Abhijit Banerjee and Esther Duflo in their essay 'What is middle class about the middle classes around the world?' continue a strand of thought that originated in their previous essay, 'The Economic Lives of the Poor'. Specifically they contrast the lives of 'the middle class' in a sample of countries worldwide, with the 'lives of the poor' that they discussed in their previous essay. In particular, they attempt to address the question: "Is there anything special about the way the middle class spend their money, earn their incomes, or bring up their children" (4).

The sample of countries on which they draw is: Guatemala, India, Indonesia, Ivory Coast, Mexico, Nicaragua, Panama, Pakistan, Papau New Guinea, Peru, South Africa, Tanzania and East Timor. This is the same sample that they used for the 'economic lives' paper, which makes comparison easier. They partition the middle class into two groups: households in which individuals earn $2-4 per person per day, or $6-10 per person per day (I don't know what happened to $5 - probably a survey construction issue). As a consequence of using the same survey, some of the criticisms that I lodged previously still apply, such as representing the 'middle class' as it was 15 years ago as representative of how it is now. I believe strongly that this is not the case for South Africa, and this comes through in some of the data that they present.

Unlike previously, Banerjee and Duflo attribute some of the differences that they observe to social norms, and therefore to culture. I think that this is an improvement. I understand that they need to be circumspect about introducing this kind of idea into economic analysis, but I believe strongly that doing so strengthens their analysis. Note that I am not going to discuss their definition of the middle class, I think many might attack this, but I don't think it warrants concern right now. It is sufficient to note that the $2-4 group is quite large in most of the countries, and the $6-$10 is somewhat smaller. This seems to identify factors typifying a middle class.

As I did for the 'economic lives' paper, I list some descriptions from the statistics they present:

• relative to the poor the middle class have a declining amount of expenditure on food, and they select tastier & more expensive food.
• no pattern for alcohol and tobacco - goes up in some countries, down in others.
• as share of income dedicated to food decreases, the amount freed for entertainment expenditure increases, more middle class individuals have TVs
• no clear pattern for education spending
• there is a definite rise in spending on health care
• the middle classes have better access to household infrastructure and basic amenities (electricity, water, latrine).
• rankings of expenditures between the countries for spending seems to be the same as for the poor.
• there are possibly common norms for expenditure, e.g. in SA extravagant expenditure on funerals by middle class seems to pressure the poor to attempt to emulate the middle classes
  
• middle class less directly connected to agriculture than the poor
• oddly, rural middle classes less likely to own land than rural poor
• rural middle class are local entrepreneurs
• share of entrepreneurs and share of employees seems roughly the same for poor and for middle class
• trying to get something "without a large resource commitment seems to infuse the middle class" (16)
• there seem to be differential returns to entrepreneurship for men and women: the return for men is greater - this could be due to women often trying to run businesses while engaged with child care, or other factors
• middle class businesses are undercapitalized, similar to the poor they do not have good access to capital
• however middle classes have better access to sources of credit
• middle class business seem to remain small even though health care spending rises dramatically - why do they not save more and invest in capital to increase returns?
• cannot really view the middle classes as particularly entrepreneurial, in fact here they are about as entrepreneurial as the poor
• distinct difference between poor and middle class is that the poor have erratic casual employment, whereas middle classes seem to have more regular salaried employment
• there seems to be a large element of luck in terms of admission to the middle class, for example a business deciding to locate a factory close to an individual household's home can produce a long term virtuous cycle of employment of one generation, education of the next, employment of the next and more education for the generation following.
• the middle classes tend to migrate for longer and to migrate for long term job prospects, reflecting either better searching for jobs, or better qualification, the problem is that we cannot disentangle the causality of admission to the middle class - could be better migrants get into middle class, or that middle classes promote permanent migration for jobs
• middle classes have lower fertility and substantially lower rates of mortality among the elderly, most likely as a consequence of dramatically increased health care spending relative to poor
• urban middle classes have higher education spending than other groups, includes spending on private tutors
  
• higher share of children enrolled in school for middle classes, although SA is anomalous here, which, I would argue, is a consequence of them using a survey from the end of Apartheid (1993).

Summing up:

• Most middle class characteristic: having a steady well-paying job
• The idea of a 'good job' is typically middle class
• The middle class work longer hours and more regularly than the poor
• Middle class seem to want to get 'good jobs' so that their children can be educated, or have talent revealed, and move up in the world
• Middle classes are not the 'source of entrepreneurship' we might be led to believe they are by news media and some politicians

As much as I think that the paper achieves a sense of what the 'middle class' is about, it does so in a manner which I think is problematic. I experienced the feeling that the paper should have, instead, been two papers. The first paper would have been a comparison of different middle class characteristics with the sample of countries that they used and the second an in depth analysis of the middle class in the Udaipur region of India where Banerjee and Duflo have incredibly detailed information from the surveys that they conducted. The problem is that it is difficult, in my opinion, to argue that the experiences in Udaipur, India in 2006 are necessarily representative of the experiences of rural South Africans in 1993, or, more importantly, in South Africa in 2008. A similar criticism applies to the 'economic lives' paper which had a similar dynamic of discussing the international patterns and then going into a deeper analysis combined with anecdotes in Udaipur.

That being said, the study is comprehensive and tries to cover a substantial amount of ground. I am surprised that something like this hasn't been done before. I would compare it to the recent work by Hertz et al (2008) on education inequality, which also takes a broad approach in order to come up with some common insights. Hopefully these insights can give us better tools with which to assess international policy differences and for international institutions such as the World Bank, while maintaining more nuanced approaches at the national level of policy.

References
Banerjee, A.V., Duflo, E. (2008). What is Middle Class about the Middle Classes around the World?. Journal of Economic Perspectives, 22(2), 3-28. DOI: 10.1257/jep.22.2.3

Hertz Tom, Tamara Jayasundera, Patrizio Piraino, Sibel Selcuk, Nicole Smith, and Alina Verashchagina (2007). 'The Inheritance of Educational Inequality: International Comparisons and Fifty-Year Trends,' The B.E. Journal of Economic Analysis & Policy: Vol. 7: Iss. 2 (Advances), Article 10.