier general interest journals and in the leading field journals. We expand the list of 45 journals used by Moore, Newman, and Turnbull (1998), adding five European journals (see Appendix). Specifically, we added the European Economic Review, Journal of Royal Statistical Society, Oxford Bulletin of Economics and Statistics, Scandinavian Journal of Economics, and Scottish Journal of Political Economy. (6) Including or deleting these five journals had no substantial impact on our results. The last publication variable, Total Other Pubs, includes all of the other publications listed on an individual's curriculum vitae. This broad publication variable includes textbooks, academic books, comments, notes, and other miscellaneous documents that have been published, but it excludes working papers and book reviews. (7) In the U.S. sample, the total publication variable differs slightly since it includes all books and refereed journal articles published and excludes comments and replies. However, there are very few comments and replies in both samples.

Chair Years is used to proxy an individual's investments in administrative skills. This variable is defined as the number of years the individual has served as chair, either currently or in the past. Ph.D. Quality is a dummy variable having a value of one if the individual received his Ph.D. degree from one of the top programs in economics. (8)

To capture the effects of promotion on salary, we included two rank variables in the model. Rank II has a value of one if the individual is an associate professor in the United States or a reader/senior lecturer in the United Kingdom and zero otherwise. Rank III has a value of one if the individual holds the rank of professor in the United States or United Kingdom. We anticipate that the rank effects will be greater in the United Kingdom because of the automatic pay progression in the higher ranks under the national salary schedules. Also, this specification provides additional information on within-rank earnings differentials and helps us determine what fraction of the higher earnings associated with experience occurs within ranks as opposed to more experienced faculty with higher ranks earning higher than average salaries.

Finally, we include a Gender variable having a value of one for males and zero for females. A number of U.S. studies of academic labor markets have reported a statistically significant salary differential that favors males (e.g., Johnson and Stafford 1974). Recently Blackaby and Frank (2000) estimated an earnings equation for a national sample of U.K. academic economists. They reported an earnings gap of 9.1% for married women and a 14.2% gap for unmarried women.

Table 2 presents the sample means for the variables discussed above. In our sample, academic economists in the United States were paid 49.8% more than U.K. economists in 1999. Part of this differential may be attributed to the fact that the U.S. sample contains a larger percentage of professors (Rank III) and a larger fraction with Ph.D. degrees from the top programs. In addition, U.S. economists in our sample have published more level 1 and level 2 articles. The subsequent analysis examines the wage differential that remains after controlling for these observed productivity differences.

5. Regression Results

Table 3 presents the OLS regressions for each country. The first two columns contain results for the baseline model, which includes rank variables. Because rank may be endogenous, in the last two columns we present results excluding rank. Focus first on the baseline model. While the two equations are fairly similar in terms of explanatory power, there are important differences between the two countries in the influence of some of the income-producing characteristics. (9) First, controlling individual productivity, we find Experience has no significant effect on salary differentials within the United States. In contrast, the experience coefficients remain statistically significant in the U.K. equations despite controls for individual productivity. We attribute this result to the automatic pay raises for lecturers and reader/ senior lecturers that are generated under the U.K. national salary schedules.

Second, it is interesting to note the differential effects of seniority between the two markets. While seniority has no observable effect on earnings differentials in the United Kingdom, it appears to have a significantly negative effect on relative earnings in the United States. The negative effect of seniority on relative earnings in U.S. academic labor market has been noted in previous studies. (10) These results, however, are puzzling in light of the robust positively sloped seniority-earnings profiles typically found in nonacademic markets. The interpretation of the negative seniority-earnings profile in academic labor markets is still an open question. Ransom (1993) argues that senior less mobile faculty members are monopsonized by universities. Moore, Newman, and Turnbull (1998) find that the negative seniority effect on earnings diminishes and eventually disappears as more comprehensive measures of productivity are included in the earnings equation. (11) Bratsberg, Ragan, and Warren (2003) argue that previous research failed to account for the quality of the job match. Accounting for the quality of the job match, they find that faculty pay falls significantly with seniority.

The absence of a significant seniority effect in the U.K. market is not surprising. Under the negotiated national salary schedules, which cover lecturers and readers/senior lecturers, new hires must be paid less than the current faculty, so salary compression is not an empirically important phenomenon in the U.K. system. It is also unlikely that monopsonization of older less mobile faculty would occur under a negotiated national salary schedule. Research in the United States indicates that unionization in higher education increases the salaries of senior faculty more than it increases the salaries of junior faculty (Brown and Stone 1977; Barbezat 1989).

Holding observed research output constant, we find that Ph.D. Quality increases salaries approximately 4.3% in the United States and 6.9% in the United Kingdom. (12) We interpret this to reflect either a halo effect or an unobserved quality differential. We expect, however, that the halo effect would dissipate over time, as an individual's quality becomes known. When our model is estimated separately for each rank, the Ph.D. Quality variable is significant only for assistant professors and lecturers. (13)

Turning to the publication variables, we find that quality research is highly rewarded in both countries. As expected, the marginal return to a paper in a top-tier journal is significantly greater than the marginal return to a publication elsewhere. Publishing a level 1 paper increases an individual's relative salary by 2.5% in the United States and by 1.1% in the United Kingdom. Publishing a level 2 paper raises an individual's relative salary another 0.7% in the United States but has no significant effect in the United Kingdom. We hesitate, however, to conclude that publications in level 2 journals do not significantly affect an individual's relative earnings. As we discuss below, when rank is deleted from the model, a publication in a level 2 journal has a significant positive effect on relative earnings in the United Kingdom. The relative magnitude of the coefficient on Level 2 Articles in the U.K. and the U.S. salary regressions remains the same whether the rank variables are included or not.

We find that publishing in other outlets (Total Other Pubs) has a small positive impact on salaries. We also tested hypotheses regarding the presence of diminishing returns to publications and the possible effect on earnings of the age distribution of an individual's stock of academic publications. (14)

Our earnings models also reveal a positive and significant return to experience as chairperson in both countries. The annual payoff is 1.9% in the United States and 1.4% in the United Kingdom. Evaluated at the sample means, the average chair payoff is approximately 16% in the United States and 14% in the United Kingdom. It is not clear what kind of productivity effects are being measured by this variable. It is possible that the chair premium reflects positive rewards to administrative skills, or compensation necessary to offset the opportunity cost of forgone research time and the atrophy of research skills while in office, or pure compensation for serving in that role. (15) Also some of the skills necessary for being selected chair may be highly correlated with other skills valued by the university, skills that are not included in our model. (16)

Finally, the rank variables in the baseline model could provide additional information on within-rank earnings differentials and help determine what fraction of the higher earnings associated with experience occurs within ranks as opposed to more experienced faculty with higher ranks earning higher than average salaries. The Rank H and Rank III variables have positive and significant marginal effects on relative salaries in both countries holding constant productivity differentials. Relative to Rank I, the Rank H premium is 13% in the United States and 14% in the United Kingdom. The Rank III premium relative to Rank I is about 38% in the United States and 34% in the United Kingdom. However, the rank premiums appear quite similar in magnitude for the United States and United Kingdom.

We recognize that rank may be endogenous and that it is likely to be influenced by many of the other explanatory variables included in our salary regressions. Therefore, we estimated the U.S. and U.K. salary regressions with Rank H and Rank III variables deleted. The results without rank are reported in the last two columns of Table 3. With the exception of other publications in the U.S. equation, the statistical significance and the magnitude of the estimated coefficients increased for all of the explanatory variables in both the U.S. and U.K. salary regressions. Not surprisingly, the experience coefficients rise when rank is omitted. However, the relative magnitude of other coefficients was unchanged by the deletion of the rank variables. In the following counterfactual exercise we present lifetime earnings gaps for specifications with and without rank.

6. The Counterfactual

The earnings models suggest that while academic salary levels in the United States and the United Kingdom are determined largely by the same factors, the marginal returns to some income-producing characteristics differ between the two markets. In addition, as we stated earlier, it is widely recognized that academic salaries in the United States are higher than in the United Kingdom. Another way to frame the question is to ask what an otherwise equivalent faculty member in the United Kingdom would earn if those skills were transported to the U.S. academic labor market. The models estimated above can be used to calculate the earnings for a typical faculty member in the United Kingdom with characteristic vector [X.sup.UK] but compensated according to vector [[beta].sup.US]. That is, the counterfactual for a typical British economist offering his bundle of skills in the United States can be calculated in the following way: [([summation][[bar.X].sup.UK][[beta].sup.US]) + [[beta].sup.US.sub.0]]. (17) Our representative U.K. economist would have a predicted U.S. salary of $70,690 as compared to an actual U.K. salary of $50,433 in 1999. That is, the average British economist would be paid about 40% more in the United States. By comparison, the World Bank (2003) reports that U.S. manufacturing labor costs exceeded that of the United Kingdom by 21% for the period 1995-1999.

Another way to view this differential is to simulate a synthetic age-earnings profile for both U.S. and U.K. economists. Consider two economists, one a very productive scholar in terms of publications and the other less successful. For simplicity we assume both are male, that neither scholar moves over a 30-year career. We also assume neither individual ever serves as department chair. In both countries we simulate career earnings profiles for two academic economists at opposite ends of the research productivity distribution.

First consider a relatively productive economist who generates a strong publication record and graduated from a top Ph.D. program. Specifically, this means publishing a level 1 article once every two years, one level 2 article, and one "other" publication each year. (18) Additionally, this relatively productive scholar receives his first promotion in year 6 and a second promotion in year 12.

Our less productive scholar never publishes a level 1 or level 2 journal article, but does publish one article per year in lower tier journals. This less productive scholar is also promoted in year 6 but is never promoted to professor.

Using the synthetic age-earnings profiles, Figures 1 and 2 contain the difference between U.S. and U.K. earnings for these two economists. Figure 1 presents the annual U.S.-U.K. earnings gap for our highly productive scholar. Figure 2 presents the corresponding annual gap for our less productive scholar. Both U.K. scholars would begin their careers earning roughly $18,000 more in the United States than in the United Kingdom. But, note that the gap widens over time for the economist with a highly productive research record. After 30 years, the U.S. salary is $58,508 higher than the U.K. salary. For the less productive scholar, the U.S.-U.K. salary gap falls over time to $6366 after a 30-year career.

[FIGURES 1-2 OMITTED]

We also performed the same counterfactual using models that did not include rank. For our less productive scholar, the earnings gap profile calculated from the model without rank scholar remains relatively flat throughout his career, staying within the $25,000 to $30,000 range. Qualitatively, the results for our highly productive scholar are unchanged--the gap continues to rise monotonically. However, with rank deleted from the model, our estimated profile lacks the discrete jump associated with promotion to full professor. Also, our highly productive scholar's salary would be about $25,000 initially and grows to a difference of $90,000 higher in the United States than in the United Kingdom over 30 years.

With regard to this exercise, several caveats deserve mention. All U.S. faculty members are associated with Ph.D. programs, while some U.K. faculty are not. Likewise, other forms of selection bias may not be captured in the model. These biases may lead to an overstatement of the premium offered in the U.S. market. One such bias may arise from our comparison of U.S. and U.K. faculty pay/productivity structures at two different points in time. To the extent that returns to productivity have changed over time, our estimates of the premium in the United States may be biased upward. For example, with the introduction and increased importance of the research assessment exercises in the U.K., research productivity has become a more important influence in salary determination. This change in research culture may suggest an overstatement of the expected earnings gap between the two labor markets for future cohorts.

However, regardless of specification, these simulated profiles highlight a substantial difference in the U.S.-U.K. earnings gap for academic economists. At the end of a 30-year career in the United Kingdom, the more productive scholar gives up just over $1 million in lifetime earnings to remain in the United Kingdom, while the less productive scholar forgoes about $323,000 in cumulative earnings to work in the United Kingdom.

7. Conclusions

Despite significant differences in compensation schemes in the United Kingdom and the United States, our human capital model does a good job explaining the variation in academic salaries for economists within each country. Differences in the compensation schemes and market forces, however, cause the timing of the rewards and the relative payoffs to human capital variables and types of publications to vary between the United States and the United Kingdom.

Our analysis suggests that academic economists in the United States are paid approximately 40% more than their counterparts in the United Kingdom, adjusting for human capital, productivity, and demographic factors. To illustrate the lifetime earnings gap implied by our models, we simulated age-earnings profiles for two hypothetical academic economists. These profiles highlight the significant cumulative earnings gap between the United States and United Kingdom, viewed over a 30-year career. We find that the earnings gap between the United Kingdom and the United States widens with experience for relatively productive research economists. On the other hand, the gap may narrow over time for relatively less productive research scholars. Nevertheless, the cumulative foregone earnings are substantial.

The U.K. earnings structure for academic economists may be viewed as more egalitarian than the United States. In the United Kingdom evidence suggests a lower earnings spread between more productive and less productive research economists, holding experience constant. It is also reflected in the United Kingdom's national pay schedule, where there is an apparent absence of salary differentials between disciplines or across productivity profiles within disciplines. Apparently, culture does matter.

Given the limitations of our data, these conclusions must be viewed as tentative. We offer a couple of caveats. First, both samples of academic economists are small and nonrandom. We believe, however, that they are adequate for identifying the major determinants of salary differentials within and between the two countries. Second, we only have data on academic salaries. There may be important differences in fringe benefits, indirect subsidies (e.g., housing allowances), and other nonpecuniary compensations (i.e., teaching loads), which may vary between the two countries. Finally, our measures of productivity are not complete. While we identify the quantity and quality of research activity in some detail, our measure of service productivity includes only administrative service (Chair Years). Since we do not have data on teaching performance, we are unable to measure, and therefore capture, the effects of teaching productivity.

We dedicate this paper to the memory of William "Jeff" Moore, our coauthor, colleague, and friend who passed away prior to publication of this paper. Special thanks to Andrew Oswald, Jeff Frank, Peter Sloane, Daniel Hamermesh, Julie Hotchkiss, and two anonymous referees for their helpful comments and suggestions.

Appendix: Level 2 Journals

1. American Economics Associate Papers and Proceedings

2. American Journal of Agricultural Economics

3. Brookings Papers on Economic. Activity

4. Canadian Journal of Economics

5. Econometric Theory

6. Economic Developmental Cultural Change

7. Economic History Review

8. Economic Inquiry

9. European Economic Review

10. History of Political Economy

11. Industrial and Labor Relations Review

12. Journal of the American Statistical Association

13. Journal of Business

14. Journal of Business and Economic Statistics

15. Journal of Comparative Economics

16. Journal of Econometrics

17. Journal of Economic Behavior and Organization

18. Journal of Economic Dynamics and Control

19. Journal of Economic History

20. Journal of Economic Literature

21. Journal of Environmental Economics and Management

22. Journal of Finance

23. Journal of Financial Economics

24. Journal of Financial and Quantitative Analysis

25. Journal of Health Economics

26. Journal of Human Resources

27. Journal of Industrial Economics

28. Journal of Institutional and Theoretical Economics

29. Journal of International Economics

30. Journal of International Money and Finance

31. Journal of Labor Economics

32. Journal of Law and Economics

33. Journal of Legal Studies

34. Journal of Macroeconomics

35. Journal of Mathematical Economics

36. Journal of Monetary Economics

37. Journal of Money Credit and Banking

38. Journal of Public Economics

39. Journal of Regional Science

40. Journal of Royal Statistical Society

41. Journal of Urban Economics

42. Kyklos

43. National Tax Journal

44. Oxford Bulletin of Economics and Statistics

45. Oxford Economic Papers

46. Public Choice

47. Rand Journal of Economics

48. Scandinavian Journal of Economics

49. Scottish Journal of Political Economy

50. Southern Economic Journal

Received October 2003; accepted March 2006.

References

American Association of University Professors. 1993. The annual report of the economic status of the profession. Academe 73:8-90.

American Association of University Professors. 2000. The annual report of the economic status of the profession. Academe 86:12-95.

Barbezat, D. A. 1989. The effect of collective bargaining on salaries in higher education. Industrial and Labor Relations Review 42:443 53.

Bett, Sir Michael. 1999. Independent review of higher education pay and conditions. London: The Stationary Office.

Blackaby, David, and Jeff Frank. 2000. Ethnic and other minority representation in UK academic economics. The Economic Journal 110:F293-F311.

Blinder, Alan. 1973. Wage discrimination: Reduced form and structural estimates. Journal of Human Resources 8:436-55.

Booth, Alison, Jonathan Burton, and Karen Mumford. 2000. The position of women in UK academic economics. The Economic Journal 110:F312-33.

Bowen, William G. 1963. University salaries: Faculty differentials. Economica 30:341 59.

Bratsberg, Bernt, James F. Ragan, and John T. Warren. 2003. Negative returns to seniority: New evidence in academic market. Industrial and Labor Relations Review 56:306 23.

Brown, W. W., and C. C. Stone. 1977. Academic unions in higher education: Impacts on faculty salary, compensation and promotions. Economic Inquiry 15:385-96.

Financial Times. 1998. Britain's top 100 universities. 29 April, pp. 1-2.

Gordon, Nancy M., Thomas E. Morton, and Ina C. Braden. 1974. Faculty salaries: Is there discrimination by sex, race, and discipline? American Economic Review 3:419-27.

Hamermesh, Daniel, George Johnson, and Burton Weisbrod. 1982. Scholarship, citations and salaries: Economic rewards in economics. Southern Economic Journal 49:472-81.

HEFCE (Higher Education Funding Council for England). 1997. The impact of the 1992 research assessment exercise on higher education institutions in England. London: HEFCE.

Hoffman, E. P. 1976. Faculty salaries: Is there discrimination by sex, race, and discipline? Additional evidence. American Economic Review 66:196-98.

Johnson, George, and Frank Stafford. 1974. Lifetime earnings in a professional labor market: Academic economists. Journal of Political Economy 82:549-69.

Kalaitzidakis, Pantelis, Theofanis Mamuneas, and Thanasis Stengos. 1999. European economics: An analysis based on publications in the core journals. European Economic Review 43:1150-68.

Katz, David A. 1973. Faculty salaries, promotions, and productivity at a large university. American Economic Review 63:469-77.

Machin, Stephen, and Andrew Oswald. 2000. UK economics and the future supply of academic economists. The Economic Journal 110:F334-49.

McDowell, John M., Larry D. Singell, Jr., and James P. Ziliak. 1999. Cracks in the glass ceiling: Gender and promotion in the economics profession. American Economic Review 89:392-96.

Metcalf, David. 1970. University salaries: Faculty differentials. Economica 37:362-72.

Moore, William J., Robert J. Newman, and Geoffrey K. Turnbull. 1998. Do academic salaries decline with seniority 9 Journal of Labor Economics 16:352-66.

Moore, William J., Robert J. Newman, and Geoffrey K. Turnbull. 2003. Internal markets for department chairs: Comparative advantage, life-cycle, and jury duty. Journal of Labor Research 24:670-82.

Oaxaca, Ronald. 1973. Male-female wage differentials in urban labor markets. International Economic Review 14:693-709.

Oaxaca, Ronald, and Michael Ransom. 1999. Identification in detailed wage decompositions. The Review of Economics and Statistics 81:154-157.

Ransom, M. R. 1993. Seniority and monopsony in the academic labor market. American Economic Review 83:221-33.

Scott, Loren C., and Peter M. Mitias. 1996. Trends in rankings of economic departments in the U.S.: An update. Economic lnquiry 34:378-400.

Ward, Melanie. 1998. Gender differences in academe. Ph.D. thesis, University of Aberdeen: Aberdeen, U.K.

Web of Science. Cited reference search. Accessed January and February 2000. Available http://www.portal.isiknowledge. com/portal.cgi/wos.

Williams, G., T. Blackstone, and D. Metcalf. 1974. The academic labour market: Economic and social aspects of a profession. Amsterdam: Elsevier Scientific Publishing Company.

World Bank. 2003. Worm development indicators. Washington, DC: World Bank.

(1) Until recently this was a national policy. But, we have been told that both the London Business School and the London School of Economics have adopted a differential pay scale for some disciplines, in particular, economics, finance, and accounting.

(2) Currently, there are three different salary schedules, which cover approximately 78% of the university teachers (Bett 1999). These include one for pre-1992 universities, one for post-1992 universities (excluding Scotland), and one for post-1992 institutions in Scotland. There are some differences between these schedules in terms of equivalent staff in different categories of institutions and in the length of the scale and number of incremental points within them.

(3) For purposes of analysis, we shall assume there are only three ranks in the U.K. system: lecturers, senior lecturers/readers, and professors. Actually, the pre-1992 universities include lecturer A and lecturer B grades, but in practice these tend to operate as a single grade. Post-1992 institutions include lecturer, senior lecturer, and principal lecturer grades. The first two of these also operate as a single grade, which we will designate as lecturers. In each category of institutions there is virtual automatic progression between the two lowest grades. In Scottish Conference institutions there is a single lecturer grade covering the equivalent range of academic staff. We treat the principal lecturer grade in post-1992 institutions as equivalent to the senior lecturer grade in pre-1992 institutions, and we combine them with readers to form our second basic rank. Senior lecturer and reader posts have long been similar in pay and status differing in little more than name. See Williams, Blackstone, and Metcalf (1974).

(4) It should be noted that U.K. academic economists may have some incentives to move since individual universities have some discretion in appointing new hires on the national salary scale.

(5) Salaries in the U.S. were obtained from published university budgets for the 1992-1993 academic year. Salaries of U.K. economist are self-reported in 1999. To facilitate a comparison of U.S. and U.K. salaries, we updated the 1992-1993 academic salaries for U.S. economists into 1999 figures based on the growth rate of salaries reported by the American Association of University Professors (1993, 2000). We transformed the 1999 U.K. salaries measured in pounds into dollars using the OECD purchasing power conversion factor.

(6) The five journals were added following the recommendations of colleagues in the U.K. The U.S. data were recoded to include the five journals in the level 2 category, which resulted in a total of 24 additional level 2 publications.

(7) One important measure of the quality of an individual's scholarly research is excluded from our comparative analysis. Previous studies confirm that citations have a positive influence on the salaries of academic economists in the United States (Hamermesh, Johnson, and Weisbrod 1982; Moore, Newman, and Turnbull 1998). We collected individual career citations data for U.K. economists from the Web of Science Citation Databases. However, our attempts to incorporate career citations into the analysis produced results that were implausible. When the citations variable was included in our U.K. earnings model, it had a significant negative influence on salaries. We altered the data set and model specification but were unable to change this implausible result, so citations were not included in our reported regressions.

(8) For the United States, the Ph.D. quality variable has a value of one if the individual received the Ph.D. from a top 13 program (see Moore, Newman, and Turnbull 1998). In the United Kingdom, the Ph.D. quality variable has a value of one if the individual received a Ph.D. from Cambridge, Oxford, or the London School of Economics, which were the premier programs for most of the observation period. We recognize that the recent 1996 RAE evaluations differ somewhat.

(9) A fully interactive specification pooling the U.S. and U.K. samples revealed significant differences in both slope and intercept coefficients between the two markets. This is consistent with the hypothesis that while earnings distributions in both countries depend on similar demographic and productivity characteristics, the marginal returns to specific variables can differ. That is, there are differences in institutional characteristics between the U.S. and U.K. academic labor markets, which are reflected in each country's earnings distribution.

(10) See Gordon, Morton, and Braden (1974); Hoffman (1976); Ransom (1993); and Moore, Newman, and Turnbull (1998).

(11) In that study the analysis was limited to tenured faculty since heterogeneity tests indicated that assistant professors should not be pooled with senior faculty. Heterogeneity tests on the U.K. data indicated that lecturers can be pooled with senior faculty. Therefore, for purposes of comparative analysis we also pooled all U.S. faculty ranks.

(12) In semilogarithmic equations, the percentage effect of a dummy variable on the dependent variable is equal to 100.[exp [beta] - 1].

(13) The separate rank equations can be obtained from the authors upon request.

(14) While evidence of diminishing returns was found among our sample of U.S. academic economists, we were unable to detect such evidence for the U.K. market. We do not wish to make too much of this result, however. Second, among professors, we attempted to determine whether the effect of publications on earnings depends on how the individual's stock of publication is distributed across a career. This was accomplished by partitioning total publications into those produced during each rank (lecturer, senior lecturer, and professor). This partitioning was performed for level 1 and level 2 publications separately. We could find no evidence that the age distribution of an individual's publications affected relative earnings, holding total publications constant. That is, an individual's relative salary does not appear to be sensitive to whether the individual's total stock of publications is relatively "new" or "old."

(15) See Moore, Newman, and Turnbull (2003) for an empirical analysis designed to sort through the sources of the chair premium.

(16) We also examined four other hypotheses using our sample of U.K. economists. First, we included a Ph.D. dummy variable for whether a faculty member had a terminal degree. Unlike the market in the United States, where virtually all faculty members have a Ph.D. degree, many U.K. faculty members do not possess a Ph.D. degree. Holding productivity constant, we found no premium for individuals with a Ph.D. degree. Second, we tested whether there were university or departmental quality effects on salary structure. To proxy university quality, we used the Financial Times ratings of U.K. universities (Financial Times, April 29, 1998). This variable never attained significance, which is not surprising under the national salary schedule. To measure department quality, we used the RAE seven-point rating scale for economics and econometric units. This variable had a significant positive effect (0.035 with a t value of 2.35) when added to the U.K. earnings equations. This result is consistent with the hypothesis that higher rated departments bid up the salaries of research scholars. Third, we attempted to test what effect mobility has on earnings by including a variable measuring the number of times an individual moved between universities subsequent to the rank of lecturer. This variable was not significant in any of the specifications. This is understandable since we do not have information on individuals who may have received outside offers that were matched by their current department. The effect on earnings of counter offers is conceptually the same as an actual move for higher pay. Finally, we included a dummy variable for post-1992 universities to capture the effects of different national pay scales. This dummy failed to attain statistical significance at conventional levels.

(17) This equation follows from a standard decomposition procedure used by Blinder (1973) and Oaxaca (1973) of the following form [DELTA]sal = [[beta.sup.US.sub.0] - [[beta].sup.UK.sub.0] + [summation][[beta].sup.US.sub.i][X.sup.US.sub.i] - [summation][[beta].sup.UK.sub.i][X.sup.UK.sub.i], with superscripts denoting regressions run separately for U.S. and U.K. samples. More recent studies have extended the decomposition procedure, offering alternative approaches, which correct for identification problems. For example, see Oaxaca and Ransom (1999), who demonstrate that a fundamental identification problem exists when one attempts to estimate the separate contributions of sets of variables to the unexplained portion of the earnings decomposition. Since we use the decomposition procedure to merely motivate the counterfactual exercise, we do not pursue these alternative decomposition procedures.

(18) These productivity rates were determined by examining the productivity profiles of the top economists in both samples. Our assumptions regarding publication flows are based on the publication flows of the most productive scholars in our samples. The most prolific scholar in our sample generated 20 level 1 journal articles in a little less than 20 years. The mean number of level 1 publications per year among full professors is 0.17, but this includes a large number of professors with zeros.

William J. Moore, Department of Economics, Louisiana State University, Baton Rouge, LA 70803, USA.

Robert J. Newman, Department of Economics, Louisiana State University, Baton Rouge, LA 70803, USA; E-mail eonewm@ lsu.edu; corresponding author.

Dek Terrell, Department of Economics, Louisiana State University, Baton Rouge, LA 70803, USA.

Table 1. Composition of a Sample and Comparison with Other Studies

 Present Blackaby Booth, Burton,
 Study and Frank and Mumford
 (2000) (2000)

Percentage within rank
 Lecturers 35.7 37.8 44.8
 Senior lecturer/reader 26.9 20.1 21.4
 Professors 37.3 25.3 18.8
 Researchers 0.0 4.7 12.0
Distribution of faculty
 by RAE score
 5+ 7.1 12.9 11.4
 5 22.8 23.7 20.8
 4 51.9 28.1 20.1
 3 (a) 14.2 25.6 34.3
 2 3.9 8.5 12.4
 1 0.0 1.3 1.0
Percent female by rank
 Lecturer 20.0 -- 19.0
 Senior lecturer/reader 9.0 -- 11.2
 Professor 0.0 -- 4.1

 Machin and HEFCE
 Oswald (1997)
 (2000)

Percentage within rank
 Lecturers 44.1 --
 Senior lecturer/reader 25.2 --
 Professors 20.1 --
 Researchers 10.6 --
Distribution of faculty
 by RAE score
 5+ -- 13.4
 5 -- 27.8
 4 -- 32.2
 3 (a) -- 21.2
 2 -- 4.2
 1 -- 1.2
Percent female by rank
 Lecturer -- --
 Senior lecturer/reader -- --
 Professor -- --

(a) This category combines "3 upper" and "3 lower."

Table 2. Model Descriptive Statistics: Variable Means

Model Variables U.K. Faculty U.S. Faculty

Salary in U.S. dollars (1999) 52,558 75,500
Experience 16.70 16.41
Seniority 11.40 12.62
Gender (% male) 0.90 0.95
Ph.D. Quality 0.21 0.41
Level 1 Articles 2.09 2.67
Level 2 Articles 4.57 6.96
Total Other Pubs 31.34 20.44
Rank II (%) 0.2698 0.2486
Rank III (%) 0.3730 0.5304
Chair Years (a) 0.96 0.86
Sample size 126 181

(a) The mean is conditioned on chair experience.

Table 3. Salary Equations for U.S. and U.K. Academic Economists

 Models with Rank Variables

Model Variables U.S. U.K.

Experience 0.0059 0.0150 **
 (0.0072) (0.0051)
Experience (2) -0.00003 -0.00032 **
 (0.00016) (0.00013)
Seniority -0.0082 ** 0.0001
 (0.0023) (0.0019)
Gender (M = 1) 0.0254 0.0664
 (0.0455) (0.0397)
Ph.D. Quality 0.0422 * 0.0669 **
 (0.0215) (0.0281)
Total Other Pubs 0.0009 * 0.0009 **
 (0.0005) (0.0003)
Level 1 Articles 0.0252 ** 0.0107 **
 (0.0035) (0.0034)
Level 2 Articles 0.0074 ** 0.0021
 (0.0020) (0.0027)
Chair Years 0.0186 ** 0.0141 **
 (0.0048) (0.0060)
Rank II 0.1218 ** 0.1323 **
 (0.0425) (0.0311)
Rank III 0.3195 ** 0.2956 **
 (0.0529) (0.0388)
Intercept 10.8589 ** 10.4123 **
 (0.0525) (0.0402)
F value 57.43 47.41
Adjusted [R.sup.2] 0.7752 0.8033
N 181 126

 Models without Rank Variables

Model Variables U.S. U.K.

Experience 0.0275 ** 0.0235 **
 (0.0062) (0.0059)
Experience (2) -0.00040 ** -0.00047 **
 (0.00015) (0.00010)
Seniority -0.0064 * 0.0007
 (0.0026) -0.0023
Gender (M = 1) 0.0255 0.0941 *
 (0.0512) (0.0480)
Ph.D. Quality 0.0660 ** 0.0764 **
 (0.0238) (0.0341)
Total Other Pubs 0.0006 0.0013 **
 (0.0006) (0.0004)
Level 1 Articles 0.0265 ** 0.0130 **
 (0.0039) (0.0041)
Level 2 Articles 0.0124 ** 0.0084 **
 (0.0021) (0.0032)
Chair Years 0.0223 ** 0.0254 **
 (0.0052) (0.0068)
Rank II

Rank III

Intercept 10.7609 ** 10.3825 **
 (0.0568) (0.0488)
F value 53.86 34.67
Adjusted [R.sup.2] 0.7255 0.7080
N 181 126

* Significant at 0.10.

** Significant at 0.05.