Sunday, December 9, 2007
An article in the New York Times on why eating local or organic food may not necessarily be green or carbon-friendly, an echo of what Tim Harford wrote a while ago. Also how difficult it is for consumers to make informed decisions. Helpfully, Tim Harford, in another article points out the way forward.
Wednesday, December 5, 2007
A superior way is to calculate the Farmers’ Suicide Rate (FSR), which is the number of suicides per 100,000 farmers. Dr Nagaraj has done that. P Sainath writes that the FSR of about 12.9 is higher than the General Suicide Rate (GSR) of above 10. The latter is the number of suicides per 100,000 population.
However, it is methodologically incorrect to compare the GSR with the FSR. This is so because the size of the work (labour) force is much smaller than the total population. While calculating GSR one takes the total number of suicides and divides it by the total population of the country. The total population includes, for example, all sub-adults including infants and children who are not part of the work force. The percentage of population below the age of 14 is about 33% (2001 census) and if one were to exclude this the GSR would jump to 16.
But while calculating FSR one takes only those (adult farmers) numbers who are part of the labour force. This is a much smaller number. Census of India 2001 classifies 400 million or approximately 40% of the population as workers. Hence FSR ought to be compared with the suicide rate for other professions and not the general population. Or FSR should be compared over time.
Also note that the Census of India definition of workers (see here and here) is different from the NCRB category of professional status of suicide victims, the latter includes the unemployed and housewives.
Dr Nagaraj also argues for making adjustments for one, the exclusion, by the NCRB, of non-title holders who are actually cultivators from the category of farmers and two, for the distinction between marginal and main workers in the Census categories. Both these matters are debatable and it is a moot point if such adjustments ought to be made. Even if done these adjustments are likely to change the suicide rate but not the variation in them over time. And even after making such an adjustment the FSR increases to 15.8 which is lower than the 16 (GSR) that I calculate excluding children from the count.
Monday, December 3, 2007
P Sainath has written extensively on the subject of suicides by farmers and thinks it is an indicator of economic distress in rural areas. Most of his stories are reports from the ground (See here and here). Of late, he has written some pieces (see 1, 2 and 3) based on analysis done by Dr. K. Nagaraj of MIDS, who in turn has used data collected by the National Crime Records Bureau (NCRB). See here for an interview with Dr Nagaraj.
Sainath's dramatic headlines - 1.5 lakh farmer suicide during 1997-2005 and a farmer's suicide every 30 minutes seem to suggest very high incidence.
Let us look at the evidence. Here is a table from Sainath's own piece:
During 1997-2005 there were 1.5 lakh suicides by farmers, as he says. But during the same period there were a total of 1 million suicides in the country. If there was a suicide every 30 minutes by a farmer there was one every 4-5 minutes by the general population.
Obviously these absolute figures have to be put in context. As the table suggests:
1) The overall incidence of suicides (suicide rate) for the country is stable at just over 10 per 100,000 population (i.e. ten suicides per lakh of population per year).
2) Suicides among farmers are also more or less constant at about 15% of the total suicides in the country.
Thus, there is hardly any increase in the proportion of suicides committed by farmers in the nine-year period.
(There is an inexplicable jump in the number of farmer suicides from 1997 to 1998, which may be a data issue as data on farmers as a category started being collected from 1995 and several states reported late. Also strictly speaking we should account for the likely declining proportion of farmers to total population but nine years is not that long a period, nevertheless the stated numbers would understate the farmers' suicide rate).
Curiously, Dr Nagaraj and P Sainath make no mention of the causes of the suicides that are part of the NCRB statistics. The NCRB data suggests that a mere 6% or so suicides are due to poverty and bankruptcy, the majority being due to illness (22%), family problems (22%), other causes (28%) and unknown (15%).
I don't have the breakdown by reasons for farmers but it can hardly be that these other reasons are absent as a cause for suicides by farmers.
If true, that just like for the general population the reasons for committing suicides by farmers are predominantly non-economic, then the whole hypothesis of Sainath and Nagaraj would take a severe knock.
Four states – Maharashtra, Karnataka, Andhra Pradesh, and Madhya Pradesh (including Chattisgarh) – have accounted for more than 60% of farmer suicides during 1997-2005. Dr. Nagaraj relates this to several factors such as the lack of support from government, aridness of the regions and relying on cash crops and bought-out inputs. He also groups the various states in several categories. See the articles above for details.
However, the patterns don't quite hold. For example, if the affected states are arid so are Gujarat and Rajasthan which have low number and incidence of suicides. Kerala has one of the highest suicide rates, both overall and among farmers but it is not arid, but quite wet. Chattisgarh and Madhya Pradesh are not arid either.
If cash crops are a factor, then one would expect Gujarat to be on the high suicides by farmers list, but it is not. And Chattisgarh hardly grows any “cash crops” and is a high suicide state. And certainly rice and wheat are cash crops for Punjab, Haryana and Uttar Pradesh but the incidence of suicides is low in these states. These states are also high-input using ones.
Tamil Nadu too is arid in many parts and the agriculture is quite “commercial” but the incidence of farmers' suicides is low though overall suicide incidence for the population is much higher than the national average. Communist ruled West Bengal has a high rate of overall suicide rate but not so for farmers.
One hardly sees a pattern, certainly none that fits the neat hypothesis of Sainath and Nagaraj.
A perusal of Sainath's field reports and other media reports suggest that farmers who took their lives are land owners (not landless), had frequently leased-in additional land, invested in irrigation and had borrowed sums which by Indian standards were not inconsiderable – several lakhs of rupees in many (most?) cases. Note, that the suicide-prone are not marginal farmers and landless labourers who one would expect to bear the brunt of agricultural distress. Also note that relatively poorer states like Bihar and Orissa have much lower rates of suicides than the four most seriously affected.
Finally, distress doesn't translate automatically and immediately to suicides. It takes a lot more than distress for a person to commit suicide. Why it happens at all and why so in select parts of India is beyond the scope of this blog and my competence. Some clues can be found in the book - The Tipping Point by Malcom Gladwell – who points out the “locally contagious epidemic” nature of suicides once a few have occurred.
P Sainath and Dr Nagaraj also point out that there are limitations to the data. I agree that there could be severe deficiencies but it is not at all certain, a priori, as to what direction revised statistics would take.
The stories, data and analysis presented of suicides by farmers are at present far from convincing.