Economics Faculty Research at Teaching Institutions: Are Historically Black Colleges Different?

Jacqueline Agesa [*]

Maury Granger [+]

Gregory N. Price [++]

This paper examines the difference in research output of economics departments at historically black colleges and universities (HBCUs) and non-HBCUs that are teaching institutions. We also examine the causal relationship

between economics faculty research and the number of an institution's baccalaureate graduates who earn doctorates in economics. Our findings suggest that economics departments at HBCUs produce less research output relative to non-HBCUs. However, research output is equally effective in producing economics doctorates at both types of institutions. These findings suggest that a plausible way to increase the stock of black Ph.D. economists is to increase economics research at HBCUs.

1. Introduction

An emerging literature on economics faculty research at undergraduate institutions underscores the fact that research universities are no longer the sole sources of scholarship in economics. This is particularly true in the case of liberal arts colleges. Analyses by Hartley and Robinson (1997) and Bodenhorn (1997) suggest that an increasing emphasis is being placed on scholarship in economics at liberal arts colleges. To the extent that the historical mission of this type of institution has been teaching, some concerns can be raised as to whether or not an emphasis on scholarship is complementary with teaching. In this context, Hartley and Robinson (1997) report that at least one measure of teaching effectiveness is complementary with scholarship: There is a positive correlation between the research productivity of an economics department and the number of its baccalaureate graduates who earn Ph.D. degrees in economics. Agesa, Granger, and Price (1998) report a similar relationship between economics faculty research productivity and the number of baccalaureate graduates earning economics doctorates at historically black colleges and universities (HBCUs).

With the exception of Hartley and Robinson (1997), who include Morehouse and Spelman Colleges in their sample (two liberal arts colleges that are HBCUs), the emerging literature on economics research at teaching colleges and universities provides no insights into economics faculty research at HBCUs relative to non-HBCUs. Ranking 125 liberal arts colleges on the basis of the number of articles published over the 1989-1994 period that were indexed in the Journal of Economic Literature (JEL), Hartley and Robinson's results show the average for the sample is 9.7 JEL-indexed articles. [1] The totals for Morehouse and Spelman, respectively, were one and three JEL-indexed publications. Because the typical institution in the sample is a non-HBCU, this suggests that relative to non-HBCUs, economics faculty at HBCUs may publish less. As Morehouse and Spelman are only two HBCUs and may or may not be representative of all HBCUs, a fundamental question is whether HBCUs in general are different from non-HBCUs with regard to economics faculty research productivity.

In this paper, we examine whether or not HBCUs are different from non-HBCUs in terms of research productivity in economics. The determinants of research productivity are estimated for a sample of 39 HBCUs and 101 non-HBCUs. Our specification allows the measurement of the expected returns to research output of faculty and institutional characteristics at these teaching institutions. We also explore whether or not research productivity at these teaching institutions has any impact upon teaching effectiveness and the extent to which it varies by institutional type. In particular, we explore whether or not, for the HBCUs and non-HBCUs in our sample, economics faculty research and teaching are complementary in a particular way--does faculty research productivity influence the number of students that ultimately earn doctorates in economics?

A comparative analysis of research output at HBCUs and non-HBCUs is of interest for several reasons. First, existing empirical analyses on faculty research in economics at teaching colleges utilize samples that consist either almost entirely of non-HBCUs (Bodenhorn 1997; Hartley and Robinson 1997) or entirely HBCUs (Agesa, Granger, and Price 1998). An analysis with a sample that includes both non-HBCUs and a substantial number of HBCUs would promote insight into the extent to which institutional type matters for faculty research productivity in economics. Second, given that HBCUs are not sufficiently sampled in existing studies, no safe generalization can be made as to whether or not HBCUs are fundamentally different than non-HBCUs with regard to faculty research productivity. This analysis provides an initial assessment of whether or not HBCUs are different. Finally, for both non-HBCUs and HBCUs, the research productivity of an economics faculty has been found to be positively correlated with the number of its baccalaureate graduates who earn doctorates in economics (Hartley and Robinson 1997; Agesa, Granger, and Price 1998). If this functional relationship is independent of institutional type (HBCU or non-HBCU), an implication is that research and teaching are complementary at both types of institutions. Moreover, given the underrepresentation of black Americans among Ph.D. economists, a causal and positive relationship between HBCU faculty research productivity and the doctoral achievements of HBCU baccalaureate graduates in economics suggests that one way to increase the number of black Ph.D. economists would be for HBCUs to commit more resources to economics faculty research. Moreover, an examination of the relationship between economics faculty research and the doctoral achievements of HBCU baccalaureate graduates will perhaps provide a better understanding of the limited supply of black Ph.D. economists and the potential of HBCU economics programs to increase the supply. [2]

The remainder of this paper is organized as follows. Section 2 provides a brief history of HBCUs. The data and descriptive statistics for HBCUs and non-HBCUs are provided in section 3. In section 4, we estimate research output equations for the sample of HBCUs and non-HBCUs utilizing a Poisson regression approach. This approach allows the measurement of differences in the expected level of economics research output between the two types of institutions. The relationship between faculty research productivity and teaching effectiveness at teaching institutions is explored in section 5. Measuring teaching effectiveness as the number of an institution's baccalaureate graduates who earn doctorates in economics, we explore the relationship between economics faculty research productivity and the probability that an institution's baccalaureate graduates will earn a doctorate in economics. In section 6, we explore scenarios under which HBCUs could be instrumental in increasing the number of black Americans earning do ctorates in economics. The last section concludes.

2. Historically Black Colleges

The history of HBCU can be traced back to the years prior to the Emancipation Proclamation. In 1837, the first HBCU, Cheyney State Training School, was founded. Financial support for HBCUs at the state level began in earnest as a reaction to the Land Grant Act of 1890. For the first half of the 20th century, a dual system of education was maintained on the basis of the logic of Plessy v. Ferguson. [3] Only later, in 1954, did the Supreme Court attempt to remedy racially segregated educational institutions and the implied funding disparities, in Brown v. Board of Education. [4] Between 1854 and 1899, a total of 81 public and private HBCUs had been created, the majority of which where located in the South. Most of the HBCUs formed during this period were agricultural and vocational institutions, and as Du Bois (1994) notes, many were, by circumstance, merely capable of providing blacks with the equivalent of a high school education. Notwithstanding such humble origins and circumstances, Du Bois reported that b etween 1875 and 1900, HBCUs produced 1056 graduates, the majority of whom were employed as teachers, clergymen, physicians, and civil servants. With respect to economics in the curriculum at HBCUs, an early analysis by Oak [5] (1938) reveals that HBCUs began offering coursework, and in some instances undergraduate degrees in economics, beginning in 1917.

Currently, there are 87 HBCUs that are four-year baccalaureate degree-granting institutions in the United States and territories. Of these, three have Carnegie classifications as doctoral institutions, and one has a classification of Research 1 [6] The capability of HBCUs in producing black baccalaureate graduates is rather impressive. In 1995, only 19% of all black undergraduates were enrolled at HBCUs, yet in that same year, HBCUs were responsible for 33% of all baccalaureate degrees awarded to blacks. Even more relevant to the analysis of this paper, approximately 25% of all doctorates in economics awarded to blacks between 1986 and 1996 went to HBCU alumni (Price 1998). Howard University is the only HBCU with a doctoral program in economics, and they appear to be especially impressive in producing black Ph.D. economists; between 1980 and 1995, 24 black baccalaureate graduates of Howard earned economics doctorates, approximately 8% of all doctorates in economics awarded to black Americans over the period. [7]

The history of HBCUs also includes contributions to the stock of prominent black economists and economic knowledge. Some of the prominent black economists that were either educated at or affiliated with HBCUs include George Edmund Haynes and Abram Harris, Jr. Haynes was a graduate of Fisk University (a HBCU) and was the first black American to receive a doctorate in economics from Columbia University. Haynes chaired the Social Science Department at Fisk for many years and later served as the Director of the Office of Negro Economics, U.S. Department of Labor. Stewart (1997) notes that Haynes' scholarly work, which included several books and refereed scholarly articles, was indeed seminal and can inform substantially contemporary investigations of black economic status. Abrams Harris Jr. earned a doctorate in economics from Columbia University in 1930 and was on the faculty of Howard University for almost two decades. He later joined the faculty at the University of Chicago. His most famous work is the classi c The Negro as Capitalist (Harris 1936). Darity (1997) notes that, overall, Harris made significant contributions to economic anthropology, black studies, institutional economics, and the history of thought. In more recent years, David Swinton, the current president of Benedict College (an HBCU) and one of the top 20 black economists in terms of citations (Medoff 1989), was a longtime faculty member in economics at Clark-Atlanta University and Jackson State University--both of which are HBCUs.

Given the history of HBCUs, it would not be an overstatement to say that they have a legacy of contributing to the stock of economics knowledge and to the supply of black economists. Below, our analysis will perhaps reveal the extent to which this legacy manifests itself regarding contemporary research and scholarship and how contemporary research at HBCUs compares to that of non-HBCUs. Our analysis of the relationship between faculty economics research and the number of an institution's baccalaureate graduates who earn doctorates in economics will reveal the extent to which the continuation of a research legacy at HBCUs has an impact on the supply of black Ph.D. economists.

3. Data and Simple Comparisons

Research output in economics was measured for the 39 nonresearch HBCUs with faculty listings in The 1996 Prentice Hall Guide to Economics Faculty (Hasselback 1996). We restrict our analysis to nonresearch colleges and universities because less than 3% of all HBCUs have Carnegie classifications above the Masters level. [8] Additionally, a random sample of 101 non-HBCUs was chosen from the population of all colleges and universities in the continental United States with Carnegie classifications of Masters (Comprehensive) Colleges and Universities 1, Masters (Comprehensive) Colleges and Universities 2, Baccalaureate (Liberal Arts) Colleges 1, and Baccalaureate Colleges 2. [9] Research output for economics faculty at these 101 non-HBCUs was also measured by using the faculty listing for these institutions in the 1996 Prentice Hall Guide. Together, this group of 39 HBCUs and 101 non-HBCUs constitute the sample observations for our data set. A complete listing of the sample institutions can be found in Appendix A.

This paper utilizes three measures of research output in the economics discipline. [10] The first is based on the number of articles and notes produced between 1983 and 1997 that are indexed in the EconLit database by faculty members listed in the 1996 Prentice Hall Guide for each institution (TOTJEL). This database is available on CD-ROM and includes all journals cataloged in JEL. Moreover, this measure is an accepted gauge of research output in the discipline of economics. The second measure is a count of the number of JEL-indexed articles produced by the faculty of each economics department between 1983 and 1997 that is listed in Laband and Piette's (1994) ranking of the quality of 130 economics journals on the basis of citations per article (TOTQA). The third measure also utilizes the Laband and Piette measure. However, instead of counting the number of ranked journal articles, we actually assign a modified version of Laband and Piette's weights on the basis of quality. Specifically, for a given economic s department, total quality adjusted publications is defined as [sigma] [[w.sup.1/2].sub.i], where [w.sub.i] is the weight assigned to each article produced by economics faculty according to the ranking of the journal in which the article was produced (WTQA). [11] Thus, the latter two measures capture the production of quality research at these institutions, with the second measure simply counting the number of quality articles produced by a department, whereas the third measure assigns quality weights to articles produced.

Because the research output of an economics department over a 15-year period (1983- 1997) is measured for the faculty listed in the Prentice Hall Guide for a single year (the year 1996), our measures of research output do not necessarily indicate that faculty members were affiliated with their institution at the time of publication. Moreover, this is a standard shortcoming of studies of research output in economics. However, it could be argued that even if individuals produced all or a portion of their research output before their employment at their 1996 institution, that output is a reflection of the institution's goals in research output because the institution made the decision to hire that individual at least partially on the basis of their research record.

Table 1 reports the descriptive statistics for our measures of research output and other variables that may determine research output. Variable definitions, along with sources, are included in Appendix B. On average, HBCUs produced 2.49 JEL-indexed articles between 1983 and 1997 (TOTJEL), whereas the average for non-HBCUs is 9.71 articles. For both HBCUs and non-HBCUs, the standard deviation of JEL output is almost twice the means. These preliminary statistics indicate that, on average, economics faculty at HBCUs produce less research output as compared to their non-HBCU counterparts. However, the comparatively large standard deviations of JEL output for each group (as compared to the means) illustrate the variation in output within each group. The other measures of research output tell a similar story. The total quality adjusted measure (TOTQA) is 1.04 and 4.97 for HBCUs and non-HBCUs, respectively, whereas the weighted quality adjusted measure (WTQA) is 0.27 and 1.68, respectively.

An interesting question that arises is, why do economics faculty at HBCUs produce less research? Is it something intrinsic about black colleges, or is it something about the characteristics of these institutions or their economics faculty that is found in greater (or lesser) abundance among the HECUs? Examining characteristics of HBCUs and the economics faculty at these institutions relative to non-HBCUs may provide some insights into these questions. Table 1 presents complete definitions of all variables, their sources, and the descriptive statistics for these variables separately for HBCU and non-HBCU institutions.

On average, faculty at HBCUs receive their doctorates from lower-ranked institutions (PEDIGREE). Indeed, the average institutional rank for faculty at HBCUs is 68.57 as compared to 56.35 for non-HBCUs (recall that in rankings, the lower the number, the better). However, past research indicates that quality of degrees has little influence on scholarly output at liberal arts schools--another group of nonresearch institutions (Bodenhorn 1997). Other distinguishing differences between HBCUs and non-HBCUs include lower average experience (AEXP) of faculty at HBCUs, lower average salaries (SALARY), and slightly lower student quality (SQUAL). Also, average research expenditure (RESEXP) at HBCUs is more than nine times that at non-HBCU institutions ($4,069,000 and $444,000, respectively). However, because this broadly defined measure is inclusive of government grants and contracts, private gifts, grants, and contracts for specific research projects and training programs, this type of expenditure may not directly res ult in refereed publications. [12]

The descriptive statistics also reveal a shortcoming of using this data set to measure the difference in research output at HBCUs and non-HBCU institutions. Economics faculty at HBCUs are less likely to be housed in departments in which economics is the sole discipline. Indeed, 30% of HBCU departments are economics only (ECON_ONLY) as opposed to 46% of non-HBCUs. If the multidisciplinary faculty of HBCU economics departments are more likely to publish in areas outside of economics, then our measures of research output would understate research activity in these departments. On the other hand, many of the multidisciplinary departments in our sample are economics and finance departments. Because many finance journals are JEL indexed, the fact that many HBCU economics departments are housed in multidisciplinary units may inflate our measures of research output for these institutions.

A second shortcoming of using this data set to measure differences in research output at the two types of institutions is that it does not allow the measurement of teaching loads for faculty at these institutions. Moreover, if faculty at HBCUs have disproportionately larger teaching loads as compared to their counterpart faculty at non-HBCUs, then teaching at HBCUs may crowd out research activity and may partially explain the difference in research output at the two types of institutions. Despite these shortcomings, an analysis of JEL research output at teaching institutions by utilizing a sizeable subsample of HBCUs provides an initial assessment of the difference in research output at HBCUs and non-HBCUs. Further, by exploring the relationship between research output and teaching effectiveness--as measured by the number of economics doctorates produced--this analysis reveals a link between teaching and research at HBCU and non-HBCU teaching institutions. The next section presents Poisson regression analyse s that control for institutional and faculty characteristics in the measurement of the difference in research output at HBCU and non-HBCU institutions.

4. Determinants of Research Output

We first examine the influence of faculty and institutional characteristics on the research output of economics departments at nonresearch institutions. In contrast to the double log regression approach of previous analysis (Bodenhorn 1997), we estimate research output equations via Poisson regressions. Two reasons motivate and justify a Poisson regression approach (Chappell, Kimenyi, and Mayer 1990). First, because the observations on research output--counts of articles--are integer valued, an econometric framework based on a discrete probability distribution is more appropriate. Second, a double logarithmic regression specification requires either excluding zero observations or incorporating ad hoc procedures to accommodate the zero values. [13] A Poisson specification for research output allows zero to be a natural outcome of the data generating process without any ad hoc adjustments on the zero observations.

To estimate the impact that institutional and faculty characteristics have on an institution's research output, the number of an economics department's publications is viewed as being generated by a Poisson probability distribution:

Prob([R.sub.ij] = n) = [exp{-[[lambda].sub.i]}[[[lambda].sup.n].sub.i]]/n! (1)

Where [R.sub.ij], a discrete random variable that measures research output for institution i using the jth measure of research (j = TOTJEL, TOTQA, or WTQA), n = 0, 1, 2,..., N, exp = 2.71828, and [[lambda].sub.i] = the expected value [14] and variance of [R.sub.ij].

Our research output specification allows the testing for a different level of research at HBCUs relative to non-HBCUs. Specifically, a Poisson regression model is formulated by specifying [[lambda].sub.i] as a function of a vector of exogenous variables:

ln [[lambda].sub.i] = [beta]X + [alpha]([HBCU.sub.i]) + [epsilon], (2)

Where X is a matrix of institutional and faculty attributes that include the average salary at the institution (SALARY), the number of library volumes at the institution (LIBVOL), the dollar amount of research expenditures at the institution (RESEXP), a dummy for institutions that are private (PRIVATE), a dummy for institutions that offer a master's degree or higher (MA), a student quality ranking index (SQUAL), a dummy for departments that are solely economics (ECON_ONLY), the number of faculty members in the department in which economics is housed with earned doctorates (NFAC), and a regional dummy (NORTHEAST) [15] Faculty attributes include the average quality rank of the institution from which economics faculty received their degrees (PEDIGREE), average faculty experience (AEXP), and its square, ([AEXP.sup.2]). The variable HBCU is a dummy which takes on the value 1 if the institution is an HBCU and 0 otherwise. Its coefficient [alpha] captures the difference in the expected research output of economics f aculty at HBCUs and non-HBCUs, holding faculty and institutional characteristics constant.

Estimation of the parameters is obtained maximizing the following log-likelihood function with respect to [beta], where [R.sub.ij] = R:

L([beta]) = [-[[lambda].sub.i] + [beta]XR - ln R!] (3)

The first-order condition in [beta] is nonlinear and is suitable for estimation via maximum likelihood or an iterative nonlinear least squares procedure. [16]

Table 2 reports the results for each measure of research output. For each specification, we also report the coefficient v, a test for mean-variance equality--a restriction required for a Poisson random variable--and Pseudo-[R.sup.2] as a goodness of fit measure. [17] In each specification reported, v is insignificant, suggesting the adequacy of a Poisson specification for the dependent variable. The first column details the results with TOTJEL as the dependent variable. Our findings suggest that having a large economics department (NFAC) significantly increases an institution's expected number of JEL-indexed publications produced. Also, private institutions (PRIVATE) located in the northeast region of the United States (NORTHEAST) with higher student quality (SQUAL) and more library volumes (LIBVOL) have significantly greater expected JEL output. This is indicated by the large and highly significant coefficients on these variables. The negative and highly significant coefficient on the ECON_ONLY variable indi cates that economics departments housed in multidisciplinary departments have significantly greater expected JEL output. By evaluating regressors at mean values, teaching institutions that house their economics departments in single disciplinary units have an expected JEL output of 2.21 fewer JEL-indexed articles as compared to multidisciplinary departments. [18] Because the multidisciplinary departments in our sample are disproportionately inclusive of finance faculty, our findings may be capturing the additional publications of finance faculty in finance journals that are JEL indexed.

Additionally, we find that teaching institutions with faculty members from higher-ranked institutions (lower number) have significantly greater expected JEL output. This is indicated by the negative and highly significant coefficient on the PEDIGREE variable. These results differ from past results on the determinants of research output for individual economists at liberal arts institutions (Bodenhorn 1997). This difference may result from differences in research structure at liberal arts schools and all teaching institutions. Or it may stem from the fact that Bodenhorn examines research output of individuals, whereas this analysis examines research output by departments. Moreover, our findings may suggest that the effect of degree quality on research output on the entire department is more significant than the degree-quality effect on individual research output. In other words, the degree-quality effect on the whole may be larger than its effect on the sum of individual parts.

Our findings for AEXP and its square significantly indicate the concavity of the experience-research profile for economics departments at teaching institutions; that is, teaching institutions with economics faculty with more average experience have significantly larger JEL output. However, the coefficient on [AEXP.sup.2] is negative and slightly significant, indicating that expected research output tapers off at higher levels of average experience. Moreover, the central finding for this analysis is the coefficient on the HBCU variable. Economics departments at HBCUs produce significantly fewer expected EL-indexed articles as compared to departments at non-HBCUs. Indeed, the expected value of HBCU TEL output is 2.028 articles below that of non-HBCUs evaluated at regressor averages).

These findings are quite stable using the other measures of research output (see columns 2 and 3 of Table 2). We find that the expected number of quality articles at HBCUs is significantly (1.553) fewer articles than non-HBCU institutions (TOTQA equation), and these articles have significantly (0.15 percentage points) less weight (regressors are evaluated at average values). These findings support the notion that economics departments at HBCUs produce less research output as compared to departments at non-HBCUs. To check the validity of these results, alternative specifications were explored. [19]

It is important to note, however, that the research output gap between non-HBCUs and HBCUs decreased tremendously after controlling for institutional characteristics. Indeed, the gap in JEL output of 7.22 using raw data (see Table 1) decreased to 2.028 after controlling for institutional attributes, suggesting that 72% of the raw gap in EL output (TOTJEL) between HBCUs and non-HBCUs is due to institutional attributes. Other research measures reveal a similar story; 60.4% of the HBCU deficit in the number of quality articles (TOTQA) and 89.4% of the deficit in quality weight (WTQA) can be attributed to institutional attributes. This suggests that institutional factors at HBCUs work against faculty research and retard research output.

In the next section we examine the influence of research output on one measure of teaching effectiveness--the production of economics doctorates among an institution's baccalaureate graduates. If increased JEL-indexed articles and quality publications increase the probability that baccalaureate graduates of an institution earn doctorates in economics, then our findings of less faculty research output at HBCUs relative to non-HBCUs may suggest that HBCUs are not utilizing economics faculty research to an extent necessary for engendering an increase in the supply of black Ph.D. economists.

5. The Relationship between Faculty Research and the Doctoral Achievements of Students

Hartley and Robinson (1997) report a positive correlation between the research productivity of an economics department of a college or university and the number of its baccalaureate graduates who earn doctorates in economics. One interpretation of this finding is that faculty research productivity in economics manifests itself as a particular kind of teaching effectiveness--a faculty member who is an active researcher is more able to cultivate an interest and aptitude among undergraduates that induces them to pursue a Ph.D. in economics. In this context, a positive correlation between faculty research and the doctoral achievements of undergraduates suggests that faculty research and teaching effectiveness are not necessarily competing objectives for colleges and universities where teaching is the most important component of the institutional mission. Hartley and Robinson (1997) argue that because measures of teaching effectiveness are hard to compute, a readily available measure of teaching effectiveness for teaching colleges and universities is the number of students who graduate and go on to earn doctorates.

Several recent analyses report findings suggesting that faculty scholarship and teaching are mutually supporting activities. McCaughey (1993) reports a positive correlation between faculty research productivity and student teaching evaluations across several academic disciplines. Lindgren and Nagelberg (1998) report that in the case of law school professors, there is a positive relationship between scholarly citations and scores on teaching evaluations. Law school professors with a high number of citations are almost twice as likely to receive above-average teaching evaluations as professors with a low number of citations. Noser, Manakyan, and Tanner (1996) find that for economics professors at the undergraduate level, there is a positive and significant relationship between research activity and teaching effectiveness. These findings suggest that faculty research productivity also enhances teaching effectiveness and quality as perceived by students.

The idea that faculty research productivity in economics is linked to teaching effectiveness seems controversial. [20] However, it has been offered as a working hypothesis by Kasper et al. (1991) in the context of examining whether or not the decline in the number of graduates of liberal arts colleges who earned doctorates in economics can be explained by the fact that undergraduate faculty at these institutions no longer have the research aptitude to encourage undergraduates to become professional economists. An even stronger position on the necessity of faculty research for teaching effectiveness was articulated by the late Nobel Prize-winning economist George Stigler, who argued that "The teacher who does not publish must have the conscience of a saint if he is not to take things easy: to pontificate instead of to reason; to conjecture rather than to know" (Stigler 1963, p. 15).

A causal relationship between faculty research productivity and the doctoral achievements of baccalaureate graduates is of particular consequence for HBCUs. It has been estimated that only 1.2% of all doctorates in economics are held by black Americans (Ruffins 1996). As HBCUs are responsible for a disproportionate number of baccalaureate degrees awarded to black Americans, any causal mechanism between faculty research productivity in economics and the doctoral achievements of economics majors at HBCUs could provide an avenue to increase the supply of black Ph.D. economists. In this context, one partial explanation for the dearth of black Ph.D. economists could be the research productivity of HBCU economics departments. [21]

It is important to realize, however, that no sweeping generalizations can be drawn from this analysis about the efficiency of HBCUs in producing black doctorates, relative to non-HBCUs. This is the case because in the random sample of 101 non-HBCUs used for this analysis, only 12 black doctorates were produced, whereas in the nonrandom sample of 39 HBCUs, 55 black doctorates were produced. However, in reality, HBCUs account for roughly 25% of black doctorates produced over this period. Thus, our data set would misleadingly indicate that HBCUs produce a substantially larger fraction of black doctorates in the United States as compared to non-HBCUs. As a consequence, this data set is not appropriate for evaluating the production of black doctorates at HBCUs relative to non-HBCUs. However, it is quite appropriate for examining the production of doctorates in general at teaching institutions. Further, it is reasonable to think of the baccalaureate graduates of HBCUs who earn doctorates as blacks, given that our data indicate that this is the case more than 95% of the time.

A positive correlation between the research productivity of faculty and the graduate school achievements of students suggests a functional relationship in which the number of an institution's baccalaureate graduates who earn doctorates in economics is a function of faculty research productivity at the institution from which the students graduated. Because the number of doctorates earned by an institution's baccalaureate graduates is integer-valued, as in the case of an institution's research output, we view it as being generated by a Poisson probability distribution:

Prob([TPHD.sub.i] = n) = [exp{-[[lambda].sub.i]}[[[lambda].sup.n].sub.i]]/n! (4)

where [TPHD.sub.i], a discrete random variable, is the number of doctorates in economics earned by baccalaureate graduates of institution i, n = 0, 1, 2, ... , N, exp = 2.71828, and [[lambda].sub.i] = expected value and variance of [TPHD.sub.i].

The corresponding Poisson regression model is formulated by specifying [[lambda].sub.i] as a function of exogenous variables, as follows:

ln [[lambda].sub.i] = [beta]X + [epsilon], (5)

where X is a vector of exogenous variables; the vector of coefficients [beta] captures the impact of these variables on the mean number of Ph.D. economists produced by an institution. Included in X are the estimated values of the expected research output of institution i (estimated in Equation 2) for the jth measure of research (j = TOTJEL, TOTQA, and WTQA). It is included as a regressor because the number of baccalaureate graduates who earn doctorates is recursively related to research output. Its coefficient captures the impact of the expected level of the jth measure of research produced by a department on the expected number of doctorates produced by that department.

Other variables used in the Poisson regressions are reported in Table 1, along with their means and standard deviations. [22] The dependent variable is the total number of doctorates earned between 1980 and 1996 by an institution's baccalaureate graduates. Although this temporal window is seemingly arbitrary to the extent that the decision of an institution's baccalaureate graduates to pursue a doctorate in economics is subject to short-run fluctuations, a 16-year exit window allows a more steady state perspective of the flow of HBCU and non-HBCU baccalaureate graduates out of economics doctorate programs.

In Table 3, the results of three Poisson regressions with a specification similar to that of Hartley and Robinson (1997) are reported in columns 1, 3, and 5. The three specifications control for student quality, faculty quality, and the number of faculty in the economics department. The specifications differ only with respect to the measure of faculty research productivity used. The insignificance of v for the specifications in Table 3 suggests the adequacy of a Poisson specification for the dependent variable.

In each case, the coefficient on the research productivity measure is positive and significant. This indicates that for the teaching colleges and universities in the sample, research productivity of economics faculty significantly increases the expected number of an institution's baccalaureate graduates who earn doctorates in economics; this is true regardless of the measure of research productivity used. These findings are consistent with the notion that research and teaching effectiveness--as measured by the number of students who get doctorates--are mutually supportive activities at teaching institutions.

The student and faculty quality variables also behave as one would expect. The positive and significant sign of SQUAL suggests that as measured student quality increases at a given institution, the effect is an increase in the expected number of an institution's baccalaureate graduates who earn doctorates in economics. PEDIGREE is negative and significant, suggesting that as the average quality ranking of the program from which Ph.D. faculty earned their doctorates gets smaller (that is, as ranking gets closer to number one), the effect is to increase the expected number of baccalaureate graduates from an institution who earn doctorates in economics.

In another specification, we included an HBCU dummy and a variable for the interaction of HBCU and the jth research measure j (j = TOTJEL, TOTQA, and WTQA). The coefficient on the HBCU variable captures differences between HBCUs and non-HBCUs in the expected number of doctorates produced. The coefficient on the interaction of HBCU and research capture HBCU and non-HBCU differences in the impact of faculty research on the expected number of doctorates produced. The results of this specification are reported in columns 2, 4, and 6 for our three research measures. The HBCU dummy is positive but insignificant in all three specifications. This suggests that the expected number of undergraduates at HBCUs that earn doctorates is no different than that of undergraduates at non-HBCUs, all things being equal. Apparently, there is no difference between HBCUs and non-HBCUs in the production of doctorates. Also, the coefficient on the interaction term is insignificant in all three specifications. These findings suggest t hat the returns to faculty economics research--the payoff to faculty research in terms of the expected number of baccalaureate graduates who earn economics doctorates--is independent of institutional type (HBCU or non-HBCU).

In general, the results reported in Table 3 lend support to the idea that the graduate school achievements of economics majors is directly influenced by the research productivity of economics faculty. Hence, these results provide further evidence that at least some teaching information is contained in faculty research activity (Hartley and Robinson 1997). For HBCUs, the reported results have a special implication. Any interventions at HBCUs that create incentives for more faculty research in economics would have the effect of enhancing teaching effectiveness and increasing the supply of black Ph.D. economists.

6. The Production of Black Economics Doctorates at HBCUs

The above findings suggest that economics departments at HBCUs produce less research output as compared to non-HBCUs and that research output of economics departments at these teaching institutions directly influences the production of doctorates. Further, we find no significant difference in HBCUs and non-HBCUs with regard to the impact of economics faculty research productivity on the production of economics doctorates.

To cast insight into the ways in which HBCUs can increase the number of black economics doctorates, we perform simulations to explore the expected number of black doctorates who would have been produced at HBCUs between 1980 and 1996 under three scenarios. The first is the expected number of black doctorates produced at HBCUs given the average characteristics of HBCUs and the research structure of HBCUs. Empirically, this implies estimating the expected research equation (Equation 2) for the sample of HBCUs (excluding the HBCU dummy), then putting the fitted values of research output at HBCUs into a production of black doctorates equation for the sample of HBCUs (Equation 5 with the number of black HBCU graduates who have earned economics doctorates as the dependent variable [BPHD]). [23] These estimates answer the following question: what is the expected number of black economics doctorates produced at an HBCU over this period, given the attributes and research structure at these institutions?

These estimates can be compared to the simulated expected production of black doctorates at HBCUs over this period under the second scenario, given the average characteristics of HBCUs and the research structure of non-HBCUs. This implies estimating the expected research equation for the sample of non-HBCUs (Equation 2, excluding the HBCU dummy), then putting the fitted values of research output at HBCUs--if they had the non-HBCU research structure--into a production of black doctorates equation for the sample of HBCUs (Equation 5 with BPHD as the dependent variable). This scenario is interesting because it allows us to constrain the attributes of HBCUs (this includes faculty salaries, student quality, research expenditure, library volumes, etc.) to their current levels while imposing the non-HBCU environment of research production on HBCUs to examine the number of black doctorates that would have been produced at HBCUs over this period. Moreover, this scenario allows us to answer the question, what is the e xpected number of black doctorates produced at HBCUs if HBCUs converted resources to research in a manner similar to non-HBCUs?

In the final scenario, we examine the number of black doctorates who would have been produced at HBCUs over this period given the average research output of non-HBCUs and the HBCU mean value of all other attributes. These estimates are found by estimating the expected number of black doctorates produced at HBCUs evaluated at the average research output of non-HBCUs (for each of our three measures of research output). This specification allows us to answer the question, what would have been the number of black doctorates produced at HBCUs if they had the average research output of non-HBCUs? These three scenarios are estimated by using each of our three measures of research output (TOTJEL, TOTQA, and WTQA).

The results are summarized in Table 4. The first row details the expected number of black doctorates produced at an HBCU, given the average characteristics of HBCUs and utilizing our three measures of research output. The expected number of black doctorates produced at an HBCU over this period, given the attributes of these institutions, is between 1.19 and 1.23 economics doctorates. However, if HBCUs converted resources to research in a manner similar to non-HBCUs, then we would expect between 1.22 and 1.25 black doctorates produced at each institution (scenario 2, second row of Table 4). This would have increased the expected number of black doctorates produced by an HBCU by between .022 and .034 percentage points, depending on the research output measure used. For the 39 HBCUs on our sample, collectively, this would have resulted in an increase in black doctorates of between 1 and 1.3 over this 16-year period. These findings suggest that HBCUs could utilize their attributes differently in the production o f research and, consequently, modestly increase the number of doctorates produced.

The scenario in row 3 indicates that if HBCUs increased economics research output to the average levels of non-HBCUs, the expected production of black doctorates at an HBCU would increase between .124 and .249 percentage points. For the 39 HBCUs in our sample, this would translate into an overall increase of between 5 and 10 black doctorates. Given that in the nation, a total of 307 doctorates in economics were awarded to black Americans over this period (Collins 1999), our simulations suggest that the 39 HBCUs in our sample could have increased black Ph.D. production in the nation by 5% over this period.

The simulations in Table 4 provide a basis for speculating about the impact of increasing faculty research in HBCU economics departments for all HBCUs that award baccalaureate degrees. Our sample includes only 39 HBCUs, but excluding Howard University, medical institutions, graduate divinity programs, and community colleges, there are 87 HBCU teaching institutions in total. If the results above hold for the HBCUs not in our sample, and if the economics faculty at these HBCUs averaged roughly two JEL-indexed publications every three years--the average for the non-HBCUs in our sample--approximately 129 additional black Ph.D. economists would have been produced under scenario 3 in Table 4. This is approximately a 42% increase in black economics doctorates produced over the period. [24]

This analysis provides simulations of the production of black Ph.D. economists at HBCUs, given non-HBCU research structure and research output. However, caution should be exercised in interpreting these results. Indeed, it is much easier to give HBCUs the research structure or output of non-HBCUs in a regression exercise, as opposed to performing this task in reality. Moreover, given HBCUs unique mission of educating students whose likelihood of college completion would be dramatically less if they attended a non-HBCU, it is not certain that all aspects of an HBCU acquisition of non-HBCU structure or research output would result in a welfare improvement--especially, say, if increased economics faculty research intensity at HBCUs reduces activities that otherwise increase the probability of college completion for HBCU students. Notwithstanding, these simulations provide an instructive starting point for understanding the potential of economics faculty research at HBCUs in producing black economics doctorates.

7. Conclusion

This paper utilizes three measures of research output to examine whether there is a difference in research output of economics departments at HBCU and non-HBCU teaching institutions. After controlling for faculty and institutional characteristics, our research output specifications allow the determination of whether economics faculty research productivity is significantly different at the two types of institutions. A second objective of this analysis is to utilize a data set in which HBCUs are well represented to determine the causal link between research productivity and a particular type of teaching effectiveness--the number of baccalaureate graduates that earn doctorates in economics.

Our findings suggest that HBCUs are different from non-HBCUs; that is, after controlling for faculty and institutional characteristics, economics departments at HBCUs produce significantly less research output relative to non-HBCUs. However, institutional attributes account for between 60 and 89% of the raw gap in research output between the two types of institutions, depending on the research measure used. Although our data set does not allow the determination of all of the environmental factors that may explain the remaining HBCU discrepancy in research output (after controlling for observed institutional factors), some factors which may hamper research activity at HBCUs may include large teaching loads, extensive office hours, and a lack of technical support and physical infrastructure required for research. [25]

Additionally, our results suggest that there are indeed payoffs to be reaped from an increase in economics faculty research at teaching institutions. It is in this context that HBCUs and non-HBCUs are similar. We find that an increase in economics faculty research productivity increases the number of an institution's baccalaureate graduates who earn doctorates in economics. Utilizing the doctoral achievements of an institution's undergraduates as a measure of teaching effectiveness, our results suggest that teaching effectiveness in economics and economics faculty research are mutually supporting activities.

There are some limitations of our analysis. First, we are not sure if the omission of some HBCUs from The Prentice Hall Guide introduces selection bias. [26] Of the 87 HBCU teaching institutions, only 39 had published rosters in the Prentice Hall directory. To the extent that the HBCUs in our sample provided roster information because the faculty members are predisposed toward publishing in refereed journals, the parameter estimates are biased. A second source of bias could be that our measures of research are narrowly confined to economics journals. In our sample, relative to non-HBCUs, HBCUs were more likely to have their economics departments housed in multidisciplinary units. To the extent that economics faculty at HBCUs are more likely to publish in refereed journals that are not JEL indexed, the parameter estimates of research output at these institutions may be downwardly biased.

The general tenor of our results suggests that the research gap that exists between HBCUs and non-HBCUs is important to the extent that it has consequences for the supply of black Ph.D. economists. Indeed, our policy simulations indicate that an increase in the research output of economics faculty at HBCUs would result in more black Americans earning doctorates in economics. Thus, any interventions that increase the capability of HBCU economics faculty to publish in refereed journals would yield a social benefit in the form of more black Ph.D. economists. Interventions suggested by our results include increased library volumes and hiring faculty with more experience and who have earned doctorates from higher-ranked institutions. [27] Notwithstanding the implications for HBCUs, our results should not be interpreted to suggest that non-HBCUs are not capable or should not be in the business of producing black Ph.D. economists--we find that the Ph.D. return to faculty research is independent of institutional typ e. Nonetheless, given the high concentration of black students at HBCUs and the dearth of black Ph.D. economists, our results suggest that HBCUs could be a strategic resource in enhancing the nation's pipeline of black Ph.D. economists.

(*.) Division of Finance and Economics, Marshall University, Huntington, WV 25755, USA; E-mail agesaj@marshall.edu; corresponding author.

(+.) Department of Economics, North Carolina A&T State University, Greensboro, NC 27411, USA; E-mail grangerm@ncat.edu.

(++.) Department of Economics, North Carolina A&T State University, Greensboro, NC 27455, USA; E-mail price@jade.ncat.edu.

During a part of this project, Agesa was a postdoctoral fellow at the University of North Carolina (UNC). She thanks the Ford Foundation for financial support and UNC for the hospitality. We are grateful to James Peoples, Alvin Headen, two anonymous referees, and the editor of this journal for valuable comments on an earlier draft. A preliminary version of this paper was presented at the 1999 Allied Social Sciences Association meetings in a joint session with the National Economic Association. We thank session participants for their comments. The usual disclaimer applies.

Received February 1999; accepted October 1999.

(1.) Hartley and Robinson (1997) ranked schools classified as national liberal arts colleges in U.S News and World Report (America's Best Colleges 1995) according to four measures of research productivity in economics. Spelman College had a rank no higher than 71, with a total of three published articles indexed in the Journal of Economic Literature (JEL), and Morehouse College had a rank no higher than 100, with a total of one JEL author-adjusted article.

(2.) Earlier analyses of the supply of black economists and HBCU economics programs have been provided by Jones (1988) and Simms and Swinton (1988).

(3.) The Supreme Court established that "separate" accommodations for white and nonwhite races were constitutional in Plessy v. Ferguson. See 163 U.S. 537 (1986).

(4.) See 347 U.S. 483 (1954).

(5.) The analysis by Oak (1938) provides an interesting historical perspective on economics education at HBCUs. Although the author does not indicate how many economics majors were produced by HBCUs during their early years, he does reveal the extent to which economics faculty in these programs were engaged in scholarship. For the 18 HBCUs in the sample that the author surveyed, only three economics professors from two HBCUs (Fisk University and Howard University) were found to have had any publications.

(6.) Source: IPEDS data at WEBCASPAR, http://www.nst.gov/srs/srsdata.htm.

(7.) Collins (1999) provides data showing that between 1980 and 1995, 286 economics doctorates were earned by black Americans.

(8.) Although Howard University is an HBCU that is listed in the 1996 Prentice Hall Guide, it was not included in this study because of its Carnegie classification as a Research I Institution.

(9.) The random sample was formed by assigning a unique number to each of the colleges and universities with a Carnegie classification of Baccalaureate or Masters level (there are over 900 institutions in these categories). Next, the random number generator from the SAS statistical package was used to generate the sample of 101 non-HBCUs.

(10.) For each measure of research output, each coauthor of multiple authored articles was given credit for one article.

(11.) The purpose of this weighting scheme is to reduce the excessive weight placed on the highest ranked journals. For instance, using Laband and Piette's ranking based on citations per article, one Journal of Economics Literature article has the weight of ten Southern Economic Journal articles. Moreover, by taking the square root of journal weights before summing by department, the ordinal ranking of journals is maintained, but the inordinate weight assigned to the highest ranked journals is reduced. This weighting scheme was used by Bodenhorn (1997) in his analysis of economics research at elite liberal arts schools.

(12.) Another reason the average research expenditures for the HBCUs in the sample dramatically exceeds that of the non-HBCUs is that 15 of the HBCUs are Land Grant Universities (LGUs), whereas none of the non-HBCUs are. LGUs receive considerable annual appropriations to fund agricultural extension research. Excluding the HBCUs that are LGUs, average research expenditures for the HBCUs decrease by approximately 50%, and the difference in average research expenditures between the HBCUs and non-HBCUs in the sample becomes less dramatic.

(13.) Bodenhorn (1997), for example, adds 0.01 to each zero observation before converting the research output measures to logarithms.

(14.) The weighted quality adjusted measure (WTQA) was made integer-valued by multiplying the quality weights (WTQA) by 100 and rounding to whole numbers.

(15.) Given the small sample size, it was not feasible to divide regions into more than two categories (the categories used were northeast or not). Moreover, this demarcation of region has been used in a previous analysis of research output in economics (Bodenhorn 1997).

(16.) See chapter 43 of LIMDEP Version 6.0: User's Manual and Reference Guide (1992).

(17.) Specifically, if dependent variable [R.sub.ij] has a Poisson distribution, in a Poisson regression the mean of dependent variable [R.sub.ij] is E[[R.sub.ij]] = [[lambda].sub.i] = [lambda]([X.sub.i], [beta]). Cameron and Trivedi (1990) show that a test for mean-variance equality is based on the hypothesis test [H.sub.0]: var([R.sub.ij]) = [[lambda].sub.i] versus the alternative [H.sub.a]: var([R.sub.ij]) [[lambda].sub.i] + vg([[lambda].sub.i]), where g([[lambda].sub.i]) is a function specified to equal 1, [[lambda].sub.i], or [[[lambda].sup.2].sub.i]. A test for mean-variance equality is a t-test for the significance of v in the auxiliary regression [sigma] [w.sub.i]g([[lambda].sub.i])[{([R.sub.ij] - [[lambda].sub.i]).sup.2] - [R.sub.ij] - vg([[lambda].sub.i])} = 0, where [sigma][w.sub.i]g([[lambda].sub.i]) is a weight based on a consistent estimate of [beta]-the fitted value, for example. If the t-test on the coefficient of v is insignificant, the null hypothesis of mean-variance equality cannot be rejec ted, suggesting the adequacy of a Poisson specification for the dependent variable. As a goodness of fit measure, we report the Pseudo-[R.sup.2] of McFadden (1974), defined as [R.sup.2] = 1 - (log [L.sub.[omega]])/log [L.sub.[omega]]) where [L.sub.[omega]] is the maximum of the likelihood function when maximized with respect to all parameters, and [L.sub.[omega]] is the maximum of the likelihood function when maximized with respect to a constant term only.

(18.) The marginal effect of the kth variable on the expected research output is estimated using [[beta].sub.k][e.sup.[[beta].sub.x]], where it is the vector of means of matrix X.

(19.) In addition to the specification described above, alternative specifications were run in which the HBCU dummy was interacted with the other independent variables. In each instance, the interaction coefficient was small and insignificant. This suggests that the influence of any institutional attribute did not significantly differ between HBCUs and non-HBCUs.

(20.) For contrary views on the relationship between faculty research activity and teaching, see Sykes (1988), Merritt (1998), and Shapiro (1998).

(21.) This presupposes, of course, that the lack of black Ph.D. economists is simply a "pipeline problem." Turner, Myers, and Creswell (1999) argue that there are other factors, such as salaries and racial or ethnic intolerance, that result in the underrepresentation of black Americans in academia.

(22.) The research productivity measures were defined previously.

(23.) BPHD is the number of black baccalaureate graduates of an HBCU who earned a doctorate in economics over the 1980-1996 period as reported in the National Science Foundations' Doctorate Records File. SOURCE: National Science Foundation Computer-Aided Science Policy and Research Database (WEBCASPAR) located on the Internet at http:/www.nsf.gov/srs/srsdata.htm.

(24.) Collins (1999) provides data showing that between 1980 and 1996, 307 economics doctorates were earned by black Americans.

(25.) Some of these factors were identified by Hudgins (1994) as determinants of HBCU research productivity in sociology.

(26.) In a telephone conversation with James Hasselback, he revealed that the primary reason that economies departments are not listed in The Prentice Hall Guide is failure to respond to his survey (after multiple requests).

(27.) Although the average number of library volumes for HBCUs in our sample is greater than the average for non-HBCUs, the variance in library volumes for HBCUs is greater than for non-HBCUs (see Table 1). This may indicate that some HBCUs have well-stocked libraries, while others have very bleak stock.

References

Agesa, Jacqueline, Maury Granger, and Gregory N. Price. 1998. Economics research at historically black colleges and universities: Rankings and effects on the supply of black economists. Review of Black Political Economy 25:41-54.

America's best colleges. 1995. U.S. News & World Report 119:108-41.

Bodernhorn, Howard. 1997. Teachers and scholars too: Economic research at elite liberal arts colleges. Journal of Economic Education 28:323-36.

Cameron, Colin A., and Pravin K. Trivedi. 1990. Regression-based tests for overdispersion in the Poisson model. Journal of Econometrics 46:347-64.

Chappell, William F., Mwangi S. Kimenyi, and Walter J. Mayer. 1990. A Poisson probability model of entry and market structure with an application to U.S. industries during 1972-77. Southern Economic Journal 56:918-27.

Collins, Susan, M., 1999. Report of the committee on the status of minority groups in the economics profession. American Economic Review 89:488-91.

Darity, William Jr. 1997. Soundings and silences on race and social change: Abram Harris Jr. in the Great Depression. In A different vision. African American economic thought. Volume 1, edited by Thomas D. Boston. New York: Routledge, pp. 230-49.

Du Bois, W E. B. 1994. The souls of black folk. New York: Gramercy.

Goldberger, Marvin L., Brendan A. Maher, and Pamela Ebert Flattau, eds. 1995. Research-doctorate programs in the United States: Continuity and change. Washington, DC: National Academy Press.

Harris, Abram L., Jr. 1936. The negro as capitalist: A study of banking and business among American negroes. Philadelphia: The American Academy of Political and Social Science.

Hartley, James E., and Michael D. Robinson. 1997. Economic research at national liberal arts colleges: School rankings. Journal of Economic Education 28:337-49.

Hasselback, James R. 1996. The 1996 Prentice Hall guide to economics faculty. Upper Saddle River, NJ: Prentice Hall.

Hudgins, John L. 1994. The segmentation of southern sociology? Social research at historically black colleges and universities. Social Forces 72:885-93.

Jones, Barbara A. P. 1988. NEA presidential address: Economics programs at historically black colleges and universities. Review of Black Political Economy 16:5-14.

Kasper, Hirschel, Geoffrey Woglom, Vernon Dixon, David Colander, Roger Kaufman, Bernard Saffran, Mark Kupperberg, Carolyn Shaw Bell, Gary Yohe, and Michael McPherson. 1991. The education of economists: From undergraduate to graduate study. Journal of Economic Literature 29:1088-109.

Laband, David N., and Michael J. Piette. 1994. The relative impacts of economics journals: 1970-1990. Journal of Economic Literature 32:640-58.

LIMDEP Version 6.0: User's manual and reference guide. 1992. Bellport, NY: Econometric Software Inc.

Lindgren, James, and Allison Nagelberg. 1998. Are scholars better teachers? Chicago-Kent Law Review 73:823-35.

McCaughey, Robert A. 1993. But can they teach? In praise of college professors who publish. Teachers College Record 95:242-57.

McFadden, Daniel. 1974. The measurement of urban travel demand. Journal of Public Economics 3:303-28.

Medoff, Marshall. 1989. The relative quality of black economists. Review of Black Political Economy 18:81-5.

Merritt, Deborah J. 1998. Research and teaching on law faculties: An empirical exploration. Chicago-Kent Law Review 73:765-821.

Noser, Thomas C., Herman Manakyan, and John R. Tanner. 1996. Research productivity and perceived teaching effectiveness: A survey of economics faculty. Research in Higher Education 37:299-323.

Oak, V. Vishnu. 1938. Evaluation of business curricula in negro colleges. Journal of Negro Education 7:19-31.

Peterson's guide to four year colleges. 28th edition. 1998. Princeton, NJ: Peterson's Guides.

Price, Gregory N. 1998. Black colleges and universities: The road to Philistia? Negro Educational Review 49:9-21.

Ruffins, Paul. 1996. Black economists: An elite clan of warrior intellectuals. Black Issues in Higher Education 13:18-24.

Shapiro, Fred R. 1998. They published, not perished, but were they good teachers? Chicago-Kent Law Review 73:835-45.

Simms, Margaret C., and David H. Swinton. 1988. A report on the supply of black economists. Review of Black Political Economy 17:67-88.

Stewart, James B. 1997. George Edmund Haynes and the office of negro economics. In A different vision: African American economic thought. Volume 1, edited by Thomas D. Boston. New York: Routledge, pp. 213-29.

Stigler, George. 1963. The intellectual and the market place and other essays. London: Collier-Macmillan.

Sykes, Charles J. 1988. Profscam: Professors and the demise of higher education. Washington, DC: Regnery and Gateway.

Turner, Caroline S. V., Samuel L. Myers, Jr., and John W. Creswell. 1999. Exploring underrepresentation: The case of faculty of color in the midwest. The Journal of Higher Education 70:27-59.

Appendix A

List of Colleges and Universities Included in the Sample

Alabama [a] A&M State University

Alabama [a] State University

Albany [a] State College

Alcorn [a] State University

Alma College

Aquinas College

Athens State College

Augsburg College

Augusta College

Augustana College (IL)

Augustana College (SD)

Austin Peay State

Belmont University

Benedict [a] College

Betbune [a] Cookman College

Bloomfield College

Brescia College

Cameron University

Campbell University

Carroll College

Carson-Newman College

Centenary College

Central [a] State University

Cheyney [a] University

Clark [a] Atlanta University

Colorado College

CUNY-Lehman

CUNY-Staten Island

David Lipscomb University

Delaware [a] State University Depauw University

Dillard [a] University

Drake University

Elmhurst College

Emory and Henry College

Fayetteville [a] State University

Ferris State University

Fisk [a] University

Florida [a] A&M University

Florida Southern College

Fort [a] Valley State College

Frostburg State University

Gannon University

Geneva College

Georgia College

Grambling [a] State University

Grand Canyon University

Greensboro College

Grinnell College

Hampton [a] University

Hartwick College

Hendrix College

Illinois Wesleyan University

Indiana Wesleyan University

Jackson [a] State University

Jacksonville University

Jersey City State College

Kentucky [a] State University

Kenyon College

Kings College

Lafayette College

Lamar University (Beaumont)

Le Moyne College

Lewis and Clark College

Lincoln [a] University (MO)

Lincoln [a] University (PA)

Louisiana State University (Shreveport)

Lycoming College

Marietta College

Marist College

Mcneese State College

Merrimack College

Middlebury College

Millersville University

Mississippi [a] Valley State University

Moorhead State University

Morehouse [a] College

Morgan [a] State University

Morningside College

Morris [a] Brown College

Mount Holyoke College

Mount Saint Marys College

Murray State University

Norfolk [a] State University

North Adams State College

North [a] Carolina A&T State University

North [a] Carolina Central University

Northeastern Illinois University

Oglethorpe University

Ohio Wesleyan University

Plymouth State College

Prairie [a] View A&M University

Providence College

Saginaw Valley State University

Saint Anselm College

Saint Cloud State University

Saint Marys College (IN)

Saint Marys University

Salve Regina University

Samford University

Simpson College

Smith College

South [a] Carolina State University

Southeastern Oklahoma State University

Southern College of Seventh Day Adventists

Southern Nazarene University

Southern [a] University (Baton Rouge)

Southern Utah University

St. Lawrence University

SUNY-Purchase

Tennessee [a] State University

Texas Lutheran College

Texas [a] Southern University

Tuskegee [a] University

Union College

University [a] of Arkansas at Pine Bluff

University of Hartford

University of Houston-Downtown

University of Michigan-Dearborn

University of Northern Iowa

University of Portland

University of Puget Sound

University of South CarolinaSpartanburg

University of Texas-Pan Am

University of Texas-Permian Basin

University of Texas-Tyler

University [a] of the District of Columbia

University of Wisconsin-Superior

University of Wisconsin-Whitewater

Utica College of Syracuse University

Virginia [a] State University

Washington and Lee University

Webster University

West [a] Virginia State College

Westminster College

Wheaton College (IL)

Wheaton College (MA)

Winston-Salem [a] State University

Woodbury University

York College (PA)

Denotes [a] that the institution is a Historically Black College or University (HBCU).

Appendix B: Variable Definitions

TOTJEL is the total number of articles and notes produced by faculty members at each institution [a] between 1983 and 1997 that are in the EconLit database. This database includes all journals abstracted in Journal of Economic Literature and is available on CD-ROM.

TOTQA is the total number of JEL-indexed articles produced by faculty members at each institution between 1983 and 1997 that are ranked by Laband and Piette (1994).

WTQA is the total weight of JEL-indexed articles produced by faculty members at each institution between 1983 and 1997, using modified Laband and Piette (1994) weights, based on citations per article.

SALARY is the average faculty salary for the institution in 1996. SOURCE: Integrated Post-secondary Education Data Systems (IPEDS) Financial Data, National Science Foundation, Web-based Computer Aided Science Policy Analysis and Research (WEBCASPAR). [b]

PEDIGREE is an average faculty quality index based on the quality of the graduate programs from which an institution's Ph.D. faculty graduated. It utilizes the National Research Councils' (NRC) ranking of doctorate-granting economics departments in 1995. The top rated programs in the NRC survey were the University of Chicago and Harvard University, and they were assigned a rank of 1.5. The lowest rated program was Northeastern University, which was assigned a rank of 107. If a faculty member had a Ph.D. from a program which was not listed in the NRC survey, a value of 108 was assigned. SOURCE: Goldberger, Maher, and Flattau (1995).

HBCU is a dummy variable indicating whether or not the institution is a Historically Black College or University.

LIBVOL is the number of volumes in library (millions). SOURCE: Peterson's Guide To Four Year Colleges (1998).

AEXP is the average number of years since the completion of the Ph.D. for all assistant, associate, and full professors listed in the 1996 Prentice Hall Guide to Economics Faculty. SOURCE: Hasselback (1996).

[AEXP.sup.2] is the square of AEXP.

SQUAL is a student quality index based on Peterson's ordinal ranking of an institution's admissions criteria. It is defined as follows: most difficult = 5, very difficult 4, moderately difficult = 3, minimally difficult = 2, and noncompetitive 1. SOURCE: Peterson's Guide To Four Year Colleges (1998).

ECON_ONLY is a dummy variable indicating whether or not the disciplinary unit in which the economics department is housed is exclusively economics. ECON_ONLY = 1 indicates the undergraduate program is exclusively economics. SOURCE: Hasselback (1996).

NORTHEAST is a dummy variable indicating whether or not the institution is located in the northeastern United States.

RESEXP is the research expenditure for the institution for the year 1995 (in thousands of dollars). SOURCE: IPEDS Financial data, National Science Foundation (WEBCASPAR). [b]

TPHD is the number of baccalaureate graduates of an institution who earned a doctorate in economics over the 1980-1996 period as reported in the National Science Foundations' Doctorate Records File. SOURCE: National Science Foundation (WEBCASPAR). [b]

NFAC is the number of faculty members with earned doctorates, excluding deans, in the department in which the economics faculty are housed, as reported in the 1996 Prentice Hall Directory. SOURCE: Hasselback (1996).

PRIVATE is a dummy variable indicating whether an institution is private or not.

MA is a dummy variable indicating whether or not an institution has Carnegie classification of Master's or above.

STU_ENROLL is the student enrollment for the institution in the Fall of 1995. SOURCE: IPEDS, National Science Foundation (WEBCASPAR). [b]

(a.) Faculty members at each institution indicates faculty members listed in The 1996 Prentice Hall Guide to Economics Faculty (Hasselback 1996).

(b.) WEBCASPAR is located at http:/www.nsf.gov/srs/srsdata.htm.