return of publications for economics faculty | My Assignment Tutor

The return of publications for economics facultyOnur Baser Elda PemaThe MEDSTAT Group Michigan State UniversityAbstractThis study uses comprehensive panel data to determine the effect of publications on thesalaries of full-time economics faculty in nine midwestern universities. The data set allowsus to control not only the volume but also the quality of publications. Recent developments inthe ISI-Web of science enable us to divide total citations per faculty member into citationsby others and self-citations. Since none of the traditional measures (citations, publicationindexes, total article pages) when used individually fully accounts for all research output, allavailable measures should be used. Our findings indicate that average number ofarticle-pages published in The American Economic Review (AER) are likely to increasesalary by %1.3 to %1.9 per year. Neither self-citations nor publications in non-rankedjournals appears to affect salary.Citation: Baser, Onur and Elda Pema, (2003) “The return of publications for economics faculty.” Economics Bulletin, Vol. 1,No. 1 pp. 1-13Submitted: May 13, 2003. Accepted: August 22, 2003.URL: are the quantifiable reward for publishing an article in the AER? Whatif the same article appeared in an alternative journal? Does an article affectsalary if it remains unpublished? Questions regarding the effect of publications on academic salaries have been posed since the 1960s, and a number ofempirical studies since then confirmed the now widely accepted opinion thatpublications have a positive impact on salary. The relative magnitude of thatimpact, however, is to dispute, mostly due to pervasive measurement problems.The objective of this study is to improve estimation by reducing measurementerror. We do this by combining factors that previous studies investigated separately: the structure of salaries in academe, the impact of publications onsalary, the quality of publications as measured by citation frequency, journalreputation, and, finally the problematic determination of objective measure ofquality in the context of academic research.Because rankings are frequently used to assess the quality of economics journals, departments, and academicians, this paper on the rewards of publicationdeals with the problem of ranking. We provide an overview of the prevalentranking methods and the rationales for considering them adequate measuresof quality. We then incorporate the standard methodologies into the salarymodel in order to determine whether academic activity indexes used to rankdepartments and journals are related to salary, and whether these methodologies for determining quality fully account for all academic activity.1 Salaries and Research ActivityAlmost all previous studies have agreed that publications, citations, by colleagues, quality of education, and teaching ability have a positive effect onearnings (e.g., Daniel S. Hamermesh et al., 1982 Johnson, and Weisbrod, 1982,Melichar, 1965, 1968; Siegfried and White, 1973; Katz 1973; Tuckman and Leahey, 1975; Tuckman, Gapinski, and Hageman, 1977; Hansen, Weisbrod, andStrauss, 1978). These studies have differed, however, in their estimation ofrelative weights of each factor.Studies using data from a national survey of faculty conducted between 1972and 1973 by the American Council of Education (ACE) in 1972-1973, werelimited to using only approximate measures of research activity (Tuckman andLeahey, 1975; Tuckman, Gapinski, and Hageman, 1977). Other studies usedtotal number of publications as a measure of professional activity (Katz, 1973;Siegfried and White, 1973; Hansen, Weisbrod, and Strauss, 1978) while ignoring the quality and length of publications. Other studies limited the numberof faculty publications by accounting for article published in a few leadingjournals or during a 3-5 year time span. Most authors recognized the need toaccount for quality and length of publications, but few of them incorporated1a quality index in their models.Hamermesh, Johnson, and Weisbrod (1982) suggested measuring the quality of scholarly research by the total number of citations. This method waslater criticized because citation measures included self-citations and creditedonly the first author in a co-authored publication. Recent changes in the waycitations are recorded avoid both problems.The key to determining academic research quality lies in assessing the qualityof scholarly publications. The most common way of accounting for differencesin quality has been to separate publications into distinct categories and/orconsider only those articles appearing in the highest rated journals in thefield.While our study incorporates techniques from previous studies, it improvesupon them by attempting to account for all publications in terms of qualityas well as length.One variable that we exclude because of data unavailability is teaching performance. This variable appears hard to model for many reasons. Katz, afterinterviewing department chairs, concluded that under graduate student evaluations were viewed with skepticism and thus promoted to use of graduatestudents as a measure of teaching quality. Siegfried and White used studentratings, while Tuckman et al., used teaching awards as a proxy for teaching quality. With the exception of the latter study, the rest found teachingperformance to be insignificant in the determination of salaries.2 RankingsOver the last few decades, numerous articles in academic economics journalshave addressed the problem of ranking economics departments, journals, andfaculty. The ranking methods used in these articles fall into one of the following three categories: (1) opinion-derived rankings, based on surveys of severalacademicians in the field (Malouin and Outerville, 1987); (2) citation-derivedrankings, based on the number of times a journal or economist is cited (Davisand Papanek, 1984; Blair et al. 1986); and (3) quantity rankings, based on thenumber of pages published per professor or department and generally weightedby the quality of journals (Graves et al., 1982). Most of these published evaluations have been criticized for the subjectivity of the ranking methods.Opinion-based rankings were considered largely inaccurate and, for the mostpart, biased. Rankings based on citation calculation raised the question ofwhether citation quantity could be considered a reasonable indicator of pro-2fessor quality. Furthermore, critics quickly asserted that the same weight toself-citations as to citations by others and that crediting only the first author of a multi-authored publications seriously flawed the method. Quantityrankings used only publications in “leading journals” and excluded alternativeforms of publication. But while the list of “leading journals” often includedas many as 25 and was proven to be heavily biased toward theoretically oriented journals (Hawkins et al., 1973). Yet, most critics of quality rankingshave neglected to consider the irrelevance of a method’s alleged subjectivityto a rating’s ultimate effects on real variables. In terms of prestige, there islittle doubt that subjective ratings are of relevance. This study attempts toinvestigate whether quality ratings affect real variables such as salary.3 Data, Preliminary Analysis, and ModelWe collected detailed information on salaries and the publication activities ofeconomics faculty from the nine big midwestern universities. Unlike previousstudies that used cross-sectional data, we have compiled a panel data consisting of the 1991-1992, 1995-1996, and 1998-1999 academic years in an attemptto control for unobserved individual effects.1. Salary. This is our dependent variable. We measured it as the totalgross salary received by the economics faculties from the employing institution during the above academic years. We did not collect informationon outside grants, consulting fees, or other sources of income (Assumingthat these do not affect the salary decisions). 1 In our regressions, salaryappears in logarithmic form (LOGSALARY). 22. Publications. Data on research activity was collected by using a combination of faculty vitas and the Econlit database. This database provides acomprehensive list of publications since 1969 and gives full credit to eachauthor for co-authored articles. Publications are measured as number ofpages per author. We sorted them into three categories:(i) Articles in journals (TOTARTPG) includes articles appearing in allacademic journals- national, regional, and specialty. The number ofcomments and replies are also included wheras reviews and correctionsare not. We added the square of this term (TOTARTPGSQ) into ourregression in order to capture possible diminishing effect.(ii) Other publications (TOTOTHER) includes books, textbooks, editedbooks, articles in collective volumes, book chapters, book reviews, gov-1 The same assumption was explicitly made by Tuckman et al. (1977) and implicitlyby all others.2 We tried the level salary first, but a Ramsey RESET test revealed functional formmisspecification.3ernment documents, conference proceedings and working papers.(iii) Non-ranked publications (TOTNONRNK) includes articles publishedin journals that received a weight of zero in our citation-based ranking.Using citation accounts of 800 journals from the Social Science CitationIndex (excluding medical, psychology, and some law journals), followingwe ranked them according to number of total cites. Table 1 lists someof the highest-ranking journals and compares them to the rankings ofLiebowitz and Palmer.3. Publication Index(PUBLINDX). Our purpose in constructing a publication index was not to develop an index more accurate than those usedin previous studies, the point rather was to reconstruct and index previously used to rank both department and journals in order to evaluatehow current perceptions of quality affect academic salaries. The publication index was constructed by multiplying the number of pages perperson (pp.article) by the weight of the journal (W ) in which the articlewas published as given by table 1; we then summed this product over allpublications:NXi=1JXj=1pp.article[i]authors ×Wj100,where i = 1, 2, …N (total articles published), j = 1, 2, …J (journal inwhich article was published).4. Citations. The reputation of researchers is often measured by number oftimes their works been cited. Citations were collected from the ISI Web ofScience site and drawn from the Social Sciences Citation Index database.Efforts were made to account for every citations per scholar by searchingboth with and without the scholar’s middle initial 3 . To reduce the risk ofmistakenly tallying of like-named authors when searching without initials,we cross referenced the cited journals for author names and addresses,since most publications list the institutional address of the researcher).We divided total citations into two parts: self-citations (SELFCITES)and citations by others (OTHCITES). We added the square of citationsby others to control for possible diminishing effect (OTHCITESSQ).5. Female. It is used to check for the possibility of discrimination in salarydeterminations since previous studies have reported such findings.6. Universities. We generated a categorical variable for each university toaccount for department-specific factors.7. Top 10. We combined several published rankings of economics depart-3 Due to personal preferences or recording practices, the ratio of author citationswithout the author’s middle initial to citations with the middle initial vary widely.A note of caution therefore to future researchers: using one of either search criterionis likely to underestimate the citation counts for at least a substantial part of thesample.4ment to construct a list of 10 universities considered to be the best (seetable 2). This variable received a value of one for having received a Ph.D.from a Top10 university and intended to capture the quality of education.8. Chair. This variable took a value of one if an individual has served as adepartment chair on a given time interval.Summary statistics are displayed in Table 3.Pooled Ordinary Least Squares (POLS) and Fixed Effects (FE) model werefitted to the data. The regression results from a POLS aimed at assessingthe general behavior of the variables in the model. The FE model controlledfor the possibility of unobservable factors that may influence the explanatoryvariables.4 FindingsTable 4 presents the results obtained by two different regressions. These resultsindicate that most variables have expected signs and that most publicationvariables are statistically significant. Both Our publication-index and articlepages variables suggest that 10 article pages in the American Econmic Review(AER) are expected to increase the salary by approximately 1.3 % according toPOLS and 1.9% according to the FE. The average salary 1999 salary $93,229would thus increase an additional $1,251 according to POLS and $1,762 according to FE. If an article is cited by others 50 times, POLS and FE estimatean approximate salary increase of 2% and 3.5% respectively. The statisticalsignificance of all three measures- namely the publication index, total article pages, and citations- suggests that none of these measures by itself fullyaccounts for all research output.According to POLS, the other forms of publications described above do seemto affect salary. Publications at non-ranked journals and self citations do notappear to affect salary according to results obtained by both regressions.We find no evidence of discrimination on the basis of gender. However, thismay well be due to the very small number of women faculty in the sample(only 21, or 8.4%, of the sample).Having received a PH.D. from one of the 10 best universities in the countrydoes not appear to affect the academic salaries of economics faculty.However, having served as department chair implies predicted a salary increaseof about 12% to 17%.55 Concluding CommentsThis paper estimates economics faculty salaries by using a panel data fromnine big midwestern universities by systematically considering several factsrelated to publications and citations. Publications, we conclude, have a positive and diminishing effect on salaries. Furthermore, quality of publication isa significant determinant of salary. Publication indexes, our result indicate, donot fully account for the all publication quantity and quality in salary determination as indicated by the significance of such variables as article-pages andcitations by others, but publication indexes still influence salaries significantly.Therefore we accept only with reservations ratings of departments and academicians that do not employ a more comprehensive measure of publicationactivity.Two limitations of the paper should be discussed. (1) It can reasonably beargued that the ranking of the Ph.D.-granting department at the time whenthe faculty member was hired is more relevant than its the current ranking.Observed salaries might be a function of old rankings. Using rankings of thePh.D.-granting departments at the time of first hire could rectify this potentialfactor. We tried to mitigate such possible discrepancies by using several different rankings at various points in time. (2) We did not use AER-equivalentpages in accounting for length of articles as is customary in most studiesabout ratings. In this sense, we consider our publication measures impreciseto a slight degree but nevertheless useful.6REFERENCESHamermesh, Daniel S., Johnson, George E. and Weisbrod, BurtonA., “Scholarship, Citations and Salaries: Economic Rewards in Economics,”Southern Economic Journal, October 1982, 49, 472–81.Siegfried, John J., White, Kenneth J., “ Financial Rewards to Researchand Teaching: A Case Study of Academic Economists Author,” American Economic Review, May 1973, 63, 309-15.Katz, David A., “Faculty Salaries, Promotion, and Productivity at a LargeUniversity, ” American Economic Review, June 1973, 63, 469-77.Tuckman, Howard P.; Leahey, Jack, “What Is an Article Worth ?,” Journal of Political Economy, October 1975, 83, 951-67.Tuckman, Howard P.; Gapinski, James H.; Hagemann, Robert P.,“Faculty Skills and the Salary Structure in Academe: A Market PerspectiveAuthor, ” American Economic Review September 1977, 67, 692-702.Hansen, W. Lee; Weisbrod, Burton A.; Strauss, Robert P. Source.,“Modelingthe Earnings and Research Productivity of Academic Economists Author,”Journal of Political Economy, August 1978, 86, 729-41.Malouin, Jean-Louis; Outreville, J.-Francois Source., “The RelativeImpact of Economics Journals: A Cross-Country Survey and Comparison Author”, Journal of Economics and Business August 1987, 39, 267-77.7Table 1. Citation-Based Ranking of Economic JournalsAbout 800 journals were ranked according to the number of cites recorded by the Social Sciences Citation Index.Lieboweitz and Palmer (1984) ranked journals the same way, and their rankings are provided on the last column for comparison.For illustration purpose, this table provides weights for only some of the journals in the 800-journal list. TOTAL CITESWeightWeightJOURNAL-1998-1998-1980AMERICAN ECONOMIC REVIEW8999100100ECONOMETRICA794388.2771.51JOURNAL OF POLITICAL ECONOMY669774.4278.45QUARTERLY JOURNAL OF ECONOMICS413845.9829.13YALE LAW JOURNAL382142.461.66JOURNAL OF FINANCE379142.1331.93ACADEMY OF MANAGEMENT JOURNAL345938.44ADMINISTRATIVE SCIENCE QUARTERLY336237.36AMERICAN POLITICAL SCIENCE REVIEW325236.14JOURNAL OF CONSUMER RESEARCH276230.69JOURNAL OF FINANCIAL ECONOMICS267629.748.02JOURNAL OF MARKETING265429.49JOURNAL OF MARKETING RESEARCH254828.31ECONOMIC JOURNAL254028.2326.64JOURNAL OF ECONOMIC THEORY251427.9419.75JOURNAL OF ECONOMETRICS247927.55REVIEW OF ECONOMIC STUDIES241126.7929.5REVIEW OF ECONOMICS AND STATISTICS233125.9STANFORD LAW REVIEW226325.15STRATEGIC MANAGEMENT JOURNAL225325.04MICHIGAN LAW REVIEW207223.0221.89HARVARD BUSINESS REVIEW200822.31UNIVERSITY OF CHICAGO LAW REVIEW189921.1JOURNAL OF MONETARY ECONOMICS186020.677.43JOURNAL OF SOCIAL ISSUES183620.4AMERICAN JOURNAL OF AGRICULTURAL ECONOMICS181220.1416.78AMERICAN JOURNAL OF POLITICAL SCIENCE160817.87JOURNAL OF ECONOMIC PERSPECTIVES158317.59JOURNAL OF LAW & ECONOMICS158017.5615.25HUMAN RELATIONS154817.2JOURNAL OF ECONOMIC LITERATURE15301710.01JOURNAL OF PUBLIC ECONOMICS143715.97WORLD DEVELOPMENT140815.655.88DEMOGRAPHY139415.4915.79JOURNAL OF LEGAL STUDIES125013.897.28ENVIRONMENT AND PLANNING A124913.88EUROPEAN ECONOMIC REVIEW124313.813.15ECONOMIST116812.98JOURNAL OF ENVIRON. ECON AND MANAGEMENT115212.8INTERNATIONAL ECONOMIC REVIEW111312.3716.75JOURNAL OF HUMAN RESOURCES111312.378.44JOURNAL OF BUSINESS108312.0313.11JOURNAL OF MANAGEMENT106311.81RAND JOURNAL OF ECONOMICS103911.55 Continued.Page 1Table 1. Concluded. JOURNAL OF BUSINESS & ECONOMIC STATISTICS98810.98INDUSTRIAL & LABOR RELATIONS REVIEW98310.927.92WORLD POLITICS97610.85JOURNAL OF HEALTH ECONOMICS95710.63ECONOMICS LETTERS93010.334.45POPULATION AND DEVELOPMENT REVIEW8889.8711.59RESEARCH IN ORGANIZATIONAL BEHAVIOR8889.87JOURNAL OF INTERNATIONAL BUSINESS STUDIES8839.81PUBLIC CHOICE8719.685.34JOURNAL OF POLITICS8579.52JOURNAL OF INTERNATIONAL ECONOMICS8389.314.9URBAN STUDIES8369.295.83JOURNAL OF MONEY CREDIT AND BANKING8349.279.32ECONOMICA8259.1714.19JOURNAL OF URBAN ECONOMICS7878.755.51JOURNAL OF ECONOMIC HISTORY7378.1910.26JOURNAL OF LABOR ECONOMICS7338.15LAND ECONOMICS7308.116.86JOURNAL OF CRIMINAL LAW & CRIMINOLOGY7238.03SOCIAL SCIENCE QUARTERLY7197.997.95JOURNAL OF DEVELOPMENT ECONOMICS7077.86REGIONAL STUDIES7037.818.22JOURNAL OF ECONOMIC BEHAVIOR & ORGANIZATION6987.76ECONOMIC INQUIRY6847.6MONTHLY LABOR REVIEW6697.4314.86JOURNAL OF MANAGEMENT STUDIES6547.27SOUTHERN ECONOMIC JOURNAL6467.1811.05JOURNAL OF ECONOMIC DYNAMICS & CONTROL6367.07OXFORD BULLETIN OF ECONOMICS AND STATISTICS6176.86HEALTH EDUCATION RESEARCH6066.73JOURNAL OF BANKING & FINANCE6026.69JOURNAL OF ACCOUNTING RESEARCH6016.686.84ECONOMIC DEVELOPMENT AND CULTURAL CHANGE5976.638.22LAW AND CONTEMPORARY PROBLEMS5896.55POPULATION STUDIES-A JOURNAL OF DEMOGRAPHY5896.55OXFORD ECONOMIC PAPERS-NEW SERIES5826.476.79APPLIED ECONOMICS5786.42CANADIAN JOURNAL OF ECONOMICS5566.185.54REVIEW OF FINANCIAL STUDIES5476.08ECONOMIC HISTORY REVIEW5456.069.62HEALTH ECONOMICS5446.05ANNALS OF THE AMERICAN ACADEMY OF POLITICAL AND SOCIALSCIENCE5426.02INDUSTRIAL RELATIONS5426.026.35JOURNAL OF THE AMERICAN PLANNING ASSOCIATION5406TRANSPORTATION SCIENCE5285.87ECONOMIC GEOGRAPHY5045.67.06NATIONAL TAX JOURNAL5025.588.73 Page 2Table 2. Ranking of Economics Departments US News Gourman Ehrenberg Dusansky Scott Conroy Golden Davis Gravesand World Report Report Hurst Vernon Mitias Dusansky et al. Papanek et al.‘98 ‘97 ‘96 Fall ‘95 Fall ‘94 Fall ‘92 ‘86 78-81 74-781HarvardChicagoChicagoPrincetonHarvardPrincetonChicagoChicagoChicago2MITMITHarvardHarvardChicagoMITJohns HopkinsHarvardHarvard3StanfordHarvardMITMITPennsylvaniaNorthwesternRochesterMITStanford4PrincetonPrincetonStanfordPennsylvaniaMITChicagoStanfordStanfordWisconsinMadison5UC-BerkeleyStanfordPrincetonNorthwesternNorthwesternHarvardHarvardPrincetonPennsylvania6ChicagoYaleYaleNYUStanfordUC-BerkeleyCornellYaleMIT7YaleUC- BerkeleyUC-BerkeleyBostonUniversityPrincetonUC-San DiegoUC-BerkeleyPennsylvaniaYale8NorthwesternPennsylvaniaPennsylvaniaYaleMichiganBoston UniversityMITWisconsinMadisonUCLA9PennsylvaniaNorthwesternNorthwesternStanfordUC-BerkeleyYaleColumbiaColumbiaUC-BerkeleyMinnesotaPennsylvania10MinnesotaMinnesotaUC-San DiegoUCLAVPIUC-BerkeleyPrincetonWisconsin-MadisonNYU Our list includes all universities appearing in the above top ten rankings.Page 3Table 3. Summary Statistics VariableSALARY(1999$)Variable DescriptionSalary indexed for 1999 dollars(using the Consumer Price Index)YearsObservationsMeanStandard Deviation199221085,38626,394199522687,39526,965199921193,22928,680199224631.2834.45PUBLINDXPublication Index199523435.8737.88199923643.644.131992245129.7123.52TOTARTPGTotal published article-pagesper author1995233156.07145.061999235194.71166.11992245178.23393.11TOTOTHERTotal pages per author of otherpublications1995233231.45545.371999235309.28765.11199224420.3133.11TOTNRPGTotal article pages per author innon-ranked journals199523226.139.97199923434.8348.741992256197.23298.07OTHCITESNumber of citationsexcluding self-citations1995244219.66297.11999246248.98344.18199225611.0416.42SELFCITESNumber of self-citations199524414.1420.57199924616.1624.34199270.03–CHAIR=1 if department chair199570.03–1999100.04–= 1 if Ph.D. was receivedin any one of the universitiesconsistently ranked in the top 1019921520.67–TOP 1019951530.67–19991750.74–1992240.09–FEMALE=1 if female1995240.10–1999230.09– Page 4TABLE 4. Regression AnalysisDependent Variable: LOG(SALARY) Independent VariablesPOLSFETOTARTPG0.00164640.0012878[0.0003463]**[0.0003630]**TOTARTPGSQ-0.0000016-0.0000012[0.0000004]**[0.0000003]**PUBLINDX0.00102840.0024635[0.0005701]+[0.0006993]**OTHCITES0.00041910.0007211[0.0001124]**[0.0003538]*OTHCITESSQ-0.0000001-0.0000005[0.0000000]**[0.0000001]**TOTOTH0.00007410.0000478[0.0000359]*[0.0000501]TOTNONRNK0.0000752-0.0000049[0.0004889][0.0005481]SELFCITES-0.00085150.0000866[0.0007136][0.0015085]FEMALE-0.0592502[0.0480023]TOP10-0.0016968[0.0438996]CHAIR0.12163540.1788828[0.0518454]*[0.0448093]**Y950.00621660.0272711[0.0135715][0.0119770]*Y990.01842830.0455945[0.0147393][0.0162584]**UNIVERISITIESYesNoCONSTANT10.940627610.9672715[0.0503200]**[0.0468648]**OBSERVATIONS616616R-SQUARED0.470.43 Robust standard errors in brackets+ significant at 10%; * significant at 5%; ** significant at 1%Page 5


Leave a Reply

Your email address will not be published.