Opinion | A comfortable number turns inconvenient

There are both classification and industry size estimation issues with Nasscom's digits

For many years now, the captains of India’s information technology (IT) services industry, and indeed many investors, have relied on numbers from the National Association of Software and Services Companies (Nasscom) to help guide growth expectations for IT services. One number that is often spoken of in guidance estimates for the industry is the yearly estimate that Nasscom puts out for Indian IT exports. I, too, have relied on this number as a benchmark.

Imagine my shock then, when I learnt last week that the Nasscom number is out of sync with India’s balance of payments (BoP) data. It turns out that while software exports recorded in BoP has been in the rather tight range of $73-77 billion over the last four years (FY14-18), the Nasscom-estimated India IT services exports has grown from $98 billion in FY14 to $125 billion in FY18. The difference between the BoP numbers, which show little to no growth, and Nasscom’s number, which shows a growth of 28% during the same period, is stark. Given compounding, Nasscom’s number exceeds the corresponding BoP number by almost 60%.

I called Viju George of JP Morgan, a fine analyst whom I have known for years, who reported this discrepancy. George clarified that Sajjid Chinoy, JP Morgan’s economist, was the first to ask why there was such a stark difference. George set about analyzing the reasons for the difference. Since this immediately begs the question about how much sanctity Nasscom’s India IT exports data has when compared with the software exports data recorded in the BoP, he confirmed that he had also spoken to Nasscom about his analysis.

While there are several nuances, and I suggest the reader try to get a copy of George’s analysis to satisfy himself or herself with all the data, here is a brief synopsis that covers the lion’s share (~75%) of what constitutes the difference:

(a) Nasscom’s definition of exports includes overseas income booked by overseas subsidiaries of India IT companies; though part of the industry, this income is not ‘exports’ as per the BoP. George estimates that this factor alone explains 50-60% of the absolute dollar gap between Nasscom industry estimate and software services exports in BoP. To my mind, this is actually a positive. It means that Indian IT services firms are now booking and deriving more of their revenues from outside India, thereby indicating a sure trend toward becoming truly global firms versus just Indian firms that operate globally.

(b) Nasscom faces limitations in obtaining revenue from all its member companies for forward estimation. As a result, Nasscom estimates revenues by using fairly aggressive pricing taken from “best-of-breed" listed firms. This pricing is then extrapolated to cover a large swathe of the remainder, including ‘captives’ or global in-house centres (GICs) operating in India. GICs don’t operate at best-of-breed margins, they merely transfer cost over to their mother companies while including a small mark-up to approximate notional ‘earnings’ to satisfy the tax authorities. If GICs continue to grow at a scorching pace, this gap will become wider.

It would appear then, that there are both classification and industry size estimation issues with Nasscom’s number.

Until FY10, the two data points had been virtually identical after which divergence crept in and has expanded ever since. The reason for the divergence is the change in methodology followed by the Reserve Bank of India (RBI) in computing software services in the BoP.

Prior to 2010, RBI used the Nasscom data as the controlling figure. In 2011, RBI recommended that for purposes of BoP, software services data be compiled entirely based on the banking channel data from receipts through the Foreign Exchange Transactions Electronic Reporting System (FETERS). Other sources (such as Nasscom) were to be used for validation purposes only. This change made the BoP data for software exports a sub-set of the Nasscom estimate as only a portion of revenues of the sector gets routed through FETERS.

To be fair, George points out that Nasscom’s estimate is likely still the most reliable industry estimate that is available, outside of RBI’s strict accounting for software BoP. Given the discrepancy and the fact that a large swathe of the investing public gauges sentiment by this number, it is probably time the methodology is revised.

Siddharth Pai is founder of Siana Capital, a venture fund
management company focused on deep science and tech in India.