Progressivity Ultimately Comes At Its Own Expense

**********

Via Raj M. Desai and Homi Kharas for Brookings (full .pdf of the working paper here)

Does an expanding middle class benefit society’s poorest? Much has been written recently about the rapid growth of the middle class as well as the rapid fall in absolute poverty (Kharas 2017; Kocharand Oates 2015; Burrows 2015). However, few studies seek to link these two trends. It is worth emphasizing at the outset that a growing middle class and a falling poverty rate are not simply two sides of the same coin; there is a large “vulnerable” (or near poor) cohort between the poorest individuals and the middle class. Additionally, the trends can be quite different. In the United States, for example, the percentage of middle-class households has steadily fallen since the 1970s, while the portion of households in the lowest income brackets has remained steady (Kochhar,Fry, and Rohal 2015). Similar trends have occurred in the European Union since the early 2000s (ILO 2015). By contrast, in sub-Saharan Africa and Latin America, most of those lifted out of poverty appear to have joined the ranks of the vulnerable rather than the middle class (Calvo-Gonzalez 2017; Chandy 2015). There, the middle class has stagnated despite reductions in poverty.

In richer nations, the industrial revolution that shaped the political identity of the working poor also gave rise to a strata of salaried professionals who helped administer the private and public services that supported manufacturing and urbanization–the so-called “middling sort” (Hunt 1996). This middle class formed political alliances with the working poor against predominantly rural elites, and was largely responsible for, among other achievements, the expansion of the voting franchise (Moore 1966; Lang 1999; Acemoglu and Robinson 2005), the legalization of trade unions (Pelling 1963), and ultimately, the creation of the modern welfare state (Hay 1975; Esping-Andersen 1990).

We examined the effect of the rise and evolution of the middle class on extreme poverty, using the World Bank’s international poverty line of $1.90 per person per day in 2011 purchasing power parity (PPP)- adjusted terms. Like the definition of poverty, the definition of the middle class used is also set in absolute terms, comprising households where per capita income or consumption lies between $11 and $110 per person per day in 2011 PPP terms–referred to as a “global,” as opposed to national, definition of the middle class (Kharas, 2017). We argue that middle-class expansion initially is pro-poor given the incentives of the emerging middle class and the working poor to cooperate on matters of social policy. As citizens join the ranks of the middle class, they lobby for programs that provide them income stability and protections against shocks (social insurance). By allying with the working poor who seek social assistance (income transfers), middle-class constituents increase their bargaining power relative to elites who seek labor flexibility and lower taxes in a competitive global economy. Over time, however, as the middle class prospers and acquires greater political influence, the balance of programs shifts increasingly toward social insurance and away from social assistance. In this way, the middle class begins to capture an increasing proportion of the benefits of social spending, leaving less for welfare services targeted exclusively at the poorest. One implication of this is that the emerging middle class has never been truly progressive, because progressivity ultimately comes at its own expense.

We attempted three separate, but related tasks. First, we investigated the effect of the middle class on the degree of extreme poverty across countries and over time. Second, we estimated the mediating effect of the middle class on poverty through social spending and through social policy choices regarding citizen eligibility for benefits. Third, we examined the effect of middle-class size on the size and distribution of benefits per capita of social assistance and social insurance programs in developing countries.

Examining cross-country, time-series (unbalanced) data covering more than 100 countries from 1870 to the present, we find a strong effect of the middle class on poverty, even correcting for country income levels, but this effect has diminished over time. Today’s rich countries had far lower levels of poverty between 1870 and 1920 than today’s developing countries have, even though per capita income levels are roughly equivalent. We also find evidence that the middle-class impact on poverty reduction is both direct and mediated through social spending, but the marginal effect of the latter declines as middle-class size expands, with social spending having no marginal effect on poverty reduction once the size of the middle class approaches 30 percent of the population. On average, half of the total effect of middle-class size on poverty is mediated through social spending.

Turning to an analysis of current developing countries since 2000, we find that the dollar benefits available to the poorest quintile within each country get larger in the presence of a larger middle class. In contrast, the size of absolute benefits to those in the top quintile (which usually includes most of the middleclass households in a developing country context) that is about three times as great as the poorest quintile, suggesting that the middle class is capturing significantly greater benefits from national social spending programs than are the poor.

Our results highlight the politically pivotal role that the middle class plays in supporting the expansion of social protection–and thus the need for cross-class solidarity–but also illustrates the disproportionate influence of the middle class in shaping a country’s social protection regime.

A major conclusion is that it is critically important to distinguish between social assistance and social insurance. In most countries, assistance programs are funded at relatively modest rates and progressivity is modest. Social insurance tends to be more universal in terms of coverage (and enjoys stronger political support from the middle class) and dispenses far larger amounts of money, but benefits are strongly oriented toward the richest quintiles. Our analysis suggests that increasing the volume of resources going toward social assistance and altering the design and regressivity of social insurance are the policy areas most likely to benefit the poor.