SELECTIVE ESSAYS IN COMPETITIVE ENVIRONMENTS

Marco Piatti BBus(Ec)

Primary Supervisor: Professor Benno Torgler

Submitted in fulfilment of the requirements for the degree of Master of Business (Research)

School of and Finance Faculty of Business Queensland University of Technology Submitted February, 2012

i

Keywords

American Economic Review, (Australian) Rugby League, Competition, Corruption, Extraordinary Wealth, Gini Coefficients, Globalisation, Impact of Team Colours, Inequality, Rankings, Red, Superrich, Superstars, Team Sports, , Winner Take All Markets.

Selective Essays in Competitive Environments i ii

Abstract

This thesis is a collection of essays that utilises descriptive and empirical tools to examine competitive environments such as in academia, superrich and sport. The essays capture different aspects of the winner-take-all phenomenon by looking at citation and publication inequality in a top tier economics journal namely the American Economic Review. How globalisation and corruption influence the accumulation of extraordinary wealth and finally, how in a fairly equal competition, that is in the National Rugby League in Australia, wearing red shirts could lead to a comparative advantage and hence, tip the balance between winning and losing. The results within academia indicate that a highly unequal distribution exist, in which only a few top authors or institutions produce the majority of output. Furthermore, the results obtained in the superrich environment indicate that corruption and globalisation enhances the accumulation of extraordinary wealth. Finally, the results in the sport environment are mixed. While we find support for a positive effect of wearing red jerseys in our descriptive analysis, we find a negative effect when we control at the team level. However, when we investigate the relative difference in the degree of redness between home and away team, we find a quite strong positive effect of wearing red shirts even after controlling at the team level.

ii Selective Essays in Competitive Environments iii

Table of Contents

Keywords ...... i Abstract ...... ii Table of Contents ...... iii List of Figures ...... v List of Tables ...... vi Dedication ...... vii Statement of Original Authorship ...... viii Acknowledgments ...... ix CHAPTER 1: INTRODUCTION ...... 1 1.1 Academia ...... 5 1.1.1 Baumol’s wish list ...... 8 1.2 Extreme wealth ...... 8 1.3 Sport ...... 9 1.4 Background Information on the Following Four Chapters ...... 11 CHAPTER 2: COMPETITION IN ACADEMIA: EVIDENCE FROM THE AMERICAN ECONOMIC REVIEW...... 13 2.1 Introduction ...... 14 2.2 Top Institutions, Top Papers, and Leading Publishing in the AER ...... 16 2.3 Institutional Rankings ...... 18 2.4 Country Rankings ...... 29 2.5 Top Papers ...... 30 2.6 Publishing Frequency ...... 31 2.7 Top Authors ...... 34 2.8 Conclusion ...... 37 CHAPTER 3: COMMENT ON WILLIAM BAUMOL’S “TOWARD A NEWER ECONOMICS: THE FUTURE LIES AHEAD!” ...... 42 3.1 Introduction ...... 43 3.2 Methodology ...... 43 3.3 Role of Mathematics ...... 44 3.4 Applied Econometrics ...... 46 3.5 Macroeconomics ...... 47 3.6 Economic History and History of Economic Thoughts ...... 49 3.7 Mathematical and Quantitative Methods ...... 50 3.8 Job Openings for Economists (JOE) ...... 51 3.9 Behavioural Economics ...... 52 3.10 Concluding Remarks ...... 53 CHAPTER 4: EXTRAORDINARY WEALTH, GLOBALIZATION, AND CORRUPTION. ... 55

Selective Essays in Competitive Environments iii iv

4.1 Introduction...... 56 4.2 Methodological Approach ...... 58 4.2.1 Data Sets and Hypotheses ...... 58 4.2.2 Specification of the Test Equation ...... 64 4.3 Empirical Results ...... 65 4.4 Concluding Remarks...... 69 4.5 Tables ...... 71 CHAPTER 5: THE RED MIST? RED SHIRTS, SUCCESS AND TEAM SPORTS...... 77 5.1 Introduction...... 78 5.2 Data and Methodology ...... 87 5.3 Results ...... 92 5.4 Conclusions...... 97 CHAPTER 6: CONCLUDING REMARKS ...... 101 6.1 Summary of Findings...... 101 6.2 Policy Implications ...... 103 6.3 Shortcomings ...... 104 6.4 Further Research ...... 105 REFERENCES ...... 109 APPENDICES ...... 119 Appendix A: Chapter 2 ...... 119 Appendix B: Chapter 4 ...... 120 Appendix C: Chapter 5 ...... 125

iv Selective Essays in Competitive Environments v

List of Figures

Figure 2.1: FLUCTUATIONS OF UNIVERSITIES POSITION OVER ONE HUNDRED YEARS ...... 26 Figure 2.2: LORENZ CURVES OF CITATIONS FOR THE DIFFERENT TIME PERIODS ...... 33 Figure 3.1: JOB OPENING FIELDS OF SPECIALIZATION (IN %, 1991-2009) ...... 52 Figure 5.1: COLOUR AND AGGRESSION USING NUMBER OF GOOGLE HITS IN MILLIONS ...... 80 Figure 5.2: COLOUR AND DOMINANCE USING NUMBER OF GOOGLE HITS IN MILLIONS ...... 81 Figure 5.3: COLOUR AND WINNING USING NUMBER OF GOOGLE HITS IN MILLIONS ...... 82 Figure 5.4: HOME TEAMS’ PERFORMANCE (IN %) WITH AND WITHOUT THE PRIMARY COLOUR RED ...... 93

Selective Essays in Competitive Environments v vi

List of Tables

Table 2.1: A SUMMARY OF INSTITUTIONAL RANKINGS ...... 20 Table 2.2: TOTAL EXPENDITURES FOR RESEARCH IN THE LATE 1930S ...... 22 Table 2.3: TOP INSTITUTIONS 1911–1920...... 22 Table 2.4: TOP INSTITUTIONS 1931–1940 ...... 22 Table 2.5: TOP INSTITUTIONS 1950–1959...... 24 Table 2.6: TOP INSTITUTIONS 1981–1990 AND 2001–2010 ...... 25 Table 2.7: TOP TEN INSTITUTIONS BASED ON CONTRIBUTORS’ PH.D. INSTITUTION ...... 27 Table 2.8: TOP TEN INSTITUTIONS BASED ON CONTRIBUTORS’ PH.D. INSTITUTION ...... 28 Table 2.9: TOP COUNTRIES PUBLISHING IN AER ...... 30 Table 2.10: TOP 10 AER PAPERS BY CITATION ...... 31 Table 2.11: DISTRIBUTION OF PUBLICATIONS AMONG AUTHORS (1911–2010) ...... 32 Table 2.12: GINI COEFFICIENT FOR US TEAM SPORTS...... 34 Table 2.13: TOP “SUPERSTARS” IN AER (12 AND MORE PUBLICATIONS) ...... 36 Table 3.1: EXPLORING THE ROLE OF MATHEMATICAL TOOLS ...... 46 Table 3.2: EXPLORING THE ROLE OF APPLIED ECONOMETRICS/EMPIRICAL ANALYSIS ...... 47 Table 3.3: SUBJECT-MATTER DISTRIBUTION OF PAPERS OVER TIME ...... 49 Table 4.1: DETERMINANTS OF EXTREME WEALTH (NBI) ...... 71 Table 4.2: DETERMINANTS OF EXTREME WEALTH (NBI) ...... 72 Table 4.3: DETERMINANTS OF EXTREME WEALTH (NBI) ...... 73 Table 4.4: DETERMINANTS OF EXTREME WEALTH (NBI) WITH CORRUPTION INDEXES INCLUDED ...... 74 Table 4.5: DETERMINANTS OF EXTREME WEALTH (NBI) WITH CORRUPTION INDEXES INCLUDED ...... 75 Table 5.1: PREVIOUS STUDIES ...... 89 Table 5.2: IMPACT OF RED JERSEYS ON MATCH SUCCESS ...... 94 Table 5.3: IMPACT OF RED DIFFERENCES ON MATCH SUCCESS ...... 96

vi Selective Essays in Competitive Environments vii

Dedication

I would like to dedicate this work to my mother who unexpectedly and much too early passed away at the beginning of November 2011. Without her love, encouragement and support I would not have had the strength or means to pursue neither tertiary education nor the fulfilling career I hope to achieve upon completion of my postgraduate research studies. Mami I dearly miss you. Love Marco.

Selective Essays in Competitive Environments vii viii

Statement of Original Authorship

The work contained in this thesis has not been previously submitted to meet requirements for an award at this or any other higher education institution. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made.

Signature: ______

Date: 15 June 2012

viii Selective Essays in Competitive Environments ix

Acknowledgments

Firstly, I would like to express my deepest appreciation to my principal supervisor, mentor and friend Professor Benno Torgler, whose determination, guidance and moral support was invaluable in completing this thesis. Secondly, I want to apologize to Manuela and at the same time promise that I will not be calling so often anymore to ask for guidance from her husband. Furthermore, I would like to thank in particular David Savage, Markus Schaffner, Jonas Fooken and Ho Fai Chan who, I do not know how often, helped me out with their knowledge and experience. I would also like to say thanks to all the other students on level eight for the great atmosphere and friendship we maintain with each other.

Additionally, I would like to thank the School of Economics and Finance, the School of Business and Queensland University of Technology staff. In particular I would like to thank the former Head of School Professor Tim Robinson and the current Head of School Michael Kidd for their help and support.

Last but not least I would like to express my deepest gratitude to my family. Specially, to my partner Padaranee Panthong, who probably had to suffer most when I was once again completely consumed by my studies and drifting off in the world of economics. Thank you for being so patient with me. And I would also like to thank my father and Rosy for their help and support when I was in need.

Selective Essays in Competitive Environments ix

Chapter 1: Introduction

Competition is the spice of sports; but if you make spice the whole meal you'll be sick.

Leonard, George

The price which society pays for the law of competition, like the price it pays for cheap comforts and luxuries, is great; but the advantages of this law are also greater still than its cost -- for it is to this law that we owe our wonderful material development, which brings improved conditions in its train. But, whether the law be benign or not, we must say of it: It is here; we cannot evade it; no substitutes for it have been found; and while the law may be sometimes hard for the individual, it is best for the race, because it ensures the survival of the fittest in every department.

Carnegie, Andrew

Competition can be heartless, cruel or brutal; nonetheless competition is a universal phenomenon, but competition may also foster innovation and technological advancements. In nature, only the strongest plants will survive, likewise in the animal kingdom only the strongest male will mate and secure the continuation of its genes. Schaller (1972 quoted in Wilson, 2000) found that territorial fights between male lions in the Serengeti often ended fatally. Natural selection, or survival of the fittest, is nature’s way of reacting to changes in environment by making a species better adapted to their new surroundings and fortifying their chances of survival (e.g. Ferrari and Chi, 1998). Human history is littered with the competition between cultures, for control of or access to resources, with an almost incalculable number of fatalities. The price for control is often genocide and subjugation, for example, the obliteration of the Incas through the Spanish lust for gold or more recently the oil dispute between Iraq and Kuwait that triggered the First Gulf War. However, competition is not always driven by the need for survival or the baser side of human

Chapter 1: Introduction 1 2 nature. Almost three thousand years ago the ancient Greeks already held sporting contests to determine the country’s best athletes and shower them with accolades. 1 Nowadays, competition is embedded in virtually every part of the modern world, it has become so ubiquitous that some are no longer even viewed as competitions but just a part of life. The number and type of different contests in which a person can participate is virtually endless and begin not long after birth. Children are compared with siblings and other children for things like age at which first steps are taken, first words spoken and even birth height and weight. This competition then graduates into schooling with competitions for grades and popularity. Even activities seen by many as recreational become competitive, music competitions and recitals even little league sporting competitions. The competition only grows fiercer as school climaxes when competition for scholarships and places at top universities are at their height. Now begins the adult life, a career and striving for jobs and promotions, the seeking of a mate and the procreation of children, which begins the cycle anew. Generally people believe that competition is a positive thing because it motivates people to extend themselves, become better and work harder. However, competition at the extremes can be detrimental for society, when there is no social benefit from the competition and winning becomes the only reason to compete. In most competitions (not all) there is usually only one winner who is quite often only marginal better than the runners up. However, the payoff structure is such that the benefits (prize money and recognition) for the winner excessively outweigh the trivially better performance by a vast margin. No one remembers who ran second in the last Olympic 100 meter sprint or who came second from your graduating year, as the old anecdotal saying would indicate “To the victor go the spoils.”

Environments such as the ones described above are also often called winner- take-all markets and are well understood as a theoretical concept of fierce competition that end with only one winner or a small group (see Berger and Bodie, 1979; Frank and Cook, 1995; Rosen, 1981). Frank and Cook (1995) state that winner-take-all markets are found in a large number of situations, whenever markets have a non-linear payoff structure, such that many individuals compete for a limited

1 http://www.britannica.com/EBchecked/topic/428005/Olympic-Games/59589/The-ancient-Olympic- Games

2 Chapter 1: Introduction 3 number of substantial prizes at the top. In non-linearly increasing compensation systems the winner’s payoff is often a multiple of the runner-up. For example in tennis, while the winner at the Australian Open 2012, Novak Djokovic, earned more than A$2.3 Million in prize money, the runners up Rafael Nadal, who was just marginally weaker, earned 50% less.2 Such reward systems are frequently found in sporting competitions or compensation arrangements in top management positions (see Ehrenberg and Bognanno, 1990, Bebchuk and Grinstein, 2005). One of the first papers to use this terminology described a situation where students participated in a portfolio selection project where only the winning portfolio won a prize while all the other participants received nothing (Berger and Bodie, 1979). However, this paper did not investigate anything in relation to the inefficiency problem in these particular markets. The nature of a winner-take-all market is inefficient because the rewards for the top performers exceed input by a huge margin and all the other contestants receive little or no return. Winner-take-all markets suffer from overcrowding in part because of a human weakness with regards to gambling (Frank and Cook, 1995) and that overcrowding largely exist due to people’s tendency to overestimate their odds of winning. Additionally, participants in winner-take-all markets compete against a largely unknown field of opponents. Fischbacher and Thöni (2008) use an experimental approach by analysing entry behaviour in winner-take-all markets. Their findings indicate that these markets are inefficient because of the excess of players (labour supply), which creates non-optimal welfare outcomes for society. For instance, every new contestant entering a winner-take-all market reduces the odds of winning for each adversary in the existing market. Moreover, they discovered that excess entrance will escalate with group size. Albers et al., (2000) state that subjects might be attracted to the tension created by the uncertainty about the outcome of a situation, such that the excitement of participating in a contest could explain the excess entry into winner-take-all markets. Rosen (1981) uses a purely theoretical approach in his investigation of superstars in such markets, defining superstars as a small group of people who earn abnormal income and dominate the markets in which they engage. He goes on to point out that a small number of universities are responsible for a large portion of doctoral degrees as well as a comparatively low number of academics account for a large portion of citations and possibly even

2 http://www.australianopen.com/en_AU/event_guide/prize_money.html

Chapter 1: Introduction 3 4 published articles. This is in line with Coaldrake and Stedman (1999) who emphasize that “a consistent finding has been that research output is highly skewed, with relatively few academics contributing to the bulk of research publications and a significant number of academics producing little or no output over prolonged periods” (p. 21). Rosen (1981) indicates that all superstar markets feature two characteristics: A highly unequal distribution of market size and compensation towards the most talented people; and a close link between personal advantages and the size of one's own market. Adler’s (1985) main argument is that markets for superstars only exist where consumption demands knowledge. The gathering of information by a consumer entails talking to other consumers, and a conversation is easier if all participants possess a similar prior knowledge. Consumption is not a momentary experience but a dynamic process: “the more you know, the more you enjoy” (p. 208). If there are artists with whom everybody is accustomed, a consumer would be better off supporting these stars even if their ability is not superior to others (a bandwagon effect). Furthermore, he points out that because it is costly to obtain information on numerous performers or authors, consumers will focus their demand on a relatively small number of individuals, who will be elevated to celebrities. Since consumers prefer popular artists, other consumers will switch to them, creating a snowball effect that leads to stardom. This can be observed in the publishing environment with a few lucky authors earning contracts for millions of dollars while a vast number of equally talented writers end up with next to nothing (Frank and Cook, 1995). According to Adler (1985) if everyone could be a star then the star would not necessarily be the most talented person but rather the one with the greatest luck (i.e. to be at the right time at the right place). The sports industry is another typical winner-take-all market. All over the world people try to become a professional sports person, however, only a few become professional athletes, while the others get little or no return. Rosen and Sanderson (2001, p. F60) state that “in the perhaps 30 new players are talented enough to make it into the 320 player roster of the NBA in a year, and those 30 players started out as more than 10,000 high school seniors”. Ehrenberg and Bognanno (1990) used tournament theory to analyse the payoff structure of professional golf tournaments and discovered that in a typical professional tournament, the winner collects hundreds of thousands of dollars or approximately 80% more than the runners up. In contrast, the difference between finishing 21st and 22nd varies by only a few hundred dollars. A

4 Chapter 1: Introduction 5 change in institutional conditions can also have a dramatic effect on shaping a superstar market. For instance, Leeds and Kowalewski (2001) investigated how a rule change in the National Football League (NFL) affected player income, when in 1993 the NFL and the National Football League Players Association (NFLPA) entered into new bargaining agreements. They found evidence that free agency and the salary cap dramatically widened income inequality between the top players (quarterbacks) and average players.

This thesis will empirically analyse three distinctive environments such as academia, superrich and sport in which competition is particular high and may lead to misallocation of human capital and a highly skewed distribution favouring the most successful competitors or institutions. Each paper investigates how certain factors influence the outcomes of these winner-take-all markets. In academia the examination looks at how top institutions may be influencing the publishing output in one of the top tier economic journals (the American Economic Review) over time. Furthermore, the case study investigates if economists have moved towards or away from a twenty-year old wish list, proposed by top author William J. Baumol, on how certain pillars of economics should change. In the superrich environment the investigation centres on if and how corruption and/or globalisation leads to an increase in superrich people through the examination of the Forbes list of billionaires for an eight-year period. And finally the sports environment investigates if wearing the colour red leads to a comparative advantage in fairly even competitions such as the Australian National Rugby League. The following sections further the discussion on these topics by providing an introductory literature review with a more detailed and topical discussion contained in each of the papers.

1.1 ACADEMIA

The vital significance of universities as being devoted to the development and diffusion of knowledge has been widely established around the globe, and is revealed in the considerable investment in higher education and research provided by government, industry and individuals (Coaldrake and Stedman, 1999). However, the academic environment is also one in which competition is extremely high. Every

Chapter 1: Introduction 5 6 year top tier economic journals receive an excess amount of articles, while only a handful will be considered worthy for publication. Torgler and Piatti (2011) found that the number of papers submitted to the American Economic Review increased substantially over time while concurrently the number of papers selected for publication decreased from 22% in 1953 to only 6.4% in 2009. Frey (2009) describes it as a ‘‘Publication Impossibility Theorem System’’ or “PITS” into which young academics are trapped. To publish in an A* Journal is exceptionally competitive and difficult. However, without having at least one top publication it is almost impossible to obtain a tenure track position or a promotion in a prestigious university. The authors of the unsuccessful papers, some of which are probably as good as the papers that are published, will have invested an excessive amount of time, sometimes years, and energy writing an article which they believed was ready for publication. Unsuccessful papers can certainly be submitted to another journal but this is associated with high transaction costs for the authors as this causes further delays in publishing, where the outcome of a submission may not be known for months or possibly years. Frey (2009) states that, “it cannot be dismissed readily that economists might perform a more useful social service if a larger number of them were induced to solve pressing and applied current problems rather than to be ‘‘wasted’’ in the useless effort of publishing articles in A-journals” (p. 339). For instance, Laband and Tollison (2003) have analysed 73 and 96 economics journals for the years 1974 and 1996 respectively, and discovered that on average 26% of articles are without any citations (dry holes). This misallocation of human capital by squandering time, money and resources which otherwise could be used elsewhere (e.g., in favour for the public) clearly describes the inefficiency aspects of winner- take-all markets in which only a handful of superstars are able to survive. Furthermore, the overconfidence of participants attempting to achieve a top publication by competing with all other participants in the market is an additional dilemma (see e.g. Camerer and Lovallo, 1999). But what differentiates these superstars from the also rans? Are they so much superior to other participants? This is unlikely as a distinctive feature of winner-take-all markets is the fact that superstars are often only marginally better than the other contestants. Therefore it is interesting to descriptively examine specific characteristics of the academic winner- take-all environment to determine if this corresponds here as well.

6 Chapter 1: Introduction 7

In the US talented students can participate in academic competition such as MathCounts, National History Day or spelling competition from an early age (Ozturk, 2008). The differentiation between winners and losers in the winner-take-all market has already started in middle school and continues on to high school, college and the work force. Having graduated from a top high school in the US will increase the probability of successfully entering into freshman classes at an Ivy League institution, which will subsequently increase the chances of obtaining a position in a prestigious firm or a place at a top university to commence further study. Frank (1999) describes a situation that clearly identifies such a dilemma:

“A friend who teaches at Harvard described to me the case of a woman from a small Florida college who had applied to Harvard’s graduate program in economics several years ago. She had scored within a few points of 800 on her GREs, both quantitative and verbal, and also had a very high score on the economics achievement test. She had straight A’s and glowing recommendations from several senior professors, who described her as the best student they’d ever encountered. The admissions committee agonized long and hard over this woman’s file, but in the end decided to reject her. They simply had too many other applicants who had compiled equally strong records at much more highly selective institutions” (p. 8).

In this particular market the results of working or having obtained a doctoral degree at a top institution on the publication output in a top tier economics journal over time are descriptively analysed. Prior studies have indicated that having obtained a Ph.D. from a top university is more important to being successful in obtaining a publication in top tier journal than being affiliated with one (see e.g. Kocher and Sutter, 2001). Many editors or co-editors of these A* journals obtained their Ph.D. at a top institution (see Hodgson and Rothman, 1999) and having been able to form networks while undertaking a doctoral degree at these top universities might play a significant role in why such a difference exists. However, it may also be a self-selection bias such that that these top universities simply attract the best students and academics who will later publish more articles in top journals.

Chapter 1: Introduction 7 8

1.1.1 BAUMOL’S WISH LIST In addition, as a case study in academics we investigated a wish list published in the Economic Journal in 1991 by one of the top two publishers in the American Economic Review namely William J. Baumol. In his article “Toward a Newer Economics: The Future Lies Ahead.” Baumol expressed his hopes for the future of certain pillars of economics such as mathematics, applied econometrics, macroeconomics, economic history and history of economic thoughts, mathematical and quantitative methods as well as behavioural economics. It is important to obtain an understanding how these pillars change over time, as they are crucial to maintain top performances in the future. This analysis attempts to determine if since the date of publication have we moved towards or away Baumol’s wish list. Keeping in mind, however, that the predictions expressed by Baumol were a forecast for a century in the future and therefore might not yet conform to the results we observe.

1.2 EXTREME WEALTH

In the modern world the life of the rich and glamorous has always attracted considerable interest from both the media and society. Numerous books exist with plans and suggestion on how to become superrich. This chapter empirically analyses the determinants of extraordinary wealth specifically focusing on how globalisation and corruption affect the accumulation of super wealth. Remarkably, only a few studies exist that have researched this topic with several of these studies focused only on one particular country. They investigated how, where and why fortunes were accumulated in different countries (AUS, US, UK and NZL) from different industries (see Siegfried and Round, 1994; Blitz and Siegfried, 1992; Siegfried and Roberts, 1991 and Hazledine and Siegfried, 1998). Goldman (1998) analysed why Russian businessmen first emerged on the Forbes list during the 1990s while concurrently Russia was seeking a loan from the IMF. Kennickell (2003) and Kopczuk and Saez (2004) both analyse the Forbes list of the 400 richest Americans and observe that wealth grew relatively strongly for the top 100 on the list over the period they analysed. Neumayer’s (2004) paper appears to be the first to investigate the issue at an international level and found that it is easier to accumulate super wealth in the wealthier, more populated countries. A country’s position to interact globally increases the possibility set for super-rich people and decreases restrictions on

8 Chapter 1: Introduction 9 efficient markets. Atkinson (2006) discovered that wealth is highly concentrated at the top, with the top 42 of 793 billionaires owning a quarter of the total wealth of this group. Several studies have analysed the relationship between globalisation and inequality (see Wade, 2004; Dreher and Gaston, 2008; Nissanke and Thorbecke, 2006; Zhang and Zhang, 2003). While people compare themselves with their surroundings and are concerned with their relative position, Frank (1999) remarks that research provides convincing proof that concerns about relative position is an innate and fundamental aspect of human nature. Wade (2004) points out that, income inequality is accompanied by increased poverty, slower economic growth as well as higher . Moreover, he points out that in the presence of higher income inequality, richer people in developed countries are more likely to compare themselves to wealthier people in developed countries and that this comparison may lead to richer people behaving more corruptly to achieve similar living standards as wealthier people in richer countries. Additionally, Gupta et al. (2002) find empirical evidence that corruption increases inequality. In a situation where corruption is out of control the distribution of resources is executed in a discretionary and unfair method. Rose-Ackerman (1999) points out that at the cost of normal people long-term connections with a few corporations are established to exploit a nation’s capital. Hence, in a government or state in which those in power exploit national resources and steal, fortune is often transferred into the hands of a small number of people. For example, Levin and Satarov (2000) investigate corruption and organisations in Russia, and raise the criticism that corruption is at the heart of Russia’s economy, therefore, to explore the connection between globalisation, corruption and extraordinary wealth we use an international perspective.

1.3 SPORT

The sports environment is a multibillion-dollar business. Every weekend millions of spectators are either watching sports competition at home on television or live at the venue. Sports are often characterised as winner-take-all markets and one feature of it is the fact the runners up are often only slightly worse. The diversity of jersey colours in sport makes such highly competitive environments interesting to analyse and investigate if marginal performance increases can be achieved by wearing a certain colour. Hill and Barton (2005) state that if the skill levels are fairly

Chapter 1: Introduction 9 10 equal, such as in close competitions, wearing red shirts could tip the balance between winning and losing. In this chapter we are in particular interested if the myth is true that teams wearing red are more successful. Colour has a wide-ranging and varied role in both nature and society. The colour red in nature and society is often utilised to signal danger (Humphrey, 1976). Gangs and gang members, besides wearing certain types of clothing and tattoos, have conventionally worn colours as a way of identifying themselves and the gang to which they belong.3 Additionally, different caste in India wear different colours (see Fehrman and Fehrman, 2004). Red has also been essential in the development of human psychology. For example, red is the first colour children learn to recognise (Garbini, 1894) or women wearing red dresses are considered more beautiful and desirable than females wearing any other colour (Elliot and Niesta, 2008). In nature the colour red is often associated with danger and dominance, not only in fish (Tinbergen, 1952) but also in higher order animals such as birds, primates and reptiles (Pryke, 2009). Hence, one can adopt the idea that in aggressive competitions the colour red could affect the outcome of sport contests. Most prior studies have examined if certain colours enhance the probabilities of winning in single competitions (see Greenlees et al., 2008; Dijkstra and Preenen 2007; Elliot et al., 2007; Ioan et al., 2007; Hill and Barton, 2005; Rowe et al., 2005; and Rehm et al., 1987) while only a few have investigated the effect of colours in team competitions (see Attrill et al., 2008; Ilie et al., 2008; Sutter and Kocher, 2008 and Hill and Barton, 2005). However, none of these studies, compared to our paper, have applied a multivariate analysis or analysed the National Rugby League (NRL) in Australia. The rugby league in Australia is particular interesting because it tries to maintain a fairly even competition by imposing a salary cap on the teams. Therefore one way to obtain a comparative advantage compared to other teams could be by wearing red jerseys as a general perception of red teams being more successful than any other team exists (e.g. Ferrari, Manchester United or Bayern Munich).

3 http://gangsorus.com/gang_colors.htm

10 Chapter 1: Introduction 11

1.4 BACKGROUND INFORMATION ON THE FOLLOWING FOUR CHAPTERS

The four essays presented here are collaborative works - three essays are co- authored with Professor Benno Torgler while one is co-authored with Professor Benno Torgler and David A. Savage. The first essay, entitled “Competition in Academia: Evidence from the American Economic Review” is an extended extract of a working paper called “A Century of American Economic Review”. Additionally the second essay, entitled “Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!”” is published in Economics Bulletin. Furthermore the third essay entitled “Extraordinary Wealth and Corruption” is revised and re-submitted to Review of Income and Wealth. The final essay included in this work entitled “The Red Mist? Red Shirts, Success and Team Sports” is forthcoming in Sport in Society.

Chapter 1: Introduction 11

Chapter 1: Introduction 12

Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Statement of Contribution of Co-Authors for Thesis by Published Paper

The authors listed below have certified* that:

1. they meet the criteria for authorship in that they have participated in the conception, execution, or interpretation, of at least that part of the publication in their field of expertise; 2. they take public responsibility for their part of the publication, except for the responsible author who accepts overall responsibility for the publication; 3. there are no other authors of the publication according to these criteria; 4. potential conflicts of interest have been disclosed to (a) granting bodies, (b) the editor or publisher of journals or other publications, and (c) the head of the responsible academic unit, and

5. they agree to the use of the publication in the student’s thesis and its publication on the QUT ePrints database consistent with any limitations set by publisher requirements.

In the case of this chapter:

Competition in Academia: Evidence from the American Economic Review is an extended extract of a working paper: A Century of American Economic Review (2011).

Contributor Statement of contribution*

Marco Piatti Has equally contributed to all aspects of this paper, including research, analysis and writing.

15/06/2012

Has equally contributed to all aspects of this paper, including research, Benno Torgler* analysis and writing.

Principal Supervisor Confirmation

I have sighted email or other correspondence from all Co-authors confirming their certifying authorship.

Benno Torgler 15 June 2012 ______Name Signature Date

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 13 14

2.1 INTRODUCTION

Academic journals play a vital part in the communication of scientific knowledge and new ideas. Academic articles exist as a reliable source for information to academics and practitioners alike (Harzing, 2002). In recent times, however, the academic system has become even more competitive, Frey (2005) describes it as incredibly time constrained for assistant professors and graduate students. In some cases it is an all or nothing proposition, either they publish in a top academic journal or, they have to bury their aspiration of becoming an academic. Academia in general has developed into a “battle of attention” due to excess of papers produced every year to which economics is no exception. Such proliferation is illustrated by the development of IDEAS, the largest bibliographic database dedicated to economics. It is freely accessible on the WEB and tries to enhance the distribution of research in economics (see http://ideas.repec.org/). This database contains information on 11,975 institutions, covering 26,852 authors registered with the RePEc Authors Service who have authored 539,619 items listed in the archive. In 2010, the information dissemination service New Economic Papers (NEP) sent out 4,448 weekly reports about new research based on 87 fields, and the RePEc service recorded almost nine million downloads and 31 million abstract views (see http://blog.repec.org/, accessed January 6, 2011). This level of activity does clearly characterize the highly competitive environment in academia.

Numerous papers have developed journal rankings as a means to determine the most prestigious journal in which to publish one’s research. In most of which (see Kalaitzidakis et al., 2010, 2003; Wall, 2009; Engemann and Wall, 2009; Kodrzycki and Yu, 2005) the American Economic Review (AER) together with the Quarterly Journal of Economics (QJE), the Journal of Political Economy (JPE) and Econometrica are considered to be the top-tier economic journals. An interesting approach to rank economics journals is the study by Axarloglou and Theoharakis (2003) who conducted an online survey among members of the American Economic Association (AEA). Their study tried to reveal an answer to the questions such as which are the best journals in economics and which of the journals are most read by economists. They observed that their respondents ranked the AER as number one followed by, JPE, Econometrica and QJE. Furthermore, their respondents also disclosed that the AER is read the most followed by the Journal of Economic

14 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Perspectives and the Journal of Economic Literature. Since the AER is considered to be one of the most prestigious as well as one of the most read journals, amongst members of the AEA, this paper takes a very close look at it. Established in 1911, the AER is the first journal of the AEA. Stigler, Stigler, and Ferdinand (1995) refer to it as the flagship of the North American economic society. The AER has considerably influenced the economics landscape over the last 100 years. A century after Adam Smith’s two publications in Europe, of An Inquiry into the Nature and Causes of the Wealth of Nations, saw an evolving of economic journals (Diamond, 1988). Previously, serious economics was only published in books or nonspecialist periodicals. The QJE, introduced in 1886, was the first fully professional outlet in the US (Stigler, Stigler, and Friedland, 1995). Two additional US journals came into being within the next fifteen years (JPE and the AER). In Europe one of the first economics journals, however not published in English, was DeEconomist founded in 1852 in the Netherlands. The Economic Journal founded in 1891 was perhaps the most prominent journal in Europe or the globe in the early 1900s. Especially since one of the most famous economists of all time, John Maynard Keynes (1911-1946), was its editor.

In 1911 the first article published in the AER was written by a distingiuished female , Katharine Coman (1857-1915), who in the early 1880s, was also the first American woman to have obtained a professorship of statistics (Vaughn, 2004). During the early 1900s, she also chaired the Department of Economics and was the Dean of Wellesley College. The first AER editor, Davis Rich Dewey, was in charge of managing the journal for a period of three decades, a longer appointment than any editor since. Since then the AER has had nine additional editors: Paul T. Homan (1941-1951), Bernard F. Haley (1952-1962), John G. Gurley (1963-1968), George H. Borts (1969-1980), Robert W. Clower (1981-1985), (1985-2001), Ben S. Bernanke (2001-2004), Robert A. Moffitt (2004-2010), and Pinelopi Koujianou Goldberg (2011-present). The appears both at the beginning and the end of the AER’s century, since both Davis Rich Dewey and Robert A. Moffitt have been affilited with Johns Hopkins during their terms as editors of the journal.

To generate our dataset, we collected data for a period of a hundred years mainly utilizing resources that should be accessible to most economists, such as AER

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 15 16 articles or articles discussing AER contributions and occasionally Journal Citation Reports.

2.2 TOP INSTITUTIONS, TOP PAPERS, AND LEADING ECONOMISTS PUBLISHING IN THE AER

Given that there seems to be a natural desire for distinction (see e.g. Frank and Cook, 1995; Frank, 1985, 1991, 1999; and Frank and Sunstein, 2001), it is not surprising that a vast amount of papers have emerged that discuss the rankings of economics departments and researchers (see Table 2.1) as well as an impressive development on the economic research on the evaluation of scientific progress (Amir and Knauff, 2008). The demand for rankings is comprehensible as academics in the field are eager to acquire knowledge on the best institutions in which to continue their research. University administrators also find rankings helpful for assessing the progress of their departments and for developing student recruitment strategies (Scott and Mitias, 1996). More important, as indicated by Laband and Tollison (2003) is the significant increase for faculties to be more research active. This emphasis on more research can not only be seen in the US but also in Canada, Australia as well as a number of countries in Europe and South America. Laband and Tollison (2003) continue to state that in the US and Canada the incentive to be more research productive can lead to a reduced teaching load, marginally higher raises for faculty who publish in peer-reviewed journals as well as financial support for conference travel. In other countries it is less likely to obtain a lower teaching load but there are financial rewards to publishing. For example, Butler (2003) states that, to universities in Australia, every published article in a peer reviewed journal is worth more than A$3,000 and a book A$15,000. Furthermore, the Excellence in Research for Australia (ERA) Initiative assesses research quality within Australia's higher education institutions using a combination of indicators and expert review by committees comprising experienced, internationally-recognized experts (Australian Research Council (ARC), 2011).4 Until just recently the ARC provided a downloadable excel file (ERA journal list) in which it categorized journals into either A* (being the best), A, B, C or not ranked. However, from 2012 the ERA journal list

4 http://www.arc.gov.au/era/ (accessed 02.11.2011)

16 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

will not provide a classification for journals anymore. In a media release in May 2011 Senator the Hon Kim Carr justifies this action by stating that5:

“There is clear and consistent evidence that the rankings were being deployed inappropriately within some quarters of the sector, in ways that could produce harmful outcomes, and based on a poor understanding of the actual role of the rankings. One common example was the setting of targets for publication in A and A* journals by institutional research managers. In light of these two factors – that ERA could work perfectly well without the rankings, and that their existence was focusing ill-informed, undesirable behaviour in the management of research – I have made the decision to remove the rankings, based on the ARC’s expert advice.”

Frey (forthcoming) argues that today “the importance of scientific idea and the position of a scholar are defined by rankings. What matters nowadays is the recognition produced by a general rankings system, normally based only on the quantity of scientific output, irrespective of quality” (p. 2). He continues to state that rankings offer straightforward measures of relative position in science, which is a particular positive characteristic for academicians from other fields and public decision makers. Nevertheless, as Osterloh and Frey (2010) point out rankings suffer from serious shortcomings. For instance, rankings based on citations do not take into consideration that articles are often referred to because they are wrong and not because they are considered to be a valuable contribution to knowledge. In addition, rankings suffer from a mis-citation bias against names from unfamiliar languages as their names are more frequently spelled incorrectly and hence, citations cannot be assigned to the corresponding authors (Kotiaho, 1999; Kotiaho et al., 1999). This contributes to an undercitation bias for names from unfamiliar languages compared to names from English speaking countries (Kotiaho et al., 1999). Furthermore, as Mayer (2004) points out there also exist so called “hat tipping” citations, which are made to please authors that could be potential referees, to demonstrate that the relevant literature has been studied or even in the hope that cited authors will do the same in return. Therefore, rankings based on citation analysis should be looked at carefully.

5 http://minister.innovation.gov.au/Carr/MediaReleases/Pages/IMPROVEMENTSTOEXCELLENCEINRESEARCHFORAUST RALIA.aspx (accessed 12.06.2012)

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 17 18

Interestingly, researchers at present devote a significant amount of time to have their research screened, which indicates the desire to publish in a top journal. The competitive environment in academia produces an excess of papers every year. As van Dalen and Klamer (2005) point out the excess of papers in science create an environment in which scientists cannot pay attention to every article that is published. Hence, to distinguish one’s work from all other articles, scientist are almost obliged to participate in the attention game to promote their work. Laband and Tollison (2003) state that in 1974, only 19% of the papers published in the AER had been presented at one or more conferences, workshops or seminars for critical commentary prior to publication. On average, the number of presentations per paper was 0.24. Twenty-five years later, the accepted papers which had privously been presented increased significantly to 73% and the mean number of pre-publication presentations has risen to 4.73. Likewise, the average number of informal contributors thanked per AER article has grown from 4.33 to 9.59.

2.3 INSTITUTIONAL RANKINGS

To produce Table 2.1, which summarizes our aggregations, we first aggregated the institutional ranking results presented in numerous previous publications.6 We looked at all the rankings in these papers and counted the number of times a university appeared in the top 10 (first results column) or the top 20 (second results column). This table may therefore provide a simple overview of institutional historical strength that takes into account the advantages and limitations of different ranking methods and approaches. Table 2.1 clearly indicates the dominance of US institutions. More specifically, in the first fifteen positions are only US universities to find and of the 23 universities listed only three are outside the US namely the Hebrew University, the University of Western Ontario and the London School of Economics (LSE). In first place is MIT, followed by Harvard and Chicago. Moreover, by comparing private with public universities it is apparent that private institutions outperformed public schools. Only two public universities are positioned in the first 10 positions and both of which are campuses from the University of California, Berkeley as eighth and Los Angeles as tenth. Private universities, even

6 In many cases, we observe single rankings that accumulate a relatively large number of journals together. For example, Loren C. Scott and Peter M. Mitias (1996) used 36 journals to develop a university ranking.

18 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

though more prestigious and admired, are not necessarily superior to public universities.7 Clearly the most observable distinction between American public and private universities is the cost of tuition, which is significantly higher at a private university.8 Assuming no financial constraints it really depends on the student himself / herself to determine which university system (public or private) is most suitable to his / her abilities. In a next step it might also be interesting to examine whether the top tier universities are located at the East or the West coast of the US. The result indicates that in the top ten, five institutions are from the East coast, MIT, Harvard, Princeton, Pennsylvania and Yale, two institutions are located at Lake Michigan (Chicago and Northwestern) and three Universities, Stanford, Berkley and Los Angeles are positioned at the West coast. As two of these studies only utilized US data,9 we present in the appendix Table A1 an adjusted ranking with these two papers excluded. The ranking is almost the same with only minor differences. For instance, Columbia University is now in the top 10 and the University of California, Berkeley has improved its position among the top 10 universities.

7 http://www.brainchild.org/publicORprivateU.html 8 http://www.brainchild.org/publicORprivateU.html 9 Richard Dusansky and Clayton J. Vernon (1998) and Loren C. Scott and Peter M. Mitias (1996).

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 19 20

Table 2.1: A SUMMARY OF INSTITUTIONAL RANKINGS Appearance as a Appearance as a University Top 10 University Top 20 University Institute of Technology 38 38 Harvard University 34 36 University of Chicago 32 36 Stanford University 31 37 Princeton University 30 36 University of Pennsylvania 28 33 Yale University 27 32 University of California, Berkeley 26 33 Northwestern University 22 32 University of California, Los Angeles 15 31 Columbia University 13 26 University of Michigan 12 26 University of Wisconsin 11 29 Carnegie Mellon University 7 17 New York University 6 25 Hebrew University 5 8 University of Washington 4 8 Rochester University 4 24 University of Western Ontario 3 10 London School of Economics 3 13 University of Minnesota 3 20 Brown University 3 10 Cornell University 3 16 Notes: Data from Tom Coupé (2003), table 2 and table 4 (covering two time periods, 1978– 1982 and 1996–2001; four different rankings); two tables from Philip E. Graves, James R. Marchand, and Randal Thompson (1982), table 1 and table 2; table 1 from Richard Dusansky and Clayton J. Vernon (1998); table 3 from Pantelis Kalaitzidakis, Theofanis P. Mamuneas, and Thanasis Stengos (2003); five from Erkin Bairam (1994), table 1 (AER 1985–90), table 2 (Econometrica 1985-90), table 3 (Economic Journal 1985–90), table 4 (JPE 1985–90) and table 5 (QJE 1985–90); table 1 from Amir and Knauff (2008); three tables from Stephen Wu (2007), table 2 (AER), table 3 (JPE), and table 4 (QJE) between for the 2000–2003 period; eight from Scott and Mitias (1996), table 1 (1984–93), table 3 (1984–93), table 4, tables 5, 6, 7 (a comparison of the Top 5 in five journals); and 12 by John J. Siegfried (1994), table 1 (AER, by decade between 1950 and 1989), table 2 (JPE, by decade between 1950 and 1989) and table 3 (QJE, by decade between 1950 and 1989) and table 2 by Jean Louis Heck (1993).

The subsequent four tables look entirely at single main articles that were published over the 100 years of the AER’s existence (1911–1920 and 1931–1940 in Table 2.3 and Table 2.4; 1950–1959 in Table 2.5; and 1981–1990 and 2001–2010 in Table 2.6).10 For the 1950–1959 period only, we rely on results of a previous published article in the AER by Cleary and Edwards (1960). To follow the Cleary and Edwards (1960) approach, when an article was multi-authored, the number of

10 When more than one author affiliation was listed, we used the author’s main affiliation.

20 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

pages was divided evenly between the authors’ affiliations. Additionally, to observe the variation of a university’s position over time we also included a column which indicates the change of position compared to the previous table.

Table 2.3 and Table 2.4, focuses on the periods 1911–1920 and 1931–1940, present universities which have published one hundred pages or more in the AER. In his 2004 paper Samuelson states that in 1935 only a few strong economic research universities existed such as Harvard, Chicago, Columbia and a few others.

As we can see in Table 2.3 and Table 2.4 besides these universities we also find Princeton, Yale, New York, Minnesota, Cornell, Illinois, Wisconsin or Ohio State at the top. These six (1911–1920) or eight (1930–1940) universities produced between 38% of all pages published in the AER in the earlier and 35% in the later period. Geiger (1986) points out that, for few universities, estimates of total expenditures for research in the late 1930s accumulated to impressive levels. As depicted in Table 2.2 most of institutions that invested a significant amount of funding to research are also the top contributors to AER in Table 2.4. The only exceptions of which are New York and Ohio State University which do not belong to this elite club of large contributors to research but nonetheless supplied a considerable number of pages to AER. Examining the changes of universities positions in Tables 3 and 4 the university which increased its ranking the most is the University of Chicago by 16 places. Yale, on the other hand, plummeted ten positions and is now ranked eleventh. The university that contributed the most pages to AER in the 1930s is Columbia, which climbed three positions.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 21 22

Table 2.2: TOTAL EXPENDITURES FOR RESEARCH IN THE LATE 1930S

More than US$ 2,000,000 US$1,000,000 – 1,500,000 California MIT Chicago Pennsylvania Columbia Harvard Illinois Michigan US$ 1,500,000 – 2,000,000 Under US$ 1,000,000 Cornell Johns Hopkins Minnesota Princeton Wisconsin Stanford Yale Caltech Note: Source Geiger (1986).

Table 2.3: TOP INSTITUTIONS 1911–1920

1911–1920

Number of Percent of Total Institutions Pages Pages Yale University 324 9.2 Harvard University 254 7.2 Princeton University 228 6.5 Columbia University 215 6.1 University of Illinois 172 4.9 Cornell University 140 4.0 Total 1333 37.9 Notes: Institutions whose total contribution to AER during the period was 100 pages or more (based on author affiliation). Only primary journal articles are counted (i.e., Papers and Proceedings are excluded).

Table 2.4: TOP INSTITUTIONS 1931–1940

1931–1940 Number of Percent of Total Change of position Institutions Pages Pages from previous Table Columbia University 248 6.8 +3 Harvard University 197 5.4 = Princeton University 176 4.8 = University of Wisconsin 141 3.9 +6 New York University 133 3.7 +2 University of Minnesota 133 3.7 +3 University of Chicago 130 3.6 +15 Ohio State University 123 3.4 +5 Total 1281 35.2 Notes: Institutions whose total contribution to AER during the period was 100 pages or more (based on author affiliation). Only primary journal articles are counted (i.e., Papers and Proceedings are excluded). Washington, D.C., although it contributed 125 pages, is excluded because no affiliation was provided.

22 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Table 2.5 presents all institution, which contributed 100 pages or more for the period between 1950 and 1959 to the AER. As depicted in Table 2.5 several new universities, such as University of California, MIT, Stanford, Johns Hopkins, Pennsylvania, Vanderbilt and Carnegie Institute of Technology appeared in the ranking alongside with the Federal Reserve System and the International Monetary Fund. Investigating the history of MIT, Stanford and Johns Hopkins a bit closer, it is of no surprise that they climbed on average by 18 positions in Table 2.5. After WWII, MIT, under the entrepreneurial leadership of Rupert MacLaurin, expanded its economics department significantly and by the 1950s it was established as one of the World’s leading centres of economic research11 hosting academics such as , Charles P. Kindleberger and . Furthermore, in the early 1960s another person of influence in economics namely joined the department at MIT. Similarly after WWII, Stanford, under the leadership of Bernard Haley, Edward Shaw, and Moses Abramovit, rose to national and international fame in economics. Additionally, a future Nobel Laureate joined the department in 1949.12 Johns Hopkins was also no exception. By the 1950s the economics department employed only seven full-time faculty members. Nonetheless, it had established itself as one of the leading economics department in the country. Evsey Domar, (who later won a Nobel Prize), and were part of the faculty. In addition, it produced two future Nobel Prize winners: Merton Miller (Ph.D. ’53) and (Ph.D. ’63).13 Furthermore, not only are the top seven institutions responsible for around one-third of the published pages, but all 17 institutions combined account for more than 60% of all the pages contributed to the AER.

As the number of universities, which contributed one hundred or more pages, increased substantially, 29 institutions in the 1980s and 42 in the last decade respectively, Table 2.6 provides an overview for institutions that published more than two hundred pages in the AER in more recent times. More specifically, we looked at the periods 1981-1990 and 2001-2010. There are at least two possible explanations why the number of institutions that contributed more than one hundred pages rose substantially. One of which is the increase in the length of papers published in the

11 http://econ-www.mit.edu/about/. 12 http://economics.stanford.edu/department. 13 http://econ.jhu.edu/about/history/.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 23 24

AER over time. Examining these two periods we observe a 60% increase in length. Another explanation appears to be the fact that the 2001–2010 data indicate a decrease of the concentration of universities on the top of the list. Whereas in the 1950s, 17 institutions together were accountable for more than 60% of all pages published in the AER, during the first decade of this century, 18 institutions provided only 46% of all pages. Moreover, for the first time we not only find institutions from the US but also from Canada (University of Western Ontario and University of British Columbia) and the United Kingdom (LSE) on the list. The University of California, if considered as one institution, is undoubtedly the institution which publishes the most pages in the AER since the 1950s.

Table 2.5: TOP INSTITUTIONS 1950–1959 Change of position Number of Percent of Total Institution from previous Pages Pages Table University of California 392 6.9 +37 Massachusetts Institute of Technology 363 6.4 +22 Stanford University 309 5.4 +14 University of Chicago 218 3.8 +3 University of Michigan 214 3.8 +7 Federal Reserve System 200 3.5 — Johns Hopkins University 199 3.5 +16 University of California, Los Angeles 197 3.5 — Harvard University 185 3.3 -7 Yale University 164 2.9 +1 University of Wisconsin 158 2.8 -7 University of Pennsylvania 135 2.4 +2 Princeton University 134 2.4 -10 University of Illinois 133 2.3 -3 Vanderbilt University 112 2.0 — Northwestern University 111 2.0 +15 Carnegie Institute of Technology 102 1.8 +33 International Monetary Fund 100 1.8 — Total 3426 60.5 Notes: Institutions whose total contribution to AER during the period was 100 pages or more (based on author affiliation), but Papers and Proceedings are excluded. Data from this time period are from Cleary and Edwards (1960: 1012). A contribution was defined to include articles, review articles, notes, communications, and memorials; book reviews are excluded.

24 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Table 2.6: TOP INSTITUTIONS 1981–1990 AND 2001–2010 1981–1990 Number of Percent of Change of position Institutions Pages Total Pages from previous Table Massachusetts Institute of Technology 539 4.7 +1 Princeton University 510 4.5 +11 Harvard University 500 4.4 +6 University of Chicago 432 3.8 = University of California, Los Angeles 333 2.9 +3 University of Pennsylvania 312 2.7 +6 University of Michigan 283 2.5 -2 Stanford University 267 2.3 -5 University of California, Berkeley 260 2.3 — University of Western Ontario 228 2.0 — Yale University 200 1.8 -1 Total 3864 33.9 2001–2010 Number of Percent of Change of position Institutions Pages Total Pages from previous Table Harvard University 1049 5.5 +2 University of California, Berkeley 875 4.6 +7 University of Chicago 758 4.0 +1 Stanford University 749 3.9 +4 Massachusetts Institute of Technology 651 3.4 -4 Yale University 493 2.6 +5 Princeton University 483 2.5 -5 Columbia University 479 2.5 +13 University of Pennsylvania 456 2.4 -3 New York University 399 2.1 +5 University of California, Los Angeles 382 2.0 -6 Northwestern University 377 2.0 = University of Michigan 343 1.8 -6 London School of Economics 299 1.6 +21 University of Maryland 267 1.4 +7 University of British Columbia 253 1.3 +16 University of California, San Diego 204 1.1 +36 Carnegie Mellon University 201 1.1 +17 Total 7899 45.9 Notes: Institutions that contributed 200 pages or more to AER (based on authors affiliation). Only primary journal articles are included (Papers and Proceedings are excluded). Figure 2.1 illustrates positional changes for selected universities over time. MIT was omitted in Figure 2.1 because we only included top institutions for which we collected data over the entire one hundred years and MIT was not ranked in the 1911 – 1920 period. As can be seen the University of Chicago increased its rank noticeably from position 22 in the first period 1911–1920 to seventh in 1931–1940 and rose further to third in the last decade. Harvard and Princeton have almost an identical shape of their lines. Both started out very high in the ranking kept their position for the first 20 years and dropped considerably in the 1950s following by sharp increase in the 1980s. In the last decade Princeton and Harvard’s curve

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 25 26 deviated, while Princeton dropped back to seventh place Harvard climbed to the number one position.

Figure 2.1: FLUCTUATIONS OF UNIVERSITIES POSITION OVER ONE HUNDRED YEARS 22 19 16 13 10 RankingPosition 7 4 1

1911-1920 1931-1940 1950-1959 1981-1990 2001-2010

Chicago Harvard Pennsylvania Princeton Yale

Having investigated the publication record based on authors’ institutions we thought that in a further step it is also interesting to examine at which institutions the authors completed their doctoral degree. By studying at one of these top institutions, a student generates human capital skills that allows a student to perform better academically in the future. Furthermore, students at a top institution are influenced by their teachers who normally are well-established economists such as Samuelson or Simon. For instance, while Samuelson was a professor at MIT, Joseph Stiglitz and Robert C. Merton, who have both won the Nobel Prize in Economics, graduated from there as well. Additionally a student generates the ability to facilitate the use of other types of capital (for example, social capital). There exists a high possibility that relationships between fellow students and/or faculty members are formed, who would most likely be editors, co-editors or board members of these top tier journals.

Given the difficulties of determining the Ph.D. affiliation of authors in the earlier periods, we decided to reduce our analysis to the last 60 years of contributions to the AER and at the same time analyse the data using shorter five year time periods

26 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

(1984–1988 and 2004–2008). The results clearly indicate that there is much stronger concentration of a few universities at the top. Researchers with a Ph.D. from the ten universities listed, supplied between 56.4% (1984–1988) and 54.4% (2004–2008) of all contributors in the AER. The top ten universities listed in the 1950s contributed around 75% of all contributors in the AER. In the more recent periods, from 1984– 1988 and 2004–2008, a clear dominance (especially in the later period) of MIT and Harvard emerges as in both periods these two institutions provided more than 20% of all contributors. In line with the results of Table 2.1, MIT is again ranked higher than Harvard. The only non-US institution appearing on one of these top 10 lists is LSE in the 2004–2008 period.

Table 2.7: TOP TEN INSTITUTIONS BASED ON CONTRIBUTORS’ PH.D. INSTITUTION

1950-1959

Number of Percent of Total Sample Institutions Contributors of 287 Contributors Harvard University 62 21.60 Columbia University 38 13.24 University of Chicago 34 11.85 University of California 20 7.00 University of Wisconsin 13 4.53 Yale University 12 4.18 University of Pennsylvania 11 3.83 University Michigan 10 3.48 Cornell University 8 2.79 Stanford University 8 2.79 Total 216 75.29 Notes: Data for the years 1950 to 1959 are from Frank R. Cleary and Daniel J. Edwards (1960). We include short papers (e.g., notes, comments, replies) but exclude Papers and Proceedings items.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 27 28

Table 2.8: TOP TEN INSTITUTIONS BASED ON CONTRIBUTORS’ PH.D. INSTITUTION

1984-1988 Percent of Total Number of Change of postion Institutions Sample of 691 Contributors from previous table Contributors Massachusetts Institute of Technology 81 11.7 — Harvard University 59 8.5 -1 University of Chicago 48 6.9 = Stanford University 37 5.4 +6 University of California, Berkeley 37 5.4 — Princeton University 31 4.5 — Yale University 27 3.9 -1 University of California, Los Angeles 26 3.8 — University of Pennsylvania 23 3.3 -2 Northwestern University 21 3.0 — Total 390 56.4 2004-2008 Percent of Total Change of position Number of Institutions Sample of 815 from previous Contributors Contributors Table Massachusetts Institute of Technology 108 13.3 = Harvard University 87 10.7 = Stanford University 43 5.3 +1 Princeton University 43 5.3 +3 University of Chicago 34 4.2 -2 University of California, Berkeley 31 3.8 -1 University of Minnesota 25 3.1 +7 Yale University 25 3.1 -1 London School of Economics 24 2.9 17 Northwestern University 23 2.8 = Total 443 54.4 Notes:We include short papers (e.g., notes, comments, replies) but exclude Papers and Proceedings items. Contrary to most rankings of economics departments Amir and Knauff (2008) propose an alternative method. They rank economics departments not according to their research productivity but on the strengths of their Ph.D. program as measured by a department’s ability to place doctoral students in top tier universities or business schools. Their ranking also indicates a clear dominance of MIT, followed by Harvard, Stanford, Princeton, Chicago, Yale, Berkeley, Oxford, Minnesota, Northwestern and LSE. Their ranking is very similar to the results presented in Table 2.8 for recent years. The Spearman correlation coefficient between these two rankings is rather strong with 95.2 percent.

28 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

2.4 COUNTRY RANKINGS

Rather than focusing only at the micro level (institutions), we thought it might also be interesting to look at the macro level. More specifically, to investigate how individual countries perform.14 In terms of the number of publication per capita15, Israel is in both periods ranked number one followed by the US. Kocher and Sutter (2001) find a similar result by examining 15 economics journals for a period that includes 1977, 1982, 1987, 1992 and 1997. Analysing the whole period at once they also observe that Israel was the most successful country followed by the US. By comparing Kocher and Sutter’s (2001) ranking for current affiliations with Table 2.9 in this thesis it can be seen that the two rankings are reasonably similar in respect to the countries included, however, the position of the individual countries, except for the first two, varies. These results indicate that when taking into consideration the population size of a country, the dominance of the US diminishes. Additionally, Australia dropped from position four in the former period (1984-88) to position 15 (not shown) in the latter period (2004-08). As Butler (2003) points out in the mid 1990s the funds to Australian universities were for the first time distributed according to a number of performance measures such as graduate student numbers or completion rates, research income and publications. Furthermore, Butler (2003) mentions that every published peer reviewed journal article (no differentiation on the quality of the journal) in Australia was worth more than A$3,000 to a tertiary institution. She continuous to state that without any endeavour to distinguish between qualities of different journals when allocating funds to universities, there is little encouragement to attempt publication in prestigious journals. It appears that this has led to a significant higher publication output in the last decade in lower-ranked journals and could be a potential reason why Australia has lost its strength looking at AER.

14 For previous studies in that area, see, e.g., Geoffrey M. Hodgson and Harry Rothman (1999) or Martin Kocher and Matthias Sutter (2001). 15 This approach assumes that academic resources devoted to economics are highly correlated with a country’s population. Such may not, however, be the case for populous developing countries (Martin Kocher and Matthias Sutter 2001). This problem can thus be mitigated by looking at only the top 10 countries.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 29 30

Table 2.9: TOP COUNTRIES PUBLISHING IN AER

Period 1984– Period 2004– Rank 1984– Rank 2004– Country 1988 2008 1988 2008 Israel 5.35 2.55 1 1 USA 3.17 2.37 2 2 Switzerland 0.61 1.33 6 3 Sweden 0.72 1.32 5 4 Norway − 1.28 − 5 United Kingdom − 1.00 − 6 Canada 2.10 0.95 3 7 Netherlands − 0.79 − 8 Ireland 0.57 0.71 8 9 Belgium 0.41 0.57 10 10 Australia 0.75 − 4 − New Zealand 0.61 − 7 − Finland 0.41 − 9 − Notes: Number of AER publications; top 10 countries in each period; per capita in millions.

2.5 TOP PAPERS

Having looked at institutions and countries in a next step we thought it might be interesting to analyse the top papers, publishing frequency and the most successful authors that publish in the AER. To analyse the former we adopt an imperfect but widely available proxy for quality, namely citations. Citations are an imperfect measure of quality because often citations do not reflect the scientific merit of the citied work. Furthermore, not all citations are worth the same. Having a citation in the AER is of much higher value than citations from less successful journals owing to the greater likelihood of additional future citations (Johnson, 1997). Further possible problems of citation analysis are self-citations, the increase in the total number of citations over time and the positive correlation between the number of authors and accumulation of citations (Bornmann and Daniel, 2008). Numerous studies (e.g. Bodenhorn, 2003; Laband, 1990) argue that self-citations are solely self-serving and leaving them in the analysis will overestimate the significance of the result. However, Medoff (2006) in his analysis does not find that self-citations have a statistically significant effect on the total number of citations an article receives. Additionally, Costas et al. (2010) states that author self-citations are central in the normal progression of scientific communication, as authors need to refer to their earlier work as an indication of continuity in their career. Furthermore, they also state that co-author self-citations are a way of indicating the transfer of knowledge between those who created the original idea. We therefore do not exclude self- citations from our analysis. More specifically we used the 2009 Journal Citation

30 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Reports16 to obtain the results presented in Table 2.10. The most cited papers to date are Armen Alchian and Harold Demsetz’s (1972) study “Production, Information Costs, and Economic Organization,” followed by Michael C. Jensen’s paper “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.”17 Remarkably, two authors have more than one paper in the top 11 list of the most cited papers; Joseph Stiglitz appers three times and Carl Shapiro two times.

Table 2.10: TOP 10 AER PAPERS BY CITATION Publication Title Authors Citations Date PRODUCTION, INFORMATION COSTS, ALCHIAN, AA 62(5) 1972 2215 AND ECONOMIC ORGANIZATION DEMSETZ, H *AGENCY COSTS OF FREE CASH FLOW, CORPORATE FINANCE, AND JENSEN, MC 76(2) 1986 2014 TAKEOVERS CREDIT RATIONING IN MARKETS STIGLITZ, JE 71(3) 1981 1462 WITH IMPERFECT INFORMATION WEISS, A MONOPOLISTIC COMPETITION AND DIXIT, AK 67(3) 1977 1421 OPTIMUM PRODUCT DIVERSITY STIGLITZ, JE THE USE OF KNOWLEDGE IN HAYEK, FA 35(4) 1945 1159 SOCIETY EQUILIBRIUM UNEMPLOYMENT AS A SHAPIRO, C 74(3) 1984 1053 WORKER DISCIPLINE DEVICE STIGLITZ, JE A SENSITIVITY ANALYSIS OF CROSS- LEVINE, R 82(4) 1992 1050 COUNTRY GROWTH REGRESSIONS RENELT, D ROLE OF MONETARY POLICY FRIEDMAN, M 58(1) 1968 1000 NETWORK EXTERNALITIES, KATZ, ML 75(3) 1985 993 COMPETITION, AND COMPATIBILITY SHAPIRO, C MIGRATION, UNEMPLOYMENT AND HARRIS, JR DEVELOPMENT —2-SECTOR 60(1) 1970 974 TODARO, MP ANALYSIS *CLIO AND THE ECONOMICS OF DAVID, PA 75(2) 1985 974 QWERTY Notes: *Articles in the Papers and Proceedings (entitled AER Papers until vol. 100, no. 4). Data based on the 2009 Journal Citation Reports including the reloading on September 22, 2010. Data accessed December 2010.

2.6 PUBLISHING FREQUENCY

Using data that covers almost the entire century of the AER we present in Table 2.11 the frequency with which authors publish.18 As depicted in Table 2.11, the majority of authors, 67%, only published once, 16% published twice and 7% published three times in the AER,

16 This dataset includes a reloading on September 22, 2011, of data originally collected in December 2010. 17 When Google Scholar is used as the search engine, these two papers are also ranked top among the papers reported in Table 7 (accessed January 25, 2011). 18 Until vol. 100, no. 4.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 31 32

Table 2.11: DISTRIBUTION OF PUBLICATIONS AMONG AUTHORS (1911–2010)

NUMBER OF PUBLICATIONS FREQUENCY PERCENT 1 4288 66.98 2 1046 16.34 3 464 7.25 4 214 3.34 5 140 2.19 6 86 1.34 7 44 0.69 8 35 0.55 9 22 0.34 10 21 0.33 11 14 0.22 >11 28 0.45 Total 6402 100 Notes: Measured up until vol. 100, no. 4; these numbers include Papers and Proceedings. indicating that only a minority (less than 10%) published more than three times in the AER. Figure 2.2 represents, what we would call, a citation Lorenz curve of a citation inequality proxy of all articles published between 1911 and 2010. This figure at the bottom right hand side indicates a highly unequal distribution of citations with a Gini coefficient of over 0.77. In other words 20% of the articles account for 80% of all citations. Investigating the Gini coefficient for the different periods, it emerges that the citation inequality was highest between 1959-59 with a Gini coefficient of 0.86, followed by 0.85 in the 1930s and 0.80 in the earliest decade 1911-1920.

32 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Figure 2.2: LORENZ CURVES OF CITATIONS FOR THE DIFFERENT TIME PERIODS

Lorenz Curve 1911 – 1920 Lorenz Curve 1931 – 1940 1 1 .8 .8 .6 .6 .4 .4 Fraction of all citation Fraction of all citation Gini Coefficient 0.80 Gini Coefficient 0.85 .2 .2

0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative publication proportion Cumulative publication proportion

Lorenz Curve 1950 – 1959 Lorenz Curve 1981 – 1990 1 1

.8 .8

.6 .6

.4 .4

Fraction of all citation Gini Coefficient 0.86 Fraction of all citation Gini Coefficient 0.70

.2 .2

0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative publication proportion Cumulative publication proportion

Lorenz Curve 2001 – 2010 Lorenz Curve 1911 – 2010 1 1 .8 .8 .6 .6 .4 .4 Fraction of all citation Gini Coefficient 0.66 Fractionof all citations Gini Coefficient 0.77 .2 .2 0 0

0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative publication proportion Cumulative publication proportion

However, as we collected the information on the total number of citations in December 2010 the newer published articles will not have accumulated a large

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 33 34 number of citations, therefore, may distort the result that we obtained for the 2001- 2010 Gini coefficient. These Gini coefficients are even higher as in the sports environment, which is normally considered to be a highly competitive arena. Fort (2003) for example, reports Gini coefficients for a tournament outcome in men’s and women’s pro golf of 0.635 (men) and 0.621 (women) but argues that in team sports, the values are lower (up to 0.5). This is in line with Scully (2004) who reports Gini coefficient for the four national sports leagues in the USA, for baseball (0.626), basketball (0.528), football (0.512) and hockey (0.458). It occurs that baseball generates the highest income inequality out of these team sports. This however, might be explained by the nature of the game as a single player (pitcher, batter) has a much greater influence on the game outcome as in the other team sports (Scully, 1995 as stated in Seaman, 2003). Moreover, as depicted in Table 2.12, in contrast to the decrease in the Gini coefficients for the citation Lorenz curve above, the team sport earning inequality is increasing.

Table 2.12: GINI COEFFICIENT FOR US TEAM SPORTS MLB NBA NFL NHL Year 1990 1998 1989 2000 1990 1998 1991 1999 Gini 0.482 0.626 0.428 0.528 0.397 0.512 0.319 0.458 coeff. Note: Source Scully (2004).

2.7 TOP AUTHORS

Because of the strong “superstar effect” illustrated in Figure 2.2, it should be interesting to analyse who the authors are that published 12 or more times in the 100 year history of the AER (including Papers and Proceedings).19 The results presented in Table 2.13 indicate that out of these 27 authors, eight have won the Nobel Prize; Joseph E. Stiglitz, James M. Buchanan, Vernon L. Smith, Franco Modigliani, , Paul A. Samuelson, George A. Akerlof as well as . In addition, out of these eight, four have also won the Medal namely Joseph E. Stiglitz, Paul A. Samuelson, Milton Friedman and James Tobin. When including Papers and Proceedings, Joseph E. Stiglitz is leading the ranking with 36 publications followed by William J. Baumol with 29, a position they maintain when

19 Data up until vol. 99, no. 5.

34 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Papers and Proceedings are excluded (Stiglitz: 24; Baumol: 22). In another column we report the total accumulated citations of the 27 authors paper published in the AER. As before Stiglitz and Baumol are leading the list. Additionally, Stiglitz is also leading the average citation obtained per article followed by George J. Stigler when Papers and Proceeding are included or Milton Friedman when they are not. Table 2.13 also reports so called ‘dry holes’, which are defined as publications that so far received no citations (Laband and Tollison, 2003). When Papers and Proceedings are included 18 out of the 27 superstars have at least one or more dry holes in their repertoire, a number that only slightly decreases from 18 to 15 when Papers and Proceedings are excluded. However, the absolute number of dry holes declines by more than half when Papers and Proceedings are omitted. Hence, not even these superstars are spared from having dry holes in their collections. For instance, Laband and Tollison (2003) have analysed 73 and 96 economics journals for the years 1974 and 1996 respectively, and discovered that on average 26% of articles are dry holes. However, not all dry holes are useless as some can terminate a particular research program or may settle or solve a puzzle in such a way that the papers are not cited (Mayer, 2004). Therefore, dry holes should not be seen as a waste but rather as a proof of healthy competition (van Dalen and Klamer 2005).

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 35

Table 2.13: TOP “SUPERSTARS” IN AER (12 AND MORE PUBLICATIONS) N AER Total Citations AVG Citations Dry N AER Total Citations AVG Citations Dry Author Publ. Nobel Prize John Bates Clark Medal Accumulated per Article Holes Publ.* Accumulated* per Article* Holes* STIGLITZ, JE 36 YES YES 6963 193.4 0 24 6161 256.7 0 BAUMOL, WJ 29 NO NO 2155 74.3 1 22 1945 88.4 1 CLARK, JM 24 NO NO 139 5.8 7 10 112 11.2 3 EISNER, R 20 NO NO 271 13.6 2 13 199 15.3 2 BUCHANAN, JM 19 YES NO 730 38.4 2 13 635 48.8 2 SMITH, VL 19 YES NO 1383 72.8 0 15 1067 71.1 0 FELDSTEIN, M 18 NO YES 573 31.8 2 9 469 52.1 0 BOULDING, KE 15 NO YES 118 7.9 2 4 20 5.0 1 MODIGLIANI, F 15 YES NO 1140 76.0 0 12 938 78.2 0 FISHER, I 14 NO NO 35 2.5 6 8 31 3.9 1 FRIEDMAN, M 14 YES YES 1288 92.0 3 9 1217 135.2 1 HELPMAN, E 14 NO NO 1273 90.9 0 14 1273 90.9 0 NELSON, RR 14 NO NO 813 58.1 2 6 281 46.8 1 ROTH, AE 14 NO NO 1098 78.4 0 14 1098 78.4 0 STIGLER, GJ 14 NO NO 1487 106.2 0 9 1151 127.9 0 HANSEN, AH 13 NO NO 119 9.2 6 7 114 16.3 3 REYNOLDS, LG 13 NO NO 83 6.4 3 6 73 12.2 0 SAMUELSON, PA 13 YES YES 891 68.5 1 10 837 83.7 1 AKERLOF, GA 12 YES NO 857 71.4 2 7 649 92.7 0 BACH, GL 12 NO NO 78 6.5 5 10 47 4.7 4 BLINDER, AS 12 NO NO 1023 85.3 0 5 665 133.0 0 FETTER, FA 12 NO NO 16 1.3 3 9 12 1.3 2 HART, AG 12 NO NO 10 0.8 7 6 6 1.0 3 SMITHIES, A 12 NO NO 26 2.2 3 5 12 2.4 1 STEIN, JL 12 NO NO 318 26.5 2 11 307 27.9 2 SUMMERS, LH 12 NO YES 862 71.8 0 6 511 85.2 0 TOBIN, J 12 YES YES 781 65.1 0 4 299 74.8 0 Notes: *AER Papers and Proceedings are excluded, as is R. G. Blakey (mainly Revenue Acts)

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 36

2.8 CONCLUSION

This paper presents several aspects of publishing within the AER over the last century. As far as we are understand this paper is the first that has analysed the skewness of citations by presenting Lorenz citation curves for the different decades over the hundred years. In our analyses we discovered that MIT is the leading institution in our generated “super ranking” (Table 2.1), followed by Harvard and Chicago. A clear dominance of institutions located in the US can be observed when analysing Table 2.1. Out of 23 institutions only three are located outside the US namely Hebrew University, London School of Economics and University of Western Ontario. However, as we are only investigating one particular journal the results may change significantly by analysing a larger sample of journals. Additionally as we were unable to obtain data for papers that got rejected we might suffer from a selection bias. Nonetheless we believe that our results are still informative and interesting to the economics profession. Additionally, we wanted to determine the best performing countries, in terms of publication per capita, in this metric we found that Israel outperformed the USA. This finding supports those of Kocher and Sutter (2001) who found the same ordering when examining 15 economics journals over a period of twenty years. Additionally, we observe that 54% of all contributions in the AER in all time periods came from the top ten institutions (see Table 2.7 and Table 2.8) based on authors Ph.D. affiliation.

These results indicate that the concentration at the top of academia in economics has been continuously decreasing from 75% of all contributions provided by authors of the top ten institutions in the 1950s to less than 55% in the last decade. This decrease of concentration is again mirrored in our citation Lorenz curves (see Figure 2.2) for all articles published in the AER over its one hundred year history. The overall Gini coefficient is 0.77, which indicates that just 20% of the articles are responsible for 80% of all citations. When we analyse the decade separately it shows some interesting changes, the citation inequality peaked in the 1950’s with a Gini coefficient of 0.86. The same decade in which the contributions of the top ten institutions based on Ph.D. was also at its highest. Interestingly, the overall Gini coefficient for our citation Lorenz curve is even higher than that observed in men’s (0.635) and women’s (0.621) pro golf (Fort, 2003) or in the four major US sports leagues MLB (0.626), NBA (0.528), NFL (0.512) and NHL (0.458) (Scully, 2004).

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 37 38

The significant skewness of citation inequality indicates that the publishing and citations in the AER may well be a market for superstars, with greater levels of inequality and superstardom than that usually reserved for the greatest sports stars.

The superstar effect has been investigated across many areas including sports economics; Hollywood economics (De Vany, 2004), cultural economics (Frey, 2000), and academia economics (Torgler and Piatti, 2011; Azoulay et al., 2010). These environments are generally characterized as winner-take-all markets, where a small heterogeneity in performance translates into large reward differences (Frank and Cook, 1995). This is exactly the situation that academics face, who’s participation in this market suffer from a distinct probability that one’s work goes unnoticed or uncited. Furthermore, publishing in a top journal is not a guarantee of success due to the strong citation inequality among AER papers. However, if an article happens to obtain attention, the compensation in terms of status, promotions and conference invitations are substantial (van Dalen and Klamer, 2005). These results provide a clear indication of the challenges grad students and early career academics have to overcome. It is as Harzing (2002) and Frey (2005) point out, they either publish or perish. Not even the most successful authors in the AER’s 100 year history have been exempt of possessing so called dry holes (printed articles with no citation) in their repertoire. As a case study in the following chapter we will be analysing a wish list put forward by one of the most prolific authors in the AER, William J. Baumol, on how the future of economics should advance. This will be done by comparing his wish list outlined in his article “Toward a Newer Economics: The Future Lies Ahead!” published in the Economic Journal in 1991 and data we collected from the AER. As May (2006) points out a tenure track position is the final objective of nearly all students who are trying to become an academic. Essentially, this achievement means a lifetime position, intellectual freedom and a kind of prestige within the profession.

There are perhaps other options, as Breneman (1997 stated in McPherson and Schapiro, 1999) indicates that numerous young academics would prefer employment arrangements other than tenure, such as term appointments coupled with the benefits typically given to tenured faculty like travel funds, sabbaticals, and so on. To avoid the pressure for graduate students and early career academics to publish or perish universities should perhaps reassess their hiring procedure and allow young

38 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

researchers to grow into their freshly started career as academics. In many other vocations beginners are given time to learn the “trade”, why is the same courtesy not extended to young academics? Not even in the sports world are rookies expected to MVP nominees in their first few seasons. So why are young academics expected to be on top of their game as soon as they leave university? To minimize these pressures for future academics it is of high importance that institutions prepare their graduate students accordingly. Adams (2002) stresses that the academic environment has changed significantly over the last twenty years but most graduate programs have not. She goes on to list five areas which in particular need attention: teaching, research, academic life, job search and academic options. Furthermore, as only a handful graduating Ph.D. students have extensive teaching experience it emerges that graduate courses are not adequately addressing a vital aspect of faculty work. Especially since Boice (1992 as stated in Adams, 2002) points out that for new faculty members teaching is the responsibility which demands the most instant attention and consumes the most time and energy. Therefore, it is crucial that doctoral graduate students are given the opportunity to get accustomed to lecturing before large classes as most of the graduating doctoral student will not find a position at a research university. Having obtained research experience while being a graduate student can increase one’s confidence to articulate a research question, design studies and write for publication (Austin, 2002), but may not be of assistance when writing exams or curriculum. Buchmueller et al. (1999) found that prior exposure to research as a graduate student is a positive predictor of subsequent research productivity. This is particularly important as time and funding allocated to research at non-research institutions might be limited, but it is research output that will be most valued. One solution to overcome both constraints might be to employ undergraduate students as research assistants, which might even further increase research productivity in following years as one gets familiarized with the skills which are needed to undertake research even earlier. An additional hurdle for young academics is that universities are substituting between the higher paid tenure track positions to the lower paid non-tenure track positions or part-time positions as a greater number of universities experience financial difficulties (Clawson, 2009).

The bottom line is to prepare new faculty for the changing environment at universities, it is the educating institutions themselves, which need to adjust or

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 39 40 modernize their graduate programs. As Adams (2002) points out the structure of academic programs could be adjusted to incorporate several strands of equally respected and supported career pathways of doctoral students. She goes on to state that to implement such strategies would require the coming together of graduates, faculty members in many institutions and professionals in other organisations. In addition to institutional support, it might be a reasonable idea for governments to take away some of the stress of young researchers by providing post-doctoral positions sponsored through scholarships. This would mean that as the position is not remunerated by the university, the institutional expectations for research could be removed and could be open to young academics of any nationality. Such arrangements could be a beneficial experience of young academics and at the same time help young researchers to develop valuable research skills in a foreign country.

40 Chapter 2: Competition in Academia: Evidence from the American Economic Review.

Chapter 2: Competition in Academia: Evidence from the American Economic Review. 41 42

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!”

Statement of Contribution of Co-Authors for Thesis by Published Paper

The authors listed below have certified* that:

1. they meet the criteria for authorship in that they have participated in the conception, execution, or interpretation, of at least that part of the publication in their field of expertise; 2. they take public responsibility for their part of the publication, except for the responsible author who accepts overall responsibility for the publication; 3. there are no other authors of the publication according to these criteria; 4. potential conflicts of interest have been disclosed to (a) granting bodies, (b) the editor or publisher of journals or other publications, and (c) the head of the responsible academic unit, and

5. they agree to the use of the publication in the student’s thesis and its publication on the QUT ePrints database consistent with any limitations set by publisher requirements.

In the case of this chapter:

Comment on William Baumol's “Toward a Newer Economics: The Future Lies Ahead!”, Economics Bulletin, Vol. 31 no.2 pp. 1304-1312.

Contributor Statement of contribution* Marco Piatti

Has equally contributed to all aspects of this paper, including research,

analysis and writing. 15/06/2012

Has equally contributed to all aspects of this paper, including research, Benno Torgler* analysis and writing.

Principal Supervisor Confirmation

I have sighted email or other correspondence from all Co-authors confirming their certifying authorship.

Benno Torgler 15 June 2012 ______Name Signature Date

42 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 43

3.1 INTRODUCTION

Twenty years ago, in January 1991, William Baumol published a very interesting article in the Economic Journal. He pointed out in the abstract that “My title is about as far as I have ever been willing to go in the way of prognostication… Despite the comforting reassurance offered to the authors here that they will certainly be dead long before their forecasts can possibly be tested against reality, I feel obliged to confess that I can offer with any degree of confidence only one prediction – that the future will surprise me” (p. 1). In his conclusion he stresses that “I remain sceptical about our ability to foresee the future - certainly about the future extending a century ahead … Research is by its nature peculiarly resistant to foresight, if only because one of the investigator’s most valued goals is to surprise the audience” (p. 8). He is more inclined to refer to his “wishes” or “hopes”.

Now, we are too impatient to wait and see what happens a century after Baumol’s contribution, so we have taken a look at what has happened with his wishes and hopes 20 years later. We clearly do not want be dead long before such forecasts can possibly be tested against reality, even though we are both still relatively young academics. Encouraged by ABBA’s catchy lyrics of “there’s no hurry any more when all is said and done”, we felt the urge to put Baumol’s wish list to the test so that we may happily and peacefully wait, independently of our success, to see what will happen with the list over the next 80 years.

3.2 METHODOLOGY

Now, how can we analyse Baumol’s wish list? We have decided to take a closer look at the publications in one of the world’s premier economics journals, namely American Economic Review20, publishing generalised and representative state-of-art economic analysis. It is important to choose a general interest journal for our analysis as it is clear that specialized journals such as Journal of Economic

20 It is worth mentioning as a side note that Baumol has been able to publish so far (till the end of 2010) 29 times in American Economic Review (leading articles and articles in the Papers and Proceedings; without the Papers and Proceedings 22 times). Only Joseph Stiglitz has more publications in AER (35 with Papers and Proceedings; 24 without). In other words, there is hardly a better way to test Baumol’s wish list! However, one can criticize that taking AER as representative of the entire economics discipline can be misleading despite the fact that the AER attempts to publish papers to the general well-tooled economist.

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 43 44

Theory are supposed to focus on mathematically-oriented approaches. We turn our attention to two separate time periods, 20 years apart, namely 1984 to 1988 and 2004 to 2008. The period 1984 to 1988 provides an overview as to the state of research before Baumol published his paper in the Economic Journal in 1991. The period 2004 to 2008 offers us an accurate picture what is happening in recent years or in other words 20 years later. In our investigation we include articles, comments or replies, but we exclude the Proceedings issues or invited contributions due to the fact that those articles do not undergo the same review process as non-invited articles or articles in other issues.

3.3 ROLE OF MATHEMATICS

Let us work through Baumol’s wish list in the order in which they are discussed in his paper. The first aspect in Baumol’s wish list is regarding the role of mathematics. He refers to the time where one was “expected to begin with a few words of apology, arguing, or at least asserting, that employment of this tool did not necessarily make the resulting work less ‘realistic’ or less relevant. Even so, it was customary for the algebra to be relegated to an appendix where it would not disturb the sensibilities of the normal reader” (p. 2). He stated in his contribution that he had worked with some determination in order to change this situation (e.g., being in favour of some grounding in mathematics as a standard of a postgraduate curriculum), but in the article he raises the criticism that “things may have gone a bit far in the opposite direction”, pointing out that “few specialised students are allowed to proceed without devoting a very considerable portion of their time to the acquisition of mathematical tools, and they often come away feeling that any piece of writing they produce will automatically be rejected as unworthy if it is not liberally sprinkled with an array of algebraic symbols” (p. 2). He acknowledges that mathematical methods have provided invaluable contributions in many economic fields and that there is “no reason to impede or discourage the work of even the most abstraction-minded and esoteric of mathematical economists” (p. 2). However, the “trouble is that if individuals are not respected for the pursuit of alternative approaches, if only those whose writings are pockmarked by algebraic symbols receive kudos, one can expect a misallocation of resources like that which always results from a distortion of relative prices” (p. 2). In addition, “not only can we

44 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 45 expect more than optimal amount of study and publication to be based on mathematical methods, but we can expect people to be induced to adopt this approach even though they are relatively poorly endowed with the requisite talents”. Graduate programmes, for example, will be burdened with a spate of dissertations that qualify primarily as mathematical (or econometric) exercises whose sole raison d’etre seems to be the opportunity they afford to their authors to display whatever facility they can muster in manipulation of the tools of abstraction. Even the most mathematically-oriented of our colleagues will undoubtedly agree that this is what has already happened” (p. 3).

Having heard Baumol’s argument for alternative approaches to economic analysis, we now use the record of publications in the AER to explore whether the role of mathematics has changed in last 20 years. Table 3.1 uses the number of equations as a proxy for mathematical tools used in papers. We also report a two- sample Wilcoxon rank-sum (Mann-Whitney) test indicating whether the differences between 1984-1988 and 2004-2008 are statistically significant. We observe that in 1984-1988 a paper had on average 11.9 equations in the main text and 1.37 equations in the appendix. In 2004-2008 we observe an increase to 14.64 in the main text and 7.49 in the appendix. In both cases the difference between 1984-1988 and 2004-2008 are statistically significant. However, this may be due to the length of the articles. The average length per article increased from 9.59 in 1984-1988 to 19.38 pages in 2004-2008. Once we correct for the length of articles, the total equations per article (main text and appendix) have demonstrated a very small decrease (although not statistically significant). Thus, it seems that the role of mathematics has hardly changed in the last 20 years, contrary to Baumol’s hopes for some changes to the degree of reliance upon mathematical tools. Now, the question therefore remains what will happen in the next 80 years, or in other words the full century after Baumol compiled his wish list. However, we suspect that the stagnation in this regard over the last 20 years would not surprise Baumol at all, although when referring to two aspects of his wish list he did indicate that “current fashions in economics, like fashion in other fields, will wane after a time. But I look for them only to wane – to give up their undisputed position at the summit of the hierarchy – not to vanish or to remain only as minor vestiges” (p. 4).

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 45 46

Table 3.1: EXPLORING THE ROLE OF MATHEMATICAL TOOLS

Variables Period 1984-1988 Period 2004-2008 z - score N = 585 N = 497 Numbers of equations in 11.90 14.64 4.881 the main text per article Number of equations in 1.37 7.49 10.006 the appendix per article Total equations per article 13.27 22.13 6.850 Total equations / length of 1.27 1.11 -0.301 the article Note: We considered an equation to be a mathematical formula that either is numbered in the article or is clearly separated from the text (placed by itself on a line).

3.4 APPLIED ECONOMETRICS

The next item on Baumol’s list was that university curriculum would put more emphasis on econometrics “stressing its techniques, practice in its use and avoidance of its pitfalls” (p. 4). To get some sort of a proxy of applied econometrics we analyse the number of published tables and figures per article, as tables and figures are usually derived using an empirical approach. Table 3.2 reports statistics of tables and figures for both time periods, and the number of tables in the main text have increased from 1.73 to 3.30 (difference is statistically significant). Similarly, we observe an increase in the use of tables in the appendix, but again, this effect might be due to an increase in the length of the articles. However, when correcting for the length of an article we still observe a statistically significant increase in the use of tables between 1984-1988 and 2004-2008. The same trend can be observed for figures. The number of figures per article increased from 0.98 to 3.07 in the main text and the appendix and the effect remains robust when controlling for the length of the article. This indicates that more applied econometrics or empirical approaches were published in American Economic Review (may also be correlated with use of the tool among economists). Thus, it looks as if we are moving closer to Baumol’s wish list. There seems to be a relative shift towards more applied econometrics compared to mathematical applications.21

21 One should note that our proxy (equation) may also cover econometric techniques. However, one can be sure that it should be a good proxy for terms that Baumol uses in his contribution (e.g., “algebraic symbols” (p. 2), (reliance upon) mathematical tools or “abstract analysis” (p. 4)).

46 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 47

Table 3.2: EXPLORING THE ROLE OF APPLIED ECONOMETRICS/EMPIRICAL ANALYSIS

Period 1984-1988 Period 2004-2008 Variables z - score N = 585 N = 497 Tables in the main text per 1.73 3.30 8.195 article Tables in the appendix per 0.06 0.22 6.405 article Total tables per article 1.79 3.52 8.834 Total tables / length of the 0.16 0.18 3.664 article Figures in main text per 0.96 3.00 13.235 article Figures in appendix per 0.03 0.07 3.101 article Total figures per article 0.98 3.07 16.586 Total figures / length of the 0.10 0.15 9.584 article

3.5 MACROECONOMICS

The next aspect of Baumol’s wish list is more challenging to investigate. He criticizes the short-run orientation of Macroeconomics, stating that there “are at least two major grounds for encouragement of increased attention to the longer run by academic economists. First, of course, is the inherent importance of developments for which substantial periods of time are required… The second reason it is incumbent upon academic economists to devote some attention to the longer run is that there probably is no one else available to do it. Business persons, politicians and civil servants all too often find themselves forced to work from crisis to crisis, and to struggle incessantly to bring today’s and tomorrow’s problem under control” (p. 3). It is difficult to empirically analyse in depth this part of his wish list. We attempt to quantify the changes by constructing Table 3.3, which offers an overview of the topics discussed in these two time periods based on JEL codes22 reported in the published papers. The next step checks some of the topics that may cover a longer macro time horizon (e.g., JEL Code O: Economic Development, Technological Change, and Growth). Looking at JEL codes provides an additional opportunity to

22 Comparing JEL codes over time might be problematic due to a substantial increase of field journals over time. However, several important field journals such as Journal of Mathematical Economics (first published in 1974), Journal of Public Economics (first published in 1972) or Journal of Labor Economics (1983) might have influenced the relative importance of the subject areas published in AER. On the other hand, other journals such as Journal of Economic Growth were first published afterwards (1996).

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 47 48 focus on further and more concrete aspects of Baumol’s wish list for “applied rather than basic economics, starting from the desire for a return to the wealth of nations as a leading focus for the economist’s research… But the past decade has shown that understanding of means that promise to achieve relatively rapid increases in productivity and per capita income are critical not only for the LDCs”. … The fact that others have come from behind and achieved growth rates greater than theirs has also drawn attention to our limited knowledge of means that can effectively stimulate growth” (p. 7). Table 3.3 shows the relative importance of the different JEL code topics in these two time periods. Interestingly, we observe a substantial and statistically significant increase of the relative importance of the area JEL code O over time (from 3.8% to 6.4%). On the other hand, the relative importance of Macroeconomics and Monetary Economics (JEL code E) has decreased over time (from 12.3% to 10.1%). This could be interpreted as an indication that we are observing a relative shift towards more long-run orientation of Macroeconomics.

48 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 49

Table 3.3: SUBJECT-MATTER DISTRIBUTION OF PAPERS OVER TIME

Period 1984-1988 Period 2004-2008 Variables N = 1490 N = 1884 z - core (584 Articles) (497 Articles) (A) – General Economics and Teaching 1.74% 0.42% -3.873 (B) - History of Economic Thought, 2.08% 0.16% -5.196 Methodology, and Heterodox Approaches (C) - Mathematical and Quantitative 3.62% 6.53% 9.644 Methods (D) - Microeconomics 20.07% 25.21% 20.905 (E) - Macroeconomics and Monetary 12.28% 10.08% -14.560 Economics (F) - International Economics 7.38% 5.79% -11.180 (G) - Financial Economics 3.96% 6.16% 9.849 (H) - Public Economics 8.32% 4.25% -10.954 (I) - Health, Education and Welfare 2.42% 5.10% 8.307 (J) - Labor and Demographic Economics 12.42% 8.97% -14.731 (K) - Law and Economics 0.34% 2.49% 5.568 (L) - Industrial Organization 13.02% 8.12% -14.177 (M) - Business Administration and Business 1.95% 1.91% -6.245 Economics; Marketing; Accounting (N) - Economic History 1.01% 2.28% 5.916 (O) - Economic Development, Technological 3.83% 6.37% 10.050 Change, and Growth (P) - Economic Systems 1.54% 1.06% -5.196 (Q) - Agricultural, Natural Resource: 2.75% 1.17% -5.745 Environmental, Ecological Economics (R) - Urban, Rural, and Regional Economics 1.28% 1.86% 5.477 (Y) Miscellaneous Categories No observations (Z) Other Special Topics 0.00% 2.07% - Total 100% 100% Note: JEL codes related to the papers. It seems that authors report more JEL codes over time.

3.6 ECONOMIC HISTORY AND HISTORY OF ECONOMIC THOUGHTS

We can also examine what became of Baumol’s hopes for the “reintroduction of emphasis on the teaching of economic history… It seems to me that many institutional areas lend themselves to study via historical materials, and in some it may not even be possible to carry out effective research without them. Besides, for those whose forte is not a high level of abstraction, history is apt to prove a very good source of ideas and is apt to contribute considerably to general understanding. It should also provide vital practice in the empirical analysis of messy and complicated problems of which economic history has an endless supply” (p. 5). When looking at Table 3.3 we indeed observe an increase in the relative importance

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 49 50 of economic history (JEL code N). The incidence of papers from code N has increased from 1.0% to 2.3%.

On the other hand, Baumol is more sceptical regarding the history of economic ideas: “Yet, though I have taught such a course for many years, I am much more sceptical about any attempt to inveigle more students in that direction. It is my belief that much attention is paid to the work of the past only in fields where there is currently little progress at the frontier… Still, there are undoubtedly matters of greater urgency demanding the student’s very scarce time, and so it is my predisposition to leave the area to those who are attracted to it (or to any other specialised research area) by what Veblen described as ‘idle curiosity’” (p. 5). Interestingly enough, Table 3.3 shows a decrease in the relative importance of History of Economic Thought (see JEL code B). He also emphasised the importance of an “intensive examination of topics such as the economics of education which have largely escaped the attention of mainstream economists. Not only concern with the LDCs requires us to understand more fully just what education contributes to growth, what types of education are critical for the purpose, and what allocation of educational expenditure can be most effective in facilitating growth. In several other countries, notably the United States, the growing proportion in the labour force of groups traditionally associated with inferior education constitutes a threat not only to themselves but also to the remainder of the society” (p. 7). Education is covered under JEL code I, and here we also observe a substantial increase in its relevance. In general, all these results indicate that the changes observed over the last 20 years are not that far away from Baumol’s wishes and hopes.

3.7 MATHEMATICAL AND QUANTITATIVE METHODS

Looking at the JEL code C (Mathematical and Quantitative Methods) in Table 3.3 we are able to get an idea about the importance of theoretically and methodologically driven articles. Baumol declares his hopes “that the future will bring some decrease in the display of technique for its own sake, with models constructed so as to increase what they tell us about the workings of the economy rather than just displaying the properties of some analytical procedure” (p. 6). On the other hand, papers in that area may reflect an aspect that Baumol was keen to see, namely the “greater emphasis on econometrics, stressing its techniques” (p. 4).

50 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 51

Looking at Table 3.3 we observe a substantial increase within the JEL code field C. There is a relative increase of papers from 3.6% to 6.5%.

3.8 JOB OPENINGS FOR ECONOMISTS (JOE)

Next, we also take a look at the number of new jobs listed per year by field of specialization focusing on the Job Openings for Economists (JOE). It helps measuring departments’ preferences and needs for particular field. Looking at the same areas, we observe that there has been a substantial relative increase in demand the area I (Health, Education and Welfare) with an increase from 4.1% in 1991 to 6.3% in 2009. This is consistent with Baumol’s wish list. We also observe a clear trend in Macroeconomics and Monetary Economics (E), namely a decrease over time (from 14.0% to 8.2%), while such a trend is not observable in the field Economic Development and Technological Change (O). Although we observe a decrease in the 1990s with the lowest value in 2000 (3.9%), its demand has again increased in the last 10 years. Thus, it seems that also here we are not that far away from Baumol’s wish list. On the other hand, we do not see substantial differences between Economic History (N) and History of Economic Thought (B) over time. Both report relatively stable values over time. In addition, we observe in line with previous results a clear upward trend for Mathematical and Quantitative methods (from around 10.7 to 14.1%).

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 51 52

Figure 3.1: JOB OPENING FIELDS OF SPECIALIZATION (IN %, 1991-2009) .15 .1 .05 0

1990 1995 2000 2005 2010 JEL Code Classification

B C E I N O

Notes: Values in % relative to all the fields. B: Methodology and History of Economic Thought; C: Mathematical and Quantitative Methods; E: Macroeconomics and Monetary Economics; I: Health, Education and Welfare; N: Economic History; O: Economic Development, Technological Change. Data derived from the annual Proceedings issues (Report of the Director, JOE).

3.9 BEHAVIOURAL ECONOMICS

When discussing subjects for tomorrow’s basic research Baumol argued that the “desire for economic pertinence of our constructs is not tantamount to a wish for unworkable complication. The contrary is apt to be closer to the truth” (p. 6). As an illustration he uses behavioural economics: “Behavioural economists have been disappointed by the quiet reception that has greeted their findings, and the fact that there has been little effort to incorporate those results in the central corpus of mainstream analysis. I believe one reason those results have tended to be ignored is that there has been too little analytic work examining whether and, if so, where, and to what extent, such behavioural anomalies can be expected to affect the behaviour and performance of markets … One should hope that the future will provide a more general theory that investigates more clearly where such behavioural considerations can be expected to make a significant difference for market behaviour and which indicates the nature of the difference it is likely to make” (p. 6). Meanwhile, 20 years later, one can be safe in pointing out that behavioural economics is a central corpus

52 Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 53 of economics. Researchers such as Samuel Bowles, Colin Camerer, Ernst Fehr, Bruno Frey, Herbert Gintis, Uri Gneezy, Charles Holt, John Kagel, Daniel Kahneman, David Laibson, George Loewenstein, John List, Matthew Rabin, Al Roth, Vernon Smith, , and Amos Tversky have each made substantial contributions, not only successfully establishing with rigour, creativity and pertinence this area of research in mainstream economics, but also influencing today’s face of economics. Together they have published 140 AER articles so far (till the end of 2010 including the Proceedings; 98 without Proceedings).

3.10 CONCLUDING REMARKS

In 2007 the movie Bucket List was released staring Jack Nicholson, a corporate billionaire called Edward Cole, and Morgan Freeman, a working class mechanic named Carter Chambers. While sharing a hospital room together they decided to do all the things they have ever wanted to do according to their bucket list before they die. They head off on a road trip with the wish list and in the process, both of them heal each other, become unlikely friends, ultimately finding joy in their lives. William Baumol is currently Professor of Economics and Director of the C.V. Starr Center for Applied Economics at New York University. Twenty years ago at the age of 68 he developed this wish list, and we have tried our best to empirically explore the outcomes. In his concluding remarks he mentioned that if he “could foresee tomorrow’s discovery I would no doubt be tempted to begin work on it at once; what better way to achieve priority! Here I have, consequently, adopted a more-modest stance, describing my wishes rather than my expectations. Yet if there is an element of rationality in the investigator’s choice of topics the two may not prove entirely unrelated. At least so one would hope” (p. 8). Looking back at the results obtained here we indeed observe such a tendency. Thus, we would like to conclude by adding a point to our wish list: We wish William Baumol many further successful years as a leading economist, full of expectations and wishes, or in the spirit of William Shakespeare: We wish him all the joy he can wish.

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 53

Chapter 3: Comment on William Baumol’s “Toward a Newer Economics: The Future Lies Ahead!” 54 55

Chapter 4: Extraordinary Wealth, Globalization, and Corruption.

Statement of Contribution of Co-Authors for Thesis by Published Paper

The authors listed below have certified* that:

1. they meet the criteria for authorship in that they have participated in the conception, execution, or interpretation, of at least that part of the publication in their field of expertise; 2. they take public responsibility for their part of the publication, except for the responsible author who accepts overall responsibility for the publication; 3. there are no other authors of the publication according to these criteria; 4. potential conflicts of interest have been disclosed to (a) granting bodies, (b) the editor or publisher of journals or other publications, and (c) the head of the responsible academic unit, and

5. they agree to the use of the publication in the student’s thesis and its publication on the QUT ePrints database consistent with any limitations set by publisher requirements.

In the case of this chapter:

Extraordinary Wealth, Globalization, and Corruption (2011). School of Economics and Finance, Queensland University of Technology. (2nd Revise and Resubmit in: Review of Income and Wealth).

Contributor Statement of contribution* Marco Piatti

Has equally contributed to all aspects of this paper, including research,

analysis and writing.

15/06/2012

Has equally contributed to all aspects of this paper, including research, Benno Torgler* analysis and writing.

Principal Supervisor Confirmation

I have sighted email or other correspondence from all Co-authors confirming their certifying authorship.

Benno Torgler 15 June 2012 ______Name Signature Date

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 55 56

4.1 INTRODUCTION

The billionaires of the world attract significant attention from both the media and the public, with some billionaires generating celebrity stardom23. The popular press sells thousands of books proposing formulas for accumulating wealth24, capitalizing on individual positional concerns due to relative judgments. Neumayer (2004, p. 793) states that the “accumulation of great fortunes creates uneasiness, envy and concern in many people”. Such concerns emerge as people compare themselves with their environment and care greatly about their relative position, which influences individual choices. Thus, it is not only the absolute level of an individual’s situation (e.g., income), but also the relative position that is important. Frank (1999) notes that research provides “compelling evidence that concern about relative position is a deep-rooted and ineradicable element in human nature” (p. 145). Relative changes may also induce envy in many different environments. Friedman (1962), referred to the following example in the academic world: “The college professor whose colleague wins a sweepstake will envy him but is unlikely to bear him any malice or to feel unjustly treated. Let the colleague receive a trivial raise that makes his salary higher than the professor’s own, and the professor is far more likely to feel aggrieved. After all, the goddess of chance, as of justice, is blind. The salary raise was deliberate judgment of relative merit” (p. 166, cited in McAdams, 1992, p. 103).

Surprisingly, only a limited number of studies have explored empirically the determinants of extraordinary wealth. It seems that Neumayer’s (2004) study was the first one to explore the issue at a global level using cross-sectional analysis. The dependent variable (number of billionaires in each country) was derived from the Forbes list. The results show a positive and statistically significant correlation between GDP per capita and population size. Thus, it is easier to accumulate great wealth in richer and more populous countries. The study also shows that the

23 William Gates III, had been the richest person on earth for more than a decade and is constantly in the media. A search for “Bill Gates” on Google on 9th June 2012 generates 94.5 million hits. If we accept that Google hits are an indication of stardom, it is telling that he returns almost four times as many hits as if we search for “Robert De Niro” who is seen as one of the greatest actors of his time (see http://www.imdb.com). 24 For example, recent releases on Amazon carry titles such as “Think Like a Billionaire, Become a Billionaire”, “Millionaire in 365 Days: The Daily Plan to Get There”, “Be a Real Estate Millionaire: Secret Strategies to Lifetime Wealth Today”, or simple how “How To Become a Millionaire”.

56 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 57 protection of property rights is positively correlated with extraordinary wealth, but in the two reported estimations the coefficient was only statistically significant at the 10% level. Morck, Strangeland, and Yeung (1998) find that economic growth depends on who owns the physical capital and not just on the stock of physical capital itself. They observe a correlation between lower rates of economic growth and entrenched family control of a nation’s capital. On the other hand, the control of capital by entrepreneur billionaires is correlated with faster rates of economic growth. Other studies have taken more a local perspective. Goldman (1998) explores why Russian businessmen first appeared in the Forbes list during the 1990s, even as Russia’s president Boris Yeltsin and its Prime Minister Sergei Kiriyenko were seeking a $20 billion IMF loan. Studies by John J. Siegfried and his co-authors discuss how, where and why fortunes arose in different countries from different industries. They analyse development in Australia (Siegfried and Round, 1994), the US (Blitz and Siegfried, 1992), the UK (Siegfried and Roberts, 1991) and New Zealand (Hazledine and Siegfried, 1998). Kennickell (2003) investigates wealth development in the US. Working with two lists, (one of which is the Forbes data on the 400 wealthiest Americans), the author concludes that wealth experienced relatively strong growth at the very top of the distribution, as did the share of total household wealth held by the listed names in the Forbes’ list. Similarly, Kopczuk and Saez (2004) discover that the Forbes 400 richest list in the US between 1982 and 2002 reveals a strong wealth gain for those wealthy individuals with concentration within the top 100 and in the years of the stock market bubble of the late 1990s. Atkinson (2006) observes from the 2006 Forbes list that wealth among the rich is highly concentrated. Of the 793 billionaires in the world, 42 own a quarter of the total wealth of this group (Gini coefficient: 0.46). He also reports major changes over time in France (concentration of estates in France 1902-94), Germany (wealth tax data covering the former German Reich 1924-35 and West Germany for the period 1953-95), the US (estate data for the period 1916 to 2000), and the UK (concentration of wealth between 1949 to 1960 using investment income data).

In this paper we use an international perspective to explore the relationship between globalization, corruption and extraordinary wealth. We also work with the Forbes list of billionaires but use a panel of 8 years between 1996 and 2003. The results indicate that individuals in more globalized countries are better able to

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 57 58 accumulate extraordinary wealth. In addition, we also find that there is a positive relationship between an increase in corruption and an increase in extraordinary wealth.

4.2 METHODOLOGICAL APPROACH

4.2.1 DATA SETS AND HYPOTHESES Using the Forbes list of billionaires namely the number of billionaires per country (published annually in the Forbes magazine), as a dependent variable, we develop an unbalanced panel of 8 years between 1996 and 2003. The advantage of such a list is that it provides information on people at the very top end of the wealth distribution. Standard sources of data such as surveys (population coverage) fail to capture the very wealthy (Davies and Shorrocks 2000). Forbes magazine has compiled the list for many years, which allows exploration of a relatively large data set. Moreover, Forbes combs through holdings of publicly traded companies, private investments, real estate, and art collections to establish a direct wealth estimate25. On the other hand, the use of sample surveys is subject to non-response and under- reporting, phenomena that might be particularly prevalent for the very rich (Davies and Shorrocks 2000). To determine the affluence of a person, stock prices are calculated using market prices and exchange rates as of market closings on a particular day. The closing day varies from year to year but it is usually scheduled in the early half of February. Atkinson (2006) discusses disadvantages of such a list. A key issue is that validity depends on the extent to which wealth holdings are public knowledge. This information is likely to vary across countries, regions and over time. However, reporters from the Forbes magazine are in close contact with billionaires and the fact that they have been developing the list for many years may indicate that they have invested substantial efforts in obtaining adequate coverage. However, Atkinson (2006) also points out that assets may be more visible than debts and that many of the assets are difficult to value. Similarly, Davies and Shorrocks (2000) stress that assets covered in the estimates are restricted to those that are more easily identified on public record. In addition, it may be that in some cases, family holdings are reported rather than individual holdings (Atkinson 2006).

25 Based on country of citizenship and not residency. For a description of the methodology see http://www.forbes.com/2008/03/05/billionaire-methodology-acknowledgements-billionaires08- cx_lk_0305thanks.html

58 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 59

We posit that the international environment facing a country might be a key factor to understanding extreme wealth accumulation. A country’s capacity to act globally by creating international networks guarantees the flow of information, goods and capital, thereby increasing the possibility set for super-rich people and reducing restrictions on efficient action. More than 100 years ago, Walkin (1907) acknowledged the importance of globalization: “Formerly isolated and outlying communities and countries, from Ceylon to the edge of the one-time ‘great American desert’, have been drawn into the swirl of exchange and suffer or prosper according to the level of prices determined in world markets… The opportunity of the business man in any line to profit by value-increase is multiplied by the increase in the breadth and in the number of exchanges. Recent economic evolution has thus greatly added to his power and importance” (pp. 62-63). Moreover: “Some large American fortunes were made by pioneers in the oriental and tropical trade. John Hancock’s fortune was made in the West Indian trade, which was also the foundation of the fortune of Stephen Girard” (p. 93). Atkinson (2006, p. 12) looked at the Forbes list in 2006 pointing out that “those at the very top are largely self-made. Bill Gates has topped the list for twelve years, and others in the top 25 in 2006 include Paul Allen, Steven Ballmer, Michael Dell and Lawrence Ellison, with Sergey Brin and Larry Page of Google at numbers 26 and 27”. Later he points out that “these forces of technological change and globalization may be expected to have left their mark on the distribution of self-made fortunes” (p. 25).

In line with Dreher (2006) we use Clark’s (2000) definition of globalization as a process of establishing networks of connections among actors in different countries, mediated through a variety of flows including people, information and ideas, capital, and goods (p. 86). Dreher’s (2006) data set is based on 23 variables and provides an overall measure of globalization that covers several dimensions of globalization, namely economic, political, and social globalization (for a detailed description see also Dreher et al., 2008)26. The data we use is based on the KOF Index of Globalization 2006, satisfying Dreher and Gaston’s (2008) emphasis on working with a globalization proxy that covers various aspects. Most studies focus on economic globalization despite having important social and political dimensions.

26 See also http://globalization.kof.ethz.ch/

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 59 60

They also refer to a sort of multiplier effect: “Since globalization encompasses several aspects that taken together have a greater effect than the sum of their constituent parts, it is logical to assess the effects together. Composite indices do exactly this since they provide a single statistic on which comparisons can be based, without the confounding effects of variation at lower levels of aggregation” (p. 518). More globalized environments are correlated with a higher degree of competitiveness and a lower level of protection against competitors from foreign countries, neither of which should hinder the creation of super fortune (Neumayer, 2004). We would therefore predict a positive correlation between an increase in globalization and an increase in extreme wealth accumulation.

Focussing on these elements is in line with studies that explore the relationship between globalization and inequality (e.g., Wade, 2004; Dreher and Gaston, 2008; Nissanke and Thorbecke, 2006; Zhang and Zhang, 2003). O’Rourke (2001) raises the criticism that “public debate on the issue can be frustratingly confused” (p. 1). Looking at within-country and between-country income distribution he notes that theoretical implications are theoretically ambiguous and therefore must be resolved empirically (see pp. 4-5). Wade (2004) states that income inequality comes together with (i) increased poverty, (ii) slower economic growth, (iii) higher unemployment, and (iv) higher crime. In general, Dreher and Gaston (2008) stress that “the proliferation of theories has yielded considerable uncertainty about what are the predicted effects of globalization on inequality in both developed and developing countries” (p. 531). They also point out the lack of empirical studies in that area calling it an “empirical fog” (p. 531). Similarly, Zhang and Zhang (2011) indicate that the literature on the relationship between globalization and inequality has mainly focused on developed countries (in particular the US). Dreher and Gaston’s (2008) study works with the same globalization variable that we use, and emphasize that the “importance of institutional factors highlights the need to have a sufficiently broad measure of globalization when investigating its effects on income inequality” (p. 517). Their work shows that globalization has exacerbated inequality with strong effects for OECD countries and no robust effect for less-developed countries. Zhang and Zhang (2003) investigate the effect of globalization on regional inequality in China, finding that globalization through foreign direct investment (FDI) has exacerbated the income disparity between regional and costal China during 1986–98.

60 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 61

Our analysis of the super-rich might bear a closer relation to inequality studies that deal with social justice. Relying solely on the Gini coefficient has been criticized as the coefficient can decrease even though the ratio of incomes at the extremes worsens: “What is received by the most and least economically privileged parts of a population can be a much better indicator, even though it does not use all of the data available on distribution among the population” (Sutcliffe 2004, p. 26).

In addition, we explore the correlation between corruption and super richness. In a state where corruption is rampant, the allocation of resources is distributed in a discretionary and unequal manner. Long-term relationships with a few firms might be established to exploit a nation’s wealth at the expense of ordinary people (Rose- Ackerman, 1999). Thus, in kleptocracies wealth is often transferred into the hands of a small group of individuals. For example, Levin and Satarov (2000), analyse corruption and institutions in Russia, and raise the criticism that corruption is an integral part of Russia’s economy. They state that the degree of corruption, in monetary terms, exceeds the total expenditures on science, education, health care, culture, and art that are distributed by the state regime. In some industries, criminal groups spend up to 50% of their revenues to bribe officials (p. 115). Goldman (1998) stresses that Russia is a unique case where various oligarchs accumulated their wealth in a short time. A large proportion of the biggest banks are linked to organized crime. For example, former deputy minister of the petroleum industry Vagit Alekperov ended up owning much of the industry he had previously supervised. Thus, Goldman (1998) concludes that the Russian case was based on expropriation of what was formerly state property and not due to the creation of new productive entities. Wade (2004) points out in the presence of higher income inequality, wealthier people in poorer countries are more likely to compare themselves to richer people in richer countries. This comparison may incline the elites to behave in a more corrupt manner, exploiting their own citizens to achieve a similar living standard as wealthier people in western countries. Gupta et al. (2002) find empirical support for the hypothesis that corruption increases inequality. We would therefore predict that a higher level of corruption may lead to more extraordinary wealth accumulation.

To test this hypothesis we utilize two measures of perceived corruption. First, we use the International Country Risk Guide (ICRG) that provides yearly data (see

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 61 62

Knack, 2001) on corruption. The corruption variable assesses the corruption within the political system as rated by experts. Lower scores indicate "high government officials are likely to demand special payments" and that "illegal payments are generally expected throughout lower levels of government" in the form of "bribes connected with import and export licenses, exchange controls, tax assessment, police protection, or loans". As a robustness check we will also use the control of corruption variable developed by Kaufmann et al. (2003) (KKM). The proxy measure is driven by the traditional notion of corruption, defined as “the exercise of public power for private gain” covering a variety of aspects ranging from the frequency of “additional payments to get things done” to the effects on the business environment (p. 8). Experts, households and firms are asked to rate corruption by allocating values that lie between –2.5 and 2.5, with higher scores corresponding to a lower level of corruption.

Tanzi (2002) notes that most of the corruption variables used in the literature measure “perceptions and not objective and quantitative measures of actual corruption” (p. 39). However, Tanzi (2002) also emphasizes: “One good feature is that the various indexes available are highly correlated among themselves” (p. 39). This has been successfully demonstrated by Treisman (2000), who posed an important question: “Why should one take seriously data that are based on perceptions rather than some directly observable measure of corruption?” (p. 410). He answers the question with two reasons. One is the high correlation mentioned beforehand, although he notes that such a high correlation “might indicate not a common perception of reality but a widely shared bias. This can never be completely ruled out. But if the ratings reflect bias, it is a bias that is remarkably widely shared” (pp. 411-412). As a second reason he points that the indexes predict various aspects of countries’ economic performance (e.g., investment and growth, foreign investment) and therefore the “perception of corruption may have as serious consequences for economic development as corruption itself” (p. 412). It affects investors’ decisions and the allocation of foreign aid (see also Treisman 2007). Meanwhile, there are several recent papers that examine the relationship between perceived and experienced corruption (e.g., Razafindrakoto and Roubaud, 2010; Mocan, 2008 and Weber Abramo, 2008). One problem of perceived corruption indices might be that countries which perform well in variables such as economic

62 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 63 growth tend to be given better scores on indicators related to corruption compared to countries that perform poorly. Such a “poor is bad” effect has received considerable attention in literature on corruption indices (Søreide, 2006 p. 7). Treisman (2007, p. 215) also points out that cross-national differences in perceived levels of corruption could be driven by differences in cynicism, the degree of public identification with the government, the perceived injustice of social and economic relations, the frequency of media reports on corruption, government anticorruption campaigns, and politically motivated accusations by opposition politicians.

Moreover, indexes are often not free of problems. Bjørnskov (2006), for example, demonstrates empirically that social capital cannot be treated as a unitary concept. The three components of social capital, namely trust, norms, and networks are three distinct phenomena and influence factors such as governance quality or life satisfaction differently.

A key question is the extent to which subjective indexes are correlated with experience-based proxies. Razafindrakoto and Roubaud (2010) conduct surveys among experts and non-experts to compare the experts’ perceptions of perceived corruption with actual corruption experienced by people in African countries. Their results indicate a small link between perceived corruption and actual corruption, showing that experts overestimate the level of corruption reported27. However, Treisman (2007) criticizes the ‘noise’ of experienced-based measures, as individuals might not frequently experience the extent of corruption happening at the state’s highest levels. Experts may know more about the extent of such corruption. In addition, survey respondents may not honestly answer questions regarding their own experiences with corruption. Reinikka and Svenson (2006) argue in favour of using expenditure tracking and service delivery surveys stressing that “with appropriate survey methods and interview techniques, it is possible to collect quantitative data on corruption at the micro-level” (p. 367).

In line with the study by Neumayer (2004), we control for the economic development (GDP per capita) and the population size of a country. The idea is that a larger population size allows for a larger number of super rich people compared to

27 However, they also state that the results are unlikely to invalidate the relevance of these corruption indices.

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 63 64 smaller population size. In addition, a higher GDP per capita is related to better infrastructure (physical and organizational) and better access to higher social and human capital. Moreover, it has been argued that it might be easier to accumulate greater wealth in an economy where people are wealthier (Neumayer, 2004). We therefore collect the GDP per capita and the population size of a country from the World Development Indicators.

There is a growing literature that describes and explores the “resource curse”, referring to the tendency for nations to fail to transform such an advantage into economic growth. Furthermore, the resource curse can induce violent conflicts, greater inequality and higher poverty, less democracy and institutional quality, and more corruption (see, e.g., Tsui, 2011; Bhattacharyya and Hodler, 2010; Lujala, 2010; Morrison, 2010; Shaxson, 2007; Nissanke and Thorbecke, 2006; Ross, 2001; Barro, 1999). An important aspect relevant to our study is the manner in which natural resources seem to reinforce patronage politics or nepotism, and the incentive to hoard as much of the endowment as possible for private benefits (Shaxson, 2007; Lujala, 2010). We will focus on oil, a resource that is usually not evenly distributed among the population within a country (Morrison 2010). Vicente (2010) explores the relationship in natural experimental setting where oil was discovered in the period 1997-1999 in West Africa. In comparison with a control group, they find support for an increase in corruption in sectors of primary importance to the political elite of the country. Dietz et al. (2007) find empirical evidence that corruption is linked to lower genuine saving rates in resource rich countries. Fjelde (2009) argues that the conversion of public funds into private payoffs can prolong poverty in oil-wealthy states. It allows rulers to target supporters (rent-based clientelism). Bhattacharyya and Hodler (2010) find that for countries with poor democratic intuitions, natural resources such as oil increase corruption while for countries with strong democratic institutions it reduces corruption. Gupta et al. (2002) find that abundant natural resources are linked with high income inequality. We therefore control for the relevance of oil by including a dummy variable for oil production.

4.2.2 SPECIFICATION OF THE TEST EQUATION To test our two hypotheses, we propose the following baseline equation:

64 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 65

NBIit = α + β1 CTRLit +β2 GLOBit +β3 CORRit+ TDt +REGIONi + εit (1)

where i indexes the countries in the sample, and NBIit denotes countries’ billionaires over the period 1996 to 2003. GLOBit is our index for globalization and

CORRit the level of corruption (higher values, lower corruption). The regressions also contain two key control variables, CTRLit, namely GDP per capita, the population size, and a dummy for oil production. To control for time as well as regional 28 invariant factors, we include fixed time, TDt, and fixed regional effects, REGIONi . 29 30 εit denotes the error term . We use several models, namely OLS regressions with time and regional fixed effects, left censored Tobit models due to a large amount of zeros in the data set, and zero inflated negative binomial models (ZINB) since our dependent variable is a count variable. We find comparable results when using a probit model where 1 measures whether a country has at least one billionaire and discuss some of the results without presenting the tables of these estimations31.

4.3 EMPIRICAL RESULTS

Table 4.1 and Table 4.2 present the first results. The first specification in both tables explores the impact of GLOB on NBI, with the coefficient GLOB always statistically significant at the 1% level. An increase of the globalization index by one unit corresponds with an increase in the number of billionaires by more than 3 people. The probit estimation also indicates that an increase in the globalization index by one unit is correlated with an increase in the number of billionaires by 37%. Thus, the effect is not at all negligible. Moreover, these simple specifications explain almost 40% of the variance in NBI.

28 We differentiate between Europe, Latin America, North America, North Africa, Sub Saharan Africa, the Pacific, Asia, the Caribbean and Australia. 29 For an overview of the countries and summary statistics see Appendix Table A2 and Table A3. 30 The relative role played by globalization vis-à-vis corruption is investigated by estimating beta or standardized regression coefficients in the OLS regression. 31 Tables are available upon request.

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 65 66

In the next two specifications we add CORR together with two control variables. First, we use the ICRG data set to measure the lack of corruption (see [2] and [9]). The negative coefficients indicate that a decrease in corruption (an increase in institutional quality) leads to a decrease in extraordinary wealth and the coefficients are statistically significant at the 1% level in both regressions. Specification [2] shows that on average, a one unit decrease in CORR reduces the number of billionaires by 1.2. The calculated marginal effects of the probit model (not reported) indicate that such an increase reduces the probability of generating a billionaire by 7%. The standardized coefficients show that globalization has a stronger influence on achieving extraordinary wealth than corruption.

The aim of the next two specifications ([3] and [10]) is to check the robustness of the relationship between CORR and NBI. We therefore use an alternative proxy for CORR, namely the KKM “control of corruption” variable. It should be noted that higher values are again related to a lower level of corruption. However, the number of observations decreases as we move from 8 to 4 years of country data. The results indicate that the previously observed findings remain robust. Both GLOB and CORR, are statistically significant, and we actually observe larger quantitative effects for CORR (although GLOB has still a larger impact on NBI).

The purpose of the next group of specifications ([4, 5] and [11, 12]) is to explore the impact of a country’s oil supplier status on the number of billionaires. The OLS specifications are the only ones in which the OIL PRODUCTION dummy is not statistically significant. In addition, the (non-reported) probit models return a statistically significant coefficient at the 1% level. The marginal effects indicate that being a country that supplies oil increases the probability of generating a billionaire by 27% and 22% respectively. The quantitative effects of globalization and corruption do not change substantially after controlling for oil production.

In the last two specifications of Table 4.1 and Table 4.2 ([6, 7] and [13, 14]) we exclude transition countries from our analysis.32 Institutional and economic conditions after the collapse of the communism and the reform processes that caused many transition countries to experience disorientation and an institutional vacuum (Gërxhani 2002). This process may have facilitated the accumulation of

32 For a list of transition countries see Table A6.

66 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 67 extraordinary wealth by a small group of individuals. Leiken (1996-1997) stresses that with “obsolete laws, a state incapable of enforcing them, and a climate of moral and social confusion, criminal organizations bred under the old regime have emerged as power brokers and patrons” (p. 62). He also reports that once the Soviet power began to crumble, leading party members set up dummy corporations abroad to transfer funds out of the country. Kaufmann (1997) points out that during the first half of the 1990s, some former Soviet Union countries experienced reform problems, whereby incomplete and poorly designed and implemented reforms generated opportunities for discretionary decisions by the government officials. Such opportunities for insider deals, and transforming public to private monopolies controlled by few shareholders, allowed the élite to accumulate financial resources. Thus, we exclude transition countries to see whether the results were driven by such conditions. Our findings remain relatively robust after excluding transition countries. It is only in the Tobit estimations (reported in Table 4.2) that the level of statistical significance has dropped for GLOB.

We also check whether there is a non-linear relationship between CORR and our dependent variable NBI. The results are mixed depending on the corruption variable used. Working with the KKM data clearly shows that there is linear relationship between CORR and NBI.

Next, we conduct a robustness test using a zero inflated negative binomial model due to the fact that the number of billionaires is a count variable. The results are presented in Table 4.3 using the ICRG corruption variable and in Table A4 using KKM corruption data. First we include all the countries (see specification [15]), before excluding transition countries in specification [16]. From the results, the US33 and Russia could be seen as outliers. Petras (2008) states that among “the newest, youngest and fastest-growing group of billionaires, the Russian oligarchy stands out for its most rapacious beginning” (p. 319). The privatizations overseen by Yeltsin allowed oligarchs to rise to the top: “the future billionaires stripped the Russian state of over a trillion dollars worth of factories, mines, metals, transport, oil, gas, iron, coal and other formerly state-owned resources” (p. 320). He criticizes the development of this situation: “Without exception, the transfers of property were

33 The US has the largest amount of billionaires (see Table A5 in the Appendix).

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 67 68 achieved through gangster tactics – assassinations, massive theft and seizure of state resources, illicit stock manipulation and buyouts” (p. 320).

Thus, we exclude both countries in specification [17], only the US in [18] and only Russia in specification [19]. As can be seen, CORR as well as GLOB remain statistically significant throughout all five estimations. GLOB also remains robust when we apply the KKM corruption proxy, while corruption loses its statistical significance in the ZINB model (see Appendix Table A4).

As a further robustness check we also conducted estimations using a time trend instead of time dummies. The time trend is mostly not statistically significant. This result can also be identified by looking at Table A5 in which we report the total number of billionaires throughout time34. The joint role played by the time dummies can also be investigated using a Wald-test for coefficient restrictions to test for joint significance. The test shows that in most of the cases the null hypothesis is rejected, meaning that time does not play a significant role in determining NBI.

In addition, Table 4.4 and Table 4.5 report our checks on whether the previous findings hold when an index of institutional quality using the same two datasets (ICRG and KKM) is included 35. One should note that there is a high correlation between our two corruption and the two institutional indexes (r=0.68 for ICRG and r=0.92 for KKM). Adding them simultaneously in the specification may lead to a general loss in the precision of the estimates. Nevertheless, both corruption proxies are statistically significant in all specifications. The coefficient for institutional quality is negative and statistically significant for the ICRG data set. In sum, we can conclude that our two hypotheses cannot be rejected and that corruption has a stronger impact on NBI than the two institutional indexes. As a further robustness check we also run regressions with standard errors adjustments where we cluster at the country level. The findings lead to the same conclusions.

34 Only in the ZINB model do we observe a statistically significant negative relationship between time and NBI. 35 The ICRG index consists of an average of the variables “bureaucratic quality”, “rule of law”, “democratic accountability”, “government stability”, “internal conflict” and “military in politics”. The KKM index is an average of “political stability”, “government effectiveness”, “regulatory quality”, and “voice and accountability”.

68 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 69

The results on the control variables indicated (in line with our predictions) that both the population size and the GDP per capita are positively correlated with NBI. We also explore how government interventions or economic freedom affect super wealth by deriving data from Economic Freedom of the World data base from 2000 to 2003 (Gwartney and Lawson, 2006). We use the “size of government” index that covers: general government consumption spending as a percentage of total consumption, transfers and subsidies as a percentage of GDP, government enterprises and investments as a share of total investment, and top marginal tax rate. These components indicate the extent to which countries rely on the political process to allocate resources and goods and services. Such interventions may prevent the generation of super wealth. The results (not reported) indicate a negative correlation between this index and NBI: thus, an increase in economic freedom is positively correlated with the accumulation of extreme wealth. However, the coefficient is hardly statistically significant. Neumayer (2004) finds a similar result when working with the US Heritage Foundation’s Index of Economic Freedom. Moreover, the picture does not change when we focus on alternative proxies such as regulatory restraints that limit the freedom of exchange in credit, labor, and product markets or the legal structure and security of property rights.

4.4 CONCLUDING REMARKS

This paper has studied the effect of globalization and corruption on the generation of extraordinary wealth. Although the media and the popular press are full of discussions on how to become rich, we only find a limited number of academic studies that have explored empirically the determinants of extraordinary wealth. However, besides the literature discussed in the paper, we do find discussions on the phenomenon of superstars. Rosen’s (1981) seminal paper has initiated a lively dialogue regarding stardom and salary structure – stressing that in many professions a relatively small number of people boast prodigious salaries and dominate the field. Since then, the superstar effect has been investigated not only in the economics of sports, but also in the entertainment industry, such as Hollywood economics (De Vany, 2004), cultural economics (Frey, 2000), and academia (Torgler and Piatti, 2011; Azoulay et al., 2010), and more generally in winner-take-all markets, where a

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 69 70 small heterogeneity in performance translates into large reward differences (Frank and Cook, 1995). Atkinson (2006) also points out: “Consideration of the origins of such fortunes suggests that many are made in ‘winner take all’ markets (as is evidenced by the fact that I am writing this paper using Microsoft Word, not WordPerfect which I used ten years ago)” (p. 25). Frank and Cook (1995) state that winner-take-all markets are found in a large number of situations, where markets have a non-linear pay structure, such that many individuals compete for a limited number of substantial prizes at the top. Bebchuk and Grinstein (2005) examine all S&P 500 firms and report that the mean compensation levels of chief executive officers and top-five executives increased between 1993 from $3.7m to $9m in 2003 (146 percent increase)36. The S&P 1500 have paid an aggregated compensation of not less than $350 billion to the top-five executives over the same period. Moreover, Bebchuk and Grinstein (2005) show empirically that the growth of compensations can only be partially explained by factors such as firm size or firm performance. Bebchuk, Fried and Walker (2002) stress that managers have considerable power to shape their own pay arrangements and that managerial power and the desire to camouflage rent extraction can explain the nature of executive compensations. Gabaix and Landier (2008) follow the spirit of Rosen (1981) and employ extreme value theory to study CEO pay increases. They point out that the six-fold increase of US CEO pay between 1980 and 2003 can be fully attributed to the six-fold increase in market capitalization of large companies.

In sum, our results indicate that globalization enhances super-richness. Countries’ capacity to create international networks guaranteeing the freedom to exchange information, goods and capital seems to be a key ingredient in enhancing the accumulation of extraordinary wealth. However, this positive relationship with creation of new productive entities is only one side of the coin. The other side of the coin shows that extraordinary wealth is also generated through corrupt activities. We find that a higher level of corruption is correlated with super-richness. It seems that in corrupt environments, wealth is often transferred into the hands of a small group of individuals. For example, experiences in Russia and Indonesia (under Suharto)

36 The compensation information from the standard ExecuComp database in public US companies provides information of executive’s salary, bonuses, long-term incentive plans, the grant-date value of restricted stock awards and the Black-Scholes value of granted options (p. 284).

70 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 71 have shown that a number of assets in the privatization and expropriation process were transferred to “insiders” of the system that was already in place. As Goldman (1998, p. 15) stresses, these people are not “Andrew Carnegies, Henry Fords, Bill Gates’ or even John D. Rockefellers”.

4.5 TABLES

Table 4.1: DETERMINANTS OF EXTREME WEALTH (NBI)

Explanatory OLS variables [1] [2] [3] [4] [5] [6] [7] 3.465*** 2.640*** 3.074*** 2.653*** 3.170*** 2.463*** 3.083*** GLOB (globalization (8.54) (6.39) (4.00) (6.26) (3.77) (4.70) (3.03) index) 0.221 0.159 0.211 0.159 0.218 0.147 0.210 -1.180*** -1.194*** -1.337*** CORR (lack of (-3.15) (-3.06) (-3.01) corruption) ICRG -0.092 -0.094 -0.100 -1.994** -2.144* -2.353** CORR (control of (-2.06) (1.96) (-2.05) corruption) KKM -0.149 -0.160 -0.172 1.354*** 1.722*** 1.361*** 1.797*** 1.761*** 2.096*** CTRL: log (GDP per capita) (4.88) (3.14) (4.82) (3.00) (4.89) (3.20) 0.133 0.197 0.134 0.205 0.172 0.237 1.07e-08*** 1.03e-08** 1.08e-08*** 1.07e-08** 2.79e-08*** 2.67e-08** CTRL: (3.00) (2.21) (2.95) (2.17) (3.15) (2.28) population size 0.102 0.108 0.103 0.111 0.177 0.185 -0.230 -0.644 -0.787 -1.209 CTRL: (-0.56) (-0.94) (-1.40) (-1.36) oil production -0.006 -0.021 -0.021 -0.038 Regional Fixed Yes Yes Yes Yes Yes Yes Yes Effects Year Fixed Yes Yes Yes Yes Yes Yes Yes Effects Transition Countries Yes Yes Yes Yes Yes No No Included R2 0.384 0.397 0.384 0.397 0.404 0.413 0.420 Prob. > F 0.000 0.000 0.000 0.000 0.000 0.000 0.000 # of observations 976 875 473 875 473 767 413 Notes: Robust standard errors, t-statistics in parentheses and beta in italics. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption.

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 71 72

Table 4.2: DETERMINANTS OF EXTREME WEALTH (NBI)

Explanatory RE Tobit (left censored) variables [8] [9] [10] [11] [12] [13] [14] 19.352*** 8.038*** 10.163*** 8.426*** 9.935*** 4.266 6.624* GLOB (globalization (10.85) (3.05) (2.92) (3.16) (2.81) (1.55) (1.82) index)

-3.379*** -2.949*** -3.840*** CORR (lack of corruption) (-3.15) (-2.71) (-3.27) ICRG

-7.515*** -6.298** -7.567** CORR (control of corruption) (-2.68) (-2.18) (-2.50) KKM

9.805*** 11.692*** 9.305*** 10.831*** 10.599*** 12.037*** CTRL: log (GDP per (5.58) (4.65) (5.24) (4.24) (5.76) (4.68) capita)

3.74e-08*** 3.87e-08*** 3.31e-08*** 3.41e-08*** 6.49e-08*** 6.43e-08*** CTRL: (6.12) (4.85) (5.36) (4.22) (7.44) (5.66) population size

15.030*** 13.014*** 14.210*** 12.246*** CTRL: (4.59) (3.17) (4.39) (3.02) oil production

Regional Fixed Yes Yes Yes Yes Yes Yes Yes Effects Year Fixed Yes Yes Yes Yes Yes Yes Yes Effects Transition Countries Yes Yes Yes Yes Yes No No Included Prob. > chi2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 # of 976 875 473 875 473 767 413 observations Notes: t-statistics in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption.

72 Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 73

Table 4.3: DETERMINANTS OF EXTREME WEALTH (NBI)

Explanatory Zero Inflated Negative Binomial (ZINB) variables [15] [16] [17] [18] [19] 0.841*** 0.535*** 0.769*** 0.845*** 0.766* GLOB (globalization (4.33) (2.85) (3.87) (4.28) (1.90) index)

-0.259*** -0.210** -0.157* -0.264*** -0.161* CORR (lack of (-3.14) (-2.52) (-1.82) (-3.13) (-1.90) corruption) ICRG

0.743*** 0.742*** 0.829*** 0.745*** 0.835*** CTRL: log (GDP per capita) (6.12) (6.30) (6.45) (5.98) (6.52)

6.16e-09*** 6.42e-09*** 4.94e-09*** 6.12e-09*** 5.37e-09*** CTRL: (4.54) (5.35) (3.94) (3.80) (4.56) population size

1.003*** 0.962*** 1.095*** 1.001*** 1.072*** CTRL: oil production (4.79) (5.03) (5.12) (4.68) (5.10)

Regional Fixed Effects Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Transition Countries Yes No Yes Yes No Included USA Excluded No No Yes Yes No RUS Excluded No No Yes No Yes Prob. > chi2 0.000 0.000 0.000 0.000 0.000 # of observations 976 767 859 867 867 Notes: z-statistics in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 73

Table 4.4: DETERMINANTS OF EXTREME WEALTH (NBI) WITH CORRUPTION INDEXES INCLUDED

OLS RE Tobit (left censored) Explanatory variables [20] [21] [22] [23] [24] [25] [26] [27] 2.640*** 3.170*** 2.497*** 3.078*** 8.460*** 9.977*** 4.397 6.640* GLOB (globalization index) (6.23) (3.17) (4.71) (3.00) (3.17) (2.82) (1.60) (1.82) 0.159 0.218 0.149 0.209 -1.055*** -1.141*** -2.049* -2.915** CORR (lack of corruption) ICRG (-2.74) (-2.71) (-1.70) (-2.26) -0.083 -0.085 -0.091* -0.125** -0.605* -0.607* INSTIT ICRG Index (-1.73) (-2.08) (-1.72) (-1.71) -0.039 -0.051 -2.501** -2.527** -10.084** -10.733** CORR (control of corruption), KKM (-2.26) (-2.32) (-2.30) (-2.40) -0.186 -0.185 0.129 0.061 1.137 1.199 INSTIT KKM Index (0.68) (0.28) (1.15) (0.97) 0.040 0.018 1.564*** 1.708*** 2.052*** 2.054*** 10.468*** 10.131*** 11.848*** 11.353*** CTRL: log (GDP per capita) (4.81) (2.73) (4.73 (2.89) (5.49) (3.88) (5.95) (4.28) 0.154 0.195 0.200 0.232 1.11e-08*** 1.06e-08** 2.87e-08*** 2.66e-08** 3.54e-08*** 3.36e-08*** 6.82e-08*** 6.25e-08*** CTRL: population size (2.97) (2.14) (3.19) (2.25) (5.59) (4.15) (7.62) (5.45) 0.107 0.111 0.182 0.184 -0.400 -0.561 -1.056* -1.169 13.760*** 14.267*** 12.842*** 13.380*** CTRL: oil production (-0.89) (-0.77) (-1.68) (-1.22) (4.11) (3.35) (3.86) (3.16) -0.011 -0.019 -0.029 -0.037 Regional Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Transition Country Included Yes Yes No No Yes Yes No No R2 0.397 0.404 0.414 0.420 Prob. > F / Prob. > chi2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 # of observations 875 473 767 413 875 473 767 413 Notes: Robust standard errors (OLS), t-statistics or z-statistics in parentheses, beta in italics. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption.

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 74

Table 4.5: DETERMINANTS OF EXTREME WEALTH (NBI) WITH CORRUPTION INDEXES INCLUDED

Explanatory Zero Inflated Negative Binomial (ZINB) variables [28] [29] [30] [31] GLOB (globalization 0.835*** 1.122*** 0.485*** 0.680** index) (4.30) (3.53) (2.56) (2.22)

CORR (lack of -0.224** -0.151* corruption) ICRG (-2.51) (-1.67)

INSTIT ICRG Index -0.022 -0.018 (-0.96) (-0.80)

CORR (control of -1.069*** -1.035*** corruption), KKM (-2.85) (-2.85)

INSTIT KKM Index 0.249*** 0.272*** (2.59) (2.82)

CTRL: log (GDP per 0.783*** 0.666*** 0.786*** 0.629*** capita) (6.11) (3.36) (6.20) (3.33)

CTRL: population 6.13e-09*** 6.46e-09*** 6.98e-09*** 8.00e-09*** size (4.64) (2.98) (5.65) (4.06)

CTRL: oil 0.974*** 1.136*** 0.900*** 1.053*** production (4.61) (3.74) (4.64) (3.68)

Regional fixed Yes Yes Yes Yes effects Year fixed effects Yes Yes Yes Yes Transition Country Yes Yes No No included Prob. > chi2 0.000 0.000 0.000 0.000 # of observations 875 473 767 413 Notes: z-statistics in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption.

Chapter 4: Extraordinary Wealth, Globalization, and Corruption. 75

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports.

Statement of Contribution of Co-Authors for Thesis by Published Paper

The authors listed below have certified* that:

1. they meet the criteria for authorship in that they have participated in the conception, execution, or interpretation, of at least that part of the publication in their field of expertise; 2. they take public responsibility for their part of the publication, except for the responsible author who accepts overall responsibility for the publication; 3. there are no other authors of the publication according to these criteria; 4. potential conflicts of interest have been disclosed to (a) granting bodies, (b) the editor or publisher of journals or other publications, and (c) the head of the responsible academic unit, and

5. they agree to the use of the publication in the student’s thesis and its publication on the QUT ePrints database consistent with any limitations set by publisher requirements.

In the case of this chapter:

The Red Mist? Red Shirts, Success and Team Sports. September (2011) - Accepted in: Sport in Society.

Contributor Statement of contribution* Marco Piatti

Has equally contributed to all aspects of this paper, including research,

analysis and writing.

Date 15/06/2012

Has equally contributed to all aspects of this paper, including research, David A. Savage* analysis and writing.

Has equally contributed to all aspects of this paper, including research, Benno Torgler* analysis and writing.

Principal Supervisor Confirmation

I have sighted email or other correspondence from all Co-authors confirming their certifying authorship.

Benno Torgler 15 June 2012 ______Name Signature Date

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 77 78

5.1 INTRODUCTION

Baron von Richthofen, Ferrari, Manchester United and the Chicago Bulls or Michael “Air” Jordan in particular have two common elements, a widely recognized level of skill and the colour red. In many urban myths teams wearing red are considered to be more successful (win more) than any other colour. But why is this colour associated with this perception of greater skill and performance? Is it coincidence that some of the world’s greatest sporting teams have worn red? Many euphemisms utilize red in context of aggression, such as seeing red or the red mist that link aggression to the colour. Are the expectations of red being more successful based upon historical evidence or just an instinct that red represents more aggression and a more likely chance at winning?

Colour has a wide ranging and varied role in both nature and society. For example, red, the first visible colour in the light spectrum, it is the colour most often utilized in nature and society as a (warning) signal or indicator of danger (Humphrey, 1976) and a ruddy complexion often signifying anger which often precedes violence or displays of dominance37. In society, red is repeatedly used as a literary device conjuring up a range of seemingly disparate emotions, from love to hate, violence and warfare. Mars the Roman god of war was associated with this colour, such that when a Roman warrior received an accolade for victories in battle, they were painted from head to foot in the colour. Groups or gangs use colours to symbolize membership (e.g., notorious street gangs such as the Crips (blue) and the Bloods (red)). The Bloods, e.g., have been historically identified as the most aggressive and violent street gang (Alonso, 1999; 2004). Red coloration has not only been associated with dominance and aggression in human societies, but also in the animal kingdom. It has been demonstrated that for male Stickleback fish, being shown a red object will always trigger an attack response (Tinbergen, 1952). The effects of aggression and dominance are not only observed among fish, but also higher order animals such as: reptiles, birds and primates (Pryke, 2009). Thus, one can assume

37 Anger has physical effects including raising the heart rate and blood pressure, often giving the face a red or ruddy complexion.

78 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 79 that in aggressive competition colour could affect the outcome of sports contests. To some extent modern societies have ritualized aggression and competition in the forms of sports. Sport is closely linked to the definition of competition in entomology, namely “the active demand by two or more of the same species or members of two or more species at the same trophic level for a common resource or requirement that is actually or potentially limiting” (Miller, 1967). In these ritualized competitions, team colours are used to signalize identity, membership and loyalty. Additionally, when the competition spills over from the field to the stadiums and out into the streets (in the form of riots or violence), once again team colours are used to signal friend from foe.

Colour also plays an important role in the organization and functioning of societies. In India, the various castes are symbolized by different colours. Red symbolizes the Kashatriyas Caste, which is second in the social hierarchy below the Brahmans (white colour) (Fehrman and Fehrman, 2004). The traditional wedding colours in Egypt, Russia, the Orient and the Balkans are of red and yellow hues.

Red has been fundamental in the development of human psychology. The development of human vision is progressive, such that the first colour that is recognisable by human children is red. Experiments have found that the order of colour perception by children as they aged was: red, green, orange, blue and violet (Garbini, 1894). Garbini reached the conclusion that perception and verbal expression follows a parallel path such that the progression of culture is that of colour perception. Ellis (1900) neatly summed up this by saying:

“Red, is the colour that fascinates our attention earliest, that we see and recognise most vividly; it remains the colour that attracts our attention most readily and that gives us the greatest emotional shock. It by no means necessarily follows that it is the most pleasurable colour. As a matter of fact, such evidence shows that very often it is not.”

Fehrman and Fehrman (2004) indicated that the native Maori people have more than a hundred words used to describe what we call red. In many of the primitive tribal groups, there exists a large vocabulary to describe red, black and white but no

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 79 80 or limited words for blue, green and violet (Ellis, 1900). Not only can red change the meaning of something, it can also change the emotive context and even the behavioural patterns or responses of individuals. Elliot and Niesta (2008) found that women wearing red are considered more attractive and sexually desirable to men than women wearing any other colour. Fast-food chains utilize red within the restaurants because of its effect on the human metabolism, which increases a customers’ appetite. Alternatively, fine dining and formal restaurants utilize the colours blue and green to promote a calm and relaxed atmosphere, increasing the likelihood of customers staying longer and spending more money (Singh, 2006). Providing single examples on red without discussing other colours in detail may bias the perception about the importance of red within our society. To get an idea how dominant the colour red is compared to other colours we provide three figures that report the number of hits generated via Google instant search (done December 3 2010), relating different colours with the wording “winning”, “dominance” and “aggression”.

Figure 5.1: COLOUR AND AGGRESSION USING NUMBER OF GOOGLE HITS IN MILLIONS 30 25 20 15 10 5 0 red blue orange yellow green black white

The Google search may provide a “window to the outside world” giving us an approximation how colours are linked with these factors within our society. Figure 5.1 to Figure 5.3 show that the colour red is relatively dominant in all three cases.

80 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 81

However, looking at “aggression” and “dominance” (Figure 5.1 and Figure 5.2) black has an even stronger effect.

Figure 5.2: COLOUR AND DOMINANCE USING NUMBER OF GOOGLE HITS IN MILLIONS 25 20 15 10 5 0 red blue orange yellow green black white

On the other hand, once we focus also on “winning” red is more dominant (see Figure 5.3). It is apparent that colour influences nearly every aspect of our daily life but do we observe a link between wearing red and improved performance?

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 81 82

Figure 5.3: COLOUR AND WINNING USING NUMBER OF GOOGLE HITS IN MILLIONS 400 350 300 250 200 150 100 50 0 red blue orange yellow green black white

In recent years this relatively under explored concept, that is the effect of colour on performance, has undergone a research explosion (see Pierce and Weinland, 1934; Pressy, 1921 and Singer et al., 1965). Some of the earliest investigations into the effect of colour investigated the physical or perceived effect on individuals. For example, it was claimed that bright colours “weigh” more than darker ones (Pierce, 1894). This idea was supported by both Quantz (1895) and Larguier des Bancels (1900) who claimed that red enlarged the apparent size of an object as opposed to a blue one of identical size. Pierce and Weinland’s (1934) investigated if the colour of the lights in an otherwise white room affected the output of workmen hired specifically for the experiment. They found that red did not inspire greater work output, but white, followed by yellow and green, had the highest impact on output levels. Féré’s (1887) experiments found that when patients were shown a red light, there was a measurable increase in pulse, breathing and an increase in muscular strength. Rehm et al. (1987) performed a psychological field experiment on fifth grade German students during physical education classes. The students were divided into two groups; the first group was issued a distinctive bright orange shirt, while the others wore whatever “personal shirts” they already wore (Singer et al., 1965). The number of aggressive acts committed by each of the competing teams was counted during a sporting task (handball) by independent observers. While the study only contained 30 separate match observations, the results showed a strong

82 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 83 variation in observed behaviour between the two teams. The results showed that individuals wearing the distinctive uniform (orange), committed significantly more acts of aggression than the others, however this result was only significant for male students. This, however, was explained as being a group identification (team) effect and not associated with colour. While this study was not specifically investigating the red effect, the identifiable colour variation did have a significant impact on performance and behaviour. Additionally, orange and red are often associated as having a similar effect, this may be a learned social signal that varies between cultures or biologically inherited. Pryke’s (2009) findings demonstrate that for birds, aggressive and dominate behaviour is not learnt from parents, but the red colouration does appear to have an effect on conflict resolution. This is an indication that red dominance may be innate in nature.

A key study on the red effect or more precisely red vs. blue was first performed by Hill and Barton (2005), who examined the urban myth that teams wearing red performed better in sporting contests. To this end they investigated four combat sports played at the Olympics, such as boxing, tae kwon do, Greco-Roman wrestling and freestyle wrestling. Given that competitors were randomly assigned either red or blue, the probability of either colour being successful should have been statistically equivalent. The result showed that across all four sports, red was significantly more likely to win than blue. Given the randomness of the colour allocation, this result provides strong supporting evidence for some performance enhancement effect. The authors stress that in nature, red is used as a signal of aggression or dominance and those wearing red could obtain some “natural” advantage. Furthermore, the results were consistent across competition rounds, where 16 of the 21 rounds were won by red and 19 of the 29 different weight classes were also won by red. The author’s state that given the undoubted role of other factors such as skill and strength this advantage is only likely in relatively balanced competitions. Where skill levels are fairly equal, such as in close competitions, wearing red would tip the balance between winning and losing. One of the shortcomings of this paper is the use of simplistic analysis tools. Utilizing a multivariate analysis with an adequate number of control factors would have provided a better chance of isolate the unique contribution of the colour red, especially in the non-balanced matchups. In addition, Hageman, Strauss and Leißing (2008) have shown that Hill and Barton’s (2005)

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 83 84 result may have been subject to a referee bias. They indicate that referees on average awarded more points to competitors wearing red than to competitors wearing blue.

Hill and Barton’s (2005) findings were further questioned by Rowe et al. (2005) who concluded that if wearing red provided some naturally occurring but unknown benefits, then by examining contests where neither combatant wear red the result should be an even distribution of results. They analyzed another Olympic sport (judo) where combatants wore either blue or white and found that competitors wearing blue was significantly more likely to be victorious over those in white. Rowe et al. concluded that blue was not by nature a dominant or aggressive colour and therefore some other explanation was attributable for the effect. One such explanation was an artefact of visibility, such that by wearing blue the combatant may be better able to blend into the crowd background whereas those wearing white would have an increased visibility. This would lead to the greater number victories for those in blue as opposed to any true colour inspired effect on performance. However, these results were later refuted by Dijkstra and Preenen (2007) in their follow up examination of the Rowe et al. (2005) judo paper. They found the conclusions to be erroneous, due to three endogenous factors including: the seeding system; the repêchage rounds; and the differences in recovery time of athletes. The allocation of judogi in judo competitions is not random as the higher ranked opponents are systematically allocated blue. Additionally, the tournaments are structured such that tournament favourites, or the “seeded players’, would not face off until the later rounds, this ensured that better-ranked players only faced easier opponents in the earlier rounds. This caused a large bias skewed towards a blue advantage and that once these factors are controlled for the blue dominance disappears and found no statistically significant advantage for either blue or white. Furthermore, this result held true for the other 72 judo tournaments examined which included: Olympics (2004), World Championships (2003, 2005), European Championships (2003, 2004 and 2005) and the Super World Cups (2005: Paris and Hamburg).

These investigations into the effect of wearing red on sports performance were limited to individual sporting events (boxing, wrestling etc.), until Sutter and Kocher (2008) expanded the scope of investigation and sought to investigate a team sport for a similar red effect, choosing to examine a single season of Bundesliga results. They

84 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 85 concluded, through binomial analysis, that no red effect was evident within this team sport, such that teams wearing red won 23% as opposed to 21% losing matches. Additionally, they concluded that teams wearing blue performed no better than those wearing red (blue being the second most common colour amongst teams). Sutter and Kocher (2008) conclude that other factors, such as teams support effects may be reducing the impact of red on team sports. However, as the examination was only performed on a single season of data these finding are far from conclusive as could have been possible if a longer series of matches were analysed. Attrill et al.’s (2008) paper did use a much longer sample of matches, albeit utilizing a relatively simplistic analysis of win counts without controlling for a number of other factors. Examination of the last 60 years of the English football league found that teams with a primary colour of red were much more successful over this period than teams wearing any other colour.

More recently, the investigation into the impact of red on performance has been extended to include non-traditional area sports arena. An interesting study done by Ilie et al. (2008) examined the world or virtual competition for any red performance biases in the realm of online first-person-shooters (FPS). Specifically the FPS examined was an extremely popular game, Unreal Tournament 2004 (UT2004), using the “Death-Match” ranking system over a three month period (1,347 observations). Here the player avatars were visually identical except for the team colour (either red or blue). The players are anonymous, such that players can be anywhere in the world and still be on the same team. Teams colour allocation to either red or blue is prior to commencement of a death match and rankings are based upon the number of wins over the period. The results showed that over the three month window, red teams were significantly more likely to win than teams assigned to blue. Red teams won 54.9% out of 1347 matches. The strength of this study is to focus only results from contests involving the top players which may rule out or may reduce the criticism that the differences could be explained by the preferences of better players for the red teams. Observed differences may also be linked to visual interference, such that seeing red may act as a distractor. Ioan et al. (2007) applied a computerized colour-word Stroop test (test for selective attention) to observe that seeing red distracts men. These finding are in line with the experimental results of Elliot et al. (2007) where participants were subjected to computerized tests involving

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 85 86 reactions, perception and colour. The ‘avoidance’ effect was documented across two nationalities, age groups and environmental settings, showing that seeing red significantly impairs performance of individuals. Elliot et al. (2007) stress that it may not be that wearing red inspires greater performance but that the colour distracts the opponent enough to elicit an advantage.

Now, if the mechanism is a psychological one, it is unclear why one team’s red uniforms would distract the members of another team, without simultaneously distracting the other members of the red team themselves. Identification and repeated interactions among teammates may reduce such a distraction, such that learning and adaption behaviours occurring within the team should prevent teammates becoming distracted. It may even have positive externalities, by improving the visibility of reference points (players) within the game.

A major criticism of some of the previously discussed studies (excluding experimental papers), stem from the difficulty of ruling out alternative explanations when applying a descriptive analysis that measures only the raw effects. Only in a randomized setting (random allocation of colours) we can rule out such problems. However, it is challenging to find in the sports environment a randomized setting. The previously discussed studies on the Olympics combat sports are nice examples for it. Furthermore, small data samples could contribute to erroneous estimations of findings, both positive and negative. It makes therefore sense to work with large datasets to better observe any red effect on performance, controlling for as many legitimate factors as possible. A single season or tournament may result in statistical anomalies that will be evened out over a longer time frame.

A criticism of the use of field data in a multiple regression context instead of non-random experimental data is that multiple regressions are not fully able to estimate without noise the effect of colour on performance as it is impossible to measure all the variables that might conceivably affect performance. Allison (1999) nicely points out

“No matter how many variables we include in a regression equation, someone can always come along and say, “Yes, but you neglected to control for variable X and I feel certain that your results would have been different if you had done so.”

86 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 87

In an attempt to reduce the numbers of uncontrolled factors in the regressions environments are needed that closely approximate an experimental setting. Goff and Tollison (1990) indicate that sporting events are such an environment, stating that:

“Sports events take place in a controlled environment, and generate outcomes that come very close to holding “other things equal.” In other words, athletic fields supply real-world laboratories for testing economic theories. The data supplied in these labs have some advantages over the data normally used in economic research (…) the economist can perform controlled experiments similar to those performed by the physical and life scientists. Sports data afford a similar opportunity. Although the laboratory is a playing field, the data generated are very “clean.” Most external influences are regularly controlled by the rules of the game.”

This allows for a large number of the exogenous (external) factors to be controlled when exploring the relationship between red and success. Thus, sport events can be seen as quasi-natural experimental environments, where subjects, in this case athletes, are acting in the natural environment instead of an artificial laboratory environment (natural incentives to perform). It has been shown that experiments performed in an environment where the test subjects are keenly aware that their behaviour is being monitored are prone to change their normal behaviour such that it is difficult to generalize the results (Levitt and List, 2009). Selection effects are also visible when recruiting subjects for (lab) experiments (e.g., “scientific do-gooders” interested in research). In addition, real field events such as the professional sporting arena are numerous and are driven by large financial incentives. Football players compete in a high stake but very controlled environment, where rules and regulations are consistently enforced by the referees.

5.2 DATA AND METHODOLOGY

This paper attempts to analyse the presence of any red effect within a team sport, specifically the Australian National Rugby League (NRL, 2009). The major advance of this paper is both the size and scope of the analysis. As previously discussed, prior papers have either had limited sample sizes and/or used simple analysis generating the raw rather than the partial effects without controlling in a multivariate analysis for additional factors (see Table 5.1). Our investigation extends

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 87 88 over 30 years and includes 5604 matches38 and additionally we have used a more complex multivariate analysis as opposed to a simpler comparative analysis.

38 Out of the 5604 matches we have 2 where official crowd numbers were not available.

88 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 89

Table 5.1: PREVIOUS STUDIES Multivariate Author Year Team Duration N Findings Analysis Home teams wearing Attrill et 1947 – 2003 Yes 56 Seasons n No red enjoy more long al.(2008) EPL Football run success Players in red were Greenlees Visual Analysis perceived to be of et al. No 1 Experiment n No of Footage greater skill than (2008) players in other colours Red teams UT2004 Ilie et al. significantly won more Tournaments Yes 3 Months 1347 No (2008) in a virtual (April-June) environment (p<5%) 2003-2005 Blue not significantly Dijkstra Judo Comps more successful than and (Olympics, No 72 Tournament 501 No white at any Preenen Worlds, Euro competition after (2007) and Super controlling for biases. Competitions) Red is a distracter on participants, Elliot et Laboratory No 6 Experiment 71 No significantly reducing al.a (2007) Experiments reaction and performance times. Males more distracted Ioan et al. Laboratory No 1 Experiment 50 No and affected by red (2007) Experiments (p<1%) Red does not Sutter and 2000/2001 significantly win more Kocher Bundesliga Yes 1 Season 306 No home or away games (2008) Football (p<5%) 2004 Olympics Wearing red win more (boxing, tae No 1 Tournament n No fights, statistically kwon do and Hill and significant (p<5%) wrestling) Barton Teams have a (2005) significantly better 2004 Euro Cup Yes 1 Tournament n No result when wearing red (p<5%) Wearing blue Rowe et al 2004 Olympics No 1 Tournament 602 No significantly wins (2005) (Judo) more fights (p<1%) Males wearing orange School Rehm et were more aggressive handball No 30 Games 30 No al.(1987) than other colours competition (p<1%) Notes: n: no information available. aElliot et al. (2007) find that red has a distracting effect, viewing red induces poorer performances. ** Ioan et al. (2007) findings that red is an interference colour on performance and only significant in males.

The key component under investigation in our analysis is the relationship between teams wearing red and match success. The investigation consists of all NRL matches over a 30-year period, starting from the 1979 season opener until the final game of 2008. This includes all normal season matches and finals series fixtures but excludes forfeited or byes, totalling 5604 individual matches. It is well documented that modern sporting teams sometimes employ secondary or even tertiary strips to be utilized at away matches, especially when the normal colour is thought to be too

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 89 90 similar to that of the home team. However, the introduction of multiple strips to the general competition is relatively new39. Moreover, these away strips are not always used. Prior to 1997 the distinctive yet simple schema used by teams did not require multiple jerseys. To avoid the uncertainty of team’s jersey colour at away games we have only examined the success of the home team in our analysis. The disadvantage of this analysis is the fact that we are not exploring the colour differences between the home and away team as factor of success. As a robustness test we will also run estimations where we explore the colour difference assuming that the away team plays with the home jersey40. The dataset, generated from NRL archives, newspaper and electronic sources, comprises a rich dataset of variables including: scores; home or away status; game venue; referees and crowd sizes. Within the investigation time frame two major structural changes have occurred: firstly points allocated for a try changed from 3 to 4 points beginning 1983 until current; and secondly a rebel breakaway competition41, known as the 'super-league', formed for one season. Changes to the scoring and/or the breakaway “Super League” may have created some unobserved influence on the way in which teams played and approached the game. It has been demonstrated that changes in scoring or point structures can affect the nature of team strategy and play as they may be able to elicit either more aggressive or defensive play depending upon the change (Guedes and Machado, 2002). Similarly, the introduction of the “Super League” may have changed the way in which teams approached matches by rewarding more flamboyant or aggressive teams with a greater following of fans. For these reasons we have included both of these factors to control for these possibilities. To control for the possible effects these changes may have on the analysis, we have created and included a dummy variable for both events (value 0 before the event, 1 after the change). Furthermore, the period under examination spans 30 years, as such season fixed effects have been used in an attempt to allow for unobserved changes over this extended time frame. Traditionally

39 Eastern and Western Sydney were some of the first teams (1983) to introduce an alternative strip in the early part of the season to combat the heat, as both sides were predominately dark coloured (Blue and Black) However, it was not until the mid 1990’s with the massive expansion of teams and colours that alternate strips became commonplace. 40 A deeper analysis would require looking at the video footage of all the 5604 games. 41 In 1997 a 10 team break away competition known as the Super League was formed in the on-going war for control of the sport. After one season the warring parties negotiated the reformation of a single competition.

90 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 91

NRL jerseys consisted of a two part colour system, consisting of a major and minor colour (this changed as more teams joined the league, teams can now have three or more colours). To allow for the multi-coloured system every team has been assigned two colour variables, a major and minor colour, furthermore a third variable has been included to denote the volume of red. Team colours are considered to have a majority of red if it has a volume of red greater than 50%, and a minor red if it contains a volume of less than 50%. For simplicity we have adopted the most generalizable and inclusive definition of red and by not focusing on any one particular shade of red. Therefore all shades of red that fall within the 600-700 nanometre wavelength of the visual light spectrum (see figure C-8 in Fehrman and Fehrman, 2004) have been considered as red regardless of hue, saturation or chromaticity42. In addition to these factors we have included controls for crowd size, referee and stadiums. There is a large literature examining the effect of crowd noise on performance in sporting events, which predominantly shows that large crowds influence referee decision in favour of the home side (for an overview see Greer, 1983; Nevill et al., 2002; Pollard, 1986 and Schwartz and Barsky, 1977). We have used the percentage of crowd size in lieu of the absolute value to control for both the stadium size and the crowd in attendance. Given that the auditory levels of a small crowd in a large stadium is minimal as opposed to a small sold out stadium. Thus, we have also included each individual referee in all games, to control for any possible biases of a referee towards any team or colour. An ordered probit model has been used for the estimation of the ordinal, dependant variable (Game Outcome), where it is assumed

* ' that there is an underlying continuous latent variable yi = xi β + ε i , were i measures

' the team and xi is a vector of explanatory variables describing the game of team i. β is a vector of parameters to be estimated and ε i is the error term, which is assumed to be normally distributed. Therefore yi is determined from the model as follows:

42 For example teams such as Brisbane Broncos and the North Sydney Bears are all deemed to be red. For a full list see Appendix A. As a further robustness measure Appendix B compares the team colours listed in the Pantone PMS colour system with the RGB and the CMYK colour systems.

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 91 92

* 0 if yi < µ1 Gameis Loss  * yii=1 ifµµ12 ≤< y Gameis Draw 2 ifµ ≤ y* GameisWin  2 i

The cut-off or boundary parameter µ j is the delineation between game outcomes, which consists of a discrete range for each. The parameters of the model are estimated using the maximum likelihood method (Wooldridge, 2002). The complete lists of explanatory variables that are used in the model are as follows. Our key independent variable is coded in the following two ways: MAJOR_RED dummy if team has red as primary colour (0=No red, 1=Red) and VOL_RED dummy indicating amount of red in team strip (0=none, 1=some43, 2=All). To isolate for further factors we use dummies for MATCH ROUND, REFEREE, STADIUM and TEAM to control for unobserved effects. We also control for the CROWD_SIZE (% of the capacity of the stadium). Moreover, the SUPER_LEAGUE dummy indicates a super league season and POINT_CHANGE (dummy) the structural change in the points system44. In addition, we present estimations with and without season fixed effects.

5.3 RESULTS

First we explore in Figure 5.4 whether red home teams with the primary colour perform better than home teams without the primary colour red. As can be seen, teams in red indeed perform better than other home teams.

43 The dummy assignation for ‘some’ relates to teams strips that contain some red but less than the 50% required to be the Major team colour. Teams such as the St. George Dragons have major colour assigned as white but with a secondary colour red. While this amount of red does not qualify as MAJOR_RED but it is picked up in this variable. 44 For the start of the 1983 season the ARL made major changes to the structure of the game including a ball handover after the 6th tackle and a change of points awarded for successfully crossing the opposition’s goal line (try). The points awarded were increased from 3 points to 4 points, thus increasing the value of this form of points scoring.

92 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 93

Figure 5.4: HOME TEAMS’ PERFORMANCE (IN %) WITH AND WITHOUT THE PRIMARY COLOUR RED

MAJOR_RED=1 MAJOR_RED=0

60

50

40

30

20

10

0 won draw lost

Notes: The performance between red teams and other teams is statistically significant. Red teams perform better (z-stat.=4.557) using a two-sample Wilcoxon rank-sum (Mann-Whitney) test.

Using the Wilcoxon rank-sum test (Mann-Whitney) indicates the difference is statistically significant. However, this purely descriptive analysis gives information about the raw effects and not the partial one. Table 5.2 therefore presents the regression results. As in the ordered probit estimation, the equation has a non-linear form; only the sign of the coefficient can be directly interpreted and not its size. Calculating the marginal effects is therefore a method to find the quantitative effect a variable has on the success of a team. The marginal effects that we report in Table 5.2 indicate the change in probability of winning the game when the independent variable changes by one unit (evaluated at the means).

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 93 94

Table 5.2: IMPACT OF RED JERSEYS ON MATCH SUCCESS

Ordered Probit (1) (2) (3) (4) (5) (6) (7) (8) 0.119*** 0.118*** 0.099** -0.557*

MAJOR_RED (2.86) (2.84) (2.05) (-1.84)

0.047 0.047 0.039 -0.212

0.047** 0.046** 0.030 -0.278*

VOL_RED (2.27) (2.20) (1.21) (-1.84)

0.019 0.018 0.012 -0.111

-0.038 -0.027 0.339 0.349 0.433 0.433

SUPER_LEAGUE (-0.29) (-0.21) (0.95) (0.98) (1.15) (1.15)

-0.015 -0.011 0.134 0.138 0.170 0.170

-0.173*** -0.169*** -0.371 -0.359 -0.707*** -0.707***

POINT_CHANGE (-3.21) (-3.15) (-1.50) (-1.46) (-2.76) (-2.76)

-0.069 -0.067 -0.147 -0.142 -0.272 -0.272

0.207*** 0.209*** 0.187** 0.185* -0.027 -0.027 CROWD_SIZE (% TOTAL (3.12) (3.15) (1.96) (1.94) (-0.27) (-0.27)

CAPACITY) 0.082 0.083 0.074 0.073 -0.011 -0.011

STADIUM Fixed No No No No Yes Yes Yes Yes Effects REFEREE Fixed No No No No Yes Yes Yes Yes Effects ROUND Fixed No No No No Yes Yes Yes Yes Effects SEASON Fixed No No No No Yes Yes Yes Yes Effects TEAM Fixed No No No No No No Yes Yes Effects N 5604 5604 5602 5602 5602 5602 5602 5602 Prob.>chi2 0.004 0.023 0.000 0.000 0.000 0.000 0.000 0.000 Pseudo R2 0.001 0.001 0.003 0.003 0.041 0.041 0.072 0.072 Note: Dependant variable is GAME_OUTCOME, an ordered dummy for game outcome (Loss = 0, Draw = 1 and Win = 2). Significance levels *, **, *** are 10%, 5% and 1% respectively. Marginal effect=highest score (win). Coefficients are in bold, z-stat in parenthesis and marginal effects in italics.

In specifications (1), (3), (5) and (7) we use MAJOR_RED as proxy for red and in specification (2), (4), (6) and (8) we use our three-point scale variable VOL_RED. We first start in specification (1) and (2) exploring only the relationship between red and performance. The results indicate a positive correlation that is statistically significant at the 1% or 5% level. However, the low R2 indicates the usefulness to control for further factors. Next, we control in specification (3) and (4) for further factors, namely CROWD SIZE, SUPER LEAGUE season and a structural change in the system (POINT CHANGE). The positive effect of red remains robust. In specification (5) and (6) we control for time effects (MATCH ROUND), the

94 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 95

REFEREE and the STADIUM. The red effect is still positive, but the coefficient for VOL_RED is not statistically significant anymore. We also observe a decrease in the marginal effects. Interestingly, once we control for team fixed effects in specification (7) and (8) the relationship becomes negative. Team dummy variables are included as it can be argued that the results are driven by unobserved team characteristics that are correlated with jersey colour and performance. It is possible that of all teams wearing red only a small number of these teams are highly successful giving an overall positive red effect. However, when we control at an individual team level, the success bias disappears and we observe on average that teams wearing red are less successful. Additionally, once we control for fixed effects we observe a significant increase in the R-squared compared to the previous regressions. Moreover, we observe that the marginal effects also increase. Specification (7) demonstrates that red as the primary jersey colour reduces (ceteris paribus) the probability of winning by around 21 percentage points. Furthermore, an increase of the volume of red in the jersey from “some” to “full” reduces the probability of winning by around 11 percentage points (see specification 8). This success bias creates the impression of a red effect which appears greater than the statistics would suggest driving the general perception that red teams are more successful.

So far we have just focused on the jersey colour of the home team. However, one can stress that the performance is driven by the relative colour differences among teams. Unfortunately, we do not have data on the jersey colour of the away team. However, in most of the cases teams use also their home jersey when playing away45. As mentioned above, prior to 1998 teams did not commonly utilize secondary strips, thus, we present in Table 5.3 estimations where we explore the relative colour differences assuming that the away team does not change its home jersey. We therefore report in Table 5.3 two new variables, namely REL_MAJOR_RED and REL_VOL_RED (calculated as red value home team – red value away team).

45 The colours of the jerseys worn by competing teams shall be easily distinguishable and, if, in the opinion of the referee similarity between the jerseys might affect the proper conduct of the game he may, at his discretion, order either team to change jerseys in accordance with the rules governing the competition in which the game is played (The Australian Rugby League Laws of the Game and Notes on the Laws, Official 2010: p. 10, section 4(e))

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 95 96

Table 5.3: IMPACT OF RED DIFFERENCES ON MATCH SUCCESS

Ordered Probit (9) (10) (11) (12) (13) (14) (15) (16) 0.114*** 0.115*** 0.109*** 0.189***

REL_MAJOR_RED (4.12) (4.16) (3.66) (3.83)

0.045 0.046 0.043 0.075

0.047*** 0.047*** 0.045*** 0.082***

REL_VOL_RED (3.42) (3.42) (3.01) (3.32)

0.019 0.019 0.018 0.033

-0.026 -0.023 0.352 0.348 0.448 0.428

SUPER_LEAGUE (-0.20) (-0.18) (0.99) (0.98) (1.19) (1.14)

-0.010 -0.009 0.139 0.138 0.176 0.169

-0.168*** -0.163*** -0.386 -0.367 -0.730*** -0.717***

POINT_CHANGE (-3.12) (-3.03) (-1.56) (-1.49) (-2.85) (-2.80)

-0.067 -0.065 -0.153 -0.145 -0.280 -0.276

0.217*** 0.219*** 0.201** 0.194** -0.003 -0.006 CROWD_SIZE (% TOTAL (3.26) (3.29) (2.11) (2.04) (-0.03) (-0.06)

CAPACITY) 0.086 0.087 0.080 0.077 -0.001 -0.002

STADIUM Fixed No No No No Yes Yes Yes Yes Effects REFEREE Fixed No No No No Yes Yes Yes Yes Effects ROUND Fixed No No No No Yes Yes Yes Yes Effects SEASON Fixed No No No No Yes Yes Yes Yes Effects TEAM Fixed No No No No No No Yes Yes Effects N 5604 5604 5602 5602 5602 5602 5602 5602 Prob.>chi2 0.000 0.001 0.000 0.000 0.000 0.000 0.000 0.000 Pseudo R2 0.002 0.001 0.004 0.003 0.042 0.041 0.074 0.074 Note: Dependant variable is GAME_OUTCOME, an ordered dummy for game outcome (Loss = 0, Draw = 1 and Win = 2). Significance levels *, **, *** are 10%, 5% and 1% respectively. Marginal effect=highest score (win). Coefficients are in bold, z-stat in parenthesis and marginal effects in italics.

The results provide strong support that the colour difference has an impact on performance. The coefficient is statistically significant in all eight estimations and the marginal effects, for example, indicate that an increase in major redness by one unit increases the probability of winning between 4.3 and 7.5 percentage points.

96 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 97

5.4 CONCLUSIONS

In this study, we investigated the validity of the urban myth that wearing of red confers performance benefits that could result in greater team success. If wearing red improves sporting performance then we would observe more wins for these teams than those wearing any other colour. Focusing on the performance of home teams we observe from the descriptive analysis that red teams are more likely to succeed. The multivariate analysis also provides some support for a positive effect. However, once we control in the first group of estimations for team fixed effects we find a negative relationship. This could account for the longevity of the myth that those teams wearing red seem to be more successful. These results may provide some indication that a team effect is in place, where highly successful teams wearing red are driving the overall results. This could explain Attrill et al.’s (2008) results in the English Premier League as Liverpool, Manchester United and Arsenal are historically some of the most successful teams and are all wearing red. However, when we explore the relative difference in the degree of redness between the home and the away team the positive relationship is quite strong, even after controlling for team effects. As previously mentioned, the disadvantage of this approach is that we must assume that the away team uses their home jersey even though in most instances they are indeed wearing home colours. This is most often true as teams are only required to change colours when the referee decides that the two teams are not easily distinguishable (The Australian Rugby League Laws of the Game and Notes on the Laws, 2010: p. 10, section 4(e)). The low R-squared values might be driven by the large micro dataset that we are using and it is worthwhile to mention that low R-squared values in regression equations are not uncommon in comparable large micro datasets. Moreover, it is worth stressing that low R-squared does not mean that regression equations are useless. It shows us the importance to conduct a multivariate analysis to get a good estimate of the ceteris paribus relationship between red and performance. Still we cannot exclude that we have an omitted variable bias. We may have excluded relevant variables providing an underspecified model. However, working with such a large dataset covering a period of 30 years as done in this study is an improvement compared to previous studies on team sports that also worked with such field data, but without keeping other factors fixed. Additionally, we did not take weather into consideration such as sunshine, rain or clouds as this could change

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 97 98 the perception or visibility of how individuals may see colours and therefore reverse the pattern of red being the dominant colour (Little and Hill, 2007). We did not control for colour specific characteristics such as saturation, brightness or chromaticity, which may have an impact on the size of the red effect. Moreover, we only investigated one sport in a single country; therefore, for a more conclusive result further countries / leagues should be investigated. The NFL could be such a league as it has more teams (some of which do wear red uniforms) and more (total) games, but like the NRL has a salary cap (also a draft) producing an even more balanced competition historically than even the NRL. Furthermore, there is a clear gender effect on the perception of wearing red and needs further investigation.

Thus far most of the data is related to male competition and gender specifity may drive some of the results. As Ioan et al. (2007) stress in many species, “red coloration is a powerful signal for quality regulating both intra-sexual (male-male competition) and inter-sexual (female choice) components of sexual competition”. Looking at the 2004 Olympic taekwondo competition they observe that men wearing red athletic uniform won 66.7% of the matches, while women only won 43.5% (difference between red athletic uniforms and other uniforms was not statistically significant).

In general, this literature is still in its infancy and more work is required to find a definitive answer to this research question. Based on the current findings it is difficult to provide a clear policy implication for clubs, teams or business managers who are setting up or creating a new entity or team. It makes sense to focus also on other factors such as financial performance or popularity outcomes. One only needs to look at one of the world’s most recognisable brands, Coke-a-Cola to see that red can clearly have significant financial success. In general, team sports, compared to individual sports are more complex to analyse. Sutter and Kocher (2008) e.g., stress that perhaps

“... the signal of a shirt’s colour is much less salient in team competitions, or team members feel much less intimidated subconsciously by an aggressive colour like red when comforted at the same time by the presence and support of team-mates.”

98 Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 99

However, our main focus was if red in general has an impact on team performance and not how any particular hue may affect performance. Future research could take this into consideration and investigate how a specific hue of red impacts upon people in society, such as the widely recognisable and famous “Ferrari red”. This could provide an interesting aspect of hue change on supporters and success, as the Ferrari Formula One team changed the shade of red used to fit with new sponsors. There are various questions that require a better elaboration. For example, is it that red may be more directly linked to aggressive behaviour, such that the Red Baron chose red to signify his intention to be more aggressive? Do we observe that red coloured teams are more successful at gathering sponsorships making them successful or is it because they are more attractive to fans, which congregate in greater numbers and are therefore more successful because of higher revenues generated from merchandising sales? Causal relationships or selections effects have not been explored intensively so far. For example, does wearing red promote greater performance or does greater sporting performance promote wearing red?

In general, we need more studies that explore how colour changes, within ‘the same team’ affect players or athletes’ performance and motivation. More evidence is required to understand whether individual heterogeneity drives the results at the aggregated team level. It would also be valuable to better isolate the relative importance of single potential factors such as psychological versus perceptual effects.

Chapter 5: The Red Mist? Red Shirts, Success and Team Sports. 99

Chapter 6: Concluding Remarks

6.1 SUMMARY OF FINDINGS

The four essays that make up this thesis have been selected to investigate competition in winner-take-all markets across three different environments such as academia, superrich and sport. Winner-take-all markets by their very nature are inefficient with resource allocation and because of the extremely high and imbalanced reward structures they are overcrowded with hopefuls attempting to be successful. This chapter provides a summary or an overview of the previous chapters.

While most people believe that competition is a good thing because it brings out the full potential of individuals, in extremes it can be disadvantageous for society. The academic environment analysed in Chapter 2 and 3 is such an inefficient occurrence. Obtaining a publication in a top-journal such as the American Economic Review is of fundamental importance for career academics. Without having obtained a top publication it is almost impossible for career advancement, such as obtaining promotions, pay rise or being considered for tenure track system in top tier universities. Our analysis of the American Economic Review demonstrates that the top ten institutions based on Ph.D. affiliation contributed more than 54% of all contributions in the journal in all three periods (i.e. 1950 – 1959, 1984 – 1988 and 2004 – 2008). However, we also found that the concentration at the top is decreasing from 75% in the earliest period to less than 55% in the latest period. This increase in competition between universities to publish in AER is in line with Torgler and Piatti (2011) who pointed out that the number of articles submitted to and the number of papers rejected by AER has risen considerably between 1953 and 2009. In 1953, 22% of the papers submitted were published, whereas in 2009 the number decreased to only 6.4%. Furthermore, our examination also detected that a considerable citation inequality exists with an overall Gini coefficients higher than in most sports competitions of 0.77 over the whole period. This result highlights the fact that the winner-take-all market in academia is a serious problem as only a few superstars, such as Joseph E. Stiglitz or William J. Baumol (see Table 2.13 for more names), generate the vast majority of citation or publication in these top journals. This is a

Chapter 6: Concluding Rewards 101 102 serious resource allocation issue as the vast majority of authors have squandered valuable resources such as time and money that could have been used elsewhere. Additionally, the concentration at the top could have a dangerous effect on the development and heterogeneity of ideas. If the market is so focused on winning the danger is that only work previously published in these journals will be pursued, abandoning independent thought and research in favour of publications. This would leave only a few independent think tanks and consequently this would result in a decline in innovative research.

Chapter 4 analysed how globalisation and corruption influence the accumulation of extraordinary wealth. Even though the general media is oversupplied with stories and suggestions on how to become superrich we only discovered a limited number of academics studies that have investigated empirically the determinants of extreme affluence. Overall, our results show that globalization enhances extraordinary wealth. Countries’ ability to establish international networks guaranteeing the freedom to exchange information, goods and capital appears to be a crucial component in enhancing the accumulation of extraordinary wealth. However, this positive relationship with creation of new productive entities is only one side of the coin. The other side of the coin shows that super-richness is also accumulated through corrupt activities. We observed that a higher level of corruption is correlated with extreme wealth. It seems that in corrupt environments, wealth is often transferred into the hands of a small group of individuals which is in line with Gupta et al. (2002) who find empirical evidence that corruption increases inequality. For example, experiences in Russia and Indonesia (under Suharto) have shown that a number of assets in the privatization and expropriation process were transferred to “insiders” of the system that was already in place.

In many urban myths teams wearing red are more successful than teams wearing any other colour. Hill and Barton (2005) state that if the skill levels are fairly equal, such as in close competitions, wearing red shirts could tip the balance between winning and losing. In the final Chapter of this thesis we utilise a multivariate analysis to examine if this is true for the Australian National Rugby League because they attempt to maintain a fairly even competition by imposing a salary cap for all teams participating in it. Our results are mixed. While we find support for a positive effect in our descriptive analysis, we find a negative effect

102 Chapter 6: Concluding Rewards 103 when we control at the team level. However, when we investigate the relative difference in the degree of redness between home and away teams, we find a quite strong positive effect even after controlling at the team level. The weakness of this method, however, is that we must assume that the away team uses their home jersey even though on most occasions they are actually wearing home colours. This is most often true as teams are only obliged to switch colours when the referee decides that the two teams’ jerseys are too alike (The Australian Rugby League Laws of the Game and Notes on the Laws, 2010: p. 10, section 4(e)). Given that our results are inconclusive more studies are required that examine the effect of colours in sports competition.

6.2 POLICY IMPLICATIONS

The detrimental stress new academics face when they commence a new position at a new university could be mitigated by a more appropriate preparation of postgraduate students before leaving their home institutions. While the academic environment has changed considerably over last two decades the Ph.D. programs have not (Adams, 2002). She calls attention to five points which in particular need improvement such as teaching, research, academic life and academic options, she continues to state that these suggestions seem impossible to implement. However, Preparing Future Faculty programs (PFF)46, have been experimenting with these ideas and recommended practices since 1994 and discovered that they are realistic, not expensive or difficult to realise. The majority of a Ph.D. time is spent researching and working towards their thesis, which results in a minority of graduates having any experience in teaching large classes. This is a major problem given that it is teaching that demands the most attention of new graduate students (Boice, 1992 as stated in Adams, 2002). It would be logical for graduate programs to provide students with the opportunity to obtain this valuable experience before starting their life as new faculty. On the other hand, it is also important that a healthy balance between universities that focus on research and universities that concentrate more on teaching is maintained as otherwise we would run into the dilemma of a decline in innovations

46 Funded by The Pew Charitable Trusts, the National Science Foundation, and The Atlantic Philanthropies

Chapter 6: Concluding Rewards 103 104 as discussed previously. While the current programs meet the learning demands of Ph.D. students, these programs are ill-suited to prepare graduates for life as an academic, couple this with the high demands to publish or perish in academia some immediate attention to Ph.D. policy may be required.

Additionally, as discussed in Chapter 4, globalisation and corruption play a vital part in increasing income inequality within countries. Transition countries, such as Russia, often experience that within a very short period of time some people accumulate extraordinary wealth through illegal means by exploiting a nation’s capital (Rose-Ackerman, 1999). Gupta et al. (2002) found empirical evidence that corruption increases inequality while Zhang and Zhang (2003) discovered that foreign direct investment has exacerbated the income disparity between regional and costal China during 1986–98. It is therefore fundamental that governments distribute profits made through foreign direct investment more equally among citizens of a country. Furthermore, fighting corruption more rigorously, for example, with help from international anti-corruption agencies in developing and transition countries can mitigate the problem of income inequality at the same time. As these countries are often not able to fight corruption alone; a greater involvement of the international community might sometimes be the only way to alleviate such problems.

Finally, in Chapter 5 derived from the present findings it is difficult to give a clear policy implication for clubs, teams or business managers who are setting up or creating a new entity or team. It makes sense to focus also on other factors such as financial performance or popularity outcomes. One only needs to consider one of the world’s most familiar brands, Coke-a-Cola to understand that red can clearly have significant financial success. However, imagine if teams wearing red jerseys were more successful than teams dressed in any other colour, why are not all teams competing in red shirts?

6.3 SHORTCOMINGS

Some of the shortcomings have already been discussed in the different chapters but we merely like to reiterate them here once more. As we only collected information on one specific journal the generalizability of our result in Chapter 2 can be questioned. In addition, our result might suffer from a selection bias as we only

104 Chapter 6: Concluding Rewards 105 observe the successful authors. Hence, we do not know how often an article from one of the top institutions is rejected. Then again, the relative share of submissions is more likely to be higher among top institutions. The question is whether the ratio (submission/acceptance) is different. It could easily be the case that the American Economic Review rejects proportionally more articles from academics of top universities as it receives a larger number of papers but we do not know for certain. Furthermore, our analysis is entirely descriptive rather than causal and retrospective rather than forward looking and will therefore not provide an answer on the probability of getting published. There exists a lively discussion on the shortcomings and advantages of utilising perceived corruption indices as discussed in detail in Chapter 4. One problem of perceived corruption is that countries that perform well in variables such as economic growth tend to obtain higher scores on indicators associated to corruption compared to countries that perform badly. Søreide (2006) points out that the “poor is bad” effect has obtained a large amount of attention in the literature on corruption indices. This is particular important as aid money is often distributed by examining rankings of such corruption indices. Moreover, since we have relied solely on perceived corruption indices this could have clearly biased our results. Finally, in Chapter 5, as we did not have the jersey colour of the away teams for certain, we had to assume that the away teams played in their home jerseys even though in the majority of instances they were actually wearing the home jerseys. Furthermore, for simplicity we did not take the different shades of red into consideration or the weather condition of the games.

6.4 FURTHER RESEARCH

The analysis in Chapter 2 of this thesis offers the possibility for a number of extensions some of which are time consuming but easily to pursue. The easiest of which is to extend the number of top tier economics journals in the analysis since we have only focused on one particular journal namely the American Economic Review. The results by including more journals could be more meaningful and generalizable to the entire profession. It is of fundamental importance that more studies will be undertaken that analyse the skewness of attention towards the superstars within the field of academia because the resulting inefficient allocation of human capital is expensive to society as the knowledge of highly-trained academics is squandered.

Chapter 6: Concluding Rewards 105 106

Furthermore, future studies could be investigating different fields in academia such as medicine, law or psychology to explore the question: do they experience similar problems as the discipline of economics? It would also be interesting to test if density of dry holes is strongly correlated with the citations Gini coefficients over decades. If they are positively correlated, then this would indicate that there are strict research trends. Those who write on not‐so‐hot topics are not cited, and those who write on hot topics get cited extremely. This will then give us a better idea of to what extent super‐stardom is owed to one’s talent and to what extent to being lucky enough to pick the next hot topic.47

The validity of studies that are trying to determine corruption or include corruption variables in their regression depend to a large degree of indices that measure perceived and not actual corruption. There are, however, recent studies that are promising in measuring real corruption. Reinikka and Svenson (2006) prefer the use of expenditure tracking and service delivery surveys stressing that “with appropriate survey methods and interview techniques, it is possible to collect quantitative data on corruption at the micro-level” (p. 367). More field experiments should be undertaken that explore new ways to measure corruption. Furthermore, most studies that have examined extraordinary wealth have strongly relied on the same data provider namely the Forbes List of Billionaires or the 400 Richest Americans supplied by Forbes. As these data are provided by a company and not by a non-government organisation (NGO) like the World Bank, this could be a possible shortcoming of the data. One way to improve research on superrich people would therefore be to have at least another reliable and respected provider, favourably a NGO, of such data.

Similarly, the effect of red shirts in team competitions could be explored in more detail. Whilst we have not distinguished between colour specific characteristics such as saturation, brightness or chromaticity, future research should incorporate it. It would be interesting to check how results would change if games that ended in a draw are ignored. That way, the dependent variable will have only two values where ordinary probit estimation can be used.48 Moreover, we only analysed one sport in a single country; therefore, for a more conclusive result additional countries should be

47 Thanks to one examiner for this suggestion. 48 See previous footnote.

106 Chapter 6: Concluding Rewards 107 considered. Furthermore, there is a clear gender effect on the perception of wearing red and needs further investigation. In general, we need more studies that investigate how colour changes, within ‘the same team’ affect players or athletes’ performance and motivation. Additionally, more evidence is needed to comprehend whether individual heterogeneity drives the results at the aggregated team level. It would also be important to better isolate the relative importance of single potential factors such as psychological versus perceptual effects.

Future research on winner-take-all markets could attempt to broaden its subject pool and, for example, utilize children to run economic experiments. It would be interesting to determine how early the problems of excess entry into such markets exist. Furthermore, we could also examine the fact that recognising famous brand names such as McDonald or Burger King (Hungry Jack’s in Australia) is already very apparent in young children (see Arredondo et al., 2009) and can clearly indicate that product placement and trying to obtain a comparitive advantage over competitors to succeed in the chosen market should perhaps start as early as possible.

Chapter 6: Concluding Rewards 107 108

108 Chapter 6: Concluding Rewards References

References

Adams, K. A. (2002). What Colleges and Universities want in New Faculty. Preparing Future Faculty Occasional Paper (7). Albers, W., Pope, R., Selten, R., & Vogt, B. (2000). Experimental Evidence for Attractions to Chance. German Economic Review, 1(2), 113-130. Alchian, A. A., & Demsetz, H. (1972). Production, Information Costs, and Economic Organization. The American Economic Review, 62(5), 777-795. Allison, P.D. (1999). Multiple Regression: A Primer. Thousand Oaks: Pine Forge Press. Alonso, A. A. (2004). Racialized Identitites and the Formation of Black Gangs in Los Angeles. Urban Geography, 25(7), 658-674. Alonso, A. A. (1999). Territoriality Among African-American Street Gangs in Los Angeles. University of Southern California. Amir, R., & Knauff, M. (2008). Ranking Economics Departments Worldwide on the Basis of PhD Placement. The Review of Economics and Statistics, 90(1), 185- 190. Arredondo, E., Castaneda, D., Elder, J., Slymen, D., & Dozier, D. (2009). Brand Name Logo Recognition of Fast Food and Healthy Food among Children. Journal of Community Health, 34(1), 73-78. Atkinson, A. B. (2006). Concentration among the Rich, Research Paper No. 2006/151, United Nations University. Attrill, M. J., Gresty, K. A., Hill, R. A., & Barton, R. A. (2008). Red Shirt Colour is Associated with Long-Term Team Success in English Football. Journal of Sports Sciences, 26(6), 577-582. Austin, A. E. (2002). Preparing the Next Generation of Faculty: Graduate School as Socialization to the Academic Career. The Journal of Higher Education, 73(1), 94-122. Axarloglou, K., & Theoharakis, V. (2003). Diversity in Economics: An Analysis of Journal Quality Perceptions. Journal of the European Economic Association, 1(6), 1402-1423. Azoulay, P., Graff Zivin, J. S., & Wang, J. (2010). Superstar Extinction. The Quarterly Journal of Economics, 125(2), 549-589. Bairam, E. I. (1994). Institutional Affiliation of Contributors to Top Economic Journals, 1985-1990. Journal of Economic Literature, 32(2), 674-679. Barro, R. J. (1999). “Determinants of Democracy,” Journal of Political Economy, 107(S6), S158–S183.

References 109 110

Baumol, W. J. (1991). Toward a Newer Economics: The Future Lies Ahead! The Economic Journal, 101(404), 1-8. Bebchuk, L., & Grinstein, Y. (2005). The Growth of Executive Pay. Oxford review of economic policy, 21(2), 283-303. Bebchuk, L. A., Fried, J. M., & Walker, D. I. (2002). Managerial Power and Rent Extraction in the Design of Executive Compensation. University of Chicago Law Review, 69, 751-846. Bhattacharyya, S., & Hodler, R. (2010). Natural Resources, Democracy and Corruption. European Economic Review, 54(4), 608-621. Bjørnskov, C. (2006). The Multiple Facets of Social Capital. European Journal of Political Economy, 22(1), 22-40. Blitz, R. C., & Siegfried, J. J. (1992). How did the Wealthiest Americans Get so Rich. Quarterly Review of Economics and Finance, 32(1), 5-26. Bodenhorn, H. (2003). Economic Scholarship at Elite Liberal Arts Colleges: A Citation Analysis with Rankings. The Journal of Economic Education, 34(4), 341-359. Bornmann, L., & Daniel, H. D. (2008). What do Citation Counts Measure? A review of studies on citing behavior. Journal of Documentation, 64(1), 45-80. Buchmueller, T. C., Dominitz, J., & Lee Hansen, W. (1999). Graduate Training and the Early Career Productivity of Ph.D. Economists. Economics of Education Review, 18(1), 65-77. Butler, L. (2003). Modifying Publication Practices in Response to Funding Formulas. Research Evaluation, 12(1), 39-46. Camerer, C., & Lovallo, D. (1999). Overconfidence and Excess Entry: An Experimental Approach. The American Economic Review, 89(1), 306-318. Clark, W. C. (2000) Environmental Globalization, in Governance in a Globalizing World (Eds) J. S. Nye and J. D. Donahue, Brookings Institution Press, Washington, DC, pp. 86–108. Clawson, D. (2009). Tenure and the Future of the University. Science, 324(5931), 1147-1148. Cleary, F. R., & Edwards, D. J. (1960). The Origins of the Contributors to the AER During the Fifties. The American Economic Review, 50(5), 1011-1014. Coats, A. (1969). The American Economic Association's Publications: An Historical Perspective. Journal of Economic Literature, 7(1), 57-68. Coaldrake, P., & Stedman, L. (1999). Academic Work in the Twenty-First Century. Occasional Paper Series, 99, 10. Coman, K. (1911). Some Unsettled Problems of Irrigation. The American Economic Review, 1(1), 1-19.

110 References 111

Costas, R., Van Leeuwen, T.N. & Bordons, M. (2010). Self-Citations at the Meso and Individual Levels: Effects Of Different Calculation Methods. Scientometrics 82, 517-537. Coupé, T. (2003). Revealed Performances: Worldwide Rankings of Economists and Economics Departments, 1990–2000. Journal of the European Economic Association, 1(6), 1309-1345. Davies, J. B., & Shorrocks, A. F. (2000). Chapter 11 The Distribution of Wealth. In B. A. Anthony & B. François (Eds.), Handbook of Income Distribution. Elsevier, pp. 605-675. De Vany, A. (2003). Hollywood Economics : How Extreme Uncertainty Shapes the Film Industry (1 ed.). London: Routledge. Diamond, A. M. (1989). The Core Journals of Economics. Current Contents, 21(1), 4-11. Dietz, S., Neumayer, E., & De Soysa, I. (2007). Corruption, the Resource Curse and Genuine Saving. Environment and Development Economics, 12(1), 33-53. Dijkstra, P. D., & Preenen, P. T. Y. (2008). No Effect of Blue on Winning Contests in Judo. Proceedings of the Royal Society B: Biological Sciences, 275(1639), 1157. Dreher, A., Gaston, N., Martens, P. (2008). Measuring Globalisation: Gauging its Consequences. New York: Springer. Dreher, A., & Gaston, N. (2008). Has Globalization Increased Inequality?. Review of International Economics, 16(3), 516-536 Dreher, A. (2006). Does Globalization Affect Growth? Evidence From a New Index of Globalization. Applied Economics, 38(10), 1091-1110. Dusansky, R., & Vernon, C. J. (1998). Rankings of US Economics Departments. The Journal of Economic Perspectives, 12(1), 157-170. Ehrenberg, R. G., & Bognanno, M. L. (1990). Do Tournaments Have Incentive Effects? Journal of Political Economy, 98(6), 1307-1324. Elliot, A. J., & Niesta, D. (2008). Romantic Red: Red Enhances Men's Attraction to Women. Journal of Personality and Social Psychology, 95(5), 1150-1164. Elliot, A. J., Maier, M. A., Moller, A. C., Friedman, R., & Meinhardt, J. (2007). Color and Psychological Functioning: The Effect of Red on Performance Attainment. Journal of Experimental Psychology: General, 136(1), 154-168. Ellis, H. (1900). The Psychology of Red. Pop. Sci. Mo, 57, 365-375. Engemann, K. M., & Wall, H. J. (2009). A Journal Ranking for the Ambitious Economist. Federal Reserve Bank of St. Louis Review, 91(3), 127-139. Fehrman, K., & Fehrman, C. (2004). Color: The Secret Influence: Upper Saddle River, NJ: Prentice Hall.

References 111 112

Fere, C. (1887). Note sur les Conditions Physiologiques des Emotions. Revue Philosophique de la France et de l'Étranger, 24, 561-581. Fere, C. (1885). Sensation and Movement. Brain, 8(2), 210-229. Ferrari, M., & Chi, M. T. H. (1998). The Nature of Naive Explanations of Natural Selection. International Journal of Science Education, 20(10), 1231-1256. Fjelde, H. (2009). Buying Peace? Oil Wealth, Corruption and Civil War, 1985-1999. Journal of Peace Research, 46(2), 199-218. Fort, R. D. (2003). Sports Economics: Upper Saddle River, NJ: Prentice Hall. Frank, R. H., & Cook, P. J. (1995). The Winner-Take-all Society. New York: Free Press. Frank, R. H. (1999). Higher Education: The Ultimate Winner-Take-all Market? Cornell Higher Education Research Institute (CHERI), 2. Frank, R. H. (1999). Luxury Fever. Why Money Fails to Satisfy in an Era of Excess: New York: The Free Press. Frank, R. H. (1985). Choosing the Right Pond: Human Behaviour and the Quest for Status. Oxford: Oxford University Press. Frank, R. H. (1991). “Positional Externalities.” In Strategy and Choice, ed. R. J. Zeckhauser, Cambridge, Mass.: MIT Press, pp. 25–47. Frank, R. H., & Sunstein, C. R. (2001). Cost-Benefit Analysis and Relative Position. The University of Chicago Law Review, 68(2), 323-374. Frey, B. S. (2000). Arts & Economics: Analysis & Cultural Policy. New York: Springer. Frey, B. (2005). Problems with Publishing: Existing State and Solutions. European Journal of Law and Economics, 19(2), 173-190. Frey, B. (2009). Economists in the PITS? International Review of Economics, 56(4), 335-346. Frey, B. S. (2010). Forthcoming. Withering Academia? Center for Research in Economics, Management and the Arts. Working Paper. Friedman, M. (1962). Capitalism and Freedom. Chicago: University of Chicago Press. Gabaix, X., & Landier, A. (2008). Why has CEO Pay Increased So Much? The Quarterly Journal of Economics, 123(1), 49-100. Garbini, A. (1894). Evoluzione del Senso Cromatico Nella Infanzia. Estratto dall' Archivio per l'Anthropologia e l'Etnologia 24. Geiger, R. L. (1986). To Advance Knowledge: The Growth of American Research Universities, 1900-1940. Cary, NC, USA: Oxford University Press.

112 References 113

Gërxhani, K. (2002). Tax Evasion in Albania: An Institutional Vacuum? Paper presented at the Annual Meeting of the European Public Choice Society, Belgirate. Goff, B., & Tollison, R. (1990). Sportometrics. Texas: Texas A&M University Press. Goldman, M. (1998). Russian Billionaire's Club. International Economy, 12, 10-15. Graves, P. E., Marchand, J. R., & Thompson, R. (1982). Economics Departmental Rankings: Research Incentives, Constraints, and Efficiency. The American Economic Review, 72(5), 1131-1141. Greer, D. L. (1983). Spectator Booing and the Home Advantage: A Study of Social Influence in the Basketball Arena. Social Psychology Quarterly, 46(3), 252-261. Guedes, J. C., & Machado, F. S. (2002). Changing Rewards in Contests: Has the Three-Point Rule Brought More Offense to Soccer? Empirical Economics, 27(4), 607-630. Gupta, S., Davoodi, H., & Alonso-Terme, R. (2002). Does Corruption Affect Income Inequality and Poverty? Economics of Governance, 3(1), 23-45. Gwartney, J., & Lawson, R. & Easterly, W. (2006). Economic Freedom of the World: 2006 Annual Report. Fraser Institute: Vancouver. Hagemann, N., Strauss, B., & Leißing, J. (2008). When the Referee Sees Red …. Psychological Science, 19(8), 769-771. Harzing, A.-W. (2002). Are our Referencing Errors Undermining our Scholarship and Credibility? The case of expatriate failure rates. Journal of Organizational Behavior, 23(1), 127-148. Hazledine, T., & Siegfried, J. (1997). How Did the Wealthiest New Zealanders Get so Rich? New Zealand Economic Papers, 31(1), 35-47. Heck, J. L. (1993). Eight Decades of Contributing Authors and Institutions to the "American Economic Review": A Historical Summary. The Journal of Economic Education, 24(2), 163-170. Hill, R. A., & Barton, R. A. (2005). Psychology: Red Enhances Human Performance in Contests. Nature, 435(7040), 293-293. Hodgson, G. M., & Rothman, H. (1999). The Editors and Authors of Economics Journals: A Case of Institutional Oligopoly? The Economic Journal, 109(453), 165-186. Humphrey, N. (1976). The Colour Currency of Nature. Colour for Architecture, ed. by T. Porter and B. Mikellides. London: Studio-Vista, pp. 95-98. Ilie, A., Ioan, S., Zagrean, L., & Moldovan, M. (2008). Better to Be Red than Blue in Virtual Competition. CyberPsychology & Behavior, 11(3), 375-377. Ioan, S., Sandulache, M., Avramescu, S., Ilie, A., Neacsu, A., Zagrean, L., & Moldovan, M. (2007). Red is a Distractor for Men in Competition. Evolution and Human Behavior, 28(4), 285-293.

References 113 114

Johnson, D. (1997). Getting Noticed in Economics: The Determinants of Academic Citations. American Economist, 41(1), 43-52. Kalaitzidakis, P., Mamuneas, T. P., & Stengos, T. (2003). Rankings of Academic Journals and Institutions in Economics. Journal of the European Economic Association, 1(6), 1346-1366. Kalaitzidakis, P., Mamuneas, T. P., & Stengos, T. (2010). An Updated Ranking of Academic Journals in Economics. Working Paper Series. Kaufmann, D. (1997) Corruption: The Facts. Foreign Policy, 107, 114-131. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2003). Governance Matters: Governance Indicators for 1996-2002 (Vol. 3106): World Bank. Kennickell A. (2006). A Rolling Tide: Changes in the Distribution of Wealth in the US: 1989–2001, In International Perspectives on Household Wealth, ed. E Wolff, Cheltenham: Edward Elgar Publishing, pp. 19-88. Knack, S. (2001). Aid Dependence and the Quality of Governance: Cross-Country Empirical Tests. Southern Economic Journal, 68(2), 310-329. Kocher, M. G., & Sutter, M. (2001). The Institutional Concentration of Authors in Top Journals of Economics during the Last Two Decades. The Economic Journal, 111(472), F405-F421. Kodrzycki, Y. K., & Yu, P. (2006). New approaches to ranking economics journals. The BE Journal of Economic Analysis & Policy, 5(1), 24. Kopczuk, W. & Saez, E. (2004). Top Wealth Shares in the United States, 1916-2000: Evidence from Estate Tax Returns. National Tax Journal, 57, 445-487. Kotiaho, J. S. (1999). Papers Vanish in Mis-Citation Black Hole. Nature, 398(6722), 19-19. Kotiaho, J. S., Tomkins, J. L., & Simmons, L. W. (1999). Unfamiliar Citations Breed Mistakes. Nature, 400(6742), 307-307. Laband, D. N. (1990). Is There Value-Added from the Review Process in Economics?: Preliminary Evidence from Authors. The Quarterly Journal of Economics, 105(2), 341-352. Laband, D. N., & Tollison, R. D. (2003). Dry Holes in Economic Research. Kyklos, 56(2), 161-173. Larguier des Bancels, J. (1900). XIII. De L'estimation des Surfaces Colorées. L'année psychologique, 7(1), 278-295. Leiken, R. S. (1996). Controlling the Global Corruption Epidemic. Foreign Policy, 105, 55-73. Levin, M., & Satarov, G. (2000). Corruption and Institutions in Russia. European Journal of Political Economy, 16(1), 113-132. Levitt, S. D., & List, J. A. (2009). Field Experiments in Economics: The Past, the Present, and the Future. European Economic Review, 53(1), 1-18.

114 References 115

Little, A., & Hill, R. (2007). Attribution to Red Suggests Special Role in Dominance Signalling. Journal of Evolutionary Psychology, 5(1), 161-168. Lujala, P. (2010). The Spoils of Nature: Armed Civil Conflict and Rebel Access to Natural Resources. Journal of Peace Research, 47(1), 15-28. May, M. (2006). Faculty Positions: A Tale of Two Systems - Tenure v. Adjunct. [10.1126/science.opms.r0600003]. Science. Mayer, T. (2004). Dry Holes in Economic Research: Comment. Kyklos, 57(4), 621- 625. McAdams, R. H. (1992). Relative Preferences. The Yale Law Journal, 102(1), 1-104. McPherson, M. S., & Morton Owen, S. (1999). Tenure Issues in Higher Education. The Journal of Economic Perspectives, 13(1), 85-98. Medoff, M. H. (2006). The Efficiency of Self-Citations in Economics. Scientometrics, 69(1), 69-84. Miller, R. S. (1967). Pattern and Process in Competition. Advances in Ecological Research, 4, 1-74. Mocan, N. (2008). What Determines Corruption? International Evidence from Microdata. Economic Inquiry, 46(4), 493-510. Morck, R. K., Stangeland, D. A., & Yeung, B. (2000). "III. Economic Effects of Concentrated Corporate Ownership: 11. Inherited Wealth, Corporate Control, and Economic Growth The Canadian Disease?," NBER Chapters, in: Concentrated Corporate Ownership, pages 319-372 National Bureau of Economic Research, Inc. Morrison, K. M. (2010). Oil, Conflict, and Stability, Mimeo, Department of Government, Cornell University. Neumayer, E. (2004). The Super-Rich in Global Perspective: A Quantitative Analysis of the Forbes List of Billionaires. Applied Economics Letters, 11(13), 793-796. Nevill, A. M., Balmer, N. J., & Williams, A. M. (2002). The Influence of Crowd Noise and Experience Upon Refereeing Decisions in Football. Psychology of Sport and Exercise, 3(4), 261-272. Nissanke, M., & Thorbecke, E. (2006). Channels and Policy Debate in the Globalization–Inequality–Poverty Nexus. World Development, 34(8), 1338- 1360. NRL. (2009) NRL - The official site of the National Rugby League. from http://www.nrl.com/. O'Rourke, K. H. (2001). Globalization and Inequality: Historical Trends: National Bureau of Economic Research. Working Paper 8339. Osterloh, M., & Frey, B. S. (2010). Academic Rankings and Research Governance. SSRN eLibrary.

References 115 116

Ozturk, M. A. D. C. (2008). Affective Benefits from Academic Competitions for Middle School Gifted Students. Gifted Child Today, 31(2), 48-53. Petras, J. (2008). Global Ruling Class: Billionaires and How They “Make it”. Journal of Contemporary Asia, 38(2), 319-329. Pierce, A. H. (1894, 1896). Aesthetics of Single Forms. Psychol. Rev. I, 1894: 483- 495, III, 1896: 270-282. Pierce, D. H., & Weinland, J. D. (1934). The Effect of Color on Workmen. Personnel Journal, 13, 34-38. Pollard, R. (1986). Home Advantage in Soccer: A Retrospective Analysis. Journal of Sports Sciences, 4(3), 237-248. Pressey, S. L. (1921). The Influence of Color Upon Mental and Motor Efficiency. The American Journal of Psychology, 32(3), 326-356. Pryke, S. R. (2009). Is Red an Innate or Learned Signal of Aggression and Intimidation? Animal Behaviour, 78(2), 393-398. Quantz, J. O. (1895). The Influence of the Color of Surfaces on Our Estimation of Their Magnitude. The American Journal of Psychology, 7(1), 26-41. Razafindrakoto, M., & Roubaud, F. (2010). Are International Databases on Corruption Reliable? A Comparison of Expert Opinion Surveys and Household Surveys in Sub-Saharan Africa. World Development, 38(8), 1057-1069. Rehm, J., Steinleitner, M., & Lilli, W. (1987). Wearing Uniforms and Aggression–A Field Experiment. European Journal of Social Psychology, 17(3), 357-360. Reinikka, R., & Svensson, J. (2006). Using Micro-Surveys to Measure and Explain Corruption. World Development, 34(2), 359-370. Rose-Ackerman, S. (1999). Corruption and Government: Causes, Consequences, and Reform. Cambridge: Cambridge University Press. Rosen, S. (1981). The Economics of Superstars. The American Economic Review, 71(5), 845-858. Rosen, S., & Sanderson, A. (2001). Labour Markets in Professional Sports. The Economic Journal, 111(469), 47-68. Rowe, C., Harris, J. M., & Roberts, S. C. (2005). Sporting Contests: Seeing red? Putting Sportswear in Context. Nature, 437(7063), E10. Schwartz, B., & Barsky, S. F. (1977). The Home Advantage. Social Forces, 55(3), 641-661. Scott, L. C., & Mitias, P. M. (1996). Trends in Rankings of Economics Departments in the U.S.: An Update. Economic Inquiry, 34(2), 378-400. Scully, G. W. (2004). Player Salary Share and the Distribution of Player Earnings. Managerial and Decision Economics, 25(2), 77-86. Seaman, B. A. (2003). Cultural and Sport Economics: Conceptual Twins? Journal of Cultural Economics, 27(2), 81-126.

116 References 117

Shaxson, N. (2007). Oil, Corruption and the Resource Curse. International Affairs, 83(6), 1123-1140. Siegfried, J. J., & Roberts, A. (1991). How did the Wealthiest Britons Get so Rich? Review of Industrial Organization, 6(1), 19-32. Siegfried, J. J., & Round, D. K. (1994). How did the Wealthiest Australians Get so Rich? Review of Income and Wealth, 40(2), 191-204. Siegfried, J. J. (1994). Trends in Institutional Affiliation of Authors Who Publish in the Three Leading General Interest Economics Journals. The Quarterly Review of Economics and Finance, 34(4), 375-386. Singer, J. E., Brush, C. A., & Lublin, S. C. (1965). Some Aspects of Deindividuation: Identification and Conformity. Journal of Experimental Social Psychology, 1(4), 356-378. Singh, S. (2006). Impact of Color on Marketing. Management Decision, 44(6), 783- 789. Søreide, T. (2006). Is it Wrong to Rank? A Critical Assessment of Corruption Indices. Chr. Michelsen Institute (CMI) Working Paper. Stigler, G. J., Stigler, S. M., & Friedland, C. (1995). The Journals of Economics. Journal of Political Economy, 103(2), 331-359. Sutcliffe, B. (2004). World Inequality and Globalization. Oxford review of economic policy, 20(1), 15-37. Sutter, M., & Kocher, M. G. (2008). Shirt Color and Team Performance in Football. In Myths and Facts about Football: The Economics and Psychology of the World's Greatest Sport, edited by P. Andersson, P. Ayton and C. Schmidt. Cambridge: Cambridge Scholars Publishing. Tanzi,V. (2002). Corruption Around the World: Causes, Consequences, Scope, and Cures, in: T. Abed and S. Gupta (eds.), Governance, Corruption & Economic Performance. Washington, D.C.: International Monetary Fund, p. 19-58. The Australian Rugby League Laws of the Game and Notes on the Laws, (2010): p. 10, section 4(e). Tinbergen, N. (1952). "Derived" Activities; Their Causation, Biological Significance, Origin, and Emancipation During Evolution. The Quarterly Review of Biology, 27(1), 1-32. Torgler, B. & Piatti, M. (2011). A Century of American Economic Review. UC Berkeley: Berkeley Program in Law and Economics, Working Paper Series. Treisman, D. (2000). The Causes of Corruption: A Cross-National Study. Journal of Public Economics, 76(3), 399-457. Treisman, D. (2007). What Have We Learned About the Causes of Corruption from Ten Years of Cross-National Empirical Research? Annual Review of Political Science, 10(1), 211-244.

References 117 118

Tsui, K. K. (2011). More Oil, Less Democracy: Evidence from Worldwide Crude Oil Discoveries. The Economic Journal, 121(551), 89-115. Van Dalen, H. P., & Klamer, A. (2005). Is Science A Case of Wasteful Competition? Kyklos, 58(3), 395-414. Vaughn, G. F. (2004). Katharine Coman: America's First Woman Institutional Economist and a Champion of Education for Citizenship. Journal of Economic Issues, 38(4), 989-1002. Wade, R. H. (2004). Is Globalization Reducing Poverty and Inequality? World Development, 32(4), 567-589. Wall, H. J. (2009). Journal Rankings in Economics: Handle with Care. SSRN eLibrary. doi: 10.2139/ssrn.1374569 Watkins, G. P. (1907). The Growth of Large Fortunes. Publications of the American Economic Association, 8(4), 1-170. Weber Abramo, C. (2008). How Much Do Perceptions of Corruption Really Tell Us? Economics: The Open-Access, Open-Assessment E-Journal, 2, 2008-3. Wilson, E. O. (2000). Sociobiology. The New Synthesis. Cambridge: Belknap Press of Harvard University Press; Twenty-Fifth Anniversary Edition. Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge, Massachusetts. London, England: The MIT Press. Wu, S. (2006). Recent Publishing Trends at the AER, JPE and QJE. Applied Economics Letters, 14(1), 59-63. Zhang, X., & Zhang, K. H. (2003). How Does Globalisation Affect Regional Inequality within A Developing Country? Evidence from China. Journal of Development Studies, 39(4), 47-67.

118 References

Appendices

APPENDIX A: CHAPTER 2

Table A1: A SUMMARY OF INSTITUTIONAL RANKINGS Appearance as a Top Appearance as a University 10 University Top 20 University Massachusetts Institute of Technology 29 29 Harvard University 27 28 University of Chicago 26 27 Stanford University 25 29 Princeton University 23 28 University of California, Berkeley 21 26 University of Pennsylvania 20 25 Yale University 20 23 Northwestern University 15 24 Columbia University 13 22 University of California, Los Angeles 9 24 University of Michigan 9 18 University of Wisconsin 9 22 New York University 5 20 Hebrew University 5 8 Carnegie Mellon University 4 10 University of Washington 4 8 Brown University 3 9 University of Western Ontario 3 10 Cornell University 3 13 London School of Economics 3 13 University of Minnesota 0 12 Rochester University 0 15 Notes: Data from Tom Coupé (2003), table 2 and table 4 (covering two time periods, 1978– 1982 and 1996–2001; four different rankings); two tables from Philip E. Graves, James R. Marchand, and Randal Thompson (1982), table 1 and table 2; table 3 from Pantelis Kalaitzidakis, Theofanis P. Mamuneas, and Thanasis Stengos (2003); five from Erkin Bairam (1994), table 1 (AER 1985–90), table 2 (Econometrica 1985-90), table 3 (Economic Journal 1985–90), table 4 (JPE 1985–90) and table 5 (QJE 1985–90); table 1 from Amir and Knauff (2008); three tables from Stephen Wu (2007), table 2 (AER), table 3 (JPE), and table 4 (QJE) between for the 2000–2003 period; and 12 by John J. Siegfried (1994), table 1 (AER, by decade between 1950 and 1989), table 2 (JPE, by decade between 1950 and 1989) and table 3 (QJE, by decade between 1950 and 1989) and table 2 by Jean Louis Heck (1993).

Appendices 119 120

APPENDIX B: CHAPTER 4

Table A2: COUNTRIES (122 COUNTRIES, BASED ON SPECIFICATION [1]) Albania Germany Nigeria Algeria Ghana Norway Argentina Greece Oman Australia Guatemala Pakistan Austria Guinea-Bissau Panama Bahamas Guyana Papua New Guinea Bahrain Haiti Paraguay Bangladesh Honduras Peru Barbados Hong Kong Philippines Belgium Hungary Poland Belize Iceland Portugal Benin India Romania Bolivia Indonesia Russian Federation Botswana Iran Rwanda Brazil Ireland Senegal Bulgaria Israel Sierra Leone Burundi Italy Singapore Cameroon Jamaica Slovakia Canada Japan Slovenia Central African Republic Jordan South Africa Chad Kenya Spain Chile South Korea Sri Lanka China Kuwait Sweden Colombia Latvia Switzerland Congo, Republic of Lithuania SYRIA Congo, the Democratic Republic of Luxembourg Tanzania, United Republic of Costa Rica Madagascar Thailand Cote D'Ivoire Malawi Togo Croatia Malaysia Trinidad and Tobago Cyprus Mali Tunisia Czech Republic Malta Turkey Denmark Mauritius Uganda Dominican Republic Mexico Ukraine Ecuador Morocco United Arab Emirates Egypt Myanmar United Kingdom El Salvador Namibia United States Estonia Nepal Uruguay Fiji Netherlands Venezuela Finland New Zealand Zambia France Nicaragua Zimbabwe Gabon Niger

120 Appendices 121

Table A3: DESCRIPTIVE STATISTICS

Variables Obs. Mean Std. Dev. Min Max NBI 2558 1.167 9.638 0 269 GLOB 976 2.323 0.982 0.722 5.420 CORR (ICRG) 1098 2.944 1.263 0 6 CORR (KKM) 719 -1.48e-10 0.997909 -2.050 2.583 CTRL: log(GDP per capita) 1447 7.530 1.575 4.085 10.751 CTRL: population size 1575 3.04e+07 1.18e+08 0.769 1.29e+09 Billionaire in the Country 2558 0.131 0.338 0 1 Oil Production Country* 1672 0.689 0.629 0 2 Transition Country 2558 0.088 0.283 0 1 World regions: Asia 1616 0.267 0.443 0 1 Australia 1616 0.010 0.099 0 1 Caribbean 1616 0.079 0.270 0 1 Europe 1616 0.208 0.406 0 1 Latin America 1616 0.109 0.312 0 1 North America 1616 0.020 0.139 0 1 North Africa 1616 0.030 0.170 0 1 Pacific 1616 0.035 0.183 0 1 Sub Saharan Africa 1616 0.243 0.429 0 1 Notes: *Data are from http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=5&pid=53&aid=1. See Table A5 for an overview of transition countries.

Appendices 121 122

Table A4: DETERMINANTS OF EXTREME WEALTH (NBI)

Explanatory Zero Inflated Negative Binomial (ZINB) variables [61] [62] [63] [64] [65] GLOB (globalization 1.023*** 0.626** 0.951*** 1.022*** 0.954*** index) (3.20) (2.01) (2.86) (3.13) (2.92)

CORR (control of -0.320 -0.265 -0.232 -0.319 -0.238 corruption) KKM (-1.37) (-1.11) (-0.93) (-1.34) (-0.97)

CTRL: log (GDP per 0.803*** 0.760*** 0.823*** 0.807*** 0.823*** capita) (4.13) (4.07) (4.08) (4.04) (4.18)

CTRL: 6.75e-09*** 8.39e-09*** 6.23e-09** 6.89e-09*** 6.40e-09*** population size (3.14) (4.17) (2.44) (2.62) (3.05)

Oil-production 0.906*** 0.813*** 0.942*** 0.902*** 0.933*** Dummy (3.09) (2.97) (3.08) (2.97) (3.16)

Regional Fixed Effects Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Transition Countries Yes No Yes Yes No included USA excluded No No Yes Yes No RUS excluded No No Yes No Yes Prob. > chi2 0.000 0.000 0.000 0.000 0.000 # of observations 473 413 465 469 469 Notes: z-statistics in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. CORR: higher values = lower level of corruption.

122 Appendices 123

Table A5: BILLIONAIRES PER COUNTRY PER YEAR

Countries 1996 1997 1998 1999 2000 2001 2002 2003 Argentina 3 3 2 3 4 4 1 1 Australia 1 2 2 3 3 3 3 3 Austria 0 0 0 1 0 0 0 1 Bahrain 1 0 0 0 0 0 0 0 Belgium 0 0 0 1 0 1 1 1 Bermuda 0 0 0 1 0 1 0 0 Brazil 10 3 3 8 9 6 7 4 Brunei 0 1 1 0 0 0 0 0 Canada 5 3 4 9 15 16 15 15 Chile 5 3 1 4 3 2 2 3 China 0 0 0 0 0 1 1 0 Colombia 3 3 3 3 1 1 1 1 Cuba 0 1 0 0 0 0 0 0 Czech Republic 0 0 0 0 0 0 0 0 Denmark 0 2 2 2 2 2 2 3 Ecuador 1 1 0 1 0 0 0 0 Egypt 0 0 0 0 1 1 0 0 France 15 9 8 16 14 15 15 13 Georgia 0 0 0 0 0 0 0 0 Germany 47 20 17 43 42 28 35 43 Greece 5 2 1 2 2 3 2 1 Hong Kong 18 8 8 13 13 14 12 11 India 3 4 2 7 9 4 5 7 Indonesia 10 7 4 3 2 2 1 1 Iraq 0 1 1 0 0 0 0 0 Ireland 1 1 1 2 2 2 2 2 Israel 3 1 1 2 2 5 3 3 Italy 6 4 4 4 6 17 13 11 Japan 40 14 12 30 43 29 25 19 Kuwait 1 2 2 1 1 1 1 1 Lebanon 2 2 2 2 1 1 1 1 Lichtenstein 1 0 0 1 1 1 1 1 Malaysia 11 6 2 4 5 4 5 4 Mexico 15 6 7 11 13 13 12 11 Monaco 0 0 0 0 0 0 0 0 Netherlands 3 4 2 4 3 2 2 4 New Zealand 0 0 0 0 0 1 0 0 Norway 0 0 0 0 1 2 1 1 Peru 1 0 0 0 0 0 0 0 Philippines 9 4 4 4 5 3 4 2 Poland 0 0 0 0 0 0 0 0 Portugal 0 0 2 3 2 2 1 1 Qatar 0 0 1 0 0 0 0 0 Russia 0 4 1 0 0 8 7 17 Saudi Arabia 7 5 6 5 6 8 9 9 Singapore 4 3 3 4 5 6 5 5 South Africa 2 2 1 2 1 2 2 2 South Korea 7 4 1 2 1 2 2 2 Spain 3 1 1 5 5 8 7 7 Sweden 5 3 3 4 5 5 6 5 Switzerland 12 6 6 14 14 16 13 9 Taiwan 7 3 5 7 6 5 5 5 Thailand 10 3 0 0 1 2 2 2 Turkey 3 3 3 3 4 5 6 5 United Arab Emirates 0 1 2 2 1 1 1 1 United Kingdom 5 6 5 10 14 12 13 14 USA 135 60 71 50 52 269 243 222 Venezuela 2 2 2 2 2 2 2 2 Total 476 497 538 322 298 209 223 422

Appendices 123 124

Table A6: CLASSIFICATION OF TRANSITION ECONOMIES Transition economies in Europe and the former Soviet Union CEE Albania, Bulgaria, Croatia, Czech Republic, FYR Macedonia, Hungary, Poland, Romania, Slovak Republic, Slovenia

Baltics Estonia, Latvia, Lithuania

CIS Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyz Republic, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine, Uzbekistan

Transition economies in Asia Cambodia, China, Laos, Vietnam Source: http://www.imf.org/external/np/exr/ib/2000/110300.htm (accessed 4th October 2011)

124 Appendices 125

APPENDIX C: CHAPTER 5

Table A7: TEAMS COLOURS Team Identifier State Primary Colour Secondary Colour Adelaide Rams SA Red Blue Auckland Warriors NZ Blue White Balmain Tigers NSW Orange Black Brisbane Broncos QLD Red Yellow (Gold) Bulldogs Bulldogs NSW White Blue Canberra Raiders ACT Green White Canterbury Bulldogs NSW White Blue Cronulla Sharks NSW Blue White East(s) Sydney/Suburbs Roosters NSW Blue White Gold Coast Chargers/Giants/Seagulls* QLD Grey(Green)* Black(Purple)* Gold Coast Titans QLD Blue White Hunter Mariners NSW Blue Yellow Illawarra Steelers NSW Red White Manly Sea Eagles NSW Red (Maroon) White Melbourne Storm VIC Purple Blue New Zealand Warriors NZ Black Grey Newcastle Knights NSW Blue Red Newtown Jets NSW Blue White North Queensland Cowboys QLD Blue Yellow Northern Eagles Eagles NSW White Red North(s) Sydney Bears NSW Red Black Parramatta Eels NSW Yellow Blue Penrith Panthers NSW Brown(Black)** White(White) Perth Reds Kangaroos WA Red Black Sharks Sharks NSW Blue Black South Queensland Crushers QLD Yellow(Gold) Blue South(s) Sydney Rabbitohs NSW Green Red St George Dragons NSW White Red St George Illawarra Dragons NSW White Red Sydney Bulldogs Bulldogs NSW White Blue Sydney City Roosters NSW Blue White Sydney Roosters Roosters NSW Blue White Sydney Tigers Tigers NSW Orange Black Western Reds Kangaroos WA Red Black West(s) Sydney/Suburbs Magpies NSW Black White Wests Tigers Tigers NSW Orange Black Notes: 1 = Black; 2 = Blue; 3 = Brown; 4 = Green; 5 = Grey; 6 = Orange; 7 = Purple; 8 = Red; 9 = White; 10 = Yellow. The Gold Coast Chargers/Giants/Seagulls did not have red as a primary or secondary colour, thus we have grouped the three incantations together. Penrith has a change of colours from a predominate Brown to Black, neither of which affects the Red Variables. Volume of red has been taken into account for teams such as the Broncos on a season by season basis.

Appendices 125 126

Table A8: ROBUSTNESS CHECK FOR TEAMS COLOUR CLASSIFICATION

Club Colour 1 R,G,B C,M,Y,K% Colour 2 R,G,B C,M,Y,K% Broncos PMS 222 112,25,61 0,77,45,56 PMS 123 255,198,30 0,22,88,0 Bulldogs Reflex Blue 12,28,140 91,80,0,45 White Raiders PMS 354 0,183,96 100,0,47,28 PMS 382 186,216,10 13,0,95,15 Cowboys PMS 289 0,38,73 100,47,0,71 PMS 109 249,214,22 0,14,91,2 Titans PMS 299 0,163,221 100,26,0,13 PMS 143 239,178,45 0,25,81,6 Manly PMS 222 112,25,61 0,77,45,56 White Storm PMS 2602 130,12,142 8,91,0,44 White Knights Reflex Blue 12,28,140 91,80,0,45 PMS 1795 214,40,40 0,81,81,16 Eels PMS 293 0,81,186 100,56,0,27 PMS 116 252,209,22 0,17,91,1 Panthers PMS Black 61,51,43 0,16,29,76 PMS 3155 0,109,117 100,6,0,54 Sharks PMS 291 96,175,221 56,20,0,13 White/Black 61,51,43 0,16,29,76 Rabbitohs PMS 348 0,135,81 100,0,40,47 PMS 186 206,17,38 0,91,81,19 Dragons PMS 485 216,30,5 0,86,97,15 White Roosters PMS 295 0,56,107 100,47,0,58 PMS 1795 214,40,40 0,81,81,16 Warriors Black 61,51,43 0,16,29,76 PMS 877 153,153,153 0,0,0,50 Wests Tigers PMS 151 247,127,0 0,48,100,3 Black 61,51,43 0,16,29,76 Note: Pantone Conversion http://euro-bags.eu/pantone?p=10&limit=40. All club colours are officially listed in the Pantone PMS colour system. RGB is the digital colour system (light system) and the CMYK is the printing colour system.

126 Appendices