The Impact of Globalization and Technological Change on Functional Income Inequality

(1)

The Impact of Globalization and Technological Change on Functional

Income Inequality

E.R. van Huizen S3201414

e.r.van.huizen@student.rug.nl

University of Groningen, Faculty of Economics and Business Supervisor: Tarek Harchaoui

(2)

2 Abstract

The second unbundling caused production chains to spread across the world. The resulting global value chains, can be sorted into different activities. This was a motivation to document how inequality evolved across activity. We used the Theil index to decompose inequality by activities based on data from the World Input-Output Database. Thereafter, the roles of globalization and technological change in determining inequality was empirically

investigated. We found that technology developments had an inequality reducing effect. The effects of globalization are small and differ per activity. Furthermore, technological change reduces inequality within marketing, manufacturing and R&D except for management activities.

(3)

3

After 1990, the revolution in information technology made it easier to coordinate business activities at a distance. This caused production chains to spread across the world, as they were no longer bound to a specific location (Baldwin, 2011a, 2011b). Countries evolved towards exporting a specialized set of activities instead of a type of product (Baldwin, 2006). The increase in trade, the expansion of global value chains and the investments in developing and emerging countries caused income levels to converge across the globe (Subramanian & Kessler, 2014).

Activities in global value chains can be visualized by examining the smile curve of value added. This graph distinguishes four types of activities: management activities, fabrication activities, marketing activities and research & development activities. These activities differ in their value added and in their positioning within the value chain. Some activities are more centred on the input side, such as research and development, whereas others contribute at the output side, for example marketing activities (Mudambi, 2008). Timmer, Miroudot and De Vries (2019) measure the extent to which countries specialize in activities. As a proxy of value added they use the income earned per type of activity, which is called functional income, obtained from the World Input-Output Database.

Their research has inspired writing this thesis, as it points out that a new perspective on the global dispersed economy is needed. Since the sources of income are spread over the globe, it makes sense to not only look at how countries behave in the global economy but to also focus on how activities have evolved over time. This approach offers a new perspective on measuring income inequality. We would no longer be limited to within- and between measures at the macro level but be able to distinguish inequality at a higher level of

granularity. Thus, inequality within management, manufacturing, marketing and research & development activities, to which we apply the Theil index of inequality. The Theil index is chosen, because it is an inequality index which allows to additively decompose inequality values within and between subgroups (Shorrocks, 1980).

After documenting the facts underlying income inequality and identifying some of its broad patterns, we need to discover how much globalization and technological change, two of the commonly mentioned factors that influence income inequality at a global level, contribute to the trends reported by income inequality across these activities. Using a panel of the

inequality groups, we estimate these two relationships with the fixed effects model.

The aim of this study is twofold. Firstly, we examine the trends of income inequality across activities. Thereafter, we search for the effects of technological change and

(5)

5

Topics related to income inequality are becoming more widely discussed in both the academic as in the public debate. As the world becomes more interconnected, a global view on the determinants of inequality is needed. By taking a macro level perspective and focus on stratification of income between countries, the underlying sources of inequality derived from the emergence of globalization are ignored. By analysing inequality from the perspective of economic activities, we would fill this gap in the academic field. Using the Word Input-Output tables furthermore offers an opportunity to move away from country-specific studies and measure determinants of inequality at a global scale and search for its aggregate affects.

This thesis evolved around two themes. The first being the decomposition of income inequality into four groups of activities. We have learned that the share of within group inequality is larger than the inequality between groups of activities. The total inequality index is furthermore decreasing, which is in contrast to existing evidence on income inequality. The second theme was the investigation on the effects of technological change and globalization on inequality. The effect of technological change on the total within inequality component of the Theil index was negative, thus an inequality reducing effect. We also found that the effect of technological change was inequality reducing within manufacturing and marketing. In addition, the effect of research & development and management is positive, thus an inequality increasing effect. The effects of globalization on inequality within activities is relatively small. Inequality within fabrication and marketing activities is reduced by an increase in globalization, whereas the increase of globalization is expected to increase inequality within management.

This thesis is structured as follows. Chapter one will start with reviewing literature on the developments of the global economy, inequality trends and its determinants after 1990. It will then zoom in on the effects of globalization and technological change on income

inequality. In chapter two, we will use functional income data and apply the Theil index to construct a different approach to measure income inequality. Chapter three explores the roles of globalization and technology in determining functional income inequality, by empirically testing the relationship between globalization and technological change. Thereafter we conclude this research by answering the research questions and discussing limitations and further research possibilities.

Chapter One Literature Review

The first chapter of this thesis will describe the changes in the global economy after 1990. The 1990s are often marked as the starting point of an era of hyper globalization (Subramanian & Kessler, 2014). It is therefore used as a starting point to set the stage for this thesis. The period after the 1990s is also called the second unbundling, which points to the “international

(6)

6

inequality. Its aim is to emphasize the impacts of globalization and technological change on inequality within and between the different sectors of the global value chain. It furthermore shall reveal the importance of separating production activities when evaluating income inequality, by looking at the existing literature and inequality trends. We will start with

describing the process of the second unbundling and its consequences for the global economy. Thereafter global inequality trends will be discussed, followed by an analysis of the impact of technological change and globalization on income inequality.

1.1 The Second Unbundling

Before discussing the second unbundling, it is useful to place this phenomenon in its historical context, in order to grasp the importance of its impact on the global economy. We therefore start with explaining a process which is called the first wave of globalization, or the first unbundling. The first unbundling started in the late 19th century. Railways and steamships caused trade costs to fall, which made the absolute necessity to manufacture goods close to where they would be consumed, disappear. This enabled countries to specialize in certain sectors in order to achieve economies of scale (Baldwin, 2011a). This led to competition between countries, sectors and firms. The most important consequence of this period was the big divergence between ‘the North’ and ‘the South’ (Baldwin, 2006). A reversal in economic powers occurred. According to Baldwin (2016), the ancient North (Middle-East, China, India & Pakistan), which accounted for 73% of world manufacturing output in 1750, and 50% in 1830 became the new South, with only 7% of world manufacturing output by 1913. Whereas the ancient South (North-America & Europe) became the new North.

The trend of falling trade costs gradually decreased until approximately 1980.

Globalization after this period was no longer mainly driven by progression in transportation, but driven by lower costs of transmission as a consequence of innovations in the field of information management. This revolution in the ICT made it easier and cheaper to coordinate business activities at a distance. These developments resulted in the second unbundling (Baldwin, 2011a, 2011b). The period from 1990 onwards was marked by falling costs of moving goods, technology, people and most importantly: ideas. The focus of competition between firms or between sectors, shifted to a whole new level: competition between tasks (Baldwin, 2006). This entailed that a worker that is performing a task, faced competition from another worker performing the same task in another country.

(7)

7

Another implication of the second unbundling is the rise of multinational corporations and the increase in flows of foreign direct investment (Subramanian and Kessler, 2014).

Multinationals account for approximately a third of world output and GDP. They furthermore are responsible for about two third of global exports and a third of imports of intermediate inputs (Backer, Miroudot, & Rigo, 2019). So, this entails that multinationals are not just affected by the second unbundling but that they also have a major influence on the trade of goods and services across borders. This can be seen in the relative decline of the share of labour and the increase in the share of capital as a consequence of more intense global competition (Autor, Levy, & Murnane, 2003).

The second unbundling also altered the country related process of industrialization. During the first wave of globalization firms included all the production stages within their factory borders. But after the second wave of globalization this was no longer required. This has been beneficial for developing countries that received labour intensive routine tasks. Multinationals entered the developing nations under favourable conditions. They had the technology, they knew how to manage processes. Developing countries had to deliver a reliable workforce and a friendly business climate (Baldwin, 2011b). This development led to convergence between developing and developed nations, since developing countries were given a chance to industrialize (Gonzalez, Kowalsky, & Achard, 2015). This leads us towards the next subchapter, which discusses the developments of income inequality starting from the second unbundling.

1.2 Inequality Experiences

The period after 1990 was characterized by an increase in trade and the rapid development of information technology. This stimulated convergence between nations, as output was

produced all over the world. This ‘democratization’ of global output has attributed

approximately one third of the total increase in trade (Subramanian and Kessler, 2014). Figure 1 and 2 show the level of real GDP per capita and its growth rate respectively. The results complement the expectations of convergence that were discussed before, as it becomes evident that low- and middle-income countries have a substantial lower level of GDP per capita, whereas in general they experience a higher growth rate than high-income countries. The gap in real GDP per capita between high-income and low- & middle-income countries is decreasing over time. Although the conclusions from these graphs seem to be quite

(8)

8 Figure 1

Real GDP Per Capita (Constant 2011 International $)

Notes. Four country groups are used for this analysis: low income, lower middle income, upper middle income

and high income. For more specifications on this categorization refer to Appendix A.

Figure 2

GDP per Capita Growth (Annual %)

Notes. Four country groups are used for this analysis: low income, lower middle income, upper middle income

and high income. For more specifications on this categorization refer to Appendix A.

These trends are also reflected in income inequality measures. Rodrik (2014) illustrates the concept of income inequality and its components by comparing the income of the average individual to average income. The average individual is a person who is earning the median income in a distribution, whereas average income is the mean of incomes in that same

distribution. If income was spread equally around the world, the income of the global average individual and the average global income would be the same. The larger the gap between the two, the higher income inequality would be, as it would point to an extra share of total income that is held by the group on the right tale of the distribution. Moreover, it would signal that a nation as a whole can grow, but that the average household might not benefit from its

economic prosperity. Table 1 illustrates this effect, it shows the median and mean disposable household income for two developed countries: the United States and Canada, and two

500 5000 50000 2011 in tern at ion al $ Year Low income

Lower middle income Upper middle income High income -8 -6 -4 -2 0 2 4 6 8 10 Year % Growth ra te Low income

(9)

9

developing or emerging economies: China and Taiwan. It points out that during the period 1991 to 2013/2017 the ratio between the median and mean has increased. It appears that overall, the median and mean have increased, however the mean has experienced a sharper growth, resulting in higher inequality within the country.

Since we lack data on global disposable household income we cannot make assumptions on the between country inequality component of overall inequality. We can however learn a little about the mechanisms of these two components by looking at China and the United States. Within the United States, the difference between the mean and median has become larger. This indicates an increase in dispersion in its national income distribution. The increase in average income is smaller for the US than for China, which signals a converging movement between the two. Thus, the within country inequality is increasing, whereas the between country inequality is decreasing. This means that a fast-growing country like China, is an important contributor to decreasing between country inequality. On the other hand, within Chinese borders, there is a majority of the population still living under poor circumstances, because only a small minority possesses the accumulated wealth (Rodrik, 2014).

Table 1

Median and Average Disposable Household Income in Selected Countries

Economy Median Mean Ratio

United States

1991 25212 30224,59 1,20

2016 52223 66871,93 1,28

Percentage increase 207,14 221,25 n.a.

China

2002 13781,4 17843,05 1,29

2013 42738,17 55821,84 1,31

Canada

1991 30851 34707,38 1,13

2017 61372,5 72525,99 1,18

Taiwan

1991 546911 629493,6 1,15

2016 705613,5 843198 1,19

Notes. Source: Luxembourg Income Study (LIS) Database, http://www.lisdatacenter.org (United States 1991,

2016; China 2002, 2013; Canada 1991, 2017; Taiwan 1991, 2016). Luxembourg: LIS.

When we evaluate income inequality across the world we should consider that global income inequality is built on two components. The first component being the within country

(10)

10

deviation from the mean for within country inequality is bigger than the decline in cross-country inequality. It is visible that after 1995 the deviation from the mean for within

inequality is increasing rapidly. Furthermore, the between inequality deviation is decreasing, albeit at a relative smaller pace. One could argue that this is a signal of increasing total inequality, since one component is growing faster than the decline of the other one. However, in absolute numbers the decrease of between country inequality is bigger than the increase in within inequality (Liberati, 2015). Peters (2016) presents the Gini coefficient of inequality from the period 1990 to 2015. He argues that total inequality has been increasing over the past 25 years. Furthermore, the levels of inequality in emerging economies are higher than global inequality and inequality in advanced countries.

Figure 3

Total, between and within inequality, normalized on their means

Note. Figure obtained from Liberati, P. (2015) ‘The World Distribution Of Income And Its Inequality,

1970-2009’, The review of income and wealth, 61(2), pp. 248–273. doi: 10.1111/roiw.12088.

Figure 4

Relative gain in real per capita income by global income level 1988-2008

Note. Figure obtained from Milanovic, B. (2016b) ‘The Rise of the Global Middle Class and Global Plutocrats’,

(11)

11

Now we gained deeper insights in how global income has developed after the 1990s and got a grasp of the trendlines of the two components of income inequality, we will have a look at the famous elephant graph of Branco Milanovic presented in figure 4. This graph illustrates global inequality (the sum of within- and between inequality) over the period 1988 to 2008. Based on this research he argues that: ‘The gains from globalization are not evenly distributed’ (Milanovic, 2016b).

In figure 4, points A, B and C are important for evaluating global inequality trends over the period 1988 to 2008. The people from the left side of the income distribution until point A all experienced a cumulative gain in real income over this period. However, those between the 40th and 60th percentile can be considered the ‘winners of globalization’ as they have experienced gains in real income around 80%. This class consists mostly of people from the middle class of emerging Asian economies, this is not surprising as we have already seen the growth rates of the lower-middle and upper-middle economies in figure 2. According to Rodrik (2014) we can learn that differences across countries average incomes (between-inequality) are the main driver behind global inequality. For that reason, the catching up of developing countries in the global economy is key to reducing global inequality. Which means that the growth experiences of the bottom 50% of the income distribution, is an important determinant for the direction of global inequality. Milanovic (2016b) follows this reasoning and adds that the converging trends of China and India are such an important element of global inequality that eliminating them from the sample would drive total inequality upwards.

Point B consists of the class of people that are situated around the 80th percentile of the global income distribution. That means that these people are richer than the ‘emerging middle class’ from point A. They have, however, witnessed a large decrease in real income growth. In general, people in this class are found at the lower half of the income distribution of OECD countries. It is remarkable that the countries that were growing so fast in the first unbundling, Japan, Germany and the US, now find almost half of their population in this class (Milanovic, 2016b). If we label the people at point A the ‘winners of globalization’ we can call the people at point B, the ‘losers of globalization’.

(12)

12

Summarizing, we can conclude that emerging- and developing economies experienced economic growth as a consequence of the second unbundling. This has resulted in

convergence between nations, which is reflected in the between-country inequality component of total inequality. However, we also found that within countries income inequality is

increasing. In sum, we have found that world inequality is increasing (measured by the Gini coefficient), with emerging economies having the highest levels of inequality. In addition, the experiences of developed countries in the development of within-country inequality differs per country, indicating that also specific domestic elements play a role in determining inequality from one country to another.

1.3 Drivers of Income Inequality

Since inequality is driven by both domestic factors and more general elements, we will provide a framework in which the complexity and interdependence between drivers of

inequality is described. The goal of providing this framework is to illustrate the complexity of explaining and predicting inequality trends. But also, to disentangle two global forces out of this network of forces: globalization and technological change. Figure 5 displays a

combination of two frameworks: a general framework of Förster & Tóth (2015) combined with a more detailed framework proposed by Nolan, Richiardi and Valenzuela (2019).

Figure 5 shows that inequality is the result of the distribution of labour income and capital income. Chen, Los, & Timmer (2018) add another income source: income from intangible assets (e.g. patents, trademarks, the ability to manage production and supplier networks). The share of total labour income would form 51.2% and is showing a decreasing trend. The share of tangible assets is 18.1% and intangible assets is 30.7%, both of which are experiencing an upward trend. Including intangible assets in such a framework as figure 5 and testing its connections with other drivers is an opening for future research.

Demand for labour and capital income is influenced by the need for production (aggregate demand) the market structure (sectors active in the economy and their participation rate) and the productivity of those sectors in the economy. Labour supply is determined by a country’s demographic and societal structure. Capital supply is determined by the earnings and savings of an economy and the functioning of the financial markets. Income from labour and capital is then transformed via taxes and governmental spending into disposable income. Through savings disposable income can be translated into capital supply. It furthermore forms, together with government expenditures and foreign demand the aggregate demand.

The majority of the elements is country specific, just the influence of institutions and legal systems alone (indicated by the red boxes and diamonds), will lead us to think that inequality is not a ‘one size fits all’ phenomenon. So intuitively the notion of Dabla-Norris et al., (2015) who argue that influences of inequality are different for emerging & developing countries and developed countries, is logical. However, we do see that technological change and

globalization are two forces that cannot be restricted by boundaries. To investigate the effects of their presence would require a country specific, in-depth research globally. We have therefore chosen to use another strategy to measure the impacts of globalization and

(13)

13 Figure 5

Drivers of Inequality

Note. Figure is based on the frameworks of Förster and Tóth (2015) & Nolan, Richiardi and Valenzuela (2019).

Please refer to Appendix B for background information on the set-up of this framework

Mudambi (2008) divides the value chain of a company in three categories. Upstream

activities, downstream activities and activities in ‘the middle’. Upstream activities generally exist of research & development (R&D) and design. Downstream activities are those activities that take care of the consumer end of the value chain. Activities include marketing,

distribution, aftersales-service etcetera. Activities in the middle comprise of the

manufacturing and assembly of products, or other standardized processes. These three groups differ in their value added in production. The smile curve, visible in figure 6, plots the

(14)

14 Figure 6

Smile Curve of Value Added

Note. Figure obtained from Mudambi, R. (2008) ‘Location, control and innovation in knowledge intensive

industries’, Journal of Economic Geography, 8(5), pp. 699–725.

In our study value added will be captured by functional income. As we have data on functional income for countries around the globe, we are able to study the global drivers of inequality in a global context. This entails that it is possible to measure the impacts of technological change and globalization on inequality within and between the production stages mentioned before. Now we understand why we are measuring global value chains by functional income; it is time to dive a little deeper in what we are measuring: the influence of technological change and globalization.

1.4 Technological Change

In the beginning of economic growth theories, technology was understood as driving total factor productivity, leading to higher wages and growing productivity for all. However, this mechanism is biased. It appears that some people benefit more from technology than others. This is translated in skill biased technological change, also called the canonical model. Skill biased technological change hypothesizes that the emergence of new technology raises the demand for high skilled workers (Card & DiNardo, 2002). Acemoglu & Autor (2011) argue that technological change is not biased with respect to skills, but they refer to task biased technological change. This means that technology has the ability to substitute certain tasks that used to be conducted by labour. This increases the demand for high-skilled labour, at the expense of low-skilled labour. This consequently leads to a skill premium and the increase of the relative wage of high-skilled labour.

(15)

15

of the income distribution will grow, signalling a higher total income inequality. Since the focus of this research is functional (wage) income we will not zoom in on this topic.

Research has pointed out that the effects of technological change appear to differ per task group (Acemoglu and Autor, 2011). That is why we can now search for which tasks are likely to be harmed and which tasks are benefited by technological change. According to Autor (2013) the task-based approach can be linked to wage inequality by looking at the comparative advantages of tasks. He argues that technological innovations can impact the wages of low- middle- and high skilled workers through three channels. First through the decrease of marginal productivity. Secondly, as automation raises productivity in the middle-skilled tasks, all other activities in the value chain will gain in productivity. It is therefore complementing overall productivity. Lastly, middle-skilled workers need to be reallocated, but they are not as productive in these new tasks. Thereby decreasing overall productivity. This causes the wages of high and low skilled workers to shift outward, indicating an increasing overall wage stratification.

Autor, Levy and Murnane (2003) built a framework in order to identify the effects of computerization on different groups. They categorized tasks in either analytic & interactive tasks or manual tasks, in addition they separated these two categories into routine- and non-routine tasks1_{. Placing the different categories of the smile curve in this classification. We}

could allocate R&D-, marketing- and management activities in the analytic & interactive and non-routine group, whereas we could include manufacturing in the routine manual tasks. Autor, Levy and Murnane (2003) find that activities that increase their usage of computers find a relative increase in cognitive/routine tasks whereas they see a decline in analytic routine tasks. This means due to the adoption of computerization the value added in tasks such as R&D, marketing and management increases, whereas it decreases in fabrication tasks. This is also reflected by a high within-industry skill upgrading rate, in the most computer intensive industries (Autor, Katz, & Krueger, 1998).

Manual non-routine and analytical non-routine can be found at opposite ends of the skill spectrum. Technological change complements those at the higher end, resulting in higher productivity and an increase in output. In contrast, the activities of low-skilled workers are difficult to be outsourced or automated, since physical presence is required onsite. Both groups have experienced employment growth at the expense of the middle class who, as we recall from figure 4 have suffered greatly. This phenomenon is called job polarization (Autor, 2015). Michaels, Natraj, & Reenen (2014) empirically tested the relationship between ICT and job polarization. They found that sectors that experienced the largest adoption of technol-ogy also demanded more high-skilled workers. This means that within a sector, the income for highly skilled workers would increase, whereas that of more routine based workers would decrease, resulting in an increase of income inequality. Both marketing and research and de-velopment activities are exposed to technological change. Marketing is heavily transformed through technological developments, take for example the increasing role of big data in mar-keting strategies (Rust & Huang, 2014). Furthermore, research and development benefits

(16)

16

greatly from the increase in knowledge transfers(Autor et al., 2003). This leads to the follow-ing hypotheses:

Hypothesis 1a: Technological developments increase functional income inequality within the marketing sector.

Hypothesis 2a: Technological developments increase functional income inequality within the research and development sector.

The influence of technological change on tasks that cannot be offshored or substituted is complex. Autor, Levy and Murnane (2002) investigate the relationship between employ-ment and technology by investigating both the manageemploy-ment functions and the routine-analyti-cal jobs in a large bank. It appears that because of the decrease in costs of sending and receiv-ing information companies can experience economies of specialization. Thus, cuttreceiv-ing a job into multiple smaller parts. Whereas on the other hand we have seen managers adopting broader jobs. Which means they embrace a larger portfolio of tasks. This resulted in a greater pay range for the incomes of the management class, but as routine-based analytical tasks were specialized this also resulted in divergence in their salary. Thus, the upshot of technological change would lead to wage divergence in both classes, management and routine-based analyt-ical task.

Hypothesis 3a: Technological developments increase functional income inequality within the management sector.

Verhoogen (2008) points out that to understand the relation between manufacturing and wage inequality one should look at trade theory. Only the most productive plants are able to pro-duce at large scale and deliver high quality products. Those firms are able to export and to survive on the world market. In addition, they are able to pay higher wages to their employ-ees. Since we know that technological change increases productivity, we could argue that an increase in technological change would induce an increase in functional income within the manufacturing industry.

To take the argument one step further, Rodrik (2012) argued that process of catching up due to technological developments in the manufacturing industry was present without con-ditions. This entailed that regardless of geography, policies or institutional influences, poorer countries would gain more through the adoption of technology than the relatively richer ones. This leads us towards the fourth hypothesis.

Hypothesis 4a: Technological developments decrease functional income inequal-ity within the manufacturing sector.

1.5 Globalization

(17)

17

specific conditions, and not just the production function but also the structure of the economy (Nolan, Richiardi and Valenzuela, 2019).

Technological change and trade do not necessarily have to come hand in hand. Autor, Dorn and Hanson (2015) show this in the absence of correlation between trade competition from China on local labour markets and specialization in routine task-intensive production. However, in the same research they argue that trade and technological change can also rein-force each-other, but at different places. It appears that offshoring due to falling trade costs, leads to an increase in productivity of the factors that remain at home (Autor et al., 2015). Grossman & Rossi-Hansberg (2008) also found this productivity effect and argue that globali-zation of tasks is beneficial for the worker whose task is offshored. Low-skilled labour, for example, gains even when some of the tasks it previously performed are being traded. This is possible because firms avoid costs of labour when they offshore tasks. This leads to higher profits, especially in the labour-intensive sector. Expansion of the labour-intensive sector then leads to an increase in demand for the low-skilled labour sector relative to the high-skilled sector. Thus, the low-skilled sector gains from offshoring through this productivity effect.

Due to the shifts described above we can argue that trade has influenced the structure of the economy. The theory of comparative advantage helps us to theoretically underpin this. It predicts that the decrease of trade barriers will lead to specialization of exports in areas for which countries have comparative advantages. This theory is grounded on comparative ad-vantages due to resource endowments. However, due to the second unbundling we can argue that countries specialize their activities. Countries become more specialized in a specific task due to resource endowments of that particular country, which leads to a global distribution of labour (Buckley et al., 2020). How countries specialize in a particular task differs and evolves at different paces. The following research will illustrate this. (Aghion et al., 2005) argue that in India firms within industries react in a heterogeneous manner on the liberalization of indus-trial licencing, which decreased the barriers of setting up businesses. Their main theoretical starting point was the Schumpeterian growth theorem linked to entry threats. They argue that some firms who are higher on the technological frontier, can innovate relatively easier and protect themselves against entry threats. However, others who are lower on the technological frontier cannot achieve this. This led to the hypothesis that within industries inequality would rise due to the increase in trade liberalization. They investigated this theorem in India over the period 1980-1997 and confirmed that delicensing increased performance inequality within the industrial sector.

(18)

18

An approach to determine which sector is tradable and which sector is not, would be to compare the value added per sector in export (Zeugner, 2013). We will use the research of Timmer, Miroudot and De Vries (2019) for these purposes. If we see the relative importance of value added per sector, it appears that the contribution of marketing and fabrication

activities take on average in the period 1999-2011 a higher share than management and R&D. We could argue that marketing and fabrication are more ‘tradable’ than the other two.

Following this line of reasoning we can expect income inequality within sectors to be higher in marketing and fabrication, than in management and R&D. So, we can hypothesize that the increase of globalization influences inequality within sectors positively, and we expect this relationship to be stronger for marketing and fabrication, than for management and R&D.

Hypothesis 5: The rise of globalization increases functional income inequality within marketing, research & development, manufacturing, and management sectors. 1.7 Summary

We can conclude this literature review by arguing that since 1990 the economy landscape has been redrawn, the emergence of information technology played a large role in this

transformation. It caused firms to disentangle their production lines and create global value chains. This led to a decrease in income inequality between countries, but to an increase of inequality within countries. The determinants of income inequality can be categorized by domestic drivers and global drivers. Existing literature only studies the effects of globalization and technological change on a macro level. Taking a country perspective limits research to study the global effects of technological change and globalization. In addition, factors that influence inequality that have risen due to the emergence of global value chain are not

captured in these macro-level models. That is the reason we will construct an inequality index that used functional income groups as different components in the analysis. That means that inequality within management, R&D, fabrication and marketing activities is captured, together with the inequality between the components. We continue this research by firstly, introducing the Theil index of inequality with data on the four functional income groups obtained from Timmer, Miroudot and De Vries (2019). Secondly, we test the effects of globalization and technological change on wage inequality within those four groups.

Chapter Two Data, Measures, Method

2.1 The Theil Index: A Measure of Inequality

In order to divide income inequality into the four components of interest: incomeinequality derived from management value added, income inequality derived from research &

development value added, income inequality through marketing value added and income inequality from manufacturing value added, we have to transform the income stemming from these four groups to an inequality index. We will start with explaining the research of

(19)

19

will test the hypotheses of the influence of technological change and globalization on the different parts of inequality.

2.1.1 Functional Specialization

While discussing the literature on the changes in the global economy from 1990 until now, we have found out that the information and technology revolution caused value chains to spread across the globe. Specialization changed from the sector level towards the task level. This gives reason to argue that trade should no longer be measured by taking the amount of goods a country exports or make inferences about the industries a country is specialized in, since upstream processes are not captured. That is the reason for Timmer, Miroudot and De Vries (2019) to come up with a new method of measuring functional specialization. They used the four stages of production illustrated by the smile curve of Mudambi, (2008) discussed in the previous chapter. The stages differ in factor inputs and in the potential to be relocated. The authors propose a measure to capture value added, based on labour income in each of the four sectors. Their method is divided in two steps, which we will expand with an extra step, in order to obtain an inequality measure.

The first step involves the World Input Output Database. This database provides time series data of the period 1995 to 2011 and covers 40 countries, which can be classified as developing emerging and developed, and a rest-of-the-world measure (Timmer, M.P., Dietzenbacher, E., et al., 2015). The database is used to calculate gross output needed to produce exports. The Leontief inverse is used to generate the output that is needed to satisfy the demand of an additional unit of exports. Equation 1 illustrates this relationship, in which 𝑦 captures output, (1 − 𝐴𝐷)−1 the Leontief inverse, of which (𝐴𝐷) is the domestic input matrix, and 𝑒 is a vector of exports.

𝑦 = (1 − 𝐴𝐷)−1𝑒 (1)

Thereafter 𝑑 is calculated. This is the amount of domestic value added needed to produce exports. 𝑉 is a matrix with the ratio of value added to gross output on the diagonal and zeros on the other places.

𝑑 = 𝑉 ∗ 𝑦 (2)

The second step of the research is to use an external matrix to derive at 𝑓, which is the value added by function 𝑘 in industry 𝑔. The 𝑑 matrix obtained in (2) is multiplied with 𝐵 which is an external matrix with the dimensions 𝑘 ∗ 𝑔 with 𝑘 being the 4 different production functions and 𝑔 the industries.

𝑓 = 𝐵 ∗ 𝑑 (3)

Now we know the value added by function 𝑘 in industry 𝑔 we can build an inequality

measure based on the four different production groups, as we now have a ‘new’ categorization of income that is reflecting the dispersed global economy.

2.1.2 From Functional Income to The Theil Index of Income Inequality

(20)

20

sectors, people work the same number of hours. The data on income per function will be used as the foundation for constructing the Theil index of inequality. This is a decomposable index, which means that it has the characteristics to provide us with a within group inequality

measure and a between group inequality measure. Thus, this decomposition will use the four functions of production, management, R&D, manufacturing and marketing as separate groups. It will generate a measure of income inequality within each of these groups e.g. some managers get higher paid than others, next to four within group inequality measures it will also give a measure that reflects the dispersion of incomes between the groups.

Francois Bourguignon (1979) describes in his work certain axioms or conditions that a reliable decomposition should adhere to.2_{It asks for an aggregative and an additive}

dimension. Aggregative reflecting that within group inequality and aggregate characteristics form total inequality and additive decomposability indicates that the sum of weighted within group inequality and the inequality between the groups will result in total inequality. Since the two variables in the inequality index are income and population, there are two options for additive decomposition. Population weighted decomposition and income weighted

decomposition. (Bourguignon, 1979) Since we are searching for an index that enables us to divide income in four separate groups, we will follow the income weighted decomposition strategy. The Theil index is an example of an inequality index that is fully decomposable, thus is able to distinguish within- and between inequality and is therefore suited for our purposes (Shorrocks, 1982).

The general formula for the total Theil index of inequality is displayed in (4). The index includes the population share of observations (1

𝑛) total income 𝑦𝑖 and the mean of total income 𝜇_𝑦. 𝑇(𝑦) = (1 𝑛) ∑ { 𝑦𝑖 𝜇𝑦log ( 𝑦𝑖 𝜇𝑦)} 𝑛 𝑖=1 (4)

Total inequality can be further decomposed as the sum of a within group and a between group inequality term, visible in (5). In this calculation, the within-term in itself is a weighted sum of subgroup 𝑘 inequality values (Shorrocks, 1980).

𝐼_𝑇𝑘 =(𝐼_𝑊𝑘) + (𝐼_𝐵𝑘) (5)

The formula to decompose inequality is presented in (6).

(21)

21 𝑇 = ∑𝑘𝑖=1𝑆𝑖𝑇𝑖+ ∑ 𝑆𝑖ln 𝑥̅𝑖 𝜇𝑦 𝑘 𝑖=1 for which 𝑠𝑖 = 𝑁𝑖 𝑁 𝑥̅𝑖 𝜇 (6)

The first term on the right-hand side of the equation illustrates the within group term. For which 𝑆𝑖 is the income share of group 𝑘. It is formed by dividing the number of observation in group 𝑖 (𝑁_𝑖) by the total number of observations (𝑁) multiplied by the division of average income of group i (𝑥̅_𝑖) by the total mean income of the total sample. The income share of group 𝑘 is multiplied by the Theil index of that subgroup, which can be calculated by (4). Summing the contributions of each of the subgroups will give the part of inequality that can be explained by within group inequality. The second term on the right-hand side is the

(22)

22 2.2 Functional Income Inequality

Table 2 Inequality Decomposition Year Within Management R&D Market-ing Fabrica-tion Total

within Between Total

1999 0,158 0,145 0,298 0,279 0,880 0,080 0,960 2000 0,163 0,147 0,306 0,281 0,897 0,078 0,975 2001 0,161 0,150 0,300 0,269 0,880 0,073 0,953 2002 0,151 0,147 0,293 0,261 0,852 0,072 0,924 2003 0,146 0,144 0,278 0,252 0,821 0,068 0,889 2004 0,147 0,142 0,264 0,248 0,801 0,062 0,863 2005 0,142 0,137 0,257 0,247 0,783 0,063 0,846 2006 0,144 0,134 0,256 0,244 0,778 0,060 0,839 2007 0,141 0,130 0,249 0,236 0,755 0,059 0,815 2008 0,134 0,125 0,235 0,232 0,727 0,061 0,787 2009 0,141 0,129 0,243 0,226 0,738 0,056 0,795 2010 0,140 0,123 0,245 0,243 0,750 0,060 0,810 2011 0,138 0,121 0,239 0,249 0,746 0,061 0,808

Table 2 provides the outcomes of calculating this for the four groups of interest: management, research & development, manufacturing and marketing. Theil index ranges from 0, which reflects total equality and can take infinite values (OECD, 2016). The general trend in total inequality is decreasing. This is in contrast with the evidence we found in the literature review. First of all, we are not using the net income of the whole world in this sample. We have calculated the functional income shares based on value added in exports. Furthermore, as we aggregated our sample in order to obtain inequality measures per activity, we have

grouped the countries together per activity. This means that a large part of inequality that is measured by the Theil index was subject to between country inequality experiences. We know that inequality within countries is growing rapidly, however we do not see this increase in inequality in this index yet. This leads us to consider whether the value added in exports captures all activities to make valid assumptions on global inequality as measured by the Gini coefficient, and whether or not we are limited by the time period of our research.

Comparing the total within contribution and the between group contribution to overall inequality, we see that the within-part is contributing more to total inequality than between group inequality is doing. Intuitively this makes sense, since we compiled the Theil index out of country data. Inequality within sectors is a comparison of functional income earned in sectors across the world. Therefore, is far more stratified than the deviation between the four sectors.

(23)

23

have a relatively higher inequality within their sectors than management and R&D. One of the explanations for this is that marketing and fabrication activities are more bound to a specific location than management or R&D (Rugman, A., 2005). The downstream activities that are included in marketing activities generally ask for a local appearance e.g. distribution network or local marketing. Furthermore, fabrication activities are bound to a plant. The degree of competition between fabrication or marketing activities globally, is therefore limited because of the rigidity of activities. Management and R&D on the other hand face a more competitive global market. Since it has become easier over time, to manage or design at a distance. This could be an explanation for the relatively large inequality within manufacturing and

marketing activities.

It appears that when we look at the indices separately, we can see that the between group difference is slightly larger in developing and emerging economies. Furthermore, the role of fabrication in determining the within group inequality is bigger in the developing and emerging countries than in the developed world. This is an interesting finding, as this entails that the wages in developing countries are more stratified than in developed countries. Here, comes the theory of Verhoogen (2008) who we discussed before, into practice. He argued that only some firms, who were highly productive, were able to export and become successful. Rodrik (2011) argues that the developing countries exhibit a larger dispersion in productivity than developed countries at the level of individual firms, but also for sectors. As we have data on the value added in exports of countries, it becomes apparent that in the developing world there is a higher degree of firms who are not productive enough to export. Creating a

relatively smaller group of winners and bigger pool of losers, than in the developed world, leading to a more dispersed income distribution.

All subgroups in table 2 show a decreasing inequality trend. However, when we separate the sample in developed and developing/emerging countries, visible in appendix D, we see different results. In developed countries the within inequality for management activities and R&D activities increases. On the contrary, in developing/emerging economies inequality within manufacturing industries increases, as discussed above.

Figure 7

Contribution of Sectoral Within Inequality to Total Within Inequality

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 0% 20% 40% 60% 80% 100% Year Con trib u tio n in %

(24)

24 2.3 Specification and Data

To study the influence of technological change and globalization on the different groups of inequality we will use the decomposed Theil coefficients as the dependent variable in our panel dataset. We measure the four sectors over the period 1999 until 2011 which means that the number of observations (𝑛) for each model will be 4 groups * 13 years = 52 observations. The model can be written in a generalized form as:

𝑦𝑖𝑡 = 𝑎𝑖𝑡+ 𝛽1𝑡𝑒𝑐ℎ𝑖𝑡+ 𝛽2𝐹𝑉𝐴𝑖𝑡+ 𝛽3𝐹𝐷𝐼𝑖𝑡+ 𝛽4𝑔𝑜𝑣𝑒𝑑𝑢2𝑖𝑡+ 𝛽5𝑛𝑢𝑚𝑖𝑡 + 𝑒𝑖𝑡 𝑖 = 1,2,3 The data on income inequality is derived from the research of Timmer, Miroudot and De Vries (2019) who extracted their data from the World Input Output Database (Timmer, M.P., Dietzenbacher, E., et al., 2015). Table 3 presents a brief overview of the explanatory variables used in this analysis. The first independent variable measured, is technological change. This is measured by the research and development expenditures as a percentage of GDP. For

globalization, we searched for an indicator that closely reflected our focus on global value chains. Therefore, as a measure of backward participation, foreign value-added share of exports is used. The control variable that captures employment rigidities is net union membership. According to Card, Limieux, & Riddell (2003) unions have a systematic influence on within and between country inequality. The indicator is derived from the total number of union membership with the exclusion of members outside the active labour force (Visser, J., 2016). Another control variable that captures one of the socio-demographic determinants of inequality is education. We used governmental expenditure on education as a percentage of total governmental expenditure. Lastly, net capital inflows as a percentage of GDP is used to capture the relationship between foreign direct investment and functional income inequality.

Table 3

Explanatory Variables Income Inequality

Determinant Variable Description Source

Technological

change tech

R&D expenditures as % of GDP

World Development Indica-tors (WDI)

Globalization FVA

Foreign value-added share of exports

OECD-WTO: Statistics on Trade in Value Added (data-base)

Financial flows FDI Net Inflows as % of GDP

Educational

poli-cies govedu2

Share of government ex-penditure on education

(25)

25 Chapter Three

Empirical Analysis

3.1 Estimation Methods

To estimate the effect of technological change and globalization on inequality within sector we will pool our data together in a cross-sectional time series dataset. Our dataset is strongly balanced and covers 4 groups: inequality within management activities, inequality within research & development activities, inequality within fabrication activities and inequality within marketing activities over 13 years (from 1999 until 2011). After performing the F-test and the Breusch-Pagan Largrange Multiplier (LM) we have chosen to apply the fixed effects regression method for our analysis. This estimation technique controls for time-invariant differences between our four groups (Park, 2011). In order to correct for serial correlation and groupwise heteroscedasticity we use robust standard errors in our estimations. Appendix E provides the summary statistics of this dataset.

In order to disentangle the separate effects of technological change and globalization on our 4 subgroups, we will make use of dummy variables in our estimation. We furthermore vary in the use of logarithmic scales and interaction effects to give a more elaborate

perspective on the patterns in of our data. Previously, we mentioned that functional income inequality was used as dependent variable. We performed the same analysis for the

logarithmic function of this variable, however due to its low explanatory power, we have chosen not to pursue this transformation in the remainder of this analysis3

3.2 Fixed Effects Model

The following equation: 𝑦_𝑖𝑡 = 𝛼 + 𝛽₁𝑡𝑒𝑐ℎ_𝑖𝑡+ 𝛽₂𝐹𝑉𝐴_𝑖𝑡+ 𝛽₃𝐹𝐷𝐼_𝑖𝑡+ 𝛽₄𝑔𝑜𝑣𝑒𝑑𝑢2_𝑖𝑡 + 𝛽₅𝑛𝑢𝑚_𝑖𝑡+ 𝑒_𝑖𝑡 forms the basis of our analysis. In order to distinguish between inequality within different activities we make use of dummy variables. We use inequality within R&D as a reference group and created 6 dummies. Three for technological change and three for

globalization. This is explained in the following models, of which the estimates can be found under table 4.

𝑦𝑖𝑡 = 𝛼 + 𝜆𝑡𝑒𝑐ℎ𝑖𝑡 + 𝛽1𝑎𝑡𝑒𝑐ℎ𝑖𝑡∗ 𝑑1 + 𝛽1𝑏𝑡𝑒𝑐ℎ𝑖𝑡∗ 𝑑2 + 𝛽1𝑐𝑡𝑒𝑐ℎ𝑖𝑡∗ 𝑑3 + φ𝐹𝑉𝐴𝑖𝑡+ 𝛽_2𝑎𝐹𝑉𝐴_𝑖𝑡∗ 𝑑4 + 𝛽_2𝑏𝐹𝑉𝐴_𝑖𝑡∗ 𝑑5 + 𝛽_2𝑐𝐹𝑉𝐴_𝑖𝑡∗ 𝑑6 + 𝛽₃𝐹𝐷𝐼_𝑖𝑡+ 𝛽₄𝑔𝑜𝑣𝑒𝑑𝑢2_𝑖𝑡+ 𝛽₅𝑛𝑢𝑚_𝑖𝑡+

𝑒_𝑖𝑡 (1)

𝑦𝑖𝑡 = 𝛼 + 𝜆log (𝑡𝑒𝑐ℎ)𝑖𝑡+ 𝛽1𝑎log (𝑡𝑒𝑐ℎ)𝑖𝑡∗ 𝑑1 + 𝛽1𝑏log (𝑡𝑒𝑐ℎ)𝑖𝑡∗ 𝑑2 + 𝛽1𝑐log (𝑡𝑒𝑐ℎ)𝑖𝑡∗ 𝑑3 + φlog (𝐹𝑉𝐴)_𝑖𝑡+ 𝛽_2𝑎log (𝐹𝑉𝐴)_𝑖𝑡∗ 𝑑4 + 𝛽_2𝑏log (𝐹𝑉𝐴)_𝑖𝑡∗ 𝑑5 + 𝛽_2𝑐log (𝐹𝑉𝐴)_𝑖𝑡∗ 𝑑6 + 𝛽₃𝐹𝐷𝐼_𝑖𝑡+ 𝛽₄𝑔𝑜𝑣𝑒𝑑𝑢2_𝑖𝑡 + 𝛽₅𝑛𝑢𝑚_𝑖𝑡+ 𝑒_𝑖𝑡 (2) 𝑦_𝑖𝑡 = 𝛼 + 𝜆𝑡𝑒𝑐ℎ_𝑖𝑡 + 𝛽_1𝑎𝑡𝑒𝑐ℎ_𝑖𝑡∗ 𝑑1 + 𝛽_1𝑏𝑡𝑒𝑐ℎ_𝑖𝑡∗ 𝑑2 + 𝛽_1𝑐𝑡𝑒𝑐ℎ_𝑖𝑡∗ 𝑑3 + φ𝐹𝑉𝐴_𝑖𝑡+ 𝛽2𝑎𝐹𝑉𝐴𝑖𝑡∗ 𝑑4 + 𝛽2𝑏𝐹𝑉𝐴𝑖𝑡∗ 𝑑5 + 𝛽2𝑐𝐹𝑉𝐴𝑖𝑡∗ 𝑑6 + 𝛽3𝐹𝐷𝐼𝑖𝑡+ 𝛽4𝑔𝑜𝑣𝑒𝑑𝑢2𝑖𝑡+ 𝛽5𝑛𝑢𝑚𝑖𝑡+ 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 + 𝑒𝑖𝑡 (3)

(26)

26

𝑦_𝑖𝑡 = 𝛼 + 𝜆log (𝑡𝑒𝑐ℎ)_𝑖𝑡+ 𝛽_1𝑎log (𝑡𝑒𝑐ℎ)_𝑖𝑡∗ 𝑑1 + 𝛽_1𝑏log (𝑡𝑒𝑐ℎ)_𝑖𝑡∗ 𝑑2 + 𝛽_1𝑐log (𝑡𝑒𝑐ℎ)_𝑖𝑡∗ 𝑑3 + φlog (𝐹𝑉𝐴)𝑖𝑡+ 𝛽2𝑎log (𝐹𝑉𝐴)𝑖𝑡∗ 𝑑4 + 𝛽2𝑏log (𝐹𝑉𝐴)𝑖𝑡∗ 𝑑5 + 𝛽2𝑐log (𝐹𝑉𝐴)𝑖𝑡∗ 𝑑6 + 𝛽₃𝐹𝐷𝐼_𝑖𝑡+ 𝛽₄𝑔𝑜𝑣𝑒𝑑𝑢2_𝑖𝑡 + 𝛽₅𝑛𝑢𝑚_𝑖𝑡+ log( 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛)𝑒_𝑖𝑡 (4)

Table 4

Influence of globalization and technological change on functional income inequality, 1999-2013: Fixed effects estimates

Independent Variables (1) (2) (3) (4) Intercept .522** .448** 1.09 1.06 (7.79) (6.93) (1.93) (1.69) Technological change -.046** -.068** -.433 -1.71 (-4.39) (-5.22) (-1.28) (-1.08) d1 (tech*fab) -.021*** -.037*** -.021*** -.04***

(-1.9e+6) (-1.3e+14) (-8.3e+6) (-9.7e+13)

d2 (tech*man) .016*** .021*** .016*** .021***

(1.4e+6) (7.0e+13) (6.1e+06) (1.6e+13)

d3 (tech*mar) -.084*** -.126*** -.084*** -.126***

(-7.6e+6) (-3.3e+14) (-3.2e+7) (-1.3e14)

Globalization .001 -.015 -.019 -.169

(0.97) (.82) (-1.11) (-1.01)

d4 (FVA*fab) -.000*** .003*** -.000*** .003***

(-4.4e+4) (4.2e+12) (-3.8e+4) (8.7e+11)

d5 (FVA*man) .000*** .006*** .000*** .006***

(1.1e+5) (7.8e+12) (9.3e+4) (1.8e+12)

d6 (FVA*mar) -.003*** -.068*** -.003*** -.068***

(-1.8e+6) (-6.6e+13) (-1.5e+6) (-1.9e+13)

FDI -.001* -.001 -.000 -.000 (-2.37) (-2.33) (-1.52) (-1.81) Educational policies -.018* -.017* -.018* -.017* (-2.47) (-2.47) (-2.42) (-2.44) Union membership .000 .000 .000 .000 (1.17) (2.06) (1.57) (2.30) Interaction .014 .492 (1.12) (1.03) Wald test .00*** .00*** .00*** .00*** R-squared .50 .39 .50 .39

Notes. N=52 (4 groups x 13 years). In brackets the corresponding T-statistics. Achieved level of significance *p

≤ .1, **p ≤ .05, *** ≤ .001.

The separate effects of technological change and globalization on inequality can be found by using both the estimator of the ‘full’ variable and the dummy variables, illustrated by (5) and (6).

Δ𝑦𝑖𝑡

Δ𝑡𝑒𝑐ℎ𝑖𝑡= 𝜆 + 𝛽1𝑎𝑑1 + 𝛽1𝑏𝑑2 + 𝛽1𝑐𝑑3 (5)

Δ𝑦𝑖𝑡

(27)

27

The separate effects of technological change and globalization are then 𝜆 for R&D activities, 𝛽1𝑎 for fabrication activities etcetera.

Table 5

Aggregate Effect of Technological Change and Globalization on Functional Income Inequality Technological change 𝜆 + 𝛽_1𝑎𝑑1 + 𝛽_1𝑏𝑑2 + 𝛽_1𝑐𝑑3 Globalization 𝜑 + 𝛽_2𝑎𝑑4 + 𝛽_2𝑏𝑑5 + 𝛽_2𝑐𝑑6 Model (1) -.135** -.002 Model (2) -.21** -.074 Model (3) -.522 -.022 Model (4) -1.742 -.228

Notes. Achieved level of significance *p ≤ .1, **p ≤ .05, *** ≤ .001.

3.3 Implications

The results obtained from table 5 show that only for model (1) and (2) the results for techno-logical change are significant for the total within functional income inequality variable. Therefore, we can assume that the effects of globalization or technological change may not always be significant when we consider a general income measure. However, when we distin-guish between the four groups of activities in value chains, it appears that for the technologi-cal change dummies, fabrication-, management- and marketing activities are highly signifi-cant across all models. Our reference group, research & development, is only signifisignifi-cant in the models without interaction affects (1) and (2). The globalization dummies for fabrication, management and marketing are highly significant for each model. The reference group, re-search & development, is not significant in any of the models.

(28)

28

We start the analysis of the effects of technological change on income inequality within the separate activities with manufacturing activities. We see a highly significant nega-tive effect of technological change on inequality within fabrication activities. A one unit in-crease in technological change would cause inequality within fabrication to dein-crease with .021 units. This is in line with our hypothesis that the increase in technological change influ-ences income inequality within management industries negatively. The effect of technological change on management activities is highly significant and shows a positive sign. A one unit increase in technological change increases inequality within management activities with .016 units. This is in line with our hypothesis that inequality within the research and development sector increases, when technological change occurs. The effect of technological change on marketing activities is highly significant and, against expectations, showing a negative sign. This entails that a one unit increase in technological change would decrease inequality within marketing activities with .003 units. Therefore, we cannot confirm our hypothesis that the in-crease in technological change influences inequality within the marketing sector positively. Further research would be needed to gain more insights into this relation. The effect of tech-nological change on activities in research and development is significant at the 5% level. This estimated relationship is also, against expectations negative. This means that a one unit in-crease in technological change would dein-crease income inequality within research and devel-opment activities with .046 units. Based on this estimation, we cannot confirm our hypothesis that the increase in technological change influences inequality within the research and devel-opment sector positively. More research is needed to find a theoretical rationale behind this relationship.

The effects of globalization on income inequality within different groups of a global value chain are relatively smaller than the effects of technological change. Evaluating the re-sult of the estimated relationship between globalization and income inequality within manu-facturing activities, we see a highly significant negative relationship, but with a very small co-efficient. We can therefore not make inferences about the magnitudes of the relationship, but only whether or not the sign is confirming our hypothesis. For globalization activities we made just one hypothesis, stating that inequality within activities would increase due to an in-crease in globalization. The estimated relationship between manufacturing and globalization is not in line with this hypothesis. The effect of globalization on management activities is highly significant and positive. Its coefficient is however very small. We can therefore not make any inferences on the size of the relationship, but we can only confirm that this result is in line with our general hypothesis. The effect of globalization on marketing activities is highly significant and negative. This implies that a one unit increase in globalization would decrease inequality within marketing activities with .003 units. This is not in line with our hy-pothesis, that globalization would increase inequality within activities. Finally, the relation-ship between globalization and inequality within research and development activities is not significant. This means that we cannot make inferences about this relationship. The results of this last estimations lead us to not confirm our hypothesis that the increase of globalization in-fluences inequality within sectors positively.

(29)

29

foreign direct investment. These forces have mixed impacts on inequality as pointed out by (Jaumotte et al., 2013). They argue that increase in trade has a tendency to decrease inequal-ity, whereas financial globalization and foreign direct investment have the propensity to in-crease it. Further research is needed that dives deeper into the relationship between globaliza-tion and income inequality. It should break down the different roles of globalizaglobaliza-tion and fig-ure out the effects on different activities, as these results show that it is not possible make a general hypothesis on this relationship.

Conclusion

The purpose of this study was to investigate the drivers of income inequality by taking an approach that better fitted the global economy. We used two themes to accomplish that. First we constructed the Theil index of inequality. We learned that the share of within group inequality is larger than the inequality between groups of activities. An explanation for this could lie in the construction of the index, as 39 developed and developing countries were assembled to obtain the inequality index. The Theil index also points out that total inequality has been decreasing for the time period 1999-2011. This is in contrast to existing evidence on income inequality. A possible explanation could be the usage of export data, however more research is needed to identify the possible causes.

After reviewing literature on the relationship between globalization and technological change and income inequality within management, manufacturing, marketing and R&D activities we proposed five hypotheses. We hypothesized that technological development would increase functional income inequality within management, marketing and research & development but decrease functional income inequality within manufacturing. Furthermore, we hypothesized that globalization would increase functional income inequality within all activities.

We were able to confirm our hypothesis for the influence of technological change on management and manufacturing activities, however for R&D and marketing this was not the case. More research is needed to explain this relationship. Furthermore, the estimations on the relationship between globalization and within inequality showed us that it is not possible to make a general hypothesis on the effects of this relationship. More research is needed to disentangle the effects of globalization on each activity.

(30)

30

manufacturing share in the world market increased for example from 2.3% in 1990 to 18.8% in 2013 (Autor, Dorn, & Hanson, 2016). This gives us reasons to believe that the

measurement of some of the variables form a limitation of the results.

Future research should examine the opportunities that decomposable income measures offer. Decomposing income inequality in the three categories proposed by Chen, Los and Timmer (2018) for example, will shed a new light on the roles of capital and non-capital income in the income distribution. Furthermore, functional income is the product of the number of workers, their productivity and the labour income share in value added (Timmer et al., 2019). These elements could be used for hierarchical decomposing income inequality. These ideas, however, could also be limited by the availability of data across countries and over time.

References

Acemoglu, D., & Autor, D. (2011). Skills, Tasks and Technologies: Implications for

Employment and Earnings. In Handbook of Labor Economics (4b ed., pp. 1044–1166). Elsevier B.V. https://doi.org/10.1016/S0169-7218(11)02410-5

Aghion, P., Burgess, R., Redding, S., & Zilibotti, F. (2005). Entry Liberalization and Inequality in Industrial Performance. Journal of the European Economic Association,

3((2-3)), 291–302.

Alvaredo, F., Atkinson, A. B., Piketty, T., & Saez, E. (2013). The Top 1 Percent in

International and Historical Perspective. Journal of Economic Perspectives, 27(3), 3–20. Atkinson, A. B. (2015). Inequality: What Can Be Done? (1st ed.). London: Harvard

University Press.

Autor, D. H. (2013). The task approach to labor markets: an overview. NATIONAL BUREAU

OF ECONOMIC RESEARCH, 18711, 1–30.

Autor, D. H., Dorn, D., & Hanson, G. H. (2015). Untangling Trade And Technology: Evidence From Local Labour Markets. The Economic Journal, 125(548), 621–646. Autor, D. H., Dorn, D., & Hanson, G. H. (2016). The China Shock: Learning From Labor

Market Adjustment To Large Changes In Trade. National Bureau Of Economic

Research, 21906. Retrieved from http://www.nber.org/papers/w21906%0A

Autor, D. H., Katz, L. F., & Krueger, A. B. (1998). Computing inequality: have computers changed the labor market? The Quarterly Journal of Economics, 113(4), 1169–1213. Autor, D. H., Levy, F., & Murnane, R. J. (2003). The Skill Content Of Recent Technological

Change: An Empirical Exploration. Quarterly Journal of Economics, 118(4), 1279– 1333.

Backer, K. de, Miroudot, S., & Rigo, D. (2019). Multinationals enterprises in the global

economy: Heavily discussed, hardly measured. Retrieved from

https://voxeu.org/article/multinational-enterprises-global-economy Baldwin, R. (2006). Globalisation: the great unbundling(s).

The Impact of Globalization and Technological Change on Functional Income Inequality

The Impact of Globalization and Technological Change on Functional

Income Inequality

Table of Contents