India Internet: The market size conundrum

Are mistakes in market size the reason for swings in valuations and investor behavior?

Disclaimers:
Views expressed here are my personal reflections and not indicative of the views of my current or previous employers.
No part of this post maybe reproduced or quoted without explicit permission.

Over the last 5 years that I have been involved with Indian internet investing, the biggest shift I have seen is in investors’ (including myself) perspectives on market size. It went from bearish in 2011 to irrationally bullish in 2014/15. Even though macro indicators look in better shape now than in 2013, estimations on market sizes and hence fundraising round sizes and valuations went up sometimes by an order of magnitude! While the triggers everyone has pointed at for irrational investor behavior are the 2014 national elections and Alibaba/JD.com’s public listing, I think these were secondary factors. There are more fundamental mistakes and challenges in market size calculations.

Before getting to some mistakes, let us first understand the challenges in estimating market size in India. Let us pick a simple enough market- hotel rooms. How many hotel rooms are booked in India in a day? If this were the US, this would take one Google query to get a trend of room nights available to be booked and utilization over time. For India, these are the numbers you get:

1. An official hotel industry report states a number of room nights per day (available; not booked) around 200k. This includes only the organized sector (mostly 3 star hotels and above).
2. Pitches from companies in the hotel space where the number of room nights available per day ranges from 500k-2M
3. Using jugaad methods i.e. looking at number of Justdial listings and multiplying by average number per hotel (a guess, no science) which gets it to 300–500k
4. Data on utilization of rooms from calls with hotel owners where the number ranges from 30% in some hotels to 80% in another. So, what is the average?

So, how many room nights are booked everyday in India? Could be 200,000-2 million. The variability is too high. In some sectors, you could narrow it down a reasonable range. In most sectors like these where unorganized retail is high, you could be off by an order of magnitude. How many intercity bus tickets are sold in India? How many cabs exist in India? How many people eat out in India everyday? Same problem.

The issue with high variability is that perception plays a much bigger role than it should. So, in a bull market, you tend to believe the upper end. In a bear market, you tend to gravitate to the lower end. VC valuations generally have limited correlation with current metrics. It is valued by identifying what a company could be worth and then working backwards. The biggest component of this is market size since that is fundamental to predicting future revenues.

So, can one company sell 1M hotel room nights a day in India? The answer in 2012 would have been no way, there aren’t that many rooms to sell in the first place. Look at this industry report which says 200k available room nights a day. The answer in 2015 would be yes, they could. There are 2M room nights, it is growing fast due to good macro conditions and everyone wants to rent their second apartment as a serviced apartment- potentially a much larger market. Huge change in perception, market size, future projections and valuations. Now that the data availability challenge is out of the way, let us look at some mistakes every investor has made at some point in the last few years.

Mistake 1- Sizing markets independent of price/margins

From microeconomics 101, we know that demand and supply depend on price. In my previous post, I had laid down some conditions for building marketplaces by selling below cost. One of the prerequisites I had mentioned there was large market size. However, these two are not independent since market size depends on unit economics (in spreadsheet parlance- a circular reference). This seems fairly obvious. So, why was there miscalculation en masse?

We were previously sizing up the market (e.g. e-commerce) using sensible gross margins. For example, I remember sizing the e-commerce market in India in 2012 assuming that contribution margins (net revenue minus all variable costs) would have to get to at least zero (if not positive) soon enough. However, when companies started selling below cost, it drastically altered calculations and made people question their own previous market size assumptions. In the e-com example, their growth over the next year made it clear that they would cross my five year projected market size (on gross merchandise value (GMV) terms) in probably 2 years. As more and more people started feeling the same way, people looked beyond current numbers. The power law nature of VC returns (where few companies return most of the capital) created massive FOMO (fear of missing out). We forgot for a brief period that the growth was what it was because gross margins were as bad as -20% and you can’t use that to project future growth.

Mistake 2- All revenue is created equal and you can hack your way to sustainable growth

There are 2 easy ways to hack revenue growth- worsening margins or worsening asset intensity (working capital). Selling below cost accomplished the first. Free returns and cash on delivery accomplished the second. There is almost a startup trilemma here- you cannot improve revenue growth, margins and asset intensity all at the same time (all else in the market being constant). When we saw Rs. 100 of topline with -20% contribution margins (CM), we could not find out what the equivalent topline would be at 20% contribution margins. Was it Rs. 90? Was it Rs. 50? What is the elasticity here? No idea but the private capital markets continued to pay revenue multiples on the Rs. 100 of topline. The big mistake was in assuming that Rs. 100 at -20% CM would go to Rs.200 at 20% CM. It is hard to achieve both this revenue growth and margin improvement. As we have now been reading news articles about e-com companies top line shrinking/flattening as they improve CM, we realize that valuing companies assuming both high revenue growth and significant margin improvement was just wrong, at least in India.

Mistake 3- Amazon/JD.com had achieved growth through similar strategies

The part about Amazon is just plain wrong (chart below). Amazon’s gross margins were always hovering around 20% (pre-fulfiment costs; fulfilment is 3–4% of revenues for Amazon)as early as 1995. JD.com did have negative margins (post-fulfilment costs) but publicly available data shows it was probably just negative (-1.4% in 2011). So, the precedents being quoted were lack of operational profitability and not gross profits. Operational profitability (EBITDA) is easier to control in the short term via reduced marketing or headcount. Gross profits indicate true pricing power and are harder to move up (see this recent protest by Flipkart sellers for raising commissions). Without strong network effects (e.g. Paypal) or some serious moat (e.g. hardware companies), it is hard to have sustainable growth when subsidizing both sides of the marketplace.

Mistake 4- Indian consumer market is similar to China

Look at the chart below from a recent Goldman Sachs report (referenced below) on the huge disparity between the Indian and Chinese consumer today.

Don’t let the 1.2B population cloud your judgment. While we have ~70M people with over ~$5,000 of annual income, China has over 400M there. No amount of smartphone and mobile internet adoption will change this. It can only change with improving macroeconomic conditions and governance- things venture capital investors and startups have limited control over. So, the market size is actually not that large. By subsiding below cost, you hypercharge growth and get to this ~70M people and then realize that there is not much market left to continue this growth and improve profitability. This forces the company to be profitable by de-growing a little or by being flat. So, instead of going from Rs. 50 at 0% CM to Rs. 75 at 10% CM, the company goes from Rs. 50 at 0% CM to Rs. 100 at -10% CM to Rs.75 at 10% CM. In the process, a lot of capital gets burnt (for the same eventual outcome) in the years where companies are selling below gross margin and forces valuation markdowns in between. Companies could have gotten to the same stage they would eventually get to while burning tens to hundreds of millions of $$$ lesser.

Mistake 5- US/China clones would have large markets here

As you saw in that chart on income levels, a large chunk of the population and the next 100M people who will come out of poverty are going to earn Rs.1.5–3 lakhs. The Goldman Sachs report (referenced below) explains how Indian middle class will continue to mean urban mass ($2–5k p.a.) unlike China where it means urban middle ($10k p.a.). At this income range, paying for efficiency/convenience (e.g. Rs.60 delivery fee for Rs. 300 order value of food) becomes a luxury and not a primary market need. While marketplaces may have attractive dynamics globally for high return on capital invested, most clones may end up just catering to the ~30M people with $10,000+ household income or to the ~70M people with $5,000+ annual income. The really large marketplaces would need to cater to people with $2,000–5,000 annual incomes with tight margins. So, selling below cost early on maybe a bad strategy -you cannot make a good return on investment since profit margins may not be high enough later on for customers in this income segment.

This also explains the common question observers ask about Indian internet companies — “Why are they in so many products/geographies? Why can they not be focused on solving one problem like US and Chinese companies?”. Well, to build the same $1 million of market cap in India, you do need to do a lot more given the market size restriction. Entrepreneur aspirations and VC fund sizes are not adjusted for this difference in market size. So it pushes capital consolidation in a few companies in each sector and forces a company to build out multiple products.

So, where are we today? Let us look at the largest internet sector i.e. e-commerce retailing. Every report on Indian e-commerce is quoting a market size ranging from $50–100B in 5 years. My guess is that a lot of them are assuming growth on top on today’s topline (one with unsustainable gross margins). Even today, it is hard to estimate what actual toplines would be when you convert it to sustainable steady state economics. We maybe continuing to make some of the mistakes stated above.

References and acknowledgements
1. Goldman Sachs Asian Consumer, The Indian Consumer Close-Up, June 2016

2. The point around efficiency based products not being a primary need for the masses in India was originally Kunal Shah’s (Founder, FreeCharge) idea which he brought up in one of our conversations about India internet.