Last week, a very important statistical event occurred in the economic history of India.
Government data, based on the additions made in payrolls from the Employees' Provident Fund Organisation (EPFO), the National Pension Scheme (NPS) and the Employees State Insurance Scheme (ESIC), say that 31.1 lakh new jobs were created in formal sector between September 2017 to February 2018.
These releases are given in agewise buckets of 18 to 21, 22 to 25, 26 to 28, 29 to 35 and above 35.
Assuming that anyone who gets a job in the organised sector will get an EPFO or an NPS or an ESIC number, we can assume that new registration in the 18-25 years category represent a new job created in the economy.
So does India finally have employment data? What wonders can we do if we get this labour data regularly? Will it greatly improve our understanding of the economy and will it help us make better policies?
To find answers to these questions CNBC-TV18's Latha Venkatesh caught up with TCA Anant, former chief statistician of India, Prof Himanshu, professor of Economics at JNU, Soumya Kanti Ghosh, chief economic advisor at SBI and Ananth Narayan, the market expert.
Last week a very important statistical event occurred in the economic history of India, the government released new registrations to the EPFO or the Employees Provident Fund, to the NPS or the National Pension Scheme and to the ESIC or the Employees State Insurance Scheme. These releases contain month wise new registrations starting from September 2017 and up to February 2018. These releases are given in agewise buckets of 18 to 21, 22 to 25, 26 to 28, 29 to 35 and above 35.
Assuming that anyone who gets a job in the organized sector will get an EPFO or an NPS or an ESIC number, we can assume that new registration in the 18-25 years category represent a new job created in the economy.
So does India finally have employment data? What wonders can we do if we get this labour data regularly? Will it greatly improve our understanding of the economy and will it help us make better policies? To find answers to these questions CNBC-TV18’s Latha Venkatesh caught up with TCA Anant, former chief statistician of India, Prof Himanshu, professor of Economics at JNU, Soumya Kanti Ghosh, chief economic advisor at SBI and Ananth Narayan the market expert.
Q: Since you have had access to this data even before it was publicly available. Tell us what are the key learnings that you have got from this six months of data?
Ghosh: It’s a good beginning which the government has made in terms of payroll data and if we just go each of the segment separately. I think first with the Employees Provident Fund Organisation (EPFO) data, the first thing is the government has given out the data from September to February and most of this data is - one, Aadhaar linked so that means there is no duplicity of the account. Second, the date is a net that means it take out the people who have left in that particular age group. Third, it comes after the amnesty scheme has ended in June 2017 and fourth, it gets the data in different age buckets. So that could give some sort of an approximation of the distinction between payroll and new job creation.
The Pensions Fund Regulatory and Development Authority (PFRDA) data is quite clean because most of the data apart from corporate accounts, which is a duplicity of account could be new payroll which have been offered but the Employees' State Insurance Corporation (ESIC) data, I am little bit skeptical because this data is still being cleaned up and if you look into the data carefully, the data is not Aadhaar linked. So all in all a good beginning but road to go.
Can we draw any conclusions about the unorganised sector from organised sector assuming we are going to get it every month?
Anant: About a year-and-a-half ago there was a report authored by the then Chairman of NITI Aayog on making effective data architecture for labour markets in India. One of the thing which we have pointed out is that a good architecture for labour data has different pillars from which labour data is drawn.
The report recommended that there should be regular system of both household and establishment surveys, there should also be a system of regular counting of establishments and there should be better use of administrative data in which the report had argued that we are not making adequate use of administrative data collected in India through organisations like EPFO, ESIC etc.
What the government has done by this release is bring the first pillar towards complete data structure for labour market studies in India and this is one of them. We will be getting the other legs of this architecture later in the year and maybe next year because you will start seeing regular data from the labour market surveys that etc.
Please understand these are complimentary things to understand labour markets. This is principally giving you inputs of people who are working in establishments which are covered under the provisions of these acts. Do they give you an indication of what happens elsewhere? The answer is yes, they can because normally we believe that economic activities correlated and there are interlinkages between formal and informal sector but what is the nature of those interlinkages will become clearer when all these time series are flowing and people will have time to study them. So I wouldn’t jump to premature conclusions about what has happen in the informal sector but in the long run this will an important pillar to understand both the formal and the informal sector.
While the NITI Aayog maybe commissioning and labour and household surveys are underway, actually the National Sample Survey Office (NSSO) surveys were stopped you were telling me. We do not have historical data for the last five years from the NSSO?
Himanshu: You are right about it, in fact that is the point I wanted to make that we do have a system and for overall economy if you want to know how much jobs have been created then there is no substitute other than the household survey and NSSO were the workforce as far as employment estimates were. Between 2004 and 2011, we had five large surveys where we could get information on what is happening to employment and unemployment and some of them were annual surveys. So we had this.
After 2011-12 is when we had the crisis, when we do not have household surveys which can supplement the information, give us information on how much jobs being created in the economy. The labour bureau had two data sets which is before 2014 and 2016 and that give the information that the overall employment has come down. So there are household surveys and they are the only indicators of what is the total amount of job created in the economy.
This is a data on number of accounts and extrapolating it to the number of jobs created in the economy may not be the right way of looking at it. At best you can say that these are the number of jobs that have been created in the organised sector where these are applicable. There are problems with ESIC and EPFO overlaps and if you look at only the data for people who have been given employment, assume that it’s a correct data then less than 25 only 2 million is the number which we can get, but remember one thing and that’s something which I want to come back to the NSSO service; between 2004-05 and 2011-12 for which NSSO service are used which everybody accepts; 49 million jobs were created in non-farm sector that is seven million jobs per year but still we say that the annual job creation was only two million per year. The reason was that 35 million jobs in the agriculture sector were lost – that is people moved from agriculture sector to non-agriculture sector.
So again in the informal sector the trend may not be exactly the same even if you assume that all the seven million jobs that have been created are in the organised sector then the information sector or construction sector or agriculture sector will be adding jobs to the same extent may not actually be true, but these are things which have been pointed out by the NITI Aayog document itself.
It says very clearly that the EPFO data does not – the accounts itself need to be interpreted very differently because if you have 19 employees and suddenly you have 20 employees then all 20 are counted as new employees whereas they are not. They were actually working in a different capacity. Same claims can be made for National Pension System (NPS), for example if you are working as an adhoc employee and suddenly you become a permanent employee, you join as a member of NPS but there is a great distinction between accounts and social security measures and the number of jobs. You certainly cannot make any comment as to how much jobs have been created in the economy just based on this data which represents less than 10 percent of the total workforce of the economy.
As professor Anant was saying the NITI Aayog is looking at this as one leg of employment data. Eventually or maybe very soon, by the end of the year we would have labour and household survey data as well and services data as well. Do you think if those also come on board, we will have a slightly better picture of employment data?
Himanshu: Definitely and PLF that is periodic labour force service which the NSSO is conducting, will give us estimates of employment on a quarterly basis and we will have annual basis employment data which will be coming out from NSSO and they cover the entire economy and they are going to be much better estimate of what is the job situation in the economy as such but I am happy that this data at least is being released and probably after five years or ten years when we have calibrated this data to the data that is coming out from the PLF and NSSO.
We would be in a better position to chalk out not just the extent of employment but also the extent of formalisation in the economy as far as jobs are concerned and in that sense I welcome the data release and the way the data has been released is something that should continue but we should marry different kind of data; enterprise data, the NSSO data and this kind of a data to have a holistic picture of what is happening to the employment estimates.
Even if the other legs that we have been talking about - labour and household survey will still be surveys. It looks like even if all the three legs that the NITI Aayog spoke about is available with us, even then we won’t have an employment number as is available say in the United States – the non-farm payroll number or we won’t have an unemployment figure like the US labour department is able to release, is that right?
Anant: No. When you look at data sources, when all of these are in place we will have exactly the same sort of information which is available in the US. It is important for me to make one correction, also slight addition to what Himanshu was saying, these do not represent a holistic picture of employment in the economy as a whole which a household survey provides which is the case all over the world.
However, what they do give you is a very valuable insight into a particular set of employment which is taking place in entities which are covered under the provisions of these act. So far as what you might call aspirational jobs in the country are concerned, they will all be contained in this set. To that extent a very high growth in this is a signal that the growth in a particular class of aspirational jobs has increased.
How does it happen in the backdrop of maybe people withdrawing from the labour force in less aspirational conditions, maybe from agriculture and so on, these are things which will only be possible when full data is available and careful analysis is done. I don’t think we should take away from the conclusion that this does represent a fairly healthy growth in what are in fact improves aspirational jobs.
Let me take Himanshu’s example and elaborate further. Supposing an ad hoc employee becomes regular and therefore gets ESIC number or a PF number, from his perspective he is better off. So, even if we set aside the detail of whether it is a new job or not, clearly there is an improvement in living standards which this increase is showing.
Will we be able to have this monthly figure sometime soon, say within a year or so, will we be able to say non-farm jobs created in India in the month of February was two lakh?
Anant: Monthly all India will not happen in a hurry because what NSSO is planning to give on a quarterly basis is only urban areas but I should point out that interpreting employment data in India is going to be a much more tricky task than interpreting it in the US. We will continue to be for a long time a highly seasonal economy.
The share of agriculture in employment has come down but the influence on agriculture in a number of activities is still quite large and the implicit seasonality which that creates will continue in Indian employment much more than it is present in the US or in Europe. So, a level of sophistication in labour market studies will be called for in Indian data which sometimes is not present.
Assuming that something like 200000 numbers were added to the EPFO last month or some such data should be available on a monthly basis, assuming in a year it will be seasonally adjusted as well, what would it add to the markets understanding of the economy?
Narayan: It will be a great step forward but we have got to be careful about how we go with this. To start with, the monetary policy framework that we have makes no mention of employment directly. It has an oblique mention to growth but it doesn’t even look at employment as a static number. The second part is, there still will be confusion as far as the market is concerned on the actual flow component of this particular number for some time to come, until there is some stability around the series.
You mentioned non-farm payroll in the US which is the most important data point which comes out every month across the globe, not just in the US, but that credibility has been built over several decades and not just over years and the whole process is now completely understood and accepted by the market place. It will take a long time for India to get to that stage. Even this debate that we had on the Ghosh paper about is this really formalisation of jobs or is it genuine new jobs being created, can you therefore extrapolate from this to unorganised sector especially post GST and demonetisation, you saw the controversy around that.
So, the market will take the number with a pinch of salt as well. Having said that use of big data of this kind to actually get some credible numbers and no market participant has looked at those NSSO numbers for employment ever. At least now we have something which maybe the market can start to get a feel around and over time if you add to this other big data, maybe using Jan Dhan account credits to see what kind of employment is coming through, then maybe we come closer to getting a more reliable number for the market participants.
Surely we can have more colour from this data, probably the next step would be to get even some sectoral classification of these EPFO, NPS, ESIC registrations, you think that should not be too difficult?
Ghosh: One of the points made by one of the earlier speaker was that NPS data is actually ad hoc employees, it is not correct because anyone who is joining the Indian pension scheme after 2004 has to mandatorily comply with the defined pension contribution. If you look into the data, it talks about the central government state government employees who are enrolling in different age buckets and it doesn’t talk about the corporate accounts which could be a duplication.
So, from that point of view it is actually not ad hoc. Having said that the specific point of whether it could be given across industries, I think yes, that could be an excellent step because the EPFO data covers 190 industries and the ESIC data covers 65 industries. So, if we can have a breakup of 190 industries on a monthly basis just as the US payroll does, we can have a fair bit of idea of where the jobs are getting added, where the jobs are getting deleted and the skilling programme which you are doing currently can be entirely geared towards making the people employable in the long run. EPFO jobs are most low paying jobs, paying salary of less than Rs 15000.
What about the point that if and when, maybe in 2-4 years, if we get some kind of a time series and all the legs of the data are available, monetary policy goals also may have to be tweaked you think?
Ghosh: Yes I believe so. I think that is a very fair point because any monetary policy document there is always a disconnect between the actual growth and the employment numbers, the inflation numbers because every time we are unable to collate how these two relates up. I am sure that over a point of time as the data builds up, it becomes credible.
Now this data needs to be credible over a point of time. The monetary policy needs to have a look into the data to understand the seasonal patterns and then come to some sort of a deseasonalised construction of the data so as to get a better deal. So, I think from that point of view the policy making should now become a little more easier than what it was earlier.
Already looking at the CSO data and the various things that are put out – the survey of industries etc, is CSO in a position to connect between organised and unorganised sector? If organised sector grows 1 percent, unorganised sector grows by say 75 basis points, is there any connect that CSO is already able to make so that we can use the EPFO, NPS, ESIC to arrive at some larger conclusions about the economic growth?
Anant: People have done that but they are not very robust because till recently informal sector estimates basically came only once every five years. They were built into our base revision exercise of the national accounts. Things which are that infrequent even over a 30 year span point, you only have 5 or 6 data points, that is not really large enough for you to build the sort of co-relational structure that you are trying to look for.
The whole point of the report and asking for a regular time series of data flowing from informal sector and formal sector, was to allow this sort of careful modelling to become possible.
Starting from the point when the recommendation is made, implemented on the field and a time series is built-up will take time.
When do you think we can get a composite number in terms of what you can call employment number and unemployment number for India?
Himanshu: That can only come after the National Sample Survey Office (NSSO) survey are released. That is something that will also give a kind of a benchmark to calibrate the data that is coming out from the EPFO data and the NPS data.
That is how many years from now?
Himanshu: The NSSO data is now going to be quarterly and it is going to be annual. So, it will take at least 4-5 years before we can make some kind of a conclusion as to how this data is matching upto the data that is based on the household survey. Not before 5 or 6 years when the data has stabilised, we will be in a position to say about the impact of these numbers and whether we should take these as a measure of job creation or should we take it only as measure of formalisation of the economy. I agree that this is a measure of formalisation but not definitely about jobs in the economy as such.
We in the market can get excited with even small slices of data, we get excited with PMI but still it can move the needle in markets and of course other things like auto sales or cement sales, all this excites the market. Do you think that if we get this month wise employment data or month wise registration numbers from the three bodies, we are going to have another interesting set of data that the market and both investment participants and policy makers can look at?
Narayan: We have always known that employment is a critical sociopolitical economic variable. Unfortunately while we have NSSO surveys coming out and giving us data, it has been too late or too sketchy for us to actually use it as a market participant. So, this is a great start. In fact maybe Soumya can do some research on a back testing basis to tell us what would have been the numbers looking like one year ago, two years ago or three year ago so that we even generate some time series as sitting today rather than waiting for it to build over time.
Likewise if we actually want to measure what the flow looks like and how much of it is formalisation versus new jobs, maybe we can also look at the numbers beyond 25 years to see how much has been added there and that most likely won’t be new jobs, so it will give you a measure of how much of formalisation is happening. So, there is data that can be tweaked and over time as the database builds up and more analysis happens, I think it can be a very important input into both our policy making as well as markets. Markets need any excuse anyway to play around, I am sure we will start playing around on this as well.