Tuesday, December 13, 2011

A vulnerability approach to the definition of the middle class

The concept of social class and specifically middle class, has been widely discussed in sociology and other social sciences, but mostly ignored in modern economics. In practice, the middle class has been defined in terms of income, consumption patterns, occupational status, or even by using self-identification.

Regardless of which definition is used, the measurement of middle class is dependent on a particular period and place and it is determined by several factors, such as history, culture and the development stage of a society. Analysis of the middle class emerges as a central issue given the strong influence this social group has on society, politics and the economy. Much literature (Easterly 2001) suggests that the middle class helps to produce economic benefits and foster economic development, through its emphasis on human capital investment, consumption and savings, which, in turn, incentivizes a virtuous circle contributing to further expansion of this social group. Other authors (Birdsall 2010) suggest that the middle class constitutes the backbone of democracy ensuring social and political stability by fostering social cohesion and mitigating tensions between the poor and the rich. As political scientists suggest, a strong and stable middle class is usually accompanied by a more ―stable democracy.

Over the past decade several countries in Latin America have accomplished important reductions in poverty and inequality and as a result policymakers are now confronting new challenges, and seek to understand problems they face from a political economy perspective. The simultaneous emergence of improved data quality and availability has sparked a surge in literature addressing measurement and dynamics of the middle class. However, most writing on the subject lacks clarity on the definition of what it means to be middle class. In the sociological literature there is a long tradition of class dynamics analysis based mainly on occupational structure following Goldthorpe‘s categories (Goldthrope 1987).

However, in the economic literature, the analysis has focused mainly on relative definitions, addressing a stratum of the income distribution rather than an analysis of class. Existing relative definitions compare different middle classes from place to place because income distributions differ across countries. An absolute approach becomes more advantageous because it identifies middle class as those households with income or consumption in a specific and comparable range. For instance, Banerjee and Duflo (2008) (B&D from here onwards) and Ravallion (2010) have suggested the use of absolute income thresholds to define lower, middle, and upper classes. However, while these absolute measures enable comparison across countries their definitions have resulted in descriptive statistics of income groups because the thresholds are defined arbitrarily.

The response to the question of why we care about the measurement, the analysis and the empirical contrast of different definitions of the middle class can found somewhere else (Birdsall et al. 2011; Hertova et al. 2011; Cruces et al. 2010). In line with Amartya Sen‘s (1983) statement: ―poverty is absolute in the realm of capabilities but relative in the realm of income, we propose a framework in which middle class is absolute in terms of the functionings that define it but relative in terms of the means through which those functionings can be achieved. In this paper we argue that vulnerability to poverty is the absolute functioning that defines the middle class. We set the lower threshold of the middle class at an absolute level: a 10 percent probability of falling into poverty. Using a regression-based approach we exploit panel data to determine the amount of comparable income associated with that probability level –using income as the relative measure of vulnerability to poverty. Based on our findings from applying this methodology to three countries, we set an absolute lower bound for the middle class of 10 dollars PPP. This creates an absolute lower threshold that can be used to measure the middle class across countries over time. We, then, apply the 10-dollar PPP absolute lower threshold to cross-sectional surveys (household data from 1992 through 2008-09) to measure the size of the middle class in Chile, Mexico and Peru and analyze its evolution over the last two decades. Lastly we contrast this new absolute definition with existing absolute definitions of the middle class presented by B&D and Ravallion.

Evidence from the application of the vulnerable-to-poverty approach shows that both the proportion of middle class households and the income share appropriated by this group have significantly increased in all three countries during the period under study –in-tandem with economic growth in these countries as we should expect. We also find that the proposed lower threshold of middle class is at or above the median of the income distribution in the countries analyzed. We show that alternative absolute definitions (B&D and Ravallion) lump people who are still vulnerable to poverty into their definition of middle class, which has resulted in counter-intuitive trends in the size of the middle class –the middle class expands during economic downturns and shrinks in times of growth– a problem the vulnerability-to-poverty measure corrects.

In summary, the vulnerable-to-poverty approach defines the middle class in absolute terms as those households with income or consumption in a specific and comparable range, and it makes two important contributions to the measurement of the middle class. First, methodologically, it ensures that no lower class or poor households are being identified as middle class. Second, conceptually, it establishes a well-defined conceptual framework for the analysis of the middle class over time.

This paper is structured as follows. Section 2 provides a brief literature review on the relationship between vulnerability and middle class. Section 3 presents an overview of existing measures of middle class and their shortcomings. Section 4 describes the data used in this paper. Section 5 describes the empirical strategy to estimate an absolute-standard for middle class analysis. Section 6 presents the results and contrasts it with other existing absolute definitions. Finally, section 7 concludes.

World Bank.Author: Lopez-Calva, Luis F. ; Ortiz-Juarez, Eduardo.Document Date:  2011/12/01.Document Type:  Policy Research Working Paper.Report Number:  WPS5902 Volume No: 1 of 1.