Initial Set-Up

To think about the ways in which inequality may be conditionally related to development (either negatively or positively), let us start with a simple model, which the standard literature employs to represent individual income (in current models of intergenerational transmission of inequality):

where k represents the assets of individual i, c is the fraction individual i consumes at period t and ɛ is some idiosyncratic shock that may affect the assets at time t.

The most general (and plausible) interpretation of this functional form is a world where there is intergenerational transmission of wealth among families or individuals. Individuals have some initial wealth k (for example, inherited from parents) that they allocate to maximize their expected wealth at the end of the period. After consuming some of their wealth, they bequeath the rest to their children.Footnote ²

Assume that, following Solow's aggregate neoclassical growth model, the world is one in which each individual invests k and where there is a common production function f (k) that is strictly increasing and concave. Assume also that, at the very beginning, and as a result of a random shock, individuals are differently endowed in k, that is, that there is some initial inequality in the distribution of assets (and income). With no credit markets in place, the expected marginal return on investment is higher for poor individuals. Hence, the average wealth of poor individuals (or families) will grow faster than the average wealth of rich families. Provided there are no shocks of the type embodied by ɛ, the wealth of every family will converge to a steady state with wealth k*. Under this very simple scenario, we will observe the two following facts at the aggregate level and over time: the average growth rate will be positive (as all individuals accumulate capital until reaching k*) and inequality will decline (and effectively disappear once they all reach k*). In fact, the highest growth rates will coincide with the fastest declines in income inequality. Overall, inequality and growth will be negatively related.

Technological Shocks and the Kuznets Curve

Consider next the possibility that at some point a fraction h of all individuals adopts a more efficient technology of production f _h(k) such that f _h(.) > f (.). This will generate a higher average income in the economy—both because the production function shifts upward for the h individuals at k* and because they start moving toward a new steady-state k _h*. This process of growth will be also correlated with a widening income distribution. Provided that the access to the new technology of production is not restricted in any permanent way, the remaining fraction of individuals (1 − h) will adopt it in due time (with, say, some temporal lag). Accordingly, all incomes will converge again. Equality will be back into place after some time.

From a dynamic point of view, we will observe a first period of both a growing average income (due to increased incomes among h individuals) and increasing inequality. That will be then followed by a second (and final) period with additional growth in per capita income (generated by the growth of the non-h individuals) and declining inequality. The insights of Kuznets's path-breaking paper on economic growth and income inequality (Kuznets Reference Kuznets1955) seem to fit this pattern. According to Kuznets, the process of economic development triggers a population shift from rural to nonagricultural sectors. Given certain assumptions about the mean and variance of income in each sector, growth leads to a rise in overall inequality. This increase should be temporary, however, as populations continue to shift to the industrial sector and the majority of individuals draw their incomes from that latter sector.Footnote ³

Notice that the key factor in this story (and one that is generally forgotten in all the existing economic models on inequality and growth) is whether the non-h individuals will be able (in fact, allowed) to adopt the new technology f _h(.) or not. If there are some barriers to entry, that is, if adopting the new technology has some costs that prevent some individuals from becoming more efficient, the process of income equalization (and, as a matter of fact, the last part of income growth) will not occur. Those barriers to entry may take several forms. They may consist in a direct prohibition (imposed by h individuals) to adopt the new technology to make sure the h sector enjoys higher (monopolistic) returns. They may derive from having thin credit markets that prevent low-income individuals to borrow to make the type of expenditure needed to adopt f _h(.).

Barriers to Entry, Complementarities, and Political Institutions

Under what conditions will these barriers be present (or absent) and thus will inequality remain in place (or disappear)? For the sake of simplicity, let us assume that access to policy (in the sense of having a decisive influence over decisions) is directly related to income so that only high-income individuals (the h individuals after the new technological shock) determine whether barriers to entry should be introduced or not. For example, individuals h impose their preferred solution by funding political campaigns in democracies. More extremely, they spend some of their resources to explicitly exclude the rest of the population from the decision-making process.

If the h entrepreneurs only care about absolute income, they will not have any incentive to block access to the new technology provided—although it leads to the gradual equalization of incomes, the spread of f _h(.) to everyone does not affect their income. This scenario matches the cyclical behavior of a typical open economy: in such an economy there are outbursts of technological change that generate faster growth rates and a broadening income gap; over time, however, individuals catch up and the economy moves back to some equality of income. (Because technological change never stops in modern economies, complete equalization never happens.)

Still, this rosy scenario of sudden growth followed by re-equilibration may not always happen. The members of the h sector will have an incentive to block or slow down the extension of new technologies of production to the remaining economic agents if they care about relative incomes, if they can capture some rents from creating non-perfectly competitive markets, if they care about governing (and their chances of governing are boosted by having more resources), and/or if having an initial technological advantage helps them to sustain some technological advantage over time (that is, if the likelihood of adopting a more efficient f (.) is positively correlated with being the first to adopt better technologies of production in previous periods).

The absence of convergence to a unique steady-state k _h* does not only happen when the h sector explicitly imposes some legal barriers to the adoption of the new technology. “Natural” barriers may also create permanent inequalities among individuals. To see why, assume that the new technology f _h only gives a higher return than f (.) if k > k. If credit markets are not developed enough, poor individuals at the time of the technological shock (those with k < k) will not be able to catch up with the h individuals. A solution to the problem would be to reform credit markets. A second solution may consist in transferring some fraction of k (in the form of money, skills, etc.) to the poor. If there are economic complementarities between h and non-h, that is, if the sector with the better technology gains from having more people adopting or knowing it, the h sector (which, so far, controls policy) will have strong incentives to devise mechanisms to overcome those natural barriers. In that instance, we should see again some equalization of conditions in place. In fact, this is the tale that Lindert (Reference Lindert2004) and Galor and Moav (Reference Galor and Moav2006) tell about the process of industrialization in parts of Europe. The technological shocks of the nineteenth century led to growing inequalities first and then to the support of industrialists for publicly financed education schemes to improve their workforce and, as a result, their own profits.

Given this general story about technological shocks and the potential emergence of legal or natural barriers, political institutions must play an important role to explain the behavior of growth and equality. As I have already pointed out, inequality will tend to persist over time if only the h individuals control the policymaking process—and there are no complementarities that may convince them to invest in those poor individuals with assets k < k. Logically, under a different political regime, one where non-h individuals decide, there will be strong incentives to block legal barriers or to establish mechanisms to reduce any natural barriers. In other words, if non-h individuals are the majority of the population, then democracy should generate more equality in the long run. (Still, democracy is neither a sufficient nor a necessary condition of growth and equalization. It is not a sufficient condition because even if there is democracy, the process of investment in poor individuals [those with k < k] will not happen if the majority of the population belongs to the h sector. It is not a necessary condition either because, as I discussed above, income convergence can happen provided the sector h only cares about absolute incomes or wants to exploit some complementarities across sectors.)

Politically Endogenous Barriers

So far we have only considered the case in which there is an exogenous technological shock and where economic agents react ex post by choosing a particular regulatory framework. However, in most instances, technological shocks (and their adoption among economic agents) do not happen randomly. Instead, precisely because they affect the income of individuals, they are endogenous to certain political configuration. Hence, as I discussed more formally in Boix (Reference Boix2009b), those sectors that control policymaking (i.e., the “decisive” voters) have a strong interest in controlling or regulating the access to the technologies of production. To put it differently, the introduction of a regulatory framework that could spur technological advances and growth will be a function of the incentives of policymakers.

Thus, for example, in a democratic setting, which is always characterized by a strong equalizing bias, voters may be tempted to block growth-enhancing technological shocks that may reduce their income or that simply increase inequality.Footnote ⁴ Moving closer to one of the main empirical facts that motivate this article, the persistence of inequality and economic stagnation in developing economies derives from the fact that their political and economic elites have very little incentives to transform their countries into open economies.

Before industrialization, wealth and income was heavily distributed on the basis of political access to power (which was, in turn, well related to military skills [Boix Reference Boix2009a]). Wealth was therefore based on the ownership of very specific assets—assets whose value hinged on regulatory barriers sustained by the state and/or wealth (such as land) that could not be easily redeployed in new, more productive economic sectors. Because wealth was so specific to the sector in which it was deployed, the wealthy strata and the governing elite (both of which overlapped to a great extent) had very little interest in changing the status quo and in accepting any technological breakthrough that could upset their dominant economic and political position. Under that political and economic context, inequality and long-run economic stagnation (underpinned by a non-democratic regime) ended up taking place simultaneously.

The same joint pattern both of differential access to the state and to regulation (and wealth) and of economic stagnation prevails in the underdeveloped world today. Although there are no good measures of differential access to economic and political power, the level of corruption may be interpreted as a proxy for all the mechanisms employed to secure personal favors and to avoid the transparency and equality nominally embodied in the legal system. Figure 1 shows the existing correlation between the level of inequality (measured through the Gini index) and the absence of corruption for a cross section of countries. Unsurprisingly, both measures are well correlated. Higher levels of corruption come jointly with more inequality.

Figure 1.

Correlation between Absence of Corruption and Level of Inequality