*Note that we refer to some of our other pieces that are behind the Members paywall throughout the piece. We do this for two reasons 1) brevity, 2) but also because we use our free pieces as an acquisition tool to convince readers to sign up for our $29 subscription (sneaky, right?) which entitles you to past reports, a new report every month, and access to the DJY community Discord! We are currently reviewing our pricing, but we will not raise prices on any existing subscribers.
Nevertheless, if you aren’t ready to join us today, we hope you still enjoy this SE update and feel free to share your thoughts with us on Twitter!
We will be periodically releasing company updates after we have put out a deep dive on a company. But rather than comment on quarterly metrics, we will try to focus more on how the results have changed our thesis (if at all). This second update comes just half a year after our original piece. You can read our full report here, and you can read the first update (re: management) here. We will assume readers are familiar with the company and our prior pieces in the updates.
We’ll also be releasing an update on Sea’s accelerating activities in LATAM and their budding ambitions in Europe. If you know any merchants who operate in either geography, or are a consumer living there, please reach out via email or Twitter, as we’d love to hear from you!
Company Update.
Since we last wrote on SE, they have continued to execute superbly with GMV now run-rating $60bn (from ~$45bn), Sea Money users increasing to 33mn (from 23mn), gaming users increasing to 725mn (+115mn), and paid penetration improving to 12.7%. On the back of strong results and a healthy stock price, SE raised $6.3bn ($3.8 equity and $2.5bn convertibles), giving them an ample war chest to continue to press their leads in ecommerce and experiment with their other new initiatives.
In this update, we will go through each of their segments as well as revise our thesis relative to our original piece.
Digital Entertainment.
A few key takeaways:
- Free Fire continues to dominate globally with a new peak record of 150mn DAUs and 725mn QAUs with 93mn paid users
- Free Fire maintained the leading position of highest grossing mobile app in SEA and Latin America for 8 consecutive quarters and for 3 in India
- Free Fire World Series was a hit with 5.2mn concurrent online users
They also rolled out “Pet Rumble”, pictured above, which is a mini-game within Free Fire. While this may seem benign, it is interesting for two reasons. 1) It shows how they are trying to expand out the Free Fire ecosystem into other gameplay. Unlike the core free-for-all PUBG/Fortnite-inspired battle royales, Pet Rumble is an “Among Us”-inspired mini-game within Free Fire where players play as their Free Fire pet (right now it is only playable with friends). This shows them inching in the direction of making Free Fire a platform with many different gaming experiences and a strong social component. Social elements are important as it makes game play much stickier. Remember, the original Garena was originally a sort of social connective tissue that served as a launching pad for other games on PC, so it is interesting to see them backing into a similar vision with Free Fire. 2) If you recall, and despite management referring to it as “self-developed”, Free Fire was actually acquired in 2017 from 111 Dots Studios when it was in beta. Additionally, Free Fire recently rolled out a collaboration with worldwide gaming phenomenon Street Fighter, where Free Fire players could play as Street Fighter characters. As such, it remains to be seen whether they have the ability to create convincing gaming experiences from scratch, but seeing well received innovation and IP partnerships—albeit even incremental—gives us confidence that they will be able to keep Free Fire fresh. We still are more restrained on their ability to develop other successful games, but they don’t need to as long as Free Fire remains relevant.
Overall, they are doing extremely well here, run-rating over $4bn of revenues with over 70% gross margins, generating cash they can plow back into their other businesses. The uptick in increased paid penetration from 11.9% at the end of 2020 to 12.7% is also encouraging. While it may seem like a minor improvement, 80bps is another ~6mn users that generate an annualized ~$250mn!
Another important thing to note is that ARPPU has inflected back up after 4 years of trending down. So not only are they getting more users to pay on Free Fire, but they are getting users who pay to spend more too. This is not something we would have expected to see as usually the most fervent users (the whales) are first to spend prodigiously and later cohorts of users tend to be less enthusiastic. The behavior we are seeing suggests that SE is succeeding in pushing more users down the funnel and converting them to loyal paid users. This is important as further growth will increasingly rely on better monetization of their current user base, rather than from adding new users. Sea’s management clearly understands this, as they’ve hired 1,000+ in-house skin developers (!) that are focused on just that.
Digital Financial Services.
A few key takeaways:
- Total payment volume exceeded $4.1bn in 2Q21 vs $7.8bn for all of 2020.
- Users increased to 33mn from 23mn at the end of 2020
Comparatively, there isn’t a lot new to report for Sea Money. While they continue to increase adoption, they still are not the clear leader with tough competition in fragmented markets that vary by country. SE disclosed that according to a “Snapcart” survey, ShopeePay was the most used and most liked mobile wallet in Indonesia, which is one of their biggest markets. While it is hard to know how significant that is without knowing more about the survey methods, concretely we saw that partnerships for off-platform usage have been increasing. Indonesians can now use ShopeePay as a payment option at Indomaret (top convenience store chain) as well as other chains like Wendy’s and Domino’s. They’ve also partnered with MasterCard to allow users to pay at the ~200k vendors that accept Mastercard Contactless in Thailand.
Off-platform usage adoption is important for them to build an ecosystem around ShopeePay because the more merchants they add, the better the adoption is for consumers, which in turn drives more merchant acceptance. Merchants are happy to have an easy and frictionless way to receive payment from consumers, but they don’t want to have to accommodate a bunch of different payment options, so there will ultimately only be a handful of “winners”. SE is trying to tie Sea Money to Shopee to increase uptake, which should work over time, but today if our buyer assumptions are directionally correct (details below), we don’t think even half of Shopee users use Sea Money today. However, if Shopee continues to maintain its leading ecommerce position, then they will have multiple opportunities to push users to Sea Money with loyalty programs.
A new feature called “Deals Near Me” or “Malls Around You Promo” shows the user deals in their vicinity. Users can buy the deal vouchers and redeem them immediately at the outlet. Management has claimed this has been very successful to drive foot traffic and is a neat value proposition to drive further merchant adoption while offering consumers something exclusive to the Sea Money platform.
Ecommerce.
A few key takeaways:
- GMV reached $15bn for 2Q21, +88% y/y
- Adj. EBITDA loss per order decreased ~20% y/y to $0.41
- A second market, Malaysia, reached EBITDA profitability (before corp. overhead), joining Taiwan
- Shopee was the most downloaded app on the Google Play store in the shopping category
- Shopee app ranked #1 by MAUs in Indonesia and Taiwan in the shopping category
This is where we have the most changed in our thesis and our numbers have moved materially higher. Since we are refreshing out view on the recent business performance, we will weave in the business updates with our revised thoughts. For more context, please refer to our original piece (which is free to access).
Our 5-year out SE estimates have in just 2 quarters proven to seem too conservative. Originally, we estimated T+5 GMV of $70bn, but they are now run-rating $60bn, 50%+ higher than the $40bn in our first piece. Frankly, we believe we fell prey to some degree of false precision, but also didn’t fully appreciate the extent to which Shopee would grow the ecommerce market. However, this is also a tricky piece to write because we want to delineate between things that were “wrong in process” vs “outcomes that surprised”, which could not have reliably been assumed a priori. To the extent stock prices follow fundamentals on a short-term basis, SE’s run up in the past half year suggests that they have positively surprised. To be clear, we are not just trying to dismiss their business outperformance as “unpredictable” or ask you to take us off the hook for our mistakes.
There were two main factors that informed our original estimate that shed some light on just how much Shopee has been a share-gainer and beneficiary of the Covid-led ecommerce adoption.
First Factor: Assumed traditional order seasonality. The chart below shows how Shopee usually slightly contracts on orders in from the 4th quarter to the 1st quarter. But instead of contracting this last year, they grew beyond the seasonal headwind. Not that we want to be quarter callers, but that small factor was a ~$8bn swing factor in the GMV base that compounded to $14bn. We expected traditional seasonality and also believed there was a Covid tailwind that would fade, making us more conservative in our estimate. Bigger picture, we should have simply noted that GMV has never contracted (as you’ll see below in a later graph), but we were reluctant to over extrapolate out a very idiosyncratic environment. Even with the power of hindsight, we are not sure it was wrong to err on the side of conservatism, but we think we were too conservative. Our mistake.
Second Factor: overfixating on an existing TAM in a nascent market. Our subscribers who read our recent Coupang piece may now be thinking that we just stepped into this trap again, but there is a difference in leaning on a TAM in a mature market vs a growing one. Shopee is many users’ first ecommerce experience, but nevertheless both could be growing the market in their own way. In Coupang’s case, a superior experience with <1 day shipping and frictionless purchasing from a trusted platform could lead to more ecommerce buying that would have been completed elsewhere. In Shopee’s case, simply introducing more people to ecommerce and slowly letting them build trust in it will increase use case over time. However, going from Korea’s 35% ecommerce penetration to even as much as 60% in the future isn’t as impactful as SEA moving from ~5% to 15%, even if the dollars are much less. This means TAM estimates are more volatile for younger markets, so less weight should be placed in them. In our second Coupang (first was TAM-based) ecommerce build we focused more on spend per buyer, which we have come to believe is a better metric (or at least you should look at both).
It is also notable that Shopee’s $60bn run rate is almost the same size as the entire SEA Ecommerce market was estimated to be just ~1 year ago. In our updated build we shifted the focus from top-down TAM to spend per buyer. However, this comes with its own challenges. GDP per capita in Indonesia, one of the richer SEA countries is still <$5k vs Shopee’s estimated spend per buyer of $771 (we will touch more on how we calculated this figure below, but it is a fairly high confidence estimate as they provided order frequency, AOV, and GMV). This led us to worry that they would hit a wall on user growth as the amount of buyers with discretionary income will become the limiting factor.
This concern was evident in one of our questions to IR:
We are backing into how many SE active buyers there are based on the disclosures and found that there are approximately ~55mn buyers, which is a tiny portion of the S.E.A. internet-using population and an even smaller portion of the total population. However, we were wondering if the immediately addressable TAM is much smaller than it appears. For instance, we know that Indonesia’s GDP per capita is less than $5k USD, but at the current monthly order rate it seems each buyer is spending ~$800 annually or about 16% of GDP per Capita. This seems large to us, especially in the context of Amazon Prime members spending only twice that (~$1,600) despite GDP per capita being ~13x higher. Is it only the wealthier people in the SEA region who can afford to use Shopee at this spend level or is this relative high spend per buyer explained by the retail environment in S.E.A. lacking in many offerings that only e-commerce can provide (so even less wealthy people will do most of their buying on e-commerce)? In short, are future buyers going to be less valuable than pre-existing ones because they will not be able to spend as much (frequency and order value keep going down)?
We didn’t get much in terms of clarity from them, but in hindsight they could have simply answered that SEA retail sales are estimated to be around $1 trillion today and as they add more categories they can expect to take a larger portion of that. It seems that most of our folly comes from trying too hard to think smart.
Nevertheless, it is still a legitimate concern that future buyers will be worth less. They could get hit with the double whammy of 1) buyer growth stalling as the potential buyer population is more saturated than it appears and 2) buyer spend stalls as those new users who are added tend to be lower income. Having said that, we aren’t going to base decisions on what is possible and instead focus on what we do know. (Separately, what we see in Brazil should also make you skeptical of this—we will explore this in a later, separate update). Ecommerce is a far better buying experience vs in-store shopping for most people, offering superior selection, price visibility, and convenience. SEA is a growing market with average incomes and GDP per capita trending up, and as people get wealthier, they will spend more money in channels that provide the best experience. Not to dismiss the competition, but today that is largely Shopee. (See our original piece on Sea for more on competition. Things are generally the same, but with Shopee having pulled out a larger lead and us more confident that they will stay there.) Shopee has grown GMV every quarter and Covid seems to have reinvigorated ecommerce growth. Y/Y growth was dropping every quarter from >200% in 4Q17 to 65% in 4Q19, largely as a byproduct of growing off of a larger base. But incredibly, Covid didn’t just inflect growth positively temporarily, but even a year and half later, SE is growing faster on a larger base. Last quarter, 2Q21, Shopee grew 88% y/y from an $8bn base. (That is a +2300bps higher growth rate on a ~2.5x higher GMV base!)
Along those lines, we see AOVs (average order value or average dollars spent on each order) falling consistently almost every year. In the past two years, AOV has dropped 40%. This can be interpreted as more people feeling less of a need to “group” items in an order to meet discount shipping requirements, but also could be evident of success in new categories like FMCG (fast moving consumer goods) which tend to be low ASPs. The AOV trend isn’t concerning (from a GMV perspective at least, it is somewhat for profitability) so long as they can offset it with order frequency, which they so far seem to be doing. In the most recent earnings call, they noted that average monthly frequency is now at 6x, up from the 3.7x disclosed in a January 2018 earnings call.
However, it is also the case that a falling AOV means much worse unit economics. On 3P sales, Shopee doesn’t pay the shipping but will often subsidize the merchant and the drop in AOVs likely has led to worse economics per package. Additionally, as they move more into 1P, they will be responsible for bearing the shipping costs. For now, they seem satisfied with a partner network, but their cost savings from volume will be limited. The trade-off between volumes and margins is nothing exclusive to Shopee and if they can make up the smaller per unit profit on more sales velocity, they will be fine. Indeed, Shopee is arguably in a better competitive position if they can dominate the low value commerce segment as these items tend to be high-frequency so consumers build habits and low margin so a competitor has to eat more losses to compete. Unit economics aside, it is encouraging to see frequency continue to increase. Where they are still susceptible though, and lagging global peers, is in their shipping times.
As seen above, Shopee estimates delivery times of ~5+ days, which is far behind other large ecommerce players who are closing in on 1 day. In our Coupang piece, we go to lengths to describe different shipping network strategies (owned vs partner), which can help inform how Shopee could develop over time. For now, they seem to be mostly leaning on 3PL with an increased emphasis on spinning up their own logistics, evidenced in part by closing a recent deal for their own Shopee branded trucks.
It seems they are doing a mix of what Alibaba did with Cainiao (logistics platform for 3PL partners; see our Alibaba report for more info) and what Amazon is doing by seeding their logistics network with their own delivery assets. Their ultimate plan isn’t clear and they don’t seem too keen on disclosing anything, but they have recently eliminated the ability for consumers to choose different delivery providers and instead just pick between standard, economy, or other. On the merchant side, they have been encouraging sellers to utilize Shopee Xpress for better prices and to reduce the risk of late shipments. It seems that Shopee is basically trying to consolidate all deliveries and have them go through their own logistics platform, whereby they can then better pick the best rates and optimize delivery times, filling any gaps with their own trucks. (To clarify, it is not clear to us if the trucks are owned and operated by SE or if it is a similar program Amazon Logistics, whereby they support 3PL operators by guaranteeing volume and arranging financing for the delivery assets). As Shopee moves merchants to just pick “Shopee delivery” and simultaneously takes away the consumers’ ability to choose, they have effectively supplanted the businesses and consumers relationship with the logistics provider. This will give Shopee more of an ability to swap out providers and play them against themselves for the best rates. While they do that, they are making Shopee delivery more robust with their own delivery assets (whether they are owned by Shopee is a different question than if they are exclusive to Shopee). Shopee supplanting the two-sided network with their own vehicles reminds us of the AMAP (Alibaba’s map app; see Alibaba report) discussion we had in our Didi piece whereby 3rd party map aggregation apps were supplanting the direct consumer relationship, and presented a risk for them to seed their own vehicles into the network (specifically in regards to AV in the piece). At the end of the day, Shopee owns the demand so they can satisfy it however they see best for their customers.
The natural extension of a delivery service is fulfillment. Fulfillment is one of the greatest competitive advantages an ecommerce platform can enjoy since it creates 1) exclusive, merchant-financed inventory, 2) increases trustworthiness of a transaction, and 3) greatly decreases delivery time. (Subscribers can refer to our Coupang and Alibaba reports for a more in-depth discussion of ecommerce competitive factors). Shopee is still in its early innings here, but it is interesting to see that they are buying trailer trucks, which are typically used for delivery from manufacturer/warehouse to retailer (cross-docking) and seldom for last mile delivery. We believe this suggests they are building out their own warehouses and potentially merchant package drop-off docks so that a seller could drop-off packages or inventory (for fulfillment) at one convenient location and Shopee could send it to a more centralized warehouse where it would be better positioned to be delivered quickly. This is similar to what Mercado Libre does in Latin America.
Shopee bought an undisclosed amount of the Isuzu N-series trucks pictured above. Below we show a side view of the truck. The detachable trailer means it is unlikely to be utilized for last mile delivery and suggest more B2B usage. This could be the beginnings of Shopee spinning up warehouses and leaning on merchants to adopt fulfillment by consolidating inventory at docks to centralize inventory for faster delivery.
Enabling better shipping will be critical to continued growth as many customers get turned off by bad shipping experiences. Coupang CEO Bom Kim noted that shipping complaints made up half of their total complaints in their early days, which is why they decided to build their own logistics network. Today, the most concerning thing for SE is if they do not do this first and a competitor outflanks them here. In Korea, Coupang started in ecommerce almost a decade after early incumbents like eBay but was able to overtake them by providing a better experience with <1 day shipping and better selection. Shopee might be in the lead today, having outcompeted Tokopedia, Lazada, and others so far, but their lock on the consumer is still tenuous. Their desire to own their own logistics operation today isn’t clear, but in order to truly get unmatchable delivery times, they need unpacked and pre-sorted inventory in their own warehouses. This is why Coupang leaned heavily into 1P (explored in the Coupang piece) but Shopee cannot as easily mimic this strategy without directly competing against their 3P merchants. While the channel conflict is common to other platforms, the order of roll out is what is different—Amazon and Coupang first focused on 1P and later 3P. Doing the reverse could mean merchants churn while they still feel they have somewhat equivalent online selling options. Losing 3P merchants must be compensated by increasing 1P selection, but product procurement is not a muscle Shopee has built yet (they have noted that they only plan to roll out 1P to fill selection where it is lacking and don’t expect it to be more than a single digit of GMV). If this turns out to be the case, 3rd party merchant adoption of fulfillment is essential to get to quick shipping (this is what Mercado Libre is doing). In our build below we do not assume they become a big player in fulfillment.
*(Members Plus members will be able to download the excel file and add to or adjust our assumptions.)
Ecommerce Build Update.
Despite the incredible sales velocity Shopee has experienced, extrapolating these growth rates too far into the future is, in our opinion, still a mistake. Nevertheless, our 10 year out GMV estimates have almost doubled from ~$140bn to ~$260bn. We want to be clear that these estimates will no doubt prove to be incorrect and that you should think about them as a distribution with the mean figure shown below. Additionally, these estimates are far from conservative—this assumes Amazon-like execution with revenues growing 30%+, something Amazon was only able to accomplish by opening up many new markets and becoming the undisputed leader in almost all of them (at least all of the high income ones). We also are not explicitly modeling fulfillment, but it is likely they will need to lean on this to achieve the level of ecommerce domination the numbers below imply.
Our old build is reprinted below:
Limiting SE’s share of ecommerce to 40% was problematic because it assumed the online retail TAM was right. We don’t think it is likely that Sea will achieve more than a 40% share in the long-term (could go higher for a while like Alibaba’s), but we also don’t know the actual size of that TAM (either way, GMV will be higher). The other thing we underappreciated was how low-income consumers could still be avid adopters of ecommerce. As evidenced by our questions to IR, we were worried that they already had a very high share of consumers with discretionary income. However, after doing our piece on Alibaba, we developed a new appreciation for how Pinduoduo grew the market by focusing on low AOVs and lower-income buyers who were historically neglected. Having studied this, we have more confidence now that lower income users will adopt ecommerce all the same. (In hindsight, this sounds like a silly thing to get hung up on, but we do often err on the side of conservatism). With this in mind, we increased buyers ~100mn from 215mn to 330mn, which implies 45% of the SEA population uses Shopee (for reference, Coupang, who is the clear leader in Korea, has about 30% of the population as users today, so this is an aggressive assumption that assumes near flawless execution and positive market developments).
If you flow the updated revenue numbers through the same P&L assumptions in the original piece, you get an $11bn NOPAT figure, up from the original $6.5bn. You can argue that they should have more operating leverage, but remember that we are not building in fulfillment, which would be margin dilutive.
In the interest of avoiding false precision, we expanded the margin sensitivity from 15% to 30%, which gets us an annual TSR range of 1-13%. Their multiple will be dictated in large part by their growth runway exiting T+10, which if at >20%+ could justify 35x, but if only low single digits, 20x is probably fairer.
Below we give some color on how we think about these return figures above, which are annualized total stock returns assuming our assumptions hold. Whether an investor finds these returns attractive will largely be predicated on three factors: 1) the investor’s confidence in the range of outcomes and whether they believe the most likely band of outcomes is tighter, 2) whether they are an absolute or relative investor, 3) the degree to which an investor believes SE will create value in ways we are not accounting for today (this can include all of the “call options” we mentioned in our original piece on nascent businesses like Latin America, Shopee Food, and other businesses that may not even exist today). This can be thought of as a “quality” factor.
To be clear, we are under no illusion that our estimates are “correct”, but rather our aim is to make reasonable assumptions that we think can help an investor make their own judgement on the opportunity. As mentioned, you should think of these estimates as existing on a probabilistic distribution. Of course, it is possible for Shopee to do far better and add 300mn new buyers instead of 100mn for instance, or totally flounder the logistics execution for a competitor like Coupang to make significant inroads (Coupang is rolling out select international operations now via “quick commerce”; see our full report for more on this and explore if this will be the Trojan Horse that lets Coupang penetrate SEA). There is a significant qualitative aspect to any figure we have, which is partially why our reports are so long. We would go so far as to say the qualitative aspects are by far more important than the estimates when you are dealing with companies whose values are highly predicated on future success. So while much of this piece has been rehashing numbers, do not forget that they are only one piece of the overall equation.
Truly great companies tend to outperform expectations, including yours. However, this doesn’t mean that you should expect things that you don’t currently expect. Doing this creates risk, as an investor is no longer guided by their own sense of judgement, but rather relying on good fortune to rationalize a valuation. When you lose your ability to tie a valuation to various plausible scenarios that you are okay paying for, you are no longer investing. (It is a bit of paradox: if you believe a company is great than they have to do more than you expect them to, but if you expect that, then they have to beat the expectation that you expect them to beat expectations. Now they have to beat the expectations of inventors who are expecting them to beat expectations and so on… Occasionally this feedback loop does happen where growth companies’ estimates move far beyond plausibility and reset after a quarterly report, usually alongside a brutal stock price reaction).
To clarify, it is worth “something” that great companies tend to outperform, but there is a difference between paying a premium for quality and willingness to pay anything because you blithely assume that after whatever scenario you can come up with they will do better. Such logic can be heard when people claim they will “figure it out”, “always beat by X%”, or perhaps most commonly not attempting to tie a valuation to cash flow metrics. This is probably the first time you’ve seen someone give a lecture and a “Mea Culpa” in the same write-up… but not doing any math isn’t a solution to fear of the math being wrong! The math will be wrong, but it can still be a better guiding post than nothing! Many early maps may have taken explorers to the wrong destinations, but there are also many incorrect maps that are good enough to navigate on (knowing the shape of the coastline isn’t essential to getting to the continent): the trick is to make sure that you get the big picture things right, and that the things you get wrong don’t matter. If you are not sure if your map is of high enough quality, the solution is simple, don’t take the trip. At DJY, we try to provide the best quality “maps” we can, but we are fallible to mistakes all the same too.
Conclusion.
Despite the length of this update, we still haven’t covered a lot of aspects! If you want us to cover ShopeeFood, Latin America, Poland, and other aspects, please share this piece so we know it is of value! Also make sure you add yourself to our free email list to be alerted when we release subsequent content.
Want more content now? Sign up to become a member and you will receive access to our full reports, which currently include Alibaba, Bilibili, Coupang, Didi, Nintendo, Sea, and a new one released every month! You will also get access to our discord to discuss our work (and others’) with fellow investors, operators, and researchers!
*Disclaimer: I have a position in Sea and 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.