The UPA government decided to make Telangana a separate state, and then decided against it, and then decided to set up a committee to study it. Since those eventful days in early December, there has been heightened interest in a very old question — do small states, or small administrative units, actually perform better in terms of governance and/or economic growth? In other words, is there an economic basis for splitting the large state of Andhra Pradesh into two? Several commentators have felt that such a case actually exists, but the evidence presented by them has been mostly of an ad hoc nature, e.g. to support their claim, they mention the small state success stories of Delhi and Haryana; but the not-so-good small state economic stories of Jharkhand or Assam or Punjab or Goa are ignored.
What size of population constitutes a small state? According to the 2001 census, the median population size of a state in India was 21 million; as reference, Telangana had a population of 30 million in 2001. If constituted, Telangana would be the 13th largest state in India, with slightly less population than Kerala and slightly more than Jharkhand. Not exactly small, and neither is Jharkhand.
Two classifications of small states were attempted for evaluation of growth performance. The first definition is the traditional one with population in 2001 being less than 10 million. The cut-off state is Uttaranchal with 8.6 million. The second definition is with the definitional cut-off at 30 million — Telangana as a small state definition.
The time-period for analysis is the near 30-year period from 1980 to 2008. Growth is measured according to state GDP per capita. Given the importance of “initial conditions” (which in Indian parlance is known as the “base effect”), a simple model of growth was estimated: growth as a function of GDP per capita in the initial year and whether it was a big or a small state. The expectation is that the lower the (log) GDP per capita in the initial year, the higher the growth rate. This expectation derives from the well-known theory of catch-up. The second expected result is that big states will grow at a slower rate because they are more difficult to manage.
First, the basic results. Whether the definitional cut-off point is 10 million or 30 million, there is no difference in the per capita growth rate of small states — average of 1.55 per cent per annum for the period 1980-2008. The large states average a slightly higher growth rate of 1.61 per cent per annum.
If now a deterministic relationship is estimated (the regression model noted above), the following results are obtained. First, regardless of definition, there is no catch-up observed among Indian states, i.e. the richer the state in the initial period, the faster its growth. Second: big states (>10 million) grow at a 0.3 per cent a year lower rate, but this result does not have statistical significance; big states (>30 million) grow at 0.5 per cent per year higher rate, and the result is very strongly significant.
The chart shows the excess growth for each state given its initial per capita level. The states are ordered according to population size. Excess growth is the growth after accounting for initial per capita income. As the chart makes clear, the excess growth of states like Andhra Pradesh, Tamil Nadu, etc. more than makes up for the slower growth of Madhya Pradesh, UP, Bihar, etc.
It is noteworthy that Andhra Pradesh, the state being sought to be divided, has the highest excess growth of all the states, including Delhi! In absolute terms, Delhi had a per capita income growth of 2.6 per cent per annum, while Andhra grew at 2.2 per cent. Given initial income (and initial income has a positive effect on subsequent growth), Andhra’s excess growth is 0.9 per cent per annum, while Delhi is a lower 0.8 per cent.
There can be valid non-economic reasons for division of a state; that UP and Maharashtra should be sub-divided into smaller units was argued by Dr Ambedkar in the 1950s. There can be valid economic reasons as well, e.g. equity, governance, growth. The evidence presented here suggests that better growth performance from smaller states is not supported by the limited evidence presented here.