Should concerns about the poor stop trade liberalisation?* L Alan Winters** Openness to trade has long been seen as an important element of good economic policy, and trade liberalisation as a necessary step for achieving it. Continuing extreme poverty, on the other hand, is perhaps the biggest blot on the contemporary global economic canvas. This note summarises some research which tries to plot the links between the two and asks, if such links exist, what should we do about them. I argue that trade liberalisation is generally an ally in the fight against poverty: it tends to increase mean incomes, providing more resources with which to tackle poverty, and, while it will generally affect income distribution, it does not do so in a systematically adverse way. I also recognise, however, that most trade reforms will hurt some one, and that some may increase overall poverty even while they boost incomes in total. Thus despite the favourable view I take of liberalisation, I recognise that there is an important public policy issue to be addressed as it is contemplated and implemented. It is important to be clear about the criteria for assessing trade policy ­ indeed, any policy. There has never been, to my knowledge, a serious proposition that trade liberalisation can not possibly cause poverty. Thus finding an instance in which it does so does not constitute a refutation of an interesting hypothesis. Rather, the important positive issues are the empirical ones of whether trade geenrally creates poverty ­ on · The work underlying this note was originally prepared as background to the World Bank's World Development Report 2000/1 ­ Winters (2000a,b,c). I am grateful to the UK Department for International Development for financial support and encouragement of the original work, to Xavier Cirera for research assistance, Shoshana Ormonde for logistical help, to Tricia Feeney, Kate Jordan, Caroline Lequesne, Michael Lipton, Neil McCulloch, Andrew McKay, Pradeep Mehta, Chris Stevens, Sally-Ann Way, and Howard White for comments and advice, and to the authors of the companion papers in the study ­ CUTS (1999), McKay et al (1999) and Oxfam-IDS (1999) - for making them readily available. **School of Social Sciences, University of Sussex, Falmer, BRIGHTON, BN1 9SN, UK Telephone: +44 (0) 1273 877273, Fax: +44 (0) 1273 673563, e-mail l.a.winters@sussex.ac.uk. Also Centre for Economic Policy Research, London, and Centre for Economic Performance, London School of Economics which we are currently completely ill-equipped to rule ­ and under what circumstances it might do so in specific cases. The related normative issue is whether the fact that some poor person loses from a trade reform should be sufficient to condemn it. I believe strongly that it should not. Rather, the identification of hardship arising from a generally desirable policy reform should stimulate the search for complementary policies to minimise any adverse consequences and reduce the hurt that they cause. Rejecting any reform that adversely affected any poor person would be a recipe for long-run stagnation, which would ultimately increase, not reduce, poverty. If trade liberalisation and poverty were both easily measured, and if we had plenty of historical instances of both to observe, we might be able to derive simple empirical regularities linking the two. Unfortunately, however, this is not so, so we are thrown back on fragmentary evidence on parts of the argument. The key to interpreting this evidence, as well as to designing policies to alleviate any ill effects, is to identify the channels through which such effects might operate. Thus ultimately the research reported here is designed to identify the information that is required to say whether reforms will have poverty impacts and see what might be done to avoid any adverse outcomes. Households, families and individuals Poverty is a characteristic of households or individuals, and so this is place to start. Here we define poverty in terms of a minimum acceptable level of real income (or consumption), and measure it in terms of the numbers of people who live below this minimum and by how much. There are many other important dimensions to poverty that deserve consideration, although, in fact, most are pretty closely correlated with our income or consumption measure. Moreover, for international economic institutions such as the World Bank and government development agencies such as DfID, who are concerned with poverty in mnay places, absolute measures seem more appropriate than relative ones. 2 The best way of thinking about poor households is in terms of a generalised notion of a household which produces goods or services as well as sells its labour and consumes. An increase in the price of something the household sells (labour, good, service) increases its real income, while a decrease reduces it. Obviously the household's ability to adjust to such shocks affects the size of any impact it suffers, but it does not generally affect its sign. Poor households typically have several sources of income, including social and communal services, remittances from absent family members and income in kind as well as wages and profits from production. The analysis of poverty asks how trade liberalisation affects all of these and causes shifts between them, as well as considering the things that income is spent on. Some researchers have tried to decompose the effects of trade liberalisation on poverty into parts working via the growth of mean incomes and via income inequality. The former figures in our analysis below (although we recognise that many criticisms of liberalisation refer to immediate rather than long-run effects), but the latter does not. Poverty and inequality are different phenomena, with poverty referring only to the lower end of the income distribution. And if one is to work out how that end is affected by trade shocks, one needs to analyse prices and incomes in exactly the way described above. In short, one's view of the lower end of income inequality derives from poverty analysis, not vice versa. It is also important to realise that shocks to households can impinge differently on different family members. Thus, women might experience worse poverty than men when household incomes change. For example, if women have to start to work outside the home as well as continuing to bear all family responsibilities, they will quite possibly be worse off, especially if they do not personally collect the returns to their labour. Such a situation can arise if male wages or profits from production fall, but it can also occur in response to an increase in the extra-household female wage. Similarly, one needs to pay special attention to the impact of shocks on investments in child welfare, such as basic 3 education and health. Thus as well as asking about households, one needs to consider whether trade shocks could impinge disproportionately on particular household members. The prices of goods and services Once we have a view of how the poor earn and spend their incomes, the first question to ask is how a trade liberalisation affects prices. Even simple economies have several stages between the border, where trade policy operates, and the poor household, so one consideration is how much of any price change gets passed through to the poor. An x% change in the border price of a good typically translates into a significantly smaller change for the farmer or consumer because the costs of distribution remain unchanged. It can even get lost completely if distribution is monopolised, as, for example, with official purchasing boards. However, in the latter case, just abolishing the purchasing board is not sufficient to ensure pass-through; one needs to ensure that private monopolies do not replace the official one ­ see box 1. More important than price changes is whether markets exist at all: trade reform can both create and destroy markets. Extreme adverse poverty shocks are mostly associated with the disappearance of a market, while strong poverty alleviation can arise when markets are created for previously untraded or unavailable goods. A major question for governments is whether markets are created or destroyed and what might be done to support them. Governments should not subsidise market activity per se, however, but ensure through suitable intervention that functioning markets exist ­ by, for example, assisting small farmers to create the institutions to aggregate their sales or purchases into viable lot sizes. The effect of a trade reform on prices obviously depends on what the original policy was achieving ­ box 1. Thus the consequences of trade liberalisations can be described no more generally than can those of the trade policies they undo, and experience suggests that governments have a wide variety of objectives for trade policy. The effects might also depend on spillovers from one market to another. As households adjust to a shock in 4 one market they shift its effect to others. If the latter are particularly narrow, these spillovers can be very significant. A major attraction of liberalising agriculture is that the direct beneficiaries ­ farmers ­ appear to spend much of their extra income on goods and services provided by the poor such as construction, personal service and simple manufactures. 5 Box 1: Markets ­ Better, Worse and Missing Markets are critical in determining the poverty impacts of trade liberalisations. Where conditions for the poor have improved this has usually been associated with the better performance of and access to markets. Where they have worsened, faulty markets are generally to blame and in the extreme, the problem is often missing markets. Cotton in Zimbabwe: Before Zimbabwe's reforms in the early 1990s, the Cotton Marketing Board imposed low producer prices on farmers in order to subsidise the textile industry. Following deregulation and privatisation, there is now substantial competition between three buyers, one of which is owned by farmers themselves. These compete with each other not only on price (which has increased significantly), but also by providing extension and input services to smallholders. While the latter are obviously reflected in the prices that the farmers receive, their provision fills a gap that would otherwise exist in small farmers' access to inputs (including information). Hence, the changes have assisted small farmers both through an increase in price and by enabling them to produce more. Maize in Zambia: Such changes are precisely what the reforms in Zambia were intended to achieve in the maize market. But here the result was very different. Initially, pan-territorial pan-seasonal pricing subsidised remote and small farmers and the whole of agriculture was subsidised by mining. The latter subsidies have now been removed, and marketing has been taken over by two private firms which vary prices over time and space but which do not appear, in fact, to compete very actively with each other. As a result, small and remote farmers are unambiguously worse off, whilst larger and less remote ones are probably also less well off, since the subsidies from mining probably exceeded the tax in favour of remote areas. But the deterioration in the situation of remote farmers is substantially worse than implied just by the removal of pan-territorial pricing. For them, functioning markets have largely disappeared, 6 because the very poor infrastructure makes it unprofitable for the private buyers to buy from them. Worse, they say that they can not even revert to subsistence farming because ten years of cash-crop farming have eroded the necessary skills and seed stocks. Wages and Employment Trade reform is also likely to have major effects on wages and employment ­ of which those of the unskilled are the most important for poverty purposes. If reform boosts the demand for labour-intensive products, it boosts the demand for labour and then either or both of wages or employment will increase. However, whether this reduces poverty depends on whether the poor are strongly represented in the type of labour for which demand has risen. If the poor are mostly in completely unskilled families, while it is semi-skilled labour that receives the boost, poverty will be unaffected ­ or maybe, indeed, worsened since the gains of one factor of production will be balanced by the losses of at least one other. It also depends on where the wage rate is relative to the poverty line. If wages are pushed up from subsistence to higher levels, or if the sectors expanding their employment offer above poverty-line wages, then poverty will be alleviated. For countries that are relatively abundant in unskilled labour, trade liberalisation will usually boost demand for this factor and ­ depending on the above ­ help poverty. However, not all developing countries are actually unskilled labour-abundant. For example, many Latin American and some African countries have very strong endowments of mineral and agricultural resources and so liberalisation will stimulate these sectors rather than labour-intensive ones. Similarly, if the unskilled (illiterate) are primarily employed in non-traded sectors, while exports draw mainly on the semi-skilled (literate) labour force, the net effects on employment may favour the latter. India's recent reform may illustrate this case ­ box 2. 7 Box 2 Trade reform and the Indian Manufacturing Labour Force Formal manufacturing sector employment in India grew faster in the four years after liberalisation (taken to be 1991) than in the four years before, and wages more slowly: employment at 3.8% and 9.4% and wages at 8.1% and 7.0% respectively. With formal wages well above poverty levels, this seems good news for poverty. However, since the formal manufacturing sector accounts for only about 1.3% of the Indian work force it is hardly surprising that it could attract extra workers without incurring higher labour costs. Employment in informal manufacturing, on the other hand, declined by more than twice as much, especially in rural areas. If these data are to be believed, the most likely explanation is that the real depreciation that accompanied liberalisation (which will have raised the prices of traded relative to non-traded goods) switched output from non-tradable manufactures to tradables and that the former are disproportionate users of the informal sector. From a poverty perspective the important question is what happened to those who lost their informal jobs. If they could move back into subsistence agriculture or other informal jobs at approximately the same wage, not much happened to them in poverty terms, and the observed increase in formal jobs seems to offer a net gain. If, on the other hand, the loss of an informal job signals a descent (deeper) into poverty, the net effects of these changes is negative for poverty alleviation. Unfortunately we just do not know, but wages in informal manufacturing are quite often below poverty levels. Government Revenue and Spending Trade reform can affect government revenue, but much less frequently and adversely than is popularly imagined. Often, simultaneously reducing tariff rates and removing tariff exemptions actually increases revenue. Even where it does not (as eventually must be true as tariffs fall to zero), it is not inevitable that the poor suffer. It is ultimately a political decision whether the new taxes necessary to make up the shortfall or the cuts in government expenditure that result from falling revenue hit the poor. In recent years some East Asian countries have protected pro-poor expenditures in the face of far greater 8 shocks than any trade reform would produce. Since trade reform will typically raise aggregate incomes, it should generally be feasible to raise revenue elsewhere than from the poor. Binding tariff reductions at the WTO constrains governments from subsequently raising them again and WTO rules also limit the use of export subsidies and various other protectionist devices. Overall these constraints are likely to have pro-poor effects because ad hoc tariff increases and export subsidies usually favour the rich and powerful. WTO hardly constrains production subsidies in poor developing countries and has no influence over consumption subsidies; so in the rare cases where subsidies might help poverty alleviation, WTO rules are unlikely to prevent them. One criticism levelled at globalisation is that it makes it more difficult to tax mobile factors of production, particularly capital. To the extent that this is true, it is a systemic feature, which requires treatment at the global level. An individual developing country is unlikely to be able to influence its own ability to tax by varying its trade or foreign investment policies. It is true that high tariffs might attract inward investment in large developing countries, but only as a means of avoiding the tariff in the first place. Thus if companies respond to cuts in tariffs by sourcing their local sales from abroad, the loss of company taxation is really just another dimension of the loss of tariff revenue. Risk and Vulnerability A common worry is that opening up the economy will expose it to increased risk. Certainly, it will expose it to new risks, but very often the net effect will be to reduce overall risk because world markets (which have many players) are more stable than domestic ones. Moreover, if trade liberalisation allows producers to mix foreign and domestic risks more efficiently (because part of their output or input relies on foreign markets and part on domestic ones) they will experience lower variability overall because it is very unlikely that domestic and foreign markets will both be very good or both be very bad together. 9 Sometimes, however, openness will increase risk either because effective official stabilisation schemes are undermined or because residents switch completely from one activity to another that offers higher average rewards but greater variability. In the latter case the recorded incidence of poverty could increase even though the average incomes of the poor have increased and everyone expects to be better off because of the change. That is, in return for better average incomes, people are prepared to run a higher risk of very bad outcomes, and sometimes get them. But there is another side to this coin. One important feature of poverty is that the ability to bear risk is very low, because an adverse event has such dire consequences. Under these circumstances, the poor may forego opportunities to raise average incomes precisely because they can not bear the higher risk of failure that goes with them. Thus they might suffer the adverse effects of a reform - e.g. higher consumption prices ­ without the compensating benefits of higher average earnings and hence be losers overall. Often these problems are associated with failures in capital or insurance markets. The unwillingness of peasants in Zambia to buy agricultural inputs on the open market rather than have them supplied (at high cost) by a purchasing company against future output is possibly one example of this problem. An important concomitant policy to trade reform under these circumstances would be to ensure that such market failures are obviated, perhaps by establishing a safety net to assure the poor that there really is a minimum below which they will not be allowed to fall. At the level of the country it is widely believed that countries that are more open face greater variability, but in fact this is not fully established empirically. First, open countries do not, on average, have more volatile terms of trade. Second, although an open country suffers worse GDP volatility from any given degree of terms of trade volatility (because its GDP depends more heavily on trade), this effect might be offset by greater ability to smooth consumption in the face of income fluctuations or by more stable domestic policy-making. 10 Economic Growth The key to sustained poverty alleviation is economic growth. Although growth can be unequalising, it has to be very strongly so if it is not to reduce the numbers in absolute poverty, and there is little reason to fear that growth induced by freer trade will fall systematically into this class. The argument that openness stimulates long-run growth has a good deal of empirical support, but has still not been completely proven; there is, however, no evidence that it is harmful to growth. Overall, openness is probably an important component of a growth-promoting policy package, not least because it places constraints on the awfulness of other policies. Thus in the growth dimension, although policy-makers must try to confirm that in their case the growth effects of a liberalisation are benign, they should not generally expect the contrary. Adjustment Costs All the above discussion refers to permanent, long-run, effects. But since the gains from trade rely largely on adjusting a country's output bundle, there is also a possibility that people will suffer temporary adverse shocks in the process of adjustment. This is particularly true of workers who suffer spells of unemployment. The initially non-poor can generally tide themselves over these periods, so that public policy should focus on whether those who are poor initially suffer such temporary setbacks and on policies to help them get out of trouble as soon as possible. It is well to remember, however, that the non-poor (middle classes) are frequently far better at articulating their concerns than the poor: the volume of complaint is not, by itself, a good indicator of policy failure. Conclusion A trade reform should not be condemned merely because it causes some poverty. Ultimately a balance has to be struck between the number of people pulled out of poverty and the number pushed into it. If trade reforms do cause poverty, however, governments can act to alleviate much of it. Specific help for adjusting to trade shocks is usually not 11 feasible or desirable in developing countries. It is too difficult to identify trade as the cause of job displacement or to make the case politically or morally that jobs lost to trade shocks are more deserving of help than others. On the other hand, a general commitment to compensate every adverse shock in an economy undermines incentives and entails huge budgetary cost. The correct approach, therefore, is to establish general safety nets to protect against extreme poverty from all causes and pursue complementary policies to help firms and individuals realise their productive potential. The overall poverty impact of a trade liberalisation is the sum of many separate effects on different sets of people. It defies generalisations about outcomes and will quite clearly vary from case to case according to quite subtle details. It is possible, however, to suggest a number of questions that policy-makers can ask to try to get a feel for the likely balance and location of different effects. These are suggested in box 3. Box 3 A Checklist for Policy-Makers This research suggests eleven key questions for policy-makers trying to predict the poverty impact of a trade reform: · Will the effects of changed border prices be passed through to the rest of the economy? · Is reform likely to destroy effective markets or create them and allow poor consumers to obtain new goods? · Is it likely to affect different household members differently? · Will its spillovers be concentrated on areas / activities of relevance to the poor? · What factors are used intensively in the most affected sectors? What will be the mix of wage and employment effects? Will wages exceed poverty levels? · Will the reform actually affect government revenue strongly? · Will it lead to discontinuous switches in activities? If so, will the new activities be riskier than the old ones? · Does the reform depend upon, or affect, the ability of poor people to take risks? 12 · If the reform is broad and systemic, will any growth it stimulates be particularly unequalising? · Will the reform imply major shocks for particular localities? · Will transitional unemployment be concentrated on the poor? References CUTS (1999) Conditions Necessary for the Liberalisation of Trade and Investment to Reduce Poverty, Final report to DfID, August. McKay, A., Milner, C., Kedir, A., and Franco, S. (1999) "Trade, Technology and Poverty: the Linkages - A Review of the Literature". A Report to the International Economic Policy Department of DFID, CREDIT, University of Nottingham Oxfam-IDS (1999) Liberalisation and Poverty, Final Report to DfID, August, Winters, L.A. (2000a) `Trade Liberalisation and Poverty', Discussion Paper No. 7, Poverty Research Unit, Sussex, University of Sussex, UK. Winters, L.A. (2000b) `Trade, Trade Policy and Poverty: What are the Links?', Discussion Paper No. 2382, Centre for Economic Policy Research, London. Winters L A (2000c) `Trade and Poverty:Is There a Connection?' in ?? World Trade Organisation, June, 2000. 13