Narratives of India’s labour market vary but the reality doesn’t
- PLFS and CMIE numbers diverge on women-at-work and employment but the latter’s gloomy picture is less than convincing. A closer look at the data may explain some of what’s going on in the country.
The start of a new column is like the birth of a child. What you are going to name it, and expectations of how it will all turn out. This new column series (following in the tradition of Looking for Logic, Beyond Logic and No Proof Required) is named Rashomon Diaries. Films and cricket were my earliest romances, and they have tested time. Kurosawa’s Rashomon was released in 1950, a year with considerably fewer computers and manifestly less fake commentary. Rashomon is a film more appropriate for the conflicted and polarized times of today. Never have truths been more contested than now. Rashomon Diaries will attempt to shed light on, decipher and seek truth in this data-rich AI deep-fake world.
In this article, we will look at politically sensitive data on the labour market, i.e. unemployment and labour force participation. Candidate Bill Clinton aptly described his 1992 US presidential campaign as one revolving around the economy; “It’s the economy, stupid" is now a household phrase the world over. India is four times more populous and N times more varied than the US. Not unexpectedly, there is so much variation in our economic data that almost anyone can come out with an observation that seems to ‘fit’ the narrative.
One such narrative is that the unemployment level in India is high, and increasing. Data provider Centre for Monitoring Indian Economy (CMIE) recently released unemployment rates according to its much-cited Consumer Pyramids Household Survey (CPHS) module. The CPHS states that the national unemployment rate increased by 3 percentage points (ppt) in just one month—from 7.1 % in September 2023 to 10.1 % in October 2023, coincidentally just before five states in India went to the polls in November. Can unemployment rise by such a large magnitude in India when the labour market is improving across the world? India hosted the G20 summit in September. Did its close lead to such a downturn in the economy? Or was it stubble burning in Punjab that led to a major shutdown in north India? The highest 4-month increase in unemployment ever recorded by CMIE-CPHS was the 2.9 ppt increase during the exceptional lockdown of January-April 2020.
If it is the economy that matters most (it does), and if the Indian economy is looking good (GDP data, the IMF and World Bank are all in agreement), then it is logical that the narrative of the opposition is going to be that the economy is not doing well, and its calling card is WYSINWYG: What you see is not what you get.
Official Indian data on unemployment goes by the Periodic Labour Force Survey (PLFS, current weekly status). It tells quite a story different from CMIE’s. Unemployment declined from 7.4% in 2017-18 (agricultural year July-June) to 4.2% in 2022-23. Coincidentally also a 3 ppt change, but this is over six years, not one month, and a decline, not an increase. The question analysts have to answer: Which version of reality approximates the truth best? To begin with, note that the PLFS-CWS unemployment rate of 4.2% for 2022-23 is the lowest since 2011-12, and not far from the 3.5-4.5% range that has prevailed since measurement began in 1983. Taking the more relevant usual status (unemployed for more than 30 days), the aggregate unemployment rate falls to less than 3%. CMIE-CPHS data (see chart) shows a steady increase from 4.7% in 2017-18 to 7.6% in 2022-23. The CMIE and PLFS show wildly divergent labour markets. What explains this divergence? Two important determinants of it are sampling design and definitions. Many have commented on the unrepresentative nature of the CMIE survey; here, let’s focus on differences in definition.
The PLFS definition of weekly status is whether you were employed for at least one hour in the previous week. This is a standard ILO definition. The CMIE definition is non-standard. It defines a person as employed if s/he “is engaged in any economic activity either on the day of the survey or the day preceding the survey, or is generally engaged in an economic activity." How do we compare unemployment rates from two different sources/definitions? By making estimates using PLFS data and CMIE’s definition. This eliminates sampling design differences, but tests for differences in definition.
Unemployment rates are higher for the CMIE definition, but only by 0.8 ppt. However, the first two days’ unemployment rate is the highest CMIE reports. If the unemployment rate is estimated as per CMIE for the third through the seventh day, then the PLFS and CMIE rates broadly match—in trend as well as levels! Conclusion: Definition differences play no role in explaining the differences between PLFS and CMIE data.
Another unrepresentative result of CMIE data is in terms of labour force participation rates (LFPR). These rates are related to labour inputs (and their changes to GDP growth). To paraphrase Jane Austen, it is now a truth universally acknowledged that Indian GDP growth is the fastest among G20 economies and expected to be so for this decade. However, according CMIE, less than 40% of the population is working in India, compared to about 44% in 2017-18. In contrast, PLFS data shows India’s LFPR to have increased from about 47% to 54%, and both the trend and magnitude are unaffected by adopting the CMIE definition of employment.
Also in the chart are unemployment and LFPR rates for women. Unemployment rates are marginally lower for women. CMIE female LFPR rates seem to be in a different universe, as is the trend. Female LFPR rates under the PLFS show a steady increase from 20.7% in 2017-18 to 31.3% in 2022-23. This is for the 15+ age group, as is all other data reported in the table. For the 25-64 working age-group, the female LFPR has increased from 26.4% in 2017-18 to 39.2% in 2022-23. In sharp contrast, CMIE-CPHS data not only shows the world’s lowest female LFPR data for India (8.9% in 2022-23), but also a steady decline from 11.6% in 2017-18. Other countries have seen sharp increases in their female LFPR.
CMIE-CPHS data presents challenges. Neither economic nor social explanations fit the CMIE narrative. We have presented data, perhaps too much of it. Data can be a useful tool for communication and interpretation. Contrary to Rashomon, we can arrive at some stable conclusions about reality.
These are the author’s personal views.
