THE EFFECT OF ECONOMIC GROWTH ON SOCIAL STRUCTURES

THE EFFECT OF ECONOMIC GROWTH ON SOCIAL
STRUCTURES 1

François Bourguignon
The World Bank


Introduction..................................................................................................................... 2
1. Statistical Relationships between Growth and Social Structures ............................... 6
2. The Effect of Economic Growth on Social Structures : theoretical considerations . 11
(a) The Sectoral Shift View ...................................................................................... 13
(b) General Equilibrium Models of the Distributional Effects of Growth............... 16
(c) Non-linear Savings Behavior ............................................................................. 22
3. The Effect of Economic Growth on Social Structures : empirical evidence ............ 28
(a) The Sectoral Shift Effect of Growth on Social Structures .................................. 28
(b) Effect of Growth on Inequality between Socio-economic Groups .................... 32
(c) Effects of Growth on Inequality among Individuals ........................................... 39
4. Conclusions.............................................................................................................. 49
(a) The Foremost Importance of Sectoral Shift Phenomena.................................... 49
(b) The Role of Market Integration........................................................................... 50
(c) Social Costs of Transitory Adjustment................................................................ 51
(d) Remarks on the Effect of Growth on Social Relations and Institutions.............. 52
REFERENCES ............................................................................................................. 55





October 23, 2004

1 Prepared for the Handbook of Economic Growth edited by Philippe Aghion and Steve Durlauf. This
paper was finished, with great difficulty, after I joined the World Bank as Chief Economist in the summer
2003. I am grateful to Philippe Aghion for useful comments. I thank for their patient and efficient help
Jean-Jacques Dethier and Victoria Levin. Views expressed here are essentially personal and do not engage
the World Bank in any manner.

1

---

INTRODUCTION

If economic growth actually resembled the ‘extended reproduction’ coined by Marx and
implicit in the steady state regimes of many contemporary growth models, one would not
expect growth to have major social consequences. All economic magnitudes, including
the standards of living of individuals or social groups, would be kept in the same
proportion to each other so that only the scale of the economy would be changing over
time.

Of course, economic growth is something more than a mere uniform change of scale of
economic magnitudes. For a host of reasons, it is in the very nature of growth to modify
economic structures and, because of this, to affect social structures and social relations.
For instance, growth may modify the sectoral structure of an economy leading firms in
one sector to close down and firms in other sectors to be created or expand. Growth
modifies the structure of prices, thus affecting the standard of living of households in a
way that depends on their consumption preferences. In other cases, growth will call on
some particular skills, increasing the remuneration of those endowed with those skills and
also, possibly, their decision-making power within society. Finally, growth may reduce
the availability of public goods like clean air or water, requiring public intervention in
order to maintain the adequate supply of environmental goods. In all these cases, it is not
only the economic structure – i.e., the relative importance of sectors, labor skills,
remuneration of factors, and size of the public sector – that may be modified by growth.
It is also the whole social structure, that is the relative weight of socio-economic groups
or the way in which individuals define themselves with respect to the rest of the society,
that is affected. As a consequence, social relations that govern how individuals in a
society interact with each other through explicit or implicit rules may also be modified by
economic growth and may in turn affect the growth process itself.

Rough evidence of such changes is provided by simple comparisons of economic and
social structures and institutions across countries which have reached different levels of
development. At the risk of caricaturing, it is sufficient to compare poor sub-Saharan

2

---

African countries today with some highly developed countries in Europe, North America
or in the South Pacific. At one end of the spectrum, one observes largely rural societies
dominated by household farms, few wage workers except in the limited urban sector,
social protection ensured by an extended family system and a relatively small public
sector often controlled by an unstable oligarchy. At the other end, one finds almost
exclusively democratic urban societies with salaried employment and private ownership
of capital as the main economic organization, with sophisticated redistribution system run
by governments the size of which is 3 to 4 times that observed in poorer countries.

It is tempting to attribute all of these differences to economic growth and to expect that
growth in the poorest countries will progressively make them comparable to developed
countries today. Unfortunately, things are not so simple. In particular, it is clear that
differences in economic and social structures and institutions cause differences in the
pace and structure of economic growth as much as they are caused by it. But, it is also the
case that some other factors may be influencing simultaneously both the process of
economic growth and social structures and institutions. To take an example, a longer life
expectancy due to technical advances in the field of health is likely to modify social
structures through the aging of the population but it is also likely to modify economic
behavior and the growth process, for instance because higher saving rates are rendered
necessary by the prospect of longer periods of inactivity. In turn, this effect on economic
growth rates and on the level of development may affect social structures and institutions
by changing the weight of particular sectors in the economy.

The effects of economic growth on social structures are more complex than what the
reduced form regressions found in recent literature may suggest. The relationships are
likely to be non-linear (as hypothesized, for instance, by Kuznets for income inequality)
and to depend on several country characteristics, including policy and institutional
variables. This chapter argues that simple statistical methods are unable to identify these
forces and these interactions, and that the limited number of observations available is a
serious hindrance for this identification. Under these conditions, the only methodological
approach able to identify the social consequences of economic growth is of a structural

3

---

nature. It first requires establishing hypotheses about the channels through which
economic growth may affect the social structures under analysis. These hypotheses
should then be empirically tested, provided of course that the data necessary to do so are
available.

The purpose of this chapter is to examine, among the changes in social structures that
may be observed along the growth path of a country or when comparing countries at
different levels of development, what may be considered to be the direct consequences of
economic growth. Other changes, thus, have to be considered as autonomous or possibly
caused by factors or initial conditions that may have also affected growth but that have
only an indirect relation to growth. An important example of such autonomous changes
would be technological progress.

As noted above, some of these direct social consequences of growth may affect the pace
and the structure of future growth and thus feed back into themselves through various
channels. For instance, growth may under some circumstances generate more inequality
in the distribution of economic resources, this increased inequality in turn affecting the
dynamics of the economy. This chapter focuses on that part of the circular argument that
goes from growth to social structures and ignores the other side of the circle. Such a
choice is made for reasons of analytical expediency. It turns out that the economic
mechanisms that lie behind the two parts of the circle are quite different; it would be too
ambitious a task to deal with them simultaneously. Readers interested in the effect of
social structures and institutions on growth should refer to other chapters in this
Handbook.

The social consequences of growth may be of diverse nature. A natural distinction to be
made is between the consequences of growth for ‘social structures’ and the consequences
for ‘social relations' and 'social institutions'. As suggested by Kuznets (1966, p. 157-8),
changes in social structures have to be understood essentially as the differential effects of
economic growth on predetermined social groups. For example, urban skilled workers
may benefit more from economic growth at some stage than unskilled workers, rentiers

4

---

more than farmers, men than women, or young people than older people. Growth may
also affect the size of those various groups. Some substantial proportion of people
migrate from the countryside to the cities under the pressure of urban growth, or more
people may be willing to acquire a secondary or tertiary educational level. In both cases,
social distances between individuals are modified. Changes in social institutions may
result from changes in these social structures or from autonomous forces. For instance,
changing the weight of specific socio-economic groups within society modifies the
dominant mode of social relations, and changing the economic distance between
individuals may modify the way they interact.

This chapter concentrates on the consequences of growth for social structures rather than
on social institutions or relations. Its ambition is to identify the role played by economic
growth in observed changes in social structures and to disentangle it from other factors.
Casual comparison of social structures in developed and developing countries shows
obvious and enormous differences. However, just because these two country groups
differ by their mean income level does not imply that observed social differences must be
exclusively attributed to economic growth per se. There may be many other reasons for
these differences. In particular, it is possible that initial or historical conditions are
responsible for some specific social evolution and for a particular growth path in a given
country. It is also possible that exogenous forces, such as technical progress, have a direct
specific impact on social structures, on the one hand, and on the pace and structure of
economic growth, on the other. Analyzing the social consequences of growth consists of
trying to isolate somehow the pure 'income effect' in the evolution of social structures.

This chapter is organized in three parts. The first part introduces the topic by examining
the nature of the statistical relationships existing between social indicators and
development across countries and/or across periods. It illustrates the differences in social
structures associated with differences in income, but it also shows the difficulty of
obtaining precise estimates of the size of the 'income effect' from this kind of evidence
and the need to rely on more structural analyses. The second part reviews theoretical
models of the effect of economic growth on social structures, with an emphasis on

5

---

several dimensions of social differentiation and on economic inequality. Finally, the
third part focuses on the empirical evidence in support of this structural view of the
consequences of growth for social structures.

1. STATISTICAL RELATIONSHIPS BETWEEN GROWTH AND SOCIAL STRUCTURES

One may think of literally thousands of aggregate characteristics of societies showing
extremely high degrees of correlation with indicators of economic development, either
when comparing different countries at different levels of development or when analyzing
the evolution of a single country over time. Collecting all existing results of this nature in
the economic and non-economic literature is beyond the scope of this chapter.2 Moreover,
it is not clear how informative these correlations are from the point of view of causality.
This section aims at showing that even the most sophisticated statistical techniques for
the analysis of the relationship between socio-economic indicators and the level of
development are unlikely to permit identifying the desired causality link between them.
Given the available evidence, identifying that link requires dealing implicitly or explicitly
with structural models, rather than with the reduced form models behind correlation
analysis, whatever the degree of technical sophistication of that analysis.

As an example of the correlation approach to the consequences of growth, Table 1 shows
the relationship between the level of economic development and a few indicators that
very roughly describe changes in societies' economic and social structure generally
associated with economic growth. As it will be seen later in this chapter, these indicators
describe important channels through which growth and development may modify social
structures. They include the size of the government, the level of urbanization, education,
health, demographic patterns, labor force participation, gender differences and income
inequality. The first three columns of the table report the results of a simple regression of

2 For an early comprehensive attempt of this type, see Adelman and Morris (1967) who argue that
development is a complex multi-causal process explained by many interactions between social, economic,
political and institutional variables and use factor analysis to reduce the large number of explanatory
variables into a small number of key categories. Zhang, Johnson, Resnick and Robinson (2004) present a
typology of development strategies applying the same technique to Sub-Saharan Africa.

6

---

these indicators on GDP per capita expressed in 2000 US dollars after correction for
purchasing power parity. The first two columns are based on pure cross-country
observations – observed country means for the 1970s and the 1990s - whereas the third
one is based on a pooling of all data available across countries and years during the
period 1960-2002.3

It can be checked there that, with two exceptions, all indicators appear to be significantly
and strongly correlated with economic growth. For instance, focusing on the pooled
regression, the GDP share of public expenditures is shown to increase by 0.5 percentage
point when GDP increases by $ 1,000 (thus confirming 'Wagner’s law') although this
coefficient is not statistically significant for the 1970 cross-section. Likewise, the
urbanization rate is shown to increase by 0.3 per cent and the literacy rates by 3 to 4
percentage points in presence of the same increase in income per capita, whereas fertility
decreases by 0.15 children; the 1970 cross-section shows slightly different results in all
of these cases. As a final example, income inequality and female gender bias appear to be
significantly and negatively correlated with growth. In the case of the former, a parabolic
regression exhibits the familiar inverted U-shape introduced by Kuznets some 50 years
ago 4– with a non-significant result occurring with the 1990 cross-section.

<Table 1 around here>

All these results are interesting. Yet, there are various reasons to think that simple
regression on a cross-section of countries or even on a pool of cross-section time-series
observations is a very crude approach to identifying the consequences of growth. On the
one hand, the existence of a correlation does not say much about the causality link
between two variables. Causality may be direct in either one direction, or possibly in
both. It may also be indirect and simply reflect the fact that the two variables under
scrutiny are both related to a common set of other variables. On the other hand, GDP per

3 All data are from World Bank's SIMA database (World Bank 2003).

4 On the basis of historical data, Kuznets (1955) proposed the hypothesis that income inequality tended to
increase in a first stage of economic development and to fall in a second one. See the discussion in section
2 below.

7

---

capita tends to increase more or less regularly over time so that there may be a confusion
between its effect on socio-economic indicators and that of other variables with a
comparable time trend.

Alternative econometric specifications permit taking into account some of the preceding
points. At the same time, however, they often modify the order of magnitude and the
significance of the preceding relationships. In a few instances, they even modify their
direction.

The next four columns of table1 show estimates of the growth sensitivity of socio-
economic indicators obtained with alternative econometric specifications. In all cases, the
sample is obtained by pooling country data over various years in the periods 1960-2002.
In column 4, a set of year dummy variables is added to the regression. This accounts for
the fact that socio-economic indicators might evolve over time under the influence of
some common factors independent of national economic growth. In column 5, it is a set
of country dummy variables that is introduced so as to control for 'fixed effects' or, in
other words, the effect of largely unobserved fixed country characteristics that might
affect both the original level of GDP per capita and that of the indicator under scrutiny.
The corresponding estimate of growth sensitivities thus abstracts from differences in
country means and takes into account only differences in the average time behavior of
GDP per capita and socio-economic indicators across countries. Column 6 combines both
approaches by abstracting from differences in country means as well as from an
exogenous non-linear time trend common to all countries. Finally the estimates in
column 7 are obtained by running the simple regressions of socio-economic indicators on
GDP per capita in decadal differences.

Adding a common non-linear time trend to the original simple model does not modify the
growth sensitivity of the socio-economic indicators in a significant way. More substantial
changes are obtained when fixed country effects are introduced. As could be expected,
growth sensitivity generally falls when cross-country differences are ignored, or more
exactly when cross-country differences are attributed to fixed characteristics that include,

8

---

inter alia, initial development levels. The effect of growth on the urbanization rate, the
literacy rate or life expectancy is divided by about 2 The growth sensitivity of the GDP
share of public expenditures becomes non-significant, the same being true of income
inequality, both with the linear and with the parabolic model. The only exception is labor
force participation of women, the effect of growth on which tends to increase when
controlling for fixed effects. Changes with respect to simple estimates are still bigger
when fixed effects are introduced both for countries and years. In some cases – as for
instance with fertility and gender life expectancy differential- the sign of growth
sensitivity is even reversed.

Of course, such a correction of the original estimates may well be excessive. Adding a
time trend is certainly bound to reduce growth sensitivity estimates, especially when
estimation abstracts from cross-country differences. The time trend is likely to pick up
those changes in the indicators which are independent from country specific economic
growth. Yet, results obtained with that method are not always very convincing. In
particular, that fertility would significantly increase as a response to growth once
independent forces are taken into account, as shown in the last two columns of table 1,
seems to be in contradiction with the intuition of most demographers.5 The results
shown in table 1 are likely to mask some heterogeneity among countries with respect to
the drop in fertility that is not dependent on economic growth.

The estimates reported in the last column of table 1 confirm the preceding results.
Restricting the analysis to correlation in decadal differences overall shows the same order
of magnitude for the growth sensitivity of socio-economic indicators, but also makes
those sensitivities often non-significant.6 In comparison with simple regressions and
correlations, estimates based on differences or on fixed effects thus suggest that the
evolution of the few general socio-economic indicators considered in table 1 probably
obeys other forces in addition to economic growth, or that the effect of growth is less

5 See for instance Easterlin (1996) chapter 8, and Lee (2003).
6 The difference in T-statistics between columns (6) and (7) is mostly due to the fact that the regression in
column (7) relies on fewer observations. Note, however, that ignoring possible correlation in the residuals
of the regressions behind column (6) for contiguous years may tend to a gross underestimation of the
variance of the estimates.

9

---

simple than implicitly assumed in these statistical models. In particular, it is quite
possible that the effects of growth on socio-economic indicators are strongly
heterogeneous across countries. Identifying that heterogeneity or the forces other than
economic growth that affect the evolution of socio-economic indicators is thus necessary
in order to identify the true social consequences of economic growth. But the
econometric approach illustrated in this section is unlikely to meet that objective.

Taking into account this heterogeneity across countries and going beyond the simple
statistical techniques used to produce the results of table 1 meets a fundamental
constraint: the limited number of observations. Estimating the preceding model country
by country on a time series basis would certainly permit to fully account for country
specificity. But it would only inform on the consequences of growth at a particular stage
of the development process of a given country. On the other hand, taking into account
observed heterogeneity by interacting growth rates or development levels with a host of
country characteristics and policy variables and by introducing non-linearity is also
bound to run into too few degrees of freedom. Social consequences of growth take time
to show up, so that the informational content of annual time series is not proportional to
the length of these series. In table 1, one can see that there is little difference between the
last two columns even though column (6) relies on full annual series whereas column (7)
uses only decadal differences. This means that the information that matters is not year-to-
year fluctuation but 'episodes' of growth characterized by uniformly high, moderate or
low growth rates. If this is the case, then available data may make it difficult to estimate
with satisfactory precision the observed heterogeneity in the consequences of growth.7

In summary, the analysis in this section suggests that other forces than economic growth
are behind the time evolution of most socio-economic indicators, even though the
absolute value of simple correlation coefficients is often very high. It is also possible that

7 Another approach to estimating the growth sensitivity of social indicators would be to estimate jointly
the difference equation in column (7) and the level equations in columns (1) and (2). Resulting estimates
would simply be midway. Working with annual series, it would also be possible to explicitly introduce
some lag in the effect of growth on social indicators and to use GMM-based Arellano-Bond or Blundell-
Bond 'system' estimates (Arellano and Bond 1991, Blundell and Bond 1998) instead of the fixed effect
model (6). Given the proximity of the estimates in columns (6) and (7), the lag is likely to be quite long and
the overall sensitivity not very different from the estimates shown in these columns.

10

---