Monthly Archives: August 2013

Measuring States: Money from Delhi

Given the plan outlay for each state is the primary mechanism through which states get money from New Delhi, one’d assume some predictability in this allocation. A reasonable guess would be that the outlay is either benchmarked to population or to the representation each state has in Parliament. Or even worse, maybe one’d expect a combination of both these factors, resulting in a situation similar to Tamil Nadu’s over-representation in the Lok Sabha. In which case the cluster of states benefitting from this outlay will be similar to our earlier classification.

But the data seems to suggest a level of arbitrariness that is very difficult to explain. While Andhra Pradesh, Karnataka, Gujarat and Uttar Pradesh have relatively high allocation, states similar and more suitable in profile to each of these both from a political perspective and a development view point — Tamil Nadu, Madhya Pradesh, Rajasthan and Bihar — have relatively low allocation. The chart below shows this distribution from 2008-2012.

Outlay Plan

Within the four southern states for instance, Kerala and Tamil Nadu have far less of an allocation compared to Andhra Pradesh and Karnataka. This is neither explained by per capita allocation nor by representation in Parliament. Development parameters and tax base don’t seem to justify the divergence either. It’s also not as if states that have opposition parties in poweer cluster together. For instance, Gujarat has a relatively high allocation while Madhya Pradesh and Rajasthan have much lower numbers. Rajasthan has had a Congress Chief Minister during this period while Madhya Pradesh has had one from the BJP. Likewise, Uttar Pradesh’s extremely high allocation is in contrast to a similarly poor, overpopulated and backward state in the Gangetic plains: Bihar. Other political considerations, such as a state’s MPs being crucial in propping up the UPA also don’t seem to cluster either. Gujarat and Karnataka have mostly BJP MPs who seem to have managed high allocation while Kerala, Tamil Nadu and Rajasthan have(/had) MPs crucial to the UPA and still these states have a thin slice of the pie.

In this regard, the relative efficiency of each state in utilising the allocation is another interesting data point to look at. After all, in this arbitrariness, a simple rule might then be: if a state is unable to even spend the money allocated to it one year, maybe allocating that state more money the next year is a bad idea. But that precisely is what is happening in the time period considered. Andhra Pradesh for instance returns a huge amount of its already huge allocation back to the central government; only to get an even bigger allocation the next year which it again returns the following year. Surprisingly, Uttar Pradesh, a state not known for an efficient bureaucracy or optimal resource management, seems to have at least spent all the money allocated to it and then some. Gujarat, another state with a relatively high allocation has an uneven record of utilisation and is possibly closer to Andhra Pradesh than it is to Uttar Pradesh. The chart below shows the fund utilisation for the period under consideration for each state.

Plan Utilization

Rajasthan and Tamil Nadu seem to consistently spend more money than what’s allocated to them. Surprisingly, a state that’s so advanced in every single development metric measured so far, Kerala, does not even manage to spend the little money it is allocated. Maharashtra, much like Andhra Pradesh, appears to be another huge drain on the resources of the country. The state gets a disproportionate share of the allocation and returns the funds the next year due to its inability to spend the money.

Perhaps there are other reasons for these allocations being skewed as they are. However, the source of the data does not seem to make those reasons explicit.

PS: The formula for allocation seems to be a modified version of this. Obviously that clarifies why Maharashtra has a high allocation given its tax base. But Andhra Pradesh and Karnataka?

Measuring States: Wages & Growth

The wages for casual labor in the various states, taken together with growth per capita, is another measure of how these states are really doing in terms of economic activity. The three charts below show the rural and urban wages for men and women along with growth rates of the last decade.

Rural Casual Wage

Urban Casual Wage

Growth Per Capita

The only significant data point that had predictive power towards daily casual wages was the state’s literacy rate.

Measuring States: Basic Services

Consider other light house type services that no one disputes the state has an obligation to provide. Such as, immunization of children, sanitation and availability of drinking water. In all these areas, the impact of a state government is direct, relatively straightforward to achieve and critical to the population’s well being. Here is how India’s states measure up.

ClaimsSureToiletsWater

Kerala, Tamil Nadu and Himachal Pradesh are consistently good in every measure we have looked at so far. In certain measures, the relatively richer states of Punjab and Haryana do just as well. The only other exception is Maharashtra — which has had a long history of agitational politics that still seems to have some effect. For example, Maharashtra is the state with maximum prevalence of public latrines well ahead of the second best state, Tamil Nadu. One assumes such a basic service requires a certain grass roots agitation that can be ascribed to Dalit activism; maybe there are other reasons.

A simple hypothesis that one is tempted to arrive at from the data so far is: it does not matter who is in power for a state to demonstrate good governance; what is important is the strength of the opposition. All the three high performing states of Kerala, TN and HP have not had a single government voted back to power in the last two decades. While the most obvious parameter common to the poorly performing states of Bihar, MP and Orissa is long periods of one party rule. UP’s atrociously poor indicators are despite a robust opposition while Rajasthan’s poor indicators seem to be improving at a pace better than the other comparably worse-off states.

Measuring States: Education

Outside of emergency situations, basic literacy is possibly the biggest responsibility a state government has. It’s also something that has a straight-forward causal mechanism to achieve the end unlike say growth rates or jobs created. In a country where basic literacy is still an issue, it should possibly be a threshold to cross before any state touts its “model”. Literacy rates of the population also has a high correlation with the casual daily labor wage and so it is likely to have an impact growth rates as well.

There are some complexities in measuring literacy and ranking states. For example, a state that was historically mediocre and had a poor literacy among adults will show a slower growth in improving the absolute numbers given illiterate adults can’t vanish regardless of a state educating every single child born from that point forward. A very useful metric to measure the “performance” of a state then becomes ‘how many teenaged children are illiterate?’ After all, these children must have been in school in the past 10 years which is therefore a good measure of the most recent state administration’s commitment to education. Another complicating factor in education is that it has a sticky effect — illiterate people are more likely to have their children remain illiterate than literate people. Therefore, along with absolute teen illiteracy, the distance from the overall population’s illiteracy to teen illiteracy appears to be an even better measure of the “effort” that state government undertook. Another minor matter of shameful detail is that there are gender disparities. So, all measurements are made within each gender. Here are two charts that attempt to measure the effort expended by each state in the last decade to educate its children.[1]

male effort      Female Effort

The data clearly shows some states having had a bad start and continuing to do poorly. For example, Bihar’s general female population and teen female population are both shockingly illiterate. Kerala as a highly literate state had very little extra to do. The real success stories are Tamil Nadu and Himachal Pradesh in terms of absolute numbers. However, other states that have done quite well relative to where they were include Jammu & Kashmir, Assam, Punjab, Uttarakhand, Chhattisgarh and Rajasthan. Jammu & Kashmir however has a horrible gender disparity problem in its achievement.

The two states that had a relatively good platform and still did not do as well as one would have expected them to do, happen to be Maharashtra and Gujarat. For example, Gujarat’s overall population has a better literacy rate compared to Tamil Nadu among men. But Tamil Nadu has done such a spectacular job educating its children that there are 6 times more illiterate male teens in Gujarat compared to Tamil Nadu. The picture is similar though significantly less severe in Maharashtra. Maharashtra probably did just enough to not appear as bad as Gujarat in terms of not squandering a head start. If comparing these states with Tamil Nadu is unfair, they don’t seem to do better even when compared with Haryana — both in absolute terms and in terms of gender disparity in literacy. The one state that has done reasonably well but has puzzlingly poor absolute numbers is Andhra Pradesh. With relatively good indicators in Health, the state has pathetic ones in education.

[1] — Data was obtained the same way as it was in the previous exercise.

Measuring States: Thought Experiment, Step 1

Dr Lawrence Goldman appeared in a Radio 4 podcast a while ago that the BBC has decided not have the archive available for. In it he mentioned something along the lines of ‘the cost of a more equal society is the lack of excellence.’ That may be a point of contention when the question of basic living does not arise between two societies such as Europe and USA. Within the Indian federal experiment, no state can claim that yet. So, a basic test of where you’d rather be born is a powerful metric to measure the polity and governance of these states.

Much like the thought experiment earlier for newly formed states, let us consider where would you rather be born if you were the mean, median and mode of each society(state) under consideration. Before one is even born, the true measure of well being in such an overpopulated country like India is possibly not being born at all. So, on that score, a low fertility rate is possibly the first metric to look at. The states[1] that have fertility rates below replacement are,

State Fertility Rate (2011)
Tamil Nadu 1.7
West Bengal 1.7
Kerala 1.8
Andhra Pradesh 1.8
Maharashtra 1.8
Punjab 1.8
Himachal Pradesh 1.8
Karnataka 1.9
J & K 1.9

This list assumes significance in multiple ways as it directly has consequence on political representation.

Now that you are going to be born, the next set of measures you may want to look at is whether you and your mother won’t die in the process. So, let’s look at IMR and MMR. The top 6[2] states you’d rather be born are,

State IMR (2011) MMR (2009)
Kerala 12 81
Tamil Nadu 22 97
Maharashtra 25 104
Punjab 30 172
North East 30 NA
West Bengal 32 145

The question here may be, what has the government done to improve my chances? Because it’s only logical for a person to be born not only in a good but also rapidly improving state. To achieve that, let’s consider the IMR values from 2005 as benchmark and then look at which states occupy the top 5 positions,

State IMR (2005) IMR (2011) Improvement %
Tamil Nadu 37 22 40.54
Punjab 44 30 31.82
Maharashtra 36 25 30.56
Karnataka 50 35 30.00
Bihar 61 44 27.87

By this yardstick, Kerala naturally falls way behind at position 21. It improved its IMR from 14 to 12 in this time period — a low enough starting point where improvement is very difficult. But what this also points to is the fact that Tamil Nadu has a clear divergence in terms of its improvement metric even amongst the best performing states despite its low absolute values. Albeit, still twice that of Kerala.

The way low IMR and MMR is achieved is by providing pregnant women care before, during and after child birth. So, the top five states on that regard are,

State Percentage of Child Births with skilled assist (2005-6) Percentage of at least 1 ANC Visit (2005-6) Post natal check up (2005-6) Care Score
Kerala 99.4 94.4 87.4 281.2
Tamil Nadu 90.6 98.6 91.3 280.5
Andhra Pradesh 74.9 94.3 73.3 242.5
Karnataka 69.7 89.3 66.9 225.9
Maharashtra 68.7 90.8 64 223.5

The care score is a simple sum of the three measures to rank the states[3]. So, as the data above suggests, if you have to be born in India and are not privileged enough to be granted Kerala pick Tamil Nadu.

 

[1] – This is from a list of states. Union territories have not been considered.

[2] – North East as a group has variance within as well.

[3] – There must be better way to do this and I’ll be glad if someone can help with an index for the next steps of this thought experiment.

Note on Data: the data above is from the NSSO and MHFW Reports that Krish Ashok helped extract into a spread sheet. I’ve also taken the liberty of manually typing the numbers into a spreadsheet from the voluminous data that Jean Dreze and Amartya Sen collated for their book. Given most of that data is from publicly available sources, one assumes that’s alright.

Delimitation & Democracy

Extending the previous post’s rationale, Avataram performed a basic population to seat bench-marking based on 1971, 2001 and 2011 Census data. You can see the results here: Delimitation 2011[1]. It’s quite disastrous if you are a citizen of Central/North/Western India.

Basically, the states that are losing representation in relative terms happen to be: Uttar Pradesh, Bihar, Rajasthan, Madhya Pradesh, Maharasthra, Haryana, Gujarat and Jharkhand. In that order. These are states that the BJP is either doing well in or has a reasonable chance of doing well or will absolutely have to do well in order to get anywhere close to majority in Lok Sabha. While those that are gaining the most in relative terms happen to be: Tamil Nadu, Kerala and Andhra Pradesh. In that order. Where political presence is restricted to the INC and regional parties (Left included). The gain of 7 seats for Tamil Nadu is staggering and add to this the state’s past 20 year history of delivering decisive mandates with overwhelming majority; the skew is almost unbelievable.

A study on all discretionary projects/fund allocations and general pork of the past 20 years and Tamil Nadu’s relative share of it will be interesting.

[1] – Here is a reworked sheet with 542 constituencies as benchmark: Delimitation-2011

Pollachi vs Mandsaur: What’s a vote worth?

Consider two states in India that usually have a bi-polar contest: Madhya Pradesh and Tamil Nadu. 

According to Census 2011, Madhya Pradesh has a population of 7.27 Crores. The Lok Sabha has 29 MPs representing these 7.27 crore people from Madhya Pradesh. In Tamil Nadu’s case, 7.21 crore people are represented by 39 MPs. That is, with the roughest type of calculation without considerations of voting ability, one can estimate that a citizen in Tamil Nadu has about 34% more representation in the Lok Sabha compared to the average citizen of Madhya Pradesh. If the principle of universal adult franchise is that everyone has one vote and hence an equal voice, that makes India an inherently undemocratic country.

There are after all genuine reasons for such a disparity in the citizen’s voice being heard in Delhi. Under Article 82 of the Indian Constitution, the electoral map of India is/was to be redrawn after every Census. In 1976 it was amended to freeze constituencies — so that states that implement Family Planning policies well weren’t punished for their success. Which one’d think is probably not the worst idea to have come out of Delhi. After all, the average Tamil woman now bears 1.7 children while her counterpart in Madhya Pradesh bears 3.2. Be that as it may, this post is not about the injustice of planned democracy. It’s about the electoral calculations political parties indulge in and metrics that may be a guide in that process.

To begin, let’s pick two representative constituencies in terms of this population and delimitation considerations. Pollachi, with about 10 lakh voters and Mandsaur with 13.7 lakh voters seem close to the mean and median values in terms of voters per constituency in their respective states. Let’s do a quick back of the envelope calculation in terms of the probability of each voter impacting the result of the election. Firstly, in Pollachi, let’s assume there is a direct contest between AIADMK and DMK. In this case, the influence of a Pollachi voter on his MP’s election turns out to be e^(-353.27). To see how I got that value, please refer to this quick calculation I made on paper and scanned it into PDF: Pollachi

Extending the same method, if we calculated the influence of each citizen in Mandsaur, he/she will have a probability of impact on the outcome as e^(-413.49). Again a simple calculation of these numbers mean that the average citizen in Pollachi is 14%[1] more likely to influence the election of that constituency’s MP than the citizen of Mandsaur. Remember, this additional power is over and above the 34% that this citizen of Pollachi anyway enjoys over the citizen of Mandsaur in representation on account of being from Tamil Nadu and not Madhya Pradesh.

That if one is a citizen of Mandsaur, one is dramatically disenfranchised compared to a Pollachi resident is now obvious. But how will a political party read this? Expending resources in Mandsaur and places like it, if one were an analyst for a political party, appears to be a poor return on investment. But the problem the BJP has is, its concentration in the states of Madhya Pradesh, Chhattisgarh, Rajasthan, Jharkhand and to a lesser extent Gujarat are all bound to provide far worse returns on the investment of political capital — be it money, ideology or muscle. The only island in this regard amongst poor states of India seems to be Uttar Pradesh. Not because it has managed to bring fertility rates down but because it offers a genuine three or four cornered contest that reduces our N if we used a quick and dirty method. So naturally, the DMK, ADMK, TDP, INC, TMC, SAD and the Left are all “richer” parties with better ‘bang for their buck’ compared to the BJP. Maharashtra and to a lesser extent Karnataka and Gujarat offer the BJP’s island of greater political effectiveness. They still pale in comparison to Kerala, Tamil Nadu, Punjab, Andhra Pradesh and West Bengal.

On the larger issue itself, if this trend of Kerala and Tamil Nadu having fertility rates less than replacement rate and high political representation with high HDI continues over the next decade, as it is bound to, and the rest of the large states continue to grow above replacement, as they are likely to, Indian polity is set for an interesting chauvinistic confrontation. Demographic dividend is a strange phenomenon, electorally. Especially when high social indicators are aided by a frozen Delimitation Committee.

[1] — that was more dirty than quick. But when the numbers become so low, as e^(-413) it hardly matters.

Smaller State = Better State?

The division of erstwhile Punjab region in 1966 into Punjab, Haryana and Himachal Pradesh worked out quite well for all three states. Punjab and Haryana are rich; Himachal Pradesh is truly a well governed state that stands up with two of India’s best performing states, Kerala and Tamil Nadu, on almost all indicators. This experience possibly drives much of the economic and governance justification that many commentators have made in the recent past for smaller states being the way forward for a better governed India. But Haryana and Punjab are rich and fertile regions that enormously benefited from green revolution. Does the ‘small is better’ argument hold up for the three new states — Chhattisgarh, Jharkhand and Uttarakhand — that were created at the turn of this century?

Let’s look at some basic information of these three States and how they compare to their original undivided State and with respect to India as a whole[1].

State Population (Million) Average Rural Household Expenditure/ Month (Rs) Average Urban Household Expenditure/ Month (Rs) Growth Rate per capita (%,2001-2011) Poverty Estimate as % of Population (2010) Population % in India’s lowest wealth quintile
Madhya Pradesh 72.6 903 1666 4.5 36.7 36.9
Chhattisgarh 25.5 784 1647 6.3 48.7 39.6
Bihar 103.8 780 1238 5 53.5 28.2
Jharkhand 33 825 1584 4.6 39.1 49.6
Uttar Pradesh 199.6 899 1574 3.9 37.7 25.3
Uttarakand 10.1 1747 1745 10 18 6
India 1210.2 1054 1984 6 29.8 20

It’s obvious that all three original states of Madhya Pradesh, Uttar Pradesh and Bihar are worse than India as a whole in every single metric. While Uttarakhand appears to have been a slightly wealthy region of UP, both Chhattisgarh and Jharkhand seem quite similar to the larger state they were a part of.

The number that jumps at us from the table is that despite recent events, Uttarakhand seems to be doing quite well from a growth perspective — growing at 10% annually for 10 years — compared to UP that grew at 3.9%. However, it must be noted that UP has 200 Million people while Uttarakhand has only 10 Million. Still, 10 years is a long enough time period to not ignore such a wide divergence. Jharkhand meanwhile was surprisingly worse off than Bihar in terms of growth in the same 10 year period while Chhattisgarh did much better than Madhya Pradesh and was closer to the national average. Two out of three small states that were carved out of larger states seem to be growing better than their parent sates. Does that mean anything?

Uttarakhand is clearly not as poor as UP and that may partly explain its growth divergence. But what’s interesting is: Chattisgargh has a poorer population compared to MP but seems to be doing better on per capita growth. Jharkhand’s population is less wretched from a poverty standpoint compared to Bihar but it does worse than Bihar. That either means something is horribly wrong with Jharkhand or possibly, even more worryingly, the large state of Madhya Pradesh is a true failure. Before we judge Jharkhand too harshly, it is worth noting that half of its population lives in the country’s lowest wealth quintile. Maybe this simply points to a large tribal population.

To understand how the society functions, a better method maybe to imagine someone being born there now and happens to be the mean, median and mode of that society. What are the chances of such a new born person? To pursue that thought experiment further, consider the following data,

State IMR (2011) MMR (2011) Fertility Rate Percentage of Births with Skilled Personnel Assisst Ante-natal Care of any kind (%)
Madhya Pradesh 59 269 3.1 32.7 79.5
Chhattisgarh 48 NA 2.7 41.6 88.5
Bihar 44 261 3.6 29.3 34.1
Jharkhand 39 NA 2.9 27.8 58.9
Uttar Pradesh 57 359 3.4 27.2 66
Uttarakand 36 NA 2.6 38.5 69.4
India 44 212 2.4 46.6 76.4

First data point that’s obvious is: in all the three newly formed states, the fertility rate is lower than their respective original states. Even Jharkand which does poorly when on other metrics seems to be doing better than Bihar in terms of reducing the number of children born to each woman. So if you were going to be born in these three places, your likelihood of being born at all is probably lesser in the newer states. More importantly, in each case, it appears your likelihood of survival after being born is better in the newer states. Even more significantly, in Chhattisgarh and Uttarakhand, your birth is more likely to have been overseen by someone skilled than in MP and UP respectively. Though in all these cases, that expectation is worse than India as a whole. Further, in every single instance of a new state, your mother is more likely to have received some ante natal care than in the respective old state.

The simple acts of delivering governance — such as providing ante natal care and assisting births with skilled personnel — seem to have an immediate effect on IMR. It’s heart warming that these new states are at least doing the easy things differently from their parent states and exasperating that Madhya Pradesh and Uttar Pradesh are doing so atrociously poorly in metrics that measure basic human decency. Their IMR and levels of skilled assists in births will shock the conscience of Sub Saharan Africa. So to the thought experiment of being born in a badly run large state versus a newly carved out smaller one, being born in the latter seems a safer idea.

Let’s extend the thought experiment and ask what if one were to grow up a bit and seek an education. Consider this,

State Female Literacy Rate Male Literacy Rate Female Illiteracy in age group 15-19 Male Illiteracy in age group 15-19
Madhya Pradesh 60 80.5 22.9 4.7
Chhattisgarh 60.6 81.5 16.7 6.7
Bihar 53.3 73.4 37.3 15
Jharkhand 56.2 78.5 29.6 12.7
Uttar Pradesh 59.3 79.2 25.1 10.5
Uttarakand 70.7 83.3 4.9 2.3
India 65.5 82.1 15.8 7.4

One can use the actual literacy rate as something of a historical legacy and look at the illiteracy of children/teenagers between 15-19 years of age as what the recent governments have actually done. Consistently, the smaller states have done more to educate their young despite having similar starting points in the case of Chhattisgarh and Jharkhand. Or conversely, just what are the governments in these large states of MP, UP and Bihar doing? That close of 40% of teen-aged girls in Bihar are illiterate can only drive one to despair. To understand this better, let’s consider what the government has actually provided,

State Percentage of 15-19 Year Old Completed 5 years of School (Female) Percentage of 15-19 Year Old Completed 5 years of School (Male) Percentage of 6-14 YO never enrolled in school (Female) Percentage of 6-14 YO never enrolled in school (Male)
Madhya Pradesh 78.9 83.9 15 11
Chhattisgarh 72.3 80.3 10 8
Bihar 69.7 75.5 31 19
Jharkhand 76 78.7 22 19
Uttar Pradesh 77.7 82.9 13 9
Uttarakand 90.6 93.9 6 7
India 83.7 86.2 12 8

Again, the older and larger states have not just failed older teens but they continue to fail younger children with the same level of moral bankruptcy, The newly formed states are at least failing their children a little less. Jharkhand particularly seems to be making an effort from a really poor start.

Creating the new state of Telangana from a reasonably governed state like Andhra Pradesh may not show such divergence. But from what the data suggests, Ms Mayawati may be correct in demanding four states out of what is now UP. After all it can’t do much worse any way. Might as well try that given the absolute lack of downside.

[1] — All data was typed into an excel sheet from publicly available information. If I have made an error in typing, please let me know.