The consequences of internationalization on the research topics in Economics Jean-François Laslier1 [January 21, 2016. Please refer to the final version that will be published in European Political Science, the professional journal of the European Consortium for Political Research (ECPR).]

Ce que l’on conçoit bien s’énonce clairement, Et les mots pour le dire arrivent aisément. Nicolas Boileau Introduction The use of English as the communication language in science has been on the rise in the hard sciences since the beginning of the XX-th century, up to a point where it became almost exclusive. In the social sciences, this movement started later and the situation is uneven, with Economics being seemingly ahead of time compared to Political Science. It is therefore natural to wonder whether the now almost complete domination of English, which has already happened in Economics, will soon arrive in Political Science. What are the precise consequences of such domination? Is this a good thing or not? What should we do to facilitate, or to resist, this movement? The gains from using a common language are clear: mutual understanding is much easier. We all enjoy the possibility of meeting and exchanging ideas with colleagues from all over the world. Conversely, many of us can cite the names of wonderful scholars of the previous generation who lack recognition because they did not published enough in English. A point which comes to mind, and which is mentioned by David Lublin (2016) in this symposium, is that a paper written in English has a larger readership than a paper written in a local language. This is obviously true, but it only means that a paper written in English has more potential readers. Starting from a situation in which isolated (national) groups of researchers read and write their own productions, switching to a single language changes the structure of the communication network. It gives the possibility to everyone to read any paper, but it does not guarantee anything. Working in a foreign language makes reading more difficult, a point which is of particular importance when the native language is linguistically remote from English, like Asian languages are. Therefore it is even possible that globalization makes researchers read less, on average. Even if the average number of papers read per reader is constant, using a unique language does not increase overall reading. Therefore, concretely, if the exclusive use of English increases the number of readers of some papers, it must be because it decreases the number of readers of some other papers. Potential readers are actual readers, so that the use of a single language does not by 1

CNRS, Paris School of Economics. École Normale Supérieure, 48, boulevard Jourdan, 75014 Paris, France. e-mail : [email protected]. Thanks to Ulrich Ammon, André Blais, Filippo Contesi, Thomas Piketty, Philippe Quirion, Karine Van der Straeten, Philippe Van Parijs and Bernard Walliser for their encouragements, discussions, remarks… and language editing. 1

itself increase communication. What it may well do is to increase the concentration of communication, if everyone, ends up reading the same papers. The two phenomena probably coexist in Economics: larger number of readable papers for everyone, but global decrease in the variety of papers actually read. Because the use of the same language facilitates one-to-one communication, the losses from using a common language are not immediately clear. A point often raised is that it is more difficult to speak, read, and write correctly in a foreign language. Native English speakers have an undue advantage which goes against basic fairness principles (Van Parijs 2011), and a glass ceiling in fluency and elegance of writing will still affect non-native writers no matter how much time and energy they devote to their writing. This point is probably less important for scientific publication than for oral communication: when writing, more than when talking, one can take the time to carefully work out the sentences in order to explain oneself in a satisfactory manner. A related point is that since English is the unique language, native English speakers have lower incentives to learn other languages. As a matter of fact it is more and more the case that native English-speaking scholars know no other language, and have access only to the English literature. English-speaking students have no incentive to practice other languages and, little by little, a whole body of knowledge written in non-English languages tends to disappear. Another point which is also often mentioned is that languages carry ways of thinking: the thought itself is shaped by grammatical structure and words, so that translations can only be approximations and tend to lack deepness when compared with original texts. This point of view, which is the starting point of Peter Kraus (2016) argument in this symposium, is deeply rooted in literary studies, and indeed one cannot contest that most authors are better read in their original languages. But, for science, the counter-argument is that science has to do with logic and facts, and both logic and facts should be independent of the language. Boileau’s phrase: “Whatever is well conceived is clearly stated, and the words to say it flow with ease” applies to scientific concepts and to scientific communication in a very particular manner. Ideally, the language of science is such that scientific concepts can flow with ease across the linguistic frontiers. In the community of the hard sciences (including mathematics and excluding the social sciences) the French mathematician Laurent Lafforgue is expressing a minority view when he defends2 the use of French in his discipline upon the argument that science feeds itself from linguistic and cultural diversity. The case of the social sciences is more complex. The question of language may be quite different for a researcher writing on a new econometric test and for her colleague doing qualitative ethnography. Formal Economic theory follows Debreu’s dictum: “Allegiance to rigor dictates the axiomatic form of the analysis where the theory, in the strict sense, is logically entirely disconnected from its interpretations.” (Debreu 1959). The “axiomatic form” is effectively free from the contingencies of languages, it even uses its own specific and universal mathematical language, but one must admit that Debreu’s epistemological stand will be of limited help for the practice of applied research, qualitative or quantitative, in the social sciences. The idea that thought itself is not independent of the language in which it is expressed, what Peter Kraus calls the universalist and positivist positon, seems to imply that using a single language will lead to an impoverishment of the thought. That might be true but, in this contribution, I will follow a different track, easier to ground on observations, and linked to the objects of science rather than to its mental process.

2

See Lafforgue (2005). 2

As Ammon (2012) notices: “the predominance of a single language, English, in international scientific communication excludes contributions from various non-Anglophone quarters and, consequently, contributes to skewed scientific development.” While preparing this talk for political scientists I recalled some remarks made by European colleagues working in applied Economics, who were mentioning that it was more and more difficult to publish research on European data in the main Economics journals. I decided to check the extent of this phenomenon. Is it true? Does it relate to language? In what follows, I will show that, far from its universal pretention (Fourcade et al. 2015), Economics tends more and more to study a narrow subject: the American economy.

A study of some published research in Economics It is clear that, in order to appreciate the language issue, one has to have in mind the pattern of incentives for academics which is driving the phenomenon. Concretely, the prevalence of the American language cannot be separated from the domination of US universities and academic institutions. This domination is well documented in science in general and in Economics in particular. It is extreme.3 For example, the Economics “section” of the French National Center for Scientific Research uses, for peer evaluation, a ranking of journals in Economics. This list distinguishes six top journals in Economics, all of which publish now exclusively in English. The ranking from this national agency essentially matches the many rankings of Economics journals that are published and used in this profession. In what follows I shall focus on three of these “top” journals. Data I consider the latest articles published in the main journals in Economics. There is a welldefined consensus about which are the most prestigious Economics outlets, with the agreement of most of the rankings based on citation counts. The four main journals are the Quarterly Journal of Economics (QJE), the Journal of Political Economy (JPE), the American Economic Review (AER), and Econometrica.4 The 5-years impact factors of these journals, as computed by the Web of Science for 2014, were respectively 9.79, 5.69, 4.95 and 5.76. These numbers are small compared to what is found in the hard sciences, but they are high according to standards in the discipline. Econometrica is mainly a theory journal (and some of its citations come from journals in Statistics rather than Economics). For the purpose of this article, which is not concerned with theory, I thus restrict attention to the three other journals.5

3

Marion Fourcade (2005, 2009) has studied in great detail, including with international comparisons, the profession of economist. She has in particular shown how the institutional organization of the profession determines the production of knowledge. 4 The other top-journals in the CNRS list are the Journal of Economic Literature, The Review of Economic Studies and several journals more specialized in Management. 5 Hamermesh (2013) studies the same journals as I do, on a longer time span. He does not study the objects of the published articles but only the authors and the methods. He documents in particular a remarkable increase, starting in 2003, of the fraction of elder persons (more than 50 years old) among the authors. 3

These three journals publish almost exclusively peer-reviewed scientific articles6 and I study a total of 171 such papers published by QJE (volumes 129 and 130, issues 1 to 3, 70 papers), JPE (volumes 122 and 123, 50 papers), and AER (volume 105, issues 6 to 11, 51 papers). Observations I classify each article as using data or not and, if data are used, in four categories, depending on the place where the data were collected: US, Other English-speaking country, Non-English, and Cross-National. In the category labelled Other English-speaking country, I include only developed countries: Australia, Canada (but not Quebec), the UK, and Singapore, but not India nor Anglophone African countries. Most data in these articles come from traditional statistical sources, but I do not exclude the papers using experimental data, even when collected in the laboratory and not in the field. For all the computations that follow, each article is normalized to 1 so that, for instance, a two-author paper in which the first author is in a US university and the second author has a double affiliation in a US and an Israeli one is counted as ¾ US and ¼ Non-English.

Journal Nb. articles QJE 70 JPE 50 AER 51 Total 171

Authors, affiliations Female USA

Data

6.91 51.49 5.83 35.00 6.75 33.51 19.50 120.00

Other Non No USA Other Non Cross En. En. data En. En. nat. 8.58 9.93 13,00 27.50 3.00 15.00 11.50 7.58 7.22 14.00 23.00 1.00 11.00 1.00 6.24 11.25 16.00 19.00 0.00 12.00 4.00 22.60 28.40 43.00 69.50 4.00 38.00 16.50

About authors: Overall the universities from English-speaking countries (as defined above, but in practice mainly American) represent 83% of the authors. One can note also the striking gender bias: male writers represent 89% of the total. About the topics and data: Leaving aside the papers which use no data (about 25% of the total) and those which study the world economy or are essentially trans-national (about 10% of the total), one is left with 111.5 papers, that can be classified as follows. articles %

USA 69.5 62%

Other English 4 4%

Rest of the world 38 34%

Total 111.5 100%

One can see, and this is the main observation of this paper, that the main Economics journals essentially publish research based on US data. The four papers in the category “other Englishspeaking countries” are three papers on the UK (one of which is historical) and one using data from Singapore. For the rest of the world, the list is the following:

6

The exceptions are prefaces, notes and replies. I do not include these short papers if they are linked to a main paper. I also do not include the proceedings of the conference of the American Economic Association. Same thing for the Nobel addresses (one published in JPE and one in AER) which cannot be considered as standard peer-reviewed articles. 4

Africa: North Europe: Mexico: South Am.: India: South Europe: China: Indonesia: Japan: Taïwan Medieval states:

10 6 5 2 4 2.5 1.5 1 1 1 2

(4xKenya, 2xSierra Leone, Benin, Ghana, Uganda, whole Africa) (3xNorway, Denmark, Netherland, Sweden) (Argentina, Columbia, Chile) (France, Spain, Switzerland)

(Venice, medieval Germany)

One can notice that a number of papers are devoted to development, and often use African data (mostly gathered in former English colonies). A glance at the table of contents of the main journals a few years back shows that this is a recent phenomenon, the revival of development Economics. Still, the prevalence of the US economy among the objects of published research is striking. Not a single paper on Eastern Europe, on Russia, on the Middle East, on North Africa, etc…

Lessons The three journals considered here are published in the US, their editorial boards are almost entirely composed of US-based scholars, the authors are working in US institutions, and they are studying the US economy. Contrary to what is often heard outside the US, these publications should not be considered as “international” journals but as national journals which, incidentally, are relatively open to international authors and topics. Professional and institutional factors The mechanism which leads to such a state of affairs is well-known in the profession, and is easy to describe. It is the effect and the cause of the intellectual leadership of the main US universities in Economics. These universities offer high wages and good working conditions, and can attract the best researchers. Because there is less sanction by reality in Economics than in the hard sciences, the discipline is largely amenable to “beauty contest” type of phenomenon by which a dominant group of scholars defines the criteria of quality, judges this quality, and benefits or loses from this judgment. Hence, the dual claims of the “superiority of Economics” and of the “superiority of Economists,” that emanates from the American Economics departments, can spread the entire discipline. This intellectual leadership is self-reinforcing. Researchers from other countries have thus a strong incentive to play according to the rules defined there. This implies working on the fashionable subjects, with the fashionable technical tools, and publishing in the main journals, citing the main authors. Breaking this vicious circle would require the leading departments to value diversity in methods and topics. It is fair to say that most are well aware of the situation, and many scholars make valuable efforts in that 5

direction. But that is easier said than done, and it is also for good reasons that academic life naturally tends to academism7. For instance, peer judgment must rest on commonly accepted rules of art, which thus are biased against the diversity of approaches.8 Therefore, all over the world, young economists try to be hired by American universities and to publish in American journals. Many of the best gifted and well educated ones manage to do so,9 increasing the quality of Economics in the US as much as they decrease it outside. The foreseeable consequence of this global brain drain is an impoverishment of the quality of research done outside the US and about non-US economies. The Economics Science business might do just like the movie industry, where the Hollywood movie industry has almost killed non-US cinema traditions. The next step could be a lowering of the quality of the data collected about non-US economies, a shortage of good studies on these economies, and a lesser quality of teaching in many universities around the world. In short, the on-going US-nationalization of Economic Science may have a simple consequence for the rest of the world: We will know less because we will know more about less. There are counter arguments to the point made above. Two of them can be seen in my small data set. First, maybe because of Americans’ distrust of government, the US government does not produce as good quality data as some other countries, notably in Scandinavia; indeed an increasing number of studies use unique data from Norway and other small Northern Europe countries. Second, international organizations invest more and more in international data sets, especially in developing countries. The role of language Are we dealing with a language problem or are we simply observing the consequences of the domination of US universities? It is useful to examine non-US, English-speaking countries. Unfortunately, the data that I collected for this paper are not quantitatively sufficient to draw definitive conclusions. A tentative remark is that other English-speaking countries do not seem to do much better than non-English-speaking ones. For example, I found no paper on Australia, Canada or New Zeeland, and the two or three papers with UK data do not mean much compared with the 69 with US data. I therefore submit that the power dynamic of language, as David Lublin calls it, is only indirect. The main point is the domination of the US academic system. In that play, the use of English has obviously an important role, even if indirect. For instance, it is not difficult for English-speaking universities to attract foreign students who are keen to learn English. This financially important point participates to the stability of the academic systems of Englishspeaking countries, and is not restricted to the USA. Still, university professors in Dublin and Auckland, who can attract students from Europe and Asia, use US data to publish in the best journals. The common use of English in research makes the global academic market more fluid; it may facilitates the brain drain described above but it is not sufficient to explain it.

7

In the French sense of the word: the intellectual sclerosis caused by traditionalism and an excessive care about conventions. 8 One can think of the famous quote attributed to Kenneth Boulding: “Mathematics brought rigor to economics. Unfortunately it also brought mortis.” 9 It would have been very interesting to record, with the same data set that I used, the authors’ countries of origin. But, unlike gender (observed by the given name) and affiliation (always indicated), this would have required specific enquiry for each author.

6

Another feature of the use of English in non-English-speaking countries is that academic work becomes un-readable outside academics. This may be a concern for teaching, for popular science, for the information of journalists, decision-makers and governments, and in general for the dissemination of scientific knowledge in society. This concern is often raised in big countries like France or Germany. The example of small countries with specific languages, such as most Nordic countries, shows that it is feasible, for researchers who usually work in English, to have a side activity of popularization, or consulting, in their own language. Still, one should keep in mind that the initial scientific publications in English are not accessible to non-English speakers. What about Political Science? From an epistemological point of view, Economics and Political Science are remarkably similar, with respect to several pertinent issues. Both are social sciences, with objects which are simultaneously universal and culture-dependent. The economic, as well as the political, organization differs from one country to the other: various histories, various levels of development, various legal frames, etc. Both disciplines have also universal —stated or not— goals. Notions like “power”, “individual rationality”, “sense of justice”, or “discrimination” are universal, even if their concretization is culture-dependent. The relative weights of the state and of the private sector, the structure of the productive sector, the laws and norms that govern work relations and contracts, differ across countries as much as their political institutions. Empirical studies in Economics are usually locally embedded, as they must be, but there is an obvious unity of Economics as a science, that one can see in the constant effort of economists to produce and rely on abstract theory. The same is true in Political Science, although the status of the pure theory is not as strong. One should therefore seriously consider the possibility for Political Science to follow the same path as Economics, with respect to the US scholarly domination, and then maybe, as argued above, with respect to the impoverishment of knowledge about Politics in the rest of the world.

Conclusion: What to do? The facts presented here are very partial and would need to be supplemented with a more systematic examination. This being said, for the conclusion of this contribution I will follow on the, possibly too pessimistic, views expressed above. It seems that many problems, which are apparently due to the exclusive use of English, are in fact primarily due to the overwhelming importance of US universities in the global academic market. Positive actions are required to correct this situation: to improve the job market conditions for academics outside the US, and to have truly international, non-US, journals and top quality universities and research centers. There already exist many truly international journals in Economics, specialized ones as well as more general ones, but, for reasons mentioned above they are often not the authors’ first choice. Therefore, the managers and editors of these journals cannot do much more that what they actually do to increase their quality and attraction. But it seems to me that more could be done by the universities in their recruitment policy. As game-theorists would say, in order to get out of a bad equilibrium, some players have at some point to take “positive” actions, which are sub-optimal in the short run. Foreign universities should not base their quality-judgment for recruitment exclusively on the publication record of candidates in US journals. And, with respect to the research topics, it makes sense for them to attach greater value to research done on national data. 7

Preservation of the world languages for the dissemination of science is another issue, related but distinct from the question of the use of English as a universal language for the production of science. For most researchers in Economics, and probably also in Political Science, the dissemination activity is in practice a distinct one: specific written texts, and specific means of communication. This activity, which is socially valuable but distinct from research, is performed in local languages. It should not be endangered by the general use of English for primary science. This may require giving attention to its status within the University: to maintain good means of dissemination (popular science journals, web sites, academic blogs…), and to value the dissemination activity as it must be. This paper has put forth a different point, based on facts about Economics but that might be valid as well for Political Science, and that is only indirectly related to the language question. It is a socially desirable goal that all topics and countries be studied. One can summarize the argument as follows: This social goal is not aligned with the incentives generated by the academic market, which is now dominated by US universities. Positive action is thus required from academic institutions to improve the incentives to do research on non-US data.

References Ulrich Ammon (2012) « Linguistic inequality and its effects on participation in scientific discourse and on global knowledge accumulation – With a closer look at the problems of the second-rank language communities » Applied Linguistic Review 3(2): 333–355. Nicolas Boileau (1674; 1970) L’art poétique. Paris: Librairie Flammarion. Gérard Debreu (1959) Theory of Value : An Axiomatic Analysis of Economic Equilibrium. New Haven and London: Yale University Press. Marion Fourcade (2006) “The Construction of a Global Profession : The Transnationalization of Economics” American Journal of Sociology, 112(1) : 145–194. Marion Fourcade (2009) Economists and Societies: Discipline and Profession in the United States, Britain, and France, 1890s to 1990s. Princeton, N.J.: Princeton University Press. Marion Fourcade, Etienne Ollion, and Yann Algan (2015) “The Superiority of Economists” Journal of Economic Perspectives, 29(1): 89–114. Daniel S. Hamermesh (2013) “Six Decades of Top Economics Publishing: Who and How?” Journal of Economic Literature 51(1): 162–172. Peter Kraus (2016) “In defense of a multilingual political science” [this symposium]. Laurent Lafforgue (2005) “Le Français au service des sciences” Pour la science, mars 2005 p. 8. David Lublin (2016) “The case for English” [this symposium]. Philippe Van Parijs (2011) Linguistic Justice for Europe and for the World. Oxford: Oxford University Press.

8

The language of political science

Chair : André Blais, Université de Montréal ([email protected])

Participants : David Lublin, American University ([email protected]) Jane Jenson, Université de Montréal ([email protected]) Peter A. Kraus, University of Augsburg ([email protected]) Jean-François Laslier, Paris School of Economics ([email protected])

The roundtable addresses three related questions: 1: Should we all publish (mostly) in English? 2. What are the consequences of the presence of a dominant language on the development of political science? 3. What is the impact of language on how we do political science? The roundtable took place on Thursday August 27, 2015, during the ECPR conference in Montreal.

9