[Paleopsych] Mancur Olson: How Bright are the Northern Lights? Some Questions about Sweden

[Premise Checker]

Here's a long and good one. Enough for today.

Mancur Olson: How Bright are the Northern Lights? Some Questions about Sweden
Institute of Economic Research, Lund University, Sweden
http://www.mobergpublications.se/other/olson.htm
[Click this on to get the original PDF, the graphs, and better tables.]

Contents

Introduction

Preface

Chapter 1: Why Isn't Sweden Worse Off? 1
Is the Standard Answer to the First Question Wrong? 3
Are Time Lags the Answer? 6
Is Swedish Culture the Explanation? 7
Government Size and Economic Growth in the West 7
The Route to an Answer to the Second Question 14

Chapter 2: International Trade, Competitive Markets, and Economic Growth 16
Some Surprisingly Strong Statistical Relationships 18
The Historical Relationship Is Also Strong 23
Wider Evidence on Free Trade, Competitive Markets, and Growth 25
Why Does Protectionism Hurt Growth More than Welfare Does? 27

Chapter 3: Explicit, Implicit, and Efficient Redistribution 28
"Implicit" and "Explicit" Redistributions 28
The Criteria for Redistributions Generate the Social Costs 29
Slower Innovation as a Deadweight Loss 31
Conditions on Explicit Redistributions to the Poor 33
The Theory of Efficient Redistribution 33
We are not Done Yet 34

Chapter 4: "Rational Ignorance" and the Bias of Collective Action 36
The Difficulties of Collective Action 38
The Inegalitarian Bias of Collective Action 40

Chapter 5: Why Implicit and Inefficient Redistribution is Commonplace 42
Will Coalitions Seek Unconditional Cash Transfers? 44
Rational Ignorance Makes Implicit Redistributions Possible 45
The Implicit Redistributions that Rational Ignorance Permits are Almost Never 
46

Efficient Redistributions
Aggregate Evidence on the Inefficiency of Implicit Redistributions 49
The Salience of the Evidence in Chapter 2 50
Sudden Increases in the Size of the Market and the Polity that Determines 52

Trade Policy

Chapter 6: The Lower Costs and Ultimate Limits of Explicit Redistribution 54

What Limits the Amount of Implicit Redistribution in Sweden? 56

Encompassing Organizations 58
Factors Lowering the Costs of Explicit Redistributions
59 A Recapitulation
62 Too Much of A Good Thing is Bad: Nonlinearities and Lags
65 How Bright Are the Northern Lights? 68


lntroduction

The Swedish welfare state has for several decades been discussed as the "middle 
way" between capitalism and communism. The recent collapse of the communist 
economies of Eastern Europe has brought new fuel to the debate on the future of 
the Swedish Model.

Mancur Olson, professor of economics at the University of Maryland, is one of 
the pioneers in the development of public choice theory.

The argument presented in this volume has been developed from his 1986 Crafoord 
lecture in Lund. Starting from concepts proposed in earlier works - "The Logic 
of Collective Action" (1965) and "The Rise and Decline of Nations" (1982)1 - 
Mancur Olson develops a framework for analysing the distinctive set of policies 
and institutions that together form the Swedish welfare state.

He shows that different types of redistributions may have quite different 
effects on economic performance. Less transparent "implicit" redistributions, 
aiming to protect certain industries, seem to have more damaging effects on 
performance than "explicit" cash transfers to low income groups. The social 
cost of income redistributions depend on the conditions attached to them - i.e. 
on their effects on incentives.

Mancur Olson's analysis provides a number of concepts, distinctions and 
questions that together may help us better understand the long term dynamics of 
the Swedish Model.

Allan T Malm
Director
Institute of Economic Research

1 Also published in Swedish, "Nationers uppgång och fall", Ratio 1978.


Preface

This little book is not intended to argue either for or against the "Swedish 
model" of public policies and institutions. The purpose is rather to look at 
the somewhat distinctive set of public policies and institutions in Sweden from 
a new angle. When one looks at the Swedish situation from this unusual 
perspective, some new questions and options for public policy emerge. I asked 
many experts in Sweden and elsewhere to criticize an earlier draft of this 
brief book, and found that there was about as much sympathy for my argument on 
the one side of the political spectrum as on the other.

I had to ask for so much help because I am not any kind of expert on Sweden. 
The argument in this essay emerged as much out of observation of other 
countries and from theoretical reflection as it did from what little knowledge 
I have of the Swedish scene. In part because of the fear that my limited 
knowledge would lead me into error, I have also given myself a lot of time for 
second thoughts. As it happened, the argument here began when Professor Ingemar 
Ståhl kindly asked me to present the Holger Crafoord lecture at the University 
of Lund in the Fall of 1986. Had those whose help I have sought not been so 
generous with their time I would not have felt able to provide a written 
version of that talk that Professors Allan Malm and Ingemar Ståhl asked of me.

Thus any value this book has is due in large part to the help of my critics. 
Thick as my file of criticisms is, it is possible that some comment has been 
misplaced. But, at a minimum, I am greatly indebted for helpful criticisms to 
each of the following: Peter Bohm, Karl-Olof Faxén, Robert J. Flanagan, Anna 
Hedborg, Jörgen Holmquist, Sten Johansson, Jan Karlsson, Peter J. Katzenstein, 
Walter Korpi, Jan-Erik Lane, Assar Lindbeck, Carl Johan Ljungberg, Per-Martin 
Meyerson, R. M. Mitra, Victor A. Pestoff, Olof Ruin, Göran Therborn, Claudio 
Vedovato, and Carl-Johan Westholm. Chris Bartlett, Brendan Kennelly, Adele 
Krokes, Richard Lewis, Venka Macintyre, and Young Park provided me 
indispensable help in doing the research for this manuscript or in getting it 
prepared for publication. All of the shortcomings of this manuscript are 
nonetheless entirely my responsibility.

Postscript of 1 November, 1990:

This essay was entirely written before - and much of it long before - the 
economic emergency of October and the Swedish government's crisis package of 
reductions in public expenditure announced on October 26. The argument of this 
book (particularly in the closing chapter) leads one to expect crises such as 
the one that has just occurred - and to fear that there may be others as well. 
Nonetheless, I see no purpose in adding a discussion of the short-term problem 
arising from fears of a devaluation of the Swedish crown to the book. This book 
is about the more fundamental long-run characteristics of the Swedish economy, 
not about the emergencies or the successes of the season. In a long term 
perspective, the crisis of October, 1990, does not, taken by itself, appear to 
be so momentous: the possibility of a single devaluation would not raise fears 
about the long-run future of the Swedish economy if there were not uneasiness 
about a series of devaluations in the past and fears that a further devaluation 
would not be the last. Thus I believe that the focus should be on the 
fundamental structural issues with which this book deals - when they are 
understood, the way out of the short run difficulties will also be reasonably 
clear.


Chapter 1: Why Isn't Sweden Worse Off?

For more than half a century, Sweden has been known the whole world over for a 
distinctive set of policies and institutions - for the "middle way" between 
communism and capitalism or the "Swedish model".1 This Swedish system includes 
an unusually generous welfare state, but the country also has some other 
distinctive arrangements that, I shall argue, are no less significant.

Are the distinctive Swedish arrangements really northern lights that nations 
can use to get a rough sense of direction when they choose economic and social 
policies? We cannot say until we know how well the distinctive Swedish 
arrangements are working in comparison with the different arrangements in other 
countries. Thus to answer our question - "How bright are the northern lights?" 
- we must first ask, "How is Sweden doing?"

Sweden has, however, several distinctive institutions and policies, each of 
which has a different impact on the country's performance. Some may be lifting 
Sweden up at the same time that others are dragging it down, so that the 
overall performance may not reveal the actual value of any one of these 
distinctive arrangements. The inadequacy of an undifferentiated or monocausal 
approach becomes clear simply from breaking the familiar question about how 
Sweden is doing into two separate questions:

1) Why isn't the Swedish economy performing better than it is?
2) Why isn't the Swedish economy performing worse than it is?

Although these questions are obviously parallel, their answers are not. There 
is a standard, straightforward answer to the first question, but no familiar or 
obvious answer to the second.

1 Erik Lundberg, "The Rise and Fall of the Swedish Model," Journal of Economic 
Literature 23, 1985.

2

The answer to the first question is so familiar that I shall merely evoke it 
here: income transfers and the public sector have often been larger in Sweden 
than in any other noncommunist country, with the government spending and 
transferring at times more than three fifths of the nation's Gross Domestic 
Product; tax rates have accordingly been exceptionally high, with large 
proportions of the population having paid half, or even two-thirds or 
four-fifths of marginal income in taxes. On top of this, an extraordinarily 
powerful union, the LO, has often reduced wage differentials, thereby 
attenuating incentives and discouraging investment in skills. Robert Flanagan, 
a leading labor economist contributing to the Brookings Institution study of 
the Swedish economy, said, "When an outside economist first views the Swedish 
labor market, with its compressed wage differentials, comparatively high 
marginal tax rates, and numerous government-financed alternatives to work, the 
first reaction tends to be amazement that the labor market works at all."2 Thus 
the standard answer to my first question is not only well known, but emphatic 
as well.

This makes the second question - why Sweden is not worse off - all the more 
puzzling. Amazed as he was, Robert Flanagan concluded that Sweden "works." In 
an important sense, this is surely true. According to the best available 
measures, Sweden is one of the richer countries in the world, with a real per 
capita income that is as high as (or a trifle higher) than that of most nations 
in Western Europe. Why does it not have a lower per-capita income than 
countries with smaller public sectors, lower tax rates, and no comprehensive 
union engaged in economy-wide wage leveling?

Why, for example, is per capita income in Sweden about double the level in 
Ireland? In the countries to which both Irish and Swedes have migrated, their 
incomes are similar, but there is no question that the Swedish economy is 
vastly more productive than the Irish.

Similarly, why are per capita incomes now higher in Sweden than in Great 
Britain? They were for a long time much higher in Britain; in 1870 Sweden was a 
relatively poor country, but Britain then apparently had the highest per capita 
income in the world.3 Yet the situation is now reversed, even though, for a 
very long time, the public sector has been larger, marginal tax rates higher, 
and union wage-leveling incomparably stronger, in Sweden than in Great Britain.

There is another country that used to have a per capita income substantially in 
excess of Sweden's but is now way behind. That country is Argentina. Now that 
Argentina lags so far behind the nations of Western Europe, it is no longer 
usually compared with them, even though its population is almost entirely of 
European descent. Yet, in the early decades of our century, Argentina 
apparently had one of the ten highest per-capita incomes in the entire world, 
and for a considerable period surely had higher standards of living than 
Sweden. In the period during which Argentina has fallen behind,

2 "Efficiency and Equality in Swedish Labor Markets," in Barry Bosworth and 
Alice Rivlin, eds., The Swedish Economy (Washington, DC: The Brookings 
Institution, 1987), p. 172.

3 With the exception of resource rich and recently settled Australia.

3

its redistributions of income to low-income people and its welfare state have 
been incomparably smaller than Sweden's; it is even questionable whether 
Argentina (protectionist and interventionist as its government is) should be 
described as having a welfare state at all.

Many people around the world would say that the Swedish economy has not only 
outperformed the three economies that have just been discussed, but also been 
one of the most successful economies anywhere. That is a much stronger 
conclusion than is required for my argument. To be on the safe side, I take the 
lowest possible estimate of Sweden's economic achievements. Even then, we must 
ask why the Swedish economy works or survives at all, why Sweden incontestably 
remains one of the developed economies, and why it has unquestionably 
outperformed the three aforementioned comparison economies with less 
egalitarian policies.

Is the Standard Answer to the First Question Wrong?

Some people may suppose that the answer to the second question (why isn't 
Sweden worse off?) is simply that the standard answer to the first question 
(why isn't Sweden doing better?) is wrong: the fact that Sweden's per capita 
income is better than that of some countries with a smaller public sector, 
lower marginal tax rates, and less egalitarian unionism merely shows that 
individuals are not very sensitive to incentives and that the distortions in 
resource allocations that agitate critics of the Swedish welfare state are not, 
in fact, a significant problem.

I will argue in this essay that the familiar answer to the first question is to 
a large degree correct, but that limiting and offsetting factors have not been 
understood. These countervailing considerations not only help to explain why 
Sweden has a high level of per capita income, but also suggest that a society, 
with the right policies, can afford a better provision for its least fortunate 
citizens than those who understand the importance of incentives have realized.

The countervailing considerations that are the main concern of this essay can 
only be understood if one first appreciates the large element of truth in the 
familiar argument that Sweden's economic growth is slowed by its high taxes and 
subsidies and its often egalitarian wage policies. This familiar argument is 
not merely ideological rhetoric; it is generally accepted by the leading 
economists on the left and the right alike.

The professional consensus on this point is obscured in part because many 
economists understandably point out that some welfare state expenditures can 
increase a society's income. Subsidized education and training for low-income 
people who (because of capital-rationing or other market failures) would not 
obtain it by themselves can increase a nation's income. This point may be 
especially pertinent in countries such as Japan and the United States, where 
the welfare state is relatively smaller than in Sweden but where unusually 
large proportions of the population nonetheless receive

4

subsidized higher education. But it does not appear, at least at first glance, 
that any extra spending on the development of education and skills in Sweden is 
the main explanation of Sweden's relatively larger welfare state or of its 
economic achievements.

The professional consensus that very high levels of welfare state transfers 
reduce a country's income is also obscured because many economists favor income 
redistributions to the poor even though they believe such redistributions 
reduce income. There is no inconsistency in this position. A transfer to 
low-income people may increase total welfare or utility even though it reduces 
the measured income of the society. This is because, as is now known, the 
marginal utility of consumption systematically decreases with consumption4 and 
people with higher incomes tend to get less utility or satisfaction from 
marginal consumption than poorer people.5 Thus, if it is not pushed to the 
point where the adverse effect on incentives makes the incomes of even the poor 
lower than they would otherwise be, a redistribution of income can still 
increase human welfare even if it reduces measured per capita income.

There is also a technical complexity involving "income effects" that leads most 
observers to underestimate the loss of social efficiency arising from high 
taxes. Consider a worker who is taxed to pay for transfers to those who cannot 
work. The tax reduces the reward to the worker from an additional hour of work, 
but it also lowers his disposable income. Just as the first or "substitution 
effect" gives the worker a reason to work less, so the latter or income effect 
implies that he can afford less leisure and this makes him work more. The 
latter effect at least partially offsets the former effect, and this is one 
reason why the number of hours individuals work does not usually change 
dramatically with changes in tax rates.6

Unfortunately, high tax rates are harmful to economic efficiency even in those 
cases where the income effect completely offsets the substitution effect so 
that the hours worked are unchanged. This last point is perhaps 
counterintuitive and even some skilled economists need to be reminded of it. 
The reason why there is a loss of social efficiency from taxes even when they 
do not change hours of work at all is, in essence, because the taxed 
individual, though mindful that taxes mean he can afford less leisure, takes no

4 Martin J. Bailey, Mancur Olson, and Paul Wonnacott, "The Marginal Utility of 
Income Does not Increase: Borrowing, Lending, and Friedman-Savage Gambles," 
American Economic Review, vol. 7 (3) (June 1980), pp. 372-79.

5 See my "Why Some Welfare-State Redistribution to the Poor Is a Great Idea," 
in Public Choice and Liberty, Essays in Honour of Gordon Tullock, ed. Charles 
K. Rowley (Oxford: Basil Blackwell, 1986).

6 The impact of a tax increase on hours work also depends on other things, such 
as the uses to which the tax proceeds are put. If the tax receipts are mainly 
given as transfers to the same people who pay them, or if they are used to make 
leisure activities more attractive, this can lead to a far greater reduction in 
work than otherwise. See, for example, Assar Lindbeck, 'Tax Effects vs. Budget 
Effects on Labor Supply," Economic Inquiry vol. XX No.4, October 1982, pp. 
473-89, James Gwartney and Richard Stroup, "Labor Supply and Tax Rates: A 
Correction of the Record," American Economic Review, June 1983, 73, 446-51, 
Cecil E. Bohanon and T. Norman Van Cott, "Labor Supply and Tax Rates: 
'Comment"', and Firouz Gahvari, "Comment," in American Economic Review, Vol. 
76, No. l, March 1986, pp. 277-283.

5

account of the gain to those who benefit from the extra taxes he would pay if 
he worked more (as is shown in the footnote).7 Since the loss of efficiency 
from higher taxes, even when large, is not directly observable from the total 
number of hours of work done, it is often under-estimated.

In short, the second question of why Sweden is not doing worse than it is 
cannot be dismissed simply by claiming that the standard answer to the first 
question is wrong. The large Swedish welfare system may be morally appropriate 
and on balance desirable, but that does not change the fact that this system, 
together with wage-leveling, is making Sweden's measured economic performance 
less good than it would otherwise be. If we leave aside some interesting 
technical questions that are of no special pertinence here, this conclusion 
does not require much more than the finding that individuals and firms respond 
to changes in incentives in the ways that economic theory predicts.8 Sweden's 
relatively high per capita income, and its far faster growth during the period 
it has followed the "middle way" than less egalitarian countries like 
Argentina, Britain, and Ireland, is an anomaly that cries out for explanation.

7 Suppose an individual, with preferences between money income and leisure 
given by the indifference curves below, is confronted with a (proportional) 
income tax such that his post-tax wage rate falls from W1 to W2, and that none 
of the proceeds of the tax receipts are devoted to the consumption of the 
individual who pays the taxes. In the case depicted the reduction in the 
individual's post-tax income reduces the amount of leisure he chooses to 
consume by just enough to offset the reduction in the amount of work he chooses 
to do because the reward to him of an extra hour of work is decreased. But the 
adverse effect on economic efficiency is still there. When we take into account 
the value to others of the extra taxes this individual would pay if he worked 
more, we see that the total social value of his work can be depicted as the 
slope of line AB, parallel to Wl, so that, if the individual had an incentive 
to take account of the value of his tax payments to others, he would now take 
only Q1 rather than Q2 hours of leisure.

8 The important qualifications in the theory of the "second best," for example, 
will be ignored here.

6

Are Time Lags the Answer?

A comparison with the United States in recent years suggests a possible 
explanation. Although the American standard of consumption has remained high 
since 1980, it is by no means an adequate basis for judging the country's 
economic performance in this period. As is well known, the United States since 
the beginning of the Reagan administration has run large deficits in its 
government budget and in the current account of its balance of payments. When 
the debt to overseas borrowers is repaid, consumption will have to be lower 
than it would otherwise be. Therefore, the level of consumption leads to an 
overstatement of the current performance of the American economy.

Surely the Swedes have not yet received the full bill for their extravagances 
either. As Swedish critics of the country's welfare state have pointed out, 
habits of behavior do not change overnight, so that when incentives change 
their full impacts come with lags, which are a pervasive feature of economic 
life. Thus there can be no doubt that lags are relevant in answering our second 
question. A subsidiary theme of this book will be that the shape or functional 
form of the relationship between tax and subsidy levels and economic 
performance makes the "lag" explanation more significant for Sweden than it 
would otherwise be.

Yet, even if Sweden is overcome by an economic catastrophe before the ink is 
dry on this little book, we should still be puzzled about why the lags are so 
long. The impacts of changes in macroeconomic policies, in investment 
incentives, in exchange rates, and in tariff levels normally show up within a 
few quarters or a few years. Sweden has been famous for its "middle way" for 
more than half a century. Most of the lags that are evident in economics are 
not even a tenth as long as that. When we are comparing economic performance of 
different countries, it is the relative size of the public sector that is most 
relevant. If we take the period since 1951 (when national income statistics 
first came to be generally available) as a whole, both government consumption 
and total government outlays as a percentage of national income have been 
larger on average in Sweden than in any other country in the Organization for 
Economic Cooperation and Development.9 No doubt some of the effects of a 
general change in the pattern will show up only after generations have passed, 
but the main effects of most changes in economic policy show up within a few 
years, so we still have a scientific puzzle.

9 Total government outlays are defined as government consumption + social 
security transfers + subsidies + interest on public debt + gross capital 
formation + purchases of land and intangibles. Over much of the 1951-87 period, 
social security transfers have been less in Sweden than in several other 
Western countries.

7

Is Swedish Culture the Explanation?

There is a tendency to "explain" each country's performance by referring to 
allegedly unique traits of its people. Every culture and every people have some 
obviously distinctive characteristics, so it is always easy to claim that these 
characteristics account for a country's performance. But after sustained 
examination these claims usually turn out to be pseudo-explanations. Although 
references to national character are as common on one side of the political 
spectrum as the other, my experience suggests that, where our second question 
is concerned, this tendency is likely to take what might be called a 
"conservative-chauvinistic" form. Sweden's economic performance is as good as 
it is, some say, because the disincentives of the uniquely large welfare state 
and the egalitarian LO are countervailed by Swedish or Nordic cultural 
tendencies to work harder, to save more, and to be more enterprising than most 
other peoples would be if confronted with similarly oppressive incentives. 
Being of Nordic descent myself, I understand that this type of argument can 
have an emotional appeal. But it cannot survive careful examination.

How can favorable cultural traits, which surely do not change quickly, also 
explain the relatively low incomes in Sweden and in Scandinavia in the 
mid-nineteenth century and for some time before that? Those who oppose a large 
welfare state also need to explain how a superior national character could be 
consistent with what they regard as uniquely bad national decisions on this 
matter. And how can enduring national characteristics explain why so many 
countries - such as Argentina, Britain, Germany, and Japan - had 
extraordinarily good economic performance in some periods and extremely poor 
economic performance in others? All that we know about relative economic 
performance indicates that ad hoc cultural and racial explanations normally do 
not prove useful in the long run. We can be confident that we have found valid 
explanations only when we have parsimonious and general theories that explain 
performance in a wide variety of settings.10

Government Size and Economic Growth in the West

The same puzzle of why, in spite of the distortion of incentives from high 
taxes and welfare state subsidies, Sweden still has a high per-capita income is 
evident, in a less extreme form, in most of the other countries with relatively 
large welfare states. The data on the developed democracies in general show 
that, in the years since World War

10 See Chapter 1, on "The Standards Satisfactory Answers Must Meet," in my Rise 
and Decline of Nations (New Haven, Conn. and London: Yale University Press, 
1982), or in the Swedish translation published by Ratio in 1984.

8

II, the countries with relatively large welfare states have tended to grow 
about as rapidly as those with less egalitarian policies. To be sure, as later 
parts of this book will demonstrate, different types of government expenditures 
can have very different effects, so we cannot solve the puzzle simply by 
looking at the relative sizes of government outlays and rates of economic 
growth in different countries. In addition, definitive empirical findings about 
the relation between the size of the public sector and the rate of economic 
growth would require not only more data than now exist, but also a full scale 
econometric study rather than the merely illustrative and descriptive display 
of data that is offered here.

Still, it is instructive to examine the data displayed in Figure 1 and in Table 
1. These data reveal that, when we take all the years since 1951 to 1987 
together, there is at least no clear association between the percentage of a 
nation's resources spent or transferred by government and its rate of economic 
growth. Any association there may be is not strong enough be conspicuous. In 
the 1950s, as Figure 2 and Table 2 reveal, there was, if anything, a faint 
tendency for the countries with larger welfare states to grow faster. In this 
decade, Sweden, though well above average in total government disbursements as 
a percentage of GDP, was by no means the leader. The most notable country was 
West Germany, which was then undergoing an economic miracle and had, at the 
same time, one of the very highest levels of social security transfers and also 
one of the highest percentages of total government disbursements. These facts 
must surely alert us to the likelihood that something besides the disincentives 
of the welfare state has a big impact on rates of economic growth. The 
ideological arguments that focus exclusively on the size of welfare state 
redistributions to the poor are, at best, incomplete.

The data also reveal that there is definitely no general or lasting tendency 
for nations with larger public sectors to grow any faster than other countries. 
Indeed, the one country that has all along devoted the smallest proportion of 
its resources to government - Japan - grew almost as rapidly as West Germany 
during the 1950s, and much more rapidly than any other developed country over 
the postwar period as a whole. If we compare historical periods in which the 
roles of government were quite different, we get the same mixed picture. In the 
last half of the nineteenth century, almost all countries came closer to having 
laissez-faire policies than at any time before or since. This was also a time 
of unprecedented economic progress, so this period clearly supports the view 
that laissez-faire improves economic performance. The period from 1945 to 1970 
offers quite a contrast. During these years all of the developed democracies 
greatly expanded their welfare states and the public sectors became far larger 
than ever before. Yet this was also a time of extraordinary economic growth, 
surpassing even the glorious last half of the nineteenth century. This period 
is, accordingly, evidence supporting the view that big government favors 
economic growth. But the pattern changed again in the 1970s and 80s: the 
welfare state grew still larger and at the same time economic performance 
turned sour.

Thus the historical record for the developed democracies as a whole, like the

9

comparison across the developed democracies since World War II, produces no 
regular pattern: in some countries and in some periods a relatively large 
public sector is associated with poor economic performance, but in other 
countries and periods we observe exactly the opposite.

Figure 1

Unfortunately, the ideological and partisan debates have so obscured judgment 
that some scholars have nonetheless been tempted to draw unjustified 
conclusions from the experience of individual countries, or from data such as 
that presented in Figures 1 or 2, or from one or the other of the historical 
periods mentioned above. This type of mistake is made on both the left and the 
right. Some scholars of the former persuasion conclude from data or information 
such as I have set out that a government that aggressively intervenes in the 
market is favorable to, or at least not harmful to, economic growth. Some 
advocates on the right can find in the same sources allegedly clear evidence of 
the pernicious effect of a large welfare state on economic growth. Statistical 
results are in these situations sensitive to variations in specification and to 
the inclusion or exclusion of particular countries. Japan is particularly 
important; it is such an outlier that it can reverse the results in a 
regression equation. Although the data on Japan clearly favor the rightist view 
- that a smaller government yields faster growth - this one country's rapid 
postwar growth obviously could be due to other reasons, so regressions that are 
not statistically significant unless Japan is included are unreliable.

10

Table 1: Average Government Size and GDP Growth, 1951-87 (percent)

Annual GDP Growth
Government consumption
Social Security Transfers
Government Expenditure
Current Disbursement
Total outlays
(1960-87)

Australia 3.96a 14.2 6.6b 20.8 23.8b 29.8
Austria 3.91 15.2 15.3 30.5 34.5 44.0
Belgium 3.20d 14.5e 15.3e 29.7e 36.5e 42.2
Canada 4.32 17.1 8.4 25.5 31.1 37.1
Denmark 3.03 19.4 11.1 30.5 35.6 44.3
Finland 4.01 15.3 7.5 22.7 27.9 34.3
France 4.00 16.3 17.9 34.2 37.0 42.7
Ireland 2.92 15.3 9.2b 24.5 33.8b 41.8c
Italy 4.10f 13.4 12.7 26.1 33.0 39.8
Japan 6.69e 9.1f 6.4f 15.5f 18.4f 25.3g
Netherlands 3.44 15.5 18.6h 34.1 39.4 48.9
Norway 3.85f 16.4 11.0 27.5 35.5 42.5
Sweden 3.00 21.5 12.0 33.5 39.8 49.3
Switzerland 2.92 11.5 9.1 20.7 23.7 25.0
United Kingdom 2.40 18.8 8.8b 27.6 34.9b 41.0c
United States 3.24 18.0 7.5 25.5 28.7 32.0
West Germany 3.84 16.8 14.2 30.9 35.2 42.6

Note: Government consumption = government final consumption expenditure for 
goods and services. Social security = social security benefits for sickness, 
old age, family allowances, etc. + social assistance grants and unfunded 
employee welfare benefits paid by general government. Government expenditure = 
government consumption + social security transfers. Current disbursement = 
government expenditure + interest on public debt + subsidies. Total outlays = 
current disbursement + gross capital formation + purchases of land and 
intangibles. All data calculated and rounded. For average annual GDP growth 
rates, geometric averages are used.

a.1956-87. b.1951-86. c.1960-86.
d.1954-87. e.1953-87. f.1952-87.
g. Average of 1968-87
h. Average of 1951-59 and 1968-87

Sources.

1. OECD, OECD National Accounts Statistics, 1950-68.
2. OECD, Economic Outlook, Historical Statistics, 1989.
3. United Nations, Yearbook of National Accounts Statistics, 1957,1964

11

Figure 2

I do not, however, want to go so far as to conclude that there is no causal 
relationship between the role of government and economic growth. My point is 
rather that the foregoing data and historical facts do not, by themselves, 
allow us to come to any definitive conclusion yet about the size or even the 
direction of the influence of the

12

welfare state on economic growth. Any statistical analysis that considers only 
the variables in Figures 1 and 2 and thus omits many other factors besides the 
role of government that affect the rate of economic growth, can do little more 
than raise questions. This problem of "omitted variables" and other 
difficulties of "specification" indicate that the results of statistical tests 
on only the foregoing data are likely to be spurious. To underline the point 
that the variables that have been presented here are not sufficient to allow a 
properly specified statistical test, I do not even report the results of the 
regressions that I have run on the foregoing data - that might obscure the 
mainly heuristic purpose of the foregoing tables.

Table 2: Average Government Size and GDP Growth, 1951-60 (percent)

Annual GDP Growth
Government consumption
Social Security Transfers
Government Expenditure
Current disbursement

Australia 3.85a 10.0 5.2 15.2 15.8
Austria 5.66 12.7 9.9 22.6 25.0
Belgium 2.99b 12.0c 9.7c 21.7c 25.4c
Canada 3.95 14.2 6.5 20.7 23.8
Denmark 3.28 12.6 6.9 19.5 21.4
Finland 4.94 12.0 5.5 17.4 21.2
France 4.77 14.1 12.7 26.8 30.4
Ireland 1.75 12.5 6.2 18.7 24.7
Italy 5.54d 11.9 9.3 21.2 24.4
Japan 8.23c 10.0d 3.6d 13.6d 14.7d
Netherlands 4.66 14.1 8.4 22.5 26.5
Norway 3.25d 13.1 6.6 19.7 25.3
Sweden 3.42 16.5 7.2 23.6 26.3
Switzerland 4.3 11.1 5.8 16.8 19.4
United Kingdom 2.78 17.4 5.6 23.1 29.5
United States 3.23 18.2 4.1 22.3 24.3
West Germany 8.47 3.7 12.2 25.9 28.4

Note: For definitions and sources, see Table 1.

a. 1956-60. b. 1954-60. c. 1953-60. d. 1952-60.

This said, we must also remember that, if causal relationships are 
overwhelmingly

13

strong, there is usually little need to worry about omitted variables and other 
misspecifications. If we test whether being in a plane crash reduces life 
expectancy, we will probably get the qualitatively correct answer even if we 
don't take into account whether the dead had been cigarette smokers or not. 
Thus we can be pretty sure that the effect of the size of the welfare state on 
economic growth, even though it is the crux of modern ideological debate, is 
evidently not large enough to overwhelm other factors that affect economic 
growth in the developed democracies.

When we go beyond the advanced democracies and include evidence from the 
developing nations, we find that the data are poorer. They are also more 
difficult to interpret, since the level of development of a country constrains 
or influences the share of the government in GDP. It is nonetheless interesting 
that the data on the developed and developing nations together also do not 
reveal any clear tendency for the size or growth of government to be associated 
with economic performance. Some studies, such as Daniel Landau's,11 find that a 
relatively larger public sector makes for slower economic growth. Other 
studies, such as those of Richard Rubinson12 and Rati Ram,13 find that a 
relatively larger size of the government generates faster economic growth. When 
different specifications lead to exactly opposing results, we may suspect that 
there is really no compelling pattern in the data. That is also what we find 
from a glance at the data I have displayed in Figure 3, which compare 
government size and growth for 121 developing and developed non-communist 
countries.

The data in the three Figures in this chapter, the conflicting conclusions of 
the prior studies of government size and economic growth, and the broad 
historical facts that have been set out here are nonetheless suggestive. Since 
we know from the observation of the behavior of individuals that high levels of 
welfare spending and taxation have adverse effects on incentives, why doesn't 
the size of the welfare state show up in comparisons a cross countries and 
historical periods? This question underlines the need for an answer to my 
second question about why Sweden, in spite of the large distortions of 
incentives due to the huge welfare state and to LO wage-leveling policies, is 
not worse off; the puzzle about Sweden is also evident, in a less extreme form, 
in the comparisons of different countries and historical periods. Similarly, we 
know that all the Soviet-type economies - the ones where, in principle, the 
state runs everything - have failed miserably. Though it is wildly wrong to 
suppose that large welfare states are inherently semi-Soviet, the collapse of 
the Soviet-type societies nonetheless does raise our second question in a more 
casual and perhaps intuitive way. Why isn't the Sweden of the "middle way" - 
the society with the relatively largest public sector in the noncommunist world 
- also "mid-way" in per capita income between the Soviet-type

11 "Government Expenditure and Economic Growth: A Cross-country Study," 
Southern Economic Journal (1983), pp. 783-92.

12 "Dependency, Government Revenue, and Economic Growth, 1955-70," Studies in 
Comparative Institutional Development, 12 (Summer 1977), pp. 3-28.

13 "Government Size and Economic Growth: A New Framework and Some Evidence from 
Cross- Section and Time-Series Data," American Economic Review 76 (1986), pp. 
191-203.

14

countries and the developed economies with the relatively smallest public 
sectors?

Figure 3

I shall try in later chapters to provide a conceptual framework that helps 
resolve these puzzles. The foregoing data and historical facts have helped us 
see that we must search for some factors that both of the familiar ideologies 
omit. They have also shown us that the effects of a larger welfare state on 
economic growth, important as they must be, are by no means sufficient to 
overwhelm other factors.

The Route to an Answer to the Second Question

The next task is to contrast the inconclusive aggregate data about the welfare 
state and economic growth with some other salient facts about economic 
performance, and especially about international trade in manufactures by small 
countries such as Sweden. As we shall see, the other facts form a strong, clear 
pattern. This striking pattern suggests that competitive markets open to 
international competition are the main source of economic dynamism and that 
protectionism and most other forms of government regulation in such markets 
have monstrously harmful impacts on economic growth. The striking pattern we 
shall find in the next chapter stands in sharp contrast to the aggregate facts 
on the size of government expenditures and transfers and economic growth that 
we have seen in this chapter, both for Sweden and for the rest of the world.

15

The contrast between these dramatic results and the ambiguous aggregate data on 
the size of the welfare state and economic growth will, as we shall see, help 
us discover some new ideas. With the aid of these ideas, we will be able to 
find an answer to our second question, about why Sweden, given the undoubted 
distortions in incentives from its uniquely large welfare state, is not 
performing worse than it is.


Chapter 2: International Trade, Competitive Markets, and Economic Growth

Serendipitously, the unfolding of history in recent years has produced a 
surprising number of unintended or "natural" experiments that have generated 
the same kind of solid information about how the world works that we have been 
accustomed to getting from controlled experiments in the physical sciences. 
These inadvertent experiments have, for some reason, been brought to light only 
recently. The truths they reveal will (in conjunction with well-established 
results from more familiar sources) take us much of the way toward an answer to 
the question of why Sweden is as well off as it is.

The natural experiments involve international trade in manufactures by smaller 
countries and major alterations in the size of the countries or other 
jurisdictions that determine trade policies. I shall consider the lucky natural 
experiments carefully and then go on to discuss briefly how the findings from 
these experiments fit in with the more familiar sources of insight. To 
appreciate these experiments, we must first note that data on international 
trade in manufactured goods can yield insights that do not emerge so clearly 
from the study of trade in primary products and services.

Data on trade in manufactures are especially instructive because manufacturing 
is usually less dependent on natural resource endowments, and therefore a bit 
more sensitive to economic institutions and policies, than extractive industry 
is. Saudi Arabia and Iran export a great deal of oil, but this does not tell us 
very much about what policies or institutions these countries have nor offer a 
sound basis for judging the efficiency of resource allocation in them; the oil 
exports of these countries tell us more about their geology than about their 
economic and political systems. In the extractive industries generally, and 
even to a considerable extent in agriculture, the pattern of production and 
international trade is quite sensitive to the natural resource or climatic 
endowments of a country. The raw materials needed for manufactures can, by 
contrast,

17

be imported. Naturally, this often entails extra costs, but since 
transportation costs decrease as technology advances, the influence of 
endowments of natural resources on manufacturing is becoming smaller over time.

Although the location of service industries is often even less restricted by 
natural resources than manufacturing, the statistics and other information on 
services are poorer for services than for manufacturing. The pattern of exports 
of manufactures can accordingly often tell us more about what types of economic 
arrangements or systems are effective for economic development and growth than 
can other types of trade.

A thought-experiment will make it clear why data on the patterns of trade and 
the levels of protectionism of smaller countries is especially illuminating. 
Imagine a country so large that it was all the world except for Luxembourg. 
Suppose that this colossal country bad prohibitive tariffs on trade with the 
rest of the world, namely Luxembourg. Obviously, this hypothetical country 
could not be affected that much by its tariffs against Luxembourg, because even 
without any protectionism most of what it purchased would in any event have 
been purchased internally. The protection, in other words, would have affected 
only a relatively small number of markets to a minor degree and thus could not 
have had any great consequence for our gigantic hypothetical country. I 
therefore conclude that in looking at the effects of protection, it is 
essential to consider the size of the jurisdiction that has the protection and 
to note that the biggest economies, like the United States and Japan, are not 
affected by protection nearly as much as smaller countries are.

This consideration also suggests that previous studies of protectionism have 
given relatively too much attention to the height of tariffs and other forms of 
protection and too little attention to their mileage or length - to the extent 
to which they confine trade. In other words, too little attention has been 
given to the size of the jurisdictions with protection; if there are many small 
jurisdictions with protection, the total length and impact of protective 
barriers will be very great, but if the only protective barriers are those that 
surround a few huge countries or common markets, the impact of protection will 
be limited.

A focus on manufacturing in smaller countries will not only generate some 
special insights, but also tell us something about certain recent developments 
in the theory of international trade. These recent developments appear to 
qualify the presumption from economic theory in favor of free trade, and to 
suggest that, when there is imperfect competition and decreasing costs, a 
country can sometimes increase its welfare with tariff protection. It is in 
manufacturing industries in smaller countries that decreasing costs and 
imperfect competition will usually be most striking, so our results should also 
help us test the practical applicability of recent developments in 
international trade.

If one goes outside the economics profession, there is, of course, often no 
presumption in favor of free trade at all. Many people suppose that successful 
manufacturing usually requires protection. It is often argued, for example, 
that the protection of infant industries will in due course give a country a 
comparative advantage in manufacturing that it would not otherwise have had, 
and that the country may profit

18

significantly from taking advantage of the new pattern of comparative advantage 
that protection has given it. Certainly, comparative advantage is not something 
that is given and static, but something that is made or achieved. One way to 
develop a competitive manufacturing industry, it is often said, is to protect 
this sector so that there will be learning-by-doing which will eventually 
enable the industry to become competitive. The data presented in this book will 
make it possible to obtain a powerful test of these familiar ideas.

Some Surprisingly Strong Statistical Relationships

In a previous paper, I was able to present calculations,1 which I shall re-use 
here, of the percentages of the manufacturing output of various smaller 
countries that are exported. In other words, for all those small and 
medium-sized countries on which we found the needed data, I had the gross value 
of each country's manufacturing exports divided by the gross value of its 
manufacturing output. (I would have liked to divide the value added in 
manufactured exports by the value added in manufacturing in a country, but we 
did not find the data needed to do this.) So what is presented in Table 3 is 
the gross value of manufacturing exports divided by the gross value of 
manufacturing output. This provides, of course, the percentage of a country's 
manufacturing that it succeeds in exporting.

To ensure that the water is not muddied by countries so large that most of 
their trade would be internal trade even without any protection, I have 
excluded Italy and all larger developed industrial countries, and above all 
Japan and the United States. If countries are undeveloped and thus small in 
industrial terms, they are included no matter how large their populations might 
be. For most years, I do not have all of the needed data on the less developed 
countries. Fortunately, Bela Balassa and his associates at the World Bank made 
the needed estimates and calculations for various less developed countries for 
1973. Thus Table 3 includes almost all small or medium-sized developed 
democratic countries and those developing countries for which we have the 
Balassa data for 1973. Table 4 contains the indexes of levels of tariffs on 
manufactured goods in the developed democracies that I had previously published 
in The Rise and Decline of Nations.

There is a striking pattern: if the countries have high levels of protection on 
manufacturing, they export very little of their manufactures. Argentina, a 
country that is extraordinarily protectionist (and whose economic performance 
during the last half century, as we have noted, makes the growth of the Swedish 
economy look awesomely good by comparison) exports only about 2.5 percent of 
its manufactures. (I have rounded all numbers to the nearest 2.5 percent to 
underline the shortcomings of the data

1 Done for me by Kim Chohan, Alfred Forline, Michael Kendix, and Young Park.

19

Table 3: Comparisons of Export Percentages for Small, Medium-sized and 
Developing Countries, 1973

Manufactured exports/manufactured output
Exports of "true" manufacturesb/total exports
Exports of manufactures & processed primary productc/total exports

Argentina 2.5* 17 66
Australiaa 7.5* 11 57
Austria 32.5 53 97
Brazil 5.0* 16 54
Canada 20.0 36 72
Chile 2.5* 1 86
Colombia 7.5* 12 31
Denmark 42.5 42 90
Finland 27.5 30 97
Greece 12.5 22 71
India 7.5 44 62
Ireland 37.5 36 83
Israel 15.0* 27 47
Korea 40.0* 64 93
Mexico 5.0* 30 64
Netherlands 45.0 33 85
New Zealand 5.0 14 80
Norway 35.0 40 91
Portugal 27.5 48 89
Singapore 42.5 37 76
Spain 16.0 43 85
Sweden 37.5 52 95
Taiwan 50.0* na 85
Turkey 2.5* 13 34
Yugoslavia 17.5* 47 91

a Average of 1972, 1973 and 1974. b Manufactured exports include International 
Standard Industrial Classification sub-categories 32 (textiles), 38 (metal 
manufactures) and 39 (other manufactures).

c Manufactured exports include all processed primary products that are 
classified as manufactures in United Nations statistics.

Sources: For column 1. United Nations Statistics, except for the asterisked 
figures, which were obtained from Bela Balassa, of The World Bank, Washington, 
D.C.; for column 2. Yearbook of International Trade Statistics, United Nations, 
New York, for 1978 and 1979; and for column 3, the same two United Nations 
yearbooks plus Economic Daily News, Taipei, and Economic Yearbook of the 
Republic of China 1980, Taipei.

20

and the approximate character of the calculations.) So it was with other highly 
protectionist countries in 1973. Chile exported only 2.5 percent of its 
manufactures; Columbia only 7.5 percent; Greece only 12.5 percent; India only 
7.5 percent; Mexico only 5 percent; Turkey only 2.5 percent. Brazil (a 
questionable inclusion because of its large size) exported only 5 percent of 
its manufactures.

Table 4 indicates that, of the developed democratic countries, New Zealand is 
the most protectionist; it exported only 5 percent of its manufacturing 
production. Australia is the second most protectionist on manufactures of the 
developed democracies, and it exported only about 7.5 percent of its 
manufactured output. Many countries allocate a great deal of labor and other 
resources to manufacturing, but can sell only a small percentage of their 
manufactured output in the competitive world market.

Let us now look at countries of similar industrial size with relatively open 
policies. Austria, which is a member of the European Free Trade Association and 
has relatively low tariffs, exports about a third of its manufactures. Denmark, 
a country singularly lacking in natural resources for manufacturing, 
nonetheless exports 42.5 percent of its manufactured output. Korea, with 
relatively open policies on manufacturing by the standards of developing 
countries, exports 40 percent of its production of manufactures. Similarly, the 
other countries with relatively little industrial protection export a large 
part of their manufactures: the Netherlands, 45 percent; Norway, 35 percent; 
Portugal, 27.5 percent; Singapore, 42.5 percent; Taiwan, 50 percent. As might 
be expected from its relatively low level of protection of manufactures, Sweden 
exported a healthy 37.5 percent of it manufactures in 1973. Most significantly, 
for the small and medium-sized countries on which I have succeeded in obtaining 
data, there is not a single exception to the rule that the countries that 
protect manufactures least, export manufactures most.

Note how this strong finding contrasts with the lack of any clear pattern in 
any direction in the data on the size of the welfare state and economic growth.

At this point, a specialist in international economics, or any economist who 
remembers to think of the tendency toward general equilibrium of the economy as 
a whole, may say there is an obvious explanation of the foregoing results. The 
country that has high levels of protection for manufactures may export little 
of its manufactures simply because protection that reduces imports also reduces 
the amount of that country's currency that is supplied to buy foreign exchange, 
so the protection tends to raise the value of the country's currency and thus 
reduce its exports. Over the long run the imports and exports of a country tend 
to balance, so countries that don't import much also won't export much.

If protection of all kinds and exports of all kinds were at issue, this 
argument could explain the foregoing results. But it is only the protection and 
trade of industrial products that has been considered. Countries such as 
Austria, Switzerland, the Common Market nations, and the Scandinavian countries 
- Sweden included - are exceptionally protectionist where agriculture is 
concerned. Yet, relatively speaking, they are not very protectionist in 
manufacturing. The high protection of agriculture in these countries

21

Table 4: Average Levels of Industrial Tariffs (notes on the upper part of next 
page)

World weightsc
No trade weighing: a simple average
Own country import weighing:b
Import weights on BTN aggregatesd
Import weights on each BTN commoditye

1976 Ave.
Finalf Ave.
1976 Ave.
FinalAve.
1976 Ave.
Final Ave.
1976 Ave.
Final Ave.

Australia
Dutiableg 28.8 28.0 29.1 28.1 27.8 26.7 26.4 25.2
Totalh 16.9 16.5 15.4 15.1 13.3 12.8 13.0 12.6

New Zealand
Dutiable 31.4 28.3 28.6 25.5 33.0 30.4 30.2 27.5
Total 24.3 21.9 19.7 17.6 20.5 18.7 18.0 16.3

EEC
Dutiable 8.8 6.0 9.8 7.2 9.5 7.0 9.6 7.1
Total 8.0 5.5 6.3 4.6 7.0 5.2 6.9 5.1

United States
Dutiable 15.6 9.2 8.3 5.7 9.2 5.5 7.6 4.8
Total 14.8 8.8 6.2 4.3 7.1 4.1 5.6 3.5

Japani
Dutiable 8.1 6.2 6.9 4.9 8.0 5.7 7.9 5.5
Total 7.3 5.6 3.2 2.3 6.1 4.4 5.8 4.1

Canada
Dutiable 13.7 7.8 13.1 8.9 12.0 7.3 12.9 8.3
Total 12.0 6.8 10.1 6.8 8.9 5.5 9.4 6.1

Austria
Dutiable 14.2 9.8 18.8 14.5 15.9 12.0 17.0 13.3
Total 11.6 8.1 14.5 11.2 10.5 7.9 10.9 8.5

Finland
Dutiable 17.0 14.6 11.6 9.2 11.2 9.0 11.5 9.1
Total 14.3 12.3 8.2 6.5 6.7 5.3 6.7 5.3

Norway
Dutiable 11.1 8.2 10.5 8.0 10.2 7.4 10.0 7.5
Total 8.5 6.3 6.4 4.9 5.8 4.3 5.8 4.4

Sweden
Dutiable 7.8 6.1 7.7 5.9 7.4 5.3 7.1 5.2
Total 6.2 4.9 6.3 4.8 4.6 3.3 4.5 3.3

Switzerland
Dutiable 3.7 2.7 4.1 3.3 4.2 3.1 4.0 3.1
Total 3.7 2.7 4.0 3.2 3.3 2.4 3.2 2.4

22

a. An average of tariff levels on the assumption that all commodities are of 
equal significance; b. The relative weight attributed to each tariff is given 
by the imports of that commodity by that country; c. The significance of each 
tariff determined by world imports of the commodity, or aggregate of 
commodities, to which the tariff applies. World imports are the imports of the 
countries listed and the EEC. For Notes d. through i. and the sources, see The 
Rise and Decline of Nations. raises the value of their currencies and reduces 
the extent of their exports of manufactures. They nonetheless export a large 
percentage of their manufactures. Similarly, many of the countries with 
extraordinarily high levels of industrial protection, such as Argentina, 
nonetheless export a fair amount of primary products.

To obtain a more general test of whether something besides exchange-rate or 
"general equilibrium" effects is operating, I turn now to some calculations 
done partly for the Holger Crafoord lecture in Lund out of which this essay 
grows. These calculations are in columns 2 and 3 of Table 3. These columns 
provide alternative measures of the proportion of a country's exports that are 
manufactured or "processed" products. Though any statistical segregation of 
manufactures here is arbitrary, the middle column is probably the better 
measure of "true" manufactures. (Fortunately, the results are probably not very 
sensitive to the definition of manufactures as the two columns are positively 
correlated.)

These data show that there is a distinct (though not a very strong) tendency 
for the countries with high levels of protection of manufactures to have a 
relatively low percentage of exports that are manufactures. This suggests that 
the protection of manufactures may well discourage efficiency 
disproportionately in the manufacturing sector.2

There is still further evidence that the failure of those small and 
medium-sized countries that lavishly protect manufactures to develop profitable 
manufacturing export industries is not due only to exchange rate effects. A 
country that changes from relatively open policies to high protection of 
manufactures may actually reduce the rate of growth of manufacturing output for 
domestic as well as international use. For example, after 1930 and especially 
under the regime of Juan Peron, Argentina increased

2 One other factor probably helps to explain the limited proportion of exports 
that are manufactures in countries that are very protective of manufacturing. 
The supply curves of many primary product industries may be relatively 
inelastic, so some of the enterprises in these industries will be able to 
produce some output at modest costs even when the country's institutions are 
not efficient. Countries with some exceptionally good mines or oil wells may be 
expected to export same of the yield of their natural resources even if the 
whole economy is badly organized. The Soviet Union fails to sell much in the 
way of manufactured goods in free foreign markets and has even lost the large 
agricultural exports it had in czarist times; it does nonetheless export 
relatively large amounts of the prodution of its mines and wells and thus may 
illustrate this point. (I am thankful to Christopher Clague for calling this 
point to my attention.)

23

its protection of manufactures to a colossal level and systematically exploited 
its agricultural export industries. As an authoritative study concludes, "the 
most ironic lesson of postwar Argentine experience is that if there had been 
less discrimination against exports, manufacturing expansion would have been 
greater. Indeed, the annual growth rate of manufacturing during 1900-29 (5.6 
percent) was higher than during 192965 (3.7 percent)."3

The Historical Relationship Is Also Strong

I have argued above that protection has a much greater impact in smaller 
countries than large ones and we found in the data on smaller countries a 
strong relationship between relative openness to imports of manufactures and 
success in manufacturing. We can corroborate or refute the foregoing results by 
looking at the historical evidence on the consequences of great increases in 
the size of countries or jurisdictions with trade barriers. If protectionism 
has a much greater impact on smaller jurisdictions, and if this impact is 
overwhelmingly harmful, then we should expect that dramatic increases in the 
size of a protectionist jurisdiction would greatly reduce the damage done by 
the protection. This is a question that I examined in The Rise and Decline of 
Nations, which offers theoretical reasons why a sudden and substantial increase 
in the size of a protectionist jurisdiction should stimulate rapid economic 
development. To facilitate a comparison of the historical evidence on the 
impact of protectionism with the cross- country evidence that has just been 
presented, I casually summarize here my evidence from Rise and Decline on 
whether the great periods of freeing of trade - by the method of increasing the 
size of the jurisdiction with protection - have in fact been associated with 
rapid economic development.

Though the explicit efforts to reduce tariffs and quotas have received more 
attention, it appears that the quantitatively most important freeing up of 
trade has, in fact, occurred when larger jurisdictions have been created and 
the mileage or length of protection thereby reduced. The most notable reduction 
in the length of tariff barriers in recent times was brought about by the 
creation of the Common Market in Europe through the Treaty of Rome in 1957. 
What happened through the Common Market has happened many times in history, 
usually through national unification that, often inadvertently, freed trade by 
creating a far larger market in which, even if there high protection around the 
newly unified country, there were no barriers to internal trade.

In the 1830s in Germany, for example, a Zollverein or customs union was 
created, and gradually extended and deepened, until it culminated in the German 
Reich that was completed in 1871. It is interesting that most of the 
German-speaking areas of Europe

3 Carlos Diaz Alejandro, Essays on the Economic History of the Argentine 
Republic (New Haven & London: Yale University Press, 1970), page 138; see also 
pages 126, 139-40, 252, 259-60, and 271-72.

24

were relatively poor in the period before the Zollverein and the German Reich 
were created. In the eighteenth and early nineteenth centuries, Germany was far 
poorer than Britain and the Netherlands, and probably also had a distinctly 
lower income than France. Nonetheless, in the second half of the nineteenth 
century and in the years up to World War I, the German economy grew at an 
extraordinary rate, so that by World War I Germany was undoubtedly one of the 
greatest industrial powers of the world.

I call phenomena such as the creation of the Common Market and German 
unification examples of "jurisdictional integration": such integration occurs 
whenever a much bigger jurisdiction is created that has internal free trade.

Japan offers another example of jurisdictional integration. Before the Meiji 
Restoration of 1867-68, Japan was divided into nearly three hundred separate 
feudal domains, each under its own feudal lord or "daimyo." Normally each of 
these domains had high levels of protection, limiting trade from that 
jurisdiction to other parts of the Japanese Archipelago. To the extent the 
Shogunate had some control over the whole of Japan, it used that control in 
part to make Japan as a whole virtually autarchic with the rest of the world, 
limiting trade and factor mobility with the outside world to a negligible 
level; even travel abroad was punishable by death.

The Meiji Restoration (or revolution) of 1867-68 created a free trade area 
within Japan. It eliminated the separate feudal jurisdictions and thus also the 
trade restrictions that went with them. At about the same time, a group of 
Western powers forced on Japan the "humiliating treaties." These treaties are 
described as "humiliating" because the Japanese were too weak to prevent their 
imposition. One of these treaties prohibited Japan from having any protective 
tariffs; for fifty years the country could have nothing more than tariffs for 
revenue only at rates of 5 percent or less. Because of the jurisdictional 
integration plus the "humiliating treaties," Japan experienced an increase in 
freedom of trade.

Japan was a poor and underdeveloped country before this process occurred. Some 
Western observers believed that the Japanese would never be able to manage 
modern economic life. Yet, not long after the Meiji Restoration, Japan began to 
grow very rapidly. One symptom of that growth, besides the evidence from the 
statistics, is that by 1904-05 Japan was already powerful enough to defeat 
Russia in a war.

At the end of the eighteenth century there was another example of 
jurisdictional integration - the United States. At the time of the Declaration 
of Independence in 1776, and for several years after, the thirteen ex-colonies 
were virtually independent countries. The U.S. government was created only in 
1789, when the U.S. Constitution went into effect. The Constitution outlawed 
the tariffs that some states, such as New York, had imposed against imports 
from other states. So the United States then became a substantial market in 
which, internally, there has been free trade to this day. To be sure, through 
most of the nineteenth century and until the 1930s, the United States was a 
highly protectionist country. Nonetheless, because of the absence of tariffs by 
states and the great growth of the U.S. over the nineteenth century, the U.S. 
has enjoyed a large and growing unrestricted internal market.

25

We see much the same phenomenon when we go back to Holland in the 17th century. 
When the United Provinces rebelled against Spain, they created an area in 
which, generally speaking, there was internal free trade. Although the 
Netherlands was not large by the standards of countries today, by the feudal 
standards of the time it was reasonably substantial. Moreover, its location and 
flat topography - much of it below sea level - meant that it was uniquely 
suited to canals, so that it had an exceptionally large area that was 
accessible to water-borne transportation. Before long, Holland entered its 
"Golden Age" and became the world's leader in economic development.

If we go back still further to the end of the Middle Ages, we find that the 
first country in Europe to establish true unification was England, or more 
precisely, England and Wales. By the sixteenth century, the parochial feudal 
system had been all but abolished in England. Some time later Scotland was 
conquered and all of Great Britain was essentially one free market. The 
semiautonomous towns and feudal fiefs with their separate trade restrictions 
were made part of a unified Britain. Though the textbooks call this a 
mercantilistic period and correctly emphasize that there were high national 
tariffs, there was nevertheless a great freeing of trade because of the 
increase in the size of the jurisdictions that restricted trade. This was also 
the period of the Commercial Revolution and of substantial economic progress, 
which was soon to be interrupted by the English civil wars of the nineteenth 
century. After this unstable and revolutionary period ended, Britain became the 
location for the epochal economic progress of the Industrial Revolution.

Thus in several periods of history protection has been dramatically reduced 
simply because a big market replaced many small protected markets. Even though 
the big markets were sometimes highly protected, there was a great freeing of 
trade, which was followed in every case by rapid economic development.

Wider Evidence on Free Trade, Competitive Markets, and Growth

I have emphasized the foregoing evidence on the value of wide and unprotected 
markets as seedbeds of economic growth because it is a new and different type 
of evidence. But more familiar types of evidence point in the same direction. 
This is not the place to go over the massive literature on international trade, 
but it may be useful to refer briefly to the discovery in recent years, by 
leading specialists on international trade and economic development, that trade 
policy has incomparably more importance for the growth of the developing 
countries than economists previously realized.

As I see it, the discovery arose because most of the less-developed countries 
have chosen levels of protection of manufactures that are vastly higher - often 
ten or twenty times higher - than those in the developed democracies, while a 
minority of the developing societies have, by contrast, turned to "outward- 
looking" policies. The performance of these less protectionist societies has 
been incomparably better than that

26

of the societies with "inward looking" policies. This is evident not only from 
many careful studies of selected sets of countries, but also from the 
systematic examination of essentially all of the developing countries on which 
there is usable data. In a study summarized in the World Development Report for 
1987,4 the World Bank examined 41 developing countries, which it classified 
according to the extent they were protectionist and inward looking, on the one 
hand, or approached relative neutrality in their treatment of imports and 
exports, on the other.

Notwithstanding the great importance of other factors, the less protectionist 
or more outward looking countries grew far more rapidly than the more inward 
looking, and usually did better by other measures of economic performance as 
well. As is well known, Hong Kong, Korea, and Singapore were distinguished both 
by their lesser use of protection and also by their rapid economic growth, but 
there was a similar if less marked variation in the degree of protection and in 
economic performance across the whole set of countries. Though it has not 
achieved anything like the per capita income of the three countries just named, 
Korea is nonetheless classified with the "gang of four" most successful 
developing countries. Some observers of Korea argue that it has had significant 
levels of protection and also that the country has by no means had a policy of 
laissez faire. There is evidence for the latter argument, but the high level of 
both imports and exports in Korea makes it clear that it has been, at least by 
the standards of most developing countries, definitely an outward-looking and 
relatively open country. In any event, a change in the classification of only 
one or a few countries would by no means eliminate the strong association 
between an outward orientation and faster growth found in data analyzed by the 
World Bank. Thus the evidence that the less protectionist developing countries 
have tended to have far better economic performance than more protectionist 
countries cannot be dismissed.

Another kind of evidence that is worth singling out is the record of the firms 
and industries in advanced countries that have been especially impressive in 
international competition. Michael E. Porter and his many associates have done 
a large scale, ten- nation study of many such firms and industries and the 
results have been published in Porter's book, The Competitive Advantage of 
Nations.5 That book examines an almost endless number of specific cases with a 
lot of convincing detail showing that the winning firms and industries in 
international competition have systematically been bred in environments in 
which there was vigorous domestic as well as international competition. In 
contrast, protectionism, cartelization, and subsidization have systematically 
failed to produce internationally impressive firms and industries. (An 
apparently disproportionate number of Porter's examples of successful firms and 
industries are Swedish, and we shall later see that this is not surprising in 
the light of my argument here.) The successful firms and industries are also 
regularly nourished and driven by what Porter calls a "cluster" of symbiotic 
and motivating activities - by a wide

4 Oxford University Press, 1987; see especially chapter 5. pp. 78-94.

5 New York: The Free Press, 1990.

27

array of competitive suppliers of inputs in the form of intermediate goods, by 
labor with the necessary specialized skills, and by pertinent and high quality 
research and education, and by demanding consumers with good alternatives. The 
combination of vigorous competition and unrestricted access to a vast variety 
of inputs generate the continued innovation and ever-increasing efficiency that 
are required for international competitive advantage.6 I conclude that there is 
no way that a small country with a high level of protection could have either 
the vigorous domestic competition or the symbiotic competitive cluster of 
activities and readily available inputs that are needed for success in 
international competition. Thus Porter's book provides yet another kind of 
evidence that uninhibited trade and large markets are decisively important for 
economic progress.

Still other kinds of evidence from all over the world (including eastern 
Europe) point in the same direction, but the time has come for us to relate the 
strong findings in this chapter to the ambiguous results in Chapter 1.

Why Does Protectionism Hurt Growth More than Welfare Does?

There is a puzzling contrast between the strong statistical and historical 
patterns suggesting that protectionism is extraordinarily damaging, especially 
in smaller countries, and the lack of any clear pattern in the raw facts on the 
size of the welfare state and economic growth. Why does government intervention 
in markets that cross international borders have such dramatic and easily 
demonstrated effects on economic performance, when the impact of the size of 
the welfare state on the rate of economic growth is difficult to discern in the 
aggregate data?

As we shall see, some forms of intervention in markets do indeed usually have 
far larger impacts on economic performance than others, in part because the 
impacts of some types of intervention are fairly closely monitored and thus 
limited, whereas others are not. The next chapter will distinguish two 
different types of redistributions of income that usually have quite different 
impacts on economic development. The type of redistribution of income that is 
more damaging to economic performance also turns out to be more complex and 
less conspicuous. Sweden has uniquely high levels of the more conspicuous but 
less costly type of redistribution, but it does not appear to have unusually 
high levels of the less conspicuous but more damaging type of redistribution. I 
shall attempt to show in this book that that is the single most important 
reason why Sweden is not doing worse than it is.

6 Independently, Christopher Clague has found by econometric methods that the 
manufactured exports of the less developed - and on average vastly more 
protectionist countries are, disproportionately, relatively self-contained 
products that can be produced without unrestricted access to a wide variety of 
intermediate goods and other inputs. See Christopher Clague, "Relative 
Efficiency, Self-Containment and Comparative Costs of Less-Developed 
Countries," Economic Development and Cultural Change, forthcoming.


Chapter 3: Explicit, Implicit, and Efficient Redistribution

Most people use the phrase "income redistribution" to refer to transfers to 
relatively low- income people through social insurance or welfare-state 
programs explicitly designed to reduce the inequality of the distribution of 
income. But many other kinds of governmental intervention also change the 
distribution of income, as does collective action of groups of firms or workers 
in the marketplace. The tariffs, import quotas, and other protectionist 
measures considered in the last chapter undoubtedly change the distribution of 
income - they change the prices of some products and thus also the returns to 
the firms and the owners of the labor and other resources that produce these 
products. Public subsidies, price supports, and tax loopholes obviously also 
change the distribution of income. Even many public policies that have little 
or no significance in the budget of the government - such as regulations that 
restrict entry, limit competition, fix prices, or mandate benefits to employees 
and others - change the distribution of income. Similarly, whenever firms or 
workers successfully combine, whether through explicit cartelization or tacit 
collusion, to change prices or wages this again changes the distribution of 
income.

"Implicit" and "Explicit" Redistributions

It is useful to label the money and the services that governments openly 
transfer or provide to low-income people, simply because they are deemed 
morally to deserve or need the assistance, as explicit redistributions. By 
contrast, when a society is persuaded to choose a policy mainly for some reason 
other than its redistributive effect - as is the case when a society protects 
manufacturers against competing imports, or regulates prices or competition in 
some industry, because it is persuaded that will further the

29

development or strength of the country - that redistribution will here be 
called an implicit redistribution of income.

In keeping with standard usage in economics, I define the social cost of a 
redistribution, whether it is explicit or implicit, as only the "deadweight 
loss" or "excess burden," that is, the reduction in the national income from 
the transfer. If one group is taxed to finance transfers to another, the social 
cost is the reduction in the income of the society that results from any 
impairment of the incentives to work, to save, to innovate, and to allocate 
resources to their most productive uses, plus the costs incurred in 
administering the transfer. The amount that the taxpayers transfer, though 
obviously a matter of interest to them, is not a loss or cost to the society 
since the recipients of the transfer are also a part of the society. This 
definition of social cost is standard in economics and is obviously also the 
one that is relevant when explaining the rate of economic growth or the level 
of per capita income in a society.1

Of the many of redistributional activities, both explicit and implicit, that 
take place, which have the greatest social cost for each dollar or crown 
redistributed? Strangely, this subject, important as it is, has been so badly 
neglected that we must work out much of the analysis from the start.

The Criteria for Redistributions Generate the Social Costs

In economic theory, there is a familiar answer to the question of what type of 
redistributions has the least social cost: the "lump-sum" transfer, by 
definition, has no impact on incentives and thus no social loss whatever 
associated with it. The concept of the lump-sum transfer is a useful 
simplifying device in theoretical discussions, but all sides agree that no 
society could in practice have a continuing program of lump-sum transfers. The 
continuing transfers would be taken into account in behavior and thus would 
affect the incentives and the income of the society. So the practical question 
is, what types of systematic transfers that are actually possible will come 
closest to the lump-sum ideal?

Although there will inevitably be some excess burden from the taxes that raise 
the money for a transfer, the deadweight loss can usually be minimized by an 
unconditional cash transfer. If the recipients of the transfer are not required 
to do or not do anything to receive the transfer, the transfer will not have 
any effect on the incentives that they face.

1 If any expenditure, whether by the private or the public sector, actually 
increases the national income, then I will classify it here as an investment 
rather than a redistribution, even if the increase in income does not spread 
evenly throughout the society. Thus a public investment in the skills of the 
poor (or, for that matter, a program for the rich) that did in fact increase 
the national income would not be considered a redistribution, no matter how 
much this shifted the distribution in favor of the poor (or in favor of the 
rich). Our problem will become intractable unless we distinguish 
redistributional activities from investments and changes in the supplies of 
productive factors.

30

If the transfer is in the form of cash, the recipient can use it as he or she 
pleases, so there is also no distortion of the consumption choices of 
recipients.

The great practical force of this point is evident if we consider an uneconomic 
industry. Assume that, say, the shipbuilding industry has lost its comparative 
advantage, so that the costs of operating the shipyards are greater than the 
revenues that can be obtained and that in the absence of any redistribution to 
the industry they would be closed down. Let us stipulate that there is no 
external economy to justify any subsidy on grounds of social efficiency and 
that any aid or protection for the industry arises only because of the 
political pressure applied by the shipbuilding companies and workers to obtain 
protection or aid for themselves.

Suppose the organized shipbuilding companies and the workers through lobbying 
obtain a ban on the purchase of foreign ships. Protectionism benefits the firms 
and workers who sought it only as long as they remain in the industry. If the 
criteria for getting any kind of government aid is (as it usually is) that the 
firms and the workers continue to remain employed in the industry, then the 
recipients will have to allocate their time and other resources to shipbuilding 
to qualify for the redistribution.

Keeping the losing industry going entails that the resources devoted to it 
produce output of less value to the society than they would have produced in 
the most attractive employment in unsubsidized sectors of the society. That is, 
these resources are partly wasted and there will therefore be a cost to the 
society beyond that of the taxes or other direct costs of the redistribution.

This argument applies not only to bailouts and aid to declining industries, but 
also to redistributions to growing industries. If there is, say, a 
redistributive tariff or subsidy that raises the price in a thriving industry 
or activity, then other resources will move to take advantage of the higher 
price and these entering resources will then usually produce less value to 
society than before. This migration will cease only when the marginal private 
return is the same in the favored activity as elsewhere, but then the return to 
society at the margin will be less than the return to identical resources in 
unsubsidized areas. The migration to subsidized areas also means that some of 
the redistribution will, from the point of view of those who sought it, be 
wasted.

Usually, redistributions of the typical kinds to any industry, occupation, or 
region, or any redistribution to the users of some input, will mean resources 
are allocated in a way that produces less value to society. That is, subsidies 
or regulations that favor a particular group rather than correct a market 
failure - and are accordingly simply redistributive2 - generate losses for the 
society. They generate social waste because

2 Though they won't matter much for the argument here, there are some 
complexities that need to be dealt with to isolate the smaller implicit 
redistributions. To determine whether a measure is "simply redistributive," it 
is necessary to specify an initial or reference set of institutions and 
policies in order to delineate changes. Any stipulated set of initial 
conditions may, of course, be controversial; there are different ideas about 
what institutions are most productive. Thus some observers might object to any 
given measure of the extent of implicit redistribution because they objected to 
the initial state in terms of which the change in policy was defined. (The note 
continues on next page.)

31

normally the criterion for receiving the redistribution to an industry or 
occupation or locality is that the relevant firms or workers must allocate 
their capital and their time to the favored sector rather than to the sector 
that would, in the absence of the redistribution, have been most profitable.

To keep the costs of the redistribution from becoming insupportably high and to 
prevent the redistribution from being "wasted" on those for whom it was not 
intended, regulations or limitations are often established. A government 
regulatory agency may be created to prevent "abusive" or "speculative" actions 
to take advantage of the subsidy and to prevent an oversupply of the subsidized 
good. Sometimes only the original firms and workers will be legally entitled to 
receive the higher price or other subsidy in question. In declining industries 
the regulatory measures may even be designed to induce some of the excess 
resources to leave the industry at the same time that the protection or 
subsidization make the industry more profitable than it would otherwise have 
been. Such measures can reduce the extent to which conditional redistributions 
distort the allocation of resources a cross industries.

Slower Innovation as a Deadweight Loss

Restrictions and regulations also have to be applied in practice and enforced, 
and this entails some administrative or regulatory involvement in the 
productive process. The costs of the bureaucracy needed for this purpose may 
not be very large, but the social costs of the extra complexity, rigidity, and 
delay are rarely small. If the benefits of a higher price or any other subsidy 
are to be restricted to the original capital and labor in the relevant 
industry, then some authority must decide what whether a given expenditure is 
just maintenance and repair or really a new investment. If the regulation is 
going to block the extra output that would otherwise result from a higher price 
or other subsidy, then the level of output of each firm must be monitored and 
controlled.

If new investment or output are controlled, the incentive to innovate is no 
longer so clear. If exploiting a new technology or idea changes the optimal 
level of investment and production, as it usually does, the new desired new 
pattern of investment or production is unlikely to get the needed regulatory 
approvals without delay. Those enterprises that would lose from a rival's 
innovation can also use the politics of the

Note, however, that the concept of implicit redistribution is applicable 
whatever initial distribution of endowments or wealth is preferred - the 
desired distribution of wealth can be obtained by explicit redistributions. To 
delineate implicit redistributions we need only distinguish productive measures 
that increase social income from those that implicitly redistribute it. Of 
course, even this can sometimes be very difficult.

Since this book is concerned only with implicit redistributions that bring 
large deadweight losses, this complication is not a major difficulty here. 
There is not much controversy among skilled specialists about which policies 
that have large excess burdens.

32

regulatory process to block the innovation.

Thus the controls that are needed to prevent waste and abuse in a subsidized 
sector normally delay innovation and slow the reallocation of resources. The 
social cost of conditional redistribution often includes a lower rate of 
innovation and a less flexible economy. Some economists speak of "excess 
burdens" and "deadweight losses" only in static contexts where innovation, 
flexibility, and productivity growth are irrelevant, but this usage is 
misleading and (since the dynamic losses are usually much larger that the 
static costs) also mischievous. In this book, the excess burden or deadweight 
loss resulting from a conditional redistribution includes the slower 
productivity growth that regulation and other complex decision-making 
procedures bring about.

If those in the protected, subsidized, or cartelized industry are simply given 
cash from the government with no strings attached, then the total cost to the 
rest of society is merely the sum transferred plus the deadweight loss of the 
taxes that obtain this sum; the recipients of the cash still have an incentive 
to allocate their capital and their labor to whatever employments offer the 
highest returns, and (in the absence of other market failures) these will be in 
the sectors with the highest marginal return to society. It is true that the 
beneficiaries of the unconditional cash subsidy will be better off because of 
the transfer to them, and the "income effect" of the transfer could make them 
take more leisure, but this does not reduce the efficiency of the society. Some 
have more income and may take more leisure, and others have less income and 
take less leisure, but (apart from the excess burden of the taxes) the 
efficiency of the society is not changed.

The new distribution of disposable income arising because of the redistribution 
may or may not be unjust, but that is a separate issue that is not relevant to 
the present question of which types of redistribution have the lower social 
costs. When a redistribution is by its nature or conditions restricted to those 
who remain in the existing pattern of activity, it is usually possible to 
provide the recipients with the same net gain, at a smaller cost, by giving 
them an unconditional cash payment. Thus no amount of moral concern about 
fairness alters the reality that it is the criteria or conditions inherent in 
protectionism and in most aid to industries, occupations, and localities that 
increase their social costs.

It is also important not to overstate the point that has just been made. It is 
possible that the excess burden of the taxes needed to finance an unconditional 
cash transfer could exceed the deadweight loss from the protection, monopoly, 
regulation, or other devices are often used for an implicit redistribution.3 In 
this case the unconditional cash transfer would not be the cheapest form of 
redistribution. Essentially, the reason for this is that tax burdens are also 
conditional - the amount of tax we have to pay is understandably conditional on 
how much income we earn and thus on how much we work and save, and this makes 
our individual choices socially less efficient. The conditionality inherent in 
taxation could cause greater distortions than the conditions on the typical 
implicit redistribution. The logical possibility that a typical measure for

3 I am grateful to Stephen Baba for reminding me of this.

33

implicit redistribution could have a lower social cost than a typical implicit 
redistribution is not, however, so important in practice. This study will go on 
to indicate reasons why the social costs of many implicit and conditional 
redistributions will, in reality, usually become far higher than an 
unconditional cash transfer would have been.

Conditions on Explicit Redistributions to the Poor

The idea of conditionality as the source of distortions can also be applied to 
explicit redistributions to low-income people. There is one criterion for most 
such transfers that causes them to have a greater deadweight loss than an 
unconditional cash transfer of equal size. One nearly universal condition on 
welfare-state grants for the poor is that they will no longer be available if 
the recipient becomes prosperous. This condition is only natural and proper, 
but we must, if we are being honest, recognize that it entails some deadweight 
loss: the poor person who somehow mitigates his or her plight thereby loses 
some transfers, and this reduces the poor person's incentive to obtain an 
income high enough to end the redistribution.

As this last illustration suggests, the point of my argument is not to advocate 
redistributions with no strings attached - it is rather to make clear that the 
criteria or conditions that qualify an individual or firm to receive 
redistributions are the main source of their social costs. This idea, and the 
distinction between explicit and implicit redistributions, are indispensable 
for a full understanding of economic efficiency and growth in the modern world.

The Theory of Efficient Redistribution

We can also now begin to assess what I like to call the theory of "efficient 
redistribution," which grows out of the work of Gary Becker, Donald Wittman, 
Earl Thompson and Roger Faith, and others.4 Although different advocates of 
this theory make somewhat different arguments, the most basic idea, at least in 
Gary Becker's formulation, is that those who lose from redistributions of 
income have an incentive to

4 Gary Becker, "A Theory of Competition Among Pressure Groups for Political 
Influence," Quarterly Journal of Economics, (August 1983), pp. 371-400, "Public 
Policies, Pressure Groups, and Dead Weight Costs," Journal of Public Economics, 
28 (1985), pp. 329-347; Earl Thompson and Roger Faith, "A Pure Theory of 
Strategic Behavior and Social Institutions," American Economic Review, vol. 71 
(June 1981) pp. 366-80; Donald Wittman, "Why Democracies Produce Efficient 
Results," Journal of Political Economy Vol. 97, No.6, (December 1989), pp. 
1395-1424; and Bruce Gardner, "Efficient Redistribution Through Commodity 
Markets," American Journal of Agricultural Economics, vol. 65 (May 1983), pp. 
225-34.

34

keep their losses to a minimum. When a redistribution to one group in a society 
increases to the point where the losses to others become large, then the 
resistance to this redistribution also increases. If the losses are 
substantial, Becker emphasizes, there will be so much political resistance that 
the redistribution is almost certain to be curtailed.

The deadweight losses from redistributions obviously add to the losses of those 
who lose from the redistribution, and they do not, of course, help the 
beneficiaries of the redistribution either. Both the gainers and the losers 
therefore have an incentive to keep the excess burden to a minimum. Indeed, 
they have an incentive to bargain with one another until they maximize the 
joint gains that they can obtain from reducing excess burdens - in other words, 
an incentive to continue bargaining until the society has achieved an efficient 
allocation of resources.

In Donald Wittman's formulation, at least, it is only the time and other 
valuable resources used up in the bargaining - only the transactions costs - 
that keep a society from achieving the most efficient state we can conceive of. 
But transactions and bargaining costs are an inescapable feature of reality, 
and the cost of the time and other resources that they use up is just as 
meaningful as any other cost. So transactions costs, like the costs of any 
productive activity, should be part of the costs we take into account in 
defining a Pareto-efficient or totally efficient state. Some of those who use 
the theory of efficient redistribution therefore conclude that the existing 
societies, notwithstanding the redistributions they often engage in, are 
essentially Pareto-efficient.

Accordingly, the theory of efficient redistribution predicts that any social 
losses from redistributions are small and that redistribution is, at least for 
the most part, "efficient redistribution." Unstinting proponents of this type 
of thinking, such as Thompson and Faith and Donald Wittman, therefore say (in 
effect) that our world could not be much more efficient - that this is, indeed, 
the best of all possible worlds.

If the theory of efficient redistribution is right, then we seem to have an 
answer to our second question: any redistribution that actually occurs anywhere 
is quite efficient and so any social costs are small. It follows that Sweden 
cannot be losing much from its large welfare state.

We Are Not Done Yet

Unfortunately, even though it contains the germs of some important truths, the 
theory of efficient redistribution is in large part wrong. We still have a ways 
to go to get an answer to our second question of why Sweden isn't worse off.

To see what is wrong about the theory of efficient redistribution - and even to 
isolate the elements of truth in it - we need a new conceptual framework. To 
construct this framework, we shall need a number of different ideas, including 
the distinction between explicit and implicit redistribution that has been 
developed in this chapter.

The new conceptual framework is presented in the next two chapters. Some of the

35

concepts in this framework are new and others are drawn from my books on The 
Logic of Collective Action5 and The Rise and Decline of Nations. I shall 
endeavor to present the argument in such a way that it will involve very little 
repetition for those who have read these books, yet be comprehensible to those 
who have not. When this framework is complete, we will not only be able to see 
what is right and wrong about the theory of efficient redistribution, but we 
shall also be able to look at modern Sweden from a new angle.

5 Cambridge, Mass.: Harvard University Press, 1965.


Chapter 4: "Rational Ignorance" and the Bias of Collective Action

No analysis of the social costs of redistributions can capture the essence of 
the matter unless it faces up to a sad and inescapable reality: "rational 
ignorance." The seemingly oxymoronic phrase "rational ignorance" is not, in 
fact, a contradiction in terms. In many circumstances, the typical citizen 
serves his or her individual interests best by allocating little or no time to 
the study of public affairs, even though this leaves the citizen ignorant of 
many matters that are important for the country and thus also for his or her 
own wellbeing.

The paradox becomes clear when one examines the situation of an average citizen 
who is deciding how much time to devote to studying the public policy choices 
facing the country. The more time the citizen spends studying public affairs, 
the greater the likelihood that his or her vote will be cast in favor of 
rational policies. The typical citizen will, however, receive only a small 
share of the gain from more effective policies and leadership; if there are a 
million citizens, an average citizen will get only one- millionth of the total 
gain. Yet that citizen bears the whole cost of whatever he or she does to 
become better informed about public affairs. Thus each citizen would be better 
off if all citizens spent more time finding out how to vote to make the country 
better serve their common interests.

The gain to a voter from studying public issues to determine the vote that is 
truly in his or her interest is the value to that one individual of the "right" 
election outcome, multiplied by the probability that a change in this one 
individual's vote will change the outcome of the election. Since the 
probability that a typical voter will change the outcome of the election is 
minuscule, the typical citizen, whether a physician or a taxi driver, is 
usually rationally ignorant about public affairs. This point was made, albeit 
less starkly, in Anthony Downs's classic 1957 book, An Economic Theory of

37

Democracy,1 and in recent years its extraordinary practical importance is 
coming to be realized.

Occasionally information about public affairs is so interesting or entertaining 
that it pays to acquire it for these reasons alone. Similarly, individuals in a 
few special vocations can receive considerable rewards in private goods if they 
acquire exceptional knowledge of public goods. Politicians, lobbyists, 
journalists, and social scientists, for example, may earn more money, power, or 
prestige from a knowledge of the public's business. Sometimes exceptional 
knowledge of public policy can generate exceptional profits in stock exchanges 
or other markets. Nevertheless, the typical citizen finds that his or her 
income and life chances are not improved by the zealous study of public 
affairs. Most people are not, of course, totally self-interested, and their 
altruistic motives make many of them study public affairs somewhat more than 
self-interest alone would justify, but the evidence nonetheless reveals that 
rational ignorance is undoubtedly the norm.

This fact - that the benefits of individual enlightenment about public goods 
are usually dispersed throughout a group or nation, rather than concentrated 
upon the individual who bears the costs of becoming enlightened - illuminates 
many other phenomena as well. It explains, for example, the "man bites dog" 
criterion of what is newsworthy. If the television newscasts were watched or 
newspapers were read solely to obtain the most important information about 
public affairs, aberrant events of little public importance would be ignored 
and typical patterns of quantitative significance would be emphasized. Since 
the news is, by contrast, largely an alternative to other forms of diversion or 
entertainment, intriguing oddities and human-interest items are commonplace. 
Similarly, events that unfold in a suspenseful way or sex scandals among public 
figures are fully covered by the media, whereas the complexities of economic 
policy or quantitative analyses of public problems receive only minimal 
attention. Public officials, often able to thrive without giving the citizens 
good value for their taxes, may fall from power because of an exceptional 
mistake that is simple and striking enough to be newsworthy. Extravagant 
statements, picturesque protests, and unruly demonstrations that offend much of 
the public are also explicable in this way: they make gripping news and thus 
call attention to interests and arguments that might otherwise be ignored. Even 
some acts of terrorism that are described as senseless can, from this 
perspective, be explained as effective means of obtaining the riveted attention 
of a public to demands about which they otherwise would remain rationally 
ignorant.

In part because of rational ignorance, there is much more implicit 
redistribution than explicit redistribution in most democratic societies. These 
implicit redistributions, moreover, normally are not efficient redistributions. 
The prevalence of implicit and exceptionally inefficient redistributions is due 
to the way that rational ignorance interacts with a bias in the pattern of 
collective action, to which we now turn.

1 New York: Harper & Row.

38

The Difficulties of Collective Action

The rational ignorance of the typical voter is an example of the general logic 
of collective action. This logic is readily evident in organizations that lobby 
a government for special-interest legislation or that cooperate in the 
marketplace to obtain higher prices or wages. Some examples are professional 
associations of physicians or lawyers, labor unions, trade associations of 
firms in individual industries, farm organizations, or oligopolistic 
collusions.

Such organizations can only be understood if we are aware of how difficult 
collective action is for large groups. It is difficult because the benefits of 
collective action go automatically to everyone in some group or category. If an 
association of firms wins a tariff, that raises the price for every firm that 
sells the commodity or product in question, regardless of whether the firm 
contributed to the effort to win the tariff. Similarly, if one group of workers 
strikes to bring a higher wage in some factory or mine, all the workers in the 
relevant factory or mine receive the benefit of the higher wage, regardless of 
whether they paid dues to the union or walked in the picket lines that made the 
strike successful. The same reasoning applies to the firms or workers 
attempting to raise prices or wages by combining to restrict the quantity 
supplied.

Because the benefits of collective action go to everyone in a category or 
group, it is not rational for an individual in a large group or class to make 
any voluntary sacrifices in the interests of the group. The individual citizen 
or firm will get the benefits of whatever actions others undertake whether or 
not he contributes anything and, in large groups, the single individual or firm 
is not able to bring about the desired results singlehandedly. The precise 
logic and the empirical evidence validating this point are set out in my book 
on The Logic of Collective Action, and in the literature that has grown out of 
that book, so it should not be necessary to go into any detail on this matter 
here.

It is, however, essential to note that the individuals in large groups do not 
voluntarily, in the absence of special arrangements I will consider below, 
contribute time and money to organizations that would lobby or fix prices or 
wages for exactly the same reason that the typical citizen remains rationally 
ignorant about many aspects of public affairs. An individual receives only a 
minuscule share of the return from any sacrifice he or she makes in the 
interest of a group, whether the sacrifice takes the form of dues paid to a 
lobbying or cartelistic organization, or research into what political outcomes 
are best for individuals like oneself. Thus many groups with common interests - 
such as consumers, taxpayers, the unemployed, and the poor - are not organized 
for collective action, and most people have only the haziest knowledge of 
public affairs.

Those large organizations to lobby the government or fix prices and wages that 
have managed to survive have special arrangements that mainly explain why they 
are able to attract dues-paying members. There are in all large and lasting 
organizations for collective action some special gimmicks, which I call 
"selective incentives," that account for most of the membership. The selective 
incentives are individualized benefits or

39

punishments that induce firms or people to participate in, or help pay the 
costs of, collective action. One example of a selective incentive is the 
element of compulsion inherent in the closed shop, the union shop, and the 
coercive picket line, but this is only the most obvious example. All large 
organizations for collective action that survive have some analogous 
arrangements. These arrangements are usually very subtle and often provide 
individual benefits to those who contribute to the organization for collective 
action, while denying the benefits to those who do not.

When the beneficiaries of collective action are few, there may be voluntary 
rational action to obtain collective goods without selective incentives. 
Consider the small number of large firms in a relatively concentrated industry. 
If there are, say, three large firms of about the same size in an industry, 
each firm will obtain about a third of the benefits of any action to get 
governmental favors or higher prices for the industry. This third of the 
benefits will usually be a sufficient incentive for considerable action in the 
interest of the industry. When the numbers in a group are small, it will also 
be true that each participant will have a noticeable effect on how well the 
common interest of the small group is served, and this will affect the 
likelihood that the others will contribute. Thus small groups will often 
bargain until they agree to act in their group interest to a complete or "group 
optimal" extent. This organizational advantage of small groups, and 
particularly of small groups of large firms has, as will be shown below, 
important implications for the pattern of redistributions that emerges in most 
societies.

Since collective action is difficult and problematical, it normally takes quite 
some time before a group can overcome the difficulties of collective action, 
even if it has the small numbers or the access to selective incentives that are 
needed. The bargaining that can make it possible for a small group to organize 
or collude to an optimal extent usually takes some time, since unanimous 
consent is needed for full-scale cooperation. Organizing large groups is 
incomparably more difficult and time-consuming. Selective incentives, even if 
potentially available, are hard to arrange. If the selective incentives are to 
be positive rewards to those who participate, there has to be some surplus 
profit or advantage somewhere that can be used as the source of the rewards, 
and this surplus will normally be devoted to the collective action only if 
there is a complementarity between the activity that generates the surplus and 
the collective action (as is the case when a lobby gets favorable legislation 
for the enterprise that provides the resources for its selective incentives). 
If the selective incentive is the punishment of those who do not share in the 
costs of the collective action, this punishment has to be organized and the 
resistance to it overcome.

The time it takes to get collective action going was illustrated when Jimmy 
Hoffa, who ultimately became a powerful American union leader, was a youth 
working in a warehouse in Michigan. On a hot June day, the warehouse company 
received a large shipment of strawberries and other fresh produce that would 
become valueless unless it reached consumers before it spoiled. Jimmy Hoffa and 
his collaborators chose that moment to organize a strike and a union, and the 
management, rather than lose the fresh produce, gave in. It is only in the 
fullness of time that many groups will have had the

40

able leadership and the favorable circumstances needed to organize for 
collective action. As we shall see, this fact has important implications for 
economic growth.

The Inegalitarian Bias of Collective Action

If my theory of collective action is correct, the capacity for collective 
action is most common among the relatively established and prosperous interests 
in society and is virtually absent among the poorest and most insecure elements 
in the population. As we have seen, collective action is less difficult with 
small numbers, and this favors the organization and collusion of such groups as 
the large firms in concentrated industries. Selective incentives are also more 
often available to "insiders" - incumbent workers and well-established people - 
than to prospective entrants and those on the lowest rangs of the social 
ladder. I have shown elsewhere that the learned professions often have the 
widest array of selective incentives.2 Incumbent workers in an enterprise have 
already assembled for work and, if they have worked together long enough to 
have established a social network, they have a good chance to organize a union. 
By contrast, the unemployed are not automatically assembled or associated in 
ways that make selective incentives available through social interaction. 
Normally, all of the poor and marginal parts of the population are without 
access to selective incentives. It is not, however, poverty or insecurity 
themselves that prevent collective action. Consumers - even consumers of luxury 
goods - are also not organized; they do not have the advantage of small 
numbers, and they make their purchases at so many scattered locations that the 
selective incentives of picket lines and social interaction available to some 
workers are also unattainable.

Some people may wonder whether my hypothesis that the well-established and 
economically powerful elements in society are usually best able to overcome the 
difficulties of collective action fits the facts. We can test this hypothesis 
by examining how long it took different groups to overcome the difficulties of 
collective action. If well-established and well-off groups are, in fact, able 
to organize more readily than humbler groups, then the first organizations for 
collective would have represented those groups, and the poorest and least 
secure parts of the population should not even now be organized.

1 Adam Smith's Wealth of Nations, published in 1776, is a superb source of 
information about some of the first groups that were able to organize for 
collective action. This book is mainly an attack on "mercantilism," or on 
government policies and collusive prices that result from the combined action 
of "merchants" and master- manufacturers. In Adam Smith's time and before, 
merchants and manufacturers were often organized in guilds. Smith emphasized 
how often merchants and master

2 Olson, Collective Action.

41

manufacturers colluded to fix prices or to influence government; he said that 
they rarely gathered, even for merriment or diversion, without conspiring to 
fix prices. He also argued that ordinary laborers, the poor, and those in 
agricultural pursuits were usually not organized and had relatively little 
influence. Adam Smith's observations, as well as a good deal of other evidence, 
support the deduction earlier in this essay that small groups, such as the 
merchants or manufacturers in a particular industry or town, find it less 
difficult to organize than large groups do.

The selective incentives that large groups need to organize are also more often 
available to those with established positions and higher incomes than to poorer 
and less secure individuals. The historical record in country after country 
shows that the learned professions tended to organize long before workers of 
lesser income and status. Similarly, skilled workers organized unions long 
before unskilled workers did. The first unions represented skilled workers in 
England. During the first half century of organized labor in the United States, 
unionized workers were called the "aristocracy of labor." Even among unskilled 
workers, it is mainly those who already have jobs, and almost never the 
unemployed or new entrants, that are organized. Men are also more often 
organized than women and individuals in relatively well-placed ethnic and 
social groups more often organized than those and those in disadvantaged 
groups. Most important of all, there is no society anywhere in which the 
poorest people or the unemployed are organized.

Experience therefore confirms the hypothesis that the capacity for collective 
action is positively correlated with income and established position. We must 
now go on to examine how this reality, in combination with rational ignorance, 
comes in time to generate large implicit redistributions, most of which are far 
indeed from being efficient redistributions.


Chapter 5: Why Implicit and Inefficient Redistribution is Commonplace

We have seen that rational ignorance is a fundamental reality grounded in 
individual rationality, that the difficulties of collective action can 
eventually be overcome by some groups but not by others, and that it is 
disproportionately the nonpoor, insider, and establishment interests that are 
organized for collective action. What will organizations representing nonpoor 
established interests have an incentive to do, given that they operate in a 
society in which most people are rationally ignorant?

The answer depends in part on the extent to which these organizations, whose 
members are also citizens of the larger society, have an incentive to take the 
interests of the society as a whole into account. Here I draw on the analysis 
in my Rise and Decline of Nations. Consider an organization that, though it 
might be large and have many members, is still only a small part of the whole 
country or society in question. For the sake of simple arithmetic, I assume an 
organization that represents 1 percent of the income-earning capacity of a 
country: for example, a labor union whose members' wages are in the aggregate 1 
percent of the country's national income, or a trade association of firms that 
together earn 1 percent of the national income.

Organizations of this kind are relatively less important in Sweden, where there 
are also what I have called "encompassing" organizations, like the LO and the 
employers' federation, representing constituents that in the aggregate earn a 
significant percentage of the Swedish GDP and thus have a large stake in the 
society. I will deal with this special feature of the Swedish organization 
scene later, but the analysis of "narrow" coalitions that represent only a tiny 
part of a nation's income-earning capacity is nonetheless important here. It is 
important partly because the logic of these narrow coalitions to some degree 
also applies, in a more complex way, in Sweden,1 and also

1 The reasons why this is so are set out in my "Appreciation of the Tests and 
Criticisms" in Scandinavian Political Studies (Spring 1986).

43

because these narrow coalitions have a decisive influence on the economic 
performance of most of the countries with which Sweden is being compared.

One logical possibility is that organized interests will use their capacity to 
make the society in which their constituents are located more efficient and 
prosperous. In general, being part of a rich society is better than being part 
of a poor one. A lobbying organization could, for example, lobby for measures 
that would make the society in which its members live and work more productive 
and successful. Would an organization for collective action that represents 
only, say, 1 percent of the society have an incentive to do this?

An organization that represents 1 percent of the society would get, on average, 
only 1 percent of the benefits from making its society more productive. If the 
national income of a country increases because it lobbies for more efficient 
public policies, the clients of the organization will get, on average, 1 
percent of the increase in the national income. Those members, however, will 
have borne the whole cost of whatever lobbying they have done to improve the 
efficiency of the economy. If they get 1 percent of the benefits of their 
action and bear the whole cost of their action, then trying to make the society 
more efficient and prosperous will pay off for them only if the benefits of 
that action to the society as a whole exceed the costs of that action by a 
hundred times or more. Only if the cost-benefit ratio is better than 100 to 1 
will the organization for collective action best serve its members by acting to 
make the society more efficient and prosperous.

How then can a special-interest group representing only a tiny part of a 
society best help its clients? If a larger slice of the pie that society 
produces can be obtained for the members of a special-interest organization, 
then the members of this organization will have this larger slice of the pie. 
Less metaphorically, if a larger percentage of the national output or national 
income that is produced in a country can be redistributed to the members of a 
special-interest group, then these members will have that larger share of the 
national income.

But the reader may now ask, "Won't lobbying for favors from government or 
combination in the marketplace to obtain monopolistic prices or wages make the 
economy less efficient and productive? And won't the members of the 
special-interest group bear part of the reduction in the national income that 
comes from the inefficiencies brought about by their effort to capture a larger 
proportion of the national income?" The answer, in most cases, is, "Yes." 
Cartelization will usually reduce the efficiency and prosperity of the society. 
Because a combination or cartel will produce and sell less and charge more, the 
society will normally be less efficient. Special-interest lobbying will 
similarly induce resources to go into the particular areas that are favored by 
the lobby-inspired legislation; resources will crowd into these areas until 
their contribution to the national income - their marginal social product - is 
lower than it would have been in other areas and the efficiency of the economy 
will thereby typically

44

be reduced. So both cartelization and lobbying to get a larger percentage of 
the national income will, in most cases, make the society less efficient and 
productive.

Remember, however, that our special-interest group represented 1 percent of the 
society. Its members bear only 1 percent of the loss in national income or 
output that occurs because of the inefficiency its activities bring about, but 
they get the whole of the amount redistributed to them. Thus it pays our 
hypothetical special-interest group to seek to redistribute income to its own 
members even if this redistribution reduces the national income by up to 100 
times the amount redistributed! Therefore, organizations that represent only a 
minute percentage of an economy's income-earning capacity are really 
"distributional coalitions" - coalitions that strive to redistribute more of 
the society's income to themselves, rather than to produce anything.

Will Coalitions Seek Unconditional Cash Transfers?

We know from Chapter 3 that unconditional cash transfers tend to have lower 
social costs than conditional subsidies: the conditions, such as continued 
production of some product in order to receive the subsidy (which may be in the 
form of a higher price for that product), distort the allocation of resources. 
By contrast, those who receive an unconditional cash subsidy have an incentive 
to allocate their resources to the most productive uses. Unless they obtain 
unconditional cash transfers that they will continue to receive even if they 
move into entirely different lines of activity, the coalitions' constituencies 
will get only a part (and often only a small part) of what society gives up.

Gary Becker emphasized that the political resistance to a redistribution will 
be greater the more the rest of society loses from it. Unconditional cash 
transfers would obviously be worth more to recipients than equal-sized 
redistributions with strings attached. If it were really true that 
unconditional cash transfers not only were less costly to the rest of society, 
but also had to overcome less political resistance, then it would follow that 
distributional coalitions would always demand unconditional cash transfers.

But they do not. Indeed, the redistributions that coalitions seek are almost 
never unconditional cash transfers. They are usually protective tariffs or 
quotas, monopoly prices or wages, price supports, regulations that restrict 
entry and competition, and subsidies for those in particular industries, 
regions, or occupations, or for those who use particular inputs. Physicians are 
not subsidized by cash grants that they receive even if they no longer practice 
medicine; unionized workers are not given government checks that they continue 
to get even if they retire or become entrepreneurs.

Typically, the redistributions sought by organized interests are, if possible, 
not directly from the government budget. Those who benefit from protection or 
monopoly do not want the tariffs or monopoly privileges replaced by checks 
drawn on the national treasury. When open subsidies from the treasury are the 
only attainable form of redistribution, they are almost always conditional on 
continued participation in some

45

industry or activity; the subsidies to money-losing industries are available 
only to the firms and workers who stay in the losing industry, agricultural 
subsidies go only to those who continue to farm, and money-losing national 
airlines keep getting subsidies only so long as they fly.

To see why the redistributions sought by organized interests are virtually 
never the unconditional cash transfers that would be the least costly 
redistributions for the society, we need to return to the earlier analysis of 
the bias of collective action. We saw that it followed from the logic of 
collective action that those who could organize for lobbying, collusion, and 
cartelization were those with small numbers (such as the large firms in 
concentrated and protected manufacturing industries) or those with access to 
selective incentives (such as members of professions and workers already 
established in jobs). Though in a long-stable society some groups of 
below-average income will be organized for collective action, the overall 
pattern of collective action will favor more prosperous and better established 
interests; it will not include groups like the unemployed and the poor.

The organized groups that are never poor (but often relatively well-off) 
usually cannot obtain redistributions by appealing to the egalitarian moral 
sentiments of the electorate. Naturally enough, the typical voter does not want 
to have his or her own standard of living lowered solely for the purpose of 
transferring income to someone else who is already as well off, or better off, 
than the voter. Accordingly, most organized groups cannot further their 
interests by appealing on grounds of need for explicit redistributions to 
themselves. Each narrow distributional coalition also represents a small 
minority of the electorate, and therefore does not have votes enough to pass an 
explicit redistribution to itself.

So how can an organized group that cannot qualify for explicit redistribution 
on egalitarian grounds, and cannot hope to outvote the majority that would lose 
from redistributions to it, obtain any redistributions?

Rational Ignorance Makes Implicit Redistributions Possible

A distributional coalition can usually obtain redistributions by exploiting the 
rational ignorance of the electorate. Lobbying and special interest pressure 
usually succeeds, and cartelization and collusion are typically tolerated, 
mainly because of rational ignorance. If voters were fully informed, they would 
not be swayed by the publicity or advertising stemming from organized interests 
and they would replace any representatives who were serving interests other 
than those of the voters. They would also not allow those who sell them goods 
and services to raise prices and wages through cartelization. In reality, the 
average citizen's knowledge is extremely limited; survey data in the United 
States reveal that about half of the electorate do not know the name of their 
congressman in the House of Representatives, much less what help he has given 
organized interests to obtain

46

campaign contributions. Diverse forms of public relations, advertising, and 
political indoctrination play large roles in forming public opinion.

In the real world organized interests have vast opportunities to persuade 
voters to accept what are, in fact, implicit redistributions to those organized 
interests, as long as these redistributions are designed in such a way that the 
rationally ignorant can be persuaded that the society as a whole gains, or are 
so inconspicuous or indirect that a majority of the electorate is not aware of 
them. A policy that can be made to appear, in a newspaper advertisement or a 
thirty-second television commercial, to have different beneficiaries than it 
actually has, can be successful. It does not matter much whether lengthy 
research would show that a policy was contrary to the interests of most voters, 
because it is not in the typical voter's interest to undertake such research, 
and any research done by those with a professional interest in the matter will 
not have a great political effect unless substantial resources are available to 
publicize the results. These resources may be available when two or more 
organized interests have opposing interests, but the difficulties of collective 
action ensure that many organized interests will not be countervailed.

As would be expected from the argument here, manufacturing firms ask for 
protection against imports, or investment incentives, or tax loopholes that 
will "strengthen or protect the national economy," or they inconspicuously 
raise prices through collusion. Physicians seek to keep out "unqualified" new 
doctors, to prevent competitive ("unethical") behavior, and lobby for 
government and insurance spending that insures "good quality care for the 
public." Lawyers seek to ensure that citizens have more rights to sue and get 
"justice" (with the paid counsel of lawyers) in the courts. Large farmers argue 
that the nation should not be dependent on foreign food supplies and thereby 
obtain tariffs or quotas, or they argue that the poorer farmer cannot live 
decently unless there are price supports, even though, in fact, the higher 
prices go mainly to the larger farmers who produce the most. Skilled workers 
established in an industry may speak in the name of the whole working class, 
yet use their cartelistic power to obtain a wage at which it does not pay the 
employers to hire the part of the working class that is unemployed. Virtually 
all organized groups seek privileges that they can persuade the public are 
beneficial to society as a whole; professors, for example, emphasize the value 
to society of academic freedom that tenure for themselves is supposed to 
provide, not the opportunity for sloth and neglect of students that their 
extreme job security makes possible.

The Implicit Redistributions that Rational Ignorance Permits are Almost Never 
Efficient Redistributions

The redistributions that can be made to appear to serve the interests of the 
society as a whole, or that are so inconspicuous and indirect that most voters 
don't notice them,

47

obviously cannot be unconditional checks from the treasury. It is obvious who 
gains from an unconditional cash subsidy and the costs to the treasury and the 
taxpayers from such straightforward subsidies are conspicuous and easily 
distinguished. To appear to serve the interests of the society as a whole, the 
implicit redistribution must promote some type of production or activity that 
rationally ignorant voters can be persuaded is advantageous to the nation. 
Unfortunately, the very fact that an implicit redistribution is easier to 
obtain if it encourages some industry usually increases its social costs. Such 
redistributions normally expand some industry or activity beyond the socially 
efficient level, and thereby impose a cost to the society as a whole that is 
distinct from the redistribution itself. Inconspicuous redistributions are 
often also more costly to society than conspicuous ones: the costs that are not 
noticed are less likely to be minimized.

Tariffs and quotas are exceptionally apt devices for implicit redistribution. 
They protect a domestic industry and it is usually relatively easy to persuade 
voters that this also protects or strengthens the economy as a whole. The 
rationally ignorant have no reason to think about the implicit discouragement 
that protection of a given industry entails for other, unprotected industries 
that must compete with the protected industry for resources or export in the 
face of a higher exchange rate for the national currency, much less any reason 
to master the demanding literature on the deadweight losses that are often 
involved. Regulation that limits competition and entry is also an admirable 
device for implicit redistribution. The rationally ignorant normally take it 
for granted that the regulation favors the consumers and the public rather than 
the firms that are regulated and have no reason to go into the logic showing 
why the regulated firms are normally organized for collective action to 
influence the regulatory process and why the consumers and the public normally 
are not. The regulation can also be appealingly described as something that 
assures "orderly markets" and prevents "destructive competition," and thus made 
to appear favorable to economic progress. By contrast, the losses to the firms 
that would have entered the industry in the absence of regulation and the 
higher costs the regulation usually imposes on consumers are relatively subtle 
and difficult to identify. So are the ways in which regulation often slows down 
innovation by complicating decision-making. Similarly, cartelization and 
collusion are also well-suited for implicit redistribution. They can be 
defended as efforts to promote cooperation against foreign competition, as 
devices that insure "orderly markets," and as self-help efforts that do not 
impose costs upon taxpayers. Since prices and wages change from time to time in 
any case, the costs to consumers of the higher prices and wages that collusion 
and cartelization bring about are not usually evident to the casual observer.

In general, the most expedient devices for implicit redistribution are those 
that do not rely principally on the government budget. Usually any 
redistribution through the government budget, even if it is not an 
unconditional cash transfer, is easier to identify than a change in prices or 
wages that arises because of protection, regulation, or cartelization. The 
changes in prices and wages arising from redistributions that bypass the public 
treasury normally entail deadweight losses and usually are tied up with 
regulations and agreements that slow down innovation, but the redistributions 
that result

48

from these price and wage changes are normally less conspicuous than 
governmental subsidies. The fact that redistributions through the public 
treasury are a poor choice for an organized interest partly explains why the 
sizes of the public sectors of countries are not well correlated with their 
economic performance.

Although all continuing redistributions by any feasible method entail some 
deadweight losses, the implicit redistributions that nonpoor organized 
interests have an incentive to seek are those that are the least 
straightforward or the least conspicuous, not those that have the lowest social 
cost. If it is proposed that a socially costly implicit redistribution to an 
organized interest be replaced by a less costly but more nearly transparent 
redistribution, the organized interest will normally object, because it is 
likely to lose the redistribution altogether if it becomes transparent. 
Therefore, industries that enjoy tariff or quota protection, for example, 
almost always oppose replacing this protection with open governmental 
transfers. Monopolies and cartels similarly do not want to give up their status 
in return for government checks.

Accordingly, we see that the theory of efficient redistribution is not true for 
implicit redistributions. It would probably be true, if other things were 
equal, that redistributions with a higher social cost would be at a political 
disadvantage. But other things are not equal - rational ignorance entails that 
the least straightforward and the least conspicuous methods of redistribution 
be chosen. There is no reason whatever to suppose that the redistributions 
chosen to meet these criteria will have the lowest social costs. Because of the 
conditionality inherent in redistributions that achieve their objectives by 
altering relative prices or reducing the degree of competition, devious 
redistributions will often have vastly higher social costs than unconditional 
and transparent subventions.

The opacity of implicit redistributions unfortunately also entails that there 
is no necessity that they should be curtailed just because their social costs 
become very high - these costs need not be perceived, much less measured, by a 
rationally ignorant electorate. Many empirical studies confirm the conclusion 
that we have arrived at by abstract argument: there are many real world 
examples of implicit redistributions with social costs that are large multiples 
of the increase in the net incomes of the beneficiaries.

Thus, contrary to one school of thought, there is no tendency for bargaining 
over public policy among the groups in a society to continue until joint gains 
are maximized and Pareto-efficiency is achieved. Some groups are not organized 
to bargain and some of the social losses are not even perceived by some large 
groups of losers.

So this is not, alas, the best of all possible worlds, nor even the most 
efficient. It contains many societies in which rational ignorance can regularly 
be exploited by narrow distributional coalitions composed of nonpoor insider 
interests. Such societies are, as I have said before, like china shops filled 
with wrestlers battling over the china - and breaking far more than they carry 
away.

49

Aggregate Evidence on the Inefficiency of Implicit Redistributions

Many quantitative studies have shown that particular government programs or 
protective tariffs or other implicit redistributions have social costs that are 
very large, even in relation to the amount redistributed. Critics may 
understandable object that these particular cases are unrepresentative; the 
studies may even have attracted economists' attention because the social costs 
were so high. There is, therefore, also a need for evidence about how well 
societies with high levels of implicit redistribution perform in comparison 
with other societies. If we can explain variations in growth rates and income 
levels across countries and regions - and especially variations that other 
theories cannot explain - with the aid of the type analysis of implicit 
redistribution by organizations for collective action that has been offered 
above, then that is strong evidence that implicit redistributions often really 
are inefficient. The theory of efficient redistribution already suffers because 
it cannot explain differences in economic performance across countries, and if 
a theory that explains poor economic performance as due to high social costs of 
redistribution is successful, then the theory of efficient redistribution is in 
real trouble.

Although I did not deal explicitly with the theory of efficient redistribution 
in The Rise and Decline of Nations, I did show there that all of the really 
remarkable examples of economic growth and stagnation since the Middle Ages 
could be explained in large part by the density of narrow coalitions for 
collective action. Thus the evidence in that book is also telling evidence 
against the theory of efficient redistribution. For the benefit of readers who 
do not know that book, I shall refer to a couple of the most dramatic pieces of 
evidence in it. Then I shall show that the new evidence about international 
trade in Chapter 2 is also aptly explained by the argument that we have just 
been through.

The argument in Rise and Decline predicts that long-stable societies have more 
groups that have been able to overcome the difficulties of collective action 
than lately unstable societies, and that the redistributions that narrow 
organizations for collective action have an incentive to seek normally have 
high social costs. The long-stable societies with many narrow distributional 
coalitions should therefore be less efficient and (for reasons spelled out in 
Rise and Decline) also less dynamic than otherwise comparable societies.

A great deal of evidence suggests that this is indeed the case. The society 
that has had the longest period of stability and immunity from invasion and 
institutional destruction is Great Britain. And Great Britain in the twentieth 
century, as the theory predicts, has the poorest economic performance of all of 
the major developed democracies. One of the smaller developed democracies, 
Ireland, has suffered from an even poorer economic performance, but Ireland has 
also never had its coalitions destroyed by upheaval and (by a process that will 
be described later in this chapter) it accumulated distributional coalitions at 
an exceptionally rapid rate during the long period when it had extremely high 
protection of manufactures.

50

The theory also has clear implications for the Axis nations defeated in World 
War

II. In Germany and Japan, and to a lesser extent in Italy, the repressive 
dictatorial governments and the allied occupations after the war eliminated 
most of the distributional coalitions. A few such organizations were created 
during or shortly after the allied occupations, but most of these were 
"encompassing" organizations. The theory that has been outlined implies that, 
after a free and stable legal order had been established, those societies 
should have grown surprisingly rapidly. And, as everyone knows, they enjoyed 
"economic miracles." With appropriate elaboration, the aforementioned theory 
also explains the general pattern of regional growth in the United States since 
World War II.2

The Salience of the Evidence in Chapter 2

We found earlier that it is inherent in the logic of collective action that 
small groups, such as the small number of firms in a concentrated industry, 
have less difficulty acting in their common interest than large groups. If a 
country (and especially a smaller country) protects its manufacturing industry 
from foreign competition, then often only a few firms will need to collude to 
fix prices in a given line of industry. A small number of firms in a 
concentrated industry will usually be able to fix prices without help from the 
government, but in any event their small numbers will ease the formation of a 
lobby to get governmental help in enforcing the price fixing. Collective action 
among the few firms that produce a given manufactured good is accordingly 
fairly common, even in politically unstable environments. Any effective 
price-fixing agreements in manufacturing that work for any length of time need 
to include some specifications or understandings about the definition and 
quality of the manufactured product whose price is being fixed, since 
individual firms can profit by getting a larger share of the cartelized market 
through subtle price reductions in the form of free add-ons and extra quality. 
In the long run, this usually means more complex agreements and regulations 
that slow down the rate of innovation.

If, by contrast, a country is completely open to foreign manufactures, then 
normally all of the firms in the world producing the relevant manufactured good 
will need to combine if prices are to be fixed. There are so many different 
manufacturing firms in diverse countries around the world, and the difficulties 
of coordinating their behavior across many national, cultural, and linguistic 
borders are so great, that successful worldwide cartelization of manufactures 
is rare. Since there is no world government, we can be certain that no 
government will be using its coercive power to enforce a world wide cartel 
agreement or responding to any lobbying for world wide

2 "The South Will Fall Again: The South as Leader and Laggard in Economic 
Growth," Southern Economic Journal, 49 (April 1983), pp. 917-32.

51

special-interest legislation.

Collusive price fixing among the manufacturing firms in each industry is 
accordingly more likely in countries (and especially small countries) with 
protection than in those with open markets.

If the manufacturing firms in a country are protected against foreign 
competition, they do not need to worry that their costs of production will 
become higher than those in other countries. If the workers are cartelized in 
the protected industry, they do not need to worry about wage demands that make 
production more costly than elsewhere. In countries with free trade in 
manufacturing, by contrast, unions are often severely constrained in the extent 
to which they can exploit their monopoly over the supply of the relevant labor 
by the competition of foreign firms. Although unionization takes longer to 
develop than collusion among manufacturing firms and is in some unstable 
societies obstructed by periods of repression, some countries that protect 
manufacturing have cartelized wage levels in the protected industries that are 
far above the levels that would be sustainable without the protection. So 
protectionism of manufactures not only leads to collusion among the protected 
manufacturing firms, but often in due course also to a sky-is-the-limit 
monopoly wage setting that can have no counterpart in manufacturing industries 
having to meet international competition. These unrestrained wage levels are 
sustainable only if there are work rules that keep employers from upgrading new 
and unskilled labor to replace the workers with the wages that are farthest 
above competitive levels. Such work rules make industrial life much more 
complex and legalistic and thereby reduce innovation.

In making any manufactured good of any complexity, many different components 
and resources - sometimes thousands of different inputs - are needed. If each 
of these can be purchased without hindrance in whichever part of the world 
offers the best value, manufactured goods of better quality or lower price can 
be produced. In the protectionist country, some of the components and inputs 
that are needed will be more expensive and harder to get because they are 
themselves subject to protection. If protection takes the form of quotas and 
exchange control, this problem can make it impossible for a manufacturer, 
especially one in a smaller country, to be competitive on the world market in 
the production of any complex manufactured product.

And, as we saw in Chapter 2, none of the highly protectionist smaller countries 
were able to sell much in the way of manufactures on competitive international 
markets, whereas those smaller manufacturing countries with relatively low 
protection of manufactures (such as Sweden) could. The conceptual framework 
that has been offered here is certainly consistent with (and I believe in large 
part explains) the extraordinarily strong association between the openness of 
smaller countries to imports of manufactures and their success in developing 
competitive manufacturing industries. Small countries with high protection of 
manufactures find, as collective action builds up behind their tariff barriers, 
that they suffer from pervasive and almost limitless implicit redistribution in 
manufacturing, and become so inefficient that they cannot sell manufactured 
goods in competitive world markets.

52

These results also bear upon some recent developments in the theory of 
international trade. As we recall from Chapter 2, these recent developments 
suggested that, when there is imperfect competition and economies of scale, a 
country can sometimes best serve its economic interests through protection of 
manufactures. Though the results here do not call the logic of the new models 
into question, they do suggest that these models have only a very limited 
pertinence for public policy. Imperfect competition and decreasing costs are 
most significant in manufacturing in smaller countries, and it is exactly those 
conditions that have been considered in this book. The results here suggest 
that it is precisely when decreasing costs and imperfect competition are 
conspicuous that protection will be especially damaging, since in these 
conditions protection especially reduces the number of producers that must act 
collectively in order to collude and cartelize. The failure of the small 
countries with high protection of manufactures to be able to export 
manufactures suggests that the damage done by the collective action that the 
protection facilitates is of much greater significance than the static gains 
the countries could sometimes receive from protection of manufactures in some 
cases where there are decreasing costs and imperfect competition.

Sudden Increases in the Size of the Market and the Polity that Determines Trade 
Policy

We also saw in Chapter 2 a dramatic tendency for rapid economic development 
when a great expansion in the size of the market occurred through national 
unification or through the creation of a common market. Why would this 
"jurisdictional integration" be so strongly correlated with economic 
development?

If the difficulties of collective action ensure that it emerges only slowly, 
and if narrowly-based distributional coalitions obtain implicit socially costly 
redistributions to themselves, then we can see why jurisdictional integration 
generates rapid economic development. The creation of a much bigger market and 
of a bigger jurisdiction for determining trade policy will undercut most of the 
existing distributional coalitions. It takes some time before new ones form, so 
there is exceptionally rapid economic growth for a time.

Consider a small protected market like a medieval town with its own walls and 
economic policies. Suppose that suddenly the protection in the small 
jurisdiction is eliminated because there is national unification or the 
creation of a common market. Then the organizations for collective action - the 
guilds that have profited from the use of their cartelistic powers and lobbying 
powers - will find that after jurisdictional integration their customers can 
purchase from other suppliers in other towns, or in the suburbs, or in the 
countryside. Suddenly, because of the creation of a wider market, the guilds 
have lost their monopoly power. Since the jurisdictional integration creates a 
much larger jurisdiction, it also requires lobbying on a far larger scale, so 
the

53

organizations that were of a size suitable to lobby the town will usually not 
be strong enough to influence the new governmental unit.

So the theory predicts that the extent of damage done by organizations for 
collective action will be much smaller than usual after there has been a big 
freeing of trade, whether through national unification, a common market, or 
unilateral freeing of trade.

Organizations for collective action will eventually emerge again on a scale 
sufficient for lobbying or cartelizing the larger jurisdiction and the larger 
markets that have been created. But if my argument is right, it takes quite 
some time to overcome the difficulties of collective action, at least when the 
groups in question are large ones. Thus for a time there can be unusually rapid 
growth.

As the theory predicts, there was unusually rapid growth after the creation of 
the EEC, after the German Zollverein, after the Meiji Revolution, after the 
Dutch rebellion against the Spanish, and after the national unifications in 
England and the United States. Although the timing of the economic growth 
certainly is consistent with the theory, this correlation is not, of course, 
necessarily sufficient to establish causation.

Happily, various special features of the pattern of growth offer striking 
support for the argument. If we look, for example, at England in the early 
modern period, we find that the main form of manufacturing was textile 
manufacturing and that, after jurisdictional integration, it came to be handled 
under the merchant-employer or "putting-out" system. Manufacturing was not done 
mainly in the cities where the guilds, the distributional coalitions of the 
times, held sway, but rather in the scattered cottages of the countryside. 
Merchants went out to the countryside to contract with cottagers to have wool 
spun into yarn or yarn woven into cloth. This system was expensive both in 
terms of transportation costs and transactions costs, but it was nonetheless 
cheaper than production under guild rules in the old towns. In addition, much 
of the economic growth was concentrated in new towns or in suburbs where guilds 
did not exist. When the bigger national markets were created, firms could 
produce wherever they found costs were lowest, and costs were lower in places 
that were not under the control of guilds.

Both the striking association in smaller countries between relatively low 
levels of protection of manufactures and success in developing a competitive 
manufacturing industry and the clear association between increases in the size 
of the market through jurisdictional integration and economic development fit 
neatly into the pattern predicted by the theory offered here. Both lead to less 
implicit redistribution.


Chapter 6: The Lower Costs and Ultimate Limits of Explicit Redistribution

The evidence presented in Chapter 2 revealed a remarkably clear and strong 
pattern. Among the 41 less developed countries studied by the World Bank, there 
was a regular tendency for the outward-looking countries to perform better than 
the inward-looking or protectionist countries. The large number of 
internationally successful industries studied by Michael Porter and his 
associates were systematically industries that had not been significantly 
subsidized or protected from either international or domestic competition. Most 
strikingly, in all smaller countries on which data were available, high 
protection of manufactures was strongly associated with the failure to export 
significant manufactures on competitive world markets. In addition, great 
increases in the size of a trading area and in the jurisdiction that determines 
trading policy were also regularly associated with great accelerations in 
economic development. The evidence on protectionism and jurisdictional 
integration was so overwhelming that some fairly strong conclusions could be 
drawn whether or not all of the other variables relevant to industrial 
development have been included in the analysis: the evidence was a bit like 
that on plane crashes whose adverse impact on the longevity of the victims is 
clear even without taking into account the many other variables that affect 
life expectancy.

This makes it all the more puzzling why the national-level evidence presented 
in Chapter 1 did not show any strong relationship in either direction between 
sizes of the governments or the extent of transfers to low-income people and 
economic growth. Because of what we know from observing what happens in 
individual markets, we should expect that properly specified statistical tests 
would show that an unusually large and growing welfare state would make a 
country's rate of economic growth (though by no means necessarily its level of 
utility or welfare) measurably lower than it would otherwise be. Yet from the 
fixation in ideological debates - and even from some leading

55

economists' contributions to these debates - we have been led to expect that 
the extent of transfers to low-income people and the size of the welfare state 
were of decisive significance for the fate of nations: many people obviously 
take it for granted that transfers to low-income people are so overwhelmingly 
important for economic growth and human welfare that their impacts would be 
clear even in analyses that neglect other relevant variables. But the effects 
of transfers and the size of the welfare state on economic growth are evidently 
not colossal enough so that, like the effects of plane crashes on longevity, 
they overwhelm other factors. So we must ask why hasn't the adverse effect of a 
larger welfare state on economic performance been strong enough to overwhelm 
all other factors and thus to produce a striking pattern in the cross-national 
and historical comparisons reported in Chapter 1?

The conceptual framework presented in the last three chapters suggests a 
possible answer. These chapters have shown that there are powerful incentives 
for organized groups to seek implicit redistributions and preferably implicit 
redistributions that achieve their objectives by altering relative prices 
rather than through the government budget. The indirect and concealed character 
of these redistributions - and the high degree of conditionality that is needed 
to conceal their redistributional purposes - usually makes them more costly to 
society.

If the argument that has been offered so far in this book is correct, there is 
no reason why the amount of redistribution to low-income people, and even the 
share of the government in GDP, should be closely correlated with the total 
social costs of

redistribution. It is true, as critics of large welfare states say, that 
redistributions generate deadweight losses, but it is wrong to jump to the 
conclusion that the countries that have the largest amount of explicit 
redistribution to low-income people, or the largest share of the government 
disbursements in GDP, lose the most from redistribution. Countries in which 
less is transferred to the poor and in which the government is smaller can 
easily lose more from redistribution than the countries with the largest 
welfare states do. That is probably the main reasons why, as we saw in Chapter 
l, there was no strong correlation, either for the developed democracies or for 
all noncommunist economies, between the relative size of the public sector and 
the rate of economic growth.

The main purpose of this study is, as the subtitle indicates, to ask some 
questions about Sweden. So what does the intellectual framework developed in 
the last three chapters tell us about Sweden?

As we try to answer this question, it is important to keep in mind that there 
is no lack of implicit redistributions in Sweden (and in the other countries 
with the most generous welfare states). The argument in the prior chapter that 
there are often large losses from implicit redistribution is, I believe, 
definitely applicable to Sweden.

But is Sweden the country that loses the most from implicit redistributions? Or 
even one of the countries that loses the most? I doubt it. We must postpone any 
final answer until there is further research - the main purpose of this study 
is to generate new questions. Yet I find it hard to imagine that anyone would 
argue Sweden was unique in the extent and costliness of its implicit 
redistributions. I have found it more difficult to

56

find examples of strikingly costly implicit redistributions in Sweden than in 
most other countries. What reason is there to think that Sweden would be losing 
more from implicit redistributions than other countries? Why would it be losing 
relatively more than Argentina? Or Ireland? Or Britain and most of the other 
English-speaking countries? Why would it be losing more than the average 
country in Western Europe (not to mention those of Eastern Europe and the 
developing nations)?

Although we must postpone any final conclusion until the experts on Sweden have 
researched the matter, my working hypothesis until then is that Sweden is not 
the country that suffers the most from implicit redistribution, nor is it 
probably even close to being in this position. I hypothesize that Sweden is 
doing as well as it is in comparison to many other countries because its 
performance is not dragged down as much by implicit redistribution as that of 
some other countries is. The losses from implicit redistribution in the 
countries which have the most of it are so large that they more than offset 
Sweden's larger losses of measured output because of its relatively large 
explicit redistributions to low-income people.

What Limits the Amount of Implicit Redistribution in Sweden?

One major factor that, I believe, keeps down the quantity of implicit 
redistribution in Sweden is the country's relatively high resistance to tariff 
and quota protection for manufacturers. Certainly Sweden does not lose as much 
from implicit redistribution through protection of manufactures as some 
countries do. The statistics and the historical evidence in Chapter 2 suggest 
that this is a matter of extraordinary quantitative importance. The argument in 
Chapter 5 suggests that this relative openness to imports of manufactures also 
reduces the amount of implicit redistribution in Sweden's labor market and in 
some other factor markets as well.

This raises another interesting question: Why is Sweden somewhat more favorable 
to free trade in manufactures than many other countries? It would take far too 
long to analyze this question adequately now, so I shall merely refer to some 
of the relevant arguments here. Perhaps one factor is the historical accident 
that modern Swedish (and Danish and Norwegian) industrial development began in 
part through primary product exports in the nineteenth century, especially to 
free-trading Britain, which was then about the most prosperous country in the 
world. The Scandinavian countries exported dairy products, timber, oats, 
shipping services, and iron ore, for example, in the period in which their 
catch-up growth began in the nineteenth century. Although I could easily be 
wrong, I sense that many modern Scandinavian manufacturing industries began 
with the processing or development of primary product exports. Logs were in 
time processed into finished lumber, then into paper, and finally into 
sophisticated paper products; iron ore exports ultimately turned into exports 
of iron, steel, and finally into exports of complex manufactured goods; exports 
of dairy products

57

from Scandinavia maybe had something to do with the invention in Sweden of the 
cream separator.1 I do not know enough about the matter to draw any final 
conclusions, but it would be useful for someone to look into the possibility 
that this apparent symbiosis between primary product production for export and 
manufacturing, especially in the context of the contemporary British example 
and advocacy of free trade, encouraged early Swedish manufacturing interests to 
be more favorable to exports and to free trade than they might otherwise have 
been.

Probably another factor is the quality and influence of professional economists 
over the course of modern Swedish history. Though more of the giants of 
economic thought have come from Britain than from Sweden, the Swedish 
contributions have been extraordinarily large in relation to the size of the 
country - Sweden is, perhaps, the country with the largest per capita 
contributions to the subject. Often Swedish economists have also had a 
considerable influence on economic policy and on the thinking of the 
intellectual class in general.

It might seem that my earlier argument about rational ignorance and the role of 
self-interest in political life would rule out any influence of ideas on 
economic policy, but that is not the case. As I have argued elsewhere,2 those 
with a professional stake in a subject do not find it rational to be ignorant 
of that subject. Although intellectuals are as susceptible to self-interest as 
other people, their selfish interests are more likely to show up mainly in 
those matters in which they have an immediate occupational stake (such as their 
own pay and tenure), rather than in the typical public policy issue; 
individuals in a variety of social roles are like a judge or a member of a jury 
in the sense that their individual self-interest does not bear in any important 
way on the matter at hand. So there are some people who have both an incentive 
to become informed about a public issue and a reason to look at it in a 
public-spirited way, and ideas can exert an influence through them.

Therefore, in spite of my emphasis on the importance of organized vested 
interests, I believe that the quality of ideas is also an important determinant 
of what policies and institutions a country chooses, and that Swedish economic 
performance over the long run has probably been helped by the country's 
strength in economics. It appears that public opinion about protectionism, for 
example, has been greatly different in Sweden, on the one hand, than in 
Argentina, on the other. This difference must be due, in part, to the 
difference in the economic professions in the two countries.

1 I examined this issue somewhat less superficially in a talk in Stockholm in 
1984 organized by PKbanken.

2 See my paper on "How Ideas Affect Societies" in Ideas, Interests & 
Consequences, (London: Institute of Economic Affairs, 1989), and also reprinted 
in the LSE Quarterly, 3:4 Winter 1989, pp. 279-304.

58

Encompassing Organizations

Another factor that probably reduces the amount of implicit redistribution in 
Sweden is the large role that what I call "encompassing" organizations have 
played, at least at times. Suppose that an organization encompasses a large 
part of the income-earning capacity of a country; its constituents earn, say, 
50 percent of the nation's GDP. Such an organization, if it truly furthers the 
interests of its clients, will act very differently from the narrow coalition 
considered in Chapter 5 that represented only 1 percent of the income-earning 
capacity of a country. If the constituents of the organization get half of the 
benefit of anything it does to increase the prosperity of a country, that will 
often be enough to give the organization an incentive to do something to make 
the country more efficient and innovative. Such an organization, if it 
optimally serves its clients, will also not seek any redistributions for its 
clients that entail a social loss that is large in relation to the amount 
redistributed. If the clients of an organization get half of the Swedish GDP, 
they will on average bear half of the social loss from any redistribution to 
themselves. Their organization, if it represents them rationally, will then 
arrange any redistributions to them in ways that hurt the society as little as 
possible, and it will also stop demanding redistribution whenever the social 
costs of a redistribution come to be twice as large as the amount that is 
redistributed. Unlike the distributional coalitions considered in the Chapter 
5, encompassing organizations have an incentive to seek only efficient 
redistributions, and bargaining costs between any pair of encompassing 
organizations may not be prohibitively high. Thus the theory of efficient 
redistribution may, at times, apply to some extent in societies with 
encompassing organizations.

Encompassing organizations have been relatively more important in Sweden than 
in most other countries. The LO (especially in the 1950s and early 1960s, but 
less so now) has represented a large proportion of the whole organized work 
force. To some extent, it has been linked with the Social Democratic Party, 
which strives to control the government by itself and thereby represent a 
majority of the electorate, and is accordingly an encompassing organization. 
The Swedish Employers' Federation represents most of the business in the 
country and is similarly an encompassing organization. In The Rise and Decline 
of Nations, I argued, in a cautious and carefully qualified way, that maybe 
some of the economic growth of Sweden and of other countries with encompassing 
organizations, such as Austria and Norway, could be attributed to the unusually 
encompassing character of their organizations. In subsequent publications, I 
have examined the strong forces that, over the long run, can make encompassing 
organizations break down, or fail to act in ways that serve their clients' 
aggregate interests, and emphasized again the dangers of considering 
encompassing organizations as an ideal or reliable solution to the problem of 
institutional sclerosis.3 I also urged more research on the matter.

Some extremely interesting research along these lines has in fact been done. 
For

3 "An Appreciation of the Tests and Criticisms," Scandinavian Political Studies 
(March 1986)

59

example, in independent articles on unemployment and macroeconomic stability, 
Lars Calmfors and John Driffill,4 on the one hand, and Bradford DeLong and Lars 
Jonung,5 on the other, have found that countries with encompassing 
organizations and countries with decentralized, competitive arrangements tended 
to have lower unemployment rates than countries with the in-between arrangement 
of a dense network of narrow distributional coalitions. While also making other 
contributions, these papers greatly extend the theory in Rise and Decline and 
they are in general accord with the argument that has been used in this essay.6 
Similarly, in another study Bernhard Heitger found that growth rates were also 
higher in countries that had encompassing organizations or relatively weak 
distributional coalitions and were lower in countries with strong but narrow 
distributional coalitions.7

One likely possibility is that, as the Calmfors-Driffill, DeLong-Jonung, and 
Bernhard Heitger papers appear to suggest, Sweden is doing as well as it is, in 
part, because it has had relatively more encompassing organizations than the 
English- speaking countries, for example. I am, however, anxious not to push 
this argument too far or to oversell the idea of encompassing organizations. 
Thus I hope that any readers who are inclined to think that encompassing 
organizations are a reliable solution will examine the extensive argument I 
have offered about how encompassing organizations tend to break down over time 
or come to be decisively influenced by small subsets or coalitions of their own 
membership.8

Factors Lowering the Costs of Explicit Redistributions

While implicit redistributions are much more important and more damaging to 
economic performance than has previously been understood, there are also 
reasons why explicit

4 In "Centralization and Wage Bargaining," Economic Policy (April 1988), pp. 
14-61.

5 In "Hysteresis, the Corridor, and the Political Economy of Unemployment, 
1955-1986" (forthcoming). For a fuller analysis of the Calmfors-Driffill and 
DeLong-Jonung papers and for a fuller statement and wider tests on how the more 
general theory at issue helps explain unemployment and macroeconomic 
performance, see Michael Kendix and Mancur Olson, "Changing Unemployment Rates 
in Europe and the USA: Institutional Structure and Regional Variation," in 
Labour Relations and Economic Performance (London: MacMillan Press Ltd, 1990), 
pp. 40-67.

6 In one respect, the technical features of the Calmfors-Driffill model are 
somewhat different from the argument in Rise and Decline, since they assume 
labor cartelization at the small enterprise level as their decentralized or 
competitive polar case, rather than completely competitive arrangements. But in 
their model the cartelistic power of workers in these enterprise unions is 
sharply limited by the competition from firms producing close substitutes, so 
that in fact their argument is, as they point out, similar in spirit to mine.

7 "Corporatism, Technological Gaps, and Growth in OECD Countries," 
Weltwirtschaftliches Archiv (1987).

8 In "An Appreciation..." cited above.

60

redistributions are, sometimes, less damaging to economic performance than 
might be supposed at first glance. Some of these reasons are merely the obverse 
of the arguments about implicit redistribution, and I shall set them out before 
going on to those that are entirely different.

As we saw earlier, there are fundamental reasons why such scattered groups as 
the poor are not able to act collectively and they are not organized anywhere. 
Public programs for low-income people are, accordingly, not due to lobbying or 
other organized action by the recipients of the transfers, but are rather 
mainly the result of the sympathy and egalitarian sentiments of the electorate 
and the political leaders they have elected. Since popular moral, sympathetic, 
and ideological motives mainly inspire income transfers to low-income people, 
there is usually no desire to conceal these transfers; political leaders may 
even point to them with moral satisfaction. Egalitarian welfare-state transfers 
to low-income people (and to the aged, handicapped, and so on) are, therefore, 
open transfers out of the public treasury.

This means that redistributions inspired by the moral judgments of the 
electorate need not (and often do not) involve any monopolization or 
protectionism, such as coalitional redistributions typically entail. Nor do 
they entail conditions on government subsidies designed to create the 
impression they are intended to serve some broad social purpose - they already 
reflect the moral purposes of the electorate. As a result, the extra social 
costs of monopolization, of protectionism, and of conditions on government 
grants that appear to rationalize them are usually not a serious problem with 
redistributions that arise because of the sympathy of the electorate.

There are also totally different reasons why the explicit redistributions 
inspired by the moral concern of the citizenry sometimes have lower costs than 
implicit redistributions. For one thing, the prototypical morally inspired 
redistributions are to the poor, the aged, the ill, the handicapped, and 
fatherless children. On average, the recipients of those redistributions 
inspired by sympathy are, after all, less productive than those who are well 
off -the same traits, such as age or handicaps, that tend to provide 
entitlement to welfare-state transfers, usually also imply low productivity. 
Some recipients of transfers would not have been working anyway, and transfers 
to such people need not have any significant deadweight loss beyond that of the 
taxation that is needed to pay for the transfer. In most cases, the people who 
are most productive and whose skills and resources are also currently prized in 
the society are not, at the same time, poor.

A society can transfer funds to individuals who would not have been working in 
any case or to individuals who would, in the absence of welfare payments, be 
domestic servants or gardeners, yet remain dynamic and productive. But it 
cannot misdirect the energies of its best workers, managers, professionals, 
entrepreneurs, or corporations without serious losses. It is mainly the former 
that are the objects of the sympathy of the electorate, and mainly the latter 
that are able to overcome the difficulties of collective action.

The recipients of transfers inspired by sympathy will, moreover, normally not 
be

61

major users of intermediate goods and accessory inputs in the way many 
beneficiaries of distributional coalitions are. To obtain an increase in net 
income through redistribution of a hundred million crowns, the members of a 
coalition will normally have to obtain or carry out a policy that misallocates 
intermediate goods and complementary resources, so the firms or workers in an 
industry will obtain only a part of any higher prices or other benefits their 
organized power brings about. Society may need to spend many times as much as a 
hundred million crowns to increase the net income of the organized group by 
this amount. By contrast, when there is, through the sympathy of the 
electorate, a desire to shift a hundred million crowns to the poor, there need 
be no misallocation of intermediate goods and the like, for the poor do not 
normally control productive processes that use a lot of accessory resources, 
and the transfer to them is likely to be a fairly straightforward transfer.

The limited involvement of the poor in the productive process also means that 
aid to them does not have much impact on the innovation that is the main source 
of economic growth. As Chapter 5 argued, the regulation and complex agreements 
that are associated with implicit redistributions delay innovation and thus 
affect the rate of increase of productivity as well as the static allocation of 
resources. By contrast, explicit redistributions to low-income people usually 
affect only the existing allocation of resources, rather than the processes by 
which innovation take place.

Yet another factor tends to make the losses in efficiency and dynamism from 
egalitarian-inspired redistribution less than those from redistributions 
obtained through the capacity to lobby or to cartelize. Although the matter is 
complicated by such factors as the organized power of those who administer 
public programs for the poor, there is still a sense in which these explicit 
transfers are limited by the preferences of the electorate. These transfers 
come out of the public treasury and their magnitude is accordingly known, and 
this means that in the long run they cannot be larger than some majority in the 
electorate is willing to accept. There is no equivalent constraint on 
redistributions whose magnitude and purpose are obscured.

Although rational ignorance always works against efficiency in redistribution, 
the theory of efficient redistribution is not so far off the mark for explicit 
redistributions. Certainly, Gary Becker's contention that the political 
opposition to a redistribution rises as its social costs increases is true for 
explicit redistributions. Since the costs of explicit transfers to low-income 
people are relatively transparent, we should expect that the opposition to them 
should increase as their social costs rise. This appears to have happened in 
many countries: as the size and excess burden of the welfare state has 
increased, so has the opposition to its growth. It is even possible that the 
design of programs to aid poor people will improve over time and that societies 
will, as experience and insight accumulates, converge on levels of explicit 
transfers that take full account of both the deadweight losses from such 
programs and their moral worth.

62

A Recapitulation

Let us recapitulate the argument of this essay and try explicitly to answer the 
second question - Why isn't Sweden worse off? My first question, about why 
Sweden is not even richer, has a standard answer: that Sweden's unmatched 
degree of egalitarianism and its uniquely large public sector impair the 
incentives to work, save, and allocate resources to their most productive uses. 
There is no standard answer to the second question. Yet we must be cautious 
about saying anything very general about the Swedish economy until we have an 
answer to the second question.

Although the performance of the Swedish economy looks much less impressive now 
than it did at the end of the 1960s, Sweden's per capita income still puts it 
in the top group of countries. The Swedish economy has outperformed not only 
the underdeveloped and Eastern European economies, but also some economies that 
were once ahead of it. Even on the lowest possible estimate of Sweden's 
performance, its economy is ahead of Argentina's, Ireland's, and Britain's, but 
all these societies have had less egalitarian redistribution and relatively 
smaller public sectors than Sweden. The puzzle is heightened by the fact that 
there is no very strong tendency for the countries or historical periods with 
the largest welfare states to grow more slowly than those with less 
redistribution to low-income people.

The only serious and intellectually honest way to tackle the second question is 
with a realization that the familiar answer to the first question is largely 
true. The reasons for believing that individuals respond to incentives in the 
way economists predict - and that tax and subsidy payments as large as those in 
Sweden must bring deadweight losses - are compelling. There is further evidence 
of the decisive importance of the familiar economic analysis of incentives in 
the data and historical information on international trade that was presented 
in Chapter 2. This chapter showed that trade in manufactures in smaller 
countries provides impressive evidence about the impact of protection: no small 
country with really high protection of manufactures has been able to develop an 
internationally competitive manufacturing sector. The great quantitative 
significance of trade policy is also made clear by the pattern of rapid growth 
after there has been a great increase in the size of jurisdictions and trading 
areas.

Taken together, the unequivocal data on protectionism and economic performance 
and the ambiguous data on the size of the welfare state and growth are 
puzzling. Why does the distortion of incentives through trade policy evidently 
have so much more quantitative significance than the distortion of incentives 
through welfare state redistributions?

At first glance, the theory of "efficient redistribution" might seem to explain 
why Sweden and other large welfare states are doing as well as they are. This 
theory holds that, if the dead-weight losses from any kind of redistribution 
rise, the political opposition to the redistribution will also rise, and 
ultimately to the point where further redistribution will cease. The social 
losses from redistribution are accordingly usually

63

fairly small. Some versions of the theory of efficient redistribution assume 
that the bargaining between the groups with conflicting interests about a 
redistribution will continue until the joint gains of the groups in question 
are maximized so that the society is fully efficient.

The theory of efficient redistribution as it stands is not satisfactory, in 
part because it fails to distinguish between two different types of 
redistributions of income. There are not only the explicit redistributions that 
are at the center of debates about the welfare state, but also implicit 
redistributions. These implicit redistributions occur when a government program 
or other collective action changes the distribution of income without 
increasing the aggregate real income of the society, but the policy is 
rationalized by alleged benefits to the nation as a whole or to groups other 
than the group that seeks the redistribution. For example, protectionist 
measures or restrictions on competition that are represented as strengthening a 
national economy, but actually change the distribution of income in favor of 
the group that seeks the protection or restriction of competition, are implicit 
redistributions.

The social loss from redistributions of income arises in large part from the 
criteria or conditions that are attached to or implicit in them. If an 
individual in an efficient economy with competitive markets is given cash with 
no strings attached, the incentives of the recipient are not impaired, because 
he or she continues to have an incentive to allocate all resources to their 
most productive uses. By contrast, a redistribution that is officially 
restricted to those in some industry, occupation, or locality - or that takes 
the form of a change in relative prices - distorts the incentives facing the 
recipients of the redistribution and adds to social costs.

The redistributions that actually occur, and the ways they are carried out in 
practice, depend dramatically on "rational ignorance" - the fact that the 
typical citizen does not serve his or her interests by spending a lot of time 
studying public affairs and therefore is relatively uninformed about public 
policy. Rational ignorance makes it possible for an organized interest to 
obtain a redistribution that the majority of the electorate would not have 
tolerated had it been fully informed. Accordingly, a redistribution that can be 
made to appear to be a measure that actually strengthens a society, or that is 
so inconspicuous that it is not noticed by the average voter, can be 
politically viable, even if the recipients of the redistribution are relatively 
well off people who would not have been able to persuade the electorate to give 
them a transfer on altruistic grounds.

The capacity for collective action is found mainly in established groups and is 
stronger at upper than at lower income levels. This is because collective 
action is possible only for groups that have small numbers, like the large 
firms in concentrated industries, or have access to "selective incentives" that 
are usually available only to insiders and relatively well established gropes. 
Those groups at the bottom of society, such as the poor and the unemployed, and 
some other groups, such as consumers and taxpayers, are virtually never able to 
act collectively.

This implies that most of those groups that have the capacity to act 
collectively

64

are not in a position to obtain explicit redistributions on grounds of need. 
They must instead use their power to get implicit redistributions. Because of 
rational ignorance, they can often get substantial redistributions through 
actions and policies that do not appear to be redistributional and that appear 
to serve the society as a whole, or are so inconspicuous that they are not 
noticed by a rationally ignorant electorate. Organized interests accordingly 
prefer redistributions that are not unconditional cash transfers, but rather 
embody conditions that make them appear to have a general social purpose. 
Ideally, a coalition wants policies that change relative prices in its favor, 
and that do not involve cash transfers from the government budget. Coalitions 
that represent only a tiny part of the income-earning capacity of a society 
have an incentive to seek such redistributions even if the social costs are 
large multiples of the amount they win in the distributional struggle.

The fact that many groups are not able to organize for collective action means 
that in most cases the losers from redistributions are not able to act 
collectively. It follows that there is usually little or no bargaining among 
gainers and losers from redistributions and thus little or no tendency for 
bargaining to reduce the social costs of redistribution. This factor, and the 
incentive for organized groups to choose untransparent and inherently 
conditional redistributions with relatively high social costs, means that the 
theory of efficient redistributions is wrong for implicit redistributions.

Societies with a high density of narrow distributional coalitions have lower 
income levels and growth rates than would otherwise be expected. Small 
countries with high protection of manufactures have particularly high levels of 
implicit redistribution, since the concentrated industries behind protectionist 
barriers are able to fix prices with relative ease. Cartelized labor forces in 
these industries can also organize redistributions to themselves with 
relatively little constraint. This helps to explain the strong findings on 
international trade in Chapter 2. The aforementioned facts, along with a number 
of quantitative studies of the social costs of particular implicit 
redistributions that have been done by other economists, support the theory 
offered in this book.

Although Sweden undoubtedly loses a good deal from implicit redistribution, 
there are many reasons for thinking that it probably does not lose as much from 
this as some other countries do. Sweden's relatively low level of protection of 
manufactures, its relatively high level of economic understanding, and its 
"encompassing" organizations suggest that implicit redistributions may not 
escape social control quite so much in Sweden as in some other countries.

Usually, implicit redistributions delay innovations more and have higher 
overall social costs than explicit redistributions. This is partly because 
implicit redistributions exploit rational ignorance and cannot be transparent, 
and therefore entail conditions or criteria that restrict the redistribution to 
those in some industry or activity, which in turn distort the allocation of 
resources. The lack of transparency of implicit redistributions also means that 
they are less likely than explicit redistributions to be curtailed when their 
social costs get out of hand. In contrast, altruistically motivated explicit 
redistributions often involve some special factors that limit their social 
costs: the recipients are

65

generally not the most productive people in the society, so the misallocation 
of their time involves less social loss; they normally do not control any 
significant resource beyond their own time, so intermediate goods and auxiliary 
factors are usually not misallocated; their limited involvement in the 
productive process implies that the aid to them does not normally affect the 
rate of innovation, which is the main determinant of the rate of economic 
growth.

Although a final answer to the second question must await further research, I 
propose a tentative answer designed to stimulate the thinking and research of 
those who know much more than I do about Sweden. Sweden may well lose somewhat 
more from explicit redistribution than most other countries, but explicit 
redistribution does not have nearly as much importance for economic performance 
as might be supposed from the ideological debates. At least in many countries, 
implicit redistribution is a more important influence on the economy. While 
Sweden's losses from implicit redistribution are no doubt substantial, there is 
no reason to suppose they are as high as in some other countries. Since the 
social losses from implicit redistributions are often much greater than those 
from explicit redistributions, the economies that have exceptionally high 
levels of implicit redistribution perform relatively badly. Sweden is therefore 
able, in spite of its high level of explicit redistribution, to surpass or at 
least match these countries. As I see it, that is probably why Sweden is not, 
relatively speaking, worse off.

An auxiliary finding of the argument here is that a society can, if it has good 
policies generally and avoids redistributions that have no moral justification, 
provide decently for its poor, yet also be a dynamic and prosperous society.

Too Much of A Good Thing is Bad: Nonlinearities and Lags

I am very concerned that my argument should be balanced, fair-minded, and 
useful to thoughtful people of all political persuasions. Thus I am worried 
about the possibility that the argument in this book will be pushed too far. 
This danger can be seen most starkly by imagining that the moral concern for 
those of below-average income were to go to the point that each person with a 
below-average income would be given a transfer sufficient to bring him or her 
to the average level of income. If no one has a below- average income, no one 
can have an above-average income either: this would imply a system of taxes and 
transfers that would eliminate all inequality of incomes. And this, of course, 
would eliminate all incentive to earn income.

This extreme case is useful in reminding us that the social loss from 
redistributions of income inspired by egalitarian motives is strikingly 
dependent on how much income is redistributed. The arguments and evidence 
earlier in this essay indicate that an open, competitive society can do a great 
deal to alleviate the misfortunes of the poor without losing its dynamism. The 
sclerosis in the Western societies is mainly not the result of efforts to 
relieve destitution, but rather of other causes.

66

Yet, after some point, additional egalitarian redistribution must bring 
disproportionately large costs to society. When transfers are sufficiently 
large, taxes must be so high that their excess burdens and their adverse 
effects on risk-taking and innovation are overwhelming. Moreover, the condition 
inherent in egalitarian redistribution - that the recipient lose entitlement to 
the transfer if he or she succeeds in earning a good income - ensures that the 
social costs of the distribution of the transfers must rise nonlinearly when 
redistribution comes close to the point of eliminating all inequalities. There 
need be virtually no loss of dynamism in a society from helping the poorest 5 
percent: their misfortunes and disabilities would have limited the extent of 
their production and innovation in any case, and (if other things are right) 
there will be a cornucopia of output from the remaining 95 percent. But if a 
society tries through transfers to bring even those who are 5 percent below the 
average income closer to the average, all incomes must be about the same and 
nearly all of the incentive to produce and innovate will have been taken away. 
As society enters anything resembling this latter range, increased transfers 
must lead to wildly disproportionate losses of efficiency and innovation.

Earlier in this book I presented some merely illustrative data to warn readers 
against the commonplace assumption that the large growth of the welfare state 
overwhelms other factors influencing economic performance. Lest data that were 
offered to motivate inquiry be interpreted recklessly or in a one-sided way, I 
present some further data that point in the opposite direction in Figure 4 and 
Table 5. From the figures on the size of government and economic growth in the 
last few years, it appears that the countries with larger public sectors have 
tended to grow more slowly than those with smaller public sectors. These 
further data, coming as they do from only a few years and being insufficient in 
other ways as well, establish nothing, but they do raise a useful question. 
They alert us to the possibility that redistribution could be having greater 
social costs in more recent times, when it has been pushed a good bit farther 
than in the 1950s and early 1960s.

What about the egalitarian redistributions in Sweden today? Are they more or 
less explained by my earlier argument, showing that egalitarian redistributions 
can have social costs that are fairly small, especially in comparison to those 
arising from the redistributions obtained by well-established and relatively 
well-off organized interests? Or have they risen into the range where the 
social costs are absurdly disproportionate?

This is a not a question that can be answered from afar, and it is in any case 
a matter for Swedes to decide. Moreover, to answer this question correctly one 
would have to go into many important aspects of the matter that I have not even 
touched on here. In general, these other aspects of the matter are dealt with 
very well in the impressive Swedish literature on the welfare state, so there 
was no reason for me to go into them here. The purpose of the present essay is 
not to settle ancient controversies or to summarize the existing literature, 
but rather to introduce some fresh perspectives that may enable people with a 
detailed knowledge of Sweden to get a better view of both sides of the matter.

67

Figure 4

Given the nonlinearity that has been described, we can also see why the 
argument about time lags discussed earlier in this essay was too simple. We can 
be reasonably certain that most of the adverse effects of the levels of 
egalitarian redistribution in Sweden in the 1930s or 1950s have already been 
felt. But it is too early to know what the full effects of the higher 
redistributions of the late 1970s and the 1980s will be.

68

Table 5: Average Government Size and GDP Growth, 1980-87 (percent)

Annual GDP Growth
Government consumption
Social Security Transfers
Government Expenditure
Current Disbursement
Total outlays

Australia 2.87 18.5 9.3a 27.8 34.2a 37.3a
Austria 1.67 18.7 20.0 38.7 45.3 51.1
Belgium 1.56 17.5 21.9 39.4 51.3 53.9
Canada 2.86 20.0 11.6 31.6 41.9 45.2
Denmark 1.83 26.4 17.0 43.4 55.9 59.1
Finland 3.27 19.5 10.0 29.5 35.9 39.8
France 1.67 19.1 21.3 40.4 47.1 50.6
Ireland 2.00 19.0 15.3a 34.3 49.0a 54.0a
Italy 2.21 16.1 16.5 32.6 43.6 48.2
Japan 3.85 9.8 11.0 20.8 26.7 33.3
Netherlands 1.12 17.0 27.0 44.0 54.9 60.2
Norway 3.26 19.3 15.2 34.5 45.0 48.3
Sweden 1.79 28.1 18.3 46.4 59.8 63.8
Switzerland 2.04 13.1 13.3 26.4 30.3 30.3
United Kingdom 1.70 21.4 13.5a 34.9 43.8a 46.9a
United States 2.55 18.2 11.2 29.4 34.6 35.9
West Germany 1.46 20.1 16.6 36.7 43.8 48.0

Note: For Definitions and Sources, see Table 1. a. 1980-86.

How Bright are the Northern Lights?

I hope this essay has succeeded in conveying my conviction that a society can, 
if its policies and institutions are intelligent, prevent destitution and even 
make fairly generous provision for its least fortunate citizens, yet still 
remain a prosperous and dynamic society. If a society opens it markets to 
imports and avoids special-interest legislation, cartelization, and collusion, 
it can be innovative and prosperous even while it significantly alleviates the 
privations of its poorer citizens. At least to a degree, this same conviction 
was part of the inspiration behind the Swedish welfare state. Thus I believe 
there really are Northern Lights. They are beautiful. They can also give 
societies a rough sense of direction. But they are not bright or stable enough 
to save a society, if it

69

rushes far ahead without taking along any further sources of light, from 
stumbling into catastrophe.