Immigration: a Consumption Impact Assessment

IMMIGRATION: A CONSUMPTION IMPACT ASSESSMENT

Sohrab Abizadeh, Department of Economics, The University of Winnipeg

Nancy Zukewich Ghalam, Statistics Canada

ABSTRACT -

Immigration has always been of great significance to Canada. This study undertakes an examination of the demand-side impact of immigration on the Canadian economy. Consumption patterns, both aggregate and disaggregate, are estimated using the data base provided by the 1986 Statistics Canada Survey of Family Expenditures. Aggregate results indicate that immigrant consumption patterns approach those of the Canadian-born over time while the disaggregate analysis provides several exceptions to this finding. While immigrants tend to behave like their Canadian-born counterparts in areas of shelter, household operation and health cars, significant differences are observed In the majority of other categories analyzed.

1. INTRODUCTiON

As a nation that has historically relied on immigrant peoples, immigration has always been an issue of great significance to Canada. Immigration undoubtedly has both social and economic ramifications. It is not our intention in this paper to deal with the social issues or to evaluate the economic costs or benefits associated with immigration. [Akbari (1989) indicates immigrants to be of not benefit to the Canadian society. In fact, post-1946 immigrants generally consume less social services in proportion to taxes paid. The Economic Council of Canada (1991) also concludes that immigration promotes economic efficiency due to economies of scale. This report also addresses the social costs; and benefits of immigration.] Our interest is in the demand-side impact of immigrants on the Canadian economy, the supply-side having already been analyzed by Seward (1988, 1987a, 1987b) and Abbott (1988). More specifically, we will study the consumption behaviour of immigrants as compared to the Canadian-born population. [It is not necessary to deal with the economic theory of migration in this paper. Those interested in the subject are referred to Barias (1989).]

Most research regarding consumption theory has been undertaken in the national context, disregarding the distinct characteristics of sub-populations, geographical regions or other such socio-economic classifications. The importance of these variables to consumption pattern analysis is currently being considered. The work of McConnell and De1javan (1983) focuses an the retired household, Falvin (1981) studies the adjustment of consumption expenditure to changing future expectations, and Chen and Chu (1982) and Jenkins (1989) investigate the effect of a changing age structure on consumption. Few studies, however, consider the consumption of immigrants versus that of the national population although Marr (1987) and Marr and McCready (1989) have done some work in this area.

The objective of this paper is to investigate the economic impact of immigration into Canada by examining consumption behaviour. The paper begins with background information and a brief review of literature. This is followed by a section discussing the data and methodology employed in the analysis. Finally, the results and conclusions are reported.

II. METHODOLOGY AND DATA

The critical step in analysing the consumption expenditure pattern of immigrants versus the Canadian-born cohort is the employment of an appropriate consumption model. Rather than adopting a consumption function on an ad hoc basis, we used a comprehensive model based on most recent received theory. This function is as follows:

EQUATION (1)

[See Dombusch, Fischer and Sparks (1989 Ch.8). They maintain that this equation "combines the main features that are emphasized by modem consumption function theory" (p. 276).]

[The survey variable defined as "income from investment (var91) is used as a proxy for the wealth variable in the empirical estimation of the consumption function.]

[The value for disposable income was obtained from the survey by first adding gross income from wages and salaries(var089) and gross income from self-employment (var090). Finally, in order to obtain a not income value, this gross labour income was reduced using an estimate of tax rate, i.e. personal taxes (var194) divided by income before taxes (var088).]

Next, this model is modified to include important socioeconomic variables in order to investigate potential differences in consumption behaviour between immigrants and the Canadian-born population. Equation (1) is then rewritten in a general linear form as:

EQUATION (2)

The crucial variables in equation (2) are the X 's; factors other than income and wealth which can influence the consumption behaviour of households. According to Jenkins (1989), age of the head of the household ( AG ) may be used as one explanatory variable. Level of education ( ED ) and family size ( FS ) are also important in any consumption behaviour analysis (McConnell and Doijavan, 1983) as is the sex of the family head (SEK) (Marr and McCready, 1989). To include these characteristics, the empirical equation is restructured as:

EQUATION (3)

It is argued that disaggregating the immigrant cohort with respect to the year of arrival captures potential variations in consumption behaviour due to the length of residence in Canada. [See Marr and McCready (1989) for evidence.] In this study, four distinct immigrant sub-groups are created. These divisions are intended to reflect significant world economic and political events as well as length of Canadian residence. The immigrant sub-groups are as follows: 1) distant group: arrival prior to 1946; 2) mid-1 group: arrival 1946-1970; 3) mid-2 group: arrival 1971-1980; 4) recent group: arrival 1981-1986. This differentiates among immigrants who have been in Canada for a relatively long time, a moderate time, a short time and a very short time. The Canadian-born group is used as a reference category.

Although useful for a general comparison, an aggregate consumption function indicates little about actual patterns of expenditure. To further facilitate the comparison of immigrant and non-immigrant consumption behaviour, specific categories of expenditure must be considered. Under the variable heading "total current consumption," Statistics Canada defines fourteen major consumption items (see Table 2). Replacing the total expenditure variable ( C ) in Equation (3) by total expenditure on each of the consumption items ( E ) allows us to estimate a function for each item across the immigrant/non-immigrant classifications:

The linear functional form is as follows:

EQUATION (4)

The other variables in (4) are as previously defined.

The coefficients in equation (4) are estimated using Ordinary Least Squares (OLS). An previously indicated, the total Canadian-born cohort serves as the principal reference category. For the disaggragated analysis, we concentrate on the rate of change of the household's expenditure on each consumption item per unit change in the total current disposable labour income (Y_t^d ). Thus the coefficient (b₁₂ ) in equation (4) indicates the fraction of any change in disposable labour income to be spent on specific consumption items, or the marginal propensity to consume (MPC) for each item. [This is somewhat similar to the method used by Chen and Chu (1982) for evaluating consumption patterns of retired people.]

To meaningfully compare the MPC's of the different immigrant groups with those of the Canadian-born cohorts, a test of the difference between the coefficient ( b₁₂ ) of each consumption category is conducted. Differences observed for the disposable labour income variable (Y_t^d ) are tested for their significance. The calculated t-statics are reported in Table 2.

The data used for this analysis is derived from the Statistics Canada Survey of Family Expenditures for the year 1986, the most recent available at the time this study was carried out. It consists of a multistage stratified clustered sample, selected from the labour force survey sampling frame. A sample of 10,356 households in Canada reported how they allocated their total consumption expenditure among a large group of consumption items. For the purpose of this survey a household is considered to be a spending unit. [Statistics Canada defines a spending unit as a group of people living in the same dwelling who depend upon a common or pooled income for major expenses or one financially independent individual living alone.] Information about each household's demographic structure was also collected. The data provided a sufficient set of information to estimate a general form of the consumption function as specified in equation (3) above and the disaggregated consumption functions (equation (4)) for each expenditure category.

III. RESULTS

A. The Aggregate Consumption Function Approach

Table 1 reports the results of the regression analysis of equation (3) using Ordinary Least Squares (OLS).

The coefficient for the real wealth ( WR₁ ) variable carries the expected positive sign in all cases and is statistically significant at the 99% confidence level. As a whole, with the exception of only the distant group, immigrants demonstrate a higher marginal propensity to consume out of real wealth than do the Canadian-born. This is to be expected, particularly during the early years of Canadian residence as new immigrants may not have yet established a permanent source of income other than what is earned as a return on initial wealth. It is interesting to note that the coefficient for the mid-2 group (0.73) is double that of recent immigrants (0.36) and that even after 16-40 years of Canadian residence, the estimate is still higher than that of the Canadian-born cohort (0.28). The relatively high marginal propensity to consume out of real wealth for the mid-2 group may be due to certain changes in economic circumstances in Canada; during that period greater emphasis was placed on immigrants' ability to financially survive on their own. Overall, the immigrant cohort has a marginal propensity to consume out of wealth which is equal to that of the Canadian-born cohort.

The coefficient for current disposable labour income( Y_t^d ), or marginal propensity to consume (MPC), carries a positive sign and is significant in all cases as reported In Table 1. As a rule, immigrants, with only one exception, demonstrate a higher MPC out of labour income than do the Canadian-born; compare MPC=0.60 for all immigrants to 0.52 for all Canadian-born.

Earlier research by Nightingale (1978) may also be relevant to the differences observed between immigrant and Canadian-born consumption behaviour. Nightingale attributes these variations to the transition effect and cultural differential. This is best reflected in our analysis of the MPC values of disposable labour income. This estimate is close to that of the Canadian-born during the early years following immigration but exceeds that of the Canadian-born over time, as immigrant consumption behaviour conforms to the Canadian-born. In addition to the cultural factors, it is plausible that the immigrants show what may be labelled as "the standard of living catch-up effect," whereby immigrants consume heavily to bring their originally lower standard of living up to the normal Canadian levels. It is evident from the values of MPC out of disposable labour income that immigrants generally demonstrate a higher MPC than the Canadian-born population. This alone can have an important implication in terms of the economic effects of immigration. Immigrants may have a more significant role to play in times of recession when the economy needs a demand induced stimulus. A negative intercept for the equation estimated for the most recent immigrant group indicates that during the early years of Canadian residence, a large portion of income is saved. This may be due to-the lack of security and permanent income. [Note that the consumption functions estimated here are short-run consumption functions and thus conclusions about long-run behaviour are only possible by comparing different groups of immigrants.]

The estimated coefficient for lagged disposable income is significant in five of the seven cases reported. In all instances, the coefficient is positive although insignificant for the two middle groups of immigrants (mid-1 and mid-2). Nonetheless, the inclusion of this variable and the results obtained support the theory of consumption behaviour: its coefficient is positive and statistically significant when all Canadian-born and/or all immigrants are considered. Thus it can be concluded that lagged disposable labour income has a similar effect on consumption expenditures for ail consumers in Canada. This is particularly true if one considers that the coefficient for Y_t-1^d is very close in value for all cases reported (with the exception of recent immigrants) again supporting the hypothesis that immigrant consumption behaviour approaches that of the Canadian-born over time.

The variable representing the age of the family head produces mixed results. The coefficient is positive and significant for the Canadian-born, mid-1 and recent immigrant groups. It carries a negative sign for distant immigrants and mid-2 immigrants but it is only significant for the former. Since the distant immigrant group generally represents those who are close to retirement or already retired, the negative sign merely indicates the overall decline in consumption expenditures exhibited by most consumers, as explained by the life-cycle hypothesis. This is due to both reduced need and disposable income. The positive coefficients support the theory that the older the head of the household (before retirement age), the higher the consumption expenditure. This conclusion is more valid for the Canadian-born (version 1) than for the immigrant cohort as a whole (version 2) where the coefficient is positive but insignificant.

The education variable produces distinct results. In the ca of Canadian-born cohort, the coefficient is highly significant and positive. If higher education for Canadians results in improved employment opportunities and higher earnings, it is expected that this would lead to higher consumption expenditures. However, this is not shown to be true in the case of the immigrants. Thus it would seem that immigrant consumption patterns are influenced by other major factors such as culture and value of financial assets (Nightingale, 1978). In addition, this result indicates that, on the average, immigrant's level of education does not vary as much as that of the Canadian-born population and thus does not significantly affect consumption.

The coefficients for family size (FS) are all positive, as expected, and are statistically significant in ail cases except for the distant immigrant group. it follows that the larger the number of children in a household, the higher the consumption expenditure, whether we consider the Canadian-born or immigrant cohort. However, immigrants as a whole tend to spend less on additional members of the family than do the Canadian-born. Considering all immigrants (version 2) and all Canadian-born households (version 1), we see that the addition of one person to the immigrant household will raise their expenditure by $1305.61 as compared to $1420.08 for the Canadian-born household. This may reflect cultural differences. That the coefficient for FS is not significant for distant immigrants is not surprising since most of those in this group are already retired, or close to retirement age.

The dummy variable for sex (SEX) assumes a value of 1 if the head of the family is a male and 2 if the head of family is female. In general, the results for the Canadian-born group indicate that when the head of the family is female, household consumption is lower. This is an indication of the well-established fact that females are paid lose than their male counterparts in the workplace.[Statistics Canada (1992) data shows that in 1990, the female/male earnings ratio was 67.6%. Since 1967 this ratio has only increased by about 15%.] In addition, most female-headed families are ]one-parent families. [Statistics Canada (1990) reports that in 1990, 37% of all female-headed families in Canada had incomes which fell below Statistics Canada's low income cut-offs.] The same conclusion can not be made about the immigrant cohort. The coefficient varies from positive to negative and from significant to insignificant. However, an interesting trend does emerge. The positive and significant coefficient for recent immigrants is similar to that of the Canadian-born cohort. The results for the mid-2 and mid-1 cohorts are both insignificant and move from positive to negative as arrival year becomes more distant. Finally the coefficient becomes positive and significant for distant immigrants. From this we can conclude that female-headed households which immigrated from 1981-1986 had higher levels of consumption than those households headed by men in 1986. However, those female-headed households with 40 or more years of Canadian residence exhibited the same consumption tendencies as the Canadian-born reference, that is, lower overall consumption than male headed households. All this provides more support for the hypothesis that over time immigrants tend to behave more and more like their Canadian-born counterparts.

B. Disaggregated Expenditure Analysis

Table 2 reports the results of equation (4), the estimated MPC out of labour income (Y,) for each of the fourteen disaggregated consumption items. The MPC for all immigrants as a whole is shown to be significantly different (at the 5% level) from that of the Canadian-born cohort for all but five categories: shelter, household operation, health cars, personal care and tobacco and alcohol products. Immigrant MPC tends to be higher for all but the household furnishings and miscellaneous items. These observations support the general findings of the aggregate consumption function analysis that immigrants as a whole have a higher MPC out of disposable labour income than do the Canadian-born population.

It is interesting to note that the hypothesis that immigrant consumption behaviour conforms to that of the Canadian-born over time does not generally hold true when consumption expenditure is disaggregated. The MPC is significantly different for the largest number of consumption categories in the mid-1 immigrant group (arrival year between 1946 and 1970) and is significant in fewer and fewer categories as immigrant arrival year becomes more recent.

Among the necessities (food, clothing, and shelter) the MPC for food and clothing are significantly higher for the mid-1 and mid-2 immigrants and for the whole immigrant cohort when compared to Canadian reference. This may indicate that more expensive clothing and food (such as designer labels and restaurant meals) is purchased by these immigrant sub-groups as they become better established in Canada and seek a higher standard of living. In the shelter category, the distant immigrant group is the only one with coefficient estimates significantly different from the Canadian-born. This lower MPC can be attributed to the overall decline in consumption as one progresses through the life-cycle.

Two related categories are the household operation group and household furnishings and equipment. The only significantly different estimate for the former is observed in the mid-2 immigrant group where MPC is higher than that of the Canadian-born reference. In the cue of household furnishings and equipment, the estimated MPC is lower for virtually all immigrant groups. The only exception is the insignificant estimate reported for the mid-2 group. The significance of furnishings and equipment estimates and the insignificance of household operations coefficients may be attributed to the nature of the items included in each.

With respect to transportation, the immigrant MPC tends to surpass that of the Canadian-born for both public and private transport. Distant immigrants, however, exhibit a lower MPC for private transportation facilities than the Canadian-born. In contrast with the distant group, the mid-1 and mid-2 immigrant estimates are higher than those of the Canadian reference. The only significant estimate for the public transportation category is observed in the mid-1 group where the MPC is higher.

Results are varied for health and personal care expenditures. The health ewe coefficient is significant only for the distant immigrant cohort: MPC is lower than that of the Canadian-born reference. Personal care yields significant estimates for the distant and mid-1 groups. The MPC is lower for distant immigrants and higher for the mid-1 group as compared to the Canadian-born.

Education estimates are significant and exceed both references for immigrants in general and the mid-1 group in particular. The same results are estimated for the recreation category. The nature of these two expenditure categories suggest that now immigrants require at least 16 years of Canadian residence before they are financially secure enough to begin allocating more of income increases to private education and "luxury" recreation items. Recreation includes expenditures on items such as leisure equipment and vehicles, entertainment and travel tours. Study supplies, private lesson and tuition fees are covered by the education category.

The estimated MPC for reading materials is significantly higher for the whole immigrant cohort and for the distant and mid-1 sub-group compared to the results obtained for the Canadian born.

Miscellaneous category is the only category where the estimated MPC is lower than the Canadian-born reference for ail immigrant groups, irrespective of arrival year.

In general, the results for the mid-1 group are almost identical to those reported previously for the immigrant cohort as a whole: the categories for which the MPC is significantly different from that of the Canadian-born are reduced to only food, clothing and private transportation when considering the mid-2 immigrants, although the coefficient for household operation becomes significant while the personal ewe figure is no longer so. In the case of recent immigrants, only two coefficients are significantly different: the household furnishings and miscellaneous categories. Finally, the disaggregated consumption analysis confirms the generally lower IVIPC exhibited by distant immigrants in the aggregate. These estimates are significantly lower for the shelter, household furnishings, private transportation, health care, personal care and miscellaneous categories.

IV. SUMMARY AND CONCLUSIONS

In this study, we undertook a detailed comparison of consumption patterns of immigrants to those of the Canadian-born cohort. Using data from the Statistics Canada Survey of Family Expenditures for the year 1986, we estimated both aggregate and disaggregate consumption functions for the Canadian-born population and four distinct groups of immigrants, clustered by arrival date.

Our findings, based on the aggregate consumption function estimated, are summarized as follows:

In general, differences do exist between the consumption pattern of the Canadian-born and the immigrant population. First, immigrants tend to demonstrate a higher marginal propensity to consume out of disposable labour income than does the Canadian-born cohort. Secondly, immigrants show a lower marginal propensity to consume out of wealth than do the Canadian-born. Thirdly, education is not a major determinant of the level of consumption by immigrant groups. Fourthly, although the number of children positively affects the level of consumption expenditure for immigrants and the Canadian-born alike, our results indicate that immigrant families generally spend less on additional family members than do their Canadian-born counterparts. Finally, our results strongly support the hypothesis that over time immigrant consumption patterns tend toward that of their Canadian-born counterparts.

Disaggregated consumption analysis suggests that immigrant consumption patterns are actually less divergent from the Canadian norm upon immigration than they are after 6 or more years of Canadian residence. The difference observed between the aggregate and disaggregate consumption analyses can be attributed to the nature of the expenditure categories for which the observed coefficients do not differ from those of the Canadian-born: the shelter and household operations categories. These represent a proportionately large share of total household expenditure, thus exerting greater influence on the coefficients observed in the aggregate. Thus upon examination of a more detailed break-down of consumption, it is clear that immigrant expenditure behaviour is significantly different. We can conclude that immigrants tend to stimulate the demand-side of the Canadian economy differently than do the Canadian-born.

Aggregate analysis supports the theory of immigrant consumption behaviour approaching the Canadian norm over time. By the same token, disaggregate investigation seems to confirm the assertion that "although some differences between Canadian- and foreign-born disappear when factors such as age, income, education, gender, family size are hold constant, many differences still remain. Some non-differences become significantly different over time." [Marr and McCready (1989), p. 6.] The findings of this study reinforce the general hypothesis that the demographic characteristics of households are as important in influencing expenditure patterns as income and other socio-economic factors.

TABLE 1

SUMMARY REGRESSION ANALYSIS OF CONSUMPTION FUNCTION

TABLE 2

ESTIMATED MARGINAL PROPENSITY TO CONSUME (MPC)

REFERENCES

Abbott, Michael G. (1988), "Immigration and Labour Market Adjustment" in Beach and Green (eds.) (1988) pp. 27-39.

Akbari, Ather H., (1989) , "The Benefits of Immigrants to Canada: Evidence on Tax and Public Services," 4. Canadian Public Policy, 5:424-435.

Bodes, George J. (1989), "Economic Theory and International Migration," International Migration Review, 23:457-485.

Chen, Yung-Ping and Chu, Kwang-Wen (1982), "Household 5. Expenditure Patterns: The Effective Age of Family Head," Journal of Family Issues, 3:233-251.

Dombusch, Rudiger, Fischer, Stanley and Sparks, Gordon R. (1989), acroeconomics third Canadian Edition, McGraw Hill Ryerson Limited, Toronto.

Economic Council of Canada, (1991), Economic and Social Impacts of Immigration, Supply and Services Canada.

Kavin, Marjorie (1981)"The Adjustment of Consumption to Changing Expectations About Future Income," Joumal of Political Economy, 89:974-1009.

Jenkins, Paul (1989), "Effects of Changing Age Structure on Consumption and Saving." Department of Finance Canada, Working Paper No. 89-05.

Marr, William L (1987), "Consumption and Saving Patterns of the Foreign-born in Canada," A discussion paper, The Institute for Research in Public Policy, 87. A.9.

Marr, William L and McCready, Douglas J. (1989), 'The Effects of Demographic Structure on Expansion 9. Patterns in Canada," A discussion paper, The Institute for Public Policy, June.

McConnell, Charles E. and Doijavan, Rrooz (1983) "Consumption Pattern of the Retired Household," Journal of Gerontology, 38:480-490.

Nightingale, J. (1978), Migrant Household Economic Behaviour, Australian National University Press.

Seward, Shirley B. (1988) "Policy Context of Immigration to Canada," in Policy Forum on the Role of Immigration in Canada: Future, Charles M. Beach and Alan G. Green (eds.).

Seward, Shirley B. (1987a) "Demographic Changes and the Canadian Economy: An Overview." Discussion Paper on the Demographic Review 87 A.0. (Ottawa: The Institute for Research on Public Policy).

Seward, Shirley B. (1987b) "The Relationship Between Immigration and the Canadian Economy" Discussion Paper on the Demographic Review 87 A-10 (Ottawa: The Institute for Research on Public Policy).

Statistics Canada, (1992) Earnings of awn and women 1990, Catalogue 13-217, Ottawa.

Statistics Canada, (1990) Income distributions by size in Canada 1989, Catalogue 13-207, Ottawa.

THE AUTHORS WOULD LIKE TO ACKNOWLEDGE COMMENTS ON THE ORIGINAL DRAFT BY PROFESSOR DAVID CHEAL AND CATHERINE PETERS.