In the early pages of any elementary economics textbook, the beginning student sees a rudimentary plumbing-like diagram that illustrates the “circular flow” in every capitalist market economy. There is a business sector and a household sector, connected by two pipes. Business firms employ workers; and these businesses arrange to use the accumulated wealth of households to finance plant, equipment, inventories, and other productive assets. Through one of the pipes passes a flow of purchasing power from firms to households, in the form of wages, rent, interest, and dividends, payments for factors of production. Households buy the goods and services produced by the business sector; so a return flow of purchasing power goes from households to firms through the other pipe. Later on, the student learns how financial intermediaries, governments, and foreigners can be included in this picture.
The flow from business sector to households represents income, of course. In a schematic picture of the whole economy, it is natural to classify incomes by their functional types, as I just did. (There are borderline cases, often created deliberately to avoid taxes, but they are analytically insignificant.) Within the household sector, however, individual families can have income from several sources, and in widely varying amounts. For some purposes, like the study of poverty, one is more interested in annual family income arrayed from the smallest to the largest amounts.
You can make a “frequency distribution” of family incomes by size, just as you can make one of heights, or weights, or SAT scores. This is a curve or bar graph showing the proportion of families whose yearly income falls between $1,000 and $5,000, between $5,000 and $7,500, and so on. There is a characteristic difference, however. Heights and weights and test scores tend to fall along a symmetrical curve, the “bell curve” of song and story. The proportion of families whose income falls into successive size classes is almost always described by a curve whose bulk and peak lie over to the left (i.e., most incomes, including the commonest, are fairly small), with a long tail stretching far to the right, describing a relatively small number of very high incomes. (See Table A for an example from the US, adapted from the excellent and authoritative book by Frank Levy, The New Dollars and Dreams.1 )
Just about every capitalist market economy we know generates some very low, often inhumanely low, individual and family incomes. They are at the extreme left of the frequency distribution just described. We say that they fall below the “poverty line.” It is not a logical, or even an economic, necessity that the market process should generate a class of people with very low incomes, but it is a fact.
Obviously, in any year in any society, except in the improbable case of universal equality, someone must have the lowest income, and someone else the next lowest. That says nothing about poverty. The lowest observed income could be fairly high. In the US, standard practice is to calculate an absolute poverty line, an income below which one is “poor” and above which one is not. The calculation is not a miracle of precision; it is built up by pricing a minimal diet and applying rules of thumb to allow for other necessary expenditures. The current poverty line stands at a bit over $16,000 a year for a family of four.
This absolute standard has its problems. In 1952, median family income (in today’s dollars) was about $20,000. If one stuck to the absolute standard, one would have to conclude that almost half of all Americans lived in poverty, no matter that many of them felt themselves to be quite well off. And in another fifty years, when middle-class family incomes might be five or six times $16,000, one would have to say that the lowest-income segment of society, with an income of $20,000 in today’s dollars, say, is not to be thought of as poor, no matter that it is excluded from nearly everything that the society values as part of a decent life.
Most of the rich countries of the world escape this paradox by defining poverty in a relative sense. The median family income—with half of all families having lower income and half higher—surely has a good claim to represent the “typical” family income. Relative poverty is then usually defined by an income less than half the median. And “deep” poverty is sometimes defined by an income less than 40 percent of the median. That is what the book under review does; and it is what has to be done for international comparisons, because there is no good way to define equivalent absolute poverty lines in, say, Denmark and Bulgaria. Clearly the concept of relative poverty mixes together the notions of absolute poverty and gross inequality, but maybe that is appropriate. Most of us would want to describe as “poor” a child whose clothing and diet are regarded as shameful or laughable by ordinary children in the class, even if she gets just enough calories and clothing to survive. Median family income in the US is currently about $45,000. By the relative standard, families would qualify as very poor if their incomes were lower than $18,000, so our poverty line actually represents a lower standard of living than what a European would call deep poverty in the relative sense.
We collect income figures on an annual basis, for administrative reasons, but a year may be too short a period for social and economic analysis of poverty. Persistent poverty is particularly difficult to characterize; very low income has a different meaning if it belongs to a welfare mother of three, an impecunious surgical resident, or to Mike Doonesbury before his IPO. The book by Goodin et al. is able to work around that deficiency, as will be seen.
The persistence of incomes so low as to be incompatible—in the eyes of ordinary people—with full citizenship poses two different problems. Why does it happen? And what, if anything, should public policy do about it? The questions are not completely separable: your view about the appropriate policy will very likely depend on your beliefs about why some people are poor. The Real Worlds of Welfare Capitalism is about the policy question. The authors are, in order, professors of philosophy and political science at Australian National University and the University of Melbourne, professor of labor economics at Tilburg University in the Netherlands, and project manager at the Dutch Central Bureau of Statistics (a famous center of serious economic research, by the way). All have written before about one aspect or another of the welfare state, and theirs is a distinctive and important book.
It is special in two ways. The first is conceptual. The goal is to compare three quite different philosophical and institutional approaches to welfare capitalism: the “liberal” (in the European sense, meaning strongly individualist and free-market oriented); the “social democratic”; and the “corporatist.” These are exemplified by the United States, the Netherlands, and (West) Germany respectively. Admittedly the Netherlands is not the best example of a social-democratic regime; there are “corporatist” elements mixed in. Sweden, say, would have been a better choice, but appropriate data were not available to the authors.
The notion of “corporatism” is not very familiar in the English-speaking, Protestant world. The authors say:
Corporatists cherish, above all else, attachment to one’s community…. Communities…are composed of groups nested within groups…. The fundamental value for the corporatist is for an individual to be integrated into a group, which alongside other groups is integrated in turn into a larger community.
The primary social group is, of course, the family, often patriarchal; but churches and occupational groups, including unions, are also part of the communitarian picture, and are expected to be the main line of protection against poverty. (The notion of “faith-based” provision of social services has surfaced recently on the Republican side of political debate. This has a “corporatist” ring, although the motivation rests less on communitarian principles and more on suspicion of government. Nor would Republicans be happy with the inclusion of labor unions and professional groups among the “social partners,” to use the standard Ger-man phrase.) Corporatism is generally “conservative,” but with one notable difference. Competition between social groups—including workers and employers—is not the preferred mode of interaction. Cooperation is the ideal, under a rough rule of unanimity, so that no group is left out or defeated. The classic phrase in Germany used to be the “social market economy.”
The second source of distinctiveness in this book is that it rests its conclusions on three ten-year panel studies, one for each country. A panel is a fixed sample of individuals that is followed and measured over a period of time. It is not a series of snapshot samples that pick up different people at different moments, but a sustained series of observations on an unchanging sample group. The great advantage of the panel study is that it avoids the trap of designating as poor just those people who happen to be poor in the year they are observed. The data for the US cover the years between 1983 and 1992; those for the other countries refer to the years between 1985 and 1994. It is a pity that more recent data could not be included. Maybe there will be a later supplement. (Eventually the sample dwindles by attrition.)
The authors mean to compare the three welfare systems or “regimes” according to their success in promoting efficiency (meaning two things: getting money to those who need it and not to others, and doing so in a way that does not lower the economy’s productivity); reducing poverty; and promoting equality, integration, stability, and autonomy. Among these criteria, “integration” and “autonomy” are unfamiliar enough to warrant some explanation. Integration is the opposite of isolation or “social exclusion,” to use a common European locution. A person is well integrated if he or she has stable connections with appropriate social groups: family, co-workers, even society at large. Autonomy has something like its normal meaning: the wherewithal to make one’s own decisions about one’s life, subject to the normal limitations felt by everyone.
They comment briefly as well on the extent to which the welfare systems of the US, the Netherlands, and Germany actually embody the stated principles of the liberal, social-democratic, and corporatist approaches to policy. They are not shy about grading each of the systems on its success in achieving each of the significant goals.
Just to sketch a broad picture and fix magnitudes, I include Table B,
covering a much wider group of rich countries. It is taken not from Real Worlds but from a different source.2 The right-hand column shows for each country in 1991 the proportion of families who would have been in what I earlier called deep poverty, if they had paid no taxes and received no transfers. Crudely speaking, we can say that this is what “the market” produced. The left-hand column gives the deep-poverty rate after taxes and transfers; it shows what the combined tax-and-transfer system does for those at the bottom of the income distribution scale. (Note that these are one-year snapshots, not panel data.)
The numbers in the left-hand column are always considerably less than the right-hand numbers: the well-off do pay taxes, and the very poor do receive welfare benefits. It is interesting that eleven of the fifteen countries have pre-tax-and-transfer deep poverty rates that are approximately the same. Twentieth-century capitalism is pretty much of a piece. But only pretty much: market outcomes are much more egalitarian in Finland and Norway, and only slightly less so in Germany and Switzerland. (I do not know if any definitional or measurement biases are at work here.) The US does not stand out in the pre-tax-and-transfer column; market outcomes are not less egalitarian here than in Europe generally. What does catch the eye is that the US eliminates much less poverty than any other country in the list. The “post-tax-and-transfer” deep poverty rate in the US is more than twice as high as the average of the other fourteen countries.
Real Worlds uses different sources of data, so it does not exactly reproduce these results; but the general picture is the same. And the availability of panel data provides deeper insights:
In all three countries pre-government poverty is surprisingly high—in the neighbourhood of 20 per cent. Furthermore, it remains stubbornly high, at least over the medium term, reducing only slightly over a full decade….
Post-government poverty, however, varies dramatically across these countries. Even just on an annual basis, the proportion of the population of post-government poor in the US is on average around 18 per cent, whereas it is less than half that in Germany and less than half that again in the Netherlands. Significant though they are, those differences are magnified further still over time. Dutch poverty rates drop to around 1 per cent over a five-yearly period, whereas American rates remain around 15 per cent and German ones around 6. And whereas post-government poverty virtually disappears (dropping to 0.5 per cent) in the Netherlands over a ten-yearly period, it remains stubbornly stuck at just under 6 per cent in Germany and at 13 per cent in the US.
It is possible that the American welfare system is less successful at reducing poverty because it is more meticulous about limiting transfer payments to the poor, and therefore tends to miss out on some of them. This is a matter of the efficiency of welfare and other transfer programs. It is hard to hit a narrowly defined target; one way is to spray the whole area, but that is wasteful of ammunition. The whole economy may suffer, especially if revenue has to be raised by taxation that dulls and distorts incentives—for instance by taxing different kinds of capital, or different sorts of labor, or different consumer goods, at substantially different rates. There is indeed some of that. For means-tested and non-means-tested programs, and for both together, a larger fraction of total US transfer expenditure goes to the poor than is the case in Germany and the Netherlands. This transfer expenditure includes everything from child-care benefits to Medicaid to earned-income tax credits. (In 1992, 42 percent of spending on all trans-fer programs in the US went to the poor, as against 29 percent in the other countries. For means-tested programs, the contrast was much sharper: 86 percent in the US, much less elsewhere.)
Intense desire for programmatic efficiency cannot be the whole story. It does not and could not justify the apparent willingness of the US to accept twice as much poverty as Canada and almost three times as much as the Netherlands. Is it then possible that excess poverty is the price we pay for a successful and rapidly growing economy with high productivity? Could too much welfarism, and the taxation necessary to pay for it, dilute incentives enough to be a serious drag on a large scale? The authors take a cursory look at the overall performance of the three countries, and conclude otherwise:
At the end of the day, though, those intermediate inefficiencies do not really seem to matter in so far as the bottom line is economic growth and prosperity. All these welfare regimes seem to produce about the same sort of economic growth and prosperity for their citizens.
This is a very large question, probably not intelligently answerable on a broadly aggregative, comparative basis. A much more comprehensive analysis by a leading European student of the welfare state comes to a similarly agnostic conclusion:
The studies of the aggregate relationship between economic performance and the size of the welfare state…do not yield conclusive evidence. The results of econometric studies of the relationship between social transfer spending and growth rates are mixed: some find that high spending on social transfers leads to lower growth, others find the reverse. The largest of the estimated effects—in either direction—do not, however, seem believable.3
There is a piece of general wisdom here: when different methods applied to different data yield different results about the effect of X on Y, the chances are that the true effect is pretty small.
I can add a different sort of evidence that comes to a similar conclusion from the opposite direction. In a number of studies, the McKinsey Global Institute has compared employment and productivity in the US, Germany, France, and elsewhere, industry by industry, and has sought to account causally for the substantial differences that were found. The kinds of disincentives for which local welfare institutions might be responsible—for example unavailability of labor or lack of discipline—did not show up as significant causal factors.
This is a striking conclusion, and it should not be misunderstood. The welfare system does have an effect on the incentive to work, and to produce. It can and does contribute to dependency. People between the ages of sixteen and sixty-four put in more labor-hours in the US than in Germany and the Netherlands. There are no doubt several reasons for that, not just the relative stinginess of the American welfare system; but that stinginess is probably one of the reasons. The authors of Real Worlds then look a little closer, and find that the work-disincentives hypothesis is substantially confirmed in their panel data. Europeans would not be surprised: one of the purposes of welfare is to free some people, mothers of small children, for instance, from the need to work or work full-time, and that does in fact happen.
The striking conclusion, however, is that the nature and scale of the effects of the welfare system incentives are simply not enough to leave noticeable traces on the overall performance of the economies studied. It can matter which economies those are. Assar Lindbeck has found good reasons to believe that in Sweden the welfare state had expanded to the point where it—along with other special characteristics—had become a significant drag on economic performance.4 That result is not incompatible with the findings of Real Worlds, because Sweden was an extreme case.
In addition to the effects of different welfare systems on poverty and efficiency, the authors propose to measure the effects on equality, integration, stability, and autonomy. What of these?
The authors realize that social equality has important dimensions besides command over goods and services. But income is what their data tell them about, so income is what they can analyze. They find that the pre-government market economy produces similar degrees of income inequality in the three countries, perhaps slightly less in the Netherlands than in the other two. The tax-and-transfer system reduces inequality in all three countries, but considerably more so in Germany and the Netherlands. (This can be quantified, but it is hard to describe the intuition behind the numbers: the Gini co-efficient—the commonest numerical measure of inequality—is reduced by a sixth in the US, and by about a third in the other countries.)
The pattern is pretty much the same if one looks at incomes cumulated over two, three, five, or ten years. The availability of panel data reveals an interesting fact. Before it is affected by taxes and government transfers, income cumulated over several years becomes more equally distributed as the years progress. This reflects the fact that some low or high incomes in any single year are only temporarily low or high. In the US, taxes and transfers reduce one-year inequality and two-, three-, and ten-year inequality in the same proportion. In Germany the reduction is ever so slightly greater for ten-year inequality; in the Netherlands the reduction in ten-year income inequality is noticeably greater. On the natural presumption that long-term income is the more important indicator of economic well-being, that has to be scored in favor of the Dutch system.
“Social integration” is an even more complicated concept to deal with, and Goodin et al. are reduced to piecing together fragmentary indicators from their data. They measure integration into households by looking at divorce rates and at the frequency of adults living alone; they measure integration into the labor force by stability of employment; “shared economic fate” by the proportion of households that manage to make fairly large transitions through the income distribution; and “social exclusion” by the proportion of households that end up at the bottom in their incomes and access to employment and education.
The overall verdict here is mixed, as perhaps befits this ragtag collection of indicators. Household stability is strongest in the Netherlands and weakest in the US (though one can doubt that the welfare regime has a lot to do with this result). The US achieves greater year-by-year access to employment; the Netherlands’ record is a bit better when one looks at a decade as the time unit. According to the criterion of “shared economic fate,” essentially a measure of income mobility, the US comes out best, in the sense that a larger fraction of people experience a wide range of incomes; the authors comment that the German tax-transfer mechanism also does a good job in this respect. And Germany is best at avoiding “social exclusion.”
Under the “stability” heading, the authors look briefly at employment stability (measured by the steady provision of near-full-time jobs), where the US does rather better than the other countries. Then they turn to income stability (measured—inversely—by the variability of annual income over the ten-year period). If I understand what they do here, their measure is flawed and their results fundamentally uninterpretable, so I will not report them.5
The chapter on “autonomy” starts off in a woolly way, but eventually achieves some clarity. Autonomy is a good thing. We would like everyone, including welfare recipients, to make their own decisions, to feel as much as possible in charge of their own lives. It is the “as much as possible” that causes problems. Everyone’s actions are constrained; but which constraints should count as limitations on “autonomy”? Some rich people choose to work at a job. That is an autonomous decision; they could have done otherwise. If a larger fraction of poorer people hold jobs, we know that is because they cannot afford to do without wages. Shall we count that as a loss of autonomy? (Keep in mind that if nearly everyone chose not to work, there would be very little production to be shared around.) Whatever one means by autonomy in this broad sense, there is not much information about it in the data available to the authors.
In the end, Real Worlds finds desirable something it calls “combined resource autonomy.” You have it if you do not work more than forty hours a week and your income is still above the poverty line. That is to say, you are not poor, and you do not have to work excessively to stay out of poverty. This does not escape definitional problems: Why is it a loss of autonomy if I decide I would rather work an extra ten hours a week and have a bigger television set and a holiday at Catalina Island? Never mind: combined resource autonomy is a nice thing to have, especially for the welfare population. For a single year, about 91 percent of the Dutch sample achieve resource autonomy, as do 77 percent of the German, and 67 percent of the American. This measure is after taxes and transfers, and adjusted for family composition. Averaged over ten years, the percentages rise to 97, 87, and 71.
Perhaps more interestingly, 58 percent of the Dutch sample achieve combined resource autonomy in every one of the ten years; the corresponding figure is 48 percent for Germany and 34 percent for the US. Whether on a one-, five-, or ten-year basis, those percentages would be about fifteen percentage points lower in Germany and the Netherlands were it not for the effects of taxes and transfers. In the US, the contribution of the state is at most five percentage points. The authors conclude that “social democratic welfare regimes—at least of the sort represented by the Netherlands in our study—provide more people with both the time and the money that form some important (if not the only) preconditions of a truly autonomous life.”
The authors’ overall assessment of the three regimes reflects what they have found along the way, with few surprises. The liberal regime works as advertised. The market generates high incomes on average, especially at the upper end, and reasonable growth. Welfare provisions are tightly targeted and ungenerous, in fact so tightly targeted as to miss some of the poor. “The liberal welfare regime succeeds in keeping costs down,” they write, “but at the cost of allowing poverty to remain comparatively high.”
The social-democratic welfare regime, Dutch version, also works broadly as advertised. It makes substantial, redistributive transfers, and does more than the other systems by way of eliminating poverty and reducing inequality.
The corporatist regime comes off oddly, in the authors’ view. The German system does a good job of providing its people with stable and secure incomes. There is a considerable amount of redistribution through the tax-transfer system. What is odd is that the system does not work according to its self-image. Nongovernmental institutions, like the family and occupational groups, are not the central instrument. “Such stability as corporatists produce is achieved instead primarily through the highly stabilizing tax-transfer activities of its state welfare sector, which in corporatist theory ought be purely a subsidiary device.”
Americans feel themselves overburdened by a welfare system that is in fact, by the standards of other rich countries, both lean and mean. The relative generosity of the European welfare regimes may cost them some excess unemployment, but that does not explain why the rich European economies did not match the American boom of the 1990s. More important reasons are excessively tight monetary policy, failures of industrial competition, and restrictive controls on the labor market. The welfare state seems to encroach on the economy only when it grows to Swedish proportions. What really distinguishes the US is the equanimity with which the majority contemplates the poverty of a minority.
Russell Sage Foundation, 1998.↩
L. Kenworthy, "Do Social-Welfare Policies Reduce Poverty? A Cross-National Assessment," Working Paper No. 188, Luxembourg Income Study, 1998.↩
A.B. Atkinson, The Economic Consequences of Rolling Back the Welfare State (MIT Press, 1999), p. 184.↩
The Swedish Experiment (Stockholm: SNS Förlag, 1997).↩
The difficulty is easily described. Consider two ten-year income streams. The first is 2,2,2, ,2 ten times, i.e., the people involved may make, say, $20,000 each year; the second is 1,2,3, ,10. The first has zero variability, quite properly. The second would be accounted by Goodin et al. as quite variable (in fact exactly as variable as 10,1,9,2,8 ), whereas it strikes me as perfectly regular and therefore meaningfully "stable." Their procedure mixes up trend and "instability," and this could affect the comparisons significantly.↩
Russell Sage Foundation, 1998.↩
L. Kenworthy, “Do Social-Welfare Policies Reduce Poverty? A Cross-National Assessment,” Working Paper No. 188, Luxembourg Income Study, 1998.↩
A.B. Atkinson, The Economic Consequences of Rolling Back the Welfare State (MIT Press, 1999), p. 184.↩
The Swedish Experiment (Stockholm: SNS Förlag, 1997).↩
The difficulty is easily described. Consider two ten-year income streams. The first is 2,2,2, ,2 ten times, i.e., the people involved may make, say, $20,000 each year; the second is 1,2,3, ,10. The first has zero variability, quite properly. The second would be accounted by Goodin et al. as quite variable (in fact exactly as variable as 10,1,9,2,8 ), whereas it strikes me as perfectly regular and therefore meaningfully “stable.” Their procedure mixes up trend and “instability,” and this could affect the comparisons significantly.↩