Labour Force Status of Aboriginal Canadians: Do human and social capital make a difference? Belayet Hossain & Laura Lamb Department of Economics Thompson Rivers University Kamloops, British Columbia, Canada Abstract The labour force status of Aboriginal Canadians is analysed with a focus on determining the role of an expanded definition of capital to include the human capital component of health status and social capital. The dataset of the 2006 Aboriginal Peoples Survey is examined using multivariate analysis to study labour force status. The results suggest that social capital and human capital, measured by education and health status, among other socio-demographic factors, are significant determinants of labour force status among Aboriginal Canadians. The findings have implications for public policy. JEL: J21, J48, O12 JEL: J15 J21 J18 O15 ?? Introduction Aboriginal Canadians continue to live at a lower level of economic development than nonAboriginal Canadians. While many socio-economic and institutional factors are believed to be determinants of the economic plight of Aboriginal Canadians, labour force status is consistently considered to be one of the critical factors. The objective of the current research is to identify the socio-economic and demographic determinants of the labour force status of Aboriginal Canadians. Most past studies have compared employment and labour force participation between Aborginal and non-Aboriginal Canadians (Sharpe, Arsenault, and Lapointe, 2007; Walters, White& Maxim, 2004; White, Maxim, Gyimah, 2003; Kuhn and Sweetman, 2002), with the exception of Drost (1994), while the current study focuses only on Aboriginal Canadians. The most common socio-economic factors examined in past labour market research include marital status, the presence of children, age, region of residence, and educational attainment. All of the past studies analysed gender differences (Walters, White& Maxim, 2004; Kuhn and Sweetman, 2002; Drost, 1994) except for White, Maxim, Gyimah (2003) who restricted their analysis to females. Some other variables considered in past studies on Aboriginal labour force participation include language, parental education, intermarriage, social assistance dependency, and other household income. In addition to many of the variables analysed in past studies on the labour force status of Aboriginal Canadians, this research considers an expanded definition of human capital to include the human capital component of health status. This research examines human capital with both educational attainment and health status, as has been done in some past labour supply research (Cai, 2010; Stephens, 2010; Latif, 2006), but not of the Canadian Aboriginal labour supply. Walters, White, & Maxim (2004) speculate on the importance of health status as a significant explanatory variable but are not able to include it in their analysis due to lack of available data. Within a human capital framework, health status is pertinent in that it affects potential labour market productivity and thus labour force participation (Mankiw & Scarth, 2011; Grossman, 1972). The role of social capital has also not been empirically anlaysed in past research on the labour force status of the Aboriginal Canadian population. White, Maxim and Gyimah (2003) mention a possible link between labour force activity and social capital in the community but do not include it in their empirical model. Social capital has been identified as a significant factor for employment status among all Canadians (Matthews, Pendakur, and Young, 2009) and Canadian immigrants (Grenier & Xue, 2009). Conceptually, the networks and social relations indicative of social capital are viewed as instrumental for labour market success (Woolcock, 2001). Human and social capital can be viewed as complements, although human capital is described as existing in individuals while social capital exists in relationships. It is argued that human capital 1 and social capital work together to increase the probability of successful employment outcomes (Woolcock, 2001). It is hypothesized that Aboriginal Canadians with higher levels of human capital, as indicated by educational attainment and health status, and a high level of social capital are more likely to have full-time employment and less likely to not be in the labour force. Data from the 2006 Aboriginal Peoples Survey is used to study the determinants of Aboriginal Canadian employment status with use of a multinomial probit model. This research is expected to make a contribution to the literature on Aboriginal economic development and Canadian economic growth. The relevance to the study of Aboriginal economic development is grounded in the connection between labour force status and economic development and well-being. Employment is seen as one of the most fundamental ways people participate in society, and the basis of self-respect and autonomy (Mendelson, 2004). In addition, measures of well-being, such as the Community Well-Being (CWB) index often incorporate labour force indicators. The CWB index, developed by Beavon and Cooke (2003) in order to analyze the disparity between Aboriginal and non-Aboriginal Canadians, is based on five indicators of socio-economic well-being which include measures of labour force activity such as participation and employment rates, among other variablesi (McHardy and O’Sullivan, 2004). The current research is relevant to the economic growth of Canada because of the persistent gap in employment experiences between Aboriginal and non-Aboriginal Canadians. While the overall employment rate fell in Canada, as it did for most countries, after the onset of recent global economic downturn that began in 2008, the employment rate fell further and for a longer duration for the Aboriginal population than it did for the non-Aboriginal population. From 2008 to 2010, the gap between the two labour force populations broadened in regard to participation rates, employment rates, and unemployment rates (Statistics Canada, 2011). This increasing gap is of concern given the relationship between employment and economic development and well-being. The identification and resolution of the barriers surrounding the employment gap is important in that it may impact Canada’s future economic growth, given the anticipated labour shortage due to demographic changes in the non-Aboriginal population. The Aboriginal population is younger with population growth projections greater than the total Canadian population. The Aboriginal population accounted for 3.9% of the Canadian population in 2006, and is expected to represent between 4.0% and 5.3% by 2031 (Statistics Canada, 2011). Closing the employment gap is a potential solution to current and expected skills shortages in some areas of the labour force. The remainder of this paper is organized as follows. The methodology section includes a description of the model, econometric techniques for analysis, the data and variables. Then the 2 results of the empirical analysis are explained, followed by a discussion of the results, policy implications and the conclusion. Methodology Multivariate analysis is used to test the hypothesis and examine the determinants of labour force status among Aboriginal Canadians. The dependent variable, labour force status, is an unordered categorical variable with four mutually exclusive groups, listed as follows: not in the labour force (NLF), unemployment (UEM), part-time employment (PTE) and full-time employmentii (FTE). While both the multinomial logit and multinomial probit models are suitable for analysis, the multinomial probit model is chosen because it allows for relaxation of the independence of irrelevant alternatives (IIA) assumption, which may be too restrictive for the unordered choice model of four employment status outcomes iii(Cameron & Trivedi, 2009; Greene, 2000). The structural equations of the multinomial probit model are as follows: (1) ๐๐๐∗ = ๐๐/ ๐ฝ๐ + ๐๐๐ ; Where ๐๐๐∗ is the latent value of the response variable, which is employment status; ๐๐ is the vector of explanatory variables; ๐ฝ๐ is the vector of parameters to be estimated; j = 1, 2, 3 ….k categories; i = 1, 2,… N sample. ๐๐ ’s follows a multivariate normal distribution with covariance matrix Σ, where Σ is not restricted to be a diagonal matrix, allowing errors to be correlated to each other. The category j is chosen if ๐๐๐∗ is the highest, that is if ๐๐ = ๐ ๐๐ ๐๐๐∗ ๐ ๐๐ ๐๐๐∗ = max (๐๐1∗ , ๐๐2∗ , … . ๐๐๐∗ ) = { 0 ๐๐กโ๐๐๐ค๐๐ ๐ The probability that category j is chosen is stated as ∗ ๐(๐๐ = ๐|๐๐ ) = ๐(๐๐๐∗ > ๐๐1∗ … … . . ; ๐๐๐∗ > ๐๐๐−1 ; ๐๐๐∗ > ๐๐๐∗ ) = ๐(๐๐๐ − ๐๐1 ) > ๐ / (๐ฝ1 − ๐ฝ๐ ) … … . (๐๐๐ − ๐๐๐ ) > ๐ / (๐ฝ๐ − ๐ฝ๐ )) ′ It is noted that only the differences between the ๐๐๐∗ ๐ are identified and hence a reference category must be assigned. 3 The likelihood function for multinomial probit is ๐ ๐ ๐ฟ = ∏∗ ∏ ๐(๐๐∗ = ๐)๐๐๐ ๐=1 ๐=1 1 ๐๐ ๐๐∗ = ๐ Where ๐๐๐ = { 0 ๐๐ ๐๐∗ ± ๐ Taking the log of both sides of the above function, we have the following log likelihood function: ๐ ๐ ๐ฟ(๐ฝ, Σ) = ∑ ∑ ๐๐๐ ๐๐๐๐(๐๐๐ = 1|๐๐ , ๐ฝ, ๐ด) ๐=1 ๐=1 The above log likelihood function is to be maximized with respect to the coefficients, variances and co-variances for the multinomial probit model. Data and variables This study uses data from the 2006 Aboriginal Peoples Survey (APS), conducted by Statistics Canada (2006) in the fall of 2006 through the spring of 2007. The Aboriginal Peoples Survey provides data on the social and economic conditions of First Nations people, over the age of six years, living on reserve as well as off reserve, Métis and Inuitiv. Only the data for those in the age range of 20 and older are used for the current analysis in order to omit the group most likely to be in school full-time. After the variables for the model are identified, 18,165 observations are used in the analysis. Labour force status is the dependent variable with four categories (NLF, UEM, PTE and FTE), as outlined above. The focus of the study is to examine the role of human and social capital on the employment status of Aboriginal Canadians. As previously mentioned, human capital consists of educational attainment and health status. Five categorical variables are specified to measure the impact of education on employment status (Educ1, Educ2, Educ3, Educ4 and Educ5). The specification of each of these five variables is described in Table 1. Following Stephens (2010) and Deschryvere (2005), the status of health is specified using a self-reported index. If a respondent reports good or excellent health, the health variable (HEALTH) is coded 1, if fair or poor health, the variable is coded 0. An indicator variable is developed to measure social capital (Sock) based on responses to three questions about networks and social relations on the APS 2006 survey. They are as follows: How often is this available to you? (1) Someone you can count on when you need advice, (2) 4 Someone to confide in or talk to about yourself or your problems, and (3) Someone you can count on to listen to you when you need to talk. If the sample member answers either all of the time or most of the time to all three questions, social capital (Sock) is considered to be strong and is coded 1; if the sample member answers either some of the time or almost none of the time to all three questions, then social capital is considered to be weak and is coded 0. Other socio-economic and demographic factors included which may influence employment status are gender, age, marital status, number of child, geographical location, registered Indian status, and presence of income support. All these factors, except for income support, have been examined in previous research. Income support indicates if there is at least one other person living in the household who contributes toward paying the household expenses. Income support can be considered a proxy for spousal income or employment, which has been included in labour supply research (Latif, 2006). The measurement and description of these variables are presented in Table 1. 5 Table 1 Description and specification of all explanatory variables used in the analysis Variable Name Measurement and description of variable Educ1 If respondent has completed high school then 1; otherwise 0.(default is less than high school) Educ2 If respondent has some non-university post-secondary then 1; otherwise 0. Educ3 If respondent has completed non-university post-secondary then 1; otherwise 0. Educ4 If respondent has some university then 1; otherwise 0. Educ5 If respondent has completed university then 1; otherwise 0. Health If respondent self-reports good or excellent health then 1; otherwise 0. Sock If respondent has a strong indicator then 1; otherwise 0. Gender If the respondent is male then 1; if female then zero. Age2 If respondent’s age is between 25 years and 34 years then 1; otherwise 0. (default is age 20 – 24) Age3 If respondent’s age is between 35 years and 44 years then 1; otherwise 0. Age4 If respondent’s age is between 45 years and 54 years then 1; otherwise 0. Age5 If respondent’s age is over 54 years then 1; otherwise 0. Marital If respondent is married then 1; otherwise 0. Child1 If respondent has either one or two children then 1; otherwise zero.(default is no children) Child2 If respondent has more than two children then 1; otherwise zero. Geo2 If respondent resides in a rural area then 1; otherwise zero. (default is urban residence) Geo3 If respondent resides in the arctic then 1; otherwise zero. Status If respondent is has registered Indian status then 1; otherwise zero. Isupport If respondent lives with another person who contributes toward paying the household bills then 1; otherwise zero. 6 Table 2 Percentage distribution of frequencies for labour force outcomes (%)(n = 18,165) NILF Unemployed Part-time Full-time 27.42 5.69 11.30 55.59 Total sample Total 100 Education < High school High school Some p.s. (non-univ) Complete p.s. (non-univ) Some university Complete university 46.58 24.03 28.02 18.55 26.40 14.65 7.26 6.26 5.92 5.84 4.37 1.96 9.99 11.65 11.94 11.12 18.40 10.48 36.17 15.12 54.13 64.49 50.83 72.91 100 100 100 100 100 100 Health Good/excellent Fair/poor Strong Weak Male Female 20-24 25-34 35-44 45-54 55+ Married Non-married 1-2 children >2 children Urban Rural Arctic Registered Not registered Other maintainers None 21.82 55.17 24.34 37.09 21.85 31.77 24.94 20.94 15.57 20.18 58.97 24.81 31.23 22.83 25.04 24.95 32.00 30.95 29.90 26.12 19.79 33.51 5.90 4.66 4.93 8.10 6.72 4.89 10.14 6.53 6.28 4.96 2.28 4.81 6.99 5.68 6.36 4.54 6.60 11.50 6.47 5.29 5.07 6.19 11.50 10.31 11.15 11.78 8.05 13.84 17.37 11.83 11.19 10.60 8.15 10.75 12.11 11.07 13.09 11.44 11.23 10.43 11.17 11.37 11.28 11.32 60.78 29.87 59.58 43.03 63.38 49.50 47.55 60.71 66.95 64.26 30.60 59.63 49.66 60.42 55.50 59.21 50.17 47.13 52.46 57.23 63.86 48.98 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Social Capital Gender Age Marital Children Location Status Support Source: Statistics Canada (2009) Aboriginal Peoples Survey, 2006. Public use microdata file (adults). 7 Results Descriptive statistics of the frequency distributions of all variables by employment status are illustrated in Table 2. The labour force consists of 73% of the sample members with 56% being employed full-time, 11% part-time and close to 6% unemployed. Full-time employment increases to 64% and 73% for those who completed postsecondary non-university and university, respectively. Almost two-thirds (61%) of sample members reporting good health are employed full-time compared to 30% reporting fair to poor health. The majority (61%) of those with strong social capital work full-time while less than half (43%) with weak social capital work full-time, and 37% with weak social capital are not in the labour force. The frequency distributions indicate that 63% of males are employed full-time compared to almost 50% of females, while 14% of females work part-time compared to only 8% of males. Over 60% of sample members between the ages of 25 and 55 are employed full-time and almost 59% of those over age 55 are not in the labour force. The proportion of married sample members employed full-time (60%) is greater than the proportion of non-married (50%) sample members employed full-time. Close to 60% of sample members living in urban areas are employed fulltime compared to 50% in rural areas and 47% in the arctic. Over half of those with registered Indian status (52%) are employed full-time and 30% are not in the labour force compared to 57% without status being employed full-time and 26% not being in the labour force. Close to two-thirds (64%) of sample members with income support are employed full-time compared to 49% of those without. One-third (33%) of those without income support are not in the labour force while only 20% of those with household maintainers are not in the labour force. Overall, the specifications of the multinomial probit model are found to be robust, as evidenced by the Wald statistic. The Wald test of joint significance reveals that all explanatory variables are jointly significant at the 1% level across the four categories of labour force status with the exception of being married (Marital) and having one or two children (Child1). The marginal effects of the independent variables of the multinomial probit estimation are presented in Table 3 and the predicted probabilities in Table 4v. The results of the base suggest that a single-female respondent, between the ages of 20 and 24, with a weak level of social capital, living in an urban area without registered status, having less than high school education, and fair to poor health is predicted to have 59% probability of being employed full-time. Her expected probability of not being in the labour force is 23%, of being unemployed is 4.7%, and of being employed part-time is 13%, as illustrated in Table 4. These are considered to be the base probabilities without any special characteristics. The probability of having full-time employment is expected to be 71% for a respondent who has completed high school education when all other characteristics remain constant. In other 8 words, the probability of attaining full-time employment increased by 12% for a person with a high school diploma compared to one without high school completion. Similarly, for those who complete non-university and university post-secondary education, their probability of attaining full-time employment further increases to 77% and 79%, respectively. On the other hand, for those who have some non-university and university post-secondary education, but not completion, their probability of being employed full-time increases less, to 68% and 63%, respectively. Thus, the completion of post-secondary education, regardless of university or non-university, appears to be important for attaining full-time employment for Aboriginal Canadians. The results suggest that health is the most important determinant of full-time employment. A person with excellent or good health has 84% probability of being employed full-time. That is, a healthy person has a 25% higher probability of being employed full-time compared to one with only fair to poor health. Social capital measured in terms of access to networks and social relations, is also found to be a significant factor affecting the probability of full-time employment. The probability of full-time employment increases from 59% to 68% (a rise of 9%) for a person with strong social capital compared to one with weak social capital. Among the socio-demographic variables affecting full-time employment, gender, age, the presence of more than two children, place of residence, status, and income support are found to be statistically significant. Males and those between the ages of 25 and 54 have a higher probability of being employed full-time than females and those between the ages of 20 and 24. Those who have more than two children are predicted to have a lower probability than those who do not have children. And a respondent living in a rural area has a lower probability of being employed full-time than those living in an urban area. 9 Table 3 Multinomial probit estimates (marginal effects) of employment status Variables Human Capital Educ1 Educ2 Educ3 Educ4 Educ5 Health Soc Capital Socio-demographic Gender Age2 Age3 Age4 Age5 Marital Child1 Child2 Geo2 Geo3 Status Isupport Wald Statistics Log Pseudo-Likelihood NLF UEM PTE FTE -0.101*** (0.013) -0.07*** (0.014) -0.171*** (0.011) -0.081*** (0.017) -0.16*** (0.012) -0.259*** (0.017) -0.056*** (0.013) -0.004 (0.006) -0.008 (0.007) -0.007 (0.005) -0.022*** (0.006) -0.03*** (0.006) 0.014*** (0.005) -0.015*** (0.005) -0.009 (0.013) -0.008 (0.014) -0.006 (0.011) 0.066*** (0.021) -0.006 (0.014) -0.003 (0.011) -0.02** (0.01) 0.114*** (0.017) 0.086*** (0.02) 0.183*** (0.015) 0.036 (0.024) 0.196*** (0.018) 0.245*** (0.017) 0.091*** (0.014) -0.127*** (0.009) -0.046*** (0.017) -0.117*** (0.016) -0.088*** (0.016) 0.242*** (0.023) -0.011 (0.011) -0.003 (0.011) 0.066*** (0.02) 0.04*** (0.01) -0.012 (0.015) 0.028*** (0.01) -0.091*** (0.01) 2284.02 -734073.07 0.008* (0.004) -0.018*** (0.006) -0.024*** (0.006) -0.032*** (0.005) -0.041*** (0.005) -0.008 (0.005) 0.011** (0.005) -0.002 (0.007) 0.02*** (0.005) 0.062*** (0.011) 0.016*** (0.005) -0.009** (0.005) -0.071*** (0.007) -0.056*** (0.012) -0.065*** (0.012) -0.069*** (0.011) -0.064*** (0.012) 0.004 (0.009) -0.004 (0.009) 0.019 (0.015) 0.001 (0.008) -0.029*** (0.011) -0.013* (0.008) -0.005 (0.008) 0.189*** (0.011) 0.12*** (0.02) 0.206*** (0.019) 0.189*** (0.019) -0.136*** (0.024) 0.015 (0.013) -0.003 (0.013) -0.083*** (0.022) -0.061*** (0.011) -0.021 (0.019) -0.03*** (0.012) 0.105*** (0.012) Note 1: ***, ** and * indicate the level of significance at 1%, 5% and 10% respectively Note2: Wald test of joint significance indicates all variables to be jointly significant at the .001 level, except for marital (p=.296) and child1 (p=.230). 10 Table 4 Predicted probabilities of employment status with selected characteristics Variables NLF UEM PTE FTE Base 0.231 0.047 0.130 0.591 Human Capital Educ1 0.130 0.044 0.121 0.705 Educ2 0.161 0.040 0.122 0.677 Educ3 0.060 0.041 0.124 0.774 Educ4 0.150 0.026 0.196 0.627 Educ5 0.071 0.018 0.124 0.787 Health -0.028 0.062 0.127 0.836 Total 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Soc Capital 0.175 0.033 0.110 0.682 1.00 Sociodemographic Gender Age2 Age3 Age4 Age5 Marital Child1 Child2 Geo2 Geo3 Status Isupport 0.104 0.185 0.114 0.143 0.473 0.220 0.228 0.297 0.271 0.219 0.259 0.140 0.056 0.030 0.024 0.016 0.007 0.039 0.059 0.046 0.068 0.110 0.064 0.039 0.059 0.074 0.065 0.061 0.066 0.134 0.126 0.149 0.131 0.101 0.117 0.125 0.780 0.711 0.797 0.780 0.455 0.606 0.588 0.508 0.530 0.570 0.561 0.696 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Note: Predicted probabilities are estimated using base values and the marginal effects in Table 3 Note: All variables are jointly significant except for Marital and Child1. The results suggest that an Aboriginal person with registered Indian status has a lower probability of attaining full-time employment than one without status. The presence of other household members who contribute to paying household expenses increases the probability of full-time employment. The likelihood of being ‘not in the labour force’ significantly decreases with an increase in the level of education. The base level suggests those without high school diplomas have a 23% probability of not being in the labour force, while those with high school diplomas experience a lower probability of 13%. The probability decreases further to 6% and 7%, respectively, for those, who completed either non-university post-secondary or university. Similarly, access to 11 social capital also helps to reduce the probability of being ‘not in the labour force’. The results suggest that the possession of strong social capital reduces the probability of not being in the labour force from 23% to 18%. Thus, both human and social capital may increase labour force participation substantially. Labour force participation is higher for males than for females. The probability of not being in the labour force is higher for those, who have two or more children (30%) compared to those, who do not have children. It is also higher for those, who live in rural areas (27%) compared to those in urban areas. Likewise, it is higher for those, who hold registered Indian status (26%) compared to those, who do not. The presence of other earning members in the household decreases the likelihood of being ‘not in the labour force’ to 14% as opposed to the base probability of 23%. The likelihood of being unemployed decreases with higher levels of education as well as with the possession of strong social capital. The completion of university reduces the probability of being unemployed from 4% to 2%. Likewise, the possession of strong social capital reduces the expected probability of being unemployed from 4.7% to 3% when all else remains constant. Social capital also reduces the expected probability of being employed part-time when all other factors are held constant. Discussion, policy implications, and conclusion The results provide support for the hypothesis that Aboriginal Canadians with higher levels of human capital, as indicated by educational attainment and good health status, and a higher level of social capital are more likely to have full-time employment and less likely to not be in the labour force. The significant role of educational attainment in Aboriginal labour force outcomes is not surprising given the evidence in not only the Aboriginal Canadian literature (Walters, White, Maxim, 2004; White, Maxim, Gyimah, 2003; Kuhn and Sweetman, 2002; Drost ,1994), but in most all labour force literature. Although neither health status nor social capital has been included in past empirical models of the Aborginal Canadian labour force, the significance of both are supported by the broader literature. Labour force research including a measure of health status found it to be significant (Cai, 2010; Stephens, 2010; Latif, 2006). Past research on employment status in Canada found a significant relationship between social capital and labour force participation (Grenier & Xue, 2009; Matthews, Pendakur, and Young, 2009), lending support to the current findings. Grenier & Xue (2009) examined the role of social capital with multiple variables such as the number of sources of meeting friends, the frequency of contact with friends, and participation in organizations, while Matthews, Pendacur, and Young (2009) assessed the role of social capital in employment outcomes with several 12 measures of individual networks, trust, and involvement in organizations. It is acknowledged that the measurement of social capital is challenging in the current research with the APS 2006 survey data and the limited number of variables available to measure networks and social relations. The findings of the socio-demographic variables are generally supported by past research on Aboriginal Canadian labour force outcomes (Walters, White& Maxim, 2004; White, Maxim, Gyimah, 2003; Kuhn and Sweetman, 2002; Drost, 1994). In regard to variables specific to the Aboriginal population, past studies confirm the negative impact of registered status on employment (Hull, 2005; Walters, White, Maxim, 2004; White, Maxim, Gyimah, 2003). It has been argued that status Indians are more likely to live on reserve than non-status Indians, and subsequently experience labour market barriers due to remoteness (Walters, White, Maxim, 2004), although recent findings show similar employment outcomes for status Indians living in urban areas, including urban reserves (Pendakur & Pendakur, 2011). The geographic variable of residence in the arctic has not been included in many Aboriginal Canadian studies because the Canadian Arctic is comprised of many very small communities and unique labour market conditions compared to the rest of Canadian Aborginal people (Pendakur & Pendakur, 2011). The significance of the variable implies that those in the arctic have a lower predicted probability of being employed, a higher probability of being unemployed, and a higher probability of not being in the labour force than those who live in urban areas. The results of income support (Isupport) are somewhat unexpected as it was thought that having other household members contribute to paying the household bills may be a disincentive for employment. In fact, the results imply that having other household maintainers may be an incentive for employment, perhaps indicating some type of employment culture within the household. Future research in this area might consider other variables identified in past studies on the Aboriginal Canadian labour force but not included in this study due to lack of available data, such as intermarriage, province of residence, and parental education (Walters, White, Maxim,2004; Kuhn and Sweetman, 2002; Drost ,1994). Additional measures of social capital are recommended in order to substantiate the current findings. Policy implications of the results suggest a greater recognition for the role of good health and social capital in policies developed to improve labour force participation of the Aboriginal Canadian population. While it is widely recognized that poor health is pervasive in many Aboriginal Canadian communities and needs to be addressed, the link to labour force participation is often not emphasized. A greater degree of inclusion of the Aboriginal population in Canadian society may lead to higher levels of social capital leading to the creation of networks and social relations necessary to improve labour force participation. In addition, 13 future policy development for Aboriginal labour force participation addressing all the relevant determinants is expected to have a positive impact on the economic growth of the Canadian economy by providing needed labour and by alleviating some pressure on social safety nets for those not in the labour force. This research makes a contribution to the literature on Aboriginal labour force participation and economic development by identifying the importance of including measures of health status and social capital in labour supply research. 14 References Beavon, D., and Cooke, M. (2003) “An Application of the United Nations Human Development Index to Registered Indians in Canada” in Aboriginal Conditions, ed. J. White, D. Beavon, and P. Maxim. Vancouver: UBC Press. Cameron, A. Colin & Pravin K. Trivedi (2009) Microeconomics Using Stata, Stata Press. Cai, Lixin (2010) “The relationship between health and labour force participation: Evidence form a panel data simulation equation model”, 17(1):77-90. Deschryvere, M. (2005) “Labour Force Behaviour of Men and Women in Elderly TwoHouseholds: Evidence from EU Countries” Social Welfare Policies, ENEPRI Research Report, Brussels. Drost, Helmar (1994) “Vocational Training and Unemployment: The Case of Canadian Aboriginals” Canadian Public Policy, 20(1):52-65. Greene, W. (2000) Econometric Analysis, Prentice Hall, New Jersey. Grenier, Gillies and Li Xue (2009) “Duration of Access of Canadian Immigrants to the First Job in Intended Occupation” University of Ottawa, Department of Economics, Working Papers: 0908E. Grossman, Michael (1972) “On the concept of health capital and the demand for health” Journal of Political Economy, 80(2): 223-255. Hull, Jeremy (2005) “Post-Secondary Education and Labour Market Outcomes: Canada, 2001” Indian Affairs and Northern Development Canada. Catalogue No. R2-399/2001E-PDF, www.ainc-inac.gc.ca. Kuhn, Peter and Arthur Sweetman (2002) “Aboriginals as Unwilling Immigrants: Contact, Assimilation and Labour Market Outcomes”, 15(2):331-355. Latif, Ehsan (2006) “Labour Supply Effects of Informal Caregiving in Canada” Canadian Public Policy, 32(4):413-429. Mankiw, N. Gregory and William Scarth (2011) Macroeconomics, Worth Publishers. Matthews, Ralph, Ravi Pendakur, and Nathan Young (2009) “Social Capital, Labour Markets, and Job-Finding in Urban and Rural Regions: comparing paths to employment in prosperous cities and stressed rural communities in Canada” The Sociological Review, 57(2): 306-330. 15 McHardy, Mindy and Erin O’Sullivan (2004) “First Nations Community Well-Being in Canada: The Community Well-Being Index, 2001” Indian and Northern Affairs Canada. Catalogue No. 0662-38016-9, www.ainc-inac.gc.ca. Mendelson, Michael (2004) “Aboriginal People in Canada’s Labour Market: Work and Unemployment, Today and Tomorrow”. Caledon Institute of Social Policy, Ottawa, ISBN 155382-090-8 Pendakur Krishna and Ravi Pendakur (2011) “Aboriginal Income Disparity in Canada” Canadian Public Policy, 37(1):61-83. Sharpe, Andrew, Jean-Francois Arsenault, and Simon Lapointe (2007) “The Potential Contribution of Aboriginal Canadians to Labour Force, Employment, Productivity and Output Growth in Canada, 2001- 2017” Centre for the living Study of Living Standards, CSLS Research Report No. 2007-04. Statistics Canada (2009) Aboriginal Peoples Survey, 2006. Public use microdata file (Adults). Statistics Canada (2011) “Study: Aboriginal people and the labour market” The Daily, Wednesday, November 23, 2011, www.statcan.gc.ca./daily. Statistics Canada (2011) “Population projections by Aboriginal identity in Canada” The Daily, Wednesday, December 7, 2011, www.statcan.gc.ca./daily. Stephens, Benjamin J. (2010) “The Determinants of Labour Force Status Among Indigenous Australians” The University of Western Australian Discussion Paper 10.11. Walters, David, Jerry White, and Paul Maxim (2004) “Does Postsecondary Education Benefit Aboriginal Canadians? An Examination of Earnings and Employment Outcomes for Recent Aboriginal Graduates” Canadian Public Policy, 30(3):283-301. White, Jerry, Paul Maxim and Stephen O. Gyimah (2003) “Labour Force Activity of Women in Canada: A Comparative Analysis of Aboriginal and Non-Aboriginal Women” Canadian Review of Sociology, 40(4):391-415. Woolcock, Michael (2001) “The Place of Social Capital in Understanding Social and Economic Outcomes” Canadian Journal of Policy Research, 2(1): 11-17. 16 i The other four indicators are education, income, housing, and life expectancy. Full-time employment is defined as working 30 hours or more per week. iii The Hausman specification test is typically used to test for independence of irrelevant alternatives (IIA), however, it cannot be used to test the estimates of the current model due to the robust estimates of the VCE (Cameron & Trivedi, 2009). Thus, we estimated the model with both the multinomial logit and multinomial probit model and found that the results are close to identical. iv In Canada, Aboriginal people consist of First Nations, Inuit, and Métis. v Predicted probabilities are calculated using the marginal effects in Table 3 and the values of the base in the first row of Table 4. The marginal effects show the change in probability of being in the respective labour force category associated with the respective explanatory variable. ii 17