The issue of inequality is one of the most important questions dividing the contemporary left. It highlights the antagonism between the more traditional left and the modernising social democrats who advocate Third Way politics. The critics see the Third Way as abandoning the core concerns of the left. Tony Blair and other left-of-centre modernisers, it is argued, accept globalisation as a given. Globalisation creates increasing inequalities, a widening chasm between the winners and losers. Third Way politicians have thrown in their lot with the winners and offer nothing to the losers: the rich get richer and the poor get poorer. Inequality, runs the argument, can be held at bay only if we shore up existing systems of welfare and the role of the state in limiting the polarising effects of markets.

The debate is being played out in many different countries. In the UK it pits the various ex-luminaries of Marxism Today against new Labour. On a broader European stage, it is at the root of the disaffection that Oskar Lafontaine, and many other members of the SPD, feel towards Gerhard Schroder.

I would question the two basic claims made by most on the old left, that inequality is everywhere on the increase and that globalisation is the main culprit. But I would also turn the inequality argument on its head. Only a Third Way approach can hope to counter the forms of inequality visible in the industrialised countries today. Only a Third Way approach can rediscover an effective role for public institutions and active government in promoting social justice. And to combat the inequalities we see around us, the old strategies are not only insufficient, but to some degree actively counterproductive.

Social democrats must revise not only their approach to, but also their concept of, equality. There is no future for the egalitarianism-at-all-costs that absorbed leftists for so long. The political philosopher Michael Walzer, writing in the winter 1998 issue of Dissent, has put the point very well: "Simple equality of that sort is the bad utopianism of the old left . . . political conflict and the competition for leadership always make for power inequalities and entrepreneurial activity always makes for economic inequalities . . . None of this can be prevented without endless tyrannical interventions in ordinary life. It was a historical mistake of large proportions, for which we [on the left] have paid heavily . . ."

Modernising social democrats have to find an approach that allows equality to co-exist with pluralism and life-style diversity. They must also recognise that classical liberals were right to see conflicts between freedom and equality. We should develop a more dynamic conception of equality, placing more stress on opportunity than the left has done in the past. Equality of opportunity, however, implies redistribution, mainly because one generation's equality of opportunity is the next generation's inequality of outcome. We should not deny to people the possibility of becoming wealthy (because to do so is to deny a motivator for exceptional talent) but we should recognise that, if wealth is stabilised in the hands of a few, opportunities for others will be reduced.

When we look at the statistics of inequality, we find a more complex picture than the old left position suggests. As measured by the usual statistics, inequalities of income have risen in most industrial countries, with the UK having one of the highest levels. However, in some countries - Italy, for example - income inequality appears to have gone down. The US is the most unequal of all developed societies in its income distribution. By some calculations, the proportion of people living in poverty in the US - defined as less than half of average income - is five times as high as Norway or Sweden, and three times as high as Germany.

Yet income inequality has started to diminish in America over the past four years, and the number of people below the poverty line has dropped. Blacks and Hispanics have improved their overall share of income by 14 per cent since 1994. There is some evidence, although it is disputed, that the circumstances of the poorest groups in Britain have also improved over the past couple of years or so. In several Continental countries, such as Germany, on the other hand, the proportion of people living in poverty has risen.

And while economic inequality, by and large, has been on the increase, other forms of inequality have not. In some respects the opposite is the case. In economic as well as social and cultural terms, women have become much more equal to men than they used to be. "Social egalitarianism", as measured in opinion surveys, has also increased. As Anne Phillips, professor of gender theory at the London School of Economics, puts it in her new book Which Equalities Matter? (Polity Press): "In my perception, people now care more rather than less about equality. They are more insistent on their standing as equals (what makes him think he is better than I? What makes her think she can tell me what to do?), less prepared to accept a subordinate position or believe everything the authorities say."

The old left blames increasing economic inequality on globalisation, especially the expansion of global free trade, which supposedly allows less skilled workers in one country to "undercut" those in another. Yet disparities of income have not grown only, or even primarily, in industries where trade is important. Technological change is almost certainly more significant. The spread of information technology creates a declining demand for less skilled workers, whose job opportunities and wages therefore decline. At the same time, qualified workers are able to increase their productivity and their earning power.

But that is not the whole story, or even most of it. One of the most systematic studies we have, from the US, suggests that most of the increased inequality there from 1990 to 1996 was due to other factors: the growth of two-earner families, the rise in the numbers of economically successful childless people and the growing value of capital assets, such as houses, shares and pension funds. According to the study, less than 30 per cent of the overall rise in inequality in the US since 1969 is accounted for by increased inequality in the earnings of working men.

Further, we need to consider recent research on poverty and the life cycle. The usual statistics about inequality and poverty are collected in the aggregate from year to year; they provide no information about what happens to individuals over time. Most approaches have assumed that poverty is a long-term condition. Even in-depth studies of individuals have nearly all been concerned with movement into poverty, rather than out of it.

Now data from a number of countries shows that, for many who experience it, poverty is not a permanent condition, demanding long-term social assistance programmes. A surprising number of people escape from poverty - but a greater number than used to be thought also experience poverty at some point in their lives. Using the definition of 50 per cent or under of average income, Lutz Leisering and Stephan Leibfried, authors of a new book on Time and Poverty in Western Welfare States (Cambridge University Press), found that over 30 per cent of West Germans were poor for at least one year between 1984 and 1994. This figure is three times the maximum number of poor in any one year. Those who moved out of poverty got, on average, to more than 30 per cent above the poverty line.

As John Hills and his colleagues at the LSE's Centre for Social Exclusion have shown, we have to be careful in interpreting these findings. They have an obvious appeal to those who want to dismiss poverty as an issue, or who look for an excuse to cut back on welfare systems. In the German research, more than half of those who escaped from poverty fell back below the poverty line for at least one year during the ten-year period studied.

Yet the implications of such research are considerable. The left has traditionally seen the poor as the victims of wider social forces. And there are indeed many who feel trapped by their life circumstances and can see no way of improving them. A poorly qualified man in his fifties who loses his job may have little chance of getting another one. Yet few even among the chronic poor just passively accept their life circumstances. Wherever possible, policy should build upon the action potential of the deprived. For instance, microcredit schemes (lending small sums, usually of a thousand dollars or less) may be more effective in allowing some people to make a fresh start than orthodox benefits.

Such research affirms the fundamental importance of dynamic labour markets that permit good access to jobs. Third Way politicians are right to place a prime emphasis on the labour market in welfare reforms, as has been done in the UK, US, Denmark and Holland. A society that has a high proportion of the population in work is able to release greater expenditure for healthcare, education and social protection for the vulnerable.

We have to see, moreover, that the state, including the welfare state, is not simply the solution to inequality, but can be part of the problem. The traditional leftist idea is that the market creates inequalities which it is up to the state to rectify. Yet the market, though it does create inequalities, can sometimes help to alleviate them - as welfare-to-work policies recognise. Conversely, even democratic and recognisably well-motivated states can generate inequalities or otherwise produce perverse outcomes. As the sociologist Claus Offe says, neo-liberal critics of the left "must be granted the point that excessive statism often inculcates dispositions of dependency, inactivity, rent-seeking, red tape, clientalism, authoritarianism, cynicism, fiscal irresponsibility, avoidance of accountability, lack of initiative and hostility to innovation, if not outright corruption - and so often on either side of the administration-client divide".

Social democrats have traditionally suggested a simple and morally compelling solution to inequality: take from the rich to give to the poor. Who could object to such a principle? I don't believe anyone on the left could or should do so, but to make the precept count is more complex than seems at first sight.

We have to decide first of all who "the rich" are. In the case of Bill Gates and other billionaires, this doesn't cause too many difficulties. So far as income tax is concerned, however, the category of "the rich" has to include large numbers of the "merely affluent" if it is to generate significant revenue and achieve substantial redistributive effects. "The rich" are no more of a homogeneous category than "the poor". Also wealth, like poverty, has to be considered in life-cycle terms. Some people become affluent early in life, others much later. People can acquire wealth, but then lose some or all of it. Wealth as much as poverty has many faces.

Wealth and inheritance taxes, plus incentives for philanthropy, must play a basic role in countering economic privilege. For some groups of the wealthy, other taxation strategies are just as relevant - such as those concerning stock options, tax loopholes and tax havens. But social democrats should rid themselves of the idea that most social problems can be resolved through increasing taxes to the greatest extent possible. In certain situations the reverse theorem applies - tax cuts can contribute to social justice. If carefully applied, tax cuts can increase supply-side investment, creating more profit and more disposable income. A bigger tax base is thereby created in the economy as a whole. Other tax-cutting strategies, such as the earned income tax credit pioneered by the New Democrats in the US, and adopted in modified form in the UK as the working families tax credit, can also be brought into play.

This emphasis is compatible with a continuing reliance on taxation as a means of redistribution. Social democrats need to sustain substantial tax revenues if public and welfare policies are to be funded and economic inequality kept under control. Progressive income tax needs to play a role in reducing inequalities, but it is neither sensible nor necessary to try to return to the steeply progressive systems of the past. In general, we should continue to move away from heavy reliance on taxes that might inhibit effort or enterprise, including income and corporate taxes. Seeking to build up the tax base through policies designed to maximise employment possibilities is a sensible approach - indeed, it is a key emphasis of Third Way politics. Social and economic policy can no longer be treated as if they were in separate compartments.

Finally, we should recognise that there are new dynamics of "social exclusion at the top" - the escape of elites from their fiscal and social responsibilities - just as there are at the bottom. A fundamental feature of the new social contract pioneered in Third Way politics is "no rights without responsibilities". This precept has to apply to everyone, rich as well as poor.

So this is how I would summarise a Third Way approach:

1. It takes a positive attitude towards globalisation, although not an uncritical one. Globalisation is not the prime source of new inequalities.

2. It concerns itself both with equality and pluralism, placing an emphasis on a dynamic model of egalitarianism.

3. It tries to respond to changing patterns of inequality. The poor today are not the same as the poor of the past: they include proportionately more children and single parents, for example, and proportionately fewer older people. Likewise, the rich are not the same as they used to be.

4. It accepts that existing welfare systems, and the broader structure of the state, are the source of problems, not only the means of resolving them.

5. It emphasises that social and economic policy are intrinsically connected. Social spending has to be assessed in terms of its consequences for the economy as a whole.

6. It places a stress on active welfare, coupled with labour market reform.

7. It concerns itself with mechanisms of exclusion at the bottom and the top. Redefining inequality in relation to exclusion at both levels is consistent with a dynamic conception of inequality.

The writer is director of the London School of Economics. This article is based on his Economic and Social Research Council annual lecture, given in London on 21 October