4 Economic impact of offshoring

View the most recent version.

In this section, we examine the impact of offshoring on economic performances. We focus on the impact of offshoring on productivity growth, shifts to high-value activities and labour markets.

Earlier studies have examined the effects of offshoring on labour markets (Feenstra and Hanson 2001 for a review). The evidence for the United States suggests that offshoring has increased the demand for more skilled workers; but there is no consensus on the size of the effects of offshoring. Feenstra and Hanson find that offshoring can account for as much as three-fifths of the increase in the relative wages of non-production and production workers in U.S. manufacturing from 1979 to 1990. Other studies conclude that offshoring and trade in intermediate inputs are not important factors behind the increase in the relative demand for more- skilled workers.

More recent studies have focused on the effects of offshoring on employment. Some estimates suggest that offshoring has led to a large reduction in employment of white-collar workers in the United States (McCarthy 2002). But most empirical studies suggest that the overall impact of offshoring on employment levels is small (Amiti and Wei 2005; Mankiw and Swagel 2006).

A number of recent studies have also examined the effects of offshoring on productivity performance. The limited empirical evidence suggests that material offshoring has a positive effect on productivity growth. The evidence on the effects of service offshoring on productivity growth is mixed. But the preponderence of evidence suggests that service offshoring has little effect on productivity growth, possibly due to low levels of service offshoring compared with the level of material offshoring (see Olsen 2006 for a review). An exception is the paper by Amiti and Wei (2005) that finds that service offshoring has a significant positive effect on productivity in U.S. manufacturing, accounting for 11% of productivity growth for the 1992-to-2000 period.

The model that motivates our analysis of offshoring starts with an aggregate production function that expresses gross output as a function of intermediate input and value added, and value added is represented in turn as a function of capital input, labour input and time. The existence of the value-added aggregate requires that time and capital and labour inputs are separable from intermediate inputs.⁷ This type of production function has been used most frequently in previous studies on productivity (see, for example, Jorgenson, Gollop and Fraumeni 1987).

The aggregate production is written as:

where Y denotes gross output, A denotes value-added, and X denotes intermediate inputs.

Value added is taken to be a function of capital input, labour input and time:

Where K denotes capital input, L denotes labour input. B(t) is multifactor productivity and measures the efficiency with which capital and labour are combined in producing value added.

We assume that the value-added function exhibits constant returns to scale, and gross output is a constant elasticity of substitution production with respect to value added and intermediate input:

where, σ is the elasticity of substitution between value added and intermediate input and α is a production parameter that indexes the share of value-added in gross output.⁸

While intermediate input in this formulation is separable from capital input, labour input and time, the composition of intermediate inputs is taken to affect the value-added function. Our hypothesis is that the share of imports in material and services inputs affects multifactor productivity B(t), and the index of the ratio of value added to gross output α .

There are several ways in which this can occur. In a world where specialization allows for superior products to be produced by independent contractors, use of these products rather than inferior, internally produced substitutes will improve efficiency. Or, the outside components may allow for superior products to be developed.

A portion of gains in productivity from offshoring should also be expected to come from plant turnover—the closedown and decline in less efficient plants that used to produce the inputs that are offshored.

Offshoring may have a positive effect on productivity performance through specialization and restructuring. Outsourcing occurs when firms opt to buy rather than make things in-house. Offshoring is a special form of outsourcing when the suppliers are located in foreign countries. Outsourcing and offshoring may lead to greater specialization and restructuring. If so, we would expect to find a positive relationship between outsourcing (offshoring) and productivity.

It is also possible that changes in input composition will allow a firm to move up the value-added chain. Production processes that are carried on within plants differ in terms of complexity—from those that require little skilled labour and/or transform raw materials at an early stage of the production process to those that require more capital, both machinery and equipment, and more highly skilled workers.

Sako (2006) makes a useful distinction between vertical disintegration of inputs and the unbundling of corporate functions in his discussion of outsourcing and offshoring. Vertical disintegration happens when a firm decides to buy rather than make, and when it is cheaper to source inputs that go into a firm's final product in the market rather than make them in-house.

For example, car manufacturers used to make many car components in-house at their parts divisions. But, by spinning out the parts divisions, car manufacturers now source major sections of the car—such as seats—from independent suppliers.

While material-input outsourcing is very much about the disintegration of inputs, service outsourcing is as much about the unbundling of corporate functions as vertical disintegration. Modern corporations have corporate functions, such as finance and accounting, human resources, sales and marketing, purchasing and supply, and research and development. While many processes in these functions are outsourceable, corporations typically outsource low value-added processes. High value-added processes are kept within the corporation.

For example, in human resource outsourcing, low value-added processes such as payroll and benefits administration are typically outsourced, while human resources strategy and other high value-added activities are not outsourced. In the finance and insurance functions, general accounting is the typical example of outsourcing that is given. But financial strategy and other high value-added activities are less likely to be outsourced.

All this suggests that an important effect of service offshoring is to shift to higher value-added activities. The shift to higher value-added activities as a result of offshoring will show up as an increase in the employment of knowledge occupations, or an increase in capital intensity— particularly high-technology capital intensity—of the operations, or a higher ratio of value added to intermediate input.

While our discussions so far have focused on the effect of service offshoring on shifts to higher value-added activities, material offshoring is also expected to have an effect on the size of higher value-added activities within firms. The effects of material offshoring on a shift to higher value- added activities may not be as strong as the effect of service offshoring. While offshoring of material inputs to developing countries involves inputs that are less capital- and skill-intensive, it is not obvious that the offshoring of material inputs to the United States and other developed countries would involve material inputs with lower value added. Since much of material offshoring in Canada is with the United States and other developed countries, the effect of material offshoring on the shift to higher value-added activities should be smaller than the effect of service offshoring.

To sum up, the offshorings of services and materials are both expected to raise productivity performance through the disintegration of inputs, product specialization and the exploitation of scale economies. Material offshoring is expected to have a larger effect on productivity growth than is service offshoring, as material offshoring is mainly about the disintegration of inputs while service offshoring is as much about the unbundling of corporate functions as the vertical disintegration of inputs.

Service offshoring is expected to result in a shift to high value-added activities as service offshoring very much involves the unbundling of corporate functions and offshoring of low value-added activities. The effect of material offshoring on the shift to higher value-added activities is less obvious.

The effects of offshoring on employment and wages are much less clear. To the extent that offshoring removes part of the production process that has specific characteristics, we might expect to see a relationship between offshoring and employment or wages in a specific subset of workers. Offshoring is often presumed to affect lower skilled workers. But we have already seen that the majority of Canadian offshoring is with the United States and the United States is not seen to have lower skill levels than Canada. When it comes to predicting the impact of offshoring on the demand for all workers, it is difficult to formulate expected hypotheses. Trade leads to lower demand for labour in industries or sectors affected by increased imports but to greater demand in areas whose exports are increasing.

4.1 The relationship between offshoring and productivity growth

To examine the effect of offshoring on productivity growth, we estimate the following equation derived from the value-added production function:

where Δ denotes first differences between periods, MFP_it denotes multifactor productivity based on value added and measures the overall efficiency of capital and labour inputs in producing value added, os_it is offshoring in industry i in period t, ICT_it is information and communications technology investment intensity, and D_i and D_t are a set of dummies for industry and period fixed effects.

The offshoring variable os is defined as the ratio of imports in the cost of total material and services inputs. To examine if there are differential effects of material and service offshoring, we introduce service offshoring and material offshoring separately in the regressions.

To estimate the regression equation, we divided the 1961-to-2002 period into eight periods of about equal length.⁹ In all specifications, we include period fixed effects and industry fixed effects. The period fixed effects control for any unobserved time-varying effects common across all industries. Industry fixed effects control for any unobserved industry characteristics that may affect offshoring and productivity growth at the same time. For example, there are high productivity-growth industries that offshore more inputs to other countries.

The main finding is that material offshoring is positively related to multifactor productivity (MFP) growth and the effect is statistically significant at the 5% level (top panel of Table 11). In contrast, service offshoring is not related to MFP growth.

Most material offshoring is in manufacturing industries. When we re-estimate the MFP equation on a sample of manufacturing industries, we find that the effect of material offshoring is larger in the manufacturing industries. The coefficient on material offshoring is 0.50 with a t-statistic of 2.31. This compares with the coefficient estimate of 0.43 on the material offshoring variable when the all-industry sample is used.

Table 11
Regression results on the effects of offshoring on multifactor productivity growth

For the manufacturing sector, material offshoring increased by 15.8 percentage points from 1961 to 2003. Our estimates suggest that the growth in material offshoring contributed 7.9% to total MFP growth over the period or 0.19 percentage points to annual MFP growth. Over this period, labour productivity increased at a rate of 2.96% per year in the manufacturing sector. The contribution of material offshoring accounted for about 6% of the labour productivity growth in the manufacturing sector.¹⁰

To examine if the effect of material offshoring changed over time, we introduced an interaction term between the material offshoring variable and a period dummy for post-1981. The coefficient on the interaction term is not statistically significant, thereby indicating that the effect of material offshoring on productivity growth changed little over time.

While ICTs are not found to be a major determinant of the growth in offshoring, ICT may enhance the productivity benefits of offshoring. To test the hypothesis, we introduced an interaction term of ICT intensity with material offshoring. The coefficient is positive and statistically significant. This is consistent with the view that ICT facilitates offshoring and increases the gains from material offshoring.

Previous empirical studies in Canada find that exports are positively related to productivity using the plant-level data (Baldwin and Gu 2003, 2004). In the bottom panel of Table 11, we introduce real export growth as an additional control variable in the productivity regression.¹¹ The coefficients on the offshoring variables are virtually unchanged. The coefficient on the export growth variable is positive and statistically significant at the 1% level, which suggests that industries with faster export growth are the industries with higher productivity growth. This is consistent with the previous plant-level evidence on the positive effect of exporting on productivity growth.

4.2 The relationship between offshoring and shifts to high value-added activities

Our discussion indicates that service offshoring is expected to result in shifts to higher value- added activities, while the effect of material offshoring on shifts to higher value-added activities is less obvious.

To examine whether this is the case, we use three alternative measures to capture shifts to high value-added activities, since there is no commonly accepted criteria as to what constitutes higher value-added activities. The first measure is changes in the ratio of value added to intermediate inputs. The second measure is the change in the knowledge-worker share of employment, and the third measure is high-tech capital investment. As we do not have a consistent measure of knowledge workers over a long period, we use the share of university-educated workers in total employment as a proxy for the high knowledge content of employment.

Under the assumptions that there are constant returns to scale to the value-added aggregate and intermediate inputs and that markets are competitive, the ratio of marginal products between the value-added aggregate and intermediate inputs equals their relative prices. When gross output is a constant elasticity of substitution function of value-added and intermediate inputs, we can derive the following equation:

where P_A, P_X are the prices of value-added and intermediate inputs.

Offshoring has an effect of raising the index for the share of value-added α . Therefore, ln (α/(1 - α)) should be a positive function of service offshoring. Taking the first difference, we obtain the following equation for examining the effects of offshoring on the ratio of value added to intermediate inputs:

The results on the effect of offshoring on the ratio of value added to intermediate inputs are contained in Table 12. Service offshoring is positively related to increases in the ratio of value added to intermediate input. Material offshoring is not related to changes in the ratio of value added to intermediate input.

Table 12
Regression results on the effects of offshoring on the ratio of value-added to intermediate inputs

The coefficient on the relative prices of value added and intermediate input is the elasticity of substitution between value added and intermediate inputs. The coefficient estimates suggest that the elasticity of substitution between value added and intermediate inputs is about 0.3. This is less than 1—the elasticity of substitution from the Cobb-Douglas production function—but accords with the estimates that Bruno (1984) reports from his review of a number of empirical studies.¹²

The effects of service offshoring on the shifts to high value-added activities may differ between goods and services-producing industries. To test this hypothesis, we introduced an interaction term between service offshoring and a dummy variable for service industries. The coefficient on the interaction term is not statistically significant. This indicates that the effects of service offshoring on shifts to higher value-added activities are similar between goods and services sectors.

The relationship between offshoring and shifts to higher value-added activities using the changes in the employment share of university-educated workers is contained in Table 13. Service offshoring is not related to increases in the share of university-educated workers. Material offshoring has increased in those industries where the share of university-educated workers has decreased.

Table 13
Regression results on the effects of offshoring on the relative employment of university-educated workers

As material offshoring is mostly in the manufacturing industries, we have re-estimated the equation for the share of university-educated workers using the sample of manufacturing industries. The results from the manufacturing sample show a similar result. Material offshoring is negatively related to the changes in the share of university-educated workers.

The negative relationship between material offshoring and the employment of more educated workers could arise for a number of reasons. First, material offshoring and the resulting specialization may reduce the demand for more skilled non-production workers. Baldwin and Gu (2007) report that non-production workers were negatively affected by trade liberalization in the early 1990s. Second, the offshoring of material inputs is mostly to the United States and probably within multinational firms. Offshored material inputs from the United States may have a higher skill content than the products that are produced in Canada. As a result, material offshoring may have reduced the skilled worker share in Canadian production.

Table 14 examines the relationship between offshoring and high-tech capital deepening as measured by the growth in information and communications technology (ICT) capital services per hour worked. Since ICT became important after the early 1980s, we estimate the regression using the data for the 1982-to-2003 period. Service offshoring is found to be related to high-tech capital deepening.

Table 14
Regression results on the effects of offshoring on information and communications technology capital deepening

To sum up, we find evidence that service offshoring is associated with a shift to high value- added activities in the Canadian industries. The growth in service offshoring is positively related to an increase in the ratio of valued added to intermediate inputs, and high-tech capital deepening.

In contrast, we find that material offshoring is not associated with shifts to higher value-added activities. The growth in material offshoring is not related to the changes in the ratio of value added to intermediate input, and high-tech capital deepening. It is negatively related to the share of university-educated workers.

The evidence on the effects of material offshoring reflects the fact that manufacturing industries are highly integrated between Canada and the United States. The main effect of material offshoring is found in productivity gains related to product specialization. But, as the manufacturing industries in Canada increasingly integrate themselves with industries in those countries with low wages, it is expected that material offshoring will lead to a shift to more high value-added activities. This could occur as the low-value added firms are displaced by the competition from cheap material inputs from the low-wage countries.

4.3 The relationship between offshoring, wages and employment

Offshoring in materials and service inputs results in increased imports from other countries. Like other imports, offshoring is expected to affect wages and employment in Canada.

To examine the effect of offshoring on employment, we estimate the following labour demand equation in first-difference form (Hamermesh 1993):

We expect that increases in output will have a positive effect on employment growth and an increase in wages will have a negative effect on employment growth.

The results from estimating the employment equation are presented in Table 15. Material and services offshoring has no effect on employment in Canadian industries. The coefficients on the output and wage variables have the expected signs and are statistically significant.

As material offshoring is mostly in the manufacturing sector, we have re-estimated the employment equation using the sample of manufacturing industries. Once again, we find that material offshoring is not related to changes in employment in the sample of manufacturing industries.

To examine the relationship between offshoring and wages, we estimate the following equation:

Table 15
Regression results on the effects of offshoring on employment

The estimating equation for changes in wages is a reduced form equation. A full examination of the effects of offshoring on worker wages requires a general-equilibrium model that distinguishes several effects: a labour-demand effect that captures the effect of reduced demand for the workers whose production activities are offshored; a terms-of-trade effect that captures the labour market implications of changes in relative prices due to changes in offshoring activities; and a productivity effect that captures the benefits of offshoring-led productivity gains on wages.¹³

The results (Table 16) indicate that material offshoring is not related to wage growth. Service offshoring is negatively related to wage growth.

Table 16
Regression results on the effects of offshoring on wages

To see if there is a difference in the effect of service offshoring growth, we have introduced an interaction term of service offshoring and a dummy variable for the service-sector industries. We find service offshoring has a negative effect on wage growth in the service sector, and has little effect on wage growth in the goods-producing sector.

Table 17 presents the results on the effect of offshoring on the wages of university-educated workers relative to non-university-educated workers. We find that the material offshoring and service offshoring are not related to changes in relative wages for this group.

Table 17
Regression results on the effects of offshoring on relative wages of university-educated workers

Our findings on the effects of offshoring on wages and employment are consistent with more aggregate recent estimates of the effects of offshoring on Canadian labour markets (Trefler 2005; Morissette and Johnson 2006). Morissette and Johnson find little employment effects of offshoring. Trefler (2005) finds that the Canadian labour markets have been adjusting to service offshoring via wages rather than employment. As discussed in Trefler (2005), this is unusual as the studies on the impact of international trade find that employment typically adjusts more than wages.¹⁴

Our findings on the impact of offshoring on skilled workers are different from Yan (2006), who examines the demand for non-production workers relative to production workers in Canadian manufacturing industries. Yan (2006) finds that material offshoring increases, not decreases, the demand for this more skilled category (non-production workers). But her study estimated a regression equation in levels, not in first differences as done here. The results in Yan (2006) suggest that industries with a high level of material offshoring tend to have a greater demand for more skilled workers. The results in this paper, using a first-differenced equation, suggest that growth in material offshoring reduces the demand for more skilled workers.

 

⁷ See Triplet and Bosworth (2004). For a dissenting opinion, see Diewert (2005).

⁸ The constant elasticity of substitution production function has been used extensively in research that examines changes in relative wages of skilled and unskilled workers and the role of skill-biased technological change (e.g., Katz and Murphy 1992).

⁹ The eight periods include seven of five years and one of seven years. The last period covers the seven years from 1996 to 2003.

¹⁰ Amiti and Wei (2005) find that material offshoring accounted for 5% of labour productivity growth in the U.S. manufacturing sector over the 1992-to-2000 period.

¹¹ Real exports are estimated from deflating nominal exports by gross output prices. The specification has been used in Bernard and Jensen (1999). The results from using nominal export growth are similar.

¹² Bruno (1984) reports most estimates are between 0.3 and 0.4.

¹³ Grossman and Rossi-Hansberg (2006) have developed a theoretical framework to examine those effects of offshoring on wages.

¹⁴ Liu and Trefler (2006) find no evidence that service offshoring reduced wages of workers in the United States.