ࡱ> qsnopq Ibjbjt+t+ FAAr:3 ]Xj@$P<Zl`r((%''''''$2KK:r:::%%::f n%>aPreliminary Draft The Economic Implications of Liberalising Mode 4 Trade by L Alan Winters School of Social Sciences University of Sussex Falmer BRIGHTON BN1 9SN UK Telephone: +44 (0) 1273 877273 Fax: +44 (0) 1273 673563 e-mail  HYPERLINK mailto:l.a.winters@sussex.ac.uk l.a.winters@sussex.ac.uk Centre for Economic Policy Research, London and Centre for Economic Performance, London School of Economics, London. 8th April 2002 This paper has been prepared for the Joint ϲʹ-World Bank Symposium on THE MOVEMENT OF NATURAL PERSONS (MODE 4) UNDER THE GATS, ϲʹ, Geneva, 11-12 April 2002 This paper draws very heavily on work financed by and shortly to be published by the Commonwealth Secretariat Winters et al (2002). I am grateful to my co-authors on that study, Terrie Walmsley, Zhen Kun Wang and Roman Grynberg for inputs and discussions on Mode 4, and to the last for initiating the study. I am also grateful to Antonia Carzaniga and J Michael Finger for comments on Mode 4, to Aaditya Mattoo for comments and advice on this paper and to Angelica Mayorga for logistical help. This paper sets out the economic case for liberalising the temporary flow of labour between countries for the purpose of providing services. This is the subject of negotiations currently underway about Mode 4 of the GATS. Despite being until now a mere bit-player in the GATS drama, Mode 4 is at last starting to command some attention from negotiators and policy makers. This paper, and some more detailed companion pieces, argue that this is long overdue and that serious efforts to liberalise the temporary movement of natural persons (TMNP) from developing to developed member countries could generate very large mutual benefits. The very heart of international trade, be it in goods or in factors, lies in exploiting differences. The larger the differences, the larger the potential gains from opening up international trade. In the case of TMNP, potentially large returns would be feasible if medium and less skilled workers, which are relatively abundant in developing countries, were allowed to move and provide their services in developed countries. The review of existing empirical studies of factor mobility and the new estimates described in this paper agree that there are huge returns to even relatively small movements of labour. An increase in developed countries quotas on the inward movements of both skilled and unskilled temporary workers equivalent to 3% of their workforces would generate an estimated increase in world welfare of over $US150 billion p.a. These gains are widely shared within the world economy. Moreover, as their populations ageing and their average levels of training and education rise, developed countries will face an increasing scarcity of less skilled labour. Given that, at least in some occupations, there is really no substitute for human labour, the demand for and benefits of allowing TMNP will increase through time. Thus while recognising the formidable political challenges it poses, TMNP actually offers a strong commonality of interest between developing and developed countries. Unlike with the mass migration of less skilled workers, fears for cultural identity, problems of assimilation and the drain on the public purse are hardly relevant to TMNP. The biggest concern it raises is its competitive challenge to local less skilled workers. This is neither more nor less than the challenge posed to such workers by imports of labour intensive goods from developing countries, which has been overcome by the weight of economic gain that trade could deliver and by policies to ease adjustment among local less skilled workers in developed countries. Applied with the same sensitivity and the same sorts of policies as trade policy reform in goods has received in the past, the temporary movement of less skilled workers between countries would offer the chance to reap some very large gains from trade. The paper comprises three parts. First, there is a discussion of ways in which we might think and model the liberalisation of Mode 4. This is based on two polar forms treating it as perfectly akin to goods trade and treating it as perfectly akin to labour migration. It then discusses ways in which these polar forms may be relaxed in future empirical exercises to try to estimate the effects of liberalisation. Part 2 summarises a new estimation exercise on the benefits of Mode 4 liberalisation treating it as akin to migration. This is argued to be a reasonable assumption in the context of the sort of models that economists have to use for this sort of exercise, and it suggests the very large economic benefits already alluded to. More details of the estimation exercise are available in the original sources. Finally, part 3 considers briefly the arguments for and technicalities of compensating domestic workers who are disadvantaged by inflows of workers from abroad. As noted above these are mainly the unskilled, who have proved adept at resisting goods market liberalisation, and I argue that Mode 4 will need to treat them with just as much consideration as did goods market liberalisations. 1. The Economic Case for Labour Mobility To date, the Temporary Movement of Natural Persons (TMNP) has defeated attempts to fit it into a robust analytical model. At one extreme it can be viewed as no different from cross-border services trade (Mode 1), which, in turn, is often argued to be analytically no different from ordinary goods trade. For example, an academic traveling to Geneva to deliver a paper in person is analytically close to that paper being sent in hard-copy form, electronic form or even delivered by video-link. Hence one part of the Mode 4 story is the trade story with which we are perfectly familiar. At the other extreme, Mode 4 has much in common with regular migration, whereby workers actually relocate from one country to another. This is particularly true where periods of stay are long or where a particular job in country B is filled by a continuous flow of temporary workers from country A, each being replaced by another as his contract expires. While such a revolving door provision has different implications for social integration, network formation and the inter-generational spill-overs from education, the basic fact that B gains a worker and A loses one is akin to migration. This model could be particularly relevant to agency-provided flows of middle-level professional workers such as nurses and teachers. Hence a second strand of thought about Mode 4 is based on the economics of factor mobility, and since this is less familiar to trade negotiations than the trade literature, I devote some space to it here. Neither of the polar models - trade or migration - seems to capture the full essence of mode 4 mobility, however, so I also devote a little of the paper to extensions and refinements to those models to see if we can devise a more satisfying analysis. Since ultimately the questions at stake are empirical, this last sub-section focuses on how we might estimate the benefits of Mode 4 liberalisation in practice. My musings on this subject are very preliminary, and improving them seems to me to be a very high priority for further research. International Trade At its simplest, trade in services is no different from trade in goods, for which there is now widespread acceptance of the benefits of a relatively liberal trading regime - for example, reaping economies of scale, the benefits of specialisation according to comparative advantage, learning by doing and developing expertise by concentrating on particular sectors, importing better technologies, and stronger competition. All of these apply equally to services as well as goods. Indeed, there are good reasons to expect greater gains from trade liberalisation in the services sectors than in goods: barriers are generally greater in services than in goods (although, of course, there are exceptions, e.g. agriculture), many barriers explicitly and dramatically reduce competition in the service sector, which can be most costly in efficiency terms, many services are necessary to efficiency and competition in other parts of the economy e.g. communications and transportation, or banking. Liberalisation in these sectors can have broad and deep, but essentially indirect, effects. For example, improved services can create completely new markets for other goods, which, as Romer (1994) shows, can induce dramatic welfare improvements: improved transport and communications can allow peripheral farmers to sell in the cities or to obtain previously unavailable credit which could dramatically increase their output, and services as a whole account for a greater share of income (and, frequently, employment) than industry and agriculture together. Thus, for example, Hertel et al (1999) suggest that, while 40% liberalisations in agriculture and manufacturing will each raise global welfare by about $70 billion p.a., a similar liberalisation in services could contribute over $300 billion. One should not take these modelling estimates too literally and TMNP is only part of service delivery, but the orders of magnitude are striking and even a small share of so large a benefit renders Mode 4 significant. One might think that the fact that TMNP currently accounts for only 1.4% of the value of services trade (Karsenty, 2000) shows that it has little promise for large gains. In fact, however, I believe that it shows the very opposite: the low figure arises from the very high barriers to TMNP, and so offers the greatest potentialreturns to liberalisation. None of this argues for wholly unregulated international trade in services. Governments will always have a fiduciary role in regulating many services, to counter the problems that arise from market failures such as moral hazard or asymmetric information. Rather, services trade liberalisation - including that of Mode 4 - calls for ensuring (Mattoo, 2000) that such regulations are geared to solving market failures rather than to protection, that they do so in trade efficient ways, and that, above all, they enhance rather than curtail competition. The parallel with goods trade liberalisation does, however, call for a very close look at arguments that services require generalised infant industry protection or that they have national security dimensions. These seem more like the traditional objections to trade liberalisation based on special interests, than analytical solutions to specific problems. It is worth reminding ourselves that although goods trade liberalisation is widely accepted as one of the key components of the policy cocktail required for growth and efficiency, it is not without its challenges. Trade reform is strongly redistributional, both between producers, governments and consumers, and within those groups. While widespread reform seems likely to benefit nearly everyone eventually (what Max Cordon, 1984, called the Hicksian optimism, there are likely to be short-term hardships and we can not rule out there being long-term casualties. There is a substantial political economy literature on the way in which these redistributions affect the prospects of reform e.g., Rodrik (1995) and a further literature discussing the need for, and design of, complementary and compensatory policies to counteract their adverse effects see, for example, DFID (2000) and Sapir (2001). McCulloch, Winters and Cirera (2001) offer a detailed discussion of the way in which trade liberalisation might affect poverty while Winters (forthcoming) considers the Doha Development Agenda and poverty. All of these issues are likely to be as relevant to Mode 4 as to goods market liberalisation and we should consider the lessons learned from the latter. Trade and Factor Rewards (Wages) The fundamental premise of the neo-classical theory of international trade is that the incentive to trade arises from differences in countries relative costs of producing different goods. These, in turn, arise from differences in the countries endowments of various factors of production. These endowments are assumed to be immobile between countries but mobile between sectors within any country. In its purest form, the theory generates the remarkable prediction that free (and costless) trade in goods between countries whose endowments are, in a technical sense, not too different would be sufficient to ensure that their factor prices are equalised the so-called Factor Price Equalisation (FPE) Theorem of Nobel Laureate Paul Samuelson (1949). If this were true, trade in goods and the movement of factors of production would be substitutes: as trade was freed, the incentives for labour migration and capital movement would decrease, ultimately to zero. Intuitively, one can think of goods as bundles of their constituent factors: then trade in goods and the migration of factors are two means to the same end. More technically, the result arises because under suitable assumptions factor prices are uniquely determined by goods prices; free costless trade equalises goods prices and, through that mechanism, factor prices. The basic idea of FPE is a powerful motivator of trade policies in the real world. For example, NAFTA, the EU-Mediterranean and the EU-Central Europe Agreements were all promoted partly as solutions to migration pressures. However, no-one (not even Samuelson) takes the prospect of complete FPE seriously the casual evidence against it is just overwhelming even in the absence of barriers to mobility such as within the European Union. Moreover, once we move beyond the strictly neo-classical theory, to allow trade to be determined by things such as technology differences or tax structures, trade and migration become complements rather than substitutes, and wage differences can persist indefinitely in the absence of factor movements (Markusen, 1983). Among the reasons that FPE may not obtain in reality is that many barriers to trade persist, some of a policy nature, but many that are natural, such as cultural and physical distance and geography. In addition, on a practical level, the productivity of factors varies systematically between countries possibly because of unobserved differences in their quality or in the set of complementary inputs that is available; economies of scale may allow larger economies to pay higher wages, and the taxes that influence rewards differ by country. At a more rarified and theoretical level, FPE may fail because there are more factors than goods (this is particularly likely if some factors are specific to particular industries, i.e. not inter-sectorally mobile, which is true of almost all in the short-run); because not all countries produce all goods (so-called complete specialisation), or because technology is such that the same goods prices are consistent with different factor prices (the case of so-called factor intensity reversals). Finally, Rausch (1999) argues that trade and migration are complementary because expatriot communities are instrumental in creating trade links, a view validated empirically by Dunlevy and Hutchison (2000). All of this analysis suggests that international factor mobility will remain an important feature of the world economy even as trade barriers decline. Indeed Wong (1986), for example, shows that, whatever the reason, trade and migration are positively associated under most sets of realistic circumstances in the USA. Williamson (1998) suggests that migration was a much greater source of economic integration in the nineteenth century than was international trade, accounting perhaps for 70% of the observed convergence of real wage-rental ratios. Of course even with perfectly unfettered migration the nineteenth century situation one would not expect the complete equality of real wages. There exist many non-policy discouragements to labour mobility including genuine preferences for home, the costs of establishing new social or production networks, and the fixed costs of migration. The recent developments in the theory of economic geography have explored some of these see for example, Fujita, Krugman and Venables (1999), and the discussion of the brain drain in Commander, Kangasniemi and Winters (2002). Factor Mobility This brings us to the second analytical approach to TMNP to treat it as parallel to migration. It is important to stress that TMNP is NOT international migration. It has none of cultural, social or political dimensions that are associated with international migration because it explicitly does not entail shifts in residence. However, its direct economic consequences can be thought of as those of migration. Workers enter a country temporarily to carry out particular jobs and thus labour inputs in one economy are reduced while those in another are increased. At its simplest the motive for a worker to work abroad is that her real wages are higher there. Corresponding to these different wages are different productivities. In reasonably competitive labour markets, workers are paid their marginal products firms pay workers the value that they generate and even where this is not true, the differences are not usually very large. Thus we can be very confident that when a worker moves from a low wage to a high wage country, her productivity increases and world aggregate output rises, offering scope for economic gains. In the extreme case in which workers from different countries are identical and productivity is purely a function of location-specific characteristics, the increment in output when a worker moves is equal to the difference in wages between the two countries involved. In an early model of this case, Hamilton and Whalley (1984) suggest that if labour were able to move between regions sufficiently to equalise wages around the world, world income could increase by 150% or more! Varying the assumptions e.g. to reflect higher dependency ratios in developing countries, different costs of living in different countries, or incomplete wage equalisation would still allow huge gains, far in excess of anything observed elsewhere in the trade liberalisation literature. In a much later back-of-the-envelope calculation of a different nature, I suggest the possibility of gains of over $300 billion p.a. from increased labour mobility - Winters, (2001). Suppose, very conservatively, that when a worker moves from a low to a high income country, she could make up only one-quarter of the productivity or wage gap between the two countries. (That is, assume that three-quarters of observed wage gaps are due to differences in individual characteristics such as health, education or culture, and hence that they would persist even after developing country workers started to work in the rich countries.) Suppose also that fifty million additional developing country workers worked abroad in any year, equivalent to an increase of about 5% in industrial countries populations. With a wage gap of, say $24,000 p.a., the gains would be $300 billion p.a.!  In the next section I discuss some new estimates combining these two methodologies based on work by Walmesley and Winters (2002). These also suggest huge returns to even relatively small movements of natural persons. Increasing developed countries quotas for incoming TMNP by 3% of their labour forces generates gains of over $150 billion p.a. Within the general heading of labour mobility, it is useful to identify three particular dimensions: the flow of unskilled workers from developing to developed countries; the flow of skilled and professional workers from developed to developing countries; and the flow of skilled, professional, and particularly business workers from developing to developed countries. Of course there are flows between pairs of developing countries and between pairs of developed countries, but they are not the North-South movements that are the concern of this paper. Developed-to-Developing Country Labour Flows The main issue here is the ability of developed country firms to send their specialists to their plants in developing countries, so-called intra-corporate transfers. In some cases it is highly skilled technical workers who are required, often at short notice and for short periods. Such workers are necessary for commissioning new plants and equipment, repairing and maintaining such equipment, and for providing intra-firm services such as accounting, designs or legal advice. In other cases, the interest is in the mobility of managers either senior managers to oversee major functions or the regular rotation of middle management. Firms already see these various flows as a means of increasing local efficiency and of integrating their operations on a global scale, and one must presume that they increase global output (as well as the multinationals profits). Indeed, they are central to the dissemination of both hard and soft technologies to the developing world, and so apart from very long-run concerns about the incentives they create or destroy for human capital formation, they are an important contributor to development. Tang and Wood (1999) have shown in a simple model that, as with most migration driven by wage differences, business mobility increases world output. Not surprisingly they find that it narrows the skills gap (the difference between skilled and unskilled wages) in the developing host countries (unskilled wages rise) while widening it in developed countries. In the latter, home, country unskilled workers suffer from having fewer skilled workers to cooperate with and from the competition from cheaper unskilled labour abroad. In the developing country, output (GDP) increases, and although part of it accrues to the mobile skilled workers who are domiciled in the developed country, part accrues at home in terms of higher unskilled wages and tax revenues. Such mobility might reduce the developing countrys skilled-wage and so reduce the incentive for education, and this is re-inforced by the fact that, once mobility is permitted, the MNC might be able to do less local training. Hence there is a possibility of longer-term costs to the developing country. This is far from certain, however, and has to my knowledge never been examined formally. Analytically it has parallels with the arguments surrounding the so-called beneficial brain drain to which I turn next. Developing to Developed Country Flows: Skilled Labour The second element of labour mobility is the flow of skilled workers from developing countries, including the so-called brain drain. The value of skilled labour to a well-functioning economy has never been plainer and in certain sectors e.g. IT, education and health developed countries are now actively seeking to recruit from abroad. Ones immediate reaction is that if the advanced economies gain, the developing countries from which these skilled workers emigrate must necessarily lose. Indeed, this is not an implausible scenario and it is one which should worry development specialists. The loss of the services of skilled people, even temporarily, reduces total output, and hence the tax base and scale economies. Depending on the extent of the skilled workers absences, it could also reduce an economys entrepreneurship, the ability to absorb new technologies, and various positive spill-overs from skilled to other workers and society in general. But, in fact, straight loss is far from inevitable, and is much less likely with TMNP than permanent migration. For example, skilled workers from developing countries are likely to be more productive and have higher earnings in advanced economies, and the share of their higher earnings that they bring home may more than fully offset the loss of their services locally. This is particularly true if the developing country had not been making optimal use of the skilled labour initially, say for bureaucratic reasons or because the necessary complementary inputs were not available. These arguments have been made previously about remittances from permanent or quasi-permanent migrants, but they apply with far more force to TMNP. Workers abroad are also likely to be a source of ideas, technology, markets or networks for those who remain, increasing their productivity and market opportunities. Again this applies to permanent migrants, but the spillovers are likely to be much stronger if the workers with foreign experience spend more time living and working in their home economies. Under these circumstances TMNP will boost local productivity as returning skilled workers instruct or inspire local colleagues. It is true, however, that TMNP may be less effective at building up foreign networks than permanent migration. Workers who are abroad only temporarily may also be subject to income tax in their home rather than their host countries. Under current law, this depends on how temporary they are, but it is potentially a major consideration in the allocation of the benefits of TMNP between host and home countries. A third possibility is that TMNP increases the returns to education and that the resulting increase in the supply of skills exceeds the actual loss of skilled inputs through TMNP, leaving the domestic economy in the developing country a net gainer of skills. Commander, Kangesniemi and Winters (2002) analyse these arguments in some detail. They find them quite plausible and uncover several reasons why again TMNP offers greater scope for gains than does permanent migration. They do caution, however, that the better is the receiving (host) country at screening potential immigrants so that it admits only the most able, the weaker the beneficial brain drain argument becomes. This is because the argument relies on the chance to migrate makes education more attractive to people who could otherwise not have taken it, and, on the whole, these will be less able than those who would find education profitable just in the domestic economy. If only the latter group are candidates for migration or TMNP, the incentives for others to acquire education are not affected and hence there is no additional human capital formation. Clearly developing countries policies towards skilled TMNP should depend heavily on the net balance of these effects, and this is currently very uncertain. Moreover, the balance is likely to vary by country. For example, highly skilled workers seem to benefit from clustering and the larger the market they operate in the greater their worth (because the more people there are to benefit from their good ideas). Thus one can imagine the following taxonomy of countries: very small economies which could never generate the market or society size to make acquiring skills very profitable; they gain from migration via remittances, network effects, etc. large economies which can reliably create the mass of skilled workers necessary for efficiency; while migration may reduce their local supplies of skills, it does so only on the margin, and its effects may be off-set by remittances, etc. medium-sized economies which, on the other hand, may be prevented by migration from reaching the critical mass of skills necessary to achieve local take-off in high skill activities; these suffer a quantum decline in local value added that no remittance or networking could ever overcome. This is all speculation, of course, and we have little idea of what small, medium and large means in this context, but it does at least warn us that attitudes to migration could differ between developing countries. Developing to Developed Country Flows: The Unskilled While not entirely frictionless, flows of skilled workers are much easier for developed countries to handle politically than is general migration. But the real gains from trade, be it in goods or in factors, come from exploiting differences. Hence it is the flow of unskilled (or, strictly, less skilled) workers from developing to developed countries that promises the larger returns. Not only is the proportionate gap in productivity between host and home countries likely to be largest here, but so too are the numbers of people available to move. The large benefits cited above come from less skilled mobility. But there are formidable political problems associated with large-scale permanent unskilled migration. Host countries fear cultural and integration problems because the unskilled are less likely to adapt to Western culture; they fear drains on the public purse; the jobs that unskilled immigrants take do not command immediate respect and appear to be at the expense of the employment of local unskilled workers. Given the ability of various lobbies to ensure the that protection in OECD countries is skewed towards supporting unskilled wages and employment, it is no surprise that the same forces have been able to resist immigration so effectively. TMNP offers a way out of this dilemma. Although it will clearly deliver only some of the economic benefits available from straight migration in terms of output and income, it avoids most of the latters political costs. Temporary movers pose no cultural or integration threat and make virtually no call on public services. Thus, the only major challenge posed by well-run TMNP schemes is the increase in competition that they pose for indigenous low-skill workers. This is neither more nor less than the challenge posed by imports of labour intensive goods from the developing world, and it has the same aggregate gains and distributional consequences (losses for the low-skilled, gains for everyone else). Formidable though it has been, the resistance to the liberalisation of imports of labour intensive goods has been at least partly overcome in the past by the weight of the economic gain that trade can deliver and by policies designed to ease adjustment among the local unskilled. Applied with the same policies and sensitivities as trade policy reform in goods has received in the past, TMNP among less skilled workers offers the chance to reap some of the large gains described above. Moreover, as their populations age and as average levels of training and education rise, the scarcity of less skilled labour in developed countries will worsen. And given that at least in some occupations there is really no substitute for human labour e.g. in the caring occupations, personal services and delivery of goods the benefits of TMNP will increase through time. Thus while recognising the formidable political challenges it poses, TMNP actually offers a strong long-run identity of interest between developing and developed countries. Developing the Empirical Model The major challenge to modelling the effects of Mode 4 liberalisation as a simple trade liberalisation is the complete absence of information about size of the barriers to services trade. (This is true of virtually all services trade, not just that delivered by TMNP.) Using TMNP to deliver services faces both para-tariffs, such as the costs of visas, additional health insurance, registering qualifications, etc., and quantitative restrictions. The latter come in both specific form, such as the failure to accept foreign qualifications as substitutes for equivalent to the domestic qualifications necessary for delivering a particular service, and in the general form of immigration restrictions. Often both will have to be relaxed before an increase in trade can occur. Since many QRs are prohibitions, they preclude the direct collection of data on their restrictive effects in the form of the differences between delivered and border prices. That is, there are no data from which we could directly infer the tariff equivalents of services QRs. Clearly one way of unlocking the empirical estimation of the costs of Mode 4 restrictions would be to seek alternative ways of quantifying these barriers for specific sectors. This approach amounts more or less to preparing a business plan for the provision of service X in market A by country B residents. Taking the wage in B for the workers concerned as given, one would need to quantify the additional costs of providing the service in A. Some would be obvious e.g. subsistence, insurance, travel while others could be less so e.g. the need for advertising, or discounts for customers uncertainty about quality or reliability. Yet more would be policy-related: current regulations may require electricians to complete a full training course in country A, but if objective criteria were applied with goodwill one might conclude that merely re-examining foreign-trained electricians was sufficient. Whether these barriers were costed at their actual or perspective rates would depend on whether one was considering their relaxation or not. Given the cost of B-residents supplying a service in A (which will clearly depend on the scale of operations assumed), one could compare this with the existing price in A. A small relaxation in the quota of TMNP would then generate rent equivalent to the difference between the two costs which would presumably to be shared between the mobile workers and the firms/institutions facilitating their mobility. If the relaxation in quota were non-marginal, one would then need estimates of the elasticities of supply of temporary workers and demand for the service in order to calculate the new equilibrium. The latter is not beyond imagination, but despite the fact that modellers occasionally make assumptions about it, the former elasticity is extremely difficult to estimate. Clearly any exercise such as this will be highly sector-specific and subject to very wide margins of error. Moreover, while it might indicate possible changes in trade volumes it would not necessarily immediately generate estimates of the overall welfare benefits of liberalisation. This is because many of the necessary monetary costs of TMNP e.g. those related to training and testing workers contains rents and/or subsidies. The per-unit rents and subsidies need to be quantified and the new volumes of activity to which they apply predicted in order to calculate the gross losses or gains they imply. In addition, the location in which the rents accrue needs to be identified. Moving to the opposite end of the spectrum, the problems with modelling TMNP as migration are (a) that the transactions costs associated with temporary mobility are ignored, (b) knowing the elasticity of supply of mobile workers, and (c) the implicit assumption that the work force of permanent and temporary workers is costlessly distributed across sectors within the economy according to labour demand. Items (a) and (b) come back to the business plan exercise discussed above; item (c) requires some discussion, however. Virtually all trade liberalisation modelling exercises involve the assumption of perfect long-run mobility of each kind of labour between sectors and the consequent equalisation of real wages for each kind across sectors. Some models allow for nominal wage difference between sectors that reflect the sectors non-pecuniary advantages or disadvantages. The latter are assumed unvarying and proportional to the wage so that, in fact, all wages move up and down together. Other models allow for upward-sloping aggregate supply curves of labour to reflect the way in which higher real wages will attract more people into the work force. None of these, however, addresses the issue that seems most pertinent to TMNT, namely the frictions on moving between sectors. Suppose some temporary workers are allowed in to work in a particular sector, j. If the labour for this sector were wholly specific to that sector, wages would fall and employment and output could rise. There would be spillovers to other sectors via the consequent changes in output prices the additional output would drive down prices and hence cut the demand for other goods but no direct spillover to other sectors labour markets. If on the other hand, labour were perfectly mobile between sectors, the incipient wage decline induced by the extra workers in the sector j would immediately drive some existing workers out of j into other sectors, reducing wages in other sectors. Ultimately, wages would fall by the same proportion in all sectors and, at least provided that demand factors were not unduly biased, all sectors would experience an increase in employment and output. The output shock would be spread throughout the economy, and the shock to the sector j would be much smaller in this case than in the sector-specific case. Which of these stories is more plausible? In the very short-run, labour is fairly sector- specific, especially if workers have sector-specific skills, so that impacts are deep and narrow. Kletzer (2001), for example, finds quite long lived unemployment and wage cuts for some of the workers losing their jobs in import-competing sectors. In the longer run, however, most economies show a good deal of flexibility and so impacts are broader and shallower. Borjas and Freeman (1992) show that US regions which attract large immigrant inflows experience corresponding declines in internal inflows or increases in worker outflows, for in the long run their work-forces seem to be no different from what would be expected in the absence of immigration. Given it is probably more costly to shift the sector than location (especially for the less-skilled) we might take this as evidence to support the single labour-market assumption. The fact that workers attitudes towards globalisation owe more to their skills levels than to their sectors of employment (Schreve and Slaughter, 2001) also suggests fair degrees of mobility. Clearly the truth lies somewhere between the two polar extremes, and while I incline to the view that, subject to suitable compensatory polices, the flexible labour markets approach is adequate for most policy-making jobs, there is obviously room for additional research on how to model inter-sectoral mobility. One approach that might yield fruit would be to examine the price relativities between different services across countries and check whether they could be related to the openness of the economies to sector specific migration or temporary mobility. If sectors differ in their openness, but relativities do not, one might take that as indicating that labour is relatively mobile between sectors. As with any price comparison, there will be difficulties in ensuring that like is being compared with like, but there should be fairly standard tasks that can be compared and one could try to choose countries that have similar levels of income and other demand conditions. One might also try to get around some of these difficulties by considering a whole range of services or occupations and estimate wage equations explaining earnings in terms of the individuals charcteristics and occupational/sectoral location. This might identify outlying sectors whose openness could be investigated and if one compared such wage gradients over countries one might be able to identify anomalies even more effeciently. For long-run analysis of the sort that trade liberalisation requires we need to identify, not so much the costs of moving sectors (a transitory phenomenon), although these affect the net benefits of liberalisation, as the costs of being in the wrong sector. That is, the permanent loss of productivity implied by a worker leaving her preferred sector for another. To my imperfect knowledge, we have no evidence on this at all, although ultimately, over the space of several generations, we might assume it to be very small. 2. The Gains from Temporary Movement - New Estimates  Who would benefit from liberalising the restrictions on the temporary movement of natural persons (TMNP) and by how much? This section summarizes some recent modeling results derived from a global applied general equilibrium model of South-North temporary movement of labour. The method is to fit a computable model to data from a base year (1997) and then asks how the outcome would have differed if there had been freer labour mobility in that year. Thus the results are not unconditional predictions of the effects of future policy changes, but rather quantitative thought experiments to suggest possible orders of magnitude. In the absence of quantifiable data on restrictions to services trade per se, we model TMNP merely in terms of the movement of workers from one country to another. This clearly overlooks a huge array of institutional details in actual and potential schemes for the temporary mobility of labour. However, as we have argued above, we believe that in terms of the effects on narrowly economic variables, it is not seriously misleading, especially in the sort of model we use. Ultimately TMNP means that fewer workers produce at home and more do so in the host country. The bottom line of the modeling exercise is that increased mobility equivalent to 3% of the receiving countries work forces would generate $156 billion per year in extra economic welfare. These gains are shared between developing and developed countries and owe more to unskilled than to skilled labour mobility. We believe these results are informative, although one can not possibly rely on the specific numbers. The Model The model and data used in Walmsley and Winters (2002) are based on the GTAP model and database developed by Hertel (1997). GTAP is a standard applied general equilibrium model which assumes perfect competition; consequentially this exercise contains none of the scale or clustering effects which often figure in the skilled migration literature. In each of several regions, a single household is assumed to allocate income across private and government consumption, and saving in fixed proportions. Demand for domestic and imported goods then depends on income and relative prices. Firms minimise the costs of production. They combine intermediate inputs, from domestic and imported sources, with primary factors to produce commodities for the domestic and export markets. Demand for factors of production (land, skilled and unskilled labour, capital and natural resources) depends on output and relative prices. Prices adjust to ensure that demand equals supply in every market. We modify the standard GTAP model and to incorporate the movement of natural persons as follows. We start by distinguishing the terms temporary migrant and temporary worker: A temporary migrant leaves his or her home region to become a temporary worker in a host region. Given that there is no bilateral information, we can say nothing about where temporary migrants from a given home region become a temporary workers, so we postulate a global labour pool, which collects up the temporary migrants from all home regions and then allocates them across host regions. The temporary workers add to the supply of labour in the host region and are allocated across sectors within the region according to labour demand. In the host country temporary workers earn a wage for their labour, related to their productivity. Part of this wage is then sent back to the home region via the global pool as remittances. Within a country, the income of permanent and temporary residents plus net remittances received is then allocated across consumption, saving and government spending to maximise utility. We characterise changes in policies towards TMNP as increases in developed countries quotas on inflows of temporary workers. Assuming that the quotas are always binding, i.e. that there is excess demand for places in the host countries, we can do this exogenously without having to model the incentives to move very precisely. We then assume that the new migrants are drawn from various home countries (mostly developing) according to the latters labour force shares Having determined the number of temporary migrants leaving the home region and the number of temporary workers entering the host regions, we need to calculate how these changes affect the effective supply of skilled and unskilled labour in terms of productivity units. Because we have no data on bilateral flows of workers, we have to assume that a temporary worker initially has the same productivity as the average temporary migrant in the pool of mobile labour. The latter merely reflects the productivities of these workers in their home countries and the shares of each home country in the overall total of temporary migrants. Once working in the host region, however, the temporary worker acquires some of the productivity of the host region, and his productivity is assumed to equal the average productivity of a temporary migrant plus half the difference between that and the host regions productivity Once the temporary workers have left their home regions and entered the work force of the host region and they are allocated across sectors in both countries. In the standard model, labour moves freely between sectors until wages are equalised across sectors for each type of worker. In the host regions, where the supply of labour has increased, wages are expected to decline, whereas in the home (sending) regions, they will rise. The extent of the change in wages will depend on the demand for labour which in turn depends on the demand for production, which is driven by prices and income (both of which depend on wages). Changes in the supply of labour and wage rates will ultimately affect the demand for other factors of production, notably capital. In the standard GTAP model, income includes all factor incomes (skilled and unskilled labour, land, capital and natural resources) net of depreciation and taxes. In our model factor incomes have to reflect the distinction between the incomes of temporary and permanent labour, and be adjusted for the formers remittances to their home countries. The income of temporary workers in an economy is assumed to comprise the income from their labour less remittances sent home (which, in turn, are assumed to be a given share of the wage). All other income, including that on land, capital etc, taxes and remittances received is assumed to be earned by permanent labour alone. Since we do not have data on bilateral labour movement or remittances, we need to base remittances received on average remittances from all temporary workers. Given that temporary workers in different host countries remit at different rates, changes in the geographical of temporary workers may therefore lead to changes in the average remittance rate. In order to calculate the effects of TMNP on the migrants from a particular home country, the income of the temporary labour by host region and labour type is aggregated across all host regions, and then distributed across home regions according to their numbers of temporary migrants in productivity equivalents. Changes in the economic welfare of permanent and temporary workers are related to their income flows deflated by prices in their place of residence (work). The welfare of temporary migrants is found by summing the welfare changes of temporary workers across host countries and sharing it out over the various home countries, according to their shares in total TMNP. Once the welfare changes of temporary migrants are determined, welfare by home region, regardless of temporary residence, can also be calculated by simply summing the relevant changes. Table 1 summarizes the way in which income and welfare changes are summed to obtain national totals. We distinguish home and host country concepts. The former refers to all people starting off in a particular country essentially a nationality concept. The latter refers to all people actually located in a country after mobility has occurred essentially a residence concept. Ata practical level the only difference is the treatment of the income retained by temporary workers, which is attributed to the host country when we use the host country concept and the home country when we use the home country concept. The Experiments The main simulation that we have conducted is of an increase in the quotas for inflows of skilled and unskilled temporary workers into developed countries. Following this, the effects of other issues, such as the relative importance of skilled versus unskilled mobility, and the sectoral allocation of the mobile workers are examined. Quotas on the movement of natural persons are assumed to increase in a number of traditionally labour importing regions, supplied by temporary migrants from a number of traditionally labour exporting countries according to their labour force shares. Table2 divides the regions used in this analysis into labour importing and labour exporting regions (non-zero entries in columns II and III respectively). The quotas are increased by an amount which would allow the quantity of labour supplied in the host (or labour importing) countries to increase by 3%. For example, in the case of the USA, the increase in the quota would amount to 2.7 million unskilled temporary workers and 2.4 million skilled temporary workers. China as a supplier of temporary workers, would then supply 2.4 million of the total 8.5 million unskilled workers required and 0.49 million of the total 8 million skilled workers required world-wide. Columns II and III of Table 1 give the assumed changes to regions labour forces. Increasing developed countries quotas on both skilled and unskilled temporary workers increases world welfare by an estimated $US156 billion about 0.6% of initial world income. Tables 3 and 4 give some geographical details, but, for these, care must be taken to distinguish home country residents those who start off in a country but some of whom move temporarily from host country residents the set of people who end up there after movement has occurred. Region A as a home country refers to As permanent workers who never leave plus its temporary migrants who work abroad -loosely speaking a nationality-based concept. A as a host country refers to the permanent workers plus the temporary workers from elsewhere who work there a residence-based concept. In aggregate terms the main gainers for liberalizing Mode 4 are the initial residents of the developing (labour exporting) economies, as we can see from Column V in Table 3 - developing countries as home countries. Most of this increase is the result of the higher incomes earned by the people who can become temporary migrants as a result of the relaxation in quotas (Column III in Table 3). They are now able to earn higher wages in the developed countries, as shown in Column II of Table 3 under welfare of temporary workers. [Recall that each mobile worker is both a temporary worker and a temporary migrant so columns II and III report the welfare of the same set of people allocated once by residence and once by nationality.] Despite the remittances they receive, permanent residents in the developing countries generally lose from the outflow of temporary migrants (Column IV in Table 3) because the decrease in labour supply reduces the returns to capital and other factors of production (Column IV in Table 4). Combining the results for permanent residents and (the few) temporary workers already located there gives the outcomes for developing countries as host countries (Column VI of Table 3); in general these economies record losses, but recall that this excludes the benefits experienced by the temporary migrants who are working abroad. The loss of these factor inputs reduces aggregate output in the developing countries, real GDP, in the labour exporting countries (Column V in Table 4), and because the outflow of labour is biased towards skilled labour, skilled workers real wages rise in developing countries (Column II in Table 4). Developing economies generally experience improvements in their terms of trade as the fall in their GDP reduces the supply of the varieties of goods that they produce and so drives up their prices. There are exceptions to these trends, however, especially in India, where the welfare of permanent residents increases because the increase in remittances outweighs the declines in labour and capital income. This increase in income increases the demand for domestic goods and attenuates the decline in production in the economy. It allows the real wages of both skilled and unskilled workers to rise. While the developing countries are the main beneficiaries of the increase in quotas, the initial residents of most of the developed countries also experience increases in welfare from the higher returns to capital and the increase in taxes collected. In fact the estimates in table 3 for developed countries permanent residents welfare are understated. The database available for this exercise shows the USA subsidizing output and factor use, so that expanding the USAs economy causes its welfare to fall! This is manifestly absurd, and if we over-ride the data and give the USA the OECD average tax structure, developed country permanent residents gain about $11 billion rather than the $7 billion shown. Real GDP increases substantially in the developed economies (Column V in Table 4), but in most cases the terms of trade decline as higher output drives down the prices of exports relative to imports. One number of note in table 3 is the strong positive effects on developed countries temporary migrants i.e. those who have left a developed country to work abroad. This reflects the fact that over half of the stock of skilled temporary migrants identified in our database comes from the Rest of the EU region (EU less than UK and Germany). With our unavoidably crude way of allocating these over destinations, many of them are allocated to developing countries (mostly in the Middle East), where they benefit from the strong increase in skilled wages. Among sectors, agriculture is the least affected by temporary labour movement in both the developed and developing economies. In developed economies, output of services (particularly trade, business services and other services) and most manufacturing sectors increase significantly with the labour inflows, while primary sectors and utilities show only small increases. The impact on production in developing countries is almost the mirror image, with the manufacturing and services sectors experiencing most of the decline due to the loss of skilled labour, and smaller changes in utilities and unskilled-intensive sectors. Table 5 considers the effects of relaxing the skilled and unskilled quotas separately. Notably, both the developed and developing countries would benefit more from the liberalisation of restrictions on unskilled labour than on skilled labour. For developing (labour exporting) countries, the reason is that, while skilled temporary migrants can greatly increase their earnings by moving, the negative effect of their loss on their home economies is considerable. Thus, for example, while the skilled temporary workers improve their welfare by $61 billion, permanent residents in the developing world lose $34 billion while those in the developed world break even (although this last figure is too pessimistic because of the USA data problem). Turning to unskilled mobility, the temporary migrants from developing countries gain $38 billion in all, and their remittances more than offset their original (low) contribution to home output, so that the welfare of those who remain behind also rises. For the developed (labour importing) regions, higher quotas on unskilled labour are also more beneficial in terms of welfare than are those on skilled workers, although most of the effect comes from the welfare of developed countries temporary migrants, an effect which is not completely convincingly modelled. The increases in supplies of unskilled labour reduce unskilled wages and stimulate most sectors (agricultural, manufactures and some services), whereas the benefits of increased supplies of skilled labour are concentrated in just a few services sectors. The main effect, however, is again on their existing migrants. Walmsley and Winters (2002) also provide some alternative estimations. They confirm that the precise details of the experiment do not undermine the basic results of their paper viz. that Mode 4 offers potentially huge gains to liberalisation. First, we confined the inflow of labour to the developed countries service sectors. We did this by postulating two separate labour markets between which there was no labour mobility and hence between which wages could vary. The inflows of labour to the services sector results in a large expansion of the services sectors in the developed economies at the expense of the other sectors. In the services sectors the wages of skilled and unskilled labour decline (by between 1 to 2%), while in the other sectors they increase substantially (by between 1 to 3%). Capital is replaced with the cheaper skilled labour in the services sector, allowing it to move to other sectors agriculture and manufacturing. Developing countries see the opposite result. Their services sectors decline by more and their non-services sectors by less than in the unrestricted increases case (or even expand). It is not surprising that restricting the sectors which can use temporary labour reduces the benefits for all the major groups we distinguish, for we are frustrating one of the routes in which economies maximise the benefits of the inflow of labour. However, the loss from the restriction is not very large - benefits of $152 billion compared with $156 billion. Moreover, the assumption of completely separate labour markets is too extreme. Thus, at least for the initial stages of a liberalisation, restricting new temporary labour to only the services sectors (as Mode 4 will do) will not seriously reduce the gains from those available under unrestricted mobility. The main results are also basically unaffected if we restrict labour flows only to traditional bilateral flows (e.g. USA from Mexico, EU from the Southern Mediterranean), and if we assume that developing countries have perfectly elastic supplies of unskilled labour. We also show that the gains are more or less proportional to two critical parameters. If we assume that incoming workers catch up only one-quarter of the gap between their home and host country productivities (rather than one-half), the benefits are roughtly halved. Similarly, if we assume a three-quarter catch-up, they are increased by 51%. Gains are similarly proportional to the size of the liberalization, very nearly doubling if quotas expand by 6% rather than 3% of the work force. Finally, by way of comparison, we find that the complete abolition of all goods trade restrictions in the model generates only $104 billion of welfare gains compared with $156 billion from a limited relaxation of restrictions on labour movement. It bears re-iterating: do not take these numbers literally. They are mere orders of magnitude. They are not subject to much uncertainty in terms of the experiments that we run, but they are broadly uncertain once we recognise the inadequacies of the original data, the crudity of treating temporary mobility as a mere change in countries labour endowments, and the extremity of the modeling assumptions. Nonetheless, the results suggest startlingly large benefits to freeing up the temporary movement of labour. Even for a limited liberalisation, they are far larger than the benefits available to a complete goods market liberalisation and they are also larger for unskilled than for skilled labour mobility. If they do nothing else, these results should challenge negotiators to think hard about the priorities they bring to the new round. 3. Compensatory Policies I have argued above that admitting less-skilled workers under a Mode 4 liberalisation is fundamentally no different from admitting imports of labour intensive goods under a GATT liberalisation. Both raise general welfare, but threaten indigenous less-skilled workers. This sub-section briefly considers how strong that parallel is and what lessons goods market liberalisation has or Mode 4. The postulated equivalence seems a good one to me so far as the nature of the shock is concerned. However, there are questions about whether the scale is equivalent. In developed countries, services are substantially larger employers than manufacturing and primaries and while they employ large numbers of highly skilled workers, they also employ a high proportion of societies less able members. Sectors such as personal care, janitorial services and much hotel work offer havens for the unskilled, and given that any society will have some people in such categories, they play a role both in terms of providing these people with income and allowing them the self respect of contributing. . Indeed, these sectors have often been characterised as providing the jobs to which displaced unskilled manufacturing employees can be moved when their sectors have been liberalised, although Kletzer (2001) suggests this is much less true in reality than in perception. However, their existence has helped to ease adjustment in goods sectors in a way that will not be feasible when they themselves come under pressure. If all low-grade service jobs in the developed countries were costlessly contestable by residents of the poorest developing countries at those countries wage rates, the indigenous unskilled would indeed be squeezed very hard. However, that is clearly too extreme a view, for the natural protection of distance, culture and experience will all maintain wage premia in the developed countries relative to the developing world. And at least in the foreseeable future one is talking only about partial liberalisation. Nonetheless, it does seem reasonable to expect that a vigorous Mode 4 liberalisation will pose significant adjustment strains. What could be done? Consider, first, the long run. It is important to realise that the displaced unskilled workers could be compensated out of the general gains from liberalisation. But that would require a willingness to be taxed and make transfers on the part of the more able majority, and the construction of a tax system that did not seriously discourage effort. If there are significant numbers of natives who could not make a decent living in the sheltered sectors, one would have to rely on more or less permanent transfers on a large scale and it is not clear that societies have yet really mastered this technically. (Think of the social dysfunction and distress among many indigenous peoples at present.) The sort of subsidies that might be necessary would be housing subsidies for nationals or income top-ups that brought a nationals wage up to acceptable levels even if his employer paid only the developing country wage. This discussion suggests that there is a huge return to trying to ensure that there are fewer nationals on the bottom rungs of the skills ladder. This, in turn, suggests an ever-increasing use of the education sector to increase individuals endowments of human capital, and might also entail finding ways of allowing all nationals to hold a reasonable stock of other capital. Capital owners are major beneficiaries of the inflow of workers from abroad, so if profit streams were equitably distributed, nationals would have reasonable living conditions even if they had low earning power. Exactly how such capital transfers were handled would need careful thought in order to balance freedom and personal responsibility with assurance that the income flow would remain in tact. This is exactly the debate that developed countries are having over pension plans, and one could indeed think of this problem as being akin to a lifetime pension. Turning to the shorter term and recognising that even the most aggressive liberalisation will proceed relatively slowly compared to the size of the overall economy, one can draw better parallels with goods trade liberalisations. There are broadly four approaches that deserve comment. First, sensitive sectors are just left out of the process of liberalisation. We all know that this is what happens, but it does not make it good economics. If we are serious about Mode 4 and we should be for the sakes of both developed and developing countries we should tackle some of the major sectors early on. Second, liberalisation can proceed slowly. By this I do not mean so much taking a small step and then waiting before deciding to take another, but rather planning a long transitional period for a known adjustment. The key here is credibility. If the liberalisation is not credible, long adjustment periods are an invitation to lobby for their reversal, but if the end point is firmly expected at a date certain in the future, the gradual introduction of change can give individuals a chance to adjust more gently. One should not believe that slow adjustment is always better, but just that unless there are obvious externalities, it is better to give private actors the information and let them decide the best speed of adjustment. Also tied up with the timing is the question of the macro-economic cycle. It is manifestly more difficult to get acceptance of liberalisation if the economy is weak and the costs of employment change is greater because transitional unemployment spells are longer in depressed economies. Thus there is something to be said for planning Mode 4 liberalisation to occur during the boom rather than the recession. This raises the question of whether a safeguards clause is necessary for Mode 4 (or other services as well). Such clauses are likely to be subject to a good deal of abuse, but in terms of political reality, they may be desirable. The third approach to compensation is specific trade-related compensation schemes. Among these the best known is the USAs Trade Adjustment Assistance Act. This offers a composite of measures to support an industry damaged by liberalisation with loans and assistance plus measures to compensate the displaced workers, including benefits to support income and training services. The TAA was established by the Kennedy administration in 1962 as a quid pro quo for the wave of liberalisation lead by the Trade Expansion Act (TEA) in the United States and the Kennedy Round in the GATT. It provides trade-displaced workers with extended unemployment benefits, relocation expenses and (compulsory) training as a bridge to a new job with similar levels of income and benefits. Several evaluations of the TAA programme have shown it provides additional income for temporarily displaced workers, many of whom obtain alternative employment relatively quickly anyway. But it fails to assist significantly those permanently displaced by trade-related closures. In addition, Decker and Corson (1995) suggest that training does not increase the future earnings of displaced workers. Nonetheless, the TAA forms the basis of the North American Free Trade Agreement Transitional Adjustment Assistance (NAFTA-TAA) programme, established in 1993. This assists workers who lose their jobs or whose hours of work and wages are reduced as a result of trade with Canada or Mexico by providing them with the opportunity to engage in long-term training while receiving income support. Canada and Australia have operated similar schemes at various times in the past - the General Adjustment Assistance Programme (GAAP) in Canada, and the Special Adjustment Assistance (SAA) in Australia - and with similarly unconvincing effects. Overall, experience with trade adjustment assistance has not been particularly happy. Schemes are often bureaucratic, providing limited benefits to a small category of workers who might well have found alternative jobs anyway, while providing little long-term assistance to the permanently displaced. Decisions on whether a worker is displaced due to government trade policy or some other shock have inevitably been rather arbitrary, leading to resentment among workers who fail to qualify for the benefit. In some cases, such schemes have assisted firms in moving to activities better reflecting comparative advantage, while in others they have inhibited such a move. It is notable that the Europeans and Japanese have no such specific compensation schemes. This is argued by Sapir (2001) to be partly because their less effective markets have given them a degree of insulation against trade shocks, and partly because of their much deeper general social protection systems. These are the fourth of the approaches to adjustment I want to mention. They avoid the problem of attributing an individuals problems to trade and of prioritising trade-related losses above those with other causes, but they are expensive and arguably bad for incentives. The preservation of the social security system is often quoted as one of the aim behind Europeans resistance to immigration, but in fact, TMNP, which offers the foreign workers no rights under the systems, helps to get around these fears. Nonetheless, the European models are under budgetary pressure and their future is not entirely secure. A major shock from a Mode 4 liberalisation, even if temporary, could pose serious problems if not appropriately anticipated. This section has argued that the adjustment stresses that Mode 4 liberalisation could engender are real. They can not be wished away because they will be both large and concentrated on a vulnerable section of society. In the short run, sensitivity about the timing and extent of liberalisation may contain the pressures and existing compensatory schemes can cope with those that actually arise. In the longer run, when deeper liberalisation has been achieved, more active redistribution will be required to try to ensure that fewer nationals of developed countries are actually in the sectors competing with foreign workers. This requires education and training as well as thought being given to asset distribution. References: Borjas, G. J. and Freeman (eds) (1992): Immigration and the Workforce: Economic Consequences for the US and Source Areas, University of Chicago Press: Chicago. Bowen, H.P., A. Hollander and J-M. Viaene (1998) Applied International Trade Analysis, University of Michigan Press, Ann Arbour. Commander S, Kangesniemi M and Winters L A (2002) The brain drain: curse or boon? A survey of the literature, paper for the International Seminar on International Trade, Stockholm, May, 2002. Corden W M (1984) The Normative Theory of International Trade, in Jones, R W. and Kenen, P. (ed.s), Handbook of International Economics, North Holland, Amsterdam, pp.63-130. Decker, P. T. and W. Corson (1995), International Trade and Worker Displacement: Evaluation of the Trade Adjustment Assistance Program, Industrial and Labor Relations Review 48, pp. 758-74. DfID (2000a), Eliminating World Poverty: Making Globalisation Work for the Poor. London: Cm 5006, Her Majesty's Stationery Office. Dunlevy J A and Hutchinson W K (2001) The pro-trade effect of immigration on American exports during the late nineteenth and early twentieth centuries, Discussion Paper 375, IZA, Bonn. Fujita, Masahisa; Krugman, Paul and Venables, Anthony, J., 1999, The Spatial Economy: Cities, Regions and International Trade, MIT Press, Cambridge MA. Hamilton, C., and J. Whalley (1984) Efficiency and distributional implications of global restrictions on labour mobility: calculations and policy implications Journal of Development Economics 14 (1-2), 61-75. Hertel, T. W. (ed), (1997): Global Trade Analysis: Modelling and Applications Cambridge: Cambridge University Press. Hertel, T.W. et al. (1999) `Agricultural and non-agricultural liberalisation in the millennium round,  HYPERLINK http://www.worldbank.org/reaserch/trade/archive.html/ www.worldbank.org/reaserch/trade/archive.html/ Kapstein, E. (1998), Trade Liberalization and the Politics of Trade Adjustment Assistance, International Labour Review 137, pp.501-16. Karsenty, G. (2000), Assessing Trade in Services by Mode of Supply, in GATS 2000: New Directions in Services Trade Liberalization, edited by P. Sauv and R. M. Stern, Washington DC: Brookings Institutions Press. Kletzer L (2001) Job Loss from Imports: Measuring the Costs, Institiute fro International Economics, Washington, DC. Markusen, J.R. (1983) Factor movements and commodity trade as compliments, Journal of International Economics Volume 14, pp. 341-357. Mattoo, A. (2000), Developing Countries in the New Round of GATS Negotiations: Towards a Pro-Active Role, World Economy 23, pp. 471-90. McCulloch, N., L.A. Winters, and X. Cirera, (2001), Trade Liberalisation and Poverty: A Handbook, CEPR, London. Rauch, J.E. (1999) Networks versus markets in international trade, Journal Of International Economics (48)1, pp. 7-35. Rodrik, D. (1995), Political Economy of Trade Policy, in Handbook of International Economics, edited by G. M. Grossman and K. Rogoff, Amsterdam: Elsevier. Romer, P. (1994) New goods, old theory, and the welfare costs of trade restrictions, Journal Of Development Economics (43)1, pp. 5-38. Samuelson P A (1949) Factor price equalisation once again, Economic Journal, Vol 59, pp181-197. Sapir A (2001) Whos afraid of globalisation? in Sauve P and Subramanian A (ed.s) Efficiency, Equity and Legitimacy: The Multilateral Trading System and the Millennium, Chicago University Press, Chicago. Shreve K E and Slaughter M J (2001) What determines individual trade policy preferences? Journal of International Economics, vol 54, pp.235-66. Tang, P.J.G. and Wood, A. (1999) Globalisation, Co-operation Costs and Wage Inequalities. January 1999, mimeo, Institute of Development Studies, University of Sussex. Walmsley, T. L. and L. A. Winters (2002) `An Analysis of the removal of restrictions on the Temporary Movement of Natural Persons, Mimeo, University of Sheffield. Williamson, J.G. (1998) ` Globalisation, labour markets and policy backlash in the past, Journal of Economic Perspectives, Vol. 12, No. 4 pp51-72. Winters, L.A. (1991) International Economics. Routledge, London. Winters, L.A. (2001) Assessing the Efficiency Gain from Further Liberalization: A Comment, in Sauve P and Subramanian A (ed.s) Efficiency, Equity and Legitimacy: The Multilateral Trading System and the Millennium, Chicago University Press, Chicago. Winters, L A, Walmsley T L, Wang ZK and Grynberg R, 2002, Negotiating the Liberalisation of the Temporary Movement of Natural Persons, Economic Paper No ??, Commonwealth Secretariat, London, forthcoming. Wong K-Y (1988) Are International Trade and Factor Mobility Substitutes? Journal of International Economics v21, n1/2 (August 1986): 25-43 Table 1: Regions and Sectors I All RegionsII Labour Importersa III Labour Exportersa IV European Union PartnersbV North American PartnerscVI SectorsVII Services SectorsUSA5165.36CropsCanada573.00LivestockMexico375.90(MeatUK1134.00DairyGermany1665.00FoodRest of EU4713.10Other Primary Products Rest of Europe234.12Wood and paper productsEastern Europe538.37(Textiles and Wearing ApparelFormer Soviet Union1563.65(Petroleum, chemical and mineral productsAustralia-New Zealand435.90MetalsChina2916.65(AutosJapan2610.00ElectronicsRest of East Asia509.80(Other ManufacturingSouth East Asia1709.02(Household Utilities(India2844.72((Construction(Rest of South Asia661.84((Trade(Brazil654.46(Transport(Rest of Latin America1139.50(Communications(Middle East and Northern Africa1259.69(Financial services(South Africa1810.28(Insurance(Rest of World546.61Business Services(Total16530.4816530.49Other Service(a. The figures in this column are the inflow or outflow assumed in the simulations below, in thousands. b. EU Partners are the group of countries/regions where most of the temporary labour in the EU currently comes from. c. North American Partners are the group of countries/regions where most of the temporary labour in North America currently comes from. Table 2 Home and Host Concepts: Accounting concepts for the Temporary Flow of Labour from Country A to Country B Income FlowHost ConceptHome ConceptPLBBBKBBBTLB retainedBTATTLB remittedAAKAAAPLAAA PLj permanent labour in country j Kj other factors in country j TLB temporary workers in B = temporary migrants from A Retained earnings retained in B by temporary workers Remitted - earnings of temporary workers remitted to country A A included in country As total for permanent residents B included in country Bs total for permanent residents AT included in As accounts under temporary migrants BT included in Bs accounts under temporary workers Table 3: Economic Welfare by Region and Class of Workera I RegionsII Welfare of temporary workersIII Welfare of temporary MigrantsIV Welfare of permanent residentsbV Welfare by home region III + IVVI Welfare by host region II + IVDeveloped Countries17596068577698275559182942Developing Countries-500298984-2068578301-25688 of whichFSU and CEECs-2512511-49917521-5017East and SE Asia-76229647-1219217456-12955South Asia-534158163772053516325Latin America-71830980-1245718523-13175Africa-344421688-742214266-10866 TOTALe170932170704-14626156078156306a. $US millions b. Permanent residents who do not move temporarily. c. Home refers to people originating in the specified country regardless of where they work or live. d. Host refers to people living in the specified country regardless of where they originated from. e. Includes a small rest of the world region. Source: Authors simulations Table 4: Percentage Changesa in the Real Wages of Skilled and Unskilled Workers I RegionsbII % change in Real Wage of Skilled LabourIII % change in Real Wage of Unskilled LabourIV % Change in Rental Price of CapitalV % Change in Real GDPcVI % Change in Terms of TradeDeveloped Countries-1.02-0.610.781.05-0.24Developing Countries5.130.12-0.52-0.910.53 of whichFSU and CEECs4.40-0.48-0.70-1.030.18East and SE Asia4.940.02-0.49-0.880.28South Asia5.920.600.59-0.484.55Latin America4.67-0.21-0.63-0.880.06Africa5.750.20-1.02-1.390.37Percentage changes in variable from base case. Weighted averages of results for the regions distinguished in the model weights are skilled workers, unskilled workers, GDP, GDP and GDP respectively for columns II-VI. c. Readers are reminded that Real GDP is not a measure of welfare. Real GDP is a measure of production, while welfare is a measure of the utility achieved from consumption, which depends among other things, on remittances received. Source: Authors simulations Table 5: Welfare Decomposed according to effects of increasing Skilled and Unskilled Quotasa I RegionII Welfare of permanent workers (unskilledb)III Welfare of permanent workers (skilledc)IV Welfare of temporary migrants (unskilledb)V Welfare of temporary migrants (skilledc)VI Welfare of home region (unskilledb) II + IVVII Welfare of home region (skilledc) III + VDeveloped Countries6,86012150,58717,98957,44718,111Developing Countries13,097-33,78137,67661,30950,77327,528 of whichFSU and CEECs205-5,1963,4189,0943,6223,898East and SE Asia2,331-14,5239,55820,08911,8895,566South Asia9,2957,0831,7892,36911,0849,452Latin America-158-12,29813,21617,76313,0595,465Africa1,424-8,8469,69511,99311,1193,147 TOTALd20,181-34,80789,34681,358109,52746,551a. $US millions b. This is the welfare of the whole population when only quotas on unskilled workers are relaxed. c. This is the welfare of the whole population when only quotas on skilled workers are relaxed. d. Includes a small rest of the world region.Source: Authors simulations  I directed a study funded by the Commonwealth Secretariat, the results of which will be published shortly - Winters et al (2002) and Walmsley and Winters (2002). This paper represents a distillation of some of that study, plus some new analysis of the economics of Mode 4 liberalisation.  Winters (1991), for example, gives an elementary text-book treatment of FPE and Bowen, Hollander and Viaene (1998) a more advanced one.  The main reason for differences between wages and (long-run) productivity is market power: if workers can restrict their supply to a firm and if that firm reaps excess profits from market power in the goods market, then workers can share the rents. Alternatively, if firms can control the number of jobs in a labour market, they can drive wages down below marginal products.  Labour costs per manufacturing worker, which are an indicator of productivity, were about $32,000 in the USA in 1990-94, compared with $1,192 in India, $1,442 in Lesotho, $5,822 in China, and $6,138 in Mexico; high-income countries population was 927 million in 1997 (World Bank, 1999). We argue in Winters et al (2002) that temporary workers should cover medical insurance privately.  This section is based on Winters et al (2002) and Walmsley and Winters (2002), from which more details of the modelling and results may be obtained. I am extremely grateful to Terrie Walmsley for her inputs to the modelling exercise.  Details of the data are available in Walmsley and Witners (2002)  Technically, we use Hicksian Equivalent Variation to measure changes in economic welfare. This expresses changes in money terms which can be summed across classes of workers to derive national aggregates.  By this we mean the number of workers (actual bodies) increases by 3% of the labour force. Because relative productivities differ this does not mean that the labour force increases by 3%, since the labour force increases by the number of equivalent workers.  Note, however, that in Table 3 even labour exporters record effects on temporary workers. The reason is that, although we do not vary their number in our experiments, most labour exporters have a number of temporary workers in their base data, and as wages change, so does the economic welfare of these workers. The changes parallel those of permanent residents in these countries. A similar explanation applies for labour importers and temporary workers.  By assumption, the relaxation of quotas reflects the skills mix of developed countries and so is substantially more skill-intensive than the typical developing countrys labour force endowment.  Winters et al (2002) discuss the reason for these effects.  Robustness to the details of the experiment is not the same as robustness with respect to the details of the model, however.  See Kapstein (1998) for a history of the TAA. PAGE 9 PAGE 32 NQ`34.^_M M< &))++q222277DDGGOO ORR3\h\]flmrppqq|¿ɽ5CJ j0JUmH  OJQJmH 6mH CJ5CJ j0JUOJQJ;5H*50JjCJU jCJUCJCJ CJCJ(CJ$@ @NQ`z56bf$$$$d @NQ`z56bf@^_OP LM<fg`a~ R!S!o!!"(">"?"""i#Ž K L   _  7_  _  _   _  =@^_OP LM<fg`a~ R!S!$$$S!o!!"(">"?"""i##,&&&))J+K++$ & F$$ $ & FF $ & F$ & F $ & F_ h$ & F_ hi##,&&&))J+K+++,,y-z-p2q22277::??AADDD D!DYFZFHHKK O Oƾ|yvspmjgdxz|}.lmuvmn()i      WX  Q    '++,,y-z-p2q22277::??AADDD D!DYFZFHH$$ $ & FHKK O OdPePRRRR2W3W1\2\3\i\j\]].`/` c cffjjl$ OdPePRRRR2W3W1\2\3\i\j\]].`/` c cffjjllmvnooqprpppssuuKzLzr|þ}zwtqnkopghuv`  `   <` =OPNO|}(llmvnooqprpppssuuKzLzr|s||_τ܇$$ & F` h  & F` h$r|s||_τ܇ab;<FG˪̪-.cd;<ϳгH |yvspm23lmpq  78`a *ab;<FG˪̪-.cd;<$$$dϳг 01tuijq"/g h   2-U{+Pp :;-_0ſůjOJQJU 6B*hnH  B*hnH OJQJ 6hnH hnH 6CJ5CJ OJQJmH 5CJ6 j0JU5CJ j0J5CJ UE<ϳгH JKpqt$JKpqtulm}~qrMN|yvs\]$%56-.12WX ѻ-tulm}~qrMNbc$bc    -.  ,-ab:;þ|ytqnkhH,-lm-.#$KLkl%&'c    -.  ,-ab$ $  $0$$b:;bc8 9   L!M!!!9":""$$$012`ac5o9     .!;!!!!!"&":"u"""/#O###$p$$%)%J%%%&& '7''$(((a))))))))))**3*4*U*V*Z*m*t***ÿCJ 5CJH*5CJ6CJ566CJCJ 6OJQJ 6hnH hnH OJQJ 0JOJQJjOJQJUjOJQJUFbc8 9   L!M!!!9":"""a#b##$$(%)%%%u&v& ' 'M'N'I(J()))))))¿zsn  h  t  v"#./hirabz{wx   klG)""a#b##$$(%)%%%u&v& ' 'M'N'I(J())))))) $$$$$$$$))))))***5*@*D*U*V*Z*b*c*d*e*k*l*m*t*{*|*}*~*********ÿzuqlgb]X                                                  8  ;  <  N  R  S  e"))))))***5*@*D*U*V*Z*\ $$$Y$$T֞cm %O)    p8$$ $$$$Z*b*c*d*e*k*l*m*t*{*|*}*~**tw $$$Y$$T֞cm %O)    p8$$$ $$$$$$ **************d~~wX$ $$$$$$ $$$Y$$T֞cm %O)    p8$$ *********++6+E+M+N+o+++++++++++, ,,,.,7,8,O,_,i,j,,,,,,,,,,,,,,,,,,,,,,,,,,- - --- -@-I-J-_-`-b-o-x-y------------E/F/G/g/5mH CJmH B*CJ5CJ j4CJCJ[*************************++++++++4+5+|wrmhc^Y            '  67  8  P  Q  R  S  [  fg  h  m  n  o  p  x                    "**************hww$$ $$$Y$$T֞cm %O)    p8$$$ $$$$ ******++++++++wY$$T֞cm %O)    p8$$$ $$$$$$ $$$ +4+5+6+E+F+M+O+P+m+n+o+++ww, $$$$$$ $$$Y$$T֞cm %O)    p8$$$ 5+6+E+F+M+O+P+m+n+o+++++++++++++++++++++++++,|wrmhc_Z  <=  >  D  F  G  O  P  VW  X  _  `  a  b  i                              "++++++++++++++w $$$Y$$T֞cm %O)    p8$$$ $$$$$$ ++++++++++, , , ,hw|$$$ $$$$$$ $$$Y$$T֞cm %O)    p8 , , , ,,,,,.,/,6,7,9,M,N,O,_,`,h,i,k,,,,,,,,,,,,,,,Ŀ{vqlhc^Y                                                    +  ,  -  .  6" ,,,,,.,/,6,7,9,M,N,O,_,ww$$ $$$Y$$T֞cm %O)    p8$$$ $$$$ _,`,h,i,k,,,,,,,,,,, $$$Y$$T֞cm %O)    p8$ $$$$$$,,,,,,,,,,,,,,,|$ $$$$$$ $$$Y$$T֞cm %O)    p8,,,,,,,,,,,,,,-- - ---- -@-A-I-K-L-_-a-b-o-p-x-z-{-{vrmhc^Y                            +  -  .  6  7  MN  P  Z  \  ]  d  e  lm  o  u  w  y",,,,-- - ---- -@-A-I-$$$$Y$$T֞cm %O)    p8 $$$$$I-K-L-_-a-b-o-p-x-z-{---$$Y$$T֞cm %O)    p8$$$ $$$$ {--------------------H..E/F/G/g//////////|ytojfa\W  a  c  gh  u       |    YZ  \  j  k  l  u  ~                    "--------------- $$$$$$$$$Y$$T֞cm %O)    p8----H..E/F/G/g//ulf`nn $ $$T4)#$Y$$T֞cm %O)    p8 $$$$$ g////////////////0000000000000!0"0'0)0+0-0K0L0N0i0k0m011O1Q11111111)2*2+282U2Y2w2z22222222233H3I3\3]3333333 4 43444:4B*OJQJhnH  B*hnH CJ 5H*mH 5mH 5H*5H*5CJ5T///////////////000$ PH$$lX y $$l+ y $$l y $$$///////00000000000#0%0'0(0)0K0i00011O111111+2-252½~{xupib    UVW01p  "  $  ()  +  -  01  3  5  BC  F  I  VW  Y  [  ^_%000000000#0%0'0(0)0K0i000 $$$l y $$l: y $$lv y $$$011O111111+2-25282U2Y2w2z22222222$$$ $ 5282U2Y2w2z2222222222233 333(3.343;3A3H3I3W3X3Y3Z3[3Ŀ{vrmhc^Y                      45  <  B  G  M  T  hi  q                       222233 333(3.343;3A3H3I3W3P$$j  f! $$$N$$Tֈ( ii"ioooooW3X3Y3Z3[3\3]3k3o3u3{3333333333333333$$$j  f! $$[3\3]3k3o3u3{33333333333333333333334 4 444 4&4|wrnid_Z  '  -  3  :;  B  H  O  U  Z  hi  o  u  {                                "33333334 4 444 4&4,434445464748494:4;4E4L4$$$j  f! $$&4,434445464748494:4;4E4L4S4Z4a4h4i4y445y555566$6'6(6P6T6|ungb]V        #  ,  .    3                                   !:4;4C4D4h4i4555556"6#6'6P6T6~6666666666 7 7=7>777777777778899:::::::::::;;;;;;;;;;;; < <<<=<v<ƷƷƷƷƷƷ 5CJH*5CJ5H*5CJ 6CJH*6CJ 5H*mH 5mH mH CJCJmH  B*H*hnH  B*hnH B*OJQJhnH HL4S4Z4a4h4i4y445y555566$6'6(6P6$$$ $ $$T4i"" $$ $$j  f! $$P6T6~6666666666677 7 $T$$O$$T4ֈ U-"$$$$T6~6666666666677 7 7"7'7,72787=7>7L7M7N7O7P7Q7R7`7e7k7q7þzuplgb]X                               %  :;  A  F  K  Q  W  kl              ! 7 7"7'7,72787=7>7MPO$$T4ֈ U-" $T$$O$$T4ֈ U-">7L7M7N7O7P7Q7R7`7e7k7q7w7|7}77777O$$T4ֈ U-" $T$$q7w7|7}777777777777777777777788 88888M8|wrmi` )a *  /  5  ;  @  E  LM  R  X  ^  d  i  wx  }                           7777777777777777778$O$$T4ֈ U-" $T$88 88888M8899a:$qe xx $$T4"" $  & Fa<$ O$$T4ֈ U-" $T$ M8899a:c:j:m::::::::;;C;K;O;q;y;z;;;;;;;;;;;;;;;;;<<<<< < < <<<$<*<0<6<<<=<N<T<\<b<i<p<v<w<<<<<<<<<<<<<<<<<<<<<== =========  a  ]a:c:j:m::::::::;;C;K;O;q;y;z;Y$$T֞5$!Y&g$$z;;;;;;;;;;;;;;;;;<<<<< < < <<T$$3 lX $ $$$<<$<*<0<6<<<=<N<T<\<b<i<p<v<w<<<<<<<<<<<$$$3 lX $ $$v<w<<<<< ======L=M=#>P>R>o>r>s>??@ @AABBBBCC D DNDODEE%F&FGGHHHHtIuIIIIIIIIIIIIIIIIŽŽ0JmH0J j0JU6j0JCJU j0JUCJCJmH CJ B*H*hnH  B*hnH B*OJQJhnH ><<<<<<<<<<<== ========== ='=/= $$$3 lX $ $$== ='=/=6===E=L=M=]== >Q>R>o>p>q>r>?@ABC DNDE%FGHHtIIIIIIIIIII     )/=6===E=L=M=]== >Q>R>o>p>q>r>?@A$xx xx $$T45Y&$ $$ $$3 lX $ $$ABC DNDE%FGHHtIIIIIIIIIIIh&`#$$ 0/ =!"#$%#0P/ =!"#$%DyK l.a.winters@sussex.ac.ukyK @mailto:l.a.winters@sussex.ac.uk?DyK /www.worldbank.org/reaserch/trade/archive.html/yK lhttp://www.worldbank.org/reaserch/trade/archive.html/ [0@0Normal CJmH <@< Heading 1$$@&CJ4@4 Heading 2 $@&CJ<@< Heading 3$$@&5CJ F@F Heading 4$$@&56CJ0mH <@< Heading 5$$@&CJ<@< Heading 6$$@&58@8 Heading 7 $$@&5CJ<@< Heading 8$$@&5CJ<A@<Default Paragraph Font<C@<Body Text Indent $, @,Footer  !&)@& Page Number>@"> Footnote Text CJmH 8&@18Footnote ReferenceH*>B@B> Body Text $ OJQJmH 0J@R0Subtitle mH 8>@b8Title$dh 5CJmH (U@q( Hyperlink>*B*,,Header  9r DOD 1Roman!$1$0H$7$8$ mH <S@<Body Text Indent 33CK~0igE #&H F  2 5 PE! 0*g/:4v<IS!+Hl<tcb")Z****+++ ,_,,,I---/002W33L4P6 7>778:;<</=AIi# Or|)*5+,,{-/52[3&4T6q7M8=I31`EXX !!"<D29*4""$$%%<'B'((j*z*7+@+A+J+,,,,---%-v0~01%1C1O1u1~1z3333446699::+;4;;;;;>>??­BEE$,԰ɰVVcc}}}}~Շш4<jqw}͓ԓlxgsal-;ưBLdl+3^l %($*:EDIOSPWp~ "DMck~  9ACGMV\`qw:@bk      !!##9#B#W#[#######r$z$%%%%&&+&4&,,),,,K,M,#.*...>00112#22266666666777777799::;;<R?Z???2@:@?@F@d@l@AAKBRBSB]BzEEEEENPbe(0,"4"''))--$013344788:::WEYE=HDHJJ|KKeLMPPSSZZ#[+[[[ccRf[fhhiivjjlm_vjv{*|m}t}ňovX9P NT_%(`c " 9*,MWSU '!9#K#$$--h1x1r:::;;>>9?@?@@FBlBDDEEEalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docalanwinG:\mode 4\wto\economics.docSpeir<\\Hudson6\Tsd\#Tsd\CTS\Mode 4\Speakers\Winters\economics.docWoodR;\\Hudson6\Tsd\#Tsd\CTS\Mode 4\website\Updated\economics.docWoodRG\\Hudson6\Tsd\#Tsd\CTS\Mode 4\website\Updated\symp_apr_02_winters_e.doc_ /{Ew2 vF='"޸m[ D˦z q  ME =;zFZ.*2U8 Fl9bn >vI oFKM9n6^5Q ={ j}{v'" 3:m[ %VLoQ w#J=;BXsX^yKQ%u"KW}lvMuyd-V^*RvQ )YA@6^3&f?!A9 z\DMF?Sy#W|Kv[}LApQ!neH/3fc< ZqGoQ#+m/ME H%H 4Si18rr'991k[&Q7k\8<;SE P@h 8^8`OJQJo(_@4Ep@GzTimes New Roman5Symbol3& zArialEMonotype Sorts9Garamond?5 zCourier New;Wingdings"qhMKdFefHDd&|-G)zt;!20di>4xTo date, the Temporary Movement of Natural Persons (TMNP) has defeated attempts to fit it into a robust analytical modelalanwinSpeirOh+'0$0@LXl |    yTo date, the Temporary Movement of Natural Persons (TMNP) has defeated attempts to fit it into a robust analytical modeleo dalanwinlanlan Normal.doteSpeir.d4eiMicrosoft Word 8.0a@F#@,@;@&|-G՜.+,D՜.+,t hp  UUniversity of Sussxexe)i>1 yTo date, the Temporary Movement of Natural Persons (TMNP) has defeated attempts to fit it into a robust analytical model Title(RZ _PID_GUID _PID_HLINKSAN{8AFC529A-47B0-11D6-9860-0060088181A6}A [6http://www.worldbank.org/reaserch/trade/archive.html/T= mailto:l.a.winters@sussex.ac.uk  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]_`abcdeghijklmrRoot Entry F}T]XtData 1Table8WordDocumentFSummaryInformation(^DocumentSummaryInformation8fCompObjjObjectPool]X]X  FMicrosoft Word Document MSWordDocWord.Document.89q