ARE DISAGREEMENTS AMONG MALE AND FEMALE ECONOMISTS MARGINAL AT BEST?: A SURVEY OF AEA MEMBERS AND THEIR VIEWS ON ECONOMICS AND ECONOMIC POLICY

Transcription

1 ARE DISAGREEMENTS AMONG MALE AND FEMALE ECONOMISTS MARGINAL AT BEST?: A SURVEY OF AEA MEMBERS AND THEIR VIEWS ON ECONOMICS AND ECONOMIC POLICY (forthcoming in Contemporary Economic Policy) ANN MARI MAY, University of Nebraska-Lincoln MARY G. MCGARVEY, University of Nebraska-Lincoln ROBERT WHAPLES,* Wake Forest University The authors survey economists in the U.S. holding membership in the American Economic Association (AEA) to determine if there are significant differences in views between male and female economists on important policy issues. Controlling for place of current employment (academic institution with graduate program, academic institution - undergraduate only, government, for-profit institution) and decade of Ph.D., the authors find many areas in which economists agree. However, important differences exist in the views of male and female economists on issues including the minimum wage, views on labor standards, health insurance, and especially on explanations for the gender wage gap and issues of equal opportunity in the labor market and the economics profession itself. These results lend support to the notion that gender diversity in policy-making circles may be an important aspect in broadening the menu of public policy choices. * The authors would like to thank Professor Julia McQuillan, Director of the Bureau for Sociological Research, University of Nebraska-Lincoln for her helpful suggestions on the survey questionnaire, Michael L. May, Swarthmore College '11 for his research assistance, and the Research Fund of the Department of Economics at the University of Nebraska-Lincoln for assistance in funding a portion of this research. We also thank three anonymous referees for their very helpful comments and suggestions. May: Professor of Economics, University of Nebraska-Lincoln, Lincoln, NE. amay1@unl.edu McGarvey: Associate Professor of Economics, University of Nebraska-Lincoln, Lincoln, NE. mmcgarvey@unl.edu Whaples: Professor of Economics, Wake Forest University, Winston-Salem, NC. whaples@wfu.edu

2 I. INTRODUCTION Few changes in the demographics of higher education in the U.S. have been as pronounced as the increasing percentage of doctoral degrees awarded to women in the past forty years. In 1975 men received the majority of bachelors, masters, and doctoral degrees; by 2000, the majority of bachelor s and master s degrees were awarded to women. In 2002, for the first time in U.S. history, more American women than American men received doctoral degrees from U.S. universities and today more doctorates are awarded to women than to men at U.S. universities. 1 Although the proportion of women in economics continues to lag behind that of other social science disciplines, data from the Survey of Earned Doctorates shows that women represent 34.5 percent of new doctorates in economics from U.S. universities in 2010, up from 27.0 percent in 2000 (Doctorate Recipients from US Universities 2010: Table 14). This increase represents a 7.5 percentage point increase over the period as compared with an average 2.9 percentage point increase for all other fields. As a result, increasing numbers of female economists are actively involved in research and policy-making on a variety of levels including the Board of Governors of the Federal Reserve Bank to the Council of Economic Advisors. But does it make any difference if men or women are at the table when economic policies are debated and alternatives considered? Or, are disagreements between male and female economists marginal at best? In this study we survey male and female economists who are members of the American Economic Association (AEA) and provide what is the first systematic study of the views of male and female economists on a wide variety of policy issues to determine if there are differences in 1

3 the views of male and female economists after controlling for place of current employment (academic institution with graduate program, academic institution - undergraduate only, government, for-profit institution) and decade of Ph.D. A Consensus and the Economics Profession Numerous studies using survey data to examine the degree of consensus among economists have been published over the past twenty-five years (Kearl et al. 1979; Frey et al. 1984; Alston, Kearl and Vaughn 1992; Fuller and Geide-Stevenson 2003; and Whaples 2006). While these surveys appeared relatively late in economics as compared with other social science disciplines, the importance of consensus within the discipline has a long history. In fact, concern over consensus dates back to the origins of the American Economic Association where the annual meeting offered a forum for developing consensus in a discipline thought to reflect a sometimes embarrassingly wide array of views on economic policy questions. Economists, in particular, had much to gain in the professionalization of higher education and the lack of consensus on economic questions served as yet one more reminder of the possibly less-thanscientific nature of a discipline aspiring to be scientific (Coats 1993). Recent studies examine consensus among economists by focusing on underlying assumptions, methodology, and policy issues. The results of these studies indicate that economists are less supportive of government intervention in the economy than their counterparts in other social science disciplines (Klein and Stern 2005 and 2007) and that a fair degree of consensus exists among economists. Kearl et al. (1979 p. 36) in their study of U.S. economists in a survey conducted in 1976 report more consensus on microeconomic than macroeconomic issues, but overall report that the perceptions of wide-spread disagreement are simply wrong. In a follow up study, Alston, Kearl, and Vaughan (1992) find that although 2

4 opinions of U.S. economists on particular issues had shifted, a similar pattern of consensus emerged with the caveat that degree vintage and subgroup (AEA membership, government economists, business economists, teachers of principles of economics courses at four year institutions, and institutional economists) are also important determinants of consensus on various questions. Alston, Kearl and Vaughn (1992) find considerable consensus among economists overall. Yet, as Fuchs et al. (1998) later point out, the seven questions about policy in Alston, Kearl and Vaughn (1992) show a very high level of disagreement. In their own study Fuchs et al. (1998) find considerable disagreement between labor and public economists about policy proposals in their areas of specialization. In an additional study, Fuchs (1996) surveyed health economists and found considerable disagreement about major issues of health policy. Other studies have examined the influence of international differences (Frey et al. 1984), age differences or age upon receipt of degree (Klein and Stern 2005 and Alston, Kearl, and Vaughn 1992), political party and voting patterns (Klein and Stern 2005 and 2007), school of thought (Alston and Vaughan 1993), and economic field (Kearl et al and Fuller and Geide- Stevenson 2003) to determine if these factors produce differences in the views of economists. Yet, absent from these discussions is any examination of gender differences among economists. In contrast to these previous studies, only a few studies examine gender differences in opinions of economists. 2 Davis (1997), in his study of AEA members, examines economists views of methodology, the social value of research, perceptions about the publication of that research, and perceptions about the influence of race and gender on economic research. He reports there are significant differences in the views of male and female economists with respect to whether author recognition among professional economists influences the probability of an article s acceptance in refereed economics journals in the U.S., whether there is a good-old- 3

5 boy network in the economics profession, and whether economists are amenable to interdisciplinary research approaches. Specifically, Davis finds that although the majority of male and female economists have similar opinions, women typically reach a much stronger consensus particularly on issues of equity and fairness in the economics profession. In her study of AEA economists conducted in 1992, Albelda (1997) examines a variety of questions concerning gender and the economics profession. According to Albelda, men were much less interested than women in having more attention paid to topics such as women s labor force participation, the impact of fiscal and monetary policies on women and the family structure, wage discrimination, and the economic status of minority women (Albelda 1997, pp ). Moreover, there was little support for the notion that feminism has impacted the discipline of economics nor did economists express much interest in seeing more feminist economic analysis. Most notable, however, were the large gender gaps in response to questions about the treatment of women in the economics profession (Albelda 1997, p. 69). As Albelda reports, almost one-third of male respondents agree with the statement that it is easier for women to get tenure and promotions than it is for men, while less than 4 percent of women agreed with the statement (Albelda 1997, p. 81). Hedengren, Klein and Milton (2010) examine gender differences in policy-oriented petitions endorsed by economists, finding that men are much more likely to sign libertyenhancing petitions, which call for a smaller role for the state, while women economists are somewhat more likely to endorse petitions calling for greater governmental intervention. Davis et al. (2011) conduct a survey on a wide range of issues (including favorite economic thinkers, economics journals, and blogs) and find that women economists are much more likely than men to be members of the Democratic Party and to favor policies that restrict individual liberty. In 4

6 addition, Stastny (2010) finds that women economists in the Czech Republic tend to favor much more government intervention than do male economists. While these studies offer important insights into gender differences on views of the economics profession, they ask a fairly narrow range of questions and are not designed to systematically explore the nuances of differences between male and female economists. Given the changing demographics of the economics profession over the past thirty years, these differences, if they exist, may have significant implications for policy-making outcomes. B. Gender and Economics in the U.S. Although the gender gap in voting and party identification has been a topic of widespread interest since the 1980 election, less attention has been paid to gender differences in policy preferences. Yet, several studies indicate that gender differences in policy preferences exist in the U.S. and have increased since the 1970s (Smith 1984; Shapiro and Mahajan 1986; Inglehart and Norris 2000; Saad 2003; and Pew Research Center 2009). While disagreement exists as to the cause of these differences in voting patterns, party identification, and public opinion surveys on policy issues, women and men appear to differ most on matters concerning violence and force and so-called compassion issues concerning aid to the poor, unemployed, sick, and others in need, and in their views on regulatory policies (Shapiro and Mahajan 1986, p ). While gender differences in the general population are well studied, gender differences in the views of economists have received little attention. In fact, no systematic study of gender differences in views of American economists on a wide variety of economic policy issues has yet 5

7 been undertaken. Our study attempts to remedy this by examining the views of male and female economists who are members of the AEA in the U.S. Using survey data on core precepts and approaches in economics, views on market solutions and government intervention, specific policy issues, and gender equality in economics and society we ask the question, Do male and female economists share the same views on underlying assumptions, methodological approaches, policy solutions, perceived problems in economics, and equal opportunity in society and the economics profession? Because survey data from non-economists indicates that there are often significant gender differences in views on policy issues (Saad 2003), we are interested to see if these differences disappear among economists with shared training and backgrounds. II. METHODOLOGY A Data and Descriptive Statistics Our goal is to estimate the extent of differences in opinions on economic methodology and policy issues between male and female economists with doctoral degrees who are members of the AEA. The population from which we sampled is the list of members in the 2007 American Economic Association Directory who received both their undergraduate and Ph.D. degrees from United States institutions. 3 The selection process yielded 202 randomly selected male members and 202 randomly-selected female members to whom the survey was mailed in early November The overall response rate was 35.4 percent which is similar to previous surveys. 4 < INSERT TABLE 1 HERE > Table 1 contains descriptive statistics (overall and disaggregated by sex) for the entire sample, for the 143 members who returned usable surveys, and for the 261 non-responders. Characteristics of the AEA members include demographic information on sex, year of Ph.D., and 6

8 current type of employment. This information was obtained from survey responses of those who returned the surveys and from the AEA directory entries of the non-responders. Additionally, Table 1 reports whether or not the AEA member chose to list her or his research interests in the 2007 AEA directory. This information was included as an indication not only of the economist s research activity but also the willingness of the individual to respond to survey questions. The original, random sample of AEA members was evenly divided between men and women by design. Approximately 67 percent of those sampled are employed in academic institutions with 41 percent teaching in graduate programs and 26 percent in undergraduate programs. Fourteen percent of those sampled work in the for-profit sector and 11 percent are employed by the government. The remaining eight percent are either retired, not in the formal labor market, or working in the non-profit sector. Although, in our sample, it is more likely for women to be employed in academic institutions than men, and more likely for men to work in the government or for-profit sectors than women, these differences in employment rates are not statistically significantly different from zero. The economists in the AEA survey earned their doctoral degrees about 22 years ago with women s degrees earned about 5 years more recently than men s and the average difference is statistically significant. It is interesting that, in our random sample of AEA members, male economists are more likely to list their research interests in the AEA directory than female economists. The ten percentage point difference is also statistically significant. A comparison of the responders and non-responders descriptive statistics shows that most of the characteristics are similar. Forty-five percent of those responding to the survey are women whereas 52 percent of the non-responders are women. The distribution of overall employment across sectors is roughly the same with the exception of employment in an 7

9 academic institution with a graduate program. Formal statistical tests find evidence that the overall proportions of economists teaching in graduate institutions differ for the responders and non-responders (.35 versus.44) and that it is the male responders who are statistically significantly less likely (by 11 percentage points) to teach in graduate institutions than the male non-responders. The proportion of economists who list their research interests in the AEA directory is 55 percent for the responders and 62 percent for the non-responders. On average, those responding to the survey received their Ph.D.s about two and a half years more recently than those not responding and this difference is statistically significant. A smaller proportion of those economists who responded to the survey received their degrees in the 1970s than those who did not respond. Most of the statistically significant male and female differences in the original random sample remain significant in the sample of responders. Male economists are more likely to list their current research in the AEA directory and received their Ph.D.s earlier than female economists. In the responders sample, male economists are more than twice as likely to be working in the governmental sector as female economists. Although this relative difference in government employment exists in the original random sample, it is larger and statistically significant among the responders. Our statistical analysis of gender differences in economists opinions on methodology and policy issues is based on the responses of those who returned their surveys; we do not know the opinions of the non-responders. It is possible that estimation using only the observable survey responses will lead to inconsistent estimates of the gender effect. We address this issue in the next section by estimating the decision to respond to the survey as a probit model and testing 8

10 for non-response bias in our estimation of gender differences in the group-level scaled answers using Heckman s (1979) procedure. B Scaled Responses to Group Questions and Test for Selection Bias The survey questions can be classified into five groups, each group designed to ascertain the respondent s views about different types of issues. The first group of questions relate to core principles in economics and economic methodology. The second through fourth groups examine views on market solutions and government intervention, government spending, taxing, and redistribution, and the environment. The fifth group of questions asks specifically about equal opportunity in society and gender equality in the economics profession. The wide variety of policy-related questions allows us to examine possible gender differences on a host of the most important and topical issues currently facing economists. Before examining gender differences in answers to specific questions, we first construct an overall measure of the economist s opinion on each of the five group topics. The Appendix describes the construction of each scaled opinion measure and the survey questions upon which each is based. Tables A1 and A2 provide descriptive statistics of the opinion measures and correlations of the responses between topics. We find that the Cronbach s Alpha values are greater than.70 for each of the scaled answers, indicating that each is a reliable measure of the overall opinion on the group topic. Our goal is to estimate the mean difference in opinion on economic methodology and policy issues of male and female Ph.D. economists who received their educational training in the United States. In order to better isolate the effect of gender on economists opinions, we need to control for those variables that are both correlated with the economist s opinions on policy and 9

11 methodology and with the economist s gender. Because female economists received their terminal degrees more recently, on average, than male economists, and opinions might be related to degree vintage, we include four binary indicators of the decade the Ph.D. was received as one of our control variables. The omitted category of economists contains those who received their Ph.D.s prior to We also include controls for the economist s place of employment. Although there were no statistically significant differences in the proportions of men and women employed in graduate or undergraduate-only academic institutions, nor in the for-profit sectors, there might be differences in types of employment by gender after accounting for degree vintage. The economists opinions on policy issues and methodology might reasonably be related to, or influenced by, their type of work. For these reasons, we also include the mutually exclusive employment category variables, graduate, for profit, government, and undergrad as control variables in the regression. The omitted category includes employment in the non-profit, nonacademic, non-government sector, not currently employed or retired. model, We base our inference on the estimated coefficient on the binary indicator, female, in the y female PhD1970s PhD1980s PhD1990s PhD2000s i, j j1 i j2 i j3 i j4 i j5 i graduate government profit undergrad j6 i j7 i j8 i j9 i j0 i, j, (1) where i denotes the ith economist and j denotes the jth question or group of questions. Because we cannot observe the survey answers of those economists who did not respond, even if the regression error is independent of the control variables, it is possible that estimation of j1 using only the responders observations will produce inconsistent and biased estimates of the gender 10

12 effect. We test the null hypothesis of no selection bias using the Heckman (1979) two-stage procedure. We assume that the decision to respond to the survey follows a probit model, Ps ( 1 x) PZ ( x ), where s i = 1 if individual i responds to the survey and s i =0, if not. i i i The (1 x k) vector of variables, x i, that influence the choice to respond to the survey includes the control variables in equation (1). We also include a binary control for whether or not the economist lists her or his research interests in the AEA 2007 directory (listed research) and the interaction of listed research with for profit to allow the effect of listed research on the probability of response to differ for those working in the for-profit sector. In this framework, the economist responds to the survey (s i = 1) if the unobservable net benefit from doing so ( s * i ) is positive, or if s x u 0, where u i is a standard normal * i i i random variable and is distributed independently of x i. Under these assumptions, consistent estimation of the gender effect on economists opinions on question j ( j1 ) can be obtained using only the responders information if the error term in equation (1) is uncorrelated with u i, the error term in the response equation. We test the hypothesis that these errors have zero correlation by adding the estimated inverse Mills ratio, IMR ˆ ˆ i= pdf z(xiα) / CDF z(xiα) as a regressor in equation (1) and testing that its coefficient equals zero. < INSERT TABLE 2 HERE > Table 2 reports the probit estimates of the response equation. The results indicate that the probability of responding to the survey is lower for female economists relative to males and for economists employed in two of the included job categories relative to males and those who are 11

13 retired or employed in non-academic, non-government, or for-profit institutions. Those economists who earned their Ph.D. in the 2000s are more likely to respond to the survey, relative to those who earned their degrees prior to Those economists who list their current research interest in the AEA directory are less likely to respond; however, those who work in for-profit institutions and list their current research are more likely to respond. < INSERT TABLE 3 HERE > Table 3 presents the ordinary least squares estimation of equation (1) for each of the five topic groups, with and without the inclusion of the estimated inverse Mills ratio. The dependent variable is the scaled response to the questions on each topic, standardized to have zero mean and unit variance, so that the coefficient on female is the average difference in female economists response relative to male economists in standard deviation units. The results indicate that male and female U.S. trained, AEA members generally agree on core economic precepts and methodology and on U.S. environmental policy but differ on issues related to market solutions and government intervention, government-backed redistribution, and the extent of gender inequity in the U.S. labor market and within the economics profession. Tests of the null hypothesis of no selection bias for each topic category provide no evidence that the selection equation error is correlated with the error in equation (1); moreover, the inclusion of the inverse Mills ratio in the regressions has little effect on the OLS point estimates and estimated standard errors. These results support the validity of our using the sample of responding economists to estimate model (1) and suggest that our inference on gender differences will not suffer from selection bias. 12

14 III. GENDER DIFFERENCES IN ECONOMISTS VIEWS ON POLICY ISSUES This section presents Seeming Unrelated Regression (SUR) estimation results based on the survey response to estimate gender differences in economists views on economic methodology and policy issues. The SUR model is relevant when the regression errors are correlated across equations and have different variances. For convenience, we reproduce our model of economist i s response to question j (or topic j) from the previous section, y female PhD1970s PhD1980s PhD1990s PhD2000s i, j j1 i j2 i j3 i j4 i j5 i graduate government profit undergrad j6 i j7 i j8 i j9 i j0 i, j, (1) where now we treat equation j as part of a SUR system of equations. The error terms represent those unobservable individual-specific factors that affect opinions but are unrelated to gender, decade in which the Ph.D. is earned, and type of employment. We allow the variances of the errors to differ by question, the errors to be correlated across questions, and the error covariance matrices to differ for males and females. The Generalized Least Squares (GLS) method increases the precision of our estimates by incorporating the estimated relationship among unobservable factors that form opinions on related topics. Our first estimation strategy is to measure gender differences in economists views on general issues addressed by the five groups of survey questions. This provides a more precise measure of opinion on the issue by incorporating repeated observations on the same topic (averaging an individual s answers to all questions within the group topic) but limits the sample to individuals who answered all questions within the five groups. Our second strategy is to estimate the extent of gender differences in average responses to specific questions within a general topic area. The dependent variable is individual i s response to question j. Although y i,j is 13

15 a less precise estimate of individual i s opinion (it is based on the response to only one question rather than an average of repeated responses on a topic), the sample size increases to the number of economists who answered all questions in one group rather than all groups. A Views on General Topic Areas The first SUR system explains the standardized, scaled answers that measure the deviations from the average economists views on the five topics: core precepts and economic methodology, market solutions and government intervention, government spending, taxing, and redistribution, U.S. environmental policies, and gender and equal opportunity. The optimal weighting scheme incorporated in the GLS systems estimation increases the efficiency of the estimates over the OLS method. The number of observations, however, is smaller for the systems estimation because it is based on the number of economists that have complete information for all questions in each of the five groups. < INSERT TABLE 4 HERE > Table 4 contains the GLS estimates of the coefficients in equation (1) for j= 1,2,...,5, corresponding to the five group topics. The dependent variables are the scaled responses, standardized to have zero mean and unit variance, so the coefficient on a binary control variable (female, employment category, or decade of Ph.D.) is the average difference (in standard deviation units) in the response of economists who fall into the respective category relative to those in the omitted category, given the other controls. The GLS system results confirm the conclusions based on the OLS and Heckit estimates of Table 3. Male and female members of the AEA with doctoral degrees from U.S. institutions appear to agree on core precepts and economic methodology whereas female economists tend to favor government-backed redistribution 14

16 policies more than males, view gender inequality as a problem in the U.S. labor market and economics profession more than males, and favor government intervention over market solutions more than their male counterparts. The mean gender differences in opinions on these topics are relatively large. The mean views of women economists on government spending, taxing, and redistribution and on gender inequality are both approximately one standard deviation away from the mean opinion of male economists. Although the divergence in magnitudes of the GLS and OLS estimated gender difference in opinion on U.S. environmental policies is not large, the GLS estimate is statistically significant. The mean response of female economists is about.45 standard deviation greater than the mean male response, indicating that women tend to favor an increase in U.S. environmental protection more than men. Generally speaking, the demographic controls contribute little to explaining views on the five group topics once one controls for the economist s gender. 5 The economists place of employment appears to have no effect on opinion once gender and degree vintage are considered. The results indicate a statistically significant and relatively large difference in opinion between those who earned their degrees prior to 1970 and those who earned their degrees more recently. Those who graduated in the 1970s or 1980s are less likely to agree with core precepts and economic methodology than economists who earned their degrees before The results also indicate that economists with degrees from the 1980s are more likely to favor government interventions over market solutions than economists with degrees prior to Interestingly, the only characteristic that matters for the scaled answer to the Gender and Equal Opportunity questions is whether or not the economist is female. The results from system GLS estimation of economists mean responses to the five groups of questions on economic methodology, the role of government (in allocating resources, 15

17 redistributing wealth, protecting the environment), and equal opportunity provide strong evidence of gender differences. Female economists average responses differ from male economists by a statistically significant margin in four of the five topic areas after controlling for type of employment and degree vintage. Mean differences in views on government-backed redistribution and on gender and equal opportunity also differ by a substantial margin - about one full standard deviation. B Views on Core Precepts and Economic Methodology We include several questions designed to address core economic precepts one to examine the question of individuals as rational utility-maximizers and another to examine the assumption that human wants are unlimited. Also included in this group are questions concerning economic methodology that attempt to ascertain views on mathematics in economic modeling, deductive versus empirical approaches to the study of economics, and two questions that examine potential differences in views on modeling of household decision-making and measuring aggregate economic output. GLS estimation of a SUR system for responses to the Group 1 questions indicate no difference in mean opinions by gender after controlling for degree vintage and place of employment. 6 Economists mean responses indicate agreement with the notions that individuals are utility-maximizers and that human wants are unlimited. Likewise, when examining methodological questions in the economics profession there appears to be no significant difference in beliefs. On average, economists agree that mathematical modeling should be an important part of economics and that research on households should include intra-household decision-making. Economists mean response to question 23 indicates a neutral view on GDP 16

18 being an insufficient measure of overall economic performance. The GLS estimation results for the response to question 39 indicate that, although there is no difference in opinion based solely on gender, those economists who work in the for-profit sector are more likely to believe that the profession relies too much on empirical analysis and not enough on deduction. It would appear that on several core concepts in economics and methodology, the beliefs of male and female economists are very similar. Male and female economists that are members of the AEA appear to start from the same assumptions about how people behave and how to approach the study of economics. C Views on Market Solutions and Government Intervention Previous studies such as Klein and Stern (2005) confirm that economists are less supportive of government intervention in the economy than their counterparts in other social science disciplines. Indeed, it is perhaps the penchant for free market solutions that is the hallmark of the discipline (Fourcade 2009). Yet, when we examine the question of support for market solutions versus government intervention, interesting differences between male and female economists begin to emerge. < INSERT TABLE 5 HERE > Despite the similarities in views on core principles and methodological approaches to economics, our study shows differences in the views of male and female economists in their support of market solutions versus government regulation. The GLS estimation results indicate that the mean responses of women economists are significantly lower than those of male economists, suggesting women economists are less supportive of market solutions over government interventions than male economists. The only exception is on the issue of whether 17

19 the government should tax unhealthy foods where both men and women appear approximately equal in their opposition. Table 5 presents GLS estimates of male and female economists expected response to each proposition for which the estimated female mean difference exceeds one half a point on the Likert-type scale. For those questions, the probability of responding disagree or strongly disagree to the proposition conditional on gender, degree vintage, and place of employment is estimated by a linear probability model. 7 The table reports the estimated probabilities of disagreeing with the proposition at the sample mean values of the Ph.D. and place of employment dummy variables for both male and female economists. After controlling for Ph.D. vintage and type of employment, we find that women are more likely to disagree that either the European Union or the United States has excessive government regulations by about 20 percentage points. Female economists are 27 percentage points more likely to disagree that parents should be given educational vouchers and 14 percentage points more likely to disagree that Wal-Mart generates positive net benefits. In other words, evidence suggests that male economists are more likely to prefer market solutions to economic problems. 8 Although the SUR estimated mean difference in opinion on whether the U.S. should drill for oil in the Arctic National Wildlife Refuge indicates a lower level of agreement for women, the 12 percentage point difference in the probability of disagreement is not statistically significant. According to the SUR results, the largest difference in the mean response is due to degree vintage. Those economists who earned their degrees in the 1980s or 1990s, regardless of their gender or place of employment, are much more likely to oppose drilling in the Arctic National Wildlife Refuge than economists whose Ph.D.s were granted prior to

20 It is interesting to us that there appears to be little disagreement between male and female economists about the question of whether the U.S. should impose taxes on unhealthy foods what many in the popular press have dubbed an example of the so-called nanny state. Both male and female economists show little support for the imposition of a tax on unhealthy foods. D Views on Government Spending, Taxing, and Redistribution Differences in views between male and female economists not only reflect variations in preferences for market solutions versus government intervention but also appear to reflect values differences some of which correspond to what have been called compassion issues. These differences may not emerge when we consider overall levels of government spending or taxation, but may emerge more clearly when we ask specific questions about the nature of government spending, income distribution, revenue financing options, as well as evaluating the impact of various government programs. The estimation results indicate statistically significant gender differences in opinions on each of the nine propositions; however, Table 6 reports only those with mean differences greater than one half. < INSERT TABLE 6 HERE > When asked their opinions about the size of the government, the average male economist s response is closest to the too large choice whereas the average female economist s response is closest to the about right response. Examination of the probability estimates shows that women are 24 percentage points more likely than men to believe the size of the government is either too small or much too small. There are also statistically significant differences in mean responses between men and women economists in their views of the composition of government spending and other questions about redistribution. For example, the mean response 19

21 to question 30 of a female economist is almost 2.0, indicating that current levels of military spending are too large while the average male economist s response is about 2.4, closer to the about right choice. 9 Conversely, when asked whether employers should be required to provide health insurance for their employees, the mean gender gap is almost one full point on the opinion scale with the average male response at 2.4 closer to the disagree category. The estimated proportion of female economists is.23 greater than the proportion of male economists who agree or strongly agree with the proposition. Several questions asked about the distribution of income and the tax structure as it relates to income distribution. The estimated mean response of male economists indicates a belief that the tax structure should remain at its current level of progressivity whereas the mean response of female economists falls closer to a belief that taxes should be more progressive. In fact, women are 41 percentage points more likely than men to favor a more progressive tax structure. A comparison of economists mean responses to the issue of whether the U.S. income distribution should be more equal shows that the average male indicates disagreement while females average response is closer to a neutral opinion. The linear probability model estimates that women are 32 percentage points more likely to agree or strongly agree with making the U.S. income distribution more equal than men. Along the same lines, there is disagreement on whether the U.S. should implement a tax on consumption and rely less on income taxes for revenue. The average male economist is closer to being neutral with a tax on consumption while the average female economist is closer to disagreeing. According to the point estimates, 54 percent of female economists either disagree or strongly disagree compared to 31 percent of male economists. This result may not be surprising given the difference between male and female economists on the question of making the tax structure more progressive. 20

22 We see the same gender differences in so-called compassion issues in questions about international trade and labor standards and the impact of the minimum wage. When asked if the U.S. should link import openness to the labor standards of its export partners, the average male economist s response is closer to disagreement whereas the average female economist s response is closer to neutral. Female economists are 30 percentage points more likely to agree or strongly agree with linking openness to labor standards than male economists. The average female economists opinion on the proposition that increasing the minimum wage leads to increased unemployment among unskilled workers is neutral whereas the average male indicates agreement. E Views on Environmental Policy Issues On environmental issues, we ask questions about energy taxes in general, taxes on emissions of greenhouse gases in particular, as well as ethanol subsidies to gauge the degree to which male and female economists agree on the use of taxes and subsidies to discourage or shape desired activities that may affect the environment. On this issue of taxes and subsidies, there appears to be no significant difference in the opinions of male and female economists. On the question of nudging environmental policy through various taxes, the average responses of both male and female economists indicate agreement that energy taxes should be increased along with taxes on emissions of greenhouse gases. Men and women also show no divergence in average opinions on U.S. policies favoring the environment over economic growth. Neither male nor female economists indicate support for an increase in ethanol taxes. 21

23 F Views on Gender and Equal Opportunities Although previous studies have not examined gender differences in views of economists on policy issues, studies by Davis (1997) and Albelda (1997) both show differences in the views of male and female economists concerning equity and equal opportunity in the economics profession. In this section we examine survey results on a variety of questions related to notions of equal opportunity and affirmative action policies in the U.S. as well as questions of gender equality in the economics profession. These questions allow us to examine in more detail the views of male and female economists on issues of equal opportunity. < INSERT TABLE 7 HERE> The results of our survey of male and female economists show that the area of largest disagreement between men and women lies in views on equal opportunity. These differences reveal themselves not only in views of gender equality in the economics profession but in society in general. Large disparities in average responses between male and female economists emerge in response to the statement, Job opportunities for men and women in the U.S. are currently approximately equal. The estimation results show the mean response of female economists is one point (a full standard deviation) lower than that of male economists after controlling for degree vintage and employment type. The average response of males is closer to agree with the statement while the average response of females is closer to neutral. The linear probability model, however, indicates that 58 percent of female economists with the average characteristics of the sample either disagree or strongly disagree that job opportunities are equal. Women are 42 percentage points more likely to display this opinion than men. 22

24 As our previous results indicate, male economists appear more likely to prefer market solutions to economic problems than do women. We are interested to learn if male economists are also more likely to see differences in outcomes, such as variations in wages between men and women in the U.S., as an outcome of choice or the free market versus other institutional factors. When we ask if respondents agree with the statement that the gender wage gap is largely explained by differences in human capital and voluntary occupational choices, the results of our survey indicate a significant difference between male and female economists on this issue. When comparing economists with the same employment status, who received their Ph.D. in the United States in the same decade, we find that the mean response of males is more than a full standard deviation higher than that of females. On average, women disagree that the gender wage gap is due to choices and productivity difference; the average man is more likely to agree. The probability of disagreeing or strongly disagreeing with the statement is 51 percentage points higher for female economists than for male economists. Rent seeking, male preferences for market solutions, and women s views on gender equity may be reflected in the gender gap in responses. Because one of the most visible and controversial solutions to the problem of discrimination and equal opportunity has been affirmative action, we ask two questions about affirmative action. Question 26 asks whether affirmative action programs for women are a good idea and Question 27 asks whether affirmative action programs for African-Americans are a good idea. Women s average responses to both questions are statistically significantly more favorable to the use of Affirmative Action than men s;; however, the magnitudes of the gender differences in opinions are only about one third of a standard deviation. 23

25 Finally, we examine views on equal opportunity in the economics profession and examine perceptions of equal opportunity for women as graduate students and as faculty. The question about graduate education asks respondents if they believe that opportunities for women graduate students in economics in the U.S. currently favor men or women. After controlling for degree vintage and type of employment, we find that the mean response of female economists is closer to the favors men a bit more choice and the mean response of male economists is closer to the is approximately equal for men and women choice. According to the point estimates, 52 percent of women and 15 percent of men believe graduate education favors men either a bit more or a lot more. On the issue of gender equality as it relates to women faculty in economics we ask respondents if they believe that opportunities for economics faculty in the U.S. currently favor men or women or neither sex. We find that the average female economist believes opportunities favor men more than women and this view reflects a full standard deviation difference in opinion from the average male economist. The average male response indicates a belief that current opportunities for economics faculty in the U.S. are equal, favoring neither men nor women. Women are 55 percentage points more likely to believe that opportunities for economics faculty favor men a bit or a lot more than women. On four of the five questions concerning gender equality, male and female members of the AEA have reached the opposite conclusion, with statistically significant and large differences of opinion. More than 50 percent of female economists believe that opportunities in the economics profession or in the job market in general favor men whereas more than 75 percent of male economists believe they are either equal or favor women. 24

26 IV. CONCLUSION Our survey of AEA economists provides a snapshot picture of the views of economists on some of the core precepts and approaches to the study of economics and a variety of important and topical policy issues facing policy-makers today. To isolate the difference in views on overall topics and specific policy issues, our statistical analysis controls for degree vintage and type of employment. We estimate the mean gender gap in opinions using system GLS methods that incorporate the estimated relationship among unobservable factors that form opinions on related topics. It is important to note that on the core precepts and methodological approaches to economics we found no significant difference in average opinion between male and female economists after controlling for the decade the Ph.D. was granted and type of employment. Statistical tests based on GLS systems composed of individual responses support the conclusion that male and female economists share the view that individuals are utility-maximizers, human wants are unlimited, and that mathematical modeling should be an important part of economic modeling. Moreover, male and female economists also embrace the notion that research on household decision-making should include analysis of intra-household decision-making. In other words, male and female economists appear to start with the same assumptions about how people are expected to behave and how to approach the study of economics. Despite their similar academic training and shared views about core precepts and methodology in economics, our study reveals differences in the views of male and female economists in their support of market solutions versus government intervention, in their interpretation of economic outcomes, and their views toward so-called compassion issues. 25

27 These results show a greater willingness among male AEA members to rely on markets and see problems from interfering with market processes views that emerge when questions are put in general terms and more clearly when they are put in terms of specific policy issues. Although male and female economists agree that market solutions are most efficient, on the issue of government regulation in general, male AEA economists, on average, see government regulation in both the EU and the U.S. as more excessive than do female economists. Moreover, male economists express greater support than women economists for reducing tariffs and express greater opposition to linking trade openness to partners labor policies and greater opposition to mandating that employers provide health insurance. Male economists report greater support for the use of vouchers in education and greater agreement that the competition provided by a Wal-Mart store benefits society and are far more likely to support drilling in the Arctic National Wildlife Refuge. Not surprisingly, male economists are more likely to see the costs associated with government intervention. For example, male economists are more likely to see a loss of employment from raising the minimum wage. Another area where we find some consensus is on the environment. Here the average response of both male and female AEA economists indicates support for increases in energy taxes and taxes on greenhouse emissions and recommend that ethanol subsidies be eliminated. It is important to recognize that value differences associated with environmental issues do emerge when we consider other areas of market solutions versus governmental solutions. Although the mean responses of both male and female AEA economists suggest that policies in the U.S. currently do too much to favor economic growth at the expense of environmental quality, the average female AEA economist disagrees more with drilling for oil in the Arctic National Wildlife Refuge than the average male. 26