You have been sharply critical of census methods in different parts of the world, in particular India. Why are you unhappy with the Indian census?
Starting from 1881, we have had a census every 10 years, except during the war in 1941, when we had a restricted census. Frankly, I can't think of any other developing country, apart from Sri Lanka, which has had such regular censuses. Even China had a break. That's the good part of the Indian census.

But even developed countries are facing tremendous problems with their censuses. In Germany and USA, people have stopped cooperating with the census; the non-response rate in USA is as high as 30 per cent. The Americans have filed litigations, contending census questions encroach upon their privacy.

My misgivings about the census is the extent of undernumeration. How many people have we missed out? I would say many. The undernumeration is ridiculously low in India. Statisticians all over the world have questioned how, in a vast country like India, which has a high degree of illiteracy, you can possibly have a high degree of enumeration. There is obviously something wrong with the census data.

And the inadequacies of census data are more acute in urban areas than rural. Overall, it is possible that the degree of undernumeration is low because mobility is low. Unlike in the West, people are not frequently changing jobs. In the villages, most people know each other. So it isn't surprising our census figures for rural areas are more dependable than those in urban areas.

But the extent of undernumeration is disturbingly high in urban areas. For example, you have situations where both husband and wife work and houses remain locked all day, because of which enumerators in mega-cities miss out people.

In the 1991 census, I feel the extent of undernumeration in the big cities was about 10 per cent. You may ask, did I conduct a sample survey to arrive at this figure? I didn't, and logistically, I can't. Soon after the figures were published, when I was making a film on the 1991 census, I spent whole nights trying to figure out how the census had accounted for pavement dwellers, who constitute a very large section of the population. I was more than convinced that my 10 per cent estimate was on the low side. Another fact it brought home to me was that there had been no slowing down of the urbanisation process.

Are you suggesting urbanisation has continued unabated? In this connection, how would you rate the 1991 census?
Statistically, one can say migration cannot go on forever, given the virtual collapse of urban infrastructures. But I cannot accept that there has been a considerable slowing down of urban growth rates.

Take Patna, for example. The census figures show a sharp fall in the growth rate, but I don't accept this. In Bihar, the census figures show a sharp fall in the sex ratio. You may try to apply all kinds of regression analyses, but I would say it is just a bad census. In the post-enumeration check-out, the census authorities do not give individual state figures, though they have these figures. The government does want to reveal the individual competence of the state governments, so the figures are given regionwise.

I would also challenge the way the census was conducted. What was the extent of undernumeration? The long and short of it is that our census figures are unreliable. According to my estimates, the extent of undernumeration is around 10 per cent in mega-cities, 5 per cent in towns and 3 per cent in rural areas. Using these estimates, I don't think there has been a slowing down of the urbanisation process.

The economic evidence, on the other hand, shows there has been a recession in the last decade, slowing down growth, because of which not many industries have come up. It is possible migration rates have slowed down. Detailed data on migration, which are not available, need to be seen before we take a final view, but questions remain. Even if the workforce has declined in the organised sector, what about the unorganised sector? I don't see any diminution in the number of rickshawalas, domestic servants or shoeshine boys throughout the country.

What have been the key factors in urban growth in India? How would you explain the paradox that urbanisation and modernisation have not brought down fertility rates?
While I maintain urbanisation has continued unabated, I must add that we should be careful of a component analysis of urban growth -- the role of natural increase in population (births minus deaths), net migration (in-migration minus out-migration) and reclassification balance (villages becoming towns and towns becoming cities). Though 1991 data is not available, but my rough estimate is that between 50-60 per cent of urban growth is due to natural increase. Migration accounts for 20-30 per cent. The rest is reclassification.

This means the urban birth rate continues to be significantly high in India, contrary to what many demographers would have us believe. The supporting evidence you can find in the Sample Registration System (SRS) Data, which are generated by the office of the registrar general. SRS data show urban birth rates are very high in Bihar, Uttar Pradesh, Madhya Pradesh and Rajasthan.

One of the elements of the demographic transition theory is that urbanisation and modernisation bring down fertility rates. But take an industrialised state like Haryana, where the urban birth rate is as high as 27.2 per 1,000. India needs a birth rate of around 21 per 1,000 by the turn of the century. In 1991, it was 24.3 -- 25 in urban areas and 30.9 in rural areas. Therefore, our expectation of a birth rate of 21 in 10 years is impractical.

Do you think urbanisation has affected the fertility rate anywhere in India?
Only in West Bengal, where the urban birth rate is the lowest. This is a point that has been completely overlooked by the media. The primacy of Calcutta and the bhadralok ethos has pushed up the age of marriage in West Bengal. But when you come to rural West Bengal, the birth rate is very high. In fact, the gap between rural and urban birth rates is the highest in West Bengal. In Tamil Nadu, demographic transition is still dependent on caste and poverty and unemployment -- the Brahmins are being driven out. But Tamil Nadu does not have the urban ethos West Bengal has. Madras is not a primate city in the regional sense that Calcutta is.

Isn't it ironical that in the decade of the girl child, India shows a declining sex ratio, with fewer females being born? How do you explain this phenomenon?
According to the 1991 census, the sex ratio in India is 927 females for every 1,000 males. In the decade of the girl child, this is shocking. No quick judgement, however, can be made because we still don't have enough data on the subject.

There are at least four factors that affect sex ratio at birth and all of them have to be considered to determine the fall in sex ratio. But we go on making simplistic assumptions concerning amniocentesis, though it is a problem in Haryana and Maharashtra.

Internationally, the average is 107 boys for every 100 girls. There is a theory that says that over the years, more and more boys are being born. We are only speculating, but there could be definite genetic reasons.

In India, the mortality schedule is such that more girls die than boys. In the West, more boys die.

I feel girls are not always enumerated in India. But this is not a new phenomenon. So, if the sex ratio is still declining, it must be because more and more girls are being left out, as happens in rural Bihar. In Orissa, too, male members are enumerated first. The census people finally grow tired and leave out the females.

How would you assess the impact of illegal migration from Bangladesh? Do you think it is being unnecessarily turned into a political issue?
This is an extremely disturbing phenomenon. The 1991 census figures for migration are not available, but even were they available, there is no way the census can indicate illegal migration. No illegal migrant will declare himself so.

According to some people, migration from Bangladesh is insignificant, but I feel it's much more than a political campaign. Bangladesh is a huge human reservoir and the potential for migration is great. I am not suggesting India is a land of milk and honey, but things are better here. And the migrants are now spreading as far as Punjab and Bombay.

We can make estimates, but it would be a misuse of statistics if we divide the migration figure by 900 million and conclude the proportion is small. At the district level, even a small addition can lead to economic problems and intense animosity towards the migrants. I call these migrants the international shudras (untouchables).

How would you assess the success of our family planning programmes?
I think our family planning data is bogus and I challenge the methodology used. All the couple protection rate (CPR) figures are fudged. And what I call community participation rate, which is my interpretation of CPR, is zero.

The provider approach of the government is meaningless. I would suggest expenditure in the area of education rather than family planning. I wonder how sex, a private sector activity, can be determined by public sector initiatives. The family planning department, to my mind, should be disbanded or merged with health.

Of course, the government will not agree, but its family programme has failed. We have done fairly well in our control of death rates, but then, death rates have come down in all Asian countries. Our infant mortality rate continues to be so high! Only Kerala has an infant mortality rate lesser than China -- 16 in the urban areas and 17 in the rural areas.

Our family planning programmes are donor-driven. We have relied on changing contraceptive technology, but devoid of a social context, technology has no meaning. In a poor country, monetary incentives can play havoc. Would it not be better to divert the Rs 6,500 crore -- the budget for family planning in the Eighth Plan -- towards primary education?

The assumption is put more money in the system and the birth rate will come down. This is sheer nonsense. The panchayats should be given power and money and they shouldn't be allowed to become sub-offices of the block development office.

What would you say about astrophysicist Vasant Gowarikar's optimism that India's population will stabilise by the turn of the century?
I think Gowarikar (Soumya, please check spelling) is indulging in too much fine-tuning. His assumption that India's population will stabilise is dependent on certain conditions that to my mind can never be fulfilled -- conditions like full literacy, power and irrigation. Gowarikar's statistical parameters are too simplistic, and the socioeconomic factors he considers necessary, and which he assumes will be there, are utopian. A clear case of an astrophysicist indulging in a bit of kite-flying.