Outsourcing and wage inequality in OECD countries

(1)

UNIVERSITY OF GRONINGEN

FACULTY OF ECONOMICS AND BUSINESS

Outsourcing and wage

inequality

in

OECD

countries

MSc Thesis International Economics and

Business

L.G. van Kessel

S2184745

l.g.van.kessel@student.rug.nl

6/14/2016

(2)

(3)

2 Abstract

The increase in wage inequality in OECD countries has attracted much attention in the literature. Several factors have been suggested as cause of this phenomenon, including globalization. One form of globalization which is not included in earlier research is the import of intermediate inputs, or outsourcing. This thesis covers this gap by looking into potential causes of wage inequality, with special focus on outsourcing. A distinction is made between outsourcing to OECD and non-OECD countries. The results indicate that outsourcing, specifically to non-OECD countries, is associated with a rise in wage inequality.

(4)

1. Introduction

Since the 1980s, inequality in terms of income and wage has increased in a large part of the member countries of The Organization for Economic Co-operation and Development (OECD) (OECD, 2011). This thesis aims to analyze the impact of outsourcing on wage inequality in OECD countries. Although the literature mostly focuses on income inequality (Dabla-Norris et al., 2015; Herzer and Nunnenkamp, 2013; Alvaredo et al., 2013) this thesis focuses on wage inequality, because effects of outsourcing are related to the composition of the labor market and therefore possibly influences wages. Since wages are an integral component of income, inequality in wages plays a significant role in determining the extent of income inequality. The wage inequality, as measured by the decile ratio of gross earnings of the top ten percent to the bottom ten percent of employees, increased in many OECD countries between 1980 and the late 2000s, with rates of between 20 to 25 percent (OECD, 2011). This increase is of major concern, because inequality has considerable implications for growth and macroeconomic stability (Dabla-Norris et al., 2015).

(6)

2

different perspective, the global value chain (GVC) analysis is a good way to look at the increasingly globalized world. A more intensive participation in a GVC is argued to result in more wage inequality, as production and service tasks can be located at cheaper places in the world (Timmer et al., 2013; Lopez Gonzalez and Kowalski, 2015). An important aspect of globalization, the effect of which on wage inequality is less understood, is outsourcing. In this case, the form of trade is in intermediate goods. This thesis will focus on the effect of outsourcing on wage inequality.

1.1 Outsourcing and wage inequality

(7)

3

effect is that the demand for skilled workers increases at the cost of unskilled workers, which increases the wage dispersion. This demand for skilled workers increases since outsourcing can relocate labor intensive parts abroad and specialize in skill intensive production (Wood, 1998). However, studies that examine the determinants of wage inequality in a panel of multiple countries do not include outsourcing in their analyses (OECD, 2011; Dredger et al., 2015).

The purpose of this thesis is to research the effect of the use of imported goods in production, here interchangeably used with outsourcing, on wage inequality in a panel of OECD economies.

Using a sample of 19 OECD countries, the effect of outsourcing on wage inequality is investigated. All other possible determinants as given in literature are taken into account. Moreover, a distinction will be made between outsourcing to OECD and non-OECD economies, as possibly the outsourcing is done with different incentives, and their potential impact could be different.

The remainder of this thesis is organized as following: In section 2, the increasing trend of wage inequality and the existing literature on determinants of wage inequality will be reviewed. Section 3 follows with the data and methodology. In section 4 the results follow and section 5 will present the conclusions.

2 Literature review

2.1 Increasing wage inequality in OECD countries

From data and literature, it becomes clear that income and wage inequality is increasing in most of the OECD countries. In earlier research, focus has been more on income inequality than on dispersions in wage. However, this thesis will focus specifically on wage inequality, as the effects of outsourcing are most likely related to the labor market. This section continues with definitions of inequality, income and wage and how these three concepts are related. After that, a brief review of the literature on inequality is given.

(8)

4 earnings, self-employment and capital income and public cash transfers net of income taxes and social security contributions paid by households. Wage is “the total remuneration … in

return for work done during the accounting period" and this can also be called earnings. When comparing these two definitions, wage is thus a reward for work delivered and is most often the dominant part of income, income itself can also have other sources. This can lead to the conclusion that wage is a part of income and therefore wage inequality is one part of the income inequality, which is related to the labor market. Changes in distribution of wages and salaries account for 75 percent of household income, which makes it a large driver (OECD, 2011). For a majority of people wage is the main source of income, therefore the distribution of wage has important implications for income inequality (Figini and Görg, 2006). From this one can conclude that income and wage inequality are highly interrelated and possibly follow the same pattern.

According to Dabla-Norris et al. (2015) inequality matters because it can be a signal of lack of income mobility and opportunity, and has significant implications for growth and macroeconomic stability. Different studies looked into how income and wage inequality has evolved over time.

Dabla-Norris et al. (2015) researched income inequality and its causes. They found that the Gini coefficient1- a dominant measure of income inequality- increases substantially since 1990 in the developed world. Furthermore, an OECD study in 2011 found that in most of the OECD economies, income inequality has risen since the mid-1980s (OECD, 2011). With an average Gini coefficient of 0.29 in the mid-1980s, it had increased with almost 10 percent to 0.316 in the late 2000s.

Similar trend has been observed by studies on wage inequality. The 90th to 10th percentile ratio of wage distribution – a commonly used measure of wage inequality -2 for full-time

1_{The Gini index measures the extent to which the distribution of income among individuals or} households within an economy deviates from a perfectly equal distribution. A Gini index of 0 represents perfect equality, while an index of 100 implies perfect inequality

(http://data.worldbank.org/indicator/SI.POV.GINI).

2

(9)

5

male earnings in eleven of twelve studied OECD countries increased between 1980 and 2011 (Autor, 2014). Also the OECD (2011) found that wage dispersion significantly increased since 1980 in most of the OECD countries tested, but the increase was mostly during the late 1990s and 2000.

Dredger et al. (2015) studied changes in wage inequality on behalf of the European Parliament and found that although there are differences in changes in wage inequality in European Union countries, for example in timing and intensity, in two-third of the EU inequality has increased. This increase is noted since the mid-1980s until 2011.

Although there are different statements on how the wage inequality at the lower part of the wage distribution has evolved, sources are consistent about the increase in wage inequality at the top end of the wage distribution (OECD, 2011; Glyn, 2001; Lopez Gonzalez and Kowalski, 2015). In terms of skilled and unskilled labor it can therefore be assumed that the demand for skilled workers has increased faster than their supply, leading to a rise of earnings of skilled people relative to the earnings of unskilled people (Snower, 1999), which increases wage dispersion.

(10)

6 2.2 Determinants of wage inequality

As stated earlier, in 2011, the OECD published a report on the rising inequality in OECD countries. The possible causes for wage inequality they tested were globalization, technological change and labor market institutions, next to some other control variables. They found that globalization (trade and FDI) on its own did not have any effect, but imports in combination with weaker employment protection can increase wage dispersion. Furthermore, technological change and weaker labor market institutions have a positive effect on wage inequality. In this study, outsourcing in the form of trade in intermediates is not taken into account. It could, however, have different effects on wage inequality than regular trade, as will be explained later.

There are several other studies that examine the possible causes of wage inequality. Explanations are mostly related to side determinants of skilled work. The demand-side effects can arise through an increase in the demand for high-skilled workers and a decrease in the demand for low-skilled workers; and the consequent change in the relative wages of skilled and unskilled workers. Technological change and globalization are examples of such demand effects (Lorenzen, 2011). However, the determinants are not necessarily related to the demand side. The remainder of this section will go deeper into the determinants of wage inequality, dividing the causes in two groups: Factors that reduce inequality and explanations that increase wage inequality. The first group includes female labor participation and education and the latter includes weaker labor market institutions and the demand-side factors technological change and globalization. Globalization covers the global value chain (GVC), trade, FDI and outsourcing.

2.2.1 Female labor participation

(11)

7 2.2.2 Education

Furthermore, education is a factor that could impact wage inequality. A higher educational attainment rises the supply for skilled labor. This makes skilled workers more abundant, which reduces the returns to skill (OECD, 2011; Topel, 1997). OECD (2011) find that indeed a higher proportion of skilled workers tend to decrease wage differentials. Despite this, it largely depends on how demand for skills develops. If the demand growth for skilled workers is greater than the growth in supply, relative wages are increasing (Topel, 1997).

2.2.3 Labor market institutions

(12)

8 2.2.4 Technological change

One factor which is often mentioned as most important factor for wage inequality equality is technological change or progress. This phenomenon is also called the skill-biased technological change (SBTC), which Card and DiNardo (2002) explain as: ‘that a burst of new technology caused a rise in the demand for highly skilled workers that in turn led to a rise in earnings inequality’. If technological progress increases, wage inequality could rise because skilled labor demand increases relative to the demand for the unskilled workers. The reason for this is that technological change shifts the production technologies in advantage of skilled workers (OECD, 2011). In first instance, low skilled workers get replaced and high skilled labor is needed to maintain the new technology. In addition, skills are needed to run the automated capital, so firms with a high level of technology will employ more skilled workers (Acemoglu, 2003; Lorenzen, 2011). This makes clear that the demand for skilled shift as a result of technological change, however this is often not in tandem with changes in the proportion of skilled and unskilled labor supply. This widens the gap between supply and demand for skills and lowers the relative wage of unskilled workers, widening the income gap. According to the OECD (2011), technological advances have a higher effect on wage inequality within countries than trade and financial factors.

2.2.5 Globalization

Another important factor that has gained much attention in the literature is globalization. The literature on the impact of globalization on inequality considers different aspects of globalization, such as trade and FDI, but also the global value chain (GVC) (Lopez Gonzalez and Kowalski, 2015; Figini and Görg, 2011; Atkinson, 2003).

Global Value Chain:

(13)

9

be exported after (Tsafack and Parsasnis, 2010). The rationale for GVC trade is an increase in productivity which benefits the final good or service, because it is exported to a greater extent with a higher level of specialization for all economies (Miroudot et al., 2009). In theory, the effect of GVC participation on wage inequality can be twofold. On the one hand, downward pressure can be exercised on low-skilled wage, as some low-skilled workers become redundant. On the other hand, an increase in low-skilled wages can arise, if productivity increases as a result of cheaper GVC trade (Lopez Gonzalez and Kowalski, 2015). Timmer et al. (2014) found that international production fragmentation is related to a decrease of low skilled jobs, which could be a source of increasing wage inequality. This effect happens since firms in developed countries relocate unskilled labor intensive activities to low wage countries and keep activities that require skilled workers at home (Timmer et al., 2014).

FDI:

(14)

10

(IMF, 2007), while technological change is expected to keep rising the demand for skilled labor.

Trade:

The impact of trade on inequality can be explained by existing trade theories, specifically the Heckscher-Ohlin (HO) and the Stolper-Samuelson models. The traditional HO model consist of two factors, namely skilled and unskilled labor, and two countries, North (developed) and South (developing). In developed countries, the relative wage for skilled labor will be lower than in developing countries, so North has a comparative advantage and it will specialize in the production of skill-intensive goods. This leads to a wage gap between skilled and unskilled labor, as the relative price of skill-intensive goods and the relative demand for skilled labor increases in North, with increasing relative wage as a consequence. For unskilled labor this is an opposite effect, so the relative wage will decrease (Wood, 1998; O’Rourke, 2003). According to this model, in developing countries wage inequality should decrease, in developed countries it should increase. This model is thus not entirely applicable anymore, because in both regions the wage inequality is increasing. However, in the case of developing countries a possible explanation within the boundaries of globalization, is the role of outsourcing. In emerging countries the outsourced work is relatively skill intensive for them, while it is less skilled labor intensive in developed countries (Cheng and Zhang, 2007). So, the HO model still accounts for the rising inequality in the North, because they are specializing in the high skilled work. The impact of outsourcing to emerging markets on inequality within the developed countries will be discussed later in this thesis.

(15)

11

However, the empirical results of the effects of trade and openness to trade are conflicting. As stated above, the OECD (2011) found no significant effect. On the other hand, Molnár et al. (2007) find that in almost all OECD countries greater international openness coincided with more wage dispersions between 1980 and 1999. As most of the OECD countries are developed and therefore relatively skill abundant, it is worthwhile to specialize in the production of goods that make relatively intensive use of skilled labor and to import goods that are intensive in unskilled labor (Snower, 1999). Therefore, a shift in demand away from unskilled labor occurred due to, among others, globalization (Atkinson, 2003).

2.2.6 Outsourcing

(16)

12

In addition, Cheng and Zhang (2007) developed two models of which one assumes symmetric production technologies of an intermediate good in both developed and developing countries and the other model assumes asymmetric production technologies. Symmetric production technologies in the sense that in both the developed and developing country the intermediate product is produced with unskilled labor and asymmetric means that the intermediate product uses unskilled labor in the developed country but skilled in the developing one. Both models predict trade in intermediate goods increases wage inequality in the developed country.

Next to these arguments there is also empirical evidence of outsourcing on wage inequality. Feenstra and Hanson (1996) argue that outsourcing has contributed substantially to the increase in the relative demand for nonproduction labor in the U.S., which in turn leads to an increase in the nonproduction wage share. According to Feenstra and Hanson (2001), outsourcing has a reducing effect on the relative demand for unskilled labor, because activities that use a large amount of unskilled labor can be outsourced as the domestic relative wage is higher than abroad. Outsourcing is also found to have an important role in increasing wage inequality in the U.K., along with skill-biased technological change (Hijzen, 2007). Tsafack and Parsasnis (2010) suggest that outsourcing account to a great extent for the decline in wage bill share for unskilled workers in France, and Munch and Skaksen (2009) arrive at similar conclusion in the case of Denmark. This decline in wage share of unskilled workers does drive the wage inequality up. This evidence shows that outsourcing can account for the increase of wage in the upper skill classes as well as a decline of wage for low-skilled workers.

(17)

13

wage at the top-end of the distribution, thus widening the gap between the two. We postulate the following hypothesis:

H1. Outsourcing of economic activities from OECD economies has a positive effect on wage inequality in OECD economies.

2.3 Outsourcing to other OECD countries

Although the theory and evidence explained above is mostly focused on trade in intermediates in general or between developing and developed countries, most of the OECD country imports of intermediate goods take place between OECD countries (Miroudot, Lanz and Ragoussis, 2009). Indeed the trade between OECD and non-OECD is also increasing, for instance between 1992 to 2004, the share of intermediate imports from non-OECD countries rose from 15% to 25% of total imports of intermediate goods (Molnár et al., 2007). Yet, the large part of intermediate trade is still among OECD countries. In figure 1it can be observed that for most of the countries in the sample indeed the share of outsourcing to other OECD of total outsourcing is increasing. It is likely that the impact of outsourcing on wage inequality could differ between outsourcing to other OECD and outsourcing to non-OECD, as the former may also include the outsourcing of high-skilled jobs, as will be explained in this section. For that reason, it may be interesting to look at outsourcing from OECD countries by distinguishing between outsourcing to OECD and non-OECD countries, to see if the impact on wage inequality differs.

(18)

14

Figure 1. Outsourcing to OECD and non-OECD countries

Moreover, next to the cost and productivity reasons for outsourcing there is also another form, namely strategic outsourcing. This is the purchasing of intermediate inputs form a rival in the final good market that is a more efficient supplier. This leads to an increase in price for both the final and the intermediate good. A tariff reduction in the intermediate good can

(19)

15

enable or enhance this effect (Chen et al., 2004). Link this to the Stolper-Samuelson model, the high skilled wage will therefore increase as the traded intermediate is likely to be skill-intensive.

One earlier research make the distinction between outsourcing to OECD and non-OECD countries. Yamashita (2010) found that for the U.S. trade in intermediates increases the wage inequality when importing from developing countries, but also found that it has no such effect when looking at the import from developed countries. However, this research is looking at one country, which need to be extended further. Summarized, outsourcing to other OECD countries covers high skilled as well as low skilled work, for which the effects possible outweigh each other.

All this together leads to the following hypothesis:

H2. Outsourcing of economic activities from one OECD country to another OECD country will have a weaker impact on wage inequality in OECD countries compared to outsourcing to a non-OECD country.

3 Methodology

The objective of this thesis is to determine the possible causes of wage inequality in OECD economies, with special focus on the role of outsourcing – distinguished between outsourcing to OECD and non-OECD. The relationship between outsourcing and wage inequality will be tested according to a panel data analysis. The reason for this approach is that the data is observed over time and consists of a group of cross-sectional units, in this case countries. The analysis conducted for 19 countries, for the period 1995-2011. Because not all data on wage inequality is available for every year, the panel is unbalanced. A Hausman test is performed to make sure the right model is assumed. The result indicate that the random effects model is the preferred choice. Results of the test can be found in appendix B.

(20)

16

of outsourcing. These control variables include FDI, technological change, labor market institutions, female work participation, education, sectoral shares, instability of the economy and GDP growth rate. As discussed in the literature, FDI, technological change and labor market are important because evidence suggest that these all are main determinants of wage inequality. Similarly, female participation, sectoral share of service, education, the output gap and change in GDP growth rate can all also impact wages and wage inequality, so they are also included as control variables.

The model that will be tested is the following:

𝑾_𝑰𝑵𝑬𝑸_𝒊𝒕 = 𝜶 + 𝜷_𝟏 𝑶𝑼𝑻𝑺_𝒊𝒕+ 𝜷_𝟐 𝑭𝑫𝑰_𝒊𝒕+ 𝜷_𝟑 𝑻𝑬𝑪𝑯_𝑪𝑯𝑨𝑵𝑮𝑬_𝒊𝒕 + 𝜷𝟒 𝑴𝑰𝑵_𝑾𝒊𝒕+ 𝜷𝟓 𝑾𝑶𝑴𝑨𝑵_𝑷𝑨𝑹𝑻𝒊𝒕+ 𝜷𝟔 𝑬𝑫𝑼𝑪𝒊𝒕

+ 𝜷_𝟕 𝑺𝑯_𝑺𝑬𝑹𝑽𝑰𝑪𝑬_𝒊𝒕+ 𝜷_𝟖𝑮𝑫𝑷_𝑮𝑹𝑶𝑾𝑻𝑯_𝒊𝒕+ 𝜷_𝟗𝑶𝑼𝑻_𝑮𝑨𝑷_𝒊𝒕 (1)

The above model allows us to understand the impact of outsourcing from OECD economies to all countries, including other OECD economies, and thus allows us to test the first hypothesis. To distinguish between the effect of outsourcing to only OECD countries and outsourcing to non-OECD countries, the baseline model will change into:

𝑾_𝑰𝑵𝑬𝑸_𝒊𝒕 = 𝜶 + 𝜷_𝟏 𝑶𝑼𝑻𝑺_𝑶𝑬𝑪𝑫_𝒊𝒕+ 𝜷_𝟐 𝑶𝑼𝑻𝑺_𝑵𝑶𝑵𝑶𝑬𝑪𝑫 + 𝜷_𝟑 𝑭𝑫𝑰_𝒊𝒕 + 𝜷𝟒 𝑻𝑬𝑪𝑯_𝑪𝑯𝑨𝑵𝑮𝑬𝒊𝒕+ 𝜷𝟓 𝑴𝑰𝑵_𝑾𝒊𝒕+ 𝜷𝟔 𝑾𝑶𝑴𝑨𝑵_𝑷𝑨𝑹𝑻𝒊𝒕 + 𝜷_𝟕 𝑬𝑫𝑼𝑪_𝒊𝒕+ 𝜷_𝟖 𝑺𝑯_𝑺𝑬𝑹𝑽𝑰𝑪𝑬_𝒊𝒕+ 𝜷_𝟗𝑮𝑫𝑷_𝑮𝑹𝑶𝑾𝑻𝑯_𝒊𝒕

+ 𝜷_𝟏𝟎𝑶𝑼𝑻_𝑮𝑨𝑷_𝒊𝒕

(2)

The sample consists of 19 OECD countries and counts 99 observations. The list of countries included is provided in Appendix A. Data on the dependent variable outsourcing is extracted from the ‘Trade in Value Added (TiVa) database’.3 It is developed by the OECD and the World Trade Organization (WTO) as a joint initiative. They used Inter-Country Input Output tables and the database included 61 economies for the years 1995, 2000, 2005 and 2008-2011. For this data availability reason, these will be the years that are taken into account in this research. Other sources that are used to collect data are: Different databases from the OECD, International Labor Organization (ILO), Barro-Lee and the Conference Board. Data on

3

(21)

17

outsourcing could also be extracted from the World Input-Output Database (WIOD). However, the OECD provided similar statistics and as for most of the variables the OECD database is used, this source is considered.

The following section outlines the variables and their construction. Furthermore, in table 1 we provide a summary of the variables and their expected relationship with inequality.

3.1 Dependent variable:

The dependent variable in this research is wage inequality (W_INEQ). Wage inequality can be measured by the ratio of the 90th to the 10th percentile of the wage distribution (Machin, 1996). Therefore, the inter-decile ratio (D90/D10) of earnings, provided by the OECD will be used.

3.2 Independent variables:

Outsourcing (OUTS): Outsourcing will be measured as the gross imports of intermediate

products as a share of total purchase of non-energy materials, following Feenstra and Hanson (1996). This measure captures what share of purchases used in production is imported from abroad. Two important features of this measurement must be noted. First, it uses purchases of inputs of all domestic firms and is not restricted to purchases of multinationals from foreign subsidiaries (Feenstra and Hanson, 1996). In addition, it includes imported goods and services. The measure of Feenstra and Hanson is focusing only on imports. It may have its limitations, for example that it do not take into account that domestically produced products can also be exported (Yamashita, 2010). However, in this thesis it is the right measure, since it specifically look into how intermediate inputs are used in the country’s production. Distinctions are made between gross imports of intermediate products from all countries and from OECD countries only, to see if differences exist between outsourcing to non-OECD countries and OECD countries.

To observe the effect of outsourcing on wage inequality the other factors which also determine wage inequality will be taken into account.

FDI (FDI): FDI is measured as inward FDI stock as percentage of GDP, following Figini and

(22)

18

Data is obtained from the OECD factbook. Since OECD countries are already developed in technology, it is expected that FDI has a negative effect on wage inequality.

Technological change (TECH_CHANGE): The proxy used for technological change is total

factor productivity (TFP) growth, which is a largely used indicator in literature (Nour, 2013). TFP is measured as a residual after allowing for the contribution of labor and capital inputs to output growth. The data is obtained from the Conference Board Total Economy database. As discussed earlier, technological change is expected to have a positive effect on wage inequality.

Labor market institutions (MIN_WAGE): We measure labor market institutions using

minimum wages. Minimum wages is a good representation of the labor market institutions as it is directly related to the wage distribution since it sets a floor for low wage workers. Furthermore, it is related to other measures of institutions, for example wage bargaining is represented in trade unions, and this wage bargaining is used to set minimum wages. Minimum wages are set relative to median wages and is expected to have a negative effect on wage inequality. The source for this variable is the OECD labor database.

Women participation (WOMEN_PART): Women participation will be measured as the share

of female employment in total employment. Data for this measure is extracted from the “population and labor force” dataset of the OECD. Data is available for total female employment and total employment. Becausedata is not available for all countries and times, it is complemented by statistics from the ILO database. Here the female share of employment is already given in the direct form. We expect female participation to have a decreasing effect on wage inequality, since a higher female workforce increases wages for women.

Education (EDUC): Education is included to account for the impact of skill levels of workers

(23)

19 Economic stability (OUT_GAP): Following the OECD (2011) and Dredger et al. (2015), we

also include cyclical fluctuations in aggregate demand, to capture stability of the economy (OECD, 2011). We proxy this variable by the output gap, which is the difference between actual and potential GDP as a percent of potential GDP(OECD Glossary, 2016). When a large output gap exists, not all production capacity is used, which causes the wage demand to be lower. This data is extracted from the Economic Outlook dataset provided by the OECD.

Sectoral share (SH_SERVICE): Since the composition of the economy can have major impact

on the distribution of wages, we also try to account for it by including the share of services sector in employment. The service sector employ mainly high to medium skilled workers (Gonzales et al., 2012) and thus is the sector where high earners are employed. So, the share of employment in the service sector is included as control variable, to capture the impact a high service sector can have on the wage distribution. Data is obtained from the OECD ‘Employment by activity and status’ dataset.

GDP growth rate (GDP_GROWTH): Change in GDP growth rate is included to account for the

(24)

20

Table 1. Measurement and expected relationship of variables

Variable Measure Expected relationship

Outsourcing Imports of intermediate

products as a share of total purchases of non-energy materials

Positive

Inward FDI Inward FDI stock as

percentage of GDP

Negative

Technological change TFP growth Positive

Labor market institutions: Minimum wage Minimum wage as share of median wages

Negative Share of women participation Female employment as

share of total employment

Negative

Education Percentage of population

completed tertiary education

Negative

Stability of the economy (output gap) Difference between actual and potential GDP as percentage of

potential GDP

Positive

Share of workers in service sector Service employment as share of total

employment

Positive

GDP growth rate Change in GDP growth

rate

Positive

4 Results

This section provides the regression results using equations 1 and 2, along with summary statistics for each variable.

4.1 Summary statistics

(25)

21

countries and over time, with a value of 0.01% as minimum and 439% as maximum, which thus means that the inward FDI stock is more than 4 times GDP. On average, TFP seems to be declining in OECD economies, registering an average of -0.6 percent, with the lowest being at -9% and the highest being at 6%. Minimum wage differs widely between 31% and 68% of median wage.

Table 2. Summary statistics

Variable Mean Std. Dev. Min Max

W_INEQ 3.57 .62 2.3 5.19 OUTS_WORLD .26 .12 .06 .60 OUT_OECD .19 .11 .03 .54 OUT_NONOECD .07 .04 .02 .18 FDI .61 .91 .00007 4.39 TECH_CHANGE -.006 .03 -.09 .06 MIN_W .47 .08 .31 .68 WOMEN_PART .44 .03 .37 .48 EDUC .17 .07 .04 .32 SH_SERVICE .69 .09 .45 .86 OUT_GAP -.003 .03 -.09 .09 GDP_GROWTH -1.35 8.14 -73.54 11.83 4.2 Statistical requirements

(26)

22

our regression results a variance inflation factor (VIF) test is performed to test the collinearity further.

Table 3. Correlation Matrix

According to Heiberger and Holland (2013), multicollinearity is a problem when the results of VIF exceed 5. In table 4 we provide the results of VIF for our data. On average, the VIF is 2.02, with 1.06 having the lowest and 4.23 having the highest. However, even the highest value is less than 5 and therefore, from the VIF test it can be concluded that multi-collinearity is not a severe problem for our regression.

Table 4. VIF test

Variable VIF SH_SERVICE 4.23 FDI 3.08 EDUC 2.42 OUT_OECD 2.26 MIN_WAGE 1.65 WOMEN_PART 1.51 OUT_NONOECD 1.37 OUT_GAP 1.32 TECH_CHANGE 1.27 GDP_GROWTH 1.06 Mean VIF 2.02 4.3 Results

(27)

23

find out whether the impact of outsourcing differs between outsourcing to OECD and to non-OECD, outsourcing is divided into outsourcing to OECD and non-OECD economies. From the results it becomes clear that the main driver of this effect is outsourcing to non-OECD countries. Outsourcing to non-OECD countries gain a larger and significant coefficient compared to outsourcing to OECD countries. These effects remain positive and significant even after inclusion of control variables. By this robustness of the coefficients is confirmed, which means that the effect does not result from exclusion of other variables. The quantitative magnitude of the non-OECD coefficient is more than double the size of OECD coefficient. Moreover, the impact of outsourcing to other OECD countries, though prevalent when we include other possible main determinants of wage inequality, disappears when including all control variables.

(28)

24

Table 5 Baseline model - whole sample

(1) (2) (3) (4) (5) (6) OUTS_WORLD 1.61*** 2.08*** 1.31** (0.61) (0.56) (0.59) OUT_OECD 1.20 1.40* 0.64 (0.87) (0.82) (0.82) OUT_NONOECD 2.43* 3.37*** 2.56** (1.37) (1.25) (1.26) FDI -0.02 -0.02 -0.06 -0.06 (0.06) (0.06) (0.07) (0.07) TECH_CHANGE -1.91*** -1.89*** -1.50** -1.45** (0.71) (0.70) (0.70) (0.70) MIN_W -2.13*** -2.21*** -2.22*** -2.33*** (0.49) (0.49) (0.51) (0.51) WOMEN_PART -3.49 -4.39* -2.42 -2.49 EDUC 2.523** 2.54** (1.19) (1.18) SH_SERVICE 1.23 1.32 (1.24) (1.24) GDP_GROWTH 0.0010 0.0011 (0.0021) (0.0020) OUT_GAP 1.03 0.92 (0.64) (0.64) _cons 3.21*** 3.23*** 4.08*** 4.16*** 5.03*** 4.61*** (0.22) (0.22) (0.30) (0.31) -1.07 -1.02 N 99 99 99 99 99 99 R-squared 0.0712 0.0721 0.3040 0.3131 0.4425 0.4582

Standard errors in parentheses

*p<0.10, ** p<0.05, *** p<0.01 4.4 Robustness

Since, our sample consists of all OECD economies, including two developing countries Chile and Mexico, in order to check the possible impact of these countries on our overall results – which otherwise is a sample of developed countries, an additional panel analysis is executed. In this model, outsourcing to Mexico and Chile is now included in non-OECD instead of OECD countries. In addition, since Mexico in this model is seen as a non-OECD country, Mexico is excluded from the sample. The results are displayed in table 6.

(29)

25

the results, even though these countries are less developed than the other members of the OECD.

Table 6 Baseline model - excluding Chile and Mexico

(7) (8) (9) (10) (11) (12) OUTS_WORLD 1.67*** 2.12*** 1.33** (0.62) (0.57) (0.59) OUT_OECD 1.16 1.31 0.55 (0.88) (0.82) (0.82) OUT_NONOECD 2.65* 3.65*** 2.81** -1.37 -1.23 -1.24 FDI -0.02 -0.02 -0.07 -0.06 (0.06) (0.06) (0.07) (0.07) TECH_CHANGE -1.82** -1.80** -1.30* -1.25* (0.74) (0.73) (0.73) (0.73) MIN_W -2.17*** -2.27*** -2.25*** -2.39*** (0.50) (0.50) (0.52) (0.52) WOMEN_PART -3.75 -4.72* -2.49 -2.54 EDUC 2.60** 2.59** (0.01) (0.01) SH_SERVICE 1.30 1.39 -1.26 -1.25 GDP_GROWTH 0.0009 0.0011 (0.0021) (0.0021) OUT_GAP 0.95 0.82 (0.66) (0.66) _cons 3.19*** 3.21*** 4.10*** 4.20*** 4.69*** 5.16*** (0.22) (0.23) (0.31) (0.32) -1.06 -1.10 N 94 94 94 94 94 94 R-sq 0.0780 0.0798 0.3117 0.3255 0.4601 0.4815

Standard errors in parentheses *p<0.10, ** p<0.05, *** p<0.01

(30)

26

Table 7 Robustness check

(13) (14) (15) (16) (17) (18) OUT_WORLD 1.34** 1.12* 1.29** (0.59) (0.61) (0.58) OUT_OECD 0.64 0.33 0.74 (0.84) (0.85) (0.82) OUT_NONOECD 2.62** 2.53* 2.27* -1.26 -1.33 -1.24 FDI_FLOW -0.14 -0.13 (0.20) (0.20) FDI -0.08 -0.07 -0.07 -0.07 (0.07) (0.07) (0.07) (0.07) TECH_CHANGE -1.37* -1.34* -1.08 -1.07 (0.73) (0.72) (0.67) (0.67) BERD -7.74 -10.13 (12.78) (12.83) MIN_W -2.24*** -2.35*** -2.09*** -2.22*** -2.26*** -2.35*** (0.51) (0.52) (0.54) (0.54) (0.49) (0.50) WOMEN_PART -4.37* -5.11** -3.83 -4.97* -3.00 -3.83 -2.41 -2.46 -2.49 -2.56 -2.41 -2.50 EDUC 2.51** 2.53** 3.04** 3.11** 1.97 2.03* (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) SH_SERVICE 1.25 1.32 1.40 1.54 1.65 1.71 -1.26 -1.25 -1.30 -1.29 -1.23 -1.23 GDP_GROWTH 0.0009 0.0010 0.0009 0.0009 0.0006 0.0007 (0.0021) (0.0020) (0.0022) (0.0022) (0.0020) (0.0020) OUT_GAP 0.96 0.86 0.88 0.76 (0.64) (0.64) (0.71) (0.71) UNEMP -1.74** -1.60** (0.71) (0.72) _cons 4.96*** 5.33*** 4.65*** 5.17*** 4.38*** 4.75*** (0.99) -1.03 -1.07 -1.11 -1.01 -1.06 N 98 98 96 96 99 99 R-sq 0.4484 0.4618 0.4182 0.4436 0.4679 0.4778

Standard errors in parentheses *p<0.10, ** p<0.05, *** p<0.01

(31)

27

However, concerning the instability of the economy, the sign and significance of the measure itself changes. This makes sense, as this measure of instability of the economy is directly related to the labor market and therefore to wage inequality, as the output gap is not. It can be seen that a higher unemployment rate lowers wage inequality. Unemployed population is not taken into account in the measure of wage inequality, since they do not earn wage. This negative effect suggest that a large share of unemployed is high skilled, since a larger share of high wage earners increase wage inequality, and high skilled workers are most likely to be high wage earners.

4.5 Discussion of the results

The results outlined in this section should give answer to the hypotheses set in the literature review of this thesis. The first hypothesis is ‘Outsourcing of economic activities from OECD economies has a positive effect on wage inequality in OECD economies.’ The results indicate that indeed the relationship between outsourcing and wage inequality is positive. This effect remains positive and significant regardless of the control variables included in the model. Therefore, it can be concluded that outsourcing is an important driver of the increasing wage inequality in OECD economies.

(32)

28

wage dispersions. On the other hand, this between-OECD country trade in intermediate does not necessarily need to be in inputs that requires high skilled labor. If low skilled intermediates are outsourced, then it can have a positive effect on wage inequality. If outsourcing to other OECD countries is done with both incentives, effects on demand for high and non-skilled labor can possibly outweigh each other.

Together, the results seem to suggest that effect of outsourcing on wage inequality in OECD economies is mostly due to outsourcing to non-OECD economies. This is in line with theory, since low skilled jobs are being outsourced to non-OECD countries. The effect of outsourcing to non-OECD countries is positive and significant in all models.

Next to outsourcing, education is positively related to wage inequality. Higher educational attainment on its own is expected to reduce wage differentials, since it offers a higher supply of skilled labor and this higher supply should lower wages for that skill group. However, since the results suggest that a higher educational attainment has a positive effect on wage inequality, the demand for skilled labor is most likely to increase at a faster pace than the supply. This causes the wage of educated workers to increase, despite the higher supply. Since technological change, minimum wage and to some extent a higher share of female employment all reduce wage inequality, education and outsourcing together are the factors behind the increase in wage inequality in OECD countries.

(33)

29

5 Conclusion

Since the 1980s, wage inequality is increasing in most OECD countries. Several studies find possible determinants for this trend, which includes globalization, technological change and labor market institutions (e.g. Dredger et al., 2015; OECD, 2011; Dabla-Norris et al., 2015). One important aspect of globalization that thus far did not gain much attention in panel data literature is outsourcing. This thesis tries to fill this gap in the literature and focusses on possible determinants of wage inequality in OECD countries, with special focus on outsourcing in the form of trade in intermediate inputs. The focus is on OECD economies as it became clear that in this group of countries wage inequality is increasing and no earlier panel analysis is done with respect to the relationship between outsourcing and wage inequality. Furthermore, since incentives to outsource to other OECD countries can differ from reasons to outsource to non-OECD countries, a distinction is made between the outsourcing to the two directions, to see if different effects exist.

Based on a literature review, it was hypothesized that outsourcing, both to the world as well as to only OECD economies, would lead to an increase in wage dispersion. Other factors taken into account are FDI, technological change, minimum wage, education, output gap, women participation rate in employment, GDP growth rate and the share of service in total employment. The findings confirm that outsourcing in general leads to more wage inequality. This finding is robust when including (different measures of) control variables. No significant evidence can be found for the positive effect of outsourcing to only OECD countries. Outsourcing to non-OECD countries do have a positive effect, from which can be concluded that specifically outsourcing to non-OECD economies account for the rise in wage inequality as a result of globalization. Altogether, the founded positive relationship between outsourcing and wage inequality is in line with earlier country specific research (e.g. Feenstra and Hanson, 1996; Tsafack and Parsasnis, 2010).

(34)

30

However, this solves only one side of the problem, as wages of high skilled labor can rise, despite what happens with lower wages. Regarding wages of high skilled workers, it will only decrease when supply of skilled workers become abundant. This can be obtained by a higher educational attainment, however this will lead to an increase in wage dispersions until demand and supply actually become balanced.

(35)

31

References

Acemoglu, D. (2003). Technology and inequality. Technology and Inequality.

Alvaredo, F., Atkinson, A., Piketty, T., & Saez, E. (2013). The Top 1 Percent in International and Historical Perspective. The Journal of Economic Perspectives, 27(3), 3-20.

Asteriou, D., Dimelis, S., & Moudatsou, A. (2014). Globalization and income inequality: A panel data econometric approach for the EU27 countries. Economic Modelling, 36, 592-599. Atkinson, A. B. (2003). Income inequality in OECD countries: Data and explanations. CESifo

Economic Studies, 49(4), 479-513.

Barro, R. and Lee, J. (2013). A New Data Set of Educational Attainment in the World, 1950-2010. Journal of Development Economics, 104, 184-198.

Bernanke, B. S., & Rotemberg, J. J. (1995). NBER macroeconomics annual 1995, volume 10 National Bureau of Economic Research; MIT Press.

Boeri, T., Conde-Ruiz, J. I., & Galasso, V. (2006). The political economy of flexicurity (No. 2006-15). FEDEA.

Card, D. (2009). Immigration and inequality. National Bureau of Economic Research.

Card, D., & DiNardo, J. E. (2002). Skill biased technological change and rising wage inequality:

some problems and puzzles. National Bureau of Economic Research.

Chen, Y., Ishikawa, J., & Yu, Z. (2004). Trade liberalization and strategic outsourcing. Journal

of International Economics, 63(2), 419-436.

Cheng, W., & Zhang, D. (2007). Does trade in intermediate goods increase or decrease wage inequality? The Singapore Economic Review, 52(02), 201-213.

Dabla-Norris, E., Kochhar, K., Suphaphiphat, N., Ricka, F., & Tsounta, E. (2015). Causes and consequences of income inequality: A global perspective (IMF staff discussion note no.

15/13). Washington, DC: International Monetary Fund.

(36)

32

Davis, D. R. (1997). Critical evidence on comparative advantage? North‐North trade in a multilateral world. Journal of Political Economy, 105(5), 1051-1060.

Davis, D. R., & Mishra, P. (2007). Stolper-Samuelson is dead: And other crimes of both theory and data. Globalization and poverty, 87-108. University of Chicago Press.

Davis, D. R., & Weinstein, D. E. (2001). Do factor endowments matter for North-North trade? National Bureau of Economic Research.

Dickens, R., & Manning, A. (2004). Has the national minimum wage reduced UK wage inequality? Journal of the Royal Statistical Society. Series A (Statistics in Society), 167(4), 613-626.

Deardorff, A.V. (2006), Terms of Trade: Glossary of International Economics, Singapore, World Scientific Publishing.

Dreger, C., López-Bazo, E., Ramos, R., Royuela, V., & SURIÑACH, J. (2015). Wage and income

inequality in the European Union.

Driffield, N., & Taylor, K. (2000). FDI and the labour market: a review of the evidence and policy implications. Oxford Review of Economic Policy, 16(3), 90-103.

Feenstra, R. C., & Hanson, G. H. (1996). Globalization, outsourcing, and wage inequality. National Bureau of Economic Research.

Feenstra, R., & Hanson, G. (2001). Global production sharing and rising inequality: A survey of trade and wages. National Bureau of Economic Research.

Figini, P., & Görg, H. (2011). Does foreign direct investment affect wage inequality? An empirical investigation. The World Economy, 34(9), 1455-1475.

Fortin, N. M., & Lemieux, T. (1997). Institutional changes and rising wage inequality: is there a linkage?. The Journal of Economic Perspectives, 11(2), 75-96.

Glyn, A. (2001). Inequalities of employment and wages in OECD countries. Oxford Bulletin of

(37)

33

Gonzales, F., Jensen, J. B., Kim, Y., & Nordås, H. K. (2012). Globalisation of services and jobs. Policy Priorities for International Trade and Jobs, 175-92.

Halliday, Timothy; Lederman, Daniel; Robertson, Raymond. 2015. Tracking wage inequality trends with prices and different trade models : evidence from Mexico. Policy Research

working paper; no. WPS 7471. Washington, D.C. : World Bank Group.

Harrison, A., J. McLaren, and M. McMillan (2011): ”Recent Perspectives on Trade and Inequality, ” Policy Research Working Paper 5754, World Bank.

Herzer, D., & Nunnenkamp, P. (2013). Inward and outward FDI and income inequality: evidence from Europe. Review of world economics, 149(2), 395-422.

Hijzen, A. (2007). International outsourcing, technological change, and wage inequality.

Review of International Economics, 15(1), 188-205.

IMF – International Monetary Fund (2007), “Globalization and Inequality”, World Economic Outlook, IMF, Washington, 31-65.

International Labour Organization (ILO), 2014. Global Wage Report 2014/15: Wages and Income Inequality (Geneva).

International Labour Office (ILO). 2013. ILOSTAT Database (Geneva).

Koeniger, W., Leonardi, M., & Nunziata, L. (2007). Labor market institutions and wage inequality. Industrial & Labor Relations Review, 60(3), 340-356.

Krugman, P. R. (1981). Intraindustry specialization and the gains from trade. Journal of

Political Economy, 89(5), 959-973.

Leamer, E. E. (1996). In search of Stolper-Samuelson effects on U.S. wages. National Bureau of Economic Research.

Lopez Gonzalez, J., Kowalski, P., & Achard, P. (2015). Trade, global value chains and wage-income inequality. Organisation for Economic Co-operation and Development.

(38)

34

Machin, S. (1996). Wage inequality in the UK. Oxford Review of Economic Policy, 12(1), 47-64.

Mankiw, N. G. (1990). A quick refresher course in macroeconomics. National Bureau of Economic Research.

Milanovic, B., & Squire, L. (2005). Does tariff liberalization increase wage inequality? some empirical evidence. National Bureau of Economic Research.

Miroudot, S., Lanz, R., & Ragoussis, A. (2009). Trade in intermediate goods and services. Organisation for Economic Co-operation and Development.

Molnár, M., N. Pain and D. Taglioni (2007), “The Internationalisation of Production, International Outsourcing and Employment in the OECD”, OECD Economics Department

Working Papers, No. 561, OECD Publishing.

Munch, J. R., & Skaksen, J. R. (2009). Specialization, outsourcing and wages. Review of World

Economics / Weltwirtschaftliches Archiv, 145(1), 57-73.

Nour, S. M. (2013). Technological change and skill development in Sudan. Springer Science & Business Media.

OECD. (2011). Divided we stand. Organisation for Economic Co-operation and Development. OECD (2016), "Earnings: Gross earnings: decile ratios", OECD Employment and Labour Market Statistics (database). DOI: http://dx.doi.org/10.1787/data-00302-en (Accessed on 07 April 2016).

OECD (2016), "Earnings: Minimum wages relative to median wages", OECD Employment and

Labour Market Statistics (database). DOI: http://dx.doi.org/10.1787/data-00313-en (Accessed on 13 April 2016).

OECD. (2015). OECD employment outlook 2015. Organisation for Economic Co-operation and Development.

OECD (2016), "GDP per capita and productivity growth", OECD Productivity Statistics (database). DOI: http://dx.doi.org/10.1787/data-00685-en (Accessed on 13 May

(39)

35

OECD (2015), "Gross domestic product (Edition 2015)", OECD National Accounts

Statistics (database).DOI: http://dx.doi.org/10.1787/32a469dd-en (Accessed on 31 May

2016).

OECD. (2016), Income inequality (indicator). doi: 10.1787/459aa7f1-en (Accessed on 05 April 2016).

OECD, Input-Output Tables. For download at http://oe.cd/i-o. Organisation for Economic Co-operation and Development, Paris, 2015. (Accessed on 31 May 2016).

OECD (2016), "Labour Force Statistics: Employment by activities and status", OECD

Employment and Labour Market Statistics (database). DOI:

http://dx.doi.org/10.1787/data-00289-en (Accessed on 10 April 2016)

OECD (2016), "Labour Force Statistics: Population and labour force", OECD Employment and

Labour Market Statistics (database). DOI: http://dx.doi.org/10.1787/data-00288-en (Accessed on 13 April 2016)

OECD (2016), "Labour: Labour market statistics", Main Economic Indicators (database). DOI: http://dx.doi.org/10.1787/data-00046-en (Accessed on 11 May 2016)

OECD (2016), "OECD Economic Outlook No. 99 (Edition 2016/1)", OECD Economic Outlook:

Statistics and Projections (database). DOI: http://dx.doi.org/10.1787/9572784d-en (Accessed

on 20 April 2016)

OECD (2016), "OECD Factbook", OECD Factbook Statistics (database).

DOI: http://dx.doi.org/10.1787/data-00590-en (Accessed on 11 May 2016)

OECD/WTO (2016), "Trade in value added", OECD-WTO: Statistics on Trade in Value

Added (database).DOI: http://dx.doi.org/10.1787/data-00648-en (Accessed on 18 May 2016)

O'Rourke, K. (2003). Heckscher-Ohlin theory and individual attitudes towards globalization. National Bureau of Economic Research.

Snower, D. J. (1999). Causes of changing earnings inequality.

(40)

36

Timmer, M. P., Erumban, A. A., Los, B., Stehrer, R., & de Vries, G. J. (2014). Slicing up global value chains. The Journal of Economic Perspectives, 28(2), 99-118.

Topel, R. H. (1997). Factor proportions and relative wages: the supply-side determinants of wage inequality. The Journal of Economic Perspectives, 11(2), 55-74.

Tsafack, E., & Parasnis, J. (2010). The Impact of Trade and Outsourcing on Skilled and

Unskilled Labour in France (No. 32-10). Monash University, Department of Economics.

Wood, A. (1998). Globalisation and the rise in labour market inequalities. The Economic

Journal, 108(450), 1463-1482.

(41)

37

Appendix A

Tabel 1. List of countries in the samples

(42)

38

Appendix B

Hausman test for fixed/random model

Model 1 ---- Coefficients ---- | (b) (B) (b-B) sqrt(diag(V_b-V_B)) | fe re Difference S.E. ---+--- out2world | 1.719172 1.611406 .1077663 .3485227 --- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg

Test: Ho: difference in coefficients not systematic

(43)

39 Test: Ho: difference in coefficients not systematic

chi2(2) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 0.17 Prob>chi2 = 0.9203 Model 3 ---- Coefficients ---- | (b) (B) (b-B) sqrt(diag(V_b-V_B)) | fe re Difference S.E. ---+--- out2world | 2.204425 2.076232 .1281932 .2721714 shareinwfd~k | .0064784 -.0156277 .0221061 .0250192 tfpgrowth | -1.978589 -1.911602 -.0669864 .1024688 minwage | -2.185786 -2.125948 -.0598377 .1345175 --- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg

chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 1.93

(44)

40 Model 4 ---- Coefficients ---- | (b) (B) (b-B) sqrt(diag(V_b-V_B)) | fe re Difference S.E. ---+--- out2oecd | 1.437 1.396216 .0407839 .6180308 out2nonoecd | 3.36063 3.373126 -.0124957 .5708518 shareinwfd~k | -.0066989 -.0164778 .0097789 .0291572 tfpgrowth | -1.940764 -1.891125 -.0496391 .1419391 minwage | -2.279269 -2.205158 -.0741114 .164016 --- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg

(45)

41 gdpgrowthc~e | .0004665 .0009783 -.0005117 .0000993

outputgap | .9766987 1.025213 -.048514 .0647073

--- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg

chi2(9) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 10.73

Prob>chi2 = 0.2948

(V_b-V_B is not positive definite)

(46)

42 Test: Ho: difference in coefficients not systematic

chi2(10) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 9.89

Prob>chi2 = 0.4504

(V_b-V_B is not positive definite) White test for heteroskedacticity

Model 1 Model 2 Total 22.93 4 0.0001 Kurtosis 8.80 1 0.0030 Skewness 10.81 1 0.0010 Heteroskedasticity 3.32 2 0.1897 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.1897 chi2(2) = 3.32

against Ha: unrestricted heteroskedasticity White's test for Ho: homoskedasticity

Total 27.18 8 0.0007 Kurtosis 4.95 1 0.0260 Skewness 17.93 2 0.0001 Heteroskedasticity 4.30 5 0.5068 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.5068 chi2(5) = 4.30

(47)

43 Model 3 Model 4 Total 39.29 19 0.0041 Kurtosis 4.87 1 0.0274 Skewness 15.09 4 0.0045 Heteroskedasticity 19.34 14 0.1525 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.1525 chi2(14) = 19.34

Total 52.22 26 0.0017 Kurtosis 5.44 1 0.0196 Skewness 17.69 5 0.0034 Heteroskedasticity 29.09 20 0.0860 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.0860 chi2(20) = 29.09

(48)

44 Model 5 Model 6 Total 107.21 64 0.0006 Kurtosis 1.02 1 0.3132 Skewness 30.90 9 0.0003 Heteroskedasticity 75.29 54 0.0294 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.0294 chi2(54) = 75.29

Total 123.60 76 0.0005 Kurtosis 0.76 1 0.3840 Skewness 35.03 10 0.0001 Heteroskedasticity 87.81 65 0.0313 Source chi2 df p Cameron & Trivedi's decomposition of IM-test Prob > chi2 = 0.0313

chi2(65) = 87.81

Outsourcing and wage inequality in OECD countries