|
Article Excerpt
Using a multinomial logit model, this article explains the initial destination choices of skilled-worker immigrants from four South Asian countries (India, Pakistan, Bangladesh, and Sri Lanka) who landed in Canada in 1992-2001, based on the micro data of Citizenship and Immigration Canada. We found that their choice pattern, which is characterized by extremely strong concentration in Ontario, was strongly affected by the attractions of (1) co-ethnic communities and (2) long-term income opportunities represented by earned income per capita. The temporal pattern of their choices was subject to the lagged effects of the fluctuations in the spatial pattern of employment opportunities in an economically sensible but relatively mild way. The enhancement of Quebec's attraction by the Canada/Quebec agreement on immigration dissipated within only a few years.
En utilisant une mode le multinomial logit, cet article fournit une explication des choix de destination initiale des immigrants qualifies de quatre pays de l'Asie du sud (Inde, Pakistan, Bangladesh et Sri Lanka) qui sont arrive au Canada dans la periode 1992 a 2001, se basant sur les donnees micro de l'Immigration et citoyennete Canada.
La caracteristique la plus pertinente de ces immigrants qualifies de ces quatre pays de l'Asie du sud s'ave re la tendance tre s forte de choisir l'Ontario en tant que destination initiale : plus de 80 % ont choisi l'Ontario. Notre analyse multi variee reve le que la raison principale de cette concentration extremement forte en Ontario fut l'influence de (1) l'attraction de leur communautes ethniques et (2) l'effet positif du niveau de revenu eleve.
Les choix de destination initiale des immigrants a l'etude furent aussi relies aux changements dans le patron spatial de la croissance de l'emploi de manie re raisonnable : leur preference fut influence a l'effet positif du taux de croissance de l'emploi. Vu les delais entre la demande du statut d'immigrant recu et la decision finale, il est raisonnable que l'effet de la croissance de l'emploi soit retarde de deux ans, et, a un moindre degre, par trios ans. Toutefois, en comparaison avec les migrants entre les provinces au Canada et les migrants entre etats aux E.-U., ces immigrants dans leurs comportements de choix de destination se sont averes beaucoup moins sensibles aux variations a court terme dans le patron d'opportunites d'emploi.
En plus de la disparition rapide de l'effet positif de l'Accord de 1991 entre la Canada et le Quebec sur la proportion des immigrants a l'etude allant au Quebec, nous avons aussi trouve que ceux ayant une competence linguistique en francais avaient une tendance plus importante de choisir le Quebec, et que ceux de plus de 60 ans avaient une tendance plus importante de choisir la Colombie-britannique. Toutefois, ces resultats tre s selectifs ne concernent que des proportions tre s restreintes des immigrants qualifies de ces pays d'Asie du sud et donc ne representent qu'une partie minime du portrait global.
Etant donne que les disparites de revenu entre les provinces nanties et non nanties s'ave rent persistantes, et etant donne que la concentration des communautes de l'Asie du sud dans l'Ontario a ete maintenue et meme renforcee par l'arrivee continue de nouveaux immigrants, nos resultats sugge rent qu'il y a peu de marge de manoeuvre pour qu'une politique d'immigration puisse faire afin d'encourager les nouveaux immigrants qualifies de ces pays d'Asie du sud a changer leur choix de destination initiale de l'Ontario vers les provinces les moins bien nanties. Un effort pour renforcer leurs preferences pour une autre province nantie, l'Alberta, aura plus de chances de reussir.
La concentration importante et continue des immigrants nouvellement arrives de l'Asie du sud en Ontario au cours des annees 1990 est en contraste marquee avec la dispersion accrue des immigrants peu qualifies nouvellement arrivees aux Etats-Unis (surtout les immigrants d'origine hispanique) a partir de la fin des annees 1980 a la fin des annees 1990, en partie reliee aux efforts de recrutement des employeurs des regions non metropolitaines qui avaient accuse des difficultes de combler leurs besoins pour des emplois dangereux, exigeants en termes de travail manuel et demandant peu de qualifications, du bassin de main d'oeuvre des travailleurs nes aux Etats-unis. D'autres recherches sont necessaires afin d'evaluer si les mecanismes d'une telle dispersion des immigrants puissent etre aussi developpes au Canada dans un avenir proche.
Introduction
The geographic distribution of immigrants has become a major policy concern in Canada in recent years (SPPR 2001). At the provincial level, immigrants are too concentrated in British Columbia and especially Ontario in the sense that the shares of immigrants by these two provinces are substantially greater than their shares of the Canadian-born population. According to the 2001 census, the shares of the nation's stock of 5.45 million immigrants by these two provinces were 18.5 % and 55.6 %, whereas their shares of the Canadian-born population were 11.8 % and 34.0 %, respectively. Furthermore, a comparison with the data of the 1996 census revealed that the strong concentration into these two provinces increased in the late 1990s; their shares of the immigration population in 1996 were 18.2 % and 54.8 %, respectively.
The very uneven distribution of immigrants in Canada has important societal implications. To the extent that immigrants are valuable human resources that help keep productivity high and labour cost low, the heavy concentration of the immigrants into Ontario and British Columbia helps strengthen the economies of these two provinces and sustain the persistent economic disparities between the "have" and "have not" provinces (Higgins and Savoie 1997). Since this concentration has also been an important factor leading to the declining shares of the Canadian population by Quebec and other "have not" provinces, it has contributed to the decline of their shares of seats in the House of Commons and hence reduced their political representation at the center of the nation's political power (Liaw and Xu 2005). Receiving a continual inflow of new immigrants in the order of 100,000 persons per year in recent years (CIC 2005a), the metropolitan area of Toronto in Ontario is under pressure to expand its infrastructures at a rapid pace and is experiencing problems of traffic congestion and shortage of public housing. In light of such implications, it is not surprising that both the federal government and some provincial governments (especially those of Quebec and Manitoba) are actively pursuing ways to induce a more dispersed distribution of immigrants (SPPR 2001; CIC 2005b).
The most important process that has resulted in the very uneven distribution of the immigrant population in Canada is the immigrants' initial destination choices (i.e. the choices made at the time of landing). For example, among the 2.2 million immigrants who landed in Canada in the ten years between 1995 and 2004, 18.3 % and 55.4 % chose British Columbia and Ontario at the times of their landing (CIC 2005a). Thus, it is important to understand how different factors affect the initial destination choices of the immigrants.
This paper is motivated by the desire to explain the initial destination choices of immigrants in Canada. We begin by noticing that these choices vary substantially with respect to immigration class and place of origin. This is best demonstrated by the relative attractiveness of British Columbia and Quebec. Among the 270,000 refugees who obtained landed immigrant status in 1995-2004, many more were received by Quebec (26.5 %) than by British Columbia (7.8 %), whereas among the 629,000 immigrants of the family class who landed in the same period, more chose British Columbia (19.3 %) than Quebec (13.1%). In the same landing period, many more of the 402,000 immigrants from European sources chose Quebec (23.4 %) than British Columbia (10.9 %), whereas many more of the 1,136,000 immigrants from Asian and Pacific sources went to British Columbia (25.6 %) than to Quebec (7.7 %) (CIC 2005a).
In light of the large differences in the initial destination choices by immigration class and place of origin, a sensible strategy to understand how they are affected by different factors is to focus on the immigrants of a specific class from some specific sources. Since we are particularly concerned with the excessive concentration in the immigrants' initial destination choices, we decide to focus on the skilled-worker immigrants from four South Asian countries: India, Pakistan, Bangladesh, and Sri Lanka. A common feature of the skilled-worker immigrants from each of these four countries is that over 80 % of them chose Ontario as their initial destination. Another feature of them is that their share of the skilled-worker immigrants from all sources showed a sharp upward trend, increasing from 9 % in 1992 to 25 % in 2001. (1)
The main purposes of this paper are (1) to identify a set of explanatory factors that can explain to a large extent the initial destination choices of the skilled-worker immigrants from these four South Asian countries and (2) to assess the relative importance of the explanatory factors in a multivariate context. In addition to providing substantive insights into the very uneven destination choice patterns of the immigrants, this paper makes a methodological contribution by presenting a new method for assessing the relative importance of explanatory factors that overlap substantially in explanatory powers.
The Sample and Statistical Model
For generating the information on the initial destination choices, we use the micro data in the 2001 version of the LIDS (Landed Immigrant Data System), which is created and maintained by Citizenship and Immigration Canada (CIC). In order to match the information on both the dependent variable and explanatory factors, the landing years of the immigrants are restricted to the period between January 1, 1992 and December 31, 2001. Since very few immigrants went to the Yukon, Northwest Territory, and Nunavut, we further restrict the immigrants to be those who landed in one of the ten provinces. Under these restrictions, our sample includes 63,468 skilled-worker immigrants who were "principal applicants" from India, Pakistan, Sri Lanka and Bangladesh. Note that principal applicants are allowed to include their spouse and dependents in their application for landed immigrant status. Excluding the 110 immigrants with missing values on one or more personal attributes used in our study, the size of the sample for our multivariate analysis becomes 63,358 persons.
To explain the initial destination choices of the skilled-worker immigrants, we use the following multinomial logit model:
(j|s, t) = exp(b' x(s, t, j))/[10.summation over k = 1] (b' x(s, t, k)) (1)
where p(j|s,t) is the probability that an immigrant with personal attributes s who lands in year t will choose province j as the destination; x(s, t, j) is a column-vector of observable explanatory variables; and b' is a row-vector of unknown coefficients.
To estimate the unknown coefficients, we first cross-tabulate ail the skilled-worker immigrants according to the following dimensions: (1)...
|
