Futility of 'high-frequency' jobs data

Of late, much has been said about the alleged paucity of statistical information on employment in India. Many commentators have lamented the non-availability of “high-frequency” data and have gone on to argue that availability of such data is very important for policy-making. Much has been left unsaid, however. Few have elaborated on the kind of “high-frequency” data that are required. And no one has said how their availability will help policy-making. The commentators seem to believe that answers to these questions are pretty much self-evident. But are they?

Suppose we start producing quarterly estimates of the unemployment rate. What would this tell us about employment conditions in a country where most people must engage in some kind of work to survive? But suppose we stop worrying about this question and believe that an observed rise (fall) in the rate of unemployment signals deterioration (improvement) in employment conditions or a slackening (tightening) of the labour market. How precisely should policy-makers respond? And what if a rise in the first quarter is followed by a fall in the second quarter?

It needs to be said, loud and clear, that India’s National Sample Survey of Employment and Unemployment (NSSEU) has been a far richer source of employment statistics than the Labour Force Surveys carried out in many countries of the world. In particular, the NSSEU has produced statistics that are appropriate for understanding the nature and evolution of employment conditions in a developing economy such as India’s. This is amply demonstrated in a recent document* that makes intensive use of the NSSEU data to develop perspectives on the evolving employment conditions, the challenge that confronts India’s policy-makers and the policies that need to be adopted to meet the challenge.

But the NSSEU has thus far been conducted at five-year intervals and this certainly has been a problem. Starting from this year, however, it will be conducted annually in rural India and quarterly in urban India. This gain in frequency, though, will come at a cost in terms of “depth and breadth” of information; for the questionnaire to be used in “high-frequency” surveys must necessarily be much shorter than that used in quinquennial surveys.

To identify the kind of “high-frequency” data that would be useful, we need to have a proper understanding of the nature of employment, unemployment and wages. India’s is a dual economy with surplus labour. The organised sector — composed of the government, public-sector establishments and large private-sector establishments — offers formal employment (i.e., in regular wage-paid jobs endowed with employment security and provisions for social protection and non-wage benefits) to a tiny proportion — around nine per cent — of the workforce. Another tiny proportion — around 13 per cent — is in regular-informal employment (i.e., in regular wage-paid jobs with no provisions for employment security, non-wage benefits or social protection), about half of them in the organised sector and the rest in the unorganised sector.

The vast majority of India’s workers are either in self-employment or in casual wage employment in the unorganised sector. In these types of employment, there is ample scope for work-sharing; the same amount of work can be performed by a varying number of workers. Such work-sharing means underemployment of the employed, which is significant among self-employed and casual wage employees. Unemployment is low because few can survive without engaging in some kind of work. 

In this setting, employment growth basically reflects labour force growth and is often accompanied by increased underemployment. Unemployment reflects queuing for formal jobs and is basically confined to educated youth from better-off households. Wages are not determined by demand-supply equilibrium — there is no such equilibrium, as supply perennially exceeds demand — and hence cannot signal slackening or tightening of labour markets.

What types of “high-frequency” data might our surveys usefully generate? My choice would be the following: Employment in the organised sector by type; the extent of underemployment among the self-employed and casual employees; and wages of casual labour in different production sectors. Enterprise surveys are required to generate data on employment in the organised sector. We already have the Annual Survey of Industries but need an Annual Survey of Services.

NSSEU, being a household survey, can generate the relevant data on self-employment, casual wage employment and casual wages. NSSEU, naturally, would also generate data on labour force, employment, types and sectors of employment in the economy, and unemployment rate. But these are slow-moving variables; they may show short-run fluctuations, which are not of much significance.

The difference between the growth of organised sector employment and labour force growth (which is the same as employment growth) in the economy tells us about movement of workers from lower-productivity to higher-productivity jobs; such movement is substantial when the former is significantly higher than the latter. Growth of underemployment indicates deterioration of employment conditions in the unorganised sector. Growth of casual wage rate indicates productivity growth in self-employment and hence improvement in employment conditions in the unorganised sector.

Availability of such data will surely help in effective monitoring of ongoing changes in employment conditions in the country. But how will they help policy-making? Here, it must be recognised that real improvement in employment conditions in India can only come from economic growth and structural change, and these are matters of long-run strategies. The only short-run policy that the “high-frequency” data can influence is one relating to safety-net programmes delivered through special employment schemes.

*Ajit K Ghose, India Employment Report 2016, New Delhi (Oxford University Press). 

The writer is Honorary Professor at the Institute for Human Development, New Delhi