What structural change is needed for a post­growth economy: a framework of analysis and empirical evidence

: 20 In order to avoid environmental catastrophe we need to move to a post-growth economy that can 21 deliver rapid reductions in environmental impacts and improve well-being, independent of GDP 22 growth. Such a move will entail considerable structural change in the economy, implying different 23 goals and strategies for different economic sectors. So far there are no systematic approaches for 24 identifying the desired shape of structural change and sectoral goals in terms of output, demand 25 and employment. We present a novel analysis that addresses this gap by classifying economic 26 sectors into groups with similar structural change goals. Our framework for the classification 27 considers sectoral characteristics along three dimensions, which are (a) the final energy intensity, 28 (b) the potential and desirability for labour productivity growth and (c) the relationship between 29 labour productivity and the energy-labour ratio. We present empirical evidence on the three 30 framework dimensions for economic sectors in the UK and Germany and derive structural change 31 goals for the four sector groups representing particular combinations of the sector characteristics. 32 Our analysis allows us to discuss the specific role of different economic sectors in the structural 33 change envisioned in the post-growth transition and the most important challenges they might be facing.

1 Introduction 5 (Jackson, 2015). Thirdly, increasing the share of sectors with low labour productivity growth can reduce 152 harmful economic growth in two ways. It reduces aggregate labour productivity growth (Jackson, 2017, p. 153 149; Jackson and Victor, 2011), and it redirects the expenditure of economic surplus towards activities that 154 do not increase productive capacity (Kallis, 2018, p. 123  The purpose of our framework is to translate the three overarching objectives into goals for specific sectors. 160 We specifically identify sector goals in four categories (Figure 1). The first category is the change in the 161 sector share in output and final demand, where the goal could be an increase or a decrease in the sector 162 share. The second category is the change in the sectoral employment share. The two categories effectively 163 break down the structural change in the economy into its sector-specific components. But it is difficult to 164 determine the goals in the in the first two categories without knowing the desired changes to sectoral labour 165 productivity and energy intensity. We therefore add a third category, which is the change in labour 166 productivity, and a fourth category, which is the change in energy intensity. Applying the framework produces 167 for each economic sector a set of goals, describing whether the sector share in output, final demand, and 168 employment is expected to increase or decrease and whether labour productivity and energy intensity are 169 expected to increase or decrease. 170 How the three overarching objectives are translated into sector-specific goals is determined by the inherent 171 characteristics of different sectors. For example the output share of energy-intensive steel production needs 172 to be reduced to achieve the overarching objective of reducing aggregate energy use. While the sectoral 173 goals might seem obvious for some sectors, difficulties arise where there are trade-offs between the different 174 goals. For example, taken on their own, energy intensity reductions are desirable in all sectors. But in some 175 sectors reductions in energy intensity might clash with the goal of labour productivity growth. To strike a 176 balance between comprehensiveness and ease of application we determine sector goals based on sector 6 characteristics in three dimensions. The three dimensions are the energy intensity, the potential and 178 desirability for future labour productivity growth and the relationship between labour productivity and the energy-labour ratio (Figure 1). 180 The three dimensions omit any assessment of the ability of a sector to contribute to the provision of basic 181 human needs and well-being. Arguably, such an ability is a key determinant for the sector goals in the 182 transition to a post-growth economy. We omit such a dimension in our framework because it cannot be 183 assessed based on economic statistics alone and requires democratic discussion. Our framework can 184 therefore give indications about the directions of the sector goals but not necessarily the desired magnitude 185 of change. For example we identify sectors in which final demand should be reduced. But in order to 186 determine by how much it should be reduced, further assessment of the contribution of such sectors to 187 human well-being is necessary. We define the direct energy intensity of a sector as the direct final energy consumption per Euro of gross 201 value added in constant prices, as is commonly done in the literature (Hammond and Norman, 2012). We 202 define the embodied energy intensity for each sector as the embodied final energy consumption per Euro of 203 final demand in constant prices. We restrict the analysis to the domestic components of final demand, as 204 price deflators for non-domestic components are not readily available. 205 For the purpose of allocating economic sectors into groups we distinguish two types of sectors, namely 206 sectors of high energy intensity and those of low energy intensity. We will refer to sectors of high energy 207 intensity as energy-intensive sectors and to sectors with low energy intensity as energy-light sectors. 208 2.2.2 Dimension 2: Potential and desirability of labour productivity growth 209 The second dimension describes the potential and desirability of labour productivity growth in a sector. We 210 include the dimension in the framework because it determines how changes in the employment and output 211 share of a sector can contribute to the overarching objectives of creating meaningful employment and 212 reducing aggregate labour productivity growth. By definition the dimension also determines how desirable 213 changes in the labour productivity are for a sector. The desirability of and potential for labour productivity 7 growth are two separate but related aspects and could therefore be treated as two separate dimensions. We Because the dimension combines two aspects, it can refer to different kinds of sectors. Some sectors have a 217 very low potential for increasing labour productivity. In such sectors the value of the output is mostly 218 determined by the labour time invested. Any reductions in labour inputs would reduce the quality of the 219 output (e.g. care services). While such sectors are often seen as desirable for creating employment in a 220 post-growth economy, they do not necessarily provide high-quality work (Druckman and Mair, 2019). In other 221 sectors it is possible to increase labour productivity but it might be undesirable. The post-growth literature 222 offers several potential reasons for the undesirability of labour productivity growth in a sector. Labour 223 productivity growth can eliminate meaningful jobs, for example if highly skilled craft work is replaced by 224 repetitive factory work Nørgård, 2013 To classify economic sectors in our framework we distinguish only two potential values for the dimension. 231 We refer to labour-light sectors as those sectors that have a high potential for labour productivity growth and 232 a high desirability to realise such potential. We refer to labour-intensive sectors as those sectors that have 233 either a low potential for labour productivity growth or a low desirability of labour productivity growth. 234 Assessing the dimension empirically presents a challenge. In our empirical analysis we assess only the 235 potential for labour productivity growth but not its desirability. To assess the desirability of labour productivity 236 growth we would need to analyse of the potential of a sector to provide meaningful work. While such an 237 analysis is possible, it lies beyond the time and resource constraints of our research project. The lack of such 238 an analysis is an important limitation of our study and of the wider post-growth literature. More empirical 239 research into the desirability of labour productivity growth in different sectors is vital for the development of a 240 post-growth strategy. 241 In order to assess the potential of future sectoral labour productivity growth we use the historic rates of 242 labour productivity growth as an indicator. Using historic rates has the advantage that they can be calculated 243 easily and consistently across economic sectors from existing data. But there are also large uncertainties in 244 how far historic rates of labour productivity growth will be similar to future rates. For example past labour 245 productivity growth might have exhausted the potential for further growth in some sectors, or the 246 development of new technologies might redistribute the potential for labour productivity growth between 247 sectors (Frey and Osborne, 2017). 248 We define the direct sectoral labour productivity as the sectoral GVA in constant prices divided by the hours 249 of direct labour inputs. We define the embodied labour productivity as the amount of final demand in 250 constant prices per embodied hour worked. We obtain the annual compound rate of growth in direct and 251 embodied labour productivity in each sector by fitting a log-linear regression model over the whole time 8 assigned based on absolute rates, with growth rates above 1% per year being considered high and growth 254 rates below 1% per year being considered low. 255 2.2.3 Dimension 3: Relationship between labour productivity and the energy-labour ratio 256 The third dimension in our framework describes the relationship between the growth in labour productivity 257 and the growth in the energy-labour ratio in different sectors. We include the dimension in the framework 258 because it is important for assessing the potential trade-offs between sector goals. The previous two 259 dimensions treat energy intensity and labour productivity separately and do not consider potential trade-offs 260 between different goals. To explore such trade-offs, it is useful to decompose the growth in energy intensity 261 into the growth of the energy-labour ratio and the growth of labour productivity (Semieniuk, 2015). The 262 relative size of the growth rates of these two variables then determines the change in energy intensity. The 263 energy-labour ratio describes the energy used per unit of work and can give an indication whether increases 264 in labour productivity have been achieved by increasing the energy utilised by workers. 265 Empirical evidence indicates that historical growth in aggregate labour productivity has been associated with 266 a growing energy-labour ratio (Kander et al., 2013;Semieniuk, 2015). On a sectoral level the evidence on 267 the relationship between labour productivity and the energy-labour ratio is limited. Two studies by Mulder 268 and de Groot (2004) and Witt and Gross (2019) suggest that there might be a correlation between growth in 269 labour productivity and in the energy-labour ratio in the manufacturing and transport sectors, but not in the 270 service sectors. 271 To classify economic sectors in our framework we divide sectors into two groups, depending on whether the 272 changes in the energy-labour ratio and labour productivity are positively or negatively correlated. It is also 273 possible that they are uncorrelated. The implications of 'no correlation' for the sector goals are either the 274 same as for 'positive correlation' or the same as for 'negative correlation', depending on the results for the 275 other framework dimensions. Sectors without a correlation will therefore be attributed to one of the two 276 groups accordingly. 277 For our empirical analysis, we calculate the rates of change in the direct and embodied energy-labour ratio. 278 The direct energy-labour ratio is the direct final energy consumption divided by the hours of direct labour 279 inputs. The embodied energy-labour ratio is the embodied final energy consumption divided by the embodied 280 amount of hours worked. We assess the relationship between the change in the energy-labour ratio and 281 labour productivity growth by examining the sign and relative magnitude of the average growth rates in both 282 variables. We do not perform a statistical assessment of the formal correlation between the variables in each 283 sector. Doing so would require the implementation of several statistical tests for each sector, the description 284 and discussion of which is beyond the scope of this article. In the future, however, it would be useful to 285 explore the relationship between the energy-labour ratio and labour productivity at a sectoral level using 286 more sophisticated econometric methods. represents a group of sectors with its own set of goals, derived from their specific characteristics. Table 1  293 provides an overview of those sectors goals for the different groups. To increase the clarity of presentation 294 and discussion, we group the eight possible combinations into four overarching groups based on the first two 295 dimensions. Each of these four groups has then 2 subgroups according to the characteristic in the third 296 dimension. 297 The goals outlined in Table 1 are derived purely from theoretical considerations. In summary, the need to 298 reduce the overall energy use in the economy suggests that output and energy use associated with sectors 299 of high energy intensity should be reduced relative to other sectors. In addition labour productivity growth in 300 the labour-light sectors should be supported. However, the share of the labour-light sectors in economic 301 output, demand and employment should be reduced relative to the labour-intensive sectors, in order to 302 reduce aggregate labour productivity growth and create meaningful employment. Energy intensity should 303 also be reduced throughout the economy, but there might be trade-offs with labour productivity goals, 304 depending on the relationship between labour productivity and the energy-labour ratio. The resolution of 305 these trade-offs depends on the other characteristics in each sector. 306 In practice, some combination of sector characteristics are likely to be more prevalent than others and some 307 might not exist at all. We therefore discuss the sector goals in more detail in section 4 in the context of our 308 empirical results. 309

Framework dimension (sector characteristics)
Sector goals We demonstrate the application of our framework by providing empirical estimates for the three framework 312 dimensions and for different economic sectors. We calculate sectoral values for final energy intensity, the 313 rate of change in labour productivity and the rate of change in the energy-labour ratio, both from a direct and 314 embodied perspective. Our empirical evidence covers sectors in the UK and Germany between 1995 to 315 2011. Group 1 includes sectors that are energy intensive and have a high potential and desirability for labour 343 productivity growth. In the empirical classification we present here, we only consider the potential of labour All of the Group 1 sectors have a direct and embodied energy intensity of more than 3 MJ/EUR, which 348 compares to direct and embodied energy intensities between 0.2 and 2.3 MJ/EUR for the sectors in Group 3 349 and Group 4 (Table 2 and 3). The only exceptions are the Machinery, Electrical, Equipment, Computers 350 sector and the Transport Equipment sector which values of direct energy intensity that are lower than 3 351 MJ/EUR and in the same range as the service sectors. While these sectors could therefore be considered 352 energy-light, they feature considerably higher values for embodied energy intensity, indicating that they rely 353 on energy-intensive inputs in the supply chain. 354 Group 1 sectors also exceed 2% annual growth in direct labour productivity in both countries. The exception 355 is the Food, Beverages and Tobacco sector which only achieves such rates in the UK (Table 2 and  embodied labour productivity, however, is generally lower than growth in direct labour productivity in Group 1 360 sectors (Table 2 and 3). In the UK, embodied labour productivity growth in all Group 1 sectors still exceeds 361 1.5% per year. In contrast, embodied labour productivity growth in most of Germany's Group 1 sectors is well 362 below 1% or even negative. It seems that the growth in direct labour productivity in Germany's Group 1 363 sectors has been offset by lower labour productivity growth in other parts of the supply chain. Given the short 364 time frame of our analysis we cannot say whether the pattern of low embodied labour-productivity growth in 365 Germany's Group 1 sectors presents a long-term trend. 366 The third dimension of the framework asks whether labour productivity growth in Group 1 sectors has been 367 associated with growth in the energy-labour ratio. Most, but not all, Group 1 sectors show a positive growth 368 rate in the direct energy-labour ratio in combination with growth in direct labour productivity (Figure 2). 369 Generally, the growth rate of the direct energy-labour ratio is below the growth rate of direct labour 370 productivity, leading to a decline in direct energy intensity. In contrast, the embodied energy-labour ratio 371 declines in most Group 1 sectors in both countries, sometimes in combination with increasing embodied 372 labour productivity and sometimes in combination with decreasing embodied labour productivity. 373  In our empirical classification Group 2 includes sectors with high energy intensity but low rates of labour 386 productivity growth. We allocate the Mineral Products and Construction sectors to Group 2. 387

UK
The Mineral Products sector is the only sector that mostly fits these characteristics. Its direct and embodied 388 energy intensity exceeds 3.8 MJ/EUR in both countries (Table 2 and 3). It also features a declining direct 389 and embodied labour productivity in the UK and only a growth of 0.3% in embodied labour productivity in 390 Germany. Only the direct labour productivity growth in Germany defies the pattern with a 3.3% annual rate of 391 growth. 392 We also allocate the Construction sector to Group 2, because it shows low rates of direct and embodied 393 labour productivity growth. The Construction sector does not strictly fit the characteristics of Group 2 394 because its energy intensity is low, with values of 1 MJ/EUR or below for direct energy intensity and values 395 below 3 MJ/EUR for embodied energy intensity. We still consider it useful to allocate the sector to Group 2 396 because it shows large environmental impacts in other aspects, particularly a high material intensity 397 (Giesekam et al., 2014). 398 The relationship between labour productivity and the energy-labour ratio varies between the two sectors. In 399 the Construction sector, the two variables consistently change in the same direction, in both countries and 400 both from a direct and embodied perspective. For the Mineral Products sector the relationship between the 401 two variables shows a diverse pattern. The two variables both grow in Germany from a direct perspective 402 and both decline in the UK from an embodied perspective. The growth in the two variables show opposite 403 signs in Germany from an embodied perspective and in the UK from a direct perspective. to Group 4 because the high rates of labour productivity growth in the UK seem to indicate that the two 439 sectors have a potential for labour productivity growth, even if it was not realised in Germany. 440 The relationship between the energy-labour ratio and labour productivity for Group 4 sectors varies between 441 countries and between the direct and embodied perspective. In the UK the direct energy-labour ratio is 442 growing in combination with growing direct labour productivity in all Group 4 sectors (Figure 2a In Section 2 we identify theoretical structural change goals for different sector groups based on different 468 combinations of characteristics in our three framework dimensions (Table 1). In Section 3 we allocate real 469 sectors from the UK and Germany to the sector groups based on empirical data (Table 2 and Table 3 Strategies to achieve the goals of reducing energy intensity and increasing labour productivity are not unique 514 to the post-growth transition and are discussed extensively in the wider economics literature. We do not 515 discuss the literature here but we want to point out an important challenge that is unique to the post-growth 516 transition. In the post-growth transition labour productivity growth in Group 1 sectors is aimed to be achieved 517 while simultaneously reducing output and final demand in the sectors. In the mainstream economics 518 literature, labour productivity growth in a sector is considered to be pre-condition or even a driver of output 519 growth (Nordhaus, 2005). Kaldor's growth laws suggest that labour productivity growth in the manufacturing 520 sectors is not only an important driver of output growth in the manufacturing sectors themselves but also in 521 the wider economy (Marconi et al., 2016;Thirlwall, 1983). Achieving labour productivity growth in Group 1 522 sectors under the conditions of contracting output might therefore pose difficulties. Or, in reverse, the 523 achievement of labour productivity growth in Group 1 sectors might jeopardise the goal of reducing demand 524 and output in such sectors. 525 4.2 Group 2: Energy-intensive and labour-intensive sectors 526 In our empirical analysis we only allocate the Construction sector and the Mineral Products sector to Group 2 527 based on potential labour productivity growth. Group 2 also includes energy-intensive sectors where it might 528 be desirable to adopt more labour-intensive production methods to create meaningful jobs, for example by 529 moving to small-scale, artisanal methods Nørgård, 2013). In Australia, the rise of artisan 530 bakeries has already been recognised to lower labour productivity growth (Ferguson, 2015). We do not 531 identify such sectors in our empirical analysis, but developing strategies for increasing labour-intensive 532 production methods in sectors where it is desired will be important for the post-growth transition. 533 As Group 2 sectors are, or could be, labour-intensive, expansion of production and consumption in Group 2 534 sectors would contribute to the overarching objectives to create meaningful employment and reduce 535 aggregate labour productivity growth. Given the environmental emergencies that society is facing, however, 536 we would suggest that the overarching objective to reduce energy intensity and energy demand should take 537 priority. In that case, the most important goal for Group 2 sectors is to reduce their share in output and final 538 demand, similar to Group 1 sectors. 539 The sector goals for the remaining production in Group 2 sectors are not clear cut, because there are 540 potential trade-offs between different goals. On the one hand, it might be desirable to reduce labour 541 productivity in order to create meaningful jobs and reduce aggregate labour productivity growth. On the other 542 hand, such reductions in labour productivity could increase the energy intensity of production. If labour 543 productivity and the energy labour-ratio are negatively correlated, reductions in labour productivity increase 544 the energy-labour ratio and energy intensity. Even in the case of increasing energy intensity, Kallis (2018, p. 545 134) suggests that the adoption of more labour-intensive production methods could be worthwhile because 546 lower aggregate labour productivity restricts the overall scale of production and environmental impact. Still, 547 there might be better ways to provide meaningful employment and lower aggregate labour productivity 548 without increasing the energy intensity in already energy-intensive sectors. If the labour productivity and the 549 energy-labour ratio are positively correlated or unrelated, the trade-offs are much smaller, especially if the 550 balance between reductions in the energy-labour ratio and labour productivity still allows for reductions in 551 energy intensity. 552 While the transformation towards labour-intensive production methods is a common theme in the post-553 growth literature, the literature does not offer a detailed discussion of its implications. No systematic analysis 554 is provided that identifies in which sectors the adoption of more labour-intensive methods would be feasible 555 and desirable. There might be many sectors, such as steel production, in which small-scale, labour-intensive 556 production is not possible or desirable. In the few sectors for which the literature identifies labour-intensive 557 production methods as desirable, there is little analysis of the consequences of a large-scale uptake of such 558 methods. For example, post-growth economists propose small-scale, labour-intensive farming techniques, 559 such as organic and permaculture approaches, on the ground that they are efficient in terms of energy and 560 land use. But the literature offers hardly any scientific assessments of how a large-scale shift towards labour-561 intensive farming would impact yields, food availability and labour requirements (Infante Amate and production using simple technologies or 3D printing, based on designs developed in a global digital 565 commons. It is not clear whether such an approach would be more or less labour-intensive or energy-566 intensive than current industrial production. 567 Once it is clearer in which sectors more labour-intensive production methods are desired for a post-growth 568 economy, achieving the adoption of such methods will require the removal of important barriers. In our 569 current market system, businesses are continuously under pressure to reduce production costs, a key driver 570 of labour productivity growth (Jackson and Victor, 2011;Shaikh, 2016). Except in niche areas, labour-571 intensive, small-scale manufacturing businesses cannot compete against the low prices of goods mass-572 produced in energy-intensive factories and by cheap labour abroad. Ecological tax reform that shifts tax 573 burdens from labour to environmental impacts have been proposed to reduce the energy intensity relative to 574 labour intensity (Daly, 2008). But in a system where competition is based on costs and prices, labour-575 intensive production methods will always struggle, even if price incentives are somewhat shifted in their 576 favour. The adoption of labour-intensive methods in Group 2 sectors requires a system that puts greater 577 value on quality, durability and fair working conditions. Johanisova et al. (2013) propose that an increase in 578 the use of social enterprises, not-for-profit organisations and other "non-market capitals" can play an 579 important part in creating such a system. In order for such organisations to flourish, however, consumers 580 would also need to be willing to shift away from mass consumption to buy fewer, more expensive and high-581 quality products. 582 Finally, a special challenge for the post-growth economy is the different treatment of labour productivity 583 growth in different sectors. While labour productivity growth is desired in some sectors (Group 1) it is not 584 desired in others (Group 2). Policies will therefore have to be tailored to achieve opposite outcomes in 585 different parts of the economy. Administration, Health Care and Other Services. The five sectors are the same as the labour-intensive 589 services already identified in Hardt et al. (2020). We identified the five sectors based on historic rates of 590 labour productivity growth, but Group 3 would also include energy-light sectors where labour productivity 591 growth is possible but not desirable. 592 The sectors in Group 3 have a low energy intensity, so that employment-related goals can take priority over 593 energy-related goals. The most important goal for Group 3 sectors is therefore to increase their share in 594 hours, or to reduced labour productivity if output is measured as number of students taught. Overall it is likely 603 that the share of Group 3 in output and final demand will increase if the employment share increases, at 604 least in current prices. 605 While Group 3 sectors have a low energy intensity relative to Group 1 and Group 2 sectors, Group 3 sectors 606 still account for a non-negligible fraction of the direct energy use and energy footprint for the UK and 607 Germany (Figure 3). Any expansion in the employment, output and final demand share of Group 3 sectors 608 therefore needs to be combined with reductions in energy intensity. Reducing energy intensity could clash 609 with the goal to reduce labour productivity, if labour productivity and the energy-labour ratio are negatively 610 correlated (Table 1). Fortunately, our evidence suggests that Group 3 sectors have shown reductions in 611 energy intensity and the energy-labour ratio despite reductions in labour productivity (Figure 2). There is 612 therefore no evidence for trade-offs between the goals for Group 3. 613 We already discuss the challenges for achieving an expansion of Group 3 sectors in Hardt et al. (2020) and 614 will only provide a brief summary here. Firstly, Group 3 sectors feature low labour productivity growth and 615 therefore face increasing relative costs compared to sectors with high labour productivity growth (Baumol, 616 2012(Baumol, 616 , 1967Baumol et al., 1985). Such a cost disadvantage has already pushed several market services 617 that are important for a post-growth economy, such as repair services, into the margins of our economy. It is difficult to determine structural change goals for Group 4 sectors, because they cannot contribute 636 strongly to any of the overarching objectives. Labour productivity growth is possible and desired, indicating 637 that these sectors are not a potential source of meaningful employment. Energy intensity is low, so there is energy footprint in the UK and Germany (Figure 3). A post-growth perspective would therefore suggest that 641 output and final demand in Group 4 sectors should be reduced, unless such output and final demand is 642 necessary for meeting basic needs or increasing wellbeing. In effect the structural change goals for Group 4 643 sectors are therefore similar to those for Group 1: reduce final demand and output where possible, reduce 644 the energy intensity of the remaining production and increase labour productivity (Table 1). 645 Another reason why it is difficult to define structural change goals for Group 4 sectors, is the fact that Group 646 4 sectors largely provide intermediate inputs into other sectors rather than final demand. As Figure 3 shows, 647 the share of Group 4 sectors in value added is much larger than their share in final demand. The group's 648 share in final demand is also dominated by the Real Estate sector, which largely consists of real and imputed 649 rent payments (Table 4). Defining and achieving structural change goals for Group 4 sectors therefore 650 requires an analysis of how production is interconnected with other sector groups. 651

653
More than other groups in our framework, Group 4 sectors highlight the limitations of the national accounts 654 and of our framework that relies on national accounts data. For many sectors in Group 4 it is difficult to 655 measure final demand and value added in constant prices. As the services delivered are intangible and 656 heterogeneous, it is difficult to separate any price increases into quality improvements or inflation (Eurostat,  For example, the income of the finance sectors has only recently been included as a productive activity 667 contributing to GDP (Christophers, 2011). It is likely that a considerable part of the income obtained in the can be considered as economic rent payments. Such rent payments have important implications for 670 inequality in the post-growth transition (Stratford, 2020). A fully review of this issue is beyond the scope of 671 this paper, but it serves to highlight the difficulties of defining output, demand and value added in many 672 sectors in this group. 673 The difficulties of defining structural change goals for Group 4 sectors do not mean that the sectors are not 674 important for the post-growth transition. On the contrary, the sectors in this group are very much at the heart 675 of many important challenges that our society is facing. Such challenges include unaffordable land and 676 housing (Kenny, 2019), the impacts of financial speculation (Jackson, 2018), the gig economy facilitated by 677 technological platforms (De Stefano, 2016) or the power of communication companies to exploit personal 678 data and influence democratic processes (Hind, 2019). Group 4 sectors present a very diverse set of 679 challenges that will require specific strategies for reform. Such strategies will undoubtedly affect the output, 680 demand and employment of Group 4 sectors, but it might be less useful to define sector goals in such terms. 681

Conclusion 682
In order to avoid environmental catastrophe, the environmental impacts from economic production and 683 consumption in high-income countries have to be reduced rapidly. Given the close coupling of GDP and 684 environmental impacts, achieving the necessary reductions in high-income countries will likely lead to lower 685 GDP growth, or even reductions in GDP. In high-income countries, we therefore need to create a post-686 growth economy that can simultaneously increase human well-being and deliver rapid reductions in 687 environmental impacts, independent of whether GDP is growing or declining. 688 The transformation to a post-growth economy will require structural change in the sectoral composition of 689 output, final demand and employment as well as strategies tailored to specific sectors. There will be winners 690 and losers, sectors that will expand, and sectors that will contract. Politicians are often not explicit about the 691 necessity of such structural change. They are especially not willing to identify sectors that will lose out in the 692 transition to a sustainable economy. Sometimes not even in obvious cases, such as the oil and gas industry. 693 As post-growth economists, we need to start defining the necessary structural change in order to stimulate a 694 discussion about which sectors need to expand and which sectors need to contract. Providing such a 695 definition is crucial for moving discussions beyond the abstract question of whether reductions in aggregate 696 GDP are desirable and feasible. 697 Our analysis starts to systematically define the structural change necessary for the transition to a post-698 growth economy. The framework and evidence presented allows for a consistent vision of structural change 699 to take shape. The production and consumption of energy-intensive goods will be reduced as much as 700 possible. Small-scale, labour-intensive production should be encouraged where feasible, while industrial, 701 efficiency-focused production will only be pursued where it makes sense from a social and environmental 702 perspective. Potential losses of employment will be offset by increasing employment in labour-intensive 703 services with high social value, while making sure that the new and existing jobs in these sectors are of high level. Especially information on the potential and desirability of labour productivity growth in different sectors, 708 and its implications for energy use, is currently lacking in the post-growth literature. More information on such 709 sector characteristics is needed to inform discussions on important normative questions: What production is 710 necessary and desirable? Where could production and consumption be reduced? Where exactly would 711 reductions in labour productivity be desirable and where is further pursuit of labour productivity sensible? 712 These questions tie into current debates about the future of automation. Research from a post-growth 713 perspective can offer something to such debates by investigating the desirability of automation and by 714 putting automation into the context of environmental challenges. 715 Even if we cannot determine all sector characteristics and structural change goals with certainty yet, the 716 preliminary outline we present already highlights some important challenges for achieving the necessary 717 structural change. The production, employment and consumption of different sectors is not distributed 718 equally across countries and across income groups. Strategies for achieving structural change need to be 719 just and equitable. Some of the sector goals we identify also go against the grain of our current economic 720 system. Business is currently dominated by pressures to reduce costs and grow markets and output. Many 721 of our sector goals would require resistance to such pressures. Achieving the goals might entail increasing 722 costs, reductions in output and the shrinking of markets and supply chains. Can markets be reformed so that 723 they support achieving such objectives? If yes, how? Do we need to find alternative ways of providing some 724 goods and services? The answer to the last question is almost certainly yes. The post-growth literature has 725 already started to develop alternative approaches but more needs to be done. For example Raworth (2017) 726 distinguishes between four domains of provisioning, the market, government, commons and the household. 727 Such a perspective could be linked with our framework to determine which sectors might be best suited to 728 which of the four domains. 729 6 Acknowledgements Industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of