Wesley Sze ECON 495 9 November 2010 Research Proposal: Is Cultural Diversity Good for the Economy? 1 Research Question I would like to examine the economic consequences of increased cultural diversity in Canadian metropolitan areas. More specifically, I would like to answer the question: what is the economic value of cultural diversity in Canada s metropolitan areas? Cultural diversity is a broad topic that crosses several disciplines, including sociology, geography, and political science. Although there is no universally agreed upon definition of cultural diversity, I will define it as the degree of non-homogeneity in the ethnic characteristics of a population (although I do acknowledge that cultural identity is made up of more than just ethnicity). I want to abstract away from the social and political implications of diversity and instead identify and quantify the economic consequences (positive or negative) that stem from cultural diversity. 2 Motivation: Issue of Interest Given the significant changes in cultural diversity over the past half-century in North America, Europe, and Australia, both the popular media and academia have become interested in better understanding the social and economic consequences of a rapid transformation of population. Issues of diversity, ethnicity, and multiculturalism are not only important in public policy debates, but also from an academic perspective: economic insights can help to better understand the factors contributing to productivity gains and overall economic well-being. A large empirical literature exists on the impact of low-skilled immigration on native-born wages in the short and medium run. Notable economists who have contributed to this particular subject include Borjas (1994, 1995), Card (1990, 2001, 2005) and Card & Di Nardo (2000). Although no definitive conclusion has been reached, these studies suggest perhaps only a very small temporary and relative negative effect of immigration on the labour market outcomes of natives. My research, however, will take on a slightly different approach. Instead of looking at short-run effects of immigration on segmented labour markets by skills, I want to investigate the long-run equilibrium effects of changes in cultural diversity as a whole. My aim is to quantify the economic value cultural diversity brings to Canadians, as expressed through long-run productivity gains. Relevant to my topic of interest, diversity has been addressed in many other overlapping disciplines and contexts. The urban studies literature has identified diversity as a critical element relating to the development of successful cities, albeit in a qualitative and non-technical approach. Diversity is considered to be essential to the growth associated with the world s global cities, contributing to a greater variety of consumer goods, services, and productive inputs that in turn result in 1
more productive and innovative societies (Florida 2002). In a different context, Lazear (1999) identifies potential net benefits stemming from the diversity of members within global firms. At the institutional level, on the other hand, diversity has been found to be correlated with more fragmentation and lower provision of public goods (Easterly and Levine, 1997; Alesina and La Ferrara, 2005; Habyarimana et al., 2007). My research will be based upon the work of Ottaviano and Peri (2005), where the authors evaluate the long-term productivity effects of diversity across metropolitan areas in the United States. They find a statistically significant correlation between increased cultural diversity and a dominant productivity effect, which they argue is consistent with a causal effect. One of the goals for my research will be to see to what extent these results hold in a Canadian context. 3 Underlying Economic Theory I have identified two main economic interpretations for how cultural diversity may conceivably impact economic productivity and utility. The first is the complementary skill set hypothesis: workers from different cultural backgrounds may bring skills and abilities that are complementary to one another. According to this hypothesis, diversity can be considered as a factor of production contributing to overall productivity. Cities with more diverse populations will have a greater mix of skill sets, creativity, and innovative ability that could result in a positive productivity effect. This positive externality effect that immigrants and other minority groups bring to a city may be one factor that contributes to the economic success of a city. The second is the consumption variety hypothesis, which posits that individuals enjoy variety in their consumption choices. Cultural diversity in a city can lead to the availability of a more varied set of consumption goods available to individuals living in those areas (e.g. ethnic restaurants, arts/entertainment events). However, a priori, one cannot conclude that a positive effect certainly exists. Although economic theory does provide some justification for a desirable productivity and utility effect, cultural diversity may also lead to counterproductive outcomes. For example, some individuals may incur disutility from living in a diverse environment. This could happen if one s cultural values feel threatened (e.g. when natives associate higher immigration with undercutting of wages and employment) or if intergroup conflict becomes problematic (e.g. racial frictions and the development of so-called ghettos or enclaves). Thus, depending on the relative strength of these effects, there could be either a positive or negative net effect of diversity. To model the mechanisms through which diversity could affect productivity and utility, I refer to Roback s model of local amenities and quality of life (1982). This model has been extended to the context of cultural diversity as a local amenity through the work of Ottaviano and Peri (2005), which I will implement in my own analysis. Through a simple Cobb-Douglas equilibrium model of workers and firms in separate cities, we model diversity as a productivity term that can have a negative or positive effect on output and utility. 1 In the long run equilibrium, we arrive at two conditions that Ottaviano and Peri refer to as the free entry (i.e. no firm has an incentive to exit or enter the market) and the free migration (i.e. no worker has an incentive to move between cities) conditions. Together, these two conditions determine the city s equilibrium wage and rent. This 1 Diversity is incorporated into the Cobb-Douglas equation as a variable affecting a total factor productivity term, A(div), in the Cobb-Douglas equation: A(div) H α L 1 α, where A/ div is positive or negative. 2
simple model will form the theoretical basis for my empirical investigation, where I will estimate equations for mean rents and wages (see Section 5 - Empirical Strategy). 4 Data Sources My research will be heavily based upon data from the Canadian Census. In particular, I will use the 1971, 1976, 1981, 1986, 1991, 1996, 2001, and 2006 Public Use Microdata Files (PUMF) to calculate my key dependent variables of average wage and rent for native-born Canadian males for each Census Metropolitan Area in each time period. I will also use the PUMF to obtain control variables for each individuals education level (hdgree). In total, between the 7 time periods and 23 CMAs, I will be using several hundred thousand individual PUMF observations to tabulate data for approximately 161 observations (23 CMAs 7 time periods). The main unit of observation will be at the CMA level. For my variable of interest, diversity, I propose to use several alternative measures of cultural diversity. The basic standard measure is to use the index of fractionalization, which is computed: M div = 1 (σ i ) 2 t (1) i=1 where σ i is the proportion of the population possessing the characteristic of interest. For example, the for index of fractionalization by country of birth, σ i would be the proportion of the population born in country i. Intuitively, the index would measure the probability that two randomly selected individuals were born in different countries. Because no single measure of diversity can capture the dynamic nature of diversity, I will use several different measures of diversity to proxy for actual diversity: Index of Fractionalization by Country of Birth Index of Fractionalization by Ethnic Minority Status Index of Fractionalization by Language Spoken at Home Index of Fractionalization by Religious Affiliation Share of Foreign Born Share of Ethnic Minorities Index of Fractionalization amongst Minorities by Country of Birth By using alternative measures for quantifying diversity, I hope to capture the full range and many dimensions of cultural diversity. The base measure I will use, however, is the index of fractionalization by country of birth. This has been the traditional measure used in the literature when dealing with significant immigrant effects and has the advantage of being easily computed and unambiguous in its response. I will separately extend my results to include the alternative measures of diversity to check for the robustness of my results. The Canadian Census publishes data for visible minority status (vismin), home language (hlano), mother tongue (mtnno) and place of birth (pob) aggregated to the CMA level. I will use this data to calculate my indices and measures for diversity in each CMA. One unique variant I will include in my empirical analysis is the inclusion of both the share of foreign born and an index of fractionalization amongst minorities. The former variable would capture to 3
value associated with the increase in the presence of the foreign born, regardless of country of origin. The latter variable would measure the additional impact of having more diversity within the minority population. This helps to separate two distinct conceptions of diversity that the index usually aggregates together (in Section 4 - Data Sources, I present a correlation matrix showing the relationship between the two). I have begun some preliminary examination of my data. In Table 1, I have reported the summary statistics for three diversity measures in 23 CMAs in Canada in 2001 based on visible minority status. %M inority is the proportion of the population identifying with a visible minority group. div is the index of fractionalization, and div min is the index of fractionalization amongst visible minorities. There is a wide range in the diversity index, ranging from a low of 0.02 in Trois Rivieres to a high of 0.58 in Toronto. The diversity index amongst minorities also shows a large range, from a low of 0.46 (Kelowna) to a high of 0.89 (Thunder Bay). Table 1: Summary Statistics for Diversity, 2001 CMA % Minority div div min Halifax 0.07 0.13 0.69 St. John 0.03 0.05 0.74 Quebec 0.02 0.03 0.80 Montreal 0.14 0.25 0.83 Trois Rivieres 0.01 0.02 0.76 Sherbrooke 0.03 0.05 0.82 Ottawa 0.14 0.26 0.84 Oshawa 0.07 0.13 0.80 Toronto 0.37 0.58 0.82 Hamilton 0.10 0.19 0.86 Kitchener 0.11 0.20 0.85 London 0.09 0.17 0.88 Windsor 0.13 0.24 0.85 Guelph 0.11 0.20 0.85 Kingston 0.05 0.09 0.85 Thunder Bay 0.02 0.04 0.89 Winnipeg 0.12 0.23 0.80 Regina 0.05 0.10 0.85 Calgary 0.17 0.31 0.82 Edmonton 0.15 0.27 0.83 Vancouver 0.37 0.56 0.71 Victoria 0.09 0.17 0.76 Kelowna 0.18 0.31 0.46 Data source: 2001 Canadian Census. To get a better idea of how these measures of diversity are related, Table 2 is a correlation matrix showing the correlations between the three measures and population. As expected, there is a very high correlation between %Minority and the div (over 90% of the diversity index is based upon the share of minorities). Interestingly, there is no statistically significant correlation between div min and div or %Minority. This suggests that we can use div and div min as two different measures for two very distinct aspects of diversity: the former captures diversity as the share of minorities, 4
whereas the latter captures the richness within the minority population. Table 2: Correlation Matrix % Minority div div min Population % Minority - - - - div 0.994*** - - - div min -0.215-0.199 - - Population 0.737*** 0.724*** 0.076 - *,**,*** different from zero at 10%, 5%, and 1% significance. Data source: 2001 Canadian Census. One challenge in working with my data will be the limited number of final observations. Because the PUMF only provides geographic data at the CMA level, I will be restricted to observations at this level. However, I have made efforts to compensate by choosing the maximum number of time periods available. 5 Empirical Strategy The theoretical model provides a simple estimation strategy to test for productivity and utility effects. The model predicts that there is a dominant positive productivity effect if and only if both wages and rents are positively correlated with diversity. A dominant utility effect exists if and only if wages are negatively correlated with diversity and rents positively correlated. This identification procedure is derived from the free entry and free migration conditions. As such, following the estimation strategy of Giovanni and Peri (2005), I will use the following OLS equations to test for productivity effects: ln(wage c,t ) = β 1 (Controls c,t ) + β 2 (div c,t ) + e c + e t + e c,t (2) ln(rent c,t ) = γ 1 (Controls c,t ) + γ 2 (div c,t ) + ε c + ε t + ε c,t (3) where wage c,t is the mean wage for Canadian born males aged 25-40 living in CMA c at time t and rent c,t is the mean rent paid by Canadian born males aged 25-40 living in CMA c at time t. In the case of wages, Controls is a vector containing some measure to control for average level of education of those workers in CMA c at time t (this will be based on the highest certificate, diploma, or degree response in the PUMF). For the rent specification, Controls will control for average income per capita, and will later include various variables to check for the robustness of my results. div is the key explanatory variable that will proxy for the cultural diversity of the CMA. In my basic model, div will be the index of fractionalization based on country of birth. However, I also plan to extend my empirical analysis to include other alternative measures of diversity. e c and ε c are CMA fixed effects, e t and ε t are time fixed effects, and e c,t and ε c,t are the error terms. If the coefficients on div are significantly positive in both equations, then there is evidence for a positive productivity effect associated with diversity. In addition to running the basic OLS estimation, I will also check for the robustness of my results by adding various controls that could conceivably affect wages or rents but that are not included in my basic specification. This should test for the stability of the result in the basic OLS regression 5
and ensure my results are not simply spurious correlation. For example, I will include variables for the schooling of foreign born (to test for human capital externalities), share born out of province (as a proxy for local amenity or productivity shocks) and employment levels (to control for the relative supply of workers). As part of my robustness checks, I will also consider excluding observations from Toronto and Vancouver in my basic regression to see how much the correlations found between diversity and wages and rents are driven by these two highly diverse CMAs. Lastly, to address potential endogeneity issues (i.e. self selection of more productive workers to more diverse cities), I will use two estimation techniques. First, I will see the predictive power of lagged variables: an increase in diversity in the previous period should have a positive correlation on wages and rents in the current period if a dominant productivity effect exists. At the same time, an increase in wages or rents in the previous period should have a weaker correlation on diversity in the current period. By comparing these lagged results, I hope to be able to identify if diversity is the primary force driving the causal effect. Another technique that has been used by Ottaviano and Peri (2005) is instrumental variable estimation, using estimated immigrant inflows based on historical data and distance from gateway cities as instruments for diversity. I will also consider using similar instruments to isolate the causal effect in my own results. However, one of the problems associated with this methodology is that it raises the standard errors of the results, possibly leading to insignificant results. Also, computing the estimated immigrant inflows to each CMA based on historical data is particularly complex. As I proceed with my research, I will look further into IV estimation to determine whether or not it will be suitable and feasible in my research. 6 Bibliography Alesina, Alberto, Reza Baqir, and William Easterly, Public Goods and Ethnic Divisions, Quarterly Journal of Economics, 114 (1999), 1243-1284. Alesina, Alberto and Eliana La Ferrara, Ethnic Diversity and Economic Performance Journal of Economic Literature, 43 (2005), 762-800. Borjas, George, The Economics of Immigration, Journal of Economic Literature, 32 (1994), 1667-1717. Borjas, George, Ethnicity, Neighborhoods, and Human-Capital Externalities, American Economic Review, 85 (1995), 365-390. Card, David, The Impact of the Mariel Boatlift on the Miami Labor Market, Industrial and Labor Relations Review, 43 (1990): 245-257. Card, David, Immigrant Inflows, Native Outflows, and the Local Market Impacts of Higher Immigration, Journal of Labor Economics, 19 (2001), 22-64. Card, David, Is the New Immigration Really So Bad?, The Economic Journal, 115 (2005), 300-322. Card, David and John Di Nardo, Do Immigrant Inflows Lead to Native Outflows? American 6
Economic Review, 90 (2000): 360-367. Collier, Paul, Ethnicity, Politics, and Economic Performance, Economics and Politics, 12 (2000), 225-245. Edin, Per-Anders, Peter Fredriksson, and Olof Aslund, Ethnic Enclaves and the Economic Success of Immigrants, Quarterly Journal of Economics, 118 (2003), 329-357. Ferraro, Paul and Ronald Cummings, Cultural Diversity, Discrimination, and Economic Outlook, Economic Inquiry, 45 (2007), 217-232. Florida, Richard, The Rise of the Creative Class, Basic Books, New York (2002). Habyarimana, James et al., Why Does Ethnic Diversity Undermine Public Goods Provision? American Political Science Review, 101 (2007), 709-725. Lazear, Edward, Globalisation and the Market for Team-Mates, Economic Journal, 109 (1999): 15-40. Lewin-Epstein, Noah and Moshe Semyonov, Local Labor Markets, Ethnic Segregation, and Income Inequality, Social Forces, 70 (1992): 1101-1119. Ottaviano, Gianmarco and Giovanni Peri, The Economic Value of Cultural Diversity: Evidence from US Cities, Journal of Economic Geography, 6 (2006), 9-44. Page, Scott, The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies, Princeton University Press (2007). Roback, Jennifer, Wages, Rents, and the Quality of Life, Journal of Political Economy, 90 (1982): 1257-1278. 7