|Title:||United States average number of economic literature journals, number of articles, articles per journal, and specialty journals for the 1970s, 1980s, and 1990s decades, and for 2000 through 2007|
Start of full article - but without data
Summary of Data from EconLit Database 1970-2007
Average Number Average Number of Journals of Articles Articles Years Per Year Per Year Per Journal
1970-1979 XXX X,XXX XX 1980-1989 XXX X,XXX XX 1990-1999 XXX XX,XXX XX 2000-2007 XXX XX,XXX XX
Number of Specialty Number of Specialty Journals (XX% Journals (XX% Years Cutoff) Cutoff)
1970-1979 1980-1989 1990-1999 XXX XXX 2000-2007 XXX XXX
A fundamental question for any academic discipline is, "What subjects are people researching?" This question can be asked in several ways. Which subjects are economists publishing? Which subjects are economists reading? Which subjects have the most impact on the literature? Which subjects are represented in well-respected general subject journals? How many specialty journals are dedicated to each subject?
Answers to these questions are useful for strategic planning, publication strategy, and curriculum design through the identification of shifting focus within the economics discipline. For instance, the research record of an economist who publishes in a widely published subject field that is little published in leading general subject journals should be evaluated differently from an economist whose fields have the opposite properties. A liberal arts college that desires to make more of an impact on the literature may find that the subjects that need to be taught at the undergraduate level are in conflict with the subjects that are being published in top journals. A subject area with a small percentage of articles published that has had a strong impact when adjusted for citations may be fertile ground for research.
Although the study of economic research has a long history, subject area focus has been little studied. Morin (1966) examined the sales of Chamberlin's Theory of Monopolistic Competition, Schumpeter's Theory of Economic Development, Samuelson's Foundations of Economic Analysis, Friedman's Essays in Positive Economics, and Becket's Economics of Discrimination to examine the changes in economics from XXXX to 1964. Figuring that most copies of these books were being read by researchers, he posited that the changes in the sales of these books indicate shifts in interests within economics. Fusfeld (1956) examined the departments and types of institutions which were represented by papers at the 1950-1954 American Economic Association meetings. He did not rank these departments but viewed the narrow group of represented schools as a flaw in the program selection.
Niemi (1975) created rankings of the top XXX economics departments and the top southern economics departments according to the number of publications and the number of pages published in a group of XX journals. Smith and Gold (1976) adjusted Niemi for the number of faculty members in each department. Graves, Marchand, and Thompson (1982) built on the work by Niemi as well as Smith and Gold by updating their statistics and providing a trend analysis to examine rankings through time. Scott and Mitias (1996) provided updated rankings by pages published using two different sets of economics journals and created a list of the top XX economists by pages published.
Davis and Papanek (1984) first proposed using citations for ranking economics departments. Laband (1986) ranked departments based upon the productivity of the graduates of the program using both pages published and citations. Bodenhorn (1997) ranked liberal arts colleges by citations. Medoff (1989) ranked individual economists based upon citations and ranks schools based upon the number of "top" economists employed there.
Although these analyses provide important insights into the productivity of economists and provide a benchmark for hiring and promotion to tenure, few have discussed the distribution of subject areas within journals. Harden, Liano, and Chan (2006) calculated the percentage of Real Estate Economics articles published. McCain (1991) used article subjects in her cocitation analysis. Durden and Ellis (1991) listed the most oft-cited articles in the AER broken down by subject area.
Kim, Morse, and Zingales (2006) examined all the articles from 1970 to 2005 that received more than XXX citations. They used the EconLit database to identify the Journal of Economic Literature (JEL) Codes associated with each article. They created XX JEL aggregate fields to see the change in most-cited fields through time.
We supplement this literature by describing the evolution of subject areas published in economics for nearly four decades. We calculate the percentage of all articles published in each JEL subject category. We do the same calculation for eight general economics journals. We rank the top XXX journals using Eigenfactor.com's Article Influence (AI) and reweight the percentage shares by AI for 1995-2006.
Some subject areas have been remarkably constant in their percentage share of articles. Other areas (such as Finance, Development, and Industrial Organization) have seen their share of total articles rise over the past tour decades while others (such as Microeconomics, Macroeconomics, and Labor) have seen their share fall. When the percentage share is reweighted by AI, a measure of how much the article is cited, some of these increases and decreases are confirmed (such as Finance and Macroeconomics), whereas the trend is not confirmed for others (such as Microeconomics and Labor).
The Mathematical and Quantitative Methods subject represents a relatively small share of articles published but has a greater representation in the eight general journals and has a higher weight when we consider AI, likely reflecting the broad use of mathematical and quantitative tools in other subdisciplines of economics. Although the growth in specialty journals in Finance and Development may explain their growth in overall percentage, the two disciplines show opposite effects when their shares are reweighted for AI. Any of these topics may prove fruitful for further research providing greater understanding of the evolution of the economics discipline.
II. DATA AND METHOD
Our database was compiled from the EconLit database published by the American Economic Association and includes every article in the database from 1969 to 2007. Each observation includes the journal name, the article name, all subject codes, and the date of publication. Although the other fields are unique, an article may have more than one subject code.
Subject codes take on two forms: letter codes (1991-2007) and number codes (1969-1999). Number codes were originally used to classify articles and were replaced by letter codes. An example of a letter code is EXXX (Money and Interest Rates; General). All subject codes of the form "Exxx" are in the "Macroeconomics and Monetary Theory" category. The first letter of the subject code describes the category. We will refer to this as the category letter. The two numbers that follow describe the subcategory. The last number has been reserved for future use. Miscellaneous Categories (Y) and Other Special Topics (Z) are not included in our analysis given the nature of these categories. (X)
Our main goal is to classify articles by broad subject areas; therefore, our analysis will consider the category letter. We convert number codes to letter codes using a method described in Appendix X, allowing us to calculate the percentage of articles for a particular subject during the 1970s and 1980s.
Some aggregate data are presented in Tables X and X. The number of journals and the number of articles per year have quadrupled since the 1970s. Our dataset comprised the XXX,XXX articles in peer-reviewed journals from 1969 to 2007. There were X,XXX unique journals during that period.
We analyze three sets of journal articles. First, we look at all journal articles published during our sample period, answering which subjects have been published overall. Second, we look at the top XXX journals as ranked by AI, a measure of time spent reading a journal based upon citations, showing us which subjects have been read. Third, we look at eight general economics journals, showing us what subjects have been published for the reader of nonspecialty journals.
The first method shows in which subjects economists are publishing and provides a benchmark for our analysis. The second method shows which subject areas economists are reading. The third method shows which subjects the "general" economist sees. The eight journals chosen for the third method are well known and general for the time period of our study so that if an economist saw these journals on the bookshelf of another economist, it would be difficult to figure out the other economist's specialty area. In this sense, these journals give us insight into what is being published for the "general" economist. The journal set is listed in Appendix X and is identical to that used by Conroy et al. (1995).
We measure the percentage subject share across articles by looking at the subject codes for each article. An article with n different codes is treated as n different articles with each assigned a weight of X/n of an article. When we look at all articles published and the eight general journals, no further weighting is employed; however, we reweight the articles in the top XXX journals by the AI obtained from Eigenfactor.com, which measures how much an article from a given journal is read. The set of top XXX journals is determined each year by ranking all journals by their AI. Bergstrom (2007) computes Eigenfactors which measure the percentage of a researcher's time spent reading a particular journal if the researcher randomly reads an article then randomly reads a citation article from that article then randomly reads a citation article from that article and so on. AI is the Eigenfactor of a journal divided by the number of articles published in that journal and is appropriate for our purposes because we are looking at the subject share on an article basis.
Suppose that three articles were published in a given year, each in a different journal. The first article has letter codes EXXX, EXXX, and GXXX. The second article has letter codes FXXX and OXXX. The third article has letter code LXXX. Two-thirds of the weight of the first article is in the letter code E while one-third is in the letter code G. One-half of the weight of the second article is in F and the other half is in O. The last article's weight is entirely in L.
For the first two methods (which ignore AI), the final percentages are determined by dividing by the total number of articles. The weights are: XX.X% in E, XX.X% in G, XX.X% in F, XX.X% in L, and XX.X% in O. For the method that includes AI, if the three journals have the same AI, the weights are unchanged. If the first journal has twice the AI of the other two, then the weights for E and G are scaled up while those of F, O, and L are scaled down. The resulting weights are: XX.X% in E, XX.X% in G, XX.X% in F, XX.X% in L, and XX.X% in O.
We also examine specialty journals. We define a specialty journal as a journal where the share of the two largest weights of letter codes is greater than XX%. Although arbitrary, this threshold level does lead to approximately XX%-XX% of all journals being considered specialty and XX%-XX% of all of the top XXX journals weighted by AI being considered specialty, which we consider reasonable. (X) We conduct this test annually, so a journal that is a specialty journal in one year may not be in the next or may have become a specialty journal in another area. The letter code with the largest share is considered to be the primary subject of that journal and the letter code with the second largest share is considered the secondary subject of that journal.
The results for all articles published from 1969 to 2007 are shown for each decade in Table X. The average annual percentage of all articles published is shown for each category letter. Note that these results are simply the percentage of all articles written without regard to the impact of the article or journal prestige.
Microeconomics (D) and Macroeconomics (E), which accounted for XX% of all articles written in the 1970s and 1980s, have declined to XX% of all articles in this decade, whereas fields such as Finance (G) and Development (O) have risen from X% to XX% each. The top five fields in the 1970s were Microeconomics (XX.X%), Labor (XX.X%), Macroeconomics (XX.X%), International (X.X%), and Development (X.X%). In 2000-2007, the top five fields were Finance (XX.X%), Development (XX.X%), Industrial Organization (X.X%), Microeconomics (X.X%), and Labor (X.X%). Some fields, such as Agricultural, Resource, and Environmental as well as Urban, Rural, and Regional have had fairly steady percentage shares of articles written while others such as Law and Economics and Economic History have fluctuated considerably, albeit off small bases.
The results for the eight general journals are shown for each decade in Table X.
Microeconomics (XX.X%), Mathematical and Quantitative Methods (XX.X%), and Labor (XX.X%) continue to dominate these journals comprising nearly half of all articles written. Similar to the results for all journals, Health, Education, and Welfare (I), Financial (G), and Industrial Organization (L) have all grown, and International (F), Public (H), and Macroeconomics (E) have all shrunk. However, unlike the results for all journals, Economic History (N), Microeconomics (D), and Labor (J) have grown, and Development (O) has shrunk.
AI is only available from 1995 to 2006. To see the impact of reweighting subject percentage shares for each subject category, Table X presents the average percentage share for each of the three methods for 1995-2006.
During 1995-2006, Microeconomics accounts for X.X% of all articles written and XX.X% of all articles in the eight general journals. Microeconomics accounts for XX.X% of all articles in the top XXX journals when reweighted by AI. The overweight of Microeconomics in the eight general journals appears to reflect the general appeal of this subject.
During 1995-2006, Mathematical and Quantitative Methods accounts for XX.X% of all articles when reweighted by AI, considerably higher than the X.X% share of all articles published. Mathematical and Quantitative Methods has a XX.X% share of articles in the eight general journals, likely indicating that editors of these journals perceive the greater impact of these articles and the interest that the "general" economist has for these tools.
During 1995-2006, Finance represents XX.X% of all articles but only X.X% of articles in the eight general journals, possibly because of the lack of a finance journal in those eight journals. However, finance represents XX.X% of all articles in the top XXX journals when reweighted by AI.
To examine whether an increase in specialty journals has led to some of the shift in subject area weights, we compiled the percentage of articles in specialty journals in Tables X and X. Table X compares the amount of journal specialization in each field in two periods, 1991-1999 and 2000-2007, showing the change in specialization of the various fields. (X) Table X compares the percentage of all articles that are in specialty journals in the various fields to the percentage of articles in specialty journals in the top XXX journals weighted by AI. If the top two letter codes comprise XX% of the total article weight for that journal, the journal is considered a specialty journal. At a XX% threshold, specialty journals comprise XX%-XX% of all journals.
Finance and Development both have a high percentage of articles in specialty journals that have grown as their overall percentage of all articles written has grown. When reweighted by AI, Finance has a much greater share of articles in 1995-2006 (XX.X% of all articles vs. XX.X% when reweighted for AI) and Finance is slightly more specialized (XX.X% of all Finance articles vs. XX.X% of Finance articles when reweighted by AI). Development shows the opposite effect with a greater share of all articles but a lower share of articles when reweighted by AI (XX.X% vs. X.X%). XX.X% of Development articles are in specialty journals. That number drops to XX.X% when we reweight by AI.
The number of specialty journals in Finance and Development averages approximately XX journals each for 2000-2006. The next largest field (Agricultural, Natural Resource, and Environmental) has XX journals. Although a greater number of specialty journals increases overall article share, the effect of specialty journals on AI appears ambiguous.
The decline in share for Macroeconomics (XX.X% of all articles in the 1970s vs. X.X% of all articles in the 2000s) is one of the more striking results of our analysis. The decline in share for Macroeconomics does not appear to be caused by an increase in specialty journals. Macroeconomics has about XX% of articles in specialty journals, one of the lowest percentages for a field. Macroeconomics has had a nearly steady XX specialty journals per year for 1991-2007, whereas the number of economics journals has doubled over that period. Rather than Macroeconomics becoming more specialized, the decline in share for Macroeconomics might be caused by the growth of macroeconomic specialty journals not keeping pace with the overall growth of specialty journals in economics.
The percentage share of subjects in economics has changed significantly over the past four decades. Finance, Development, and Industrial Organization have seen significant increases in share, whereas Macroeconomics, Microeconomics, and Labor have seen declines. Although an increase in specialty journals helps increase the overall share for a particular subject, the effect is ambiguous on share that is reweighted by AI.
Besides providing insight into the changing nature of the economics discipline, this analysis is useful for strategic planning, publication strategy, and curriculum design through the identification of shifting focus within the economics discipline.
AI: Article Influence