RAGS TO NOTHING

India’s middle class is almost impossible to define

Quartz india
Quartz india

On Dec. 21, we published an article on the complications that arise when India’s middle class is defined in relative terms. The following piece discusses the problems with defining middle class in absolute terms.

While defining the middle class in relative terms may thus gain us insights into the actual condition of the masses of people, it cannot be used to show the growth of some income group—here, a “middle class” defined by some income level—over time. For that reason, most of the studies reviewed here have taken, not a relative, but an absolute definition. That is, “the middle class is the section that earns or spends at least so many and at most so many rupees per day (or month, or year).”

And to compare the size of the middle class across different countries, the absolute definition must be in terms of a single currency. However, this project runs into some problems.

In the first place, there is no accepted definition of middle class status in terms of an income range. That is not surprising, as there is no generally accepted definition of poverty; different countries have their own, very different, poverty lines, and these are fiercely contested.  In the first place, there is no accepted definition of middle class status in terms of an income range. The World Bank’s two international poverty lines are—more or less arbitrarily—based on a per capita income/consumption in different countries equivalent to what one could buy in the US for $2 and for $1.25 per day, i.e., in purchasing power parity (PPP) dollars. The first is the World Bank’s definition of poverty; the second its definition of extreme poverty. It uses the latter definition ($1.25 PPP) to measure progress in achieving the United Nations’ Millenium Development Goals related to poverty.

PPP and poverty

We all know that, even though the exchange rate of the rupee is, say, Rs66 per $1, in fact one can buy more goods and services in India with Rs66 than one can buy in the US with $1. In other words, the purchasing power of the rupee within India is more than its external value. So, in order to compare incomes across different countries, the World Bank uses local price data from each country, and then tries to work out how much of each local currency is required to buy the equivalent of the goods and services one can buy for $1 in the US. In this way, local currency is translated into PPP dollars.

One PPP dollar is thus worth much fewer units of the local currency than the dollar exchange rate. For example, in 2011, when the official exchange rate of the dollar was Rs45, the PPP dollar was only Rs 15, i.e., in India one could buy for Rs 15 what one could buy in the US for $1—at least, that is the claim the World Bank makes. On this basis, the World Bank finds out how many people earn less than PPP $1.25, or PPP $2, in each country.

However, the World Bank’s PPP-based poverty line is untenable. First, we all know it is not possible for a person to survive in the US on $1.25 a day, or even $2 a day. The fact that such a measure is being widely used shows that the definition of poverty is not anchored in any definition of minimum human requirements. Sanjay Reddy and Thomas Pogge demonstrate (pdf) that the very concept of PPP is based on methodological errors.

The fundamental problem is that the World Bank poverty lines do not even attempt to ascertain whether the persons with incomes/consumption expenditures over these lines can achieve certain minimum requirements of dignified human existence—such as minimum nutrition, water, housing, sanitation, clothing, health care, education, and electricity. It is possible to avoid all problems of international comparison by simply calculating the rough sum (in the local currency) required in each country to achieve a common minimum set of requirements. And then to see what share of the population in fact has incomes lower than that sum.

Such an “achievements-based” exercise may find that much larger numbers than indicated by the World Bank’s measures do not, in fact, enjoy the income needed to purchase these minimum requirements; this is indeed why the number of those suffering multiple deprivations is much higher than those officially defined “poor.” Instead of the achievements-based poverty line proposed by Reddy and Pogge, the World Bank pushes a purely arbitrary money-metric international poverty line that is more or less useless for any real understanding of the well-being of people.

Even within the logic of the World Bank’s own PPPs, there remains the question of which cut-off level to choose. The very low level at which the World Bank sets its lower line—at $1.25—helps show progress in poverty reduction over time. Had the line been set instead at $2.50 (pdf), it would have shown a worsening trend over 1990-2005, i.e., an increasing share of the world’s population falling below the line over the period.

Estimates differ vastly according to definitions

Almost all the studies we looked at define the middle class in PPP dollar terms; they differ in the specific range they select as “middle class.” Broadly, they adopt one of two approaches. Some choose to set the lower threshold very low, and thus find that the middle class is already very large. A second approach is to set the lower threshold much higher, and find that the middle class is small.

Some studies in the latter group nevertheless proceed to project that the middle class will grow phenomenally in coming years, and constitute the bulk of the population.

India’s $2 middle class

The problems with the World Bank poverty line reproduce themselves in the studies in the first group. These set the lower threshold (for entry into the middle class) at $2 PPP day (i.e. above the World Bank’s definition of poverty). For example, the Asian Development Bank (ADB) takes those with consumption expenditures of $2-20 PPP per day as middle class. The World Bank pushes a purely arbitrary money-metric international poverty line. On this basis, it finds that 25% (using the World Bank database) or 38.1% (using India’s NSS data) of India’s population qualify as middle class. Indeed, 56% of the population of Asian developing countries qualify, too, making it, presumably, a “middle-class continent.”

However, the ADB does seem to have some qualms about its definition. It, therefore, divides its middle class into three segments, $2-4, $4-10, and $10-20, and finds that the overwhelming bulk are in the first segment, “leaving them highly vulnerable to slipping back into poverty due to economic shocks.” In India’s case, of the 38.1% figure, 22.3% fall in the first category, 12.3% in the second category, and just 3.5% in the third category.

The World Bank’s Martin Ravallion—the man who devised the World Bank’s definition of poverty—faces a similar problem (pdf). He too takes $2 PPP per day as the lower threshold, and $13 PPP per day, the US poverty line at the time, as the upper threshold. In other words, if we were to take the PPP concept seriously, all those who qualify in Ravallion’s “middle class” would be deemed poor in the US. He explains that he distinguishes between a “developing world middle class” and a “Western middle class”; entry into the latter is on crossing the $13 level.

On this basis, Ravallion finds that in 2005, India’s middle class was 263.7 million, or 24.1% of the population.

The merit of the study by Abhijit Banerjee and Esther Duflo (pdf) is that they try to find out what their defined “middle class” do for a living. Taking the middle class as those with a consumption of $2-10 PPP a day, Banerjee and Duflo carried out surveys of 13 developing countries to discover “what is middle class about the middle classes”.

Note that their upper threshold is $3 below the then US poverty line. Given the low bar they set for defining the middle class, it is not surprising that they find that the middle class are much like the poor:

…it is striking how much the poor in a particular country have in common with the middle class, in terms of how budgets are allocated…. At first blush, the occupational patterns of the middle class seem surprisingly similar to that of the poor…. in urban areas, the broad occupational patterns are remarkably similar between the poor and the middle class…

The businesses of the Banerjee-Duflo middle class “are severely under-capitalised, because the middle class, much like the poor, does not have particularly good access to capital.” The two economists are puzzled in the failure of their “middle class” to save from their income: “The striking fact about business investments, especially given the differences in potential to save, is how little difference there is between those of the middle class and those of the poor.” What are these middle class “businesses?”

…[T]he type of business they operate is also not very different from that of the poor. The number of employees who are not family members is still tiny: specifically, the businesses of those with daily per capita expenditures between $6 and $10 have on average only 0.5 to 1 more paid employee. Businesses owned by the middle class still seem to operate with very little in the way of assets, such as machinery or a form of transport.

Little wonder, as the Banerjee-Duflo “middle class” businesses include fruit and vegetable vending, rag-picking, selling milk, and collecting cow-dung. They find that “middle class” businesses “are also the most common businesses among those with consumption under $2 a day…” Perhaps the Banerjee-Duflo developing-country “middle class” behave like the poor because they are poor, and they are unaware that, having been defined as “not poor” by the World Bank, they should behave differently.

Finally, Banerjee-Duflo conclude that “The single most important characteristic of the middle class seems to be that they are more likely to be holding a steady job.” The instance they cite from their field trips is of a relatively better-off village in Udaipur, Rajasthan:

There are very few people who live on more than $4 per day in our Udaipur sample but we accidentally met several of them on one of our trips…. Signs of their relative well-being were apparent: a corrugated metal roof, two motorcycles in the courtyard, and a teenager in a starched school uniform. It turns out that, in the families we interviewed in the village, everyone of working age was working in the local zinc factory.

It would appear that zinc factory workers lie in the upper range of the Banerjee-Duflo middle class. Perhaps that should have alerted them to the fact that something was seriously amiss with their definition.

India’s 6% middle class

The second approach, of setting the lower threshold at around $10 PPP, is taken by a number of studies.

Some estimates of India’s middle class

Author Definition of middle class* (Consumption in $PPP) Estimated size of India’s middle class (for — year) Projected future size of India’s middle class
Kharas $10-100 per capita per day 5-10% of population (2010) 90% in 2040
Ernst & Young $10-100 per capita per day 5% of population (2010) 475 million by 2030
Goldman Sachs $16-82 per capita per day 5% of population (2008) “Vast majority” by 2040
McKinsey Global Institute $64-322 per household per day 5% (2005) 41%, or 583 million by 2025
NCAER (2004) $8.2-41 per capita per day** 5.7% (2001-02)
Meyer and Birdsall $10-50 per capita per day 6% or 70 million people (2009/10) No projection
Milanovic (2012) $10-22 per capita per day Very small (2005) No projection
Edward and Sumner ‘Global secure consumption layer’ $10-50 per capita per day Less than 1% (2010) No projection

* Except for Edward and Sumner.
** Household income of Rs2,00,000-10,00,000 per year, converted to per capita per day $PPP for 2001-02.

As can be seen from the above table, these studies estimate the present size of India’s “middle class” at 6% or less of the population. The last two, using the World Bank database rather than India’s NSS or NCAER, arrive at even lower figures than the rest. (Note: Edward and Sumner emphatically do not use the phrase “middle class”, because, as they correctly note, “class’ is a social and political identity not necessarily linked to estimates of expenditures per capita.” Instead, they refer to “consumption layers”, of which the layer “global secure” approximates the category being described by the other studies.) Using this measure, India’s “middle class” is to be found in the top 10%, or even less, of the population. 

These studies arrive at low estimates for the size of India’s middle class, despite the fact that $10 PPP per day is quite a low international cut-off for defining the “middle class”: it is the average of poverty lines in Portugal and Italy, and is considerably below the poverty line of the US. Nevertheless, using this measure, India’s “middle class” is to be found in the top 10%, or even less, of the population. “In other words,” the World Bank notes (pdf), “being classified as middle class at the global level is equivalent to being at the top of the distribution in many low-income countries.” Meyer and Birdsall conclude that:

If to be middle class is to be reasonably secure in material terms, then India’s middle class constitutes less than 100 million people, and is crowded into the top decile along with the much smaller number of “rich” households. In that sense India does not yet look much like the middle class “societies” of Latin America, let alone of the mature western democracies.

Even more emphatically, Milanovic says:

Clearly, there are millionaires in India as well as other people who are quite rich, and the same graph with percentiles (rather than ventiles [i.e., groups of 5%]) would have shown the top end of India’s income distribution to be a little bit higher, but even in that case it would not go past the global 80th percentile. So these rich Indians, as a group, barely match the average income of middle-class Americans. Note that these are indeed very large groups of people and that the averages may conceal some very high individual incomes: if I use ventiles, each Indian ventile consists of some 60 million people, if I use percentile each percentile is 12 million people. The latter figure is equal to the population of the municipality of Mumbai. But the key point is that although there are in India some very rich, and even some extravagantly rich people, their numbers are not statistically significant, and the number of people who have the standard of living of the American middle class is still very limited.

While Kharas, Ernst & Young, Goldman Sachs, and McKinsey do estimate India’s middle class at present to be small, the main focus of their studies is their projection that India’s middle class will expand very rapidly in the coming years. How do they arrive at these projections?

Typically, they take the existing distribution of consumption; they project some rate of growth of national income; and finally they assume that the income distribution will not become more unequal, so that the entire curve simply shifts to the right. Since the vast majority of people are at the lower end of the existing income distribution, the number of those spending over $10 PPP per day must at some point suddenly boom, given the assumptions made.

In passing, it is worth noting that, if indeed India were to develop a giant middle class with the monstrous pattern of consumption typical of its counterparts in the developed world, it would trigger many other political and environmental consequences for the world. Yet, none of the studies celebrating the rise of the Indian middle class feels the need to explore these implications.

It is unlikely that anyone 30 years hence will check whether or not these scenarios were borne out; and even if someone were to so check, no one gets penalised for wrong projections, so it is safe to predict almost anything fashionable today.

Also read: India’s middle class is actually the world’s poor

A version of this post first appeared in Aspects of India’s Economy, published by the R.U.P.E. We welcome your comments on ideas.india@qz.com.

home our picks popular latest obsessions search