Digisation and the continuous advancement of Artificial Intelligence (AI) incites projections about automation altering occupations over many sectors and countries. With increasing availability of digital infrastructure, many parts of emerging economies will open their doors and become subject to digitisation, robots and AI simultaneously (Messina 2016, Das and Hilgenstock 2018), which for advanced economies have appeared slower, with less established power structures and at different times. With high uncertainty around and little literature on the subject, the societal implications of automation in emerging economies are ambiguous, with potentials of automation boosting economies and leveling the playing field of development one one hand, but also leading to exploitative dynamics, in which companies with intellectual property rights and know-how on automation technologies inflict harm on labour markets, economies and thus many societal layers of emerging economies. While a multitude of indicators on job transformation through digitisation, robots or AI and their respective implications on employment emerged and were discussed within the literature for advanced economies, lack of abundant data causes ongoing uncertainty for emerging economies. This thesis addresses the scientific gap of who is at risk of automation in emerging economies. The literature review entails an overview on the field of labour automation, the social effects of automation, and the (projected) case for emerging economies. The quantitative analysis entails the calculation of two automation indicators, Routine Task Intensity for automation through digitisation and robots, and Suitability for Machine Learning (SML) for automation through AI for emerging economies. SML and RTI are compared among occupations and socio-demographic groups as well as their relationship towards each other. To identify patterns to susceptible socio-demographic groups within the labour markets of emerging economies, a cluster analysis is performed on the most pertinent and yet orthogonal demographic parameters. The attained insights on socio-demographic clusters and their susceptibility towards automation are discussed to develop are broader picture on how automation affects emerging economies in order to educate policymakers and aid in rethinking policy approaches in response to changes in skill demands. This thesis finds ground for concern that emerging economies will be subject to a not just broader but also likely more sudden wave of digital automation than advanced economies, potentially magnified through historic power imbalances and global market dynamics. Our results demonstrate that also for the task compositions of emerging economies, it can be expected that with the advent of AI a substantively larger share of tasks, occupations and socio-demographic groups is susceptible to automation as compared to automation through digitisation and robots. Workers with low and high education show higher automation susceptibilities than mid-educated workers, although through different technologies. The finding that also high-educated workers are susceptible motivates the conclusion that education systems might see their traditional societal role of enablers of social mobility endangered. Finally, this thesis finds that women are systematically more susceptible to AI-enabled automation and the observation that men are more susceptible to automation through digitisation and robots in advanced economies does not hold true for emerging economies. This thesis concludes by calling for particular consideration of emerging economies in the field of automation in labour markets, the merging of more granular data for emerging economies and an institutional model of automation.

Massively reducing Greenhouse Gas (GHG) emissions is one of the biggest challenges of the 21th century. An important deployment of Renewable Energies (RE) can be a key element in attaining this objective. Setting a (high) carbon price would greatly help as it would send a powerful signal to RE investors that it is time to switch from high-emitting firms to free-emitting ones. Carbon pricing is implemented through the setting of either a carbon tax or an Emissions Trading Scheme (ETS). The European Union (EU) implemented its own ETS, the EU ETS, in 2005. Since then however, it faced low permit prices (partly due to an oversupply of both carbon permits and carbon offset credits), which raises a crucial question: how could the EU ETS and its carbon offsets market be improved to generate higher permit prices, thus promoting RE investment? Two policy measures seem relevant today: implementing a floor price for permits (not examined in this work) and/or modifying the rules of the carbon offset credits market of the EU ETS. This master thesis will focus on offsets policy instruments the EU ETS could implement, drawn from a paper published by Bento et al. (2015), that aim at better designing its carbon offsets market and resulting in a higher permit price. The policy instruments analyzed are: stricter baselines, trade ratios and limits. This work will then discuss the efficiency (in regard to society welfare optimization) of these measures.

Over the course of the past centuries, humanity’s influence on the Earth’s atmosphere and climate system has become so significant as to induce geophysical feedback effects, specifically in the form of a self-reinforcing process of global warming. In order to contain the global warming and prevent the expected impacts of climate change from happening, it has become imperative to undertake drastic steps directed towards reducing emissions of greenhouse gases (including CO2). Action needs to be taken soon, as with the growing population, the demand for carbon-intensive (conventional) energy sources is growing and thus greenhouse gas emissions (the main cause for global warming) are increasing. To slow down global warming, energy conservation, energy efficiency improvement and renewable energy generation are the only practical solutions available to mankind...

The residential energy consumption and energy affordability are crucial in policy design, with regard to energy efficiency and energy poverty. This has become even more relevant now, as the energy transition is taking off and the energy prices are soaring. Given the inequality in the energy consumption and energy affordability, it is therefore critical to know which factors are impacting these phenomena, and how they are impacting these phenomena. The housing-related factors are the least understood factors impacting the energy consumption and energy affordability, even though these factors are key pillars of energy policy. Thus, this research assesses how housing-related factors impact the residential energy consumption and energy affordability in the Netherlands. For this assessment, multiple data sources are combined over the years 2016 to 2018, covering a total of more than 8 million Dutch households. A regression analysis is performed on a postal code 6 level using the housing-related factors as predictors, while controlling for socio-demographic and weather factors. This revealed that the housing-related factors impacting energy consumption and energy affordability are relatively similar. This study finds that the WOZ-value is a strong predictor for both phenomena as houses with a higher WOZ-value have a lower energy efficiency. Furthermore, it shows that rental households consume less energy but spend a similar percentage of their income on energy, and that the impact of the year of construction on both phenomena has increased significantly, when compared to earlier studies. Based on this, it is concluded that rental houses and houses with an older year of construction deserve more attention in Dutch energy policy. Furthermore, low-income households should be targeted in short-term energy policy, as this study shows that the current soaring energy prices have a severe impact on the energy affordability in the Netherlands. Insights are however broader, and can be pivotal in energy demand and energy affordability projections, both in the Netherlands and the rest of Western Europe.

National policies that seek to stimulate the energy transition require attention with the requirement for generation sources that can compensate for moments when wind and solar plants are not available. Electricity supply needs to adjust to demand in real-time, at least until energy storage becomes commercially viable, otherwise the energy security is compromised. Hence, the dilemma for energy policies is to continue to promote the expansion of renewables at the same time they secure the provision of flexible sources. One well-known technology for such is the combined-cycle gas turbine (CCGT). It has been used for a long time, however now there is concern over its economic viability due to the merit order effect of renewables. Germany is a front-runner in the energy transition and, recently, it enacted the Energy Market 2.0 to lay out a market design to accommodate the increasing intermittency. The underlying belief is that a competitive environment will allow market forces to provide the necessary flexibility in an economic efficient manner. Hence, the success of the Energiewende in this new policy framework depends on how the generation firms perceive the market conditions and investment incentives. This research investigates how these market actors and the federal government will perceive CCGT generation competitiveness, using generation costs as a criterion and external costs to account for social efficiency. A tailored Levelized Cost of Energy (LCOE) calculation is used to evaluate a CCGT power-plant in the context of a high presence of renewables. The method is broadly used and conveys the complexities of investment appraisal in a relatively simple manner, however there is no prior research that applies it to the described scenario. Energy storage is also investigated for comparison purposes. This research finds that CCGT generation firms in Germany will not have sufficient incentive to invest in new power-plants between 2020 and 2030. It is also found that the technology is not a cost-effective option for flexibility from a societal perspective either. Energy storage is more cost-advantageous than CCGT, especially when external costs are considered. Therefore, CCGT plants are expected not to be attractive for generation actors under the German energy policy framework. Vanadium batteries pose as a promising alternative for the system after 2030, but its uncertain cost development requires public investment to guarantee it effectively becomes competitive. Policies to incentivize CCGT generation seem economic inefficient and are not advisable, as the implementation of new plants during the next ten years would create long-term lock-ins.

Since the last decade, climate change and environmental awareness has received increasingly more attention from politicians and companies. In order to preserve earth’s climate, a lot has to be changed in current industry standards. In 2018, heavy-duty road transport was responsible for 680 million tonnes kg in transported commodities in the Netherlands (CBS, 2019). This resulted in 10.176 million tonnes of emitted CO2 in the same year (CBS, 2020). As these transportation activities are a big part of the current pollution, there is a lot to be gained in this area. Therefore, industries are investigating to make their transportation activities more sustainable.Large manufacturing plants often apply specific supply chain strategies in order to reduce waste. One of those wastes are warehousing costs. Just-in-Time (JIT) and Just-in-Sequence (JIS) are supply chain strategies that aim to reduce these costs (Bányai & Bányai, 2017). A JIT supply chain strategy refers to a strategy where commonly used parts are delivered at the plant when needed, therefore reducing storage capacity. A JIS strategy goes even further. When products are customizable, a JISstrategy ensures that the required parts are delivered in the sequence of production.Upon till now, a lot has been researched with regard to alternative fuels in heavy-duty transport, especially in the hydrogen and battery electric area. However, these studies often stay on a general logistics level. Specific concepts such as JIT and JIS may influence choices for alternative fuels due to more important time constraints such as a predefined and strict pearl chain horizon. Therefore, the aim of this thesis was to research the feasiblity ranking of alternative fuels (hydrogen, battery electric, HVO and LNG) in a JIT/JIS supply chain.

This study defines who is at risk of automation and discusses policies to ensure the vulnerable groups are seen. The study of Nedelkoska and Quintini (2018) is taken as the role model: The risk of automation for individuals across OECD countries is calculated by associating the expert assessment conducted by Frey and Osborne (2013) with individuals’ skills used at work collected by PIAAC. The analysis is improved by training the model with different country datasets and including additional skills into the analysis. 14% of the total employment of 33 countries is found to be at significantly high risk of automation. Workers at the highest risk of losing their jobs are more likely to be less-educated, low-income earners who perform unskilled jobs. The risk of automation declines as the level of education increases. Therefore, this study highlights the importance of training and reskilling the risky-groups to cope with the possible adverse effects of technological progress.

This research provides insight into the implementation of CCS in industrial production from a consumer perspective (i.e., the cost impact on the final product). The impact of implementing CCS in two different industries (e.g., cement and steel) to produce a final product is analyzed. Also, this research provides insight into CCS implementation in industries with high consumption products. It can be concluded that implementing CCS in industrial processes offers a solution to decarbonize industrial processes with a small impact on the final product cost.

With the rise of globalisation, countries have become more connected financially and global cross-border investment flows have become more common. FDI is an important form of cross-border flow which is responsible for the spread of technology across countries and is the main source of external finance for emerging countries. In the last two decades, FDI has increased tremendously. But this has been accompanied by fears about outward FDI taking away production activities and jobs away from the home country. I look at how outward FDI affects home country investment. One can intuitively understand that a dollar of money spent abroad means a dollar less to invest in the domestic economy. Based on the theory of the financially constrained firm, I hypothesize that outward FDI reduces domestic fixed capital investment and R&D spending. I also develop a theoretical framework to distinguish the varying effects of outward FDI on domestic investment across traditional and R&D-intensive industries.By using industry-level panel data for 18 OECD countries covering the period 1995-2009, I regressed the shares of Gross Fixed Capital Formation (GFCF) and R&D spending separately on the share of outward FDI both for all industries as well as specifically for traditional and R&D-intensive industries. While, outward FDI had a negative effect on domestic capital investments at the aggregate level, it did not have any significant effect while looking at specific industry types. This could be because of the reduced sample size in the individual types. Outward FDI had a negative effect on domestic R&D spending at the aggregate level and for R&D-intensive industries, but it had a positive effect for traditional industries. Thus, while fears about outward FDI taking away domestic fixed capital investments are valid, outward FDI can have both a positive and negative effect on R&D expenditure, depending on the type of industry. These results can help MNCs make strategic investment decisions taking into account their effect on their home country industry. It can also help policymakers formulate tax and industrial policies to promote home country investments.

This thesis examines the competitiveness of renewable energy technology compared to conventional power generation methods by examining the relative cost of solar PV, wind, and coal. The findings of this analysis seek to answer the primary hypothesis: The unique application of Levelized Cost of Electricity (LCOE), entailing the manipulation of CAPEX and CO2 variables, reveals the points in time and the exact LCOE values where it becomes cost competitiveness to switch from a coal asset to a renewable asset within the context of the German electricity market. The results illustrate that cost competitiveness between new renewables (solar PV, onshore wind, and offshore wind) and coal in the German electricity market does cross at certain points. The research seeks to determine the precise scenarios under which new renewable plants are more cost competitive than existing coal plants. The LCOE calculation has been chosen as the methodology to estimate the relative cost competitiveness of these differing power generation methods, under predefined scenarios for both CAPEX-reduction for renewables and CO2-price increase for coal. New assets are compared to existing assets; therefore, LCOE has been chosen in order to compare operational costs (OPEX) plus the capital cost (CAPEX) of new assets, with only the operational cost (OPEX) for existing assets. i.e. the investment cost for coal-fired assets is denoted as a sunk cost. This allowed for the development of a simplified framework to analyze an electricity-producing firm involved in an investment decision, whether a new solar or wind project or the continuation of an existing coal power plant...

Accurately forecasting charging demand is a major factor in energy providers’ ability to meet said demand. A number of methods have been proposed in the literature. This master thesis aims to provide a framework for energy providers to forecast charging demand due to increased EV adoption. Specifically, this research examines the effects that rental EV adoption has on charging demand at Icelandic energy provider ON Power’s fast charging stations. Further, it establishes how increased technological developments and infrastructure spending can be balanced to avoid unnecessary costs, and if ON Power’s charging infrastructure is prepared to meet the electricity demand that a fully electrified rental car fleet would bring. To answer these questions, an agent-based modeling framework was constructed to simulate charging demand in ON Power’s network. The model simulates all tourists that use rental cars in driving the Icelandic ring road as if the entire fleet consisted of EVs. A case study is conducted where the Icelandic ring road is split into nine zones, and six scenarios that each reflect different technological developments are simulated using the modeling framework. The research findings indicate a significant change in charging demand. Firstly, rental EVs affect seasonal charging demand with more than half of the demand occurring in the summer months of June-August. Secondly, geographical charging demand shifts from mostly being in the greater capital area, to mostly being outside of it. Thirdly, a sufficient charging infrastructure at Keflavík Airport is a major factor in the electrification of the Icelandic rental car fleet. The research findings further indicate that with the right strategy, it is well within reach for ON Power to handle the rental EV charging demand. A strategy is proposed for ON power to avoid unnecessary spending while still meeting charging demand, which considers current demand, future demand, and how new charging station locations should be determined.

By signing the Paris Climate Agreement, The Netherlands has committed itself to curtail its CO2- emissions in order to keep global warming well below 2°C above pre-industrial levels. Pivotal to achieve these targets is the phase-out of CO2-intensive electricity generation technologies and investments in renewables. Current investments aiming to decarbonise the electricity system are predominantly allocated to solar PV and off- and onshore wind energy. However, this trajectory might change as the development of Dynamic Tidal power (DTP) might soon enable The Netherlands to harvest the North Sea’s currently unutilised tidal currents to generate clean electricity. DTP uses a 30-70 kilometre long dam perpendicular to the coast to capture the North Sea’s tidal currents. The alternating currents that proceed parallel to the coast create a hydraulic head over the dam, which turbines in the dam convert into electricity. However, the variable availability of solar irradiation, wind, and tidal currents makes it increasingly complex and costly to match electricity supply and demand as their shares in the electricity mix increase. To ensure electricity security, freely dispatchable energy generators (e.g. natural gas, biomass or coal turbines) are deployed to cover the electricity load that could not be met by the variable renewable energy (VRE) generators. However, due to the fuel used to generate electricity, dispatchable energy sources tend to emit CO2 and bear higher marginal energy generation costs than VRES. To address this problem, the portfolio shares of solar PV, offshore wind, onshore wind and DTP that minimise the need for dispatchable backup capacity and the energy generation costs were computed in this study. Due to its novelty, special attention was paid to the effect of DTP on a VRE system’s ability to efficiently match supply and demand. In order to achieve these objectives, a novel application of the Modern Portfolio Theory (MPT) was used. The MPT originates from the stock market but is often used to comprise VRE portfolios that maximise the electricity output. However, as the aim was to minimise the residual load, demand-variability was introduced into the MPT. Existing literature had not covered this topic. Hence, by assessing to what extent including demand-variability in the MPT affects the selection of efficient VRE portfolios, this study filled a gap in the literature and serves as a starting point for future research into the application of the MPT. In total, three different electricity demand scenarios were optimised; the contemporary load profile (1), increased peak loads due to an extreme penetration of electric vehicles (EV) and electric heat pumps (HP) (2), and a flat load profile due to an extreme penetration of demand-responsive measures (3). The most efficient portfolio shares for each demand scenario were found by computing how 35GW should be distributed among solar PV, offshore wind, onshore wind and DTP in order to minimise the need for backup capacity and energy generation costs. The results of this study indicate that regardless of the demand profile, The Netherlands’ VRE system would be most cost-efficient in meeting demand when comprised of approximately 75% DTP and 25% offshore wind. However, only under the condition that the DTP-dams are geographically dispersed. Geographically dispersed DTP-dams, for example, located in Zeeland (south of The Netherlands) and off the coast of Texel (north of The Netherlands), cancels out the volatility in the electricity output caused by high and low tide. This reduces a VRE system’s volatility in electricity output and stabilises the electricity output from the dispatchable backup system, which minimises the system’s electricity generation costs. However, in terms of the amount of backup capacity required, it was found that a system comprised entirely of DTP with an integrated battery storage system would be even more efficient. The battery storage system eliminates all variability in the electricity output from DTP, which minimises the need for dispatchable backup capacity. However, as battery storage costs are significant (approx. 100,000 €/MWh), it is unlikely that a DTP-dam with a storage system large enough to flatten its electricity output is economically viable under the current market circumstances. This might change when DTP is combined with other storage systems or if battery storage costs reduce due to learning-effect, economies of scale or technical innovations. As DTP is still in its initial development phase, it is not certain DTP will successfully penetrate the power generation market. If DTP fails to enter the market, the current VRE system (61% solar PV, 8% offshore wind and 31% onshore wind) is fairly cost-efficient. However, if the aim is to reduce the amount of dispatchable backup capacity required to ensure energy security, future investments should be allocated to offshore wind. The overarching conclusion is that there is no unequivocal VRE portfolio that is most efficient to meet demand as it depends on the perspective taken (required backup capacity versus electricity generation costs). However, this study shows that the current VRE system benefits from the inclusion of DTP, both in terms of the required amount of backup capacity and the system’s energy generation costs. Only if DTP is combined with a large battery storage system to flatten its electricity output do the system’s energy generation costs surpass the costs of the current VRE system. In regards to the effect of including demand-variability in the MPT, it was found that demand variability has a limited effect on the selection of efficient VRE portfolio shares. Only in a scenario (2) with increased peak loads did the share of offshore wind slightly increase. This is due to the fact that the peaks in electricity supply from offshore wind coincide with the peaks in demand from electric HP (winter).

Climate change has increased the interest around energy intensity to many researchers. Understanding the drivers of energy intensity is essential to produce a fitting policy. While structural changes have been acknowledged as drivers of economy-wide energy intensity, it has not been recognised as an influential factor on industry-level energy intensity because there is simply not enough data to do so. To generate a more reliable understanding of the mechanism that drives industries’ energy intensity, I explore the role of international trade on inter-country differences in the energy intensity of industries. The approach in this thesis shows with multiple panel data regressions how changes in trade specialisation in products explain inter-country differences in energy intensity of industries. The results estimate that an industry’s energy intensity increases when it develops a comparative advantage in products whose production requires lots of energy, ceteris paribus, and decreases when it develops a comparative advantage in products whose production requires little energy. Since the effect of trade specialisation in one product goes on behalf of the effect of trade-specialisation in other products, there is a strong suspicion that there are changing compositions of industries. For policy changes, these results can provide insights into how a centralised approach in trade agreements and a decentralised approach in industry policies can tackle the industry’s energy-intensity problems. In future research, it might be possible to accurately predict the composition effects by explaining the changes in the energy intensity of products with trade specialisation in products.

Several studies have posed that international cooperative initiatives lead by businesses could aid in mitigating global warming in ways that are additional to national policies. However, evaluations on their progress are rare due to many informational issues. This research gap is worrisome since such initiatives can also have negative effects on global commitment due to effort fragmentation and greenwashing. This thesis evaluates progress in two initiatives that have featured prominently in the literature: the Science Based Targets initiative (SBTi) and RE100. The group of companies evaluated was reduced to the Fortune Global 500, a ranking of the largest firms by revenue. Large datasets were created with the targets set by these companies, their emissions and energy use, among others. Progress was evaluated in four aspects: the ambition of targets, the robustness of their disclosure of sustainability metrics, the implementation of changes to their energy profiles, and finally the substantive progress seen in their collective carbon footprint. It was found that among the two initiatives, SBTi appears posed for larger emissions mitigation, while RE100 is better at promoting more effective renewable energy purchasing practices. However, direct emission reductions are mostly concentrated on a reduced number of firms in energy intensive sectors, with the remaining members focusing on reductions through indirect methods that might not be additional at the global level. Similarly, it was found that informational barriers remain high even in these firms, with a plethora of inconsistencies that complicate year-by-year comparisons.

Climate change is a pressing issue and risk in our current society caused by greenhouse gas (GHG) emissions. To reduce the GHG-emissions more renewable energy is needed instead of electricity and energy production by fossil fuels, which are high in GHG-emissions. The electricity production by renewable sources has increased in the last couple of years in the Netherlands. However, renewable energy production via wind and solar energy is supply based, i.e. can only be supplied when wind and sun is available. Energy storage is needed to overcome the energy needs during low or no renewable energy in sustainable way. However, there barely seems to be any storage in the Netherlands. The market does currently not value all the services that storage can provide. Research into the value of storage is needed to understand the lack of energy storage and the current valuation of energy. Energy storage can have different levels and services that it can provide. In this thesis the focus will be on grid-scale energy storage. In the Netherlands there are no grid-scale energy storages. In other countries there are grid-scale energy storages, often in the form of Pumped Hydro Storage. In the Netherlands there is a plan, called Delta21, that wants to expand the Delta Works and improve water safety by using pumps. In the plan the pumps can also be used for a Pumped Hydro Storage to store energy. It is important to understand the value of storages like Delta21 in the Dutch electricity system.

The Netherlands is subject to labour scarcity across a variety of industries. While labour scarcity has occurred in the past, only now the consequences of higher educational attainment have become prevalent in the Netherlands. Due to a shortage of vocational workers and a population that is getting older, circumstances ask for adequate policy-making. Although policies like increasing the formal retirement age are put in place to negate labour scarcity, it remains difficult to understand the long-term outcomes of policy implications made in the present. Understanding this long-term behaviour of the dutch labour system is deemed useful to develop strategies that result in a well-functioning labour market. Since the labour market is intertwined with other systems (e.g. population, educational system and the economy) and is subject to uncertain future developments it results in increased complexity of understanding long-term dynamics. This research aims to develop a better understanding of the long-term behaviour of these systems in deeply uncertain futures.

Network measurments are mostly used to studynetwork topology, performance and security. The thesis uses network measurements in relation to large Internet corporations like Salesforce, Netflix and Snapchat. The thesis attempts to verify whether it is possible to predict company performance by monitoring network measurements. For the three corporations considered, the measurements provide insider information on the company, before it is made available to the public or released by the corporation itself. Network measurements can be used by managers effectively as a tool to understand and predict company perofrmance.

The world is experiencing a new technological revolution that can have profound implications for the nature of work. Currently, attention is particularly focused on the implication of the increased adoption of industrial robots. Multiple studies predict that especially developing countries are at risk to be negatively affected by robotization in the near future, in ways such as unemployment and greater income inequality. To a large extent, developing countries became integrated in the world economy through their involvement in global value chains (GVCs). The establishment of GVCs, in turn, were a direct result of an upward trend in offshoring. However, because both offshoring and industrial robots primarily affect routine and labor-intensive tasks, the labor-cost advantage of developing countries is quickly eroding and puts at risk their strong export position. Furthermore, in recent year world trade is indeed observed to slow down, which has further fueled concerns. This thesis aims to quantify how much the adoption of industrial robots in developed countries contributed to the decline in world trade experienced in recent years. We do this by regressing a measure of offshoring on the density of industrial robots and other control variables. Offshoring is quantified in terms of the offshoring index, which is constructed using two datasets by the Organization for Economic Co-operation and Development (OECD). To get data on the adoption of industrial robots, we use a dataset that has been made available by the International Federation of Robotics (IFR). The final unbalanced panel dataset consist of 29 countries, 15 industries and spans the period 1993-2015. The other factors that will be controlled for in the models include: labor intensity, wages, year-dummies, country-trends and industry-trends. Furthermore, we estimate the models using fixed-effects. Our regression estimates provide evidence for a negative and statistically significant relationship between the adoption of industrial robots and offshoring intensity. We estimate that if the density of industrial robots increases by 10% in OECD countries, then offshoring decreases by 0.29%. Furthermore, we investigate if the effects are particularly strong for certain industries. We find that the effect between robotization and offshoring particularly holds true for industries that have already robotized the most in relative terms. We conclude by discussing several policies that can help developing countries to tackle the disruptive effects of automation and safeguard their future economic development.

Our planet is witnessing an unprecedented increase in temperature that is causing frequent damages such as extreme weather conditions, acidification of oceans and extinction of species that may soon become irreparable. Various studies have reported that greenhouse gas emissions (GHG) from human activities in industries that produce 'material goods' like transportation, energy, manufacturing and agriculture have been the chief drivers of the climate change crisis. Even as greener and renewable energy technologies are increasingly being adopted, since consumption of material goods in the rich parts of the world continue to increase, it may not be enough to solve the environmental problems in time. However, we may be able to solve them by adopting a different way of living. In order to do so, we need to understand the purpose of the economy. According to Aristotle, Eudaimonia or the 'good life’ which is the full development of human capabilities or virtues is the ultimate end of human life and the purpose of the economy or the production and consumption of material goods by people are simply a means towards this ultimate end. A similar idea that human beings have an ultimate end towards which all actions are directed can also be found with the more popularly known Maslow’s hierarchy of needs. This essentially means that human beings need material wealth only insofar that it supports the fulfilment of their non-material needs: the development of capacities/virtues. 'Non-material' can refer to needs such as enjoying a classical dance performance, or studying philosophy or engaging in therapy. This perspective allows us to rethink growth in terms of a shift from material to non-material needs, which could also help solve the world's environmental problems as long as non-material needs can be met in less polluting ways (as compared to how material needs are met). Following this, economic activities are classified as physical (or goods-producing) or non-physical. Activities in the physical economy meet material needs and have a higher potential for productivity growth. Activities in the non-physical economy, including health care, education, research and arts, meet non-material needs and have a lower potential for productivity growth since human work is generally the final output. Based on this classification, it is observed that, in the advanced economies, the share of the physical economy in total demand is getting smaller, while the share of the non-physical economy is getting bigger over time, reflecting human development from material to non-material growth. If the non-physical economy − which consists mainly of human work such as a doctor's advice, a lesson, a concert − is relatively less polluting, such a shift could mean good news for the environment. While this looks promising, a consequence of the lower potential for productivity growth in the non-physical economy is that their activities can be expensive. Baumol (1993), termed this the 'cost disease' and argued that people in the rich parts of the world can afford the expensive 'personal services' when funds resulting from the productivity gains in the physical economy (due to labour-saving innovations) are transferred to the non-physical economy. Yet, nowadays, the standard response to the 'cost disease' is to replace human work with technology through standardisation, computerisation and robotisation of hitherto human tasks as a way to minimise costs. Due to the nature of work in the non-physical economy, the growing technology/ material-intensity can affect the quality of the non-material value created, while it may also increase the pressure on the environment. This led me to investigate the impact of the rising technology-intensity in the non-physical economy on the quality of outcome and on the environment. One of the main activities in the non-physical economy, health care, is 'an art and a science' in the sense that it takes care of the health of a human being in order to enable him to develop intellectually and psychologically and it is not just about 'fixing' what is broken. Doctors, therapists and nurses are people who possess besides clinical knowledge, interpersonal skills such as understanding and empathy to carry out complex human interactions with patients. Yet, in recent times, especially in the United States, technologies such as electronic health records (EHR), e-prescribing, tele-medicine and health apps are increasingly being adopted to displace some of the creative work performed by health care professionals. The main drivers of such trends identified are: the privatisation of health care, the systems of 'Managed Care' and government measures such as the American Recovery and Reinvestment Act (ARRA) of 2009 that allocated funds to incorporate health information technologies in health care facilities across the country. Some of the common experiences found with the adoption of technology in health care are: a supplier-induced demand for drugs and medical technologies due to a 'fee-for-service' payment model for doctors, a 'technology arms race' between hospitals and ‘direct-to-consumer’ advertising of medicines and medical technologies. Some of the consequences of these trends found are: unnecessary testing leading to increased overall health care costs, 'physician deskilling' due to decreased clinical knowledge and psychological and health implications for patients due to less physician-patient interactions and more 'end-of-pipe' solutions. From a brief review, it is also found that the health care system in Canada is less technology-/material-intensive than the health care system in the U.S. with (roughly) the same quality of service which shows that different kinds of health care systems can coexist with one another. A second activity of the non-physical economy, education, is one that enhances the intellectual and spiritual development of students, guided by a curriculum. According to various studies, teachers are indispensable in this activity in terms of their personal knowledge, their pedagogical capacities and their ability to instill curiosity, enthusiasm, sympathy, and morality in students through complex and dynamic interactions. Yet, similar to health care, even the creative work of teachers are being displaced with Information and Communication Technologies (ICT) such as: talent management software, digital learning technologies, MOOCs, allegedly to improve quality of education. In the United States, some of the trends commonly found to promote such technologies are: a pay-for-performance model for teachers to improve productivity of student grades, a common core standard curriculum and standardized tests for students to get them ready for college and careers. Some of the consequences of the growing technology intensity in education are found to be: teacher deskilling, lack of evidence of improved student performances and psychological implications for students such as diminished social skills, lack of creative and original thinking etc. These findings suggest that the growing technology-intensity in health care and education may have not significantly improved the quality of service provided, especially since the nature of work in these activities are different. Besides this, they could also lead to a higher environmental burden. In this thesis, I investigate whether the latter is the case, which is done in two steps. First, I investigate whether the non-physical economy is less polluting than the physical economy. Next, I examine the environmental consequences of the rising technology-intensity of the non-physical economy. The empirical research method applied is the Environmentally Extended Input-Output (EE-IO) analysis, which is used to compute the direct and indirect environmental effects of the physical and the non-physical economy. Direct effects refer to the pollution recorded at the place where it arises (for example, emissions due to health care). Indirect effects refer to the pollution caused by industries that supply inputs to health care, and for which the health care sector (as sector of use) is (indirectly) responsible. Through an EE-IO analysis, environmental effects (such as emissions) are ascribed to the sector of use, by reallocating emissions from the sector where they originate (sector of origin) to the sector that uses the goods or services supplied by this sector. The EE-IO analysis makes use of an Input-Output (I-O) table which records intermediate deliveries of goods and services between sectors in an economy, as well as each sector's deliveries to final demand. In addition, it uses the environmental accounts of each sector to compute the total emissions for which a sector can be held responsible, which consist of its own (direct) emissions plus the emissions that are related to the inputs it purchases from other industries. The two equations that guide this computation are B=b*(1-A)¯¹ and E=B*f where B= a vector of total (direct+indirect) emission intensities (per sector); b=direct emission intensity vector; (1-A)¯¹= the 'Leontief Inverse matrix' (derived from an I-O table) that represents the technical coefficients or the total (direct+indirect) inputs required per unit of a sector's output; f=final demand and E=a vector of total (direct + indirect) emissions per sector for a given level of final demand f. The results found were that the non-physical economy in the United States is less polluting in terms of global warming and acidification potential, energy use and tropospheric ozone gas emissions, even if emissions are re-allocated from sector of origin to sector of use. This result suggests that the impact on the environment will be lower if human beings become increasingly interested in non-material rather than material growth. However, this result may not come about if the non-physical economy gets more technology-intensive. To empirically investigate the increase in technology-intensity in health care and its corresponding environmental burden, two comparative studies based on the EE-IO analysis were carried out. The first one compared the health care system in the United States between 1995 and 2015. After adjusting both I-O tables for inflation (using Miller & Blair’s (2009) "double deflation" method) and regrouping them for comparable sector classifications, it is shown that the health care sector's technical (Leontief inverse) coefficients increased for 'computer, electronic & optical equipment', 'post & telecommunications', 'electricity & water supply', and 'renting & other business services' − indicating that the technology- or material-intensity (in terms of medical technology, ICT, administrative technology) as well as the electricity-intensity of health care increased from 1995 to 2015. Although the use of 'chemicals & pharmaceutical products' in the health care sector itself decreased, final demand for chemicals & pharmaceuticals per capita almost doubled between 1995 and 2015. In sum, the results found showed that technology/material-intensity in terms of ICT, medical technologies, energy use per unit of output and medicines per capita increased from 1995 to 2015. These data support, or at least do not falsify, the hypothesis of a trend towards higher technology-intensity of health care. However, the result for the environmental burden of the increased technology-intensity was found to be mixed. Firstly, total CO2 emission intensity of health care and the indirect contribution from 'post & telecommunications', 'computer, electronic & optical equipment' and 'renting & other business services' decreased from 1995 to 2015, largely due to an economy-wide switch to less carbon-intensive sources of energy. Although the shift from coal to natural gas reduced CO2 emissions, it also led to other types of environmental problems, such as those associated with increased production of shale gas. Secondly, the indirect emission contribution from 'electricity & water supply' industry to health care's carbon intensity increased due to the increase in energy use per unit of output (possibly due to increased use of medical equipment and other technologies in hospitals). This also led to higher total (direct+indirect) emissions from the health care sector in 2015 than in 1995 for the respective levels of final demand. Therefore, as long as the technology-intensity of health care keeps rising, this is likely to involve rising pressure on the environment (not only in terms of emissions but also in terms of increased use of earth's finite resources). The second study compared the health care system in the United States with the health care system in Canada for the year 2014. The results showed that the U.S. health care sector's technical coefficients were higher than those of Canada's health care for the relevant sectors examined in the previous study. This shows that U.S. health care is more technology-/ material-intensive in terms of medicines, medical technology, ICT and administrative technology than Canada's health care while reliance on human work is lower. Next, the total (direct+indirect) carbon-intensity of U.S. health care was also found to be higher than that of Canada's health care. This is mainly because the indirect emission contribution from sectors, namely, 'chemical & pharmaceuticals', 'renting & other business services', 'electricity & water supply' 'computer programming & information services' to health care are higher in the U.S.. However, the indirect emission contribution from sectors 'telecommunications' and 'computer, electronic & optical products' are found to be lower in the U.S. (in spite of a higher use of their goods per unit of output), possibly due to a lower carbon-intensity of these two sectors in the U.S. than in Canada. All in all, the important insight from these two studies is that even though increased purchase of medicines, ICT and related medical & administrative technologies by health care activities in the U.S. between 1995 and 2015 has not led to higher total CO2 emission intensity of health care, we know that as of 2014, high-tech health care in the U.S. is still more polluting than Canada's health care. This supports the argument that replacement of human work with technological solutions in the non-physical economy may lead to a higher environmental burden. In conclusion, since there is lack of conclusive evidence that the promotion of general, country-wide transition to high-tech health care (or education and arts) reduces costs while improving quality and reducing environmental burden, it may be desirable that free choice of technology in the non-physical economy is encouraged. The resulting diversity, or mix of low-tech and high-tech approaches in health care, education, arts, etc. would be less polluting than a linear high-tech approach in the non-physical economy, while leaving providers as well as recipients of services in the non-physical economy free to explore different technological paths.