Where is growth in India coming from?
TL;DR: Sector-specific data, satellite imagery, and vehicle registration data provide some answers
Growth in India appears to be slowing right now — real GDP growth rate has dropped every quarter since March 2018, and was a dismal 5.83% in the Q4 (Jan-Mar) 2018–19 quarter. This was the lowest GDP growth since Q4 2013–14. Moreover, real annual growth in FY 2018–19 was a relatively tepid 6.81%, the lowest it has been since 2013–14.
This is obviously concerning. But GDP growth is a coarse measure. It does not tell us how individual regions and industries are growing.
For investors and corporate planners, this is irksome. India’s overall GDP growth has little bearing on — say — the growth of a region like Assam or the growth of an industry like Pharmaceuticals. While state-level GDP figures are available, they are released more than 18 months after they are recorded. They also happen to be fairly unreliable.
At Loki.ai, we tried to think through alternative metrics for estimating industry-, state-, and city-level growth. They have some flaws — but offer perspectives that aggregate metrics do not.
These metrics were divided into two categories — those provided by government or private agencies, and those obtained through measures like satellite imagery, vehicle registration data, and electronic payments data.
A. Government/Agency provided
A1. Index of Industrial Production
The Index of Industrial Production is a monthly index created by MOSPI that measures the production volume in different industries. It is updated monthly, but does have some drawbacks.
First, it only covers the manufacturing industry and completely misses out on the service sector. Second, it only looks at production volume and not value. As such, it will not capture the impact of a shift from high-volume low-value production to low-volume high-value production (or vice-versa).
These caveats notwithstanding, there seems to have been a recent production boon in industries like wearing apparel and food items, while other industries like motor vehicles, furniture, and paper have taken a hit.
A2. RBI Consumer Confidence Survey
The Reserve Bank of India (RBI) conducts regular surveys in understand how urban households feel about their economic situation. This survey has north of 5,000 randomly-selected respondents. However, it does not look at smaller cities and rural areas.
The Consumer Confidence Survey paints a grim picture of the Indian economy. Since demonetisation in November 2016, household’s perception of changes in in employment prospects has been overwhelmingly negative.
A similar trend is observed in people’s perception of their own income levels. While the numbers are still positive, they have seen a significant drop from the pre-demonetisation days.
However, city-level data shows that these trends are not evenly distributed throughout the country. For instance, the cities of Guwahati, Bengaluru, and Hyderabad saw strong employment prospects in Dec 2018. On the other hand, Kolkata, Chennai, and Delhi saw employment prospects that were below All India levels.
A3. Naukri.com Jobspeak Index
Another measure of growth is number of jobs created by Industry. While official data for these is not available, Naukri.com’s Jobspeak index is a good proxy for formal sector jobs. This measures the number of jobs posted on Naukri.com by industry.
This shows that sectors like Automotives and Industrial Products had negative growth in jobs posted. The same was true for the telco sector.
However, sectors like software, accounting, and advertising continued to show strong growth in jobs posted.
B. Data observed in the wild
Data mentioned in Section A is of high quality, but is only limited to the formal sector. Moreover, it disproportionately covers urban India and is unable to quantify growth in more rural parts of the country.
To counter this, we tried to create metrics from data observed in the wild.
B1. Satellite-based Nightlight Intensity
NASA releases images of nighlight intensity in different parts of the world every month. These images can be used as a proxy to see how developed a region is. An example of one such image is show below.
We created an algorithm to quantify the intensity of these nightlights at a city level. As expected, this intensity was highest in big metropolitican areas like Delhi, Mumbai, Bengaluru, Hyderabad, Chennai, Kolkata, Lucknow, and others.
We then tracked how nightlight intensity changed over time. It’s important to note that an increase in growth is not always accompanied by an increase in nightlight intensity. For instance, banks making more money in Mumbai will cause high growth but are unlikely to change the nighlight intensity of Mumbai. However, increase in nightlight intensity does correlate with more factories set up, more houses created, and more roads built.
We noticed that the highest increases in nightlight intensity from 2014 to 2018 came from under-developed regions like Bihar and eastern Uttar Pradesh. The city with the highest increase in nightlight intensity in India was Purbi Champaran in Bihar (388% increase from 2014 to 2018). In Uttar Pradesh, Domariaganj had the highest increase (160% increase from 2014 to 2018).
B2. Vehicle Registration Data
In addition to nightlight intensity, we also obtained information about monthly scooter/motorcycle and car registrations in individual Indian cities. Using this, we were able to quantify how these are changing throughout the country.
For instance, Bihar saw stupendous growth in 2-wheeler registrations in the first half of 2018. On the other hand, growth in Karnataka has been mostly flat, and had faced a sustained decline in recent months.
Curiously, we also found that the city of Amethi, Uttar Pradesh (the former parliamentary constituency of Rahul Gandhi) had an absurdly high growth in 2-wheeler registrations throughout 2013. We will leave speculations around the reasons for this to the reader.
In the last 6 months (Feb — Jul 2019), we have found that areas in the North-East, and parts of Bihar, West Bengal, and Chhattisgarh have had the highest growth in two-wheeler registrations.
Conversely, parts of Punjab, Gujarat, and Maharashtra had the lowest growth in two-wheeler registrations.
This data is not completely indicative of growth. Changes in two-wheeler registrations might be also be influenced by availability of public transportation and ride-sharing services. However, we believe that it is a strong indicator of growth and changes in income levels in less urban areas of the country.
B3. Electronic Payments Data
Electronic Payments are the least useful indicator in this post for measuring economic health. They are still being adopted by much of the country, and changes in their trends holds little value for measuring economic health right now. However, we believe that they will begin to have significant predictive value once they have been widely adopted.
We have included some charts of payment indicators in India over the last 5 years in this post. It’s interesting to note that India transacts 30x more value in mobile-banking payments compared to mobile-wallet ones.
These indicators are far from perfect — but offer perspectives that official numbers are not able to. We will improve them over time, and will soon create a dashboard that allows users to track all of these in close to real-time. If you would like early access to the dashboard, please do drop me a note at rishabh@loki.ai.
Interested in learning more? Check out Loki.ai, drop me a note on Twitter (rishdotblog), or send an email at rishabh@loki.ai.
