ARCHIVED – Social Capital and Employment Entry of Recent Immigrants to Canada

6. Estimation and results

The estimation of the employment probability is undertaken in cross-sectional models as well as longitudinal ones. As discussed before, pooled estimation is subject to biased and inconsistent estimates when observed individual effects are present. However, for the purpose of this paper, the results of cross-sectional regressions (Wave 1, 2, 3 respectively and pooled Waves 1, 2 and 3) are presented in Appendix A for comparison.

The estimated effects of the non-social capital variables are consistent with the theoretical explanations and the findings of other related empirical studies. These results are not discussed in detail in the paper. Several findings related to the non-social capital variables are briefly mentioned here.

It is worth noting that controlling for other characteristics, male immigrants landed in the classes other than family class are less likely to find employment in the initial period of settlement and integration, regardless of the quality or earnings of the employment, which is showed by the negative coefficients of the immigration category dummy variables in the regressions (see Table A.6 in Appendix A). While it looks surprising in terms of the less employment likelihood of skilled worker principal applicants than family class, this result is consistent across models with social capital indicators or without. This finding is inline with relevant research on integration challenges of recent immigrants. Xue (2007) finds that skilled workers, both principal applicants and spouses and dependants, were most likely to report problems finding employment across all three waves of the LSIC. It may be associated with skilled workers’ higher expectations and reservation wages, and problems relating to recognition of foreign qualifications or experience. Family class immigrants may have lower expectations about wages and occupational status, which leads them to more easily settle for an initially unsatisfactory job. Skilled workers may choose to further education or get foreign credentials recognized before settling for an underemployment.

The population group dummy variables coefficients confirm to some extent what the descriptive statistics showed, i.e., Chinese male immigrants seem to be less likely to find employment and immigrants from Philippines perform better (see Table A.6 and A.7 in Appendix A).

6.1 Probability of employment

Conditional on the labour force participation,9 the observations that are not in the labour force (neither employed nor looking for a job) are excluded for the estimations. The analysis separates the sample into male and female groups. Table 1 and Table 2 below show the estimation results for the social capital variables of the employment likelihood of male and female immigrants respectively, from cross sectional models to panel models.  

Table 1: Estimation of employment likelihood of male immigrants in the labour force in the initial 4 years in Canada
Sample coverage: Male immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Ordinary Logit Models (No individual specific effects) Panel Models (With individual specific effects)
Wave 1 (6 months after landing) Wave 2 (2 years after landing) Wave 3 (4 years after landing) Pooled Logit Model (W1, 2 & 3) Fixed-effects Logit Model Random-effects Logit Model GEE Population-averaged Model
Coef. Std.Err. Coef. Std.Err. Coef. Std.Err Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
Social networks  
_Family and relatives
Spouse currently employed 0.392c 0.123 0.599c 0.118 0.148 0.172 0.383c 0.073 0.45c 0.098 0.486c 0.078 0.368c 0.073
Number of relatives in Canada 0.121 0.088 -0.137a 0.080 -0.108 0.087 -0.048 0.049 -1.075c 0.392 -0.076 0.056 -0.046 0.053
Relatives living nearby upon landing 0.115 0.157 0.319b 0.152 0.184 0.167 0.246c 0.090     0.278c 0.107 0.252b 0.100
Frequency of contact with family sponsors 0.222 0.239 0.174 0.220 -0.224 0.215 0.16 0.125 0.594b 0.249 0.389c 0.142 0.203 0.129
_Friends  
Friends living nearby upon landing 0.183 0.112 0.049 0.107 0.125 0.117 0.119a 0.063     0.168b 0.075 0.123a 0.070
Number of sources meeting friends -0.159c 0.040 -0.041 0.030 -0.097b 0.043 -0.109c 0.019 -0.071b 0.030 -0.131c 0.022 -0.103c 0.019
Ethnic diversity of friends 0.84c 0.213 -0.337 0.255 -0.537a 0.274 0.291b 0.121 0.275 0.172 0.283b 0.126 0.283b 0.124
Cohort diversity of friends     0.743c 0.232 0.865c 0.245                
Frequency of contact with friends 0.155 0.200 0.097 0.210 -0.344 0.401 0.235a 0.130 0.206 0.198 0.298b 0.143 0.251a 0.129
_Groups and organizational network  
Number of organizations participated in 0.189 0.229 -0.18 0.185 -0.209 0.200 -0.083 0.115 -0.076 0.172 -0.056 0.123 -0.077 0.115
Ethnic diversity of organizational network 0.19 3.557 3.673 3.016 -0.512 3.285 1.685 1.825 1.324 2.860 1.766 1.996 1.549 1.812
Frequency of activity with organizations -0.251 0.303 -0.211 0.282 0.379 0.307 -0.085 0.164 -0.284 0.238 -0.207 0.177 -0.111 0.167
Volunteered time for organizations -0.086 0.206 -0.087 0.191 0.404a 0.221 0.086 0.113 0.353b 0.162 0.177 0.120 0.122 0.112
_cons -4.51a 2.448 24.726b 10.835 0.274 28.116 -1.067b 0.513     -1.336b 0.594 -1.024a 0.529
No. of observations 2996 3360 3450 9843 3880 9843 9843
No. of groups (for panel models)         1379 3659 3659
Pseudo R2 0.196 0.124 0.1227 0.145     0.186
Percent correctly predicted 0.753 0.789   0.789 0.584 0.780 0.790
Prob > Chi-Square 
Joint test for network size (number of relatives in Canada, number of sources meeting friends and number of organizations participated in) = 0 0.0006 0.1157 0.0311 0.0000 0.0041 0.0000 0.0000
Joint test for network diversity (ethnic diversity of friends, ethnic diversity of organizational network) = 0, if applicable 0.0004 0.0074 0.0051 0.0262 0.2399 0.0403 0.0447
Joint test for network density (frequency of contact with family sponsors, frequency of contact with friends and frequency of activity with organizations)= 0 0.5319 0.7052 0.3831 0.1579 0.0454 0.0050 0.0823

a p<0.1; b p<0.05; c p<0.01.
Reference categories are in brackets.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Data source: Longitudinal survey of immigrants to Canada (2005).

Table 3 and Table 4 present estimation results across multiple model specifications, the final specifications and the marginal effects for male and female immigrants respectively, based on the GEE population-averaged panel models.

The estimations find evidence of some significant relationships between social networks and labour market outcomes. The findings are robust for different time periods and different statistical models. Again, the directions of the relationships between social capital indicators within various types of social networks and labour market outcomes are mixed. While a more diverse network is associated with higher employment probability, the absolute number of sources meeting friends has a small but negative impact on immigrants’ employment likelihood.

Table 2: Estimation of employment likelihood of female immigrants in the labour force in the initial 4 years in Canada
Sample coverage: Female immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Ordinary Logit Models (No individual specific effects) Panel Models (With individual specific effects)
Wave 1 (6 months after landing) Wave 2 (2 years after landing) Wave 3 (4 years after landing) Pooled Logit Model (W1, 2 & 3) Fixed-effects Logit Model Random-effects Logit Model GEE Population-averaged Model
Coef. Std.Err. Coef. Std.Err. Coef. Std.Err Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
Social networks  
_Family and relatives
Spouse currently employed 0.336b 0.130 0.255b 0.127 -0.022 0.148 0.142b 0.071 0.077 0.090 0.191c 0.072 0.125a 0.069
Number of relatives in Canada 0.055 0.083 0.019 0.072 0.119 0.085 0.069 0.045 -0.126 0.374 0.052 0.051 0.075 0.050
Relatives living nearby upon landing 0.189 0.160 0.331b 0.140 -0.021 0.160 0.17a 0.087     0.273c 0.097 0.169a 0.095
Frequency of contact with family sponsors 0.07 0.258 -0.123 0.204 -0.056 0.186 0.011 0.114 0.357 0.238 -0.013 0.124 0.012 0.117
_Friends  
Friends living nearby upon landing 0.304b 0.125 0.382c 0.111 0.104 0.112 0.289c 0.065     0.337c 0.073 0.283c 0.069
Number of sources meeting friends -0.124c 0.044 -0.16c 0.031 -0.177c 0.040 -0.154c 0.020 -0.086c 0.031 -0.168c 0.022 -0.148c 0.021
Ethnic diversity of friends 0.258 0.226 0.55b 0.250 0.332 0.243 0.45c 0.118 0.367b 0.186 0.479c 0.124 0.454c 0.118
Cohort diversity of friends     0.333 0.214 0.081 0.210                
Frequency of contact with friends 0.337 0.217 0.52b 0.207 0.89b 0.352 0.517c 0.129 0.413b 0.195 0.527c 0.136 0.499c 0.130
_Groups and organizational network  
Number of organizations participated in 0.083 0.257 0.127 0.215 0.308 0.227 0.159 0.127 0.143 0.187 0.212 0.131 0.144 0.127
Ethnic diversity of organizational network -2.616 3.694 2.548 2.962 -1.071 3.339 0.025 1.875 -0.242 2.795 -0.395 1.932 0.129 1.860
Frequency of activity with organizations 0.067 0.319 -0.379 0.316 -0.531 0.332 -0.262 0.183 -0.05 0.252 -0.308a 0.183 -0.234 0.185
Volunteered time for organizations 0.219 0.237 -0.068 0.209 -0.09 0.208 -0.037 0.120 -0.102 0.169 -0.089 0.124 -0.047 0.122
_cons -2.937 2.736 5.552 9.247 -18.864 20.570 0.309 0.526     0.31 0.551 0.277 0.538
No. of observations 2070 2724 2851 7674 3477 7674 7674
No. of groups (for panel models)         1284 3201 3201
Pseudo R2 1.109 0.123 0.1046 0.112      
Percent correctly predicted 0.669 0.716   0.721 0.584 0.703 0.719
Prob > Chi-Square
Joint test for network size (number of relatives in Canada, number of sources meeting friends and number of organizations participated in) = 0 0.0399 0.0000 0.0000 0.0000 0.0391 0.0000 0.0000
Joint test for network diversity (ethnic diversity of friends, ethnic diversity of organizational network) = 0, if applicable 0.4541 0.0008 0.2298 0.0006 0.1426 0.0006 0.0006
Joint test for network density (frequency of contact with family sponsors, frequency of contact with friends and frequency of activity with organizations)= 0 0.4632 0.0484 0.0329 0.0005 0.0971 0.0007 0.0012

a p<0.1; b p<0.05; c p<0.01.
Reference categories are in brackets.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Data source: Longitudinal survey of immigrants to Canada (2005).

Table 3: GEE population-averaged estimations of employment likelihood of male immigrants in the initial 4 years in Canada
Sample coverage: Male immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Model M.1 Model M.2 Model M.3 Model M.4 Model M.5 Marginal Effects
No social capital All social capital indicators Kinship and friendship indicators Social capital and time effects Final specification Final specification
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. dy/dx1 Std. Err.
Social networks  
_Family and relatives
Spouse currently employed 0.36c 0.072 0.368c 0.073 0.367c 0.073 0.393c 0.073 0.372c 0.073 0.055c 0.010
Number of relatives in Canada     -0.046 0.053 -0.046 0.053            
Relatives living nearby upon landing     0.252b 0.100 0.248b 0.100 0.184b 0.076 0.183b 0.076 0.028b 0.012
Frequency of contact with family sponsors     0.203 0.129 0.206 0.129            
_Friends  
Friends living nearby upon landing     0.123a 0.070 0.122a 0.069            
Number of sources meeting friends     -0.103c 0.019 -0.103c 0.018 -0.08c 0.019 -0.097c 0.018 -0.015c 0.003
Ethnic diversity of friends     0.283b 0.124 0.298b 0.124 0.301b 0.124 0.288b 0.123 0.045b 0.019
Frequency of contact with friends     0.251a 0.129 0.244a 0.129 0.185 0.129 0.256b 0.128 0.04b 0.020
_Groups and organizational network  
Number of organizations participated in     -0.077 0.115                
Ethnic diversity of organizational network     1.549 1.812                
Frequency of activity with organizations     -0.111 0.167                
Volunteered time for organizations     0.122 0.112                
Time effects  
Wave2             -0.213 0.373        
Wave3             0.196 0.418        
_cons -0.275 0.500 -1.024a 0.529 -1.029a 0.529 -1.507b 0.620 -0.826 0.518    
No. of observations 9843 9843 9843 9843 9843 9843
No. of groups 3659 3659 3659 3659 3659 3659
Percent correctly predicted 0.787 0.790 0.789 0.791 0.788  

a p<0.1; b p<0.05; c p<0.01.
1 Marginal effects for dummy variables are for discrete change from 0 to 1.
Reference categories are in brackets.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Data source: Longitudinal survey of immigrants to Canada (2005).

The coefficients of estimations for male immigrants’ employment probability (Table 1 and 3) are to a large extent consistent with what are expected from relevant literature. The last columns in Table 3 report marginal effects of the social capital variables.

For male newcomers, relatives’ network has few impacts on employment status except the positive impacts of the employment status of spouses and geographic closeness of relatives upon landing. Having a spouse who is currently working is associated with a 5% higher employment likelihood while having a relative living nearby at landing is related to a probability of 2.8% higher to find a job than having no kinship or relatives living far.

Friendship network has a mixed impact on males’ employment likelihood. Friendship size, ethnic diversity and frequency of contact with friends are the main elements affecting male immigrants’ employment probability, with marginal effects of -1.5% and 4.5% and 4% respectively.

There is no evidence of a linkage between organizational network and employment likelihood for male immigrants.

Table 4: GEE population-averaged estimations of employment likelihood of female immigrants in the initial 4 years in Canada
Sample coverage: Female immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Model F.1 Model F.2 Model F.3 Model F.4 Model F.5 Marginal Effects
No social capital All social capital indicators Kinship and friendship indicators Social capital and time effects Final specification Final specification
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. dy/dx1 Std. Err.
Social networks  
_Family and relatives
Spouse currently employed 0.079 0.067 0.125a 0.069 0.125a 0.069 0.148b 0.072 0.125a 0.068 0.026a 0.014
Number of relatives in Canada     0.075 0.050 0.073 0.050            
Relatives living nearby upon landing     0.169a 0.095 0.169a 0.095 0.254c 0.075 0.255c 0.074 0.053c 0.016
Frequency of contact with family sponsors     0.012 0.117 0.014 0.117            
_Friends  
Friends living nearby upon landing     0.283c 0.069 0.276c 0.068 0.278c 0.069 0.279c 0.068 0.057c 0.014
Number of sources meeting friends     -0.148c 0.021 -0.143c 0.020 -0.136c 0.020 -0.142c 0.020 -0.03c 0.004
Ethnic diversity of friends     0.454c 0.118 0.455c 0.118 0.457c 0.117 0.458c 0.117 0.095c 0.024
Frequency of contact with friends     0.499c 0.130 0.487c 0.129 0.463c 0.130 0.484c 0.129 0.101c 0.027
_Groups and organizational network  
Number of organizations participated in     0.144 0.127                
Ethnic diversity of organizational network     0.129 1.860                
Frequency of activity with organizations     -0.234 0.185                
Volunteered time for organizations     -0.047 0.122                
Wave2             -0.569 0.410        
Wave3             -0.462 0.455        
_cons 1.213b 0.501 0.277 0.538 0.304 0.536 -0.274 0.646 0.382 0.518    
No. of observations 7674 7674 7674 7674 7674 7674
No. of groups 3201 3201 3201 3201 3201 3201
Percent correctly predicted 0.711 0.719 0.720 0.719 0.720  

a p<0.1; b p<0.05; c p<0.01.
1 Marginal effects for dummy variables are for discrete change from 0 to 1.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Reference categories are in brackets.
Data source: Longitudinal survey of immigrants to Canada (2005).

Tables 2 and 4 indicate that more social network indicators relate to female immigrants’ employment probability than to males: Besides spouse’s employment status, geographic closeness of relatives upon landing, ethnic diversity, size of friendship network and frequency of contacts with friends which are significantly related to male immigrants’ labour market entry, there is also a significant positive association between geographic closeness of existing friends upon landing and female newcomers’ employment likelihood in the labour force.

It is worth mentioning that social capital indicators affect female immigrants to a greater degree than males. As indicated by the last columns in Table 3 and 4 with marginal effects of the social capital variables, geographic closeness of relatives upon landing increases males’ employment likelihood by 2.8% while raising the chance for female immigrants to find a job by 5.3%. Ethnic diversity and frequency of contact with friends both could increase the employment likelihood for females by about 10% while they only work for male immigrants at around 4%. 

Like the results for male immigrants, no convincing evidence was found to suggest that organizational network affects the probability of employment for female immigrants.

The joint Wald tests for network structure components and quality (at the bottoms of Table 1 and 2) further reveal that network size (joint test for three independent variables – number of relatives living in Canada, number of sources meeting with friends and number of organizations participated in), diversity (joint test for two variables – ethnic diversity of friends and ethnic diversity of organizational network) and density (joint test for three explanatory variables – frequency of contact with family sponsors, frequency of contact with friends and frequency of activities with organizations) are all significantly related to females’ labour market outcomes, whereas density or frequency of contact tends to have no relationship with male immigrants’ employment outcomes.

The results from cross sectional models and panel models tell a consistent story. And the likelihood ratio tests for the presence of panel-level variance component reject the null hypothesis of no panel-level variance component for both male and female estimations.10 So panel models with individual specific effects are more appropriate for my analysis.

Comparing the results from panel models, it is not surprising to see the differences between fixed effects estimator with other panel data model estimators. The much smaller sample size in fixed effects model compared to the random effects and the GEE population averaged estimators results from dropping subjects with same outcomes across periods (i.e. employed at 3 waves or unemployed at 3 waves). The fixed effects estimator is very inefficient since, as noted in the Table 1 and Table 2, six out of 10 units whose outcome were all 0 or 1 cannot contribute to the analysis. The random effects estimator and the GEE population averaged estimator are more efficient and tend to yield smaller standard errors leading to smaller p-values.

6.2 Time effects

As the cross sectional regressions for different periods (Wave 1, 2 and 3) showed in Table 1 and Table 2, the coefficients are quite different across periods. There may be differential impacts of the second wave (two years after landing) or the third wave (four years after landing) on the immigrants’ employment probability relative to the first wave (six months after landing). The panel model (or individual-specific model) (1) can be extended to include a time-specific effect as well.

(5) Mathematical equation     i = 1,…, n, t=1,…, Ti.

Mathematical equation

where st captures the omitted variables that are related to immigrants’ employment outcomes, varying over time but not across individuals, such as the changes in national economic conditions.

When time effects are incorporated into the employment probability models (Models 4 in Table 3 and Table 4), it shows that time effects are not significantly present on the employment probability for both female and male immigrants. 

6.3 Differential ethnic diversity effect

As already noted in the above analysis, ethnic diversity of friendship network appears to be the main factor within social network indicators linked to labour market outcomes for both male and female immigrants. Based on the GEE population averaged model, the ethnic diversity effects are further examined across different explanatory variables by interacting friendship diversity indicator with a variety of independent variables, to find some interesting patterns (Table 5 and Table 6):

Table 5: GEE population-averaged estimations of employment likelihood of male immigrants in the initial 4 years in Canada with interaction effects
Sample coverage: Male immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Model M.int.1 Model M.int.2 Model M.int.3 Model M.int.4 Model M.int.5
Time effect interactions Immigration category interactions Ethnic group interactions Education interactions Official language interactions
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Social networks  
Spouse currently employed 0.392c 0.073 0.376c 0.073 0.37c 0.073 0.373c 0.073 0.372c 0.073
Relatives living nearby upon landing 0.183b 0.075 0.183b 0.076 0.18b 0.076 0.182b 0.075 0.183b 0.076
Number of sources meeting friends -0.081c 0.019 -0.095c 0.018 -0.098c 0.018 -0.097c 0.018 -0.097c 0.018
Ethnic diversity of friends 0.584c 0.175 -0.334 0.281 0.448b 0.219 0.332a 0.192 0.323 0.324
Frequency of contact with friends 0.145 0.130 0.312b 0.131 0.267b 0.130 0.269b 0.128 0.256b 0.129
Time effects  
Wave2 -0.055 0.395                
Wave3 0.437 0.431                
Interactions  
Wave2 * Ethnic diversity of friends -0.355 0.242                
Wave3 * Ethnic diversity of friends -0.526b 0.243                
SWPA * Ethnic diversity of friends     0.831c 0.317            
SWSD * Ethnic diversity of friends     0.824b 0.407            
Refugees * Ethnic diversity of friends     0.342 0.382            
Others * Ethnic diversity of friends     0.447 0.466            
Chinese * Ethnic diversity of friends         0.426 0.372        
South Asian * Ethnic diversity of friends         -0.194 0.328        
Black * Ethnic diversity of friends         -0.955b 0.461        
Filipino * Ethnic diversity of friends         -1.521b 0.653        
Latin * Ethnic diversity of friends         0.639 0.658        
West Asian and Arab * Ethnic diversity of friends       0.018 0.341        
Other Asian * Ethnic diversity of friends         -1.021b 0.504        
Other Visible Minority * Ethnic diversity of friends       -0.571 1.091        
High school diploma or less * Ethnic diversity of friends           -0.286 0.289    
Some post-secondary education * Ethnic diversity of friends         0.139 0.525    
College diploma or some university * Ethnic diversity of friends         0.03 0.341    
Master's degree or above * Ethnic diversity of friends           0.027 0.315    
English * Ethnic diversity of friends                 -0.055 0.321
French * Ethnic diversity of friends                 0.05 0.259
_cons -1.611b 0.624 -0.659 0.520 -0.927a 0.544 -0.831 0.521 -0.834 0.524
No. of observations 9843 9843 9843 9843 9843
No. of groups 3659 3659 3659 3659 3659
Percent correctly predicted 0.792 0.789 0.789 0.789 0.788

a p<0.1; b p<0.05; c p<0.01.
Reference categories are in brackets.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Data source: Longitudinal survey of immigrants to Canada (2005).

Table 6: GEE population-averaged estimations of employment likelihood of female immigrants in the initial 4 years in Canada with interaction effects
Sample coverage: Female immigrants who participated in the labour force.
Dependent variable em (conditional on the participation in the labour force)
  Models
Independent variables Model F.int.1 Model F.int.2 Model F.int.3 Model F.int.4 Model F.int.5
Time effect interactions Immigration category interactions Ethnic group interactions Education interactions Official language interactions
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Social networks  
Spouse currently employed 0.145b 0.072 0.12a 0.069 0.13a 0.069 0.125a 0.069 0.127a 0.068
Relatives living nearby upon landing 0.255c 0.075 0.257c 0.074 0.252c 0.075 0.253c 0.075 0.254c 0.074
Friends living nearby upon landing 0.279c 0.069 0.281c 0.068 0.283c 0.069 0.277c 0.069 0.278c 0.068
Number of sources meeting friends -0.134c 0.020 -0.141c 0.020 -0.143c 0.020 -0.142c 0.020 -0.141c 0.020
Ethnic diversity of friends -0.094 0.193 0.12 0.206 0.303 0.204 0.612c 0.189 0.263 0.303
Frequency of contact with friends 0.543c 0.133 0.515c 0.130 0.482c 0.130 0.483c 0.129 0.487c 0.130
Time effects  
Wave2 -1.006b 0.424                
Wave3 -0.827a 0.466                
Interactions  
Wave2 * Ethnic diversity of friends 0.89c 0.252                
Wave3 * Ethnic diversity of friends 0.731c 0.246                
SWPA * Ethnic diversity of friends     0.597a 0.323            
SWSD * Ethnic diversity of friends     0.437a 0.263            
Refugees * Ethnic diversity of friends     0.427 0.359            
Others * Ethnic diversity of friends     0.85a 0.462            
Chinese * Ethnic diversity of friends         0.249 0.377        
South Asian * Ethnic diversity of friends         0.183 0.288        
Black * Ethnic diversity of friends         -0.664 0.439        
Filipino * Ethnic diversity of friends         1.301a 0.680        
Latin * Ethnic diversity of friends         0.165 0.654        
West Asian and Arab * Ethnic diversity of friends       0.455 0.378        
Other Asian * Ethnic diversity of friends         -0.329 0.497        
Other Visible Minority * Ethnic diversity of friends       1.559a 0.864        
High school diploma or less * Ethnic diversity of friends           -0.132 0.269    
Some post-secondary education * Ethnic diversity of friends         -0.08 0.443    
College diploma or some university * Ethnic diversity of friends         -0.197 0.303    
Master's degree or above * Ethnic diversity of friends           -0.473 0.328    
English * Ethnic diversity of friends                 0.307 0.299
French * Ethnic diversity of friends                 -0.3 0.262
_cons -0.152 0.648 0.459 0.521 0.497 0.537 0.312 0.523 0.431 0.521
No. of observations 7674 7674 7674 7674 7674
No. of groups 3201 3201 3201 3201 3201
Percent correctly predicted 0.720 0.721 0.722 0.720 0.721

a p<0.1; b p<0.05; c p<0.01.
Reference categories are in brackets.
Note: The estimations also include control variables for immigration category, demographic and household characteristics (age, marital status, number of children), province of residence, region of birth, ethnic group, education, official language skills, previous experience or attachment in Canada. See Appendix for the complete results.
Data source: Longitudinal survey of immigrants to Canada (2005).

1) Interactions of friendship ethnic diversity with time effects (Model M.int.1 and F.int.1).

When interactions of time effects with ethnic diversity of friendship (see Model M.int.1 in Table 5 and Model F.int.1 in Table 6) are added to the model, the time effects significantly show for female immigrants. Female immigrants are less likely to find employment in Wave 2 (i.e. from six months to two years in Canada) and Wave 3 (i.e. from two years to four years after landing) relative to the base period – Wave 1 (the first six months in Canada). As the LSIC target population is the immigrants who landed from abroad from October 2000 to September 2001, the Wave 2 and 3 time periods cover the time span from April 2001 to late 2005 when the macroeconomic condition was generally getting worse after the economic downturn. Furthermore, for female newcomers, friendship diversity has more impact on employment likelihood in Wave 2 and 3 than in Wave 1, showed by the positive and statistically significant coefficient of the interaction terms.

While time effect is not presented for male immigrants’ employment probability, the ethnic diversity of friendship network displays significantly less effect on employment in Wave 3 than in Wave 1 for this group. This result suggests that as time goes on, the effect of ethnic diversity decreases for male immigrants. In other words, the effect of ethnic concentration increases. If the first six months are treated as short run and the second and third waves are treated as longer run, this result seems to be consistent with the longer run implication of Calvó-Armengol and Jackson’s theoretical model (Calvó-Armengol and Jackson 2004). In the short run, newcomers compete against each other. In the longer run, after some members get jobs, members of the same ethnic group are more likely to pass information to each other so that the effect of ethnic diversity decreases with the time spent in Canada.

2) Interaction of friendship ethnic diversity with immigration class (Model M.int.2 and F.int.2).

Ethnic diversity of friendship has more impacts on the probability finding employment for male skilled workers (both principal applicants and spouses and dependants) than for the reference group – family class male immigrants. For female immigrants, friendship ethnic diversity does not show significantly different impacts across immigration categories.

3) Interaction of friendship ethnic diversity with population group (Model M.int.3 and F.int.3).

When looking across the various visible minority groups, the directions of the effects of ethnic diversity are quite different. For the Chinese male immigrants, an ethnic diversified network has more impact on employment probability than for the reference group – White immigrants, but the effect is not statistically significant. For Black, Filipino and Southeast Asian and Japanese and Korean male immigrants, the effect of friendship diversity on employment entry is less relative to White immigrants, indicated by significant negative coefficients. For female immigrants, ethnic diversity plays a stronger and marginally significant role only for Filipino and other visible minority newcomers in the labour market relative to the reference group --- White female immigrants.

4) Interactions of friendship ethnic diversity with education and official languages ability (Model M.int.4, M.int.5 and F.int.4, F.int.5).

In the literature, it is argued that social capital substitutes for human capital and has stronger effects for those disadvantaged in human capital (e.g. Livingston 2006). Thus human capital variables – education level and official languages proficiency are interacted with friendship diversity to see if social capital serves as a (non-perfect) substitute for human capital. If the argument holds, it is expected to see that social capital has stronger effects for those with lower education or less knowledge of official languages.

However, the results do not show any evidence of the substitute role of social capital on employment probability. Friendship diversity has no different effects for immigrants with different official language abilities and education levels on the likelihood of employment entry.

In summary, for female immigrants, ethnic diversity effects are mostly captured by the interaction terms, showed by the insignificant coefficient of the diversity indicator and significant coefficients of interaction terms for some specific groups. A diversified friendship network is especially important for female economic class immigrants and Filipino newcomers in the Canadian labour market. For male immigrants, social capital is more related to employment status of some specific groups. After controlling for disparity within different groups, a diverse friendship has stronger effects on the employment likelihood of male skilled workers but weaker impacts on employment entry of Black, Filipino and Southeast Asian, Japanese and Korean.

 

Notes

9. As the outcome – employment status – is conditional on the participation in the labour force, I also estimate the probability of participating in the labour force using various indicators of the stock of social capital along with other socioeconomic variables. The results from the labour force participation estimations are available upon request.

10. See Appendix B for a detailed discussion of the test.

Page details

Date modified: