Q3 GDP puzzles economists; TCA Anant provides clarity on nos

Macroeconomic data released on Tuesday left economists in India scratching their head as the numbers didn't seem to reflect the impact of demonetisation.

Speaking to CNBC-TV18, TCA Anant, Chief Statistician of India said: "One should keep in mind that anecdotal evidence is not the same as the estimates derived over indicators which are aggregated over a large number of sources."

On the argument that the gross domestic product (GDP) number might be overstated as formal sector data is used as a proxy for the informal sector, Anant said that CSO accounts for 70-75 percent informal sector from data sourcing from means other than the formal sector activity.

When asked about the growth estimates for the fourth quarter Anant refused to give any forecast and said that he would wait for the numbers to come.Below is the verbatim transcript of TCA Anant’s interview to Latha Venkatesh on CNBC-TV18. Q: Let me start exactly with that, the fact that some of the numbers were puzzling because anecdotal data or micro indicators were indicating more slowdown. The specific consumption part of the data is in that trade hotels segment. Now we had evidence that retail trade and wholesale trade were indicating that demand was very poor, how come your trade and hotels indicated a 7.3 percent growth? A: There are details given in the press note and you can take a look at them. However, I can refer to one of the indicators which is used when we compute this segment which is sales tax collections. As you are aware, sales tax is in fact collected on the basis of goods sold and we from that assess a volume indicator for trade and that is one of the factors behind it. So, one of the things which one must keep in mind, anecdotal evidence is not the same as estimates derived from indicators which are aggregated over a large number of sources. Q: I just wanted to know is there any specific part, since we had hard numbers on trade, freight rates fell because goods were not moving and therefore was there any element in that which did better, which made up for something else. A: They build it up from component but I don’t have the access to the detailed component wise calculation in front of me. So, I can’t give you an answer on that. However, as I told you, one indicator which does feed into that calculation is also available to you and you can see that it is significantly correlated with trade volumes. Q: Let me look at another thing which puzzled me, your overall full year gross value added (GVA) is 6.7 percent and as of now you have the quarterly breakup of 6.7, 6.8 and in this quarter 6.7 percent. Now, if I subtract, then the current quarter, Q4 GVA should come at 6.5 percent; if you just do the math. Now, what is puzzling is, if you looked at the micro data, for instance two wheeler sales, two wheeler sales were terrible in November and even worst in December. They have improved rather dramatically in the January and February numbers. However, your GVA numbers are so sedate, 6.6 percent in Q3 and by subtraction 6.5 percent in Q4, yours is very ironed out. The anecdotal data is very volatile. A: Let us not build from scooters to the overall, but let us try to -- the manner in which we do the exercise, we first make the annual estimate in all of these. In fact because our indicators set for quarterly tends to be fairly limited, so, what we do is we compile the annual and then the annual is allocated to the quarters using the indicators which are available on higher frequency for quarterly calculations. So, for that we also need to make an assessment of what the full year movement, for example, IIP. We have about now nine months data for IIP, so, we make a projection of the nine months onto 12 months and then using that rule of projection, allocate the quarterly IIP to get a sense of what the full year break down is. Now, this is admittedly the quarterly assessment, is therefore a littler coarser and we have pointed this out earlier also when we had discussions on quarterly estimates and discrepancies in quarterly estimates and I pointed out, if you remember last year, we had discussion and you were saying why is the discrepancy in quarterly is so high, I said quarterly is a much coarser allocation. We don’t have that degree of granularity in quarterly allocations. The annual tends to be a little more robust partly because of the law of larger numbers as well. Q: You have such an intimate understanding of the numbers, is it possible that the Q4 will be lower than your estimates? A: At this point of time, Q4 is still unfolding. Q: Two thirds has already unfolded. A: We don’t even have the IIP figure for the month of February as yet. Two thirds has certainly unfolded, but I have to go again by estimates and at the moment, in another months’ time, or another 10 days’ time, you will get the IIP figure for January. Now my point is the Q4 will be built up from all these numbers, so, in terms of the realm of speculation, people are of course free to speculate it in every possible way they like. However, at the moment, all that we have is the figures we have and from that we build the best possible assessment. Q: Another thing that puzzled me, your revision of the previous quarter. Look at the quarterly breakup of FY16 – Q1 is 7.8 percent, Q2 is 8.4 percent, Q3 is 7 percent and Q4 again 8.4 percent. It looks like Q3 very conveniently has gone down to make the current Q3 look good. How is that you have 8.4 percent in Q2 last year, 8.4 percent in Q4 but suddenly a depression of 7 percent in Q3? A: Allocation of last years numbers to quarters is based on a number of indicators, some of which have remained the same, particularly IIP and so on but some have been revised based on the actual output. I will give you one example – agriculture. If you look at the detailed data comparing last year and this year, you will find that the agricultural estimate for Q3 has changed. Why did that happen? What has happened is agriculture actually comes on a crop calendar but the output doesn’t come in quarterly terms. Full year output is allocated in quarterly terms by the agriculture ministry according to a certain harvesting and cropping calendar. Now the final estimates of agriculture, remember what feeds into Q3 and Q4 estimates were finalised only after our last years estimates were closed. At the time when we made the PE we were still working with third estimates of agriculture. So, these updated figures were incorporated in our RE which was released in January and once they had been incorporated it worked out like this. Now you can turn around and say this is what happened in agriculture. In other segments look at it segmentwise you will get the story emerging of what happened in Q3. Q: I just wanted you to explain to us, what is the percentage of informal sector, is it that informal sector is 45-48 percent out of the GDP? A: Informal sector in the parlance which is used in national accountings term is the household segment is about 40-45 percent. Q: The Chief Economic Advisor was explaining after the economic survey that you all have the value added formula, the input output formula for the formal sector and you extrapolate it for the informal sector. Is it not possible that in the quarter gone by the informal sector could not have been as good as the formal sector because it was more cash dependent. So, are the final figures likely to look lower? A: One of the things about informal sector is there is no direct data available to anybody. You can assess this in one of two ways. You can do anecdotes which is one approach to it but let me give you the approach which is followed in CSO and argue that it is a little more robust. I will break up the informal sector into four components, one of the largest components is agriculture which is pretty much all informal. In agriculture we assess GVA by a combination of data on output which is generated from data on acreage and yield, for which there are well established regular survey procedures which everybody is aware of and those are then conflated with value added data which comes from the cost of cultivation survey to give us an agriculture value added. This is about the most precise you can get for agriculture for anybody. So, that is a chunk of the informal sector. The second chunk of the informal sector is in trade. In trade we take an indirect assessment of the informal sector by generating a volume measure of the non-corporate sales tax collection, that is sales tax collection emanating from non-corporate sources, generate a volume measure from that and get it. Again it is an indicator of what the volume of trade in the informal sector is. It is an indirect measure but these people are paying sales tax by the fact that their goods are tax paid. The third area where informal sector activity is significantly large is in construction. In construction we use an indirect measure of assessing informal sector activity by the offtake of consumption or consumption of steel and cement which again is available to us but it is an indicator of volume. The only place where we apply formal sector value added to assess informal sector activity is in the realm of manufacturing, where we take the IIP for the two digit categories which overlap with the informal sector, admittedly those are in the formal segment and apply those two digit categories growth to the informal sector. However in the manner in which I have described to you, please note we have accounted for about 70-75 percent of the informal sector by means other than formal sector activity.Q: The short point, is that when the final numbers come, it need not necessarily be that the Q3 number is going to be revised lower? A: It need not necessarily be; that is correct. However, let me explain what are the major sources of data which we are missing, and that is something that everybody may want to keep in mind, and it is a very important data which will be coming in next year, at the moment our data is based on the advanced filing information of listed companies and as many people agree, these are the larger companies which do advanced estimates. We will get a much fuller corporate estimate when the larger set of companies prepare and finalise their accounts after the close of the financial year and file these accounts with the ministry of corporate affairs. This will take place in period starting after September 30th usually which is the last date for this purpose. Now, that will give us information about the corporate segment which is smaller than the listed companies and the large companies. How does that picture pan out? We will only get to see when that number comes out. It is anybody’s call which way that can go. It is a big difference, in the past for example when we did the revision in 2015-16, that led to a significant upward revision. It can happen, so, let us see how it comes about. Q: That revision happened because of the formal sector or informal sector? A: These are MSMEs, they are corporates, they are essentially, if you like, the middle segment of the corporate economy which is very large. However, they are formal because they maintain accounts, they are not bidi patri wala, but they are – want to call them formal -- I think the correct word is they are formal, they are organised, they are registered as companies, but they are smaller companies. Q: Finally when are we getting the back series, we have only from 2011 onwards, we wanted to know from today’s standards, what was that 9 percent GDP of 2008 likely to be in the new series? A: I appreciate the econometricians demand for a back series because he wants consistent time series. The challenge for us is because some of the data sources which we have used, in particular the corporate database, because we are now using 5 lakh companies, we are finding that the challenge of extrapolating from the visible set which is the few large companies onto the 5 lakh companies is proving to be quite difficult and there are anomalies which keep cropping up in analysis. We are trying to see what we can do best. However, the point is, if you wish and you can certainly do that, the data sources and methodology of old the series is available. Please understand the back series is not merely about using MCA data versus using the Reserve Bank of India (RBI) sample, you are more than welcome to compare the RBI sample to the MCA data and you will see which way the growth goes. Q: I only want to know how worse we are compared to the 2003-2008 period. A: I think the answer of worse or better is a much more complicated thing and my advice to all of you and particularly financial analysts, in your rush to if you like decry a set of numbers, don’t look for things which may not be found elsewhere. These are very robust numbers. Q: Not decry at all, it is only to understand the economies trajectory, have we genuinely slowed, have we not slowed? A: Let me put it this way, the economic trajectory, you have three years of overlapping data available, the relationship between the old series and the new series are available to you in the three years data. What will be clear to you is the directional movement is more or less the same. Now, if the directional movement is the same, this estimate which we have got will certainly bare out whatever characteristics you have seen in the Q3. People would like to state numbers according to preferences, but my suggestion is rather than trying to make the numbers fit your preferences, it may be better to make the analysis fit the numbers. Q: If you are a betting man, last year’s 7.6 percent became 7.9 percent, this quarter 6.6 percent will become what, will it become 6 percent or will it become 7 percent? A: If you had asked me this last year, it would have been very interesting because I remember reading one academic paper where a person convincingly argued that we invariably revise downwards every year to make the growth of the current year look better and this was based on our exercise of analysis of two years. Fortunately for me, the third year turned out to be different. So, let us say that it could go either way. Q: What is your sense, since you know the numbers like no one else, is there a chance? A: I unfortunately don’t talk about forecasting, so, I will wait for the numbers to come out. I am not going to speculate on what they are. I simply don’t have a long enough time series to speculate.