Hi everyone – this is Part 3 of our Coupang writeup. In this part, we cover our revenue build and valuation, as well as share what we consider to be risks.
If you haven’t already, you can read Part 1 here and Part 2 here.
Hey there! Welcome to DJY Research. Every month, we produce an extensive research report on a global company. So far, we’ve covered companies like Tencent, Alibaba, Nintendo, and Grab, and will be releasing our 10th report shortly. We have decided to share a sample report for free, and will be releasing our Coupang report (Oct. 2021) in 3 parts.
This is Part 3 of our Coupang research report. In this deep dive, we will cover our revenue build and valuation, as well as share what we consider risks.
In Part 1, we covered Coupang’s background history, explain their business, give context on the market they operate in, and explore competitive dynamics.
In Part 2, we covered Coupang’s Flywheel and their many other businesses and initiatives.
If you want to get access to the full report now, along with research reports (and downloadable excel worksheets) on Tencent, Alibaba, JD, Nintendo, and Grab, among many others, and access to our private members-only Discord, subscribe now for as little as $28/month!
For our revenue build we took a more top-down approach assuming market share, but buttressed that by extrapolating out the implications on a spend per buyer basis. Our hope is that by keeping both frameworks in mind, we can approach a reasonable estimate that doesn’t implicitly assume something that is impossible (which is common if you are too high level or too granular—frankly, we think in our SE piece we tried to force precision where it wasn’t helpful and that led to a weaker estimate). We will spend most of our time valuing their core ecommerce business, but conclude with a Coupang Eats sketch and some commentary on their new markets.
The exhibit below takes our TAM estimates from the Market Background section in Part 1. We apply a 40% market share assumption to GMV, which is informed by global peers’ market share. Amazon is around 40-50% depending on what you include in the TAM and Alibaba used to be 60%+. Given all of the competitive factors we discussed above, we think they will continue to gain share and take a disproportionate amount of ecommerce growth. In a decade, we think that a 40% market share is reasonable, driven by new buyers and spend per buyer growing. This gets a ~$120bn GMV figure for 2030 (includes grocery) which implies a 17% growth CAGR. With an ultimate 50/50 split between 1P and 3P and applying our 3P take-rate estimate (explained below), we get a revenue figure of ~$80bn or a 18% growth CAGR. However, given the 1P vs 3P distortion and various different 3P margin services, it is not a very informative figure. (We have seen some investors look at GMV or revenue multiples vs peers which is a nonsense comparison as all of these streams of revenue have wildly different profit profiles even if they are all “ecommerce”.)
In the exhibit below, we take our above GMV estimates and apply them on a per capita and household basis. We assume Coupang buyers will eventually reach 60% of the population (About in-line with Amazon in the US, but Alibaba is higher if you look at buyers as % of China’s internet-enabled consumers). This implies there is ~1.25x Coupang buyers per household, which is not at all unreasonable if you look at how many users Alibaba or Pinduoduo has in China per household, which is over 2x. This shows that in order for Coupang to reach our $120bn GMV figure, we need spend per buyer to increase from ~$1,500 to $3,800, or at about a 10% CAGR. Prima facie this seems fair, but it is very hard to set a confidence interval on this (at least for us). We then compared what that spend per Coupang buyer is when compared to our total average online shopping per buyer. This showed that for those that do shop on Coupang, they will eventually spend 2/3rds of all of their online spend on Coupang. This number does feel high, but if they can sell you almost anything you can buy–including groceries–and do so better than anyone else, then it seems plausible. A critical piece of achieving this would be Coupang increasing Rocket Wow members. Rocket Wow members, similar to Amazon Prime members, are much more likely to shift virtually all of the spend they can to the online platform. Amazon Prime members spend ~4x as much as nonprime members at roughly ~$2k annually (remember that online spend in Korea is higher than the US).
As we showed earlier, we estimate blended take-rate around ~11% and that increasing through higher adoption of seller services will be an important driver to revenue growth.
Our estimated take-rate is based off the following math: First, we apply an average 10% seller fee to every 3P sale. (We do not believe they will raise seller fees in the future.) Then, we assume sellers spend an average of 20% of their revenue on Coupang advertising with 50% adoption. Informing our 20% figure is various conversations with merchants, but we acknowledge our sample size was not large enough to have full confidence in it. However, as we noted in our Alibaba report:
Amazon is estimated to have advertising revenues of ~5% of GMV, with 1P estimated to be anywhere from 45-60% of that. For conservatism, assume 3P is 55%. That implies that their ad revenues are 9% of GMV in addition (!) to seller fees.
That means Amazon is today where we roughly see Coupang being in 10 years, which seems conservative and could be a source of upside, so net-net we are okay with our assumptions. With ROAS (return on ad spend) of 400-500% today, we are confident there is a long runway for advertising growth. Together, that gets us a blended take-rate of 20% before fulfillment.
The second piece is FBC, or Fulfillment by Coupang. The merchants we spoke with that used FBC (it is a newer initiative) note their fulfillment costs run ~30% of the items’ costs, but there are fixed costs and volume-based factors that conflate applying that average across their whole 3P base. We looked to Amazon, who disclosed fulfillment costs, but since it includes a few other things like payment processing and they consolidate sales of other segments like hardware and Prime, we cannot get a totally clean number (we can only back out AWS). However, we can still get a rough sense if we are off base. Using that fulfillment figure and adjusting revenue as best we can (revenues are split between 1P/3P about the same as Coupang), we get a fulfillment cost as a % of revenue of ~16-17%. If you back out revenues for their other non-retail segments and assume they make some profit of their fulfillment services, then an average of ~20% seems fair. With that, we will proceed to taking our ecommerce revenue build and estimate a cost structure.
Below, we take our revenue estimates and layer them in with gross margin assumptions and Rocket Wow revenues. We assume the 1P business can be similarly profitable to other large retailers and the 3P business will have higher margins, in-line with other global marketplace businesses (See our SE piece for more discussion on 3P marketplace gross margins). Dragging down the high margin nature of the 3P business is 3P fulfillment services, which are much lower margin due to the capital-intensive nature of that businesses. We assume a 20% “gross” margin on fulfillment based on global logistics companies with the caveat that logistics companies typically only report operating items and no cost of revenues. Starting with operating margins, which are high single digit to ~10% for these logistics companies, we estimated overhead and other non-directly revenue-generating expenses to be ~10% as well, which gets us to a 20% gross margin for fulfillment.
Coupang only reports one operating expense line item, so we have no idea how costs are allocated. The exhibit below uses just one catch-all operating expense bucket. In a decade, we see revenue growth of 18% per annum coupled with gross margin expansion leading to gross profits compounded at ~25%, while opex increases at just 16% CAGR, providing enough operating leverage to generate a significant profit of ~$5bn.
The valuation sketch below sensitizes around NOPAT for operating margins between 5-15%. Achieving $140bn in GMV and an 8% EBIT margin, at 15-25x implies a 7-9% annual return. There are few things to keep in mind at this point (there are other risks, but these are some focal ones): 1) our GMV assumptions assume Coupang becomes a clear market leader domestically and all customers who use Coupang use it almost exclusively (2/3rds of all online spend for Coupang users), 2) 1P gross margins are able to continue to expand and they do not need to rebate the margin expansion back to the consumer in the form of lower prices to keep the flywheel going, 3) 3P advertising services–$6bn of revenue and $4bn of gross profits—do not hit a ROAS wall anytime soon meaning adoption and spend will be much lower, and 4) fulfillment services–$1.6bn gross profits–are indeed eventually a profit center and not priced to just meet costs or money losing to spur increased 3P adoption.
Generally speaking, all of these comments suggest that there is a tradeoff between how profitable they ultimately become and what their developed consumer usage looks like. The more Coupang gives back to the consumer, the larger the consumer surplus, the higher we can expect adoption to be. The more profits Coupang takes, the weaker the flywheel will become than otherwise. This is a natural trade-off where Amazon literally for decades opted to just speed the flywheel up as much as possible by reinvesting and not taking margin (remember: “your margin is my opportunity”). This is all to say that perhaps profits will be lower in the future, but you will have a higher quality business. However, at some point, no amount of business quality can make up for forgone profits. It is hard to know where to optimize on this scale, but we have confidence they will figure it out.
The valuation above is only considering Coupang’s core businesses. As we have mentioned, we do not have a lot of info on their other initiatives as they are all very nascent, often barely launched. However, we take a stab at a high-level valuation of what their food delivery could become if it is successful. We still consider it a free call option, but below is what that option could be worth if things go particularly well for them.
We use a 2030 TAM estimate of $45bn for food delivery services and apply a range of market share figures. A 25-35% position is what a top 2 player could expect to garner, with 50%+ being a clear market leader, which is tough given their delayed start. At 20x, which assumes in 10 years it is growing low to mid-single digits, we get a valuation range of $5-12bn based on a 3-5% EBIT margin. The margin assumptions are informed by looking at other global players which are largely still unprofitable on a consolidated basis (some have middling profit on a contribution margin basis, which is before taking into account fixed costs). We actually think there are reasons to believe South Korea, with high population density, could be a better than average market. So perhaps their margin structure could be a bit better, which is especially true if they have a large market share.
Other notable call options includes 1) operations in new markets, 2) MyShops, 3) Coupay, 4) other new initiatives that have yet to be made public. See Part 2 for a refresher on what these businesses are.
While we have already touched on four main risks, we will touch on a few more below:
- Key-man risk. If Bom Kim left Coupang, we think that would be a negative for the company’s future. He has done a tremendous job of growing the company into something special. After IPOing he stepped down from CEO of the Korean operations, though he still is the CEO of the global operations. It sounds like this is so he can be in a bigger-picture, more strategic position and help guide their new endeavors as their core operations are in a strong position today. However, if Bom Kim ever stepped down entirely, we think Coupang would be worse off for it.
- Regulation. We know that historically Korea has been okay with large, market dominant firms so long as they are good partners to all stake-holders, including employees. In 2020, Coupang added more than 25,000 jobs, making them Korea’s #1 job creator. This is a clearly a positive, but it will be critical how the dialogue around treatment of their fulfillment and delivery workers turns in coming years. After a warehouse fire earlier this year renewed interest in the treatment of Coupang’s workers, there is more talk about whether Coupang is a good employer. There was also a story of a delivery man who had a heart attack while on the job delivering packages, which some news stories tried to skew as being a result of the demanding delivery schedule. However, this seems unfair to claim (Coupang responded that employees have statistically fewer heart attacks than the general Korean population). Coupang does already offer good pay and full-time benefits to all of their delivery drivers, which is better than what they get at competitor logistics companies. However, as they continue to grow in size, scrutiny will only continue to increase. More pro-labor regulations could either hurt margins or perhaps more harmfully, hurt their ability to relentlessly serve the customer by pushing employees. Additionally, merchants could also be a source of complaints as Coupang competes against them with 1P on their platform. There is nothing we know of today that concerns us here though.
- Ecommerce penetration could turn out to be lower than we expect and even if Coupang is a market leader, the TAM is much smaller than assumed and thus they could stop growing sooner than we anticipate.
- New ecommerce entrants with a more social commerce angle and gamification. While under such a scenario we still think Coupang’s value prop is strong, a new entrant that could capitalize on the areas Coupang is weak in could eat into the wallet share they can eventually demand from the buyer. We have seen Shopee have a lot of success with gamifying their offerings and Pinduoduo’s social aspect was also very popular. In short, this is another way of saying competition in Korea has seemed a bit stale and we have seen things work elsewhere globally that could be a threat to Coupang in Korea on margin.
- New initiatives are a larger source of cash than expected and Coupang needs to raise money or slow investment. Coupang doesn’t have an AWS like Amazon or Garena like SE, so their retail operations have to support both investment in the core infrastructure and new initiatives. Similarly, new initiatives could turn out to have a negative NPV.
Below is our summary model for CPNG, designed to just illustrate how different items flow through to EPS. Of course, Members Plus have access to all of our exhibits including the below model and can change any assumption they wish.
We hope you enjoyed the final Part 3 of our Coupang report! As a reminder, you can read Part 1 here and Part 2 here.
If you want to get access to the full report now, along with research reports (and downloadable excel worksheets) on Tencent, Alibaba, JD, Nintendo, and Grab, among many others, and access to our private members-only Discord, subscribe now!
Not quite ready to subscribe? Drop your email below to be updated when we publish our next piece!
We hoped you enjoyed and thanks for reading! Disagree with anything we said or have questions? Please feel free to kick off any discussions in our members-only discord in the Coupang channel!
*Disclaimer: People who have helped craft this analysis may individually (or through funds they work at) have a position in the company. Absolutely nothing in this report is investment advice nor should it be construed as such. We make no claims to the veracity of all facts and figures presented in this report. Certain figures could be stated erroneously and certain analysis could be incorrect. Please see the full disclaimer linked above.
DJY Research, its author(s), any of its contributors, and any funds they may advise at may own securities in the companies, or related companies, written about on this website.
All rights reserved. No content on the DJY Research website can be redistributed without the express written consent of DJY Research.