Вы находитесь на странице: 1из 14

Economic Assimilation of Immigrants to Canada over the 20th Century

May 1, 2013 Submitted to: Ben Sand Empirical Labour Economics ECON 5620 Submitted by: Marshall Boyd

1. Introduction
Immigration has played always played an important role in Canadas history. Canadian governments have always encouraged immigration as a way to bolster the population. For much of the twentieth century immigration has been an important factor in Canadas labour policy and starting in the mid-1960s Canada adopted an objective points system designed to bring in more skilled immigrants and tailor immigration to the labour needs of the economy. The question of how well do immigrants perform in the labour force compared to native workers has been a question of much interest in both Canada and the United States. Over the past three decades there has been much work trying to answer this question. Chiswick (1978) began research into the earnings and labour market assimilation of immigrants using the 1970 census to examine the effects on white, male immigrants in the United States. His research supported his hypothesis that initial earnings of immigrants would be less than native workers because skills are not perfectly mobile across countries. The returns to immigration are expected to be higher for those that have unobservable skills, ability, intelligence, or chutzpa; immigrants possessing these skills are expected to self-select for immigration and have higher returns than native workers holding all other observable characteristics. For this reason, over time, as skills are gained the immigrant is expected to earn more quickly and catch up or surpass the earnings of native workers. Chiswick found that the crossover period, when immigrants began to earn more than native workers, was between 10-15 years after immigration. This crossover effect is expected to be weaker if the self-selection is weaker, such as when immigration is due to refugee status or for family reunification. Borjas (1985) questions these results by examining the earnings of immigrants across both the 1970 and 1980 census. The main critique by Borjas is that the estimate of the effect of 1

years since immigration on earnings is biased by the changing quality of immigrants. Over time, if immigrants are not doing as well in the United States they can emigrate back to the country of origin where they will presumably perform better. This will improve the overall quality of earlier immigration cohorts as they self-select to stay or re-migrate based on labour market performance. A second bias can result from assuming that each cohort has the same quality of immigrants to begin with. This may not be true depending on changing immigration policies or political disturbances around the world. These critiques are addressed by splitting immigrants into cohorts based on the period of immigration and measuring the cohort effect across time by using the two censuses as semi-longitudinal data. By looking at cohorts across time, Borjas finds that the earnings growth of immigrants is greatly exaggerated by cross-sectional studies of immigrant earnings. The total growth of

earnings is much slower and the growth of immigrants earnings compared to native workers is even lower and sometimes negative. This means that the crossover period will be much longer than 10-15 years if it occurs at all. This result is found to hold for Canada, as well, through the 1971, 1981, and 1986 censuses (Bloom, Genier, and Gunderson, 1995). Using the same empirical method as Borjas (1985) to create semi-longitudinal data from multiple censuses, Bloom et al. find that the entry level earnings and the rate of labour market assimilation of immigrants has eroded over time. Earlier immigrants to Canada experience faster and more complete assimilation while more recent immigrants start in a worse position and are unlikely to ever reach complete assimilation in the labour market.

These findings are supported by Baker and Benjamin (1994) who also use the 1971, 1981, and 1986 Canadian censuses to examine the labour market assimilation of immigrants. They also note that immigrants will receive lower returns to education than native workers. This has been found previously in studies of American immigration but has been explained by the lower skills immigrants to the USA have relative to native workers. This explanation does not work in Canada as immigrants have, on average, more education than native workers. The decreased returns to education will counteract any assimilation into the workforce by immigrants, guaranteeing that they will receive lower earnings for their working life. In his 1995 paper, Borjas updates the findings of his 1985 paper by adding the 1990 census data to his analysis and finds that the trend in decreasing returns to immigrants continues throughout the 1980s. Each successive cohort of immigrants starts off in a worse position and improves less over time. In 1970 the average immigrant earned 1% more than a native worker by 1980 this had become nearly 10% less and in 1990 16.5% less. Further, it is unlikely that immigrant cohorts will ever reach parity with native workers and will spend most of their working life earning 15-20% less than native counterparts (Borjas, 1995). Grant (1999) uses the 1981, 1986, and 1991 Canadian censuses to look at continuing trends in immigrant assimilation. She finds that immigrants in the 1980s experienced far higher assimilation than earlier immigrants. She also found that immigrants in the later 1980s had similar entry earnings to those in the later 1980s suggesting that the downward trend of entry earnings may be ending. In this paper I will be using the 1971, 1981, 1986, 1991, 1996, and 2001 Canadian censuses to look at the assimilation of immigrants to Canada over time. Most studies have

indicated a downward trend in immigrant earnings with Grant hypothesizing that this trend ended in the 1980s, using data after this period I will ascertain whether this trend actually ended or if the 1980s were simply an anomalous period.

2. Model
The basic equation that will be examined in this paper is
, = + , +

(1)

wages and self-employment earning, is a vector of observable characteristics, , is the cohort

Where is the log of annual earnings of cohort i in period t as found by summing

effect for immigration cohort i in period t, and is an error term. Lalonde and Topel (1992) define the error term as , = , + , +

(2)

in period t. This effect varies over time as the cohort gains human capital relevant to the

The first component, , is the cohort effect and measures the assimilation of the cohort i

Canadian labour market. , is a coefficient that measures all the unobservable macro effects that variable is a fixed variable that represents the quality of each cohort.

are occurring in the Canadian labour market that would cause a change in earnings. The final

Equation (1) is estimated using the pooled data from the 1971, 1981, 1986, 1991, 1996, and 2001 Canadian censuses to act as semi-longitudinal data. In each census there is no way to track who is being surveyed but the immigration cohort is used to simulate longitudinal

information.

The immigration cohorts are immigrants before 1946, 1946-1955, and each five

year period thereafter. Equation (1) is estimated as: + , , = , (3)

From this we can estimate a cross-section of growth by taking the difference of cohort i , for both cohorts. In this way the difference is and cohort i+k using the same average level of

, +, under the assumption that , +, = 0 and + = 0. The first assumption is easy to make since it relates the unobserved macro effects over a period of time. These effects will affect both cohorts, though some might argue that a cohort that has been in the country longer will be better able to weather negative effects and take advantage of positive effects leading to a bias in this estimate. The second assumption is harder to make as it seems unlikely that each cohort of immigrants will all have equal quality. Borjas (1985) argues that the quality of cohorts has decreased but we see in Grant (1999) that the level of both education and experience has increased in recent cohorts. Either way it reveals that there is some level of bias (probably negative for Canadian immigration) in the level of assimilation of immigrants. A quasi-longitudinal estimate of growth can be estimated by comparing the same cohort across censuses. Using equation (3) again we take the difference between cohort i in time t and and . The , so that the difference is again just in t+k again using the same average level of difference in can again be netted out to , ,+ with the assumptions , ,+ = 0 and quality; however, over time the cohort quality could easily change as the less skilled and less 5 = 0. The second assumption is easier to make since we assumed to be a fixed cohort

, and the unobserved , . The difference in can be netted out to limited to only the change in

successful immigrates migrate back to their country of origin or some other country with the hopes of success. The first assumption is definitely not true as the unobserved macro effects on the economy will undoubtedly be changing over time. The difference in potential earnings using equation (3) can be decomposed following Borjas (1985) method: , +, = ( , , ) + ( , +, ) (4)

The first term on the right hand side of equation (4) is the within cohort growth of earnings of cohort i. This is the change in earnings for cohort i across two censuses. This is what Grant (1999) refers to as the true assimilation. The second term is the across cohort growth. This is the change in earnings of cohort i in year t-k compared to the earnings of cohort i+k in year t; this compares the earnings of cohorts after spending the same number of years in Canada.

3. Results
Equation (1) is estimated controlling for the observable characteristics weeks worked in the reference year (the year prior to the census year), hours worked in the reference week (the week prior to being surveyed), marital status, years of schooling, experience, experience squared, and the period of immigration into Canada. The dependent variable is found by taking the log of earnings, wages and self-employed earnings, in the reference year. Table 1 shows the estimated return to immigration for each immigration period. From this table we can see that earlier immigration cohorts, those arriving before 1965 performed much better in the Canadian labour market. We cannot see what the entry level earnings

Table 1 Returns to Immigration Census Year 1971 1981 1986 1991 0.057* 0.073* 0.006 -0.073 (0.017) (0.022) (0.03) (0.054) 0.063* 0.042* 0.033* 0.066* (0.012) (0.01) (0.012) (0.013) 0.006 0.009 0.053* 0.069* (0.016) (0.015) (0.013) (0.013) 0.016 -0.019 0.009 0.422* (0.021) (0.015) (0.016) (0.015) -0.021 -0.022* -0.043* 0.029* (0.016) (0.011) (0.012) (0.011) -0.105* -0.108* -0.066* (0.013) (0.013) (0.011) -0.139* -0.147* -0.103* (0.013) (0.015) (0.013) -0.319* -0.152* (0.019) (0.014) -0.303* (0.013) -

2001 0.205 (0.181) 1946-1955 0.093* (0.023) 1956-1960 0.008 (0.021) 1961-1965 0.051* (0.022) 1966-1970 0.024 (0.015) 1971-1975 -0.009 (0.013) 1976-1980 -0.085* (0.015) 1981-1985 -0.180* (0.016) 1986-1990 -0.249* (0.013) 1991-1995 -0.324* (0.012) 1996-2000 -0.366* (0.014) Note The standard errors are the values reported in parenthesis. Values denoted by an asterisk are significant at the 5% level. Period of Immigration Pre 1946 differential was before the 1966/1970 cohort was but we can see that after only five years in the labour force the 1961/1965 cohort was already at parity with native Canadian workers. The earlier cohorts were also performing either at parity or above the average Canadian. After the 1966/1970 cohort the entry level earnings of each cohort fell. The 1976/1980 cohort was 14% below the Canadian average, throughout the 1980s it was 30% below, the 1991/1995 cohort was an astonishing 45% below, and the 1996/2000 cohort improved to 36% below Canadian native workers.

1996 0.110 (0.121) 0.418* (0.018) 0.027 (0.018) 0.029 (0.019) 0.002 (0.13) -0.050* (0.013) -0.153* (0.014) -0.203* (0.016) -0.311* (0.013) -0.456* (0.014) -

Table 2 Decomposed Effects of Returns to Immigration 1971/1981 1981/1986 1986/1991 Period of CrossWithin Across CrossWithin Across CrossWithin Across Immigration Section Growth Growth Section Growth Growth Section Growth Growth Pre 1946 0.064 0.017 0.048 -0.027 -0.067 0.040 -0.139 -0.079 -0.060 1946-1955 0.062 -0.021 0.082 -0.020 -0.009 -0.011 -0.003 0.033 -0.036 1956-1960 0.032 0.003 0.028 0.044 0.044 0.000 -0.353 0.016 -0.368 1961-1965 0.086 -0.035 0.122 0.053 0.029 0.024 0.392 0.412 -0.020 1966-1970 0.117 -0.002 0.118 0.065 -0.021 0.086 0.095 0.073 0.022 1971-1975 0.039 -0.003 0.042 0.037 0.042 -0.005 1976-1980 0.172 -0.008 0.180 0.049 0.044 0.005 1981-1985 0.151 0.167 -0.016 1986-1990 1991-1995 1991/1996 1996/2001 CrossWithin Across CrossWithin Across Section Growth Growth Section Growth Growth Pre 1946 -0.308 0.183 -0.491 0.112 0.094 0.017 1946-1955 0.392 0.352 0.040 0.086 -0.325 0.411 1956-1960 -0.002 -0.042 0.040 -0.043 -0.019 -0.024 1961-1965 0.027 -0.393 0.420 0.026 0.022 0.004 1966-1970 0.052 -0.028 0.080 0.034 0.023 0.011 1971-1975 0.102 0.015 0.087 0.076 0.041 0.035 1976-1980 0.050 -0.049 0.100 0.095 0.068 0.028 1981-1985 0.108 -0.051 0.159 0.069 0.023 0.046 1986-1990 0.145 -0.008 0.153 0.075 0.062 0.013 1991-1995 0.042 0.132 -0.089 Assimilation is reported across two censuses. The change is broken down into growth within the cohort, true assimilation, and growth in returns compared to a cohort that spent the equal number of years since migration.

Table 2 shows the rate of assimilation of immigrants into the Canadian labour force. The cross-section column shows the estimated cross section assimilation from equation (4). This is the returns to earnings for the five year period (ten years for the 1971/1981 censuses) between the two censuses. This rate of return is always positive and varies across the years reaching a high of 17% in the 1981/1986 censuses and a low of 4% in the 1996/2001 censuses. This number however is made up of the within cohort growth, the true rate of assimilation, and the across cohort growth, the rate of assimilation relative to the previous cohort. Looking at the within growth column shows us how each cohort has actually assimilated into the Canadian labour force. With the exception of 1986-1991 and 1996-2001 the true rate of assimilation has been essentially zero for each cohorts first five years in the country. In these two periods we see much higher rates of assimilation across most of the immigration cohorts compared to other time periods. The cohorts that are not experiencing higher rates of

assimilation are the earlier immigration cohorts with 40 years or more in Canada. The across growth column shows the relative performance of each cohort with the one following it. The final row in this column compares the entry level earnings of these cohorts. It is interesting that in periods of slow assimilation the following cohort begins with much worse entry level earnings. In the 1986/1991 census period the across growth of 1981-1985 is negative and close to 0 indicating that the immigration cohort of 1986-1990 had similar or slightly higher entry level earnings. In the 1991/1996 census period the across growth of 1986-1990 was 15%, while the assimilation rate had dropped to close to zero, indicating that the immigration cohort of 1991-1995 had entry level earnings much lower.

4. Discussion
The returns to immigration seem to follow a boom and bust cycle. In some periods immigrants experience very little or negative assimilation, the bust, and in others all immigrants experience periods of increased assimilation. The overall assimilation rate is quite low with gains made during a boom often being lost during a bust. The highest rates of assimilation are for the immigrant cohort spending its first five years in Canada during a boom. The initial rate of assimilation is quite high for those cohorts and they better weather the following bust period. This finding of a boom and bust cycle corresponds with and ties together the findings of both Baker and Benjamin (1994) and Grant (1999). Baker and Benjamin examined the censuses from 1971, 1981, and 1986, all periods of a bust, and found that rates of assimilation ranged between -3 and 3%. Along with this low rate of assimilation, they found deteriorating entry level earnings. Grant used the 1981, 1986, and 1991, the last belonging to a boom period, censuses and found that for the immigration cohort 1981-1985 the rate of assimilation was much higher and that the entry level earnings of the following cohort, 1986-1990, had not fallen below the levels of the 1981-1985 cohort. From these findings she predicted that entry level earnings had levelled off and immigrants were beginning to perform better in the Canadian labour force. By taking a longer view and using more recent data, I have corroborated the results from both of these papers and found a slightly different trend. The entry level earnings of immigrants is continuing to deteriorate but during boom periods entry level earnings will improve or remain constant between immigration cohorts. Baker and Benjamin used data from a bust period and found a downward trend in return to immigrants to Canada. Grant added data from a boom period and found rightly that immigrant earnings and rate of assimilation improved over this period. However, it didnt remain at this level; the period of 1991/1996 shows another sharp 10

decline in entry level earnings and rate of assimilation. The following period, 1996/2001, was a boom period and while the entry level earnings of immigrants improved it did not return to the levels of the 1986/1991 boom period. The boom and bust periods do not necessarily correspond to the macro-economic boom and bust cycles of the economy. These booms and busts are related only to the level of earnings of immigrants. The boom could be caused by an immigrant cohort that is more highly endowed with some unobservable characteristic that allowed them to garner higher earnings. It could also be some specific effect of particular periods that favours immigrants over native Canadian workers. Based on these findings, current immigrants to Canada can expect to spend at least 20-25 years of their working lives earning less than native Canadians. Low rates of assimilation coupled with deteriorating entry level earnings means more recent immigrants may never reach parity with native Canadian counterparts. There are many potential causes to why the earnings of more recent immigrants are so far below native Canadian workers. Since the 1960s when changes were made to the Canadian immigration system there has been less emphasis placed on the skills immigrants have and more emphasis placed on family reunification. 1 The emphasis on skills now applied mainly to the independent class of immigrants but not to families. By placing less emphasis on skills and more on family reunion it could be that more recent immigrants are lacking in the unobserved qualities, ambition, talent, etc., that allow immigrants to assimilate into the work force more fully.
1

The changes to immigration in the 1960s also changed the ethnic composition of

Bloom, Grenier, and Gunderson, 4-5. 11

immigrants. There were fewer immigrants from the UK and Europe and more immigrants from Asia and Africa. Discrimination on the part of employers could be the cause that immigrants are earning less than native Canadians. Newer immigrants could also just be placing less emphasis themselves on earnings. The definition of success or the motivating factors for working could be different between earlier and more recent immigrants and so the level of earnings is not as important. Regardless of what causes the differential in earnings, it will be interesting and important to continue monitoring the earnings of immigrants. The immigrant population continues to rise and is a very important piece of Canadian labour policy. If the earnings of immigrants continue to fall in the future this key policy may be in jeopardy as immigrants will choose to live in other countries that can offer them a better return. It is also important that as the labour market is made up of more immigrants the decreased earnings does not have larger adverse effects on the economy.

12

Works Cited
Baker, Michael, and Dwayne Benjamin. "The performance of immigrants in the Canadian labor market." Journal of labor economics (1994): 369-405. Bloom, D. E., G. Grenier, and M. Gunderson. "The changing labour market position of Canadian immigrants." The Canadian Journal of Economics. 28, no. 4b (1995): 987-1005. Borjas, George J. "Assimilation, changes in cohort quality, and the earnings of immigrants." Journal of labor Economics (1985): 463-489. --, "Assimilation and changes in cohort quality revisited: what happened to immigrant earnings in the 1980s?." Journal of labor economics 13, no. 2 (1995): 201-245. Chiswick, Barry R. "The effect of Americanization on the earnings of foreign-born men." The Journal of Political Economy (1978): 897-921. Grant, Mary L. "Evidence of new immigrant assimilation in Canada." Canadian Journal of Economics (1999): 930-955.

13

Вам также может понравиться