How reliable are Unemployment Statistics?

According to the latest annual report released by Ministry of Labour and Employment, our current unemployment rate is 9.4 percent. Unemployment is higher in rural areas than in urban areas – 10.1 percent versus 7.3 percent. These numbers have increased from the previous year and perhaps supply evidence that our economy, buffeted by the twin forces of global weakness in demand and government paralysis, is slowing down. But what exactly do these numbers mean? And how reliable are they?

The meaning depends on how unemployment and related terms are defined. As these definitions change, or are interpreted differently, population and unemployment statistics also change. Unemployment rates may rise or fall merely because of a change in definition, rather than because of actual changes in the labour market.

Reliability is affected by the coverage of the survey, sampling techniques, the training and experience of the surveyors and the categorization of surveyed individuals by employment status. Going through the definition of unemployment as listed in the ministry report, it struck me that the definition was rather non-specific: the unemployment rate is the ratio between those out of work but looking for a job and those in the labour force, which is the sum of the employed and the unemployed-seeking-work (italics mine).

                The labour force, by definition, is always less than the total population because it excludes those without work and not seeking employment: small children, retirees and stay-at-home housewives. The non-specificity is apparent if we ask, “How do we identify those looking for a job?”  The ministry defines job seekers as those “looking / available for work for a relatively longer part of the reference period” (which was the 2009-10 financial year for the report).

                This definition raises two sets of questions. First, what must a person do to be looking / available for work? Must he periodically importune employers in his area for jobs? Actively network and attend job fairs? Register with the government-run unemployment exchanges? Second, what is the “relatively longer part” of the reference period? If, for example, it is greater than half the period, why not be more specific and just say so?If one is “looking for work” only if s/he registers with the government-run exchanges – and this would be the simplest definition from the data collection point of view – then unemployment would be underreported because the statistics will not capture job seekers who did not register with the exchanges. But including unregistered job seekers in the definition brings along a different set of problems.

                The surveying agency must then devise sampling methods that ensure representative coverage of the population. The survey instrument must also be well-designed such that it elicits accurate and honest answers from respondents. In practice, self-reporting by individuals about their employment status would introduce bias and error into any estimates. Respondents might choose to misrepresent their job status. They might be unaware of their status. They might be unaware of how long they have been searching. These factors would affect their placement into the appropriate (un)employment categories.To complicate matters further, even the difference between employed and unemployed is not always clear.

                Consider part-time employment. An individual might do odd jobs a few hours every week while seeking a full time job. Should s/he be classified as employed or unemployed? To solve this problem, one might define employment according to the number of hours worked per week. But then, what should this number be? If the number is too high, then some legitimately employed people will be mistakenly classified as unemployed. If too low, then the opposite will happen. The latter will be preferred by a government, especially around election time.

                Given this preference, there will be an institutional bias towards using relaxed definitions of employment and thereby painting an overly optimistic picture of the labour market.This bias, however, will be counterbalanced by other biases. We have a significant underground economy and its members are less likely to report employment to government authorities than those with legitimate employment.

                This under-reporting will make the unemployment rate seem higher than it actually is. (Under-reporting is also often offered as an explanation for the high youth unemployment rates in southern Europe.)

                A final comment on the Indian labour force: Our labour force participation rate, i.e. the proportion of the total population who comprise the labour force, is much lower than in OECD countries. Currently, our participation rate is about 36 percent, while it is 64 percent in the US and 58 percent in Germany.

                Only 425 million Indians are either employed or seeking employment. The remaining 750 million may be thought of as dependents. The low participation rate can be ascribed to the low proportion of women actively seeking employment. Whereas almost 54 percent of all males are included in the labour force, the participation rate for females is only about 16 percent. This in turn may be ascribed to the conservative, traditional structure of family relationships in India.

                However, surprisingly, participation rates for women are lower in the cities than in rural areas even though rural areas tend to be more conservative. This contradiction may be explained by higher economic necessity among rural households and different data collection standards in rural and urban areas, among other factors. 

Source: The Economic Times, January 16, 2012