The economics of the welfare state in today s world

Transcription

1 Int Tax Public Finance (2008) 15: 5 17 DOI /s x The economics of the welfare state in today s world A.B. Atkinson Published online: 24 October 2007 Springer Science+Business Media, LLC 2007 Abstract This article is concerned both with the substantive policy issue of the implications of the European welfare state in a global setting and with the way in which economists analyse the welfare state. Economics has made a major contribution to our understanding of the welfare state through the provision of formal models. These have allowed us to see the implications of social protection for countries increasingly open to international competition. These models, however, leave out essential elements, and the standard Heckscher Ohlin 2-good, 2-factor, 2-country assumptions impose too tight a straitjacket. We do not observe full factor price equalisation. The paper considers how we might relax this straitjacket to incorporate elements that are important in the public debate, while preserving tractability. The resulting model is used to investigate the impact of globalisation on the welfare state, contrasting Europe and the US. Keywords Welfare state Globalisation Employment JEL Classification H87 The first version of this paper was written while I was visiting the Economic Research Department of the Bank of Italy. I am most grateful to the Research Department for their hospitality, but make clear that the contents of the paper are solely my responsibility. The paper was presented at the April 2006 Netspar Conference in the Hague. I thank the discussants, André de Jong and Cees Oudshoorn, and the conference participants, for their helpful comments. The revision has benefited from the valuable suggestions of the referees and editors, and the final version has been greatly improved as a result of a conversation with Peter Neary. None of the above are to be held responsible for the remaining shortcomings of the paper. A.B. Atkinson ( ) Nuffield College, Oxford OX1 1NF, UK Tony.Atkinson@nuffield.ox.ac.uk

2 6 A.B. Atkinson 1 Introduction This article is concerned with the substantive policy issue of the impact of the European welfare state in a global setting and with the way in which economists analyse the welfare state. Both are important. Globalisation is widely seen as challenging Europe s social institutions. In redesigning their systems of social protection, the countries of Europe have to take account of increased international competition. But the underlying economic analysis is also crucial, since policy recommendations may follow as much from the way in which the problem is formulated as from the analysis itself. The welfare states of Europe face a number of challenges today. Here, the focus is on just one of these: that the European welfare state is under threat on account of increased international competition, acting together with skill-biased technological change to reduce the demand for low-skilled workers. A contrast is often drawn in this respect between Europe and the United States. The inflexibilities in Europe s labour market, it is argued, cause higher unemployment, and hence higher benefit costs and a smaller tax base. Economics has made a major contribution to our understanding of the welfare state (after neglecting it for much of the first 100 years of its existence) 1 through the provision of formal models. These have, in particular, allowed us to see the implications of social protection for countries increasingly open to international competition, as analysed in the models of international trade summarised in Sect. 2. The same models help us understand the impact of skill-biased technological change. However, the 2-good, 2-factor, 2-country Heckscher Ohlin model imposes too tight a straitjacket. Therefore, in Sect. 3, we consider how to relax this straitjacket and to incorporate elements that are important in the public debate. The resulting model is used in Sect. 4 to investigate the impact of globalisation on Europe and the US. The extended model highlights the role of demand for non-traded services in modifying the standard conclusions, the impact of the delocalisation of capital, and the extent to which investment in skills can help meet the challenge to the European welfare states. At the same time, there remain significant missing elements, as discussed in the concluding Sect Modelling the impact of the welfare state in a global context The impact of globalisation has featured prominently in the debate about the future of Europe s welfare states. Conversely, the extent of social protection has been taken as an important factor influencing how countries are affected by increased international competition. The common approach to modelling the interaction between globalisation and the welfare state is the Heckscher Ohlin theory of trade between countries that differ in their factor endowments but not in their technologies. The model has been used to considerable effect by Krugman (1994, 1995). He noted that what is 1 In the eight year period from 1995 to 2002, the American Economic Review published no fewer than 33 articles making reference to the welfare state, whereas from 1945 to 1984 the average was one per year.

3 The economics of the welfare state in today s world 7 needed is a consistent picture of the interaction between labour market developments in the high-wage countries and the growth of exports from the low-wage countries (1995, page 344), and that this can be provided using standard trade theory. As he said, general equilibrium trade theory has produced textbook models whose time has come (1995, page 362). Krugman applied the Heckscher Ohlin model, with two factors (skilled and unskilled labour) and two goods (X and Y ), to examine the implications of increasing global trade. The growth of the Newly Industrialising Countries (NICs) is represented by assuming that the OECD countries face a shifting offer curve (showing the amount of imports (into OECD) of X offered in exchange for exports of Y ). Within the OECD, he contrasted a country with clearing competitive labour markets (the US) and a country with a wage floor governed by minimum wage legislation or unemployment benefit (continental Europe). The inverted commas are important, since these are stylisations. The opening up of trade with NICs shifts demand away from products that use unskilled labour intensively (from X to Y ). If the OECD is modelled in a US fashion, the relative wage of unskilled workers falls, causing increased wage dispersion. From the Stolper Samuelson theorem, we know that losses are borne by the factor that is used intensively in the sector facing increased competition. In contrast, if the OECD is modelled with European assumptions, then relative wages are fixed, and the result is unemployment for unskilled workers. The loss is again borne by the unskilled, but it takes a different form. In this way, we appear to have a unified explanation as to how a single cause increased trade with NICs has a differential impact on the US and on continental Europe. It is the welfare state and/or minimum wage legislation that are responsible for the difference. As Krugman describes the position, the taxes and regulations imposed by Europe s elaborate welfare states have made employers reluctant to create new jobs, while the relatively generous level of welfare benefits has made workers unwilling to accept the kind of low-wage jobs that help keep unemployment comparatively low in the US (1994, page 1). Krugman (1995) cautions that the quantitative impact may not in fact be large, but it is natural that many policymakers have drawn the conclusion that continental Europe can only emulate the US employment performance by cutting the wages of the unskilled and reducing the degree of social protection. A scaled-down welfare state and increased wage dispersion are the necessary price for reducing the unemployment induced by globalisation. 2.1 However, we do not have two parallel universes Since these conclusions have an obvious resonance with many policy-makers, it is important that the underlying economic model is soundly based. When, however, we look more closely, it turns out that the conclusions described above do not follow directly from the textbook model. As Davis (1998a, 1998b) observed, we cannot look at two parallel universes with two trading regions (in one case, US and NIC, and in the other, Europe and NIC). We need to analyse simultaneously a minimum of three trading regions (US, NIC and Europe). Europe and the US are not just two alternative modelling assumptions; they co-exist. If, in a three country analysis, the US and Europe both produce the goods that faces NIC competition, then with the standard

4 8 A.B. Atkinson assumptions, 2 the (real) wage floor in Europe determines the relative prices of both goods, and the minimum wage prevents the relative price from falling. The US is, therefore, unaffected by increased trade. Europe bears the brunt in terms of unemployment. As it was put by Davis, European unemployment props up US wages. On the other hand, Europe may be protected to a degree because it has already become specialised in goods that use skilled labour intensively. This possibility, clearly recognised by Krugman (1995), has different consequences in the three country case. The relative prices of the two goods are no longer fixed by the European welfare state. If the entry of the NICs drives down the price of the good that uses unskilled labour intensively, this causes the relative wage of unskilled workers to fall in the US, which produces both goods. We then have the US side of the conclusion, but not the European side. Either way, as shown in Atkinson (2000), one or another part of the unified explanation cannot apply. The impact of globalisation, and the interaction with the welfare state, cannot be adequately explained by simple application of the standard Heckscher Ohlin model. We need a richer model. The extension to a three country situation the US, Europe and the NIC is a first step in that direction, but we also need to consider the number of goods and the number of factors. In the next section, it is argued that a model where all goods (and services) are traded internationally gives too much weight to global elements. We have to introduce a non-traded good. Equally, we cannot consider the macro-economics of the welfare state without considering the role of capital. The next section completes, therefore, the transition from a model to a model. 3 Toward a richer model The analysis just presented makes clear the straitjacket imposed by the theoretical structure. In the non-specialised case, everything is driven by factor price equalisation. There is a one-to-one relation between the skilled/unskilled wage ratio and the relative prices of the two goods. If the ratio is assumed fixed by the institutions of the welfare state, then all else follows. As Neary has noted, the assumption of fixed relative wages in this highly simplified model imposes an implausible degree of structure on the world economy (2001, note 3). One way in which this can be relaxed, he suggests, is by introducing non-traded goods. One obvious candidate for such a non-trade sector is that of personal services. While there has been a considerable expansion of trade in services, with ICT having opened up to trade whole areas of activity such as banking and business services, and retailing, the same developments have also highlighted the limits to such outsourcing of personal services, and the fact that there remains an essential local ingredient (such as delivery). The second extension to the standard model proposed here is, therefore, 2 The standard assumptions are those sufficient to ensure factor price equalisation where all goods are produced by all countries (see, for example, Dixit and Norman 1980). Specifically, the following are sufficient: equal numbers of goods and factors, countries have identical constant returns to scale production functions, and there are no factor-intensity reversals.

5 The economics of the welfare state in today s world 9 to assume a third sector (Z) which produces services, using only unskilled labour, which are not traded and the prices of services can, therefore, differ from country to country. This extension of the model allows us to examine how differences in the structure of demand can affect the degree to which the loss to unskilled workers from increased global competition in manufacturing sector X is offset by increased service sector employment. 3.1 Intermediate goods We have, therefore, moved from 2 to 3 sectors, but we need to vary the assumptions made about the definition of the manufacturing sectors. It has long been recognised that it is unsatisfactory to model international trade purely in terms of final consumption goods, when much of world trade concerns intermediate goods. This seems particularly the case when considering the impact of globalisation, since it is argued that this is changing the degree of vertical integration. Sinn (2006), for example, has argued that Germany is becoming a bazaar economy, shifting from being a producer to a merchant. He suggests that Germany is increasingly specialising in activities that are close to customers, while out-sourcing an ever-larger share of its high valueadded manufacturing to low-wage countries (2006, page 1). At the same time, it should be noted that the activities close to customers may involve high levels of skill in design and marketing. Therefore, it is assumed that the good X is an intermediate good, produced by OECD countries but also imported from NICs. It is an input into the production of the final good, Y, which is both exported and consumed at home. 3.2 Bringing capital back into the story The distributional impact of globalisation has typically been treated in terms of its differential effect on skilled and unskilled workers, but we have also to consider the differential effect on labour and capital. For instance, a fall in the share of wages in national income reduces the tax base for a payroll funded welfare state. The same applies to technical change. The recent literature has emphasised the bias in favour of skilled workers, but earlier writing by economists was concerned with the bias in favour of labour or capital. Meade (1964), for example, argued that automation involved technical progress with a marked labour-saving bias, and that this would change the distribution between wages and profits. In view of this, it seems essential to allow for three factors of production skilled labour (total supply fixed at L s ), unskilled labour (total supply fixed at L u ) and capital (total supply K). The distinction between skilled and unskilled labour is necessary to capture the differential effect of globalisation on different workers. But we cannot ignore capital and the changes that have taken place over time in real rates of return to capital. The introduction of capital allows us to consider another aspect of globalisation. In the case of capital, the Heckscher Ohlin assumption of complete factor immobility is not appropriate. Delocalisation is not just about unskilled workers losing their jobs, but also about the movement of capital abroad. In moving to a 3-factor model, I assume that at any time, the capital stock is fixed, but we can see the implications of different levels of K.

6 10 A.B. Atkinson 3.3 Factor intensities In the richer model considered here, we have three countries (the US, Europe and the NICs); we have three sectors: intermediate manufactures (X), final manufactures (Y ), and services (Z); we have three factors of production: skilled labour (L s ), unskilled labour (L u ) and capital (K). In full generality, this is quite complex. Therefore, we shall take a special version, making strong simplifying assumptions with regard to the pattern of factor use, while recognising that the assumptions are far from innocuous. It is assumed that the X sector uses unskilled labour in quantity L ux, and specific capital K X to produce a quantity X; that the Y sector uses skilled labour in quantity L s, and a quantity (X + M) of the intermediate good, where M denote imports, to produce a quantity Y of the final output, and that the service Z sector only uses unskilled labour, L uz, to produce a quantity Z. Capital is, therefore, a specific factor in sector X, and skilled labour a specific factor in sector Y, while unskilled labour can move freely between the X and Z sectors. 3 The production functions are assumed to be constant returns to scale Cobb Douglas for X and Y, and proportional for Z: X = AK 1 β X Lβ ux, (1a) Y = BL γ s (X + M) 1 γ, (1b) Z = L uz, (1c) where β and γ are parameters that lie between 0 and 1, and A and B are constants representing the level of technology. In effect, this is a specific factor model, in that capital is specific to the X sector, 4 and skilled labour to the Y sector, with unskilled labour mobile between the X and Z sectors. These assumptions drive a number of the conclusions, and their appropriateness can be debated (for example, that skilled labour should not be treated as a specific factor). At the same time, they are chosen to capture certain features: for example, the assumption that X is capital-intensive corresponds to the conception of X as physical production and of Y as design and marketing of the final product. 3.4 Technical change Skill-biased technical change has typically been examined in the context of an aggregate production function. The multi-sector model proposed here allows a richer treatment. The constants A and B in (1) allow for the introduction of technical progress that is specific to the X and Y sectors. Sector-specific technical progress may be a more relevant than the more commonly assumed factor-specific technological change (see Haskel and Slaughter 2002). (With the Cobb Douglas assumptions, the technical progress cannot be attributed to either factor within the sectors.) 3 The three sector model can be specified in many different ways. For example, the model of Corden and Neary (1982) provides an interesting contrast. They have two traded goods and non-traded services, but allow capital mobility, although they do not have intermediate goods, and do not distinguish between skilled and unskilled workers. 4 It would be straightforward to introduce capital as a specific factor in the Y sector as well.

7 The economics of the welfare state in today s world 11 From the cost functions, we can see that where all three goods are produced (there is no specialisation) p X = r 1 β wu β /A, (2a) p Y = ws γ p 1 γ X /B, (2b) p z = w u, (2c) where r is the rate of return to capital in the X sector, w s is the skilled wage rate, and w u the unskilled wage rate. Technical progress increases A and B, which are proportional to A and B. If the prices of the two traded goods are determined by world trade, the three pricing equations contain four unknowns (p Z, r, w s and w u ), so that there is a degree of freedom allowing the factor prices to take different values in different countries. Factor price equalisation is not imposed. In particular, if the welfare state in Europe sets a floor to the unskilled wage rate, this does not determine the relative skilled/unskilled wage rates in the US. The domestic demand for goods arises from household consumption determined by maximising a Cobb Douglas utility function such that a proportion (1 μ) of total income, R, is spent on the good Y and a proportion μ on services. The resulting consumer price index is given by π, where π = p 1 μ Y p μ Z. The wage rates shown above are gross wage rates, from which tax is deducted at rate τ to finance social protection in a revenue neutral manner. If there are H recipients of transfers who are outside the labour market (for example, pensioners), and U unemployed, and the level of social protection benefits is set in real terms at bπ, then the government budget constraint is 3.5 Determination of equilibrium τ [ L s w s + (L ux + L uz )w u ] = bπ[h + U]. (3) The three countries (the US, Europe and the NIC) are linked by world trade, facing the same relative prices for the traded goods, represented here by the terms of trade: ρ p Y /p X. The assumptions described above are assumed to apply to both the US and to Europe, although they differ in the parameters, in factor endowments, and in levels of technology. The assumptions ensure that in equilibrium all three goods are produced. In the Y sector, the pricing equation implies that the skilled wage rate is determined by the world prices of the two traded goods. From (2b), the product wage may be written as w s /p Y = (B ) 1/γ ρ (1 γ)/γ. (4) Equilibrium in the skilled labour market means that L s is equal to the labour supply, l s, assumed fixed. From the profit-maximising input condition, we can determine the demand for intermediate input (X + M), and hence the supply of Y, as increasing functions of the terms of trade, ρ.inthex sector, the fixed capital stock implies that the demand for unskilled labour is a declining function of the product wage, w u /p X, and a corresponding supply function for X: L ux = K X (w u /Aβp X ) 1/(1 β) (5a)

8 12 A.B. Atkinson and X = A 1/(1 β) K X (w u /βp X ) β/(1 β). (5b) In what follows, we will work with the ratio of the unskilled wage to the consumer price index, ω w u /π, referred to as the real unskilled wage. From the expression for the consumer price index, π, and the fact that p Z = w u, it may be seen that w u /p Y is equal to ω 1/(1 μ). The ratio of the skilled to unskilled wage rates is: w s /w u = (B ρ) 1/γ ( ρω 1/(1 μ)) 1. (6) The wage differential is a declining function of the real unskilled wage, but also depends on ρ: it is increasing with improvements in the terms of trade. Where w u is unconstrained, it adjusts to clear the market for unskilled labour, assumed to be in fixed supply, l u. Since the demand for unskilled labour in the service sector is μr/w u, the labour market clearing equation is Total income is given by l u = L ux + μr/w u. (7) R = w s l s + rk X + w u {L ux + L uz }. (8) Or, subtracting the value of service production from both sides (1 μ)r = w s l s + p X X. (9) Using (4) and (5), and substituting for R from (9), we can write the equilibrium condition (7) of the model in terms of the real unskilled wage: ( l u = K X ρω 1/(1 μ) /Aβ ) 1/(1 β)[ { }] 1 + μ/ β(1 μ) + μ/(1 μ)l s (B ρ) 1/γ ( ρω 1/(1 μ)) 1. (10) The right-hand side is a declining function of the real unskilled wage, and can take on any positive value, so that a unique equilibrium exists, determining the real unskilled wage, and hence the wage differential (from (6)). (The value of r/p X then follows from (2a).) Now the model is used to analyse the effect of globalisation on first the flexible US labour market and then the European welfare state with a floor to the wages of unskilled workers. 4 The impact of globalisation: US and Europe contrasted The effect of globalisation on the economy described above, with flexible wages (i.e. the US), may be seen from the impact of a change in the terms of trade, ρ. Suppose that the NICs are able to supply the good X at a lower price. If there were no service sector (μ = 0), the wage of unskilled workers would be fully determined by the prices

9 The economics of the welfare state in today s world 13 of traded goods. From (5), we can see that full employment could only be maintained if the real unskilled wage (which is then equal to w u /p Y ) falls by the same percentage as the terms of trade have improved. The product wage in the sector affected by NIC competition has to stay the same, so that w u falls in line with p X (ρω is constant). From (6), we can see that the ratio of skilled to unskilled wages rises. We have the standard conclusion. The same applies to the effects of technical progress in the sector employing skilled labour. An increase in B leaves (10) unchanged if μ = 0, and hence has no effect on unskilled wages, but raises the skilled wage (from (6)). Again, we have the standard conclusion. Technical progress favouring the sector using skilled workers increases the wage premium. The model here represents a departure in the introduction of the service sector. This modifies the term we have just been discussing and brings into play two further terms in (10). The modification to the first term involves ρω 1/(1 μ), in place of ρω,so that a given percentage improvement in the terms of trade requires a smaller adjustment in the real unskilled wage. If services absorbed half of total spending, the wage would fall as the square root of the terms of trade, so that a 21% improvement would require an 11% fall in real wages. Of the additional terms, the first is of the same form as before, but in the last ρ also appears positively in a way that tends to moderate the required reduction in the real unskilled wage. The effect is that ρ (1 μ) ω can now rise. The terms of trade improvement, and the decline in the unskilled wage rate, and hence p Z, both operate to increase the real income of the skilled workers, and hence increase their demand for services. This tends to offset the decline in demand for unskilled labour in manufacturing. From (10), we can see that the equilibrium real unskilled wage does not need to fall to the same extent as p X, and the size of the fall is smaller, the larger the propensity to consume services (even allowing for the appearance of μ/(1 μ) in the first term of (10)). We can also note that while technical progress in the sector using skilled labour continues to widen the wage ratio, it also raises the real unskilled wage as a result of the increased demand for services. Technical progress has effects on the demand side as well as the supply side, as in the case of a resource boom discussed by Corden and Neary (1982). 4.1 A European welfare state Now consider the European case where the level of social protection benefits, b,is set in real terms and this sets a lower floor to the net of tax wage of unskilled workers: w u (1 τ) bπ. (11) Or ω ( b/(1 τ) ). (12) If the constraint is binding, then (given the terms of trade) all factor prices are determined independently of the quantities. This means that we have to replace (10) by ( U = l u K X ρω 1/(1 μ) /Aβ ) 1/(1 β)[ { }] 1 + μ/ β(1 μ) μ/(1 μ)l s (B ρ) 1/γ ( ρω 1/(1 μ)) 1 (10EU)

10 14 A.B. Atkinson where U denotes the level of unemployment among the unskilled and ω is fixed by the constraint (12) holding with equality. If there were no service sector (μ = 0), then a rise in the terms of trade (cheaper imports of X) means that U rises (since ω is fixed). We have up to a point the standard conclusion. But in this richer model, the constraint (12) does not determine the wage differential independently of the terms of trade. From (6), an improvement in the terms of trade will widen the wage differential. The widening is less in Europe than in the US, since it rises with ρ (1/γ 1), rather than ρ 1/γ, but nonetheless accompanies the rise in unemployment. Again the impact of globalisation is moderated to the extent that μ>0, in that labour laid off in X is partly absorbed by the increased demand in the service sector. Cheaper imports of X do not mean that the real unskilled wage falls, but they have to reallocate from manufacturing to the service sector. The ease of such mobility affects the extent of unemployment. We may also note that technical progress in the production of X (via an increase in A) reduces the unemployment of the unskilled, the real wage (and the wage gap) being unaffected. Moreover, where μ>0, technical progress in the sector employing skilled workers (via B ) leads to lower unemployment. 4.2 Impact on the government budget constraint The government budget constraint is affected on both sides by the change in the terms of trade. The tax base (total labour income) is affected by the changes in factor prices, and on the expenditure side the outlay on benefits indexed to prices depends on the movement in the consumer price index, π. The constraint (3) may be re-written by dividing by π and setting L s = l s τ [ l s w s /w u + (l u U) ] ω = b[h + U]. (13) In the European case, both tax receipts and outlays are affected by the rise in unemployment. In the US case (with U = 0), only the left-hand side is affected. Tax receipts fall on account of the fall in ω but the rise in the wage differential operates in the opposite direction. A country with a sufficiently large proportion of skilled workers may see a rise in its tax receipts. 4.3 Conclusions from the richer model The first conclusion concerns the extent to which this fuller model can, unlike the version used by Krugman, produce the unified explanation. Increased international competition, in the form of cheaper imports of the intermediate good, can simultaneously cause widening wage dispersion in the US and increased unemployment of unskilled workers in Europe (where the constraint on the real unskilled wage is binding). The difference between the US and continental Europe in terms of employment performance may, therefore, be related to differences in the generosity of their welfare states. But it should be noted that the model also predicts a rise in wage dispersion in Europe. The rise in the wage differential is, other things equal, less than in the US, but we do not have exactly the Krugman conclusion.

11 The economics of the welfare state in today s world 15 The second conclusion is that there may be other mechanisms in operation. In particular, the extended model has served to highlight the role potentially played by differences in consumer demand. A larger demand in the US for personal services may have cushioned the impact of globalisation. This is not a new point. The work of Freeman and Schettkat (2001, 2005), and others, has drawn attention to the fact that a key difference between the US and Europe lies in the higher employment in the service sector in the US. Americans may spend increased real income on additional services, whereas Europeans prefer extra leisure (home production). Or, the relative rise in earnings at the top of the distribution in Anglo-Saxon countries may be fuelling demand for personal services (Manning and Goos 2003). 5 The difference in demand can be seen as a challenge to policy-makers. While this paper does not discuss the demographic dimension, one implication of an aging population is that there may be an increased demand for personal services. For this demand to be effective, however, pensioners have to possess the purchasing power. A scaling back of state pensions may, in this regard, reduce employment opportunities. On the other hand, the demand for services need not be purely private. Policy-makers in Europe have stressed the employment potential of publicly purchased services de proximité. A third conclusion relates to the delocalisation of capital. From (2a), the rate of return to capital specific to the X sector may be written [r/π] 1 β = (A /ρ)ω β μ/(1 μ). (14) It may be seen that an improvement in the terms of trade reduces the rate of return in the import competitive sector. This may lead over time to a decline in K X.From (10EU), we can see that a decline in K X leads directly to a rise in domestic unemployment, which where there is a floor to wages, can only be offset by an expansion of the service sector. If the domestic capital stock is being eroded, then this corresponds to the creation of the bazaar economy described by Sinn (2006). Moreover, the effect is also found in the US case. It is true that the magnitude is less if the unskilled real wage falls by a comparable amount, leaving the product wage (ρω) unchanged, but there is still a terms of trade effect. On the other hand, if the location of investment depends on the relative rates of profit in the US and Europe, and capital moves to the US, then this may be a different sense in which European economies are propping up US wages. A fourth conclusion is that, in the absence of factor price equalisation, the factor prices reflect factor endowments. In particular, the mix of skilled and unskilled labour affects the factor prices. From (10), we can see that, other things equal, a country with a higher ratio of skilled workers has a higher unskilled real wage, a smaller wage differential, and a lower rate of profitability in the X sector. From (13), such a country is less likely to see an erosion of its tax base as a result of globalisation. This is of policy relevance given the emphasis placed on investing in education and the acquisition of skills. Such a policy may indeed help protect the tax base of the European welfare state, although it may also accelerate the move to a bazaar economy. It also 5 The model described above assumes that demands are proportional to income, so that total demand is independent of the distribution of income, but it would be possible to allow for an income elasticity of demand for services in excess of 1, or for different demand functions for skilled and unskilled workers.

12 16 A.B. Atkinson has implications for labour migration, which is a potentially important aspect of the welfare state see Sinn (2005). A policy of encouraging inward migration of skilled workers will raise the unskilled real wage and reduce the wage differential. 5 Concluding comments It has been argued that the standard theory model cannot be applied without modification to the differential impact of globalisation on the US and Europe. Modification is necessary to incorporate elements that play an important role in the public debate, such as the employment potential of non-traded services or the delocalisation of manufacturing capital. I have suggested how we can move from a 2 2 2textbook Heckscher Ohlin model to a version that incorporates some of these features. Even with the aid of strong (probably over-strong) assumptions about the pattern of factor use, the resulting model is more complex. The model does, however, provides insights not available from the textbook model, such as highlighting the possibility that a differential impact of globalisation on the US and Europe may be due, not only to differences in the welfare state, but also to differences between the US and Europe in the demand for services. The most striking feature of the simple model used here is how much remains missing. We are still a long way from having a satisfactory general equilibrium framework to address the central issues. Here, we simply point to two missing ingredients. First, there is no proper treatment of the institutions of the welfare state. All that we have modelled is a benefit, available unconditionally to all those not in work, which enters the reservation wage of the unskilled. No account is taken of the features of real-world unemployment insurance and unemployment assistance that were designed to minimise the disincentive effects of the transfer (see Atkinson 1999). The second missing ingredient is any explanation of unemployment apart from the existence of unemployment benefit itself. It is politically attractive to say that unemployment benefit is the problem not the solution, but it is hard to believe that all unemployment would disappear if benefits were sufficiently reduced. The model as it stands does not allow for a positive probability of job loss, nor allow us to examine the possibility that globalisation may have increased the degree of job precariousness. These missing ingredients mean that we need to elaborate the model still further if we are to be able to examine the scope for reforming the welfare state to make it both more effective and more employment-friendly. References Atkinson, A. B. (1999). The economic consequences of rolling back the welfare state. Cambridge: MIT Press. Atkinson, A. B. (2000). Is rising inequality inevitable? A critique of the transatlantic consensus. WIDER Annual Lecture 3, WIDER, Helsinki. Corden, W. M., & Neary, J. P. (1982). Booming sector and de-industrialisation in a small open economt. Economic Journal, 92, Davis, D. R. (1998a). Does European unemployment prop up American wages? National labor markets and global trade. American Economic Review, 88,

13 The economics of the welfare state in today s world 17 Davis, D. R. (1998b). Technology, unemployment and relative wages in a global economy. European Economic Review, 42, Dixit, A. K., & Norman, V. (1980). Theory of international trade. Cambridge: Cambridge University Press. Freeman, R., & Schettkat, R. (2001). Marketisation of production and the US Europe employment gap. Oxford Bulletin of Economics and Statistics, 63, Freeman, R., & Schettkat, R. (2005). Marketisation of household production and the Europe US gap in work. Economic Policy, 20(41), Haskel, J., & Slaughter, M. E. (2002). Does the sector bias of skill-biased technical change explain changing wage inequality? European Economic Review, 40, Krugman, P. (1994). Past and prospective causes of high unemployment. In Reducing unemployment: current issues and policy options. Kansas City: Federal Reserve Bank of Kansas City. Krugman, P. (1995). Growing world trade: causes and consequences. Brookings Papers on Economic Activity, 1995(1), Manning, A., & Goos, M. (2003). Lousy and lovely jobs: the rising polarisation of work in Britain (CEP Discussion Paper 604). Meade, J. E. (1964). Efficiency, equality and the ownership of property. London: Allen and Unwin. Neary, J. P. (2001). Competition, trade and wages (Research Paper 2001/03). Globalisation and Labour Markets Programme, University of Nottingham. Sinn, H.-W. (2005). Migration and social replacement incomes. International Tax and Public Finance, 12, Sinn, H.-W. (2006). The pathological export boom and the bazaar effect: how to solve the German puzzle (CESifo Working Paper No. 1708). University of Munich.