Economic impacts of skilled immigration JOHN SALERIAN ∗ Productivity Commission Melbourne, Australia Abstract In this paper, we report on an economic analysis of the effects of a hypothetical increase in skilled immigration. Some of the effects are more amenable to measurement and estimation than others. Effects that cannot be reliably measured or estimated might still be significant. We find that positive effects arise from higher participation rates, slightly higher hours worked per worker and the improvement in skill composition of the workforce. Some of the economy-wide consequences contribute negatively to per capita income, such as a lower capital to labour ratio and a decrease in the terms of trade. The overall effect of an increase in skilled immigration on average per capita income is estimated to be positive. Address for correspondence Dr John Salerian Assistant Commissioner Productivity Commission LB2 Collins Street East Post Office Melbourne Victoria 8003 Telephone 61-3-9653-2190 Facsimile 61-3-9653-2302 E-mail [email protected]
∗ Paper presented at the 35th Australian Conference of Economists, 25-27 September 2006, Curtin University of Technology, Perth, Western Australia. The views expressed in this paper are those of the author and do not necessarily reflect those of the Productivity Commission. This paper draws on the work undertaken by the Productivity Commission for its study into the Economic Impacts of Migration and Population Growth (PC 2006). The content of this paper reflects the work of the following current and former Commission staff: J Sloan, M Kirby, J Pincus, P Jomini, S Bailey, D Forwood, J Fry, C Decker, M Forbes, M Kohlhaas, A Montgomery, J Novak, S Sandhu, T Simovski, A Barker, G Cuxson. J Giesecke and T Meagher at the Centre of Policy Studies (Monash University) also contributed through their consultancy with the Commission to run the Monash Model. Finally, P Miller (UWA), P Robertson (UNSW) and R Tyers (ANU) contributed through their participation as independent referees on the modelling.
In recent times, policy makers have been concerned about the effects of an ageing and slower growing population on future growth in labour supply and productivity. For example, the Australian Government asked the Productivity Commission to examine the productivity, labour and fiscal implications of likely demographic trends over the next 40 years (PC 2005). Concern about the economic effects of an ageing population has also stimulated interest in evaluating policy options to deal with it. Immigration, and skilled immigration in particular, is often cited as one such policy. In this paper, we present an analysis of the overall economic impact of a hypothetical permanent increase in the level of skilled immigration. The analysis reported in this paper was undertaken by the Productivity Commission for its recent report on the economic impacts of migration and population growth (PC 2006). The nature of Australia’s immigration has changed over the last decade. The number of immigrants arriving for permanent settlement under the Migration Program 1 has steadily increased. Between 1997-98 and 2005-06, the number of visas granted for permanent settlement increased from about 74 000 to about 140 000. Over this period, the Australian Government also increased the proportion of immigrants entering under skilled visas. Between 1997-98 and 2005-06, the number of immigrants in the Skilled Program increased from about 35 000 to about 98 000. Over the same period, the number of immigrants in the Family Program increased from about 31 000 to about 42 000. There is also an increasing number of skilled workers on temporary business visas. Between 1996-97 and 2004-05, the number of persons entering Australia annually on temporary business visas increased from about 314 000 to about 390 000. In order to evaluate the economic impacts of an increase in immigration, particularly skilled immigration, on productivity and economic growth, a framework is required that sets out the potential link. The framework used by the Commission is outlined in sections II and III. II
Definitions of productivity and growth concepts
Positive net migration is likely to result in a faster growing population and economy. It could also change the structure of the economy, depending on the extent to which the characteristics of immigrants differ from those of the pre-existing residents. Given the focus on skills in Australia’s immigration program, 1 Humanitarian immigrants (about 13 000 in 2004-05) are not included in the Australian Government’s definition of the Migration Program. Similarly, the Migration Program excludes immigrants from New Zealand, which were about 22 000 in 2004-05.
the economic effects of immigration are also likely to vary across industry, occupations and geographical locations. However, the changing size and structure of the economy, per se, is not the central issue for this study. The absolute sizes of the economy and population are less relevant when considering living standards 2 . Growth measured in per capita terms is most relevant to the standard of living of individuals and is more informative for policy making purposes. Therefore, the focus in this study is explicitly about how immigration is linked to income per capita and productivity. These links also provide insights into how immigration affects the distribution of income between pre-existing residents and new immigrants. For the purpose of this study, living standards are measured using indicators of real income per capita and real consumption per capita. The measures used include real Gross Domestic Product (GDP) per capita, real Gross National Product (GNP) per capita and real aggregate consumption per capita. GNP is equal to GDP plus Net Foreign Income and provides a measure of the income accruing to Australian residents. Using GDP and related measures allows us to identify and quantify many of the economic impacts of migration, notwithstanding the limitations of using such measures as indicators of welfare 3 . More specifically, productivity growth is expressed as labour productivity, typically measured as real GDP per hour worked. This follows the conventional approach used to study economic growth at the macroeconomic level. (i)
Whose living standards are we assessing?
The economic effects of immigration could be assessed from the perspective of various groups in the Australian population, including Australian-born, pre-existing immigrants, pre-existing residents (Australian-born plus pre-existing immigrants), new immigrants, or all of the above. The effect of new immigration on pre-existing residents is an appropriate policy focus. However, estimating the income and wealth distribution consequences of immigration for pre-existing residents and new immigrants is more complex than estimating the effects on average income per capita. Income changes can occur through many pathways including the systems of transfer payments, provision of government services (health and education), all forms of taxation, the market determination of the returns to capital and labour, and changes in wealth. Estimating the income distribution effects on pre-existing
2 For other matters, such as national security, the size of the economy and the population might be more relevant. 3 Migration can also have important social and cultural effects on society, which are not considered in this study.
residents and new immigrants would require ‘tracking’ the outcomes for both groups. This was not possible within the time and resources available for this study. So, the principle focus in this study is on the average income per capita, including new immigrants. III
Theoretically linking immigration to productivity and income per capita
It is a complex task to identify the ways in which immigration is linked to productivity and income per capita. The approach used in this study is to synthesise the theory and principles found in the fields of economic growth, macroeconomics and the economics of labour markets. Adopting this approach also assists in clustering the link into thematic groups. The main links through which migration can affect productivity and living standards are briefly discussed below. The thematic grouping of the links discussed below reveals a large number of complex factors. However, the relevance of each link is essentially an empirical matter. (i)
Supply and composition of labour
Skilled immigration increases both the total supply of labour and the skill or occupational composition of labour in the economy. Immigration also contributes to an increase in the size of the population. Growth theory suggests that it is labour supply per head of population that is relevant to understanding growth in income per capita (Weil 2005). The effects on labour supply per capita can be understood more easily by expressing hours worked per capita as the product of the ratios shown in equation (1). Employment Hours worked Hours worked Working age population Labour force = Population Population Working age population Labour force Employment
Immigration could alter these ratios with the net result dependent upon the labour market outcomes for both new immigrants and the pre-existing population.
In addition to changing the hours worked per capita, skilled immigration can change income measured as GDP per capita by changing the skill composition and labour productivity of labour supply as shown in equation (2). GDP Hours worked Wage adjusted hours worked GDP = Population Population Hours worked Wage adjusted hours worked = average hours worked × skill composition × labour productivity
Increases in the level of skilled immigration would be expected to lead to an increase in the labour value added per hour worked (a ‘skill composition effect’) assuming that skilled immigrants earn salaries and wages commensurate with their
higher skill levels. This effect would also require new immigrants to be more highly skilled on average than the pre-existing workforce. The overall outcomes for labour supply arising from immigration are determined by three principal factors, outlined below. •
The number of immigrants arriving in various visa categories relative to the size of the pre-existing population.
The extent of differences in characteristics (demographic and economic) between immigrants of various visa categories and the pre-existing population. The self-selection of immigrants entering under various visa categories (Skilled, Family or Humanitarian) can modify the labour market attributes of immigrants because the motivation for immigration can differ between visa categories (Chiswick, B. 1999).
The labour market adjustment to the increase in the supply of labour arising from immigration. In Australia, immigration initially changes the supply of labour by occupation, educational attainment, skills and region of immigrant settlement. The demand for the various types of labour is also likely to differ across industries.
The labour market adjustments across industry sectors could vary depending on: •
the labour intensity of the industrial sectors
substitution possibilities between new immigrants and pre-existing workers (Chiswick, C.U. 1989; Grossman 1982)
substitution possibilities between skilled and (Chiswick, C.U. 1989; Hamermesh and Grant 1979)
substitution possibilities between occupations
the relative growth of sectors arising from the reallocation of sectoral activity (Appleyard and Field 1992) as the economy adjusts to the increase in labour supply and its composition
any labour market rigidities.
To reach a new equilibrium with a higher ratio of skilled to unskilled workers in the economy, the marginal physical product (and real wage) of skilled workers would be expected to decrease, all else equal. Correspondingly, the marginal physical product (and real wage) of unskilled workers would be expected to rise. An analysis of the labour market adjustment arising from immigration should consider a relatively high degree of labour market segmentation (for example, labour differentiated by skill and occupation) and industry segmentation. This is
especially the case for Australia, where immigration policy targets immigrants in specified occupations and skills. (ii)
Growth theory suggests that macroeconomic conditions regarding savings and investment are important determinants of the level of labour productivity and income per worker (Barro 1997; Benge and Wells 2002; Romer 2001; Weil 2005). If the rate of net investment is unable to sustain the capital to labour ratio with a faster growing workforce arising from immigration, then labour productivity and real wages could decrease, all else equal. This process is referred to as capital dilution. Immigration can increase the supply of labour in diverse ways across skill levels, occupations and industries. The resulting adjustment of the capital to labour ratio is likely to differ across industries. The extent of the adjustment depends on the initial capital to labour ratios (labour intensity), the elasticities of substitution between labour and capital, and the sectoral reallocation of economic activity. A decrease in the capital to labour ratio has often been used to argue that immigration leads to a decrease in real wages (Borjas 1995). To bring about equilibrium in the economy with a lower capital to labour ratio, the marginal physical product of labour and real wages decrease, all else equal. Correspondingly, the marginal physical product of capital and the rate of return on capital increase. The decrease in the real wage does not necessarily mean that the income per capita of pre-existing residents decreases (Borjas 1995). The higher return on capital can contribute to higher income per pre-existing resident (to the extent that the capital is owned by pre-existing residents). The net result of these two effects determines the overall outcome. The effect of capital dilution is likely to be smaller for a small, open economy, such as Australia. A higher rate of investment can be met from foreign investment if the rate of domestic saving and investment is too low. In this environment, there would be more limited long-term effects on real wages and rates of return on capital, all else equal. There is another way that the capital to labour ratio could be affected. There is likely to be an expansion in exports arising from the larger economy. This expansion in exports could lower the prices received for exports, and lead to a decrease in the terms of trade. To the extent that domestic rates of return on capital are determined by international markets, the marginal physical product of capital would need to rise in order to raise the marginal value product of capital. In turn,
this would decrease the marginal physical product of labour, real wages, the capital to labour ratio, and labour productivity. The accumulation of capital can also be affected by the amount of capital that immigrants bring with them, and the extent to which they send remittances to their country of origin (Brown 2005; Lopez-Cordova and Olmedo 2005). In addition, the marginal value product of capital might increase as a result of changes in the construction price of capital. For example, the cost of construction could increase if wage rates in the construction industry increase in response to the increase in demand for investment. Finally, the faster rate of net investment could provide an opportunity to adopt new technology at a faster rate, thereby contributing to higher productivity and living standards. (iii)
Economies of scale and competition
Immigration contributes to a larger population. If economies of scale are present, then the increase in population size could contribute to an increase in productivity and income per capita. This source of productivity is often cited as a potential benefit of immigration (BCA 2004; Corden 2003; Foster 1996; Garnaut 2002; and Withers 2003). An important issue is the extent to which the increase in population translates into an increase in demand for the goods or services of those sectors that exhibit economies of scale. Corden (2003) suggests that the link between economies of scale and the size of the population is most relevant in sectors producing goods and services that are not traded internationally (for example domestic transport, communications and public administration). Population size (and domestic demand) could be less relevant to internationally traded goods and services sectors because economies of scale can be achieved through international trade. A caveat here, raised by both Corden (2003) and Withers (2003), is the extent to which transport costs are impeding exports. Australia’s transport costs could be high due to the ‘tyranny of distance’. In this case, a larger domestic market might provide a platform to lower unit costs of supply, thus offsetting high transport costs and providing a platform for breaking into export markets. Whether this occurs depends, in part, on the extent to which economies of scale arise from a modest increase in the size of the population based on levels of immigration observed in Australia.
There are other ways in which a larger population could affect productivity. These effects arise from those externalities that are themselves manifested as economies of scale. The externalities arise from the higher density of population and economic activity, and are often referred to as ‘agglomeration’ or ‘thick markets’ effects. Agglomeration of economic activity can: •
increase knowledge spillovers between firms (Morrison-Paul and Siegel 1999)
lead to economies of localised industry arising from shared inputs in production (repair, accounting, legal) (Quigley 1998)
exploit greater specialisation between firms and any economies of scale in local transport networks (Ciconne and Hall 1996)
reduce transaction costs (matching work skills to job requirements) (Quigley 1998)
increase the adoption of new technology by increasing the domestic capability to undertake research and development (Keller 2002)
reduce risk arising from diversity of customers and non-coincident fluctuations in sales (Quigley 1998).
Finally, a larger population (and domestic market) might improve productivity by enhancing competition in domestic markets, by supporting a greater number of firms which compete against each other (BCA 2004; Corden 2003). This could reduce the misallocation of resources arising from monopolistic pricing behaviour. (iv) Natural resources, land and environmental externalities
Increasing the size of the economy increases the demand for natural resources (fixed and renewable) and land. Growth theory suggests that if the supply of these resources is limited, then it can result in lower labour productivity and income per capita (Romer 2001). The effect is similar to that of capital dilution described above. Linking migration to the use of natural resources and its productivity effects raises similar issues to those discussed in economies of scale. Natural resources and land can be used in the production of most goods and services. However, in the context of immigration, it is their use in the production of non-traded goods and services that is most relevant. Immigration and a larger population contribute to an increase in demand for non-traded goods and services. An exception arises if immigration leads to an expansion of exports. This could increase the demands for natural resources embodied in the exports.
If the supply of natural resources cannot keep pace with growth in their demand (regardless of the source of demand), then the price for natural resources rises. Over time, the increase in price encourages substitution to alternative inputs and the development of new technologies. These developments allow the economy to decrease the intensity of the use of natural resources (Romer 2001; Weil 2005). Substitution and the development of new technologies can reduce the drag on productivity arising from the scarcity of natural resources and land. Nevertheless, a drag remains, and its size depends on the ease of substitution between natural resources/land and other factors of production. The potentially lower productivity arising from limited natural resources and land needs to be balanced against the other productivity effects of migration and population growth. The links between population size and environmental externalities, such as congestion (traffic congestion, congestion in national parks) and pollution, are similar to those described above for natural resources. The main difference is that the effects of these externalities are not necessarily reflected in market prices (Clarke and Ng 1993; Weil 2005). Thus, government intervention might be required to internalise these effects, such as establishing property rights, pollution taxes, subsidies or regulations. The avoidance of pollution and congestion can moderate the decrease in living standards and productivity. However, the abatement and enforcement costs must also be considered. The overall effect depends on both internalising the externality and the ease of substitution of technologies that have lower congestion and pollution (costs of abatement). Once again, this decrease in productivity from environmental externalities needs to be balanced against the other productivity effects of immigration and population growth. The income redistribution effects arising from the increasing relative scarcity of natural resources and land are likely to favour pre-existing residents. They accrue as increasing resource rents to pre-existing residents, if they own the resources (Clarke and Ng 1993). The exception would be the dilution of per capita resource rents for those resources embodied in exports, where prices are set by international markets (Garnaut 2002). The income redistribution effects arising from internalising environmental externalities depend on the level of the externality, the policies implemented to internalise them and their costs of abatement. In other words, it is difficult to determine the effects without knowledge about the specific policies implemented by government to solve the problem.
Government expenditure on services, transfer payments and taxation
The extent to which the characteristics of the new migrants differ from the pre-existing population has the potential to alter: •
the mix of goods and services provided by all levels of government (for example, education and health)
the mix and levels of the various transfer payments made by all levels of government (for example, social security)
the level of taxation revenue collected by all levels of government.
Governments use fiscal policy to achieve a number of objectives, such as macroeconomic management of the economy, income redistribution and economic growth. However, only the efficiency effects of changes to government expenditures and revenues are relevant to productivity and income per capita. According to growth theory, it is possible for fiscal policy (taxation and government expenditure) to have long-run implications for economic growth (Zagler and Durnecker 2004). For example, positive externalities could exist in education, health, research and development, and infrastructure. In such circumstances, there could be a role for government to facilitate provision so that the benefits of these externalities are captured (Weil 2005; Zagler and Durnecker 2004). On the revenue side, taxes can distort private decisions with respect to labour supply and capital accumulation. These distortions can alter the productivity of the economy (Diewert and Lawrence 1995; Freebairn 1995). On the other hand, distortionary taxes can also be used to internalise any externalities arising from private decisions regarding the accumulation of human and physical capital, and research and development. To illustrate how immigration might affect efficiency, consider an increase in skilled immigration. This could increase income tax revenue because a larger workforce per head of population is earning higher wages. It could also reduce transfer payments through a reduction in the proportion of the population receiving social security. Offsetting this would be an increase in the provision of goods and services. If the net result were a surplus, then the government might consider reducing tax rates to maintain the budget balance. This could reduce the deadweight loss of taxation and improve productivity. Alternatively, the government might choose to increase the provision of services, increase savings or retire debt. The efficiency
effects depend on the government response to the initial changes in expenditures and revenues, and the extent of any externalities and distortions. The nature of the taxation and transfer payment systems is also pertinent to income redistribution effects. In the above example, pre-existing residents would benefit from a reduction in the tax rate. (vi) Sectoral reallocation of economic activity
The labour supply characteristics of immigrants can lead to a reallocation of economic activity between the various sectors of the economy. The change in the relative size of each sector can affect the growth rate in productivity at the aggregate level. This is because the aggregate rate of productivity growth is a weighted average of the sectoral rates of growth in productivity, with the weights equal to the relative size of each sector. Generally, rates of productivity growth differ across sectors. The reallocation of economic activity across sectors changes the weights assigned to each sector. The aggregate rate of productivity growth could increase if the reallocation of activity were to result in higher weights for sectors with higher rates of productivity growth (Gollop 1985; Weil 2005). It is also possible for the restructuring of the economy to contribute to a lower aggregate rate of growth in productivity. (vii) Trade and technology transfer
Immigrants, through their links to and knowledge of international markets, might facilitate access to more valuable markets, causing a reallocation of the nation’s resources to exports that have a higher marginal value. Immigrants might also possess skills that facilitate the importation and adoption of new technologies, thereby raising productivity and income per capita. It might also be possible for migrants to transfer their skills to existing residents (Stromback 1994). The skill transfer could occur through training provided by skilled migrants, cross-cultural skills and the imparting of attitudes favourable to increased efficiency. IV
From the preceding discussion, there are many factors to consider in evaluating the effects of an increase in skilled immigration on productivity and growth in per capita income. The overall outcome depends on the sum of the contributions, with the magnitude of the effects essentially being an empirical matter.
Partial analysis of each link provides insights into the effects of each factor. However, such partial analysis is generally inappropriate for drawing inferences about the overall economic effects. To incorporate as many of the links as practicable, the Productivity Commission developed a series of integrated analyses and models to estimate the overall economic effects based on plausible assumptions about the relevant links. The analytical framework included labour market analysis, demographic modelling and general equilibrium modelling. However, quantifying all of the links was not practicable. Consequently, there remain a number of important factors that could contribute to an increase or decrease in the estimates reported. To help understand the economy-wide effects of immigration, particularly skilled immigration, the Commission simulated the effects on the economy of a hypothetical permanent increase in the annual intake of skilled migrants (about 39 000 or 50 per cent of the level in 2004-05). The three main steps in the analysis are set out in figure 1. Figure 1
Analytical framework used to estimate the economic effects of a hypothetical increase in skilled immigration Step 1 (Labour market analysis)
Estimate of the effects of demography, educational attainment, visa category and other migrant specific factors on participation rates, unemployment rates, hours worked and hourly income of immigrants. The analysis is used to calibrate the demographic and labour supply model used in step 2.
Step 2 (New Arrival Tracker demographic and labour supply model) Make cumulative projections of the increase in labour supply (hours worked) and population arising from a permanent increase in the level of skilled immigration, disaggregated into 67 categories based on levels and fields (disciplines) of education.
Step 3 (Dynamic General Equilibrium Model) Use the Monash Model to estimate the cumulative effects of the increase in labour supply from step 2 on the economy as a whole (GDP and GNP per capita), as well as employment and wages in occupations and industries, the capital to labour ratio across industries and in the aggregate, and the allocation of economic activity across industries.
Labour market analysis of immigrants
To gain insight into the factors influencing participation rates, unemployment rates, hours worked and hourly income, the Commission undertook regression analysis (using categorical independent variables). In particular, the objective was to shed light on general factors affecting labour supply and whether immigrants have
outcomes different to Australian-born, all else equal. The regression analyses were based on unpublished Census data. The categorical variables in the regression equation were based on human capital theory and previous empirical work (Borjas 1986; Chiswick, B. 1978; McDonald and Worswick 1999; Stromback 1983; Tran-Nam and Nevile 1988). The range of explanatory variables considered were age (as a proxy for work experience), level of post-school education, gender, region of residence (‘capital city’ or ‘other’), industry of employment, occupation of employment, immigrant status, and specific immigrant characteristics such as English ability and length of residency in Australia. (i)
Age profile of immigrants
As is demonstrated below, age and gender are important demographic characteristics affecting labour supply. One way in which skilled immigration contributes to changes in labour supply per head of population is by their age profile being different to the pre-existing resident population. The age distributions of the Australian-born and new-immigrant populations are shown in table 1. In the skilled visa group, there is a high proportion of persons of prime working age as well as a significant proportion of children. Table 1
Immigrant population Australian-born 2001 population 2001
Age Under 15 15 to 24 25 to 44 45 to 64 65 plus
Age distribution of immigrant and Australian-born populations Migration and humanitarian programs 2004-05 Skilled
5.2 9.5 33.4 34.2 17.7
24.8 14.9 29.3 20.1 10.9
21.7 15.4 55.6 7.2 0.2
13.6 16.6 50.1 13.6 6.1
37.4 23.6 31.2 6.9 0.9
20.7 16.6 51.4 9.2 2.1
Participation rates of immigrants
The factors influencing participation and differences between participation rates of immigrants and Australian-born persons are shown in table 2. The analysis indicates that participation rates for immigrants depend on compositional factors such as education, location, gender and age. They also depend on immigrant specific factors, such as English ability, years of residency in Australia and simply being an immigrant. Participation rates for skilled immigrants are likely to be higher compared with the Australian-born population because such immigrants are more highly educated, more likely to live in cities and more likely to be male. However,
lower proficiency in English of some immigrants can reduce participation. The participation rate can also be lower in the early years after arrival, indicating that it can take time to adjust to conditions in Australia. Table 2
Participation rate regression resultsa,b
Adjusted R-square Participation rate for 25 to 44 year old males, Australian-born, no post-school qualifications, in a capital city Age 15 to 24 45 to 64 65 and over Education Postgraduate level Bachelor degree/diploma level Certificate level Other demographics Live in a regional area Female Immigrant English ability of immigrants Not well or not at all Very well or well Years since immigrants’ arrival in Australia Recent (