Labour

The Makeover of India's Labour Bureau Requires More Than a New Logo

The process by which the LB collects and sources data is flawed, resulting in a poor storehouse of labour statistics.

On August 20, 2020, the Indian government unveiled an official logo for the Labour Bureau (LB), nearly eight decades after the organisation first came into existence.

Director General D.P.S. Negi explained helpfully that the logo represents “three goals that Labour Bureau aims to achieve in producing quality data – accuracy, validity and reliability”. In the picture of the logo, the blue cog wheel represents work, the graph represents statistics, the three colours signify the national flag and the ears of wheat signify the output of India’s farmers.

At the launch event, India’s labour minister emphasised the “importance” of databases in “policy making” especially for employment generation and underscored the role of artificial intelligence and digitisation in the collection and analysis of large data sets quickly. Labour secretary Heeralal Samariya also announced that the LB will be the single point of labour statistics and will be given statutory powers in the proposed labour codes.

Since October 1946, the LB has been “engaged in collection, compilation, analysis and dissemination of statistics on different facets of labour at All India level”. It comes under the supervision of the Ministry of Labour and Employment (MoLE).  The bureau’s vision is to make it “a premier agency in the field of Labour and Price Statistics” and its mission include compilation and timely release of price indices (which it has been doing well) and to “serve as the storehouse in the field of Labour Statistics by making the latest Labour Statistics available to stakeholders”.

The LB claims that its publications give “authoritative and up-to-date statistics on various facets of Labour and on current Labour scene in the country”.

While the bureau certainly deserves a ‘logo’ – even if it is several decades late – what it requires is a complete makeover as it has become contrary to its vision and mission. Indeed, one may describe it as a storehouse of incomplete, deficient, narrowly constructed and poorly conceived data that only revolves around a meagre segment of the so-called organised sector.

Labour statistics in India and role of the LB

India ratified the ILO Labour Statistics Convention, 1985 (No. 160) in 1992 and Labour Inspection Convention, 1948 (081) in 1949 and is bound to publish certain key labour statistics as provided under the agreements.

The LB’s statistics is mostly based on administrative sources and works as follows. State labour departments (SLDs) headed by the commissioner of labour or directorate of factories through their divisional/district offices for the state sphere and the central labour commissioners’ office (CLCs) for the central sphere (under the MoLE) collect data through statutory returns. They also collect data on a mostly voluntarily basis on a number of issues – for example, industrial disputes – from employers and trade unions under various labour laws.

Also read: The Coronavirus Lockdown Has Been a War on India’s Informal Labour

These offices, in turn, compile and collate this data and send it to the to LB, which publishes them suitably. On its own, the bureau also conducts ad hoc, special field studies and surveys (employment-unemployment (E-U) surveys or evaluation of impact of laws, occupational wage surveys etc.

First, the good

The bureau has a wide mandate and brings out multiple publications. It performs 18 regular functions, five new initiatives and classifies its activities under five heads – labour intelligence (price and wage indices), labour research, monitoring and evaluation, regular publications and training. It also publishes statistics on numerous variables relating to labour such as wages, earnings, productivity, absenteeism, labour turn-over and industrial relations, and publishes them in its comprehensive outlets (in both physical and electronic modes) of varying frequencies, monthly (Indian Labour Journal), annual [Indian Labour Year Book (ILYB), Indian Labour Statistics (ILS), Pocketbook of Labour Statistics (PBLS)], and special reports (annual reports on the working of major labour laws).

The organisation’s mandate has been huge and additional functions assigned to it only have proved burdensome. But, despite this, it has managed to produce a reasonably rich database for decades; notwithstanding the problems it has always faced – resource crunches, untrained and poorly paid field personnel, failures in data supply chain, etc.

It is also the competent authority under the Minimum Wages Act (MWA) to provide data on CPI to be used for determining special/cost of living allowance under it and till 2012 was the sole agency for generating Consumer Price Index (CPI) for industrial (CPI-IW), rural and agricultural labourers (CPI-AL). Its contribution to construction of a database on industrial relations for decades is praiseworthy. It processes data from other statistical agencies like the Central Statistical Organisation’s (CSO) Annual Survey of Industries (ASI) and publishes data related to Part-II of ASI.  Its ‘Occupational Wage Surveys’ (since 1958-59) provide data on wages by occupation (skills), specific industries and services which are not available from other sources.

The LB also filled in the gaps left by NSSO by providing annual data on E-U during 2010-13 and also conducted 24 Quick Quarterly Surveys during October 2008-December 2014 to assess the impact of economic slowdown on employment.

Bad to ugly

Decades ago, the First National Commission on Labour (NCL 1969) identified deficiencies in LB statistics including an absence of accuracy and reliability due to data collection dynamics and delayed publication of statistics among other issues and made elaborate recommendations on it. Later, several bodies were appointed by the government to review labour statistics which invariably dealt with LB. This includes the Working Group headed by T.S. Sankaran (1975), the K.C. Seal Committee (1981) and the Study Group chaired by Professor Lalit Deshpande (1999).

More recently, the ILO commissioned a study on labour statistics by Professor T.S. Paopla (2014). But SNCL observed that “the implementation of [these] recommendations have been partial, and many of them remain unimplemented.” Undeterred by these sombre observations SNCL on its part added to the pile of recommendations. But the LB went from bad to worse.

Un-dynamic website

When one searches for “Labour Bureau” on Google or through the MoLE, it takes us to its old website which provides data on CPI among a few others and then there is a direction to the new website, which is annoying. The revised website of LB leaves a lot to be desired. Some of the links do not lead to any material information and instead direct the user to the home page. Several of the sub-links provide rather brief and incomplete and unhelpful information with no further links. The website, for example, could provide valuable historical documents (say the Committees’ and Study Groups’ concerning labour statistics) and databases under, say, a proper ‘archives’ section which does exist but has a limited coverage.

Also read: What Job Losses in the Formal Sector Tell Us About the Lockdown’s Impact on Economy

Under the “Labour Statistics (PUB)” section of the Industrial Disputes link, I found half-baked data. Under “Whats New”, I found one item each for 2015 and 2016, three for 2017, seven for 2018, 10 for 2019 and 4 for 2020 (understandably lower due to COVID-19).

Lack of focus

Indeed, the bureau’s mandate needs to be trimmed and be given a focus primarily revolving around labour statistics only. For example, the listed items under “labour research” will not legitimately belong to it — this includes the compilation of labour part of ASI, surveys on E-U, occupational wage, etc. This entire function could be better left to its fellow-agency, the V.V. Giri National Labour Institute (VVGNLI). In fact the Second National Commission on Labour (SCNL, 2002)  wondered whether LB and VVGNLI could be merged given “duplication of functions” and mutual benefits arising out of synergies.

Though there need not be a merger, there is certainly more scope for productive interfaces between both arms of the MoLE.

A worker cuts iron rods outside a workshop at an iron and steel market in an industrial area in New Delhi, India, December 12, 2017. Photo: Reuters/Adnan Abidi/File Photo

CPI

During the post-Independent period, the LB adopted CPI (IW) series based on working class income and expenditure surveys with 1949, 1960, 1982, 2001 as base years. According to ILO Labour Statistics Recommendation, 1985 (No. 170) income and expenditure surveys should be conducted and the weights used for computing CPI once every ten years. The current series pertains to 2001 introduced in 2006 replacing 1982 series in 1988, a gap of 18 years. The efforts to replace 2001 index series is on and probably the new series with a base year of 2016 might be introduced soon.

The delays in updation of CPI affect labour welfare as outdated series based on old income-expenditure patterns will affect MW and DA for working class.

Poor conceptual base and static data coverage

The data generated by LB ultimately depends on the country’s labour laws, which as of now apply mostly to the organised sector. Since most laws have thresholds of applicability, the coverage of organised sector is further narrowed. Glaring omissions on data collection and publication include information on child labour, migrant workers, unorganised workers, building and other construction workers, service sector, occupational diseases, among others. Indeed, the COVID-19 pandemic recently exposed the government’s data limitations when it comes to vulnerable migrant, construction and unorganised workers.

Though macro data published under the Shops and Establishments Act (state laws) is available (which are of little use), on employment and wages we depend on the NSSO’s data or ad-hoc surveys. Like the ASI, why can the Labour Bureau not think to carry out Triennial Survey of Service Sector (TSSS)? In 2015, the sector had 8.05 million establishments as opposed to a meagre 0.24 million in the factory sector.

While the databases are narrow because they only cover some segments of the organised sector, the LB worsens the data coverage on its own. For example, it has been publishing statistics relating to industrial injuries in registered factories, mines, railways, and ports for a long time. But it has not expanded its sectoral coverage by including plantations, construction (a high injury prone sector), transport, hotel and restaurants. Countries like Sri Lanka provide this data for most sectors. Furthermore, LB should have expanded the definition of industrial injuries to include modern forms of occupational injuries in the service sector. The categories for which ILYB/ILS provide statistics have remained virtually the same for decades. It may not be much different even if one went back to the 1990s or earlier. There is no dynamism, nor innovation in conception and execution. LB excels in repeating like a parrot the same old confines of data.

The lack of ‘all-India’ data

It has become quite difficult to construct all-India database for  many of the variables concerning factories and employment therein.  This is because of the deficits in the data supply chain. trade unions and employers do not submit complete and usable returns in time. Even if they do, the division or district level offices of the SLDs do not transmit in time to its headquarters, i.e. the Labour Commissioner’s office (LCO).

Then, sometimes, the LCO may fail to send usable and complete returns to LB as it struggles to collect and compile data owing to aforementioned problems and due to its own failings (which may be due to person-power and infrastructural deficits, multi-tasking, etc.) Finally, the LB, due to its wide mandate and due to its own problems, merely publishes whatever data it has.

Two problems arise from this. Firstly, there is no all-India data for many categories and even state-wise time series data cannot be constructed as defaults by states are not consistent and even the submitted data do not represent the universe.

As an example of this, consider the graphs below, the data for which comes from the flagship publication of the Labour Bureau.

The ‘decline’ in key variables in the graphs are not due to genuine factors in the sectors concerned – or representative of any particular trend – but are statistical artefacts that can attributed to an inefficient functioning of the data collection system in India.

Figure 1. Workers’ Trade Unions in India, 2005-2015

Note: RTU – Number of Registered Trade Unions; TUSR – Number of RTU Submitting Returns; TUM – Member of TUSR

Trade unions sometimes fail to submit timely returns or submit unusable – incomplete or non-matching data – returns. When this happens, the labour administration at the state or district level  is slack in following-up and sending the data along the data supply chain. The trends in Figure 1 largely depend on the compliance-level of trade unions and states.  For example, even though the number of registered trade unions declined steeply from 40,175 in 2007 to 27,063 in 2008, since the proportion of registered trade unions submitting returns improved from 18.43 to 35.85, union membership increased from 7.87 million to 9.57 million.

On the other hand, even though the number of registered trade unions declined sharply between 2005-06 (-13.39%), due to a marginal rise in trade unions submitting returns (+2.61%), union membership rose marginally (+2.87%). Again, despite a decline in both registered and returns submitting trade unions between 2014-15, union membership rose marginally due to marginal rise in compliance by relatively larger trade unions. As a result of these complications, it is impossible to estimate “union density” or “union coverage” in India and researchers use the NSSO data which is based on information provided by members of the households and to that extent it is weaker.

Figure 2. Number of Working Factories (Submitting Returns) and Estimated Average Daily Employment (000s), 2005-14 

Indeed, the rise or decline in variables is due often due to the non-compliance of a number of states or UTs – the bigger the state, the bigger the data discrepancy. The dip seen in 2013 in ‘Figure 2’ above is primarily due to a high number of defaulting states (19), but also that ‘big’ industrial states like Tamil Nadu, Maharashtra, Gujarat, Kerala did not report data for that year.

On the other hand, the compliance improved for 2014 – with only 13 states/UTs not reporting data – and thus there was a small uptick. 

Figure 3. Industrial Injuries in Factories, 2005-2014  

In ‘Figure 3’ above, it appears as if 2010 was a bad year for industrial injuries. This is primarily because there were fewer defaulters (13) and also a high level of reporting from industrially significant states (Gujarat and Maharashtra together accounted for 41.64% of injuries that year). On the other hand, from 2012-14, major states like Gujarat, Kerala, Tamil Nadu, West Bengal were defaulters and that’s why industrial injuries appear to go down in this graph.

It is ironical that major industrialised and important states like Gujarat, Kerala, Tamil Nadu, West Bengal, Karnataka, Kerala, Uttar Pradesh, Madhya Pradesh have been defaulters for several years. In a statistical sense, if the same states/UTs continued to regularly not report data, then the observed trends will actually be less affected. But, alas, there are variations. Consequently, in a very real sense, vast volumes of statistics put out by the Labour Bureau is simply not that useful. Indeed, there is no such thing called all-India Data” in many publications put out by the Labour Bureau. 

Labour law reforms and no database

Under Chapter V-B of the Industrial Disputes Act, 1947,  industrial establishments employing 100 or more workers must get prior permission from the government before effecting lay-offs, retrenching workers or closing establishments. Employers have long alleged that state governments often do not give permissions for retrenchment and closure for fear of political backlash. As a result, industries have asked for the abolition of this chapter changing the threshold from 100 employees to 300 employees. It was on this basis that ten state governments changed the threshold from 100 to 300.

But the problem is that we have no statistical basis for the decision. We do not have state-level or national-level data on the applications received for layoffs or closures and then sanctioned or rejected by the government. After all, if we go by Labour Bureau data, we must be forced to believe that between 2005 and 2014, only 12,777 workers were retrenched in 206 units (an average of 62 per unit), and 779 units were closed, affecting 41,192 workers (an average of 54 per unit).

Publications and reports

LB’s publication record is mixed. 2017 is the most recent year of its flagship publications, even while a few others carry data that is fairly recent. However, if we look closely into some publications, several problems emerge is a result of poor data collection and collation. Many reports are incomplete and carry dated data. For example, for most of the variables like factories, shops and establishments, wages, trade unions, industrial disputes, etc. information for the latest year relate to 2013 to 2015.

Another drawback is that availability of LB data is far tougher than that of CSO, electronic or otherwise. To be sure, the CLC (central sphere) does a much better job in data collection than states as data on ESI, EPF, mines, etc. are more recent than what is put out by SLDs.

The ministry of labour and employment claims that “the feedback of accurate, timely and detailed statistics and actionable research on various aspects of labour activities is necessary for taking effective policy decisions”, but it will have to do much more than designing a new logo for LB.

The deficits in the data supply chain involve all actors concerned — trade unions, employers, district or divisional level offices, SLDs/CLCs and the LB.

What is galling is that LB nor these agencies did not care to rectify the “historical blunders” which is exacting a high cost in terms of policy-making. It is a pity that trade unions, who are the quickest to cry victim, are also defaulters in terms of reporting data. Why does their care for Indian labour not carry over to labour statistics?

India is “politely” asked by the ILO to maintain and provide “labour inspection statistics” as required under Art.21 of the Labour Inspection Convention. It is clear, in my opinion, that the Labour Bureau suffers from institutional sclerosis and needs strong doses of vitality. No logo, no matter how appealing, can fix this.

K.R. Shyam Sundar is a Professor at XLRI, Xavier School of Management, Jamshedpur