Yves here. Get a cup of coffee. This post walks through some of the recent efforts to defend Uber against the evidence Hubert Horan presented in earlier posts in his series. One basic issue the loyalist fail to deal with is that the long history of transportation fleet management has shown that there is a tradeoff, in terms of capacity management (how big a fleet to have) between profitability and customer convenience (how much capacity to offer to allow customers more availability/more rapid response). Thus Uber’s massive subsidies are the reason customers may find that it offers more rapid “fills” of their orders for service than traditional cabs. Remove the subsidies, as Uber eventually must, and the fact that it is inherently a high cost producer, it will offer an inferior cost/speed of response tradeoff to other providers.
By Hubert Horan, who has 40 years of experience in the management and regulation of transportation companies (primarily airlines). Horan has no financial links with any urban car service industry competitors, investors or regulators, or any firms that work on behalf of industry participants
Uber’s narrative for its growth and expected industry dominance is devoid of any economic evidence
Uber has managed to get its own story about its eventual dominance and its unprecedented $69 billion valuation widely accepted despite a lack of supporting evidence. Uber depicts itself as similar to a handful of other blockbuster Silicon Valley startups: an incredibly powerful business model based on cutting edge technological innovation that would work in any city in any country. Like those other well-known startups, losses will soon give way to robust profits and the enormous power of this business model will inevitably lead to global industry dominance, and could possibly even lead to Uber significantly displacing private car ownership.
Unfortunately, nothing in the Uber narrative is based on objective evidence or actual industry economics. Neither Uber nor anyone else has provided any verifiable evidence that Uber’s business model can even achieve sustainable profits. Similarly, there is no proof of massive efficiency advantages that would be necessary to achieve inevitable industry dominance in the near term or major reductions in private car ownership in the longer term.
In fact, the spectacular failure of Uber China directly refutes the claim that Uber has a clear path to global dominance. Urban car services have none of the powerful scale economies that allowed other digitally-based tech companies to rapidly “grow into profitability.” No one has demonstrated that any of Uber’s claimed innovations have a material impact on efficiency or profitability. Uber users may like their fares and service levels, but no one can explain where Uber will suddenly find billions in efficiencies and scale economies that would be required to make its business self-supporting.
Uber has already imposed major unilateral cuts in the driver compensation levels it used to attract the drivers it needed to support its rapid growth; no one can explain why Uber wouldn’t similarly reduce passenger benefits as it approaches industry dominance. There is absolutely no evidence showing how a profitable Uber could sustainably provide cities with higher quality and lower cost transport service.
The Naked Capitalism series’ finding that Uber’s business model could not earn sustainable returns without exploiting anti-competitive market power was based on a wide range of evidence about the competitive economics of the urban car service industry
This series sought to establish an alternative to Uber’s narrative, using economic evidence to address the question of whether Uber’s growth and anticipated dominance will improve overall economic welfare. It has used traditional, well-accepted, widely used analytic techniques that a consultant or financial analyst would use to understand the market potential and/or competitive threat of a new entrant, or that an economist or an urban transport planner would use to consider whether an alternate taxi competitive/regulatory structure might improve service levels or industry efficiency.
If an Uber-dominated industry is to improve overall economic welfare, it would meet four tests
(1) the ability to earn sustainable profits in competitive markets large enough to provide returns on its investment base
(2) The ability to provide service at significantly lower cost than existing competitors, or the ability to produce much higher quality service at similar costs
(3) The creation of new sources of sustainable competitive advantages through major product redesigns and technology/process innovations that incumbent producers could not readily match
(4) Evidence that the newly dominant company will have strong incentive to pass on a significant share of those efficiency gains to consumers in the form of lower prices and/or increased service.
Since Uber’s investors have been explicitly pursuing total worldwide industry dominance, the evidence of profits, efficiency, competitive advantages and consumer welfare benefits would need to be overwhelming and obvious.
Part one of this series focused on the evidence that Uber was a fundamentally uneconomic enterprise (e.g.2015 operating margins of negative 140% and $2 billion in actual losses) and that all of Uber’s growth to date was due to billions in predatory investor subsidies that competitors could not match.
Part six presented new evidence confirming those findings, including reports that Uber’s 2016 losses were likely to exceed $3 billion, and data showing all margin improvement had been due to unilaterally imposed cuts in driver compensation. It also discussed Uber’s spectacular failure in China, which demonstrated that its business model could not be easily leveraged worldwide.
Part two gave data showing that Uber was the industry’s high cost producer and lacked the scale/network economies other startups have used to “grow into profitability”.
Part three showed how various claims that Uber’s growth had been driven by competitively powerful innovations were readily refuted by evidence of actual industry economics. As a result, there was no basis to expect Uber could ever earn sustainable profits in a competitive market.
Part four discussed how Uber’s investors had always seen quasi-monopoly industry dominance and the elimination of legal/regulatory barriers to the exploitation of anti-competitive market power as the key to earning returns on their $13 billion investment.
None of the findings rested on any single data point, and each of the conclusions was complemented and reinforced by evidence from the separate analytical approaches (financial results, cost structure, competitive advantage, industry dynamics). To encourage a discussion that could help improve the analysis part five of the series responded to many of the readers who had posed comments, questions and criticisms about precious posts. The finding that Uber would reduce economic welfare was based on the combination of its lack of efficiency advantage (so that its growth reflected an allocation of resources to less efficient uses) and the risk that a dominant Uber could exercise anti-competitive market power.
One objective of the Naked Capitalism series was to begin the process of refocusing the public discussion about Uber around economic evidence
As discussed in part three using the example of Amazon, the media coverage and public discussion of Uber has been markedly different from past startups.
Every major startup has had a PR story it wanted to tell to investors, consumers and competitors, but media coverage of past startups treated their reasonableness and accuracy as an open question. Amazon wanted to emphasize the power of its potential competitive advantages, but reporters sought input from independent experts in relevant fields (ecommerce, warehousing, distribution, traditional retailing) so that readers could better understand whether the claims were valid. Even as the power of Amazon’s many efficiency advantages became widely understood, the question of whether Amazon could ever achieve sustainable profits remained widely debated.
Press reports on Uber, by contrast, show no effort to investigate whether claims about competitive efficiencies and technological innovations make any sense. Independent sources with actual knowledge of the economics of taxis or urban transport are never quoted.
Similarly, leaked information showing Uber’s multi-billion dollar operating P&L losses first became available in mid-2015, but the overwhelming majority of media coverage completely ignored the question of whether the company will ever be profitable.
Journalists depict Uber’s unstoppable march to industry dominance as established truth, although the basis for their certainty about Uber’s overwhelming is never explained. Competitors like Lyft (with over $2 billion in funding) are treated as minor inconveniences, and every company that had served the industry prior to 2010 is treated as completely irrelevant. Even those who have been highly critical of specific Uber practices (false advertising, driver exploitation, journalist harassment, refusal to obey insurance and safety laws) typically assume that Uber’s story is true. Worse, they take the view that the costs and/or business model changes needed to fix these problems are not significant, and that Uber would retain its yet-to-be-demonstrated efficiency advantages.
Later in this series, we will discuss in detail how Uber was able to dominate media coverage and effectively block any serious scrutiny of its business model. In simple terms, pro-Uber media coverage falls into two major categories. The first category includes advocacy directly financed by Uber’s huge PR machine and/or provided by active Uber supporters with the help of Uber PR. Unlike any previous Silicon Valley funded startup, PR was a major strategic and spending priority for Uber since day one. So far, Uber has been able to overwhelm independent or dissenting views. These efforts included paying academics to publish major pro-Uber claims that do not hold up to even cursory scrutiny (including the Steven Levitt papers discussed in part six). Yet the press reports them as if they were independent, highly rigorous research. The active Uber supporters usually have political/ideological affinities with the company (i.e. favoring the elimination of any form of regulation including barriers to monopolization and the ability to exploit anti-competitive market power) and write pro-Uber papers to advance those objectives.
The second category is technology industry journalists and analysts. As in many other parts of the media, their ability to provide readers with critical insights may be compromised by things like their dependence on access to the senior executives of the prominent companies they cover. In its early startup days Amazon felt it had a powerful competitive story to tell and welcomed the business press.
By contrast, Uber went to great extremes to show it would ruthlessly attack any outsiders that dared question their official story. At a more basic level, Uber’s narrative was designed to tap into the technology writers’ predisposition that “tech” is a powerful force for good in society and a major driver of “progress” and economic growth. There are obviously many examples of where technology companies have increased overall economic welfare, but many more cases where they haven’t; you need to look at the economic facts in each situation.
But by wrapping itself in the mantles of “technological innovator” and “industry disrupter,” and emphasizing the contrast between Yellow Cab companies based out of garages in unsavory locations and a glamourous startup funded by the Silicon Valley’s best and brightest, Uber convinced most technology journalists that they didn’t need to talk to anyone familiar with those garages, or to investigate the sources and power of Uber’s claimed innovations and disruption.
Thus the biggest challenge here is not assembling and explaining the economic evidence, but to refocus the public discussion. There are technology journalists and analysts who aren’t on the Uber payroll and don’t see Uber as part of a broader ideological crusade. Can they respond to the type of economic analysis presented in the Naked Capitalism or has Uber’s narrative become too deeply ingrained?
Unfortunately, most of the technology and the mainstream business media have been valorizing Uber for seven years. Even when confronted with hard evidence of $3 billion annual operating losses, they may be reluctant to suggest that they might have gotten the story badly wrong.
Criticism of the Naked Capitalism analysis could be an entirely constructive response as long as the criticism was based on economic evidence, for example finding more detailed Uber P&L data, new evidence about Uber’s cost competitiveness, scale economies, or the efficiency impacts of specific business model features.
The two articles considered below illustrate the difficulty in truly shifting public discussion away from Uber’s narrative and towards constructive debate based on economic evidence. They represent “best-case” examples of reporters who have been covering tech industry issues for years.
While neither author had previously published anything remotely critical of Uber, both are free of direct Uber influence and are not focused on an external political/ideological agenda.
The first was by Timothy Lee in Vox. Over the years, Lee has also covered the technology industry for Forbes and the Washington Post. The second was by Ben Thompson in his tech industry newsletter Stratechery, which has been published since 2013. Both reports started by recognizing that the recent reporting about Uber’s continuing massive losses and the Naked Capitalism series represented major challenge to conventional wisdom about Uber. Both told their readers why they disagree with the major findings of the Naked Capitalism series, but provided links so their readers could examine the detailed arguments for themselves. But none of those disagreements were backed by any verifiable economic evidence that would materially affect any of the findings they disagreed with.
Thompson’s counter-arguments: the P&L evidence is incontrovertibly false, it is impossible to draw any conclusions about Uber’s cost competitiveness, and Uber can easily exploit scale economies to grow into profitability
Thompson’s most emphatic assertion pertains to the issue (that I discussed in both part one and six) of whether the leaked 2002-06 Uber Global P&L data fully incorporates all the operating revenue and expenses from Uber China.
First, Part 1 insisted that the financial figures excluded Uber China, which was incontrovertibly false. (Thompson’s emphasis)That’s a pretty big problem considering the scale of Uber’s losses in China! Not only did that error make the all of the numbers (very) wrong, it also raised serious questions about the analysis as a whole, particularly given the fact it remains uncorrected.
This assertion grossly misrepresents what I wrote. Uber has refused to explain what is included or not included in the leaked numbers (or any other aspect of its financial results). I did not claim that the exclusion of China was a proven fact, but laid out multiple data points supporting my interpretation of the financial results. For example, the quarterly EBITAR contributions, as reported by Bloomberg
|2016||Reported Uber Global EBITAR contribution||Uber Global shareholding in Uber China|
|2nd Q||($ 800) million||Entire quarter|
|3rd Q||($ 800) million||Half of quarter|
|4th Q||($ 800) million||None (sold in August)|
Uber has never published hard data about the losses of the separately organized Uber China operation, but multiple sources have reported claims that the losses were much worse than Uber experienced elsewhere, perhaps on the order of $1 billion. My assumption was that if the P&L results of an associated company with a different ownership structure making humongous losses had been fully incorporated in the parent’s operating P&L, those results would show sudden, dramatic improvements when the subsidiary was sold, probably on the order of a couple hundred million dollars per quarter in this case. You not only don’t see that in the 2016 numbers, but you don’t see any of the 2014-16 margin volatility you would have expected to see as Uber China rapidly grew. And incorporating the P&L results for Uber China would have been a major violation of GAAP accounting rules. Uber not only had no incentive to violate GAAP rules, but would have had ample reason to keep the huge losses and awful margins separate. Would this constitute “proof” in a court of law? Of course not, but it is a fair attempt to back up an argument using data and logic.
Does Thompson back up his alternate view with data or logic? Nope. He ignores my evidence, presents no data of his own, and then makes the totally indefensible claim that his alternate view is an incontrovertibly proven fact. Why does he think his view is incontrovertibly true and mine is incontrovertibly false? Because of a single tweet from the Bloomberg reporter, suggesting that the Uber numbers might include China. A claim that the reporter didn’t think was reliable enough to ever include in any of his Bloomberg articles.
No one has disputed the bottom-line operating P&L loss numbers (e.g. $3 billion in 2016). If Thompson believes that China losses are included, and distort attempts to evaluate whether Uber’s business model is working well elsewhere, what alternate number should we use? He doesn’t say.
The worst case guess would only reduce the 2016 losses to $2.5 billion (Uber China was sold in the middle of 2016). Thompson doesn’t explain why this would significantly change any of the other findings about Uber’s competitiveness.
How Uber did its accounting here is a simple factual question, and at some point we’ll have a clear answer. It would be one thing if Thompson said that his strong hunch was that I was wrong and the numbers did include Uber China but neither of us yet had “proof”. But Thompson is willfully ignoring the available evidence, insisting that his gut feel constitutes absolute truth, and telling his readers the contradictory evidence on this one point I’ve presented raises “serious questions” about the legitimacy of the entire Naked Capitalism series.
Thompson’s second line of attack is to claim that it is impossible to draw any meaningful conclusions about Uber’s competitive economics from aggregate P&L data.
Part 1 analyzed Uber’s profit and losses at the company level. While that is certainly an important data point — as should be obvious, in the long run all companies need to be profitable — it’s not particularly helpful when it comes to a company investing heavily in growth, particularly if they are investing in pursuit of an opportunity governed by not just network effects but also scale effects. What is required is an evaluation of unit costs: how much does Uber earn or lose on an individual ride? And, per the previous point, how did that number differ by city, level of market maturation, etc.?…. The issue with analyzing Uber is that we — including the author of these blog posts — simply don’t know what the company’s unit economics are; that the author attempted to make pronouncements about the company’s financial viability anyways was, in my estimation, an error just as significant as falsely claiming that the numbers did not include Uber China.
This objection is simply silly. Would it be nice to base an Uber analysis on detailed, carefully audited tables of unit cost data broken down by city, product and time of day? Well yes, but neither Uber or any other privately owned startup in history has ever provided that kind of data. Analysts routinely evaluate the growth potential of new products and companies where there’s less information available than there is here.
Thompson is also ignoring Uber’s actual strategy and narrative. Uber did not get $13 billion in funding in the hope that its business model might be able to make money in some kinds of markets under some kinds of conditions. Uber raised $13 billion because it told investors that it had an invincibly powerful business model based on cutting edge technology that would work in any city in any country, and would rapidly drive its inevitable march to global industry dominance. Thus the Uber Global P&L data I focus on are entirely appropriate. That Uber Global’s losses have been substantially greater than any venture capital funded startup in history—on both an absolute and margin basis–provides a powerful first clue that their business model might be terribly flawed.
Note that when Thompson doesn’t like my analysis showing Uber’s costs are uncompetitive, he insists that nobody should be allowed to draw conclusions unless supported by extremely detailed internal data that will never be made available, but when Thompson doesn’t like that my finding that Uber’s business model is fundamentally uneconomic is backed by both hard P&L data and industry cost analysis, he dismisses everything on the basis of a single tweet.
The third and most important area of disagreement is that although Thompson accepts my data analysis of the traditional industry cost structure he rejects my use of that same data to support the argument that Uber lacks the scale/network economies that would be needed to quickly achieve sustainable operating profits. My finding that Uber has no significant scale/network economies is central to my argument that Uber has been engaging in predatory competition in pursuit of monopoly. Unless one can demonstrate the growth economics that will quickly convert $3 billion annual losses to sustainable profits, Uber’s behavior cannot be interpreted any other way.
Powerful scale economies only exist when the cost structure has a huge fixed cost component, so that growth rapidly drives down unit costs. I provided the example of Amazon, which did have significant fixed costs (its entire IT marketing infrastructure, warehousing/distribution) and thus could expand to new geographic and product markets at very low marginal costs. I demonstrated that fixed costs were a tiny piece of urban car services total costs (85% are the totally variable costs of drivers, vehicles and fuel) and only a small portion of the 15% corporate costs are truly fixed. Uber cannot expand into new markets at very low cost since it faces unique driver recruitment, political lobbying and competitive marketing challenges in each city. I also pointed out that in the hundred years prior to Uber there had been absolutely no evidence of a tendency towards market concentration (either within or across cities) that would naturally occur if significant scale economies existed in this industry
Does Thompson’s criticism based on any evidence showing flaws in my analysis or showing where Uber had suddenly discovered significant scale economies? Not really. His criticism is entirely based on an argument that demonstrates that Thompson doesn’t understand the difference between short-term tactical marketing issues and the question of longer-term strategic/financial viability. He correctly points out that marketing managers in a company with powerful growth economics (like Amazon) would rationally price below total cost in order to accelerate those growth economics. But he completely ignores the question of whether Uber actually has Amazon-like growth economics, and if so, where do they come from and how strong are they?
Thompson concludes with several other assertions that ignore the evidence I’d provided about those questions and are not backed by any supporting counter-evidence whatsoever. In each case he’s doing nothing more than repeating portions of Uber’s narrative.
He insists Uber achieved important efficiencies from the “leverage to rework decades-old regulations that artificially limited taxi service to the benefit of incumbents.” Zero explanation beyond this single sentence. The economics of taxi regulations are somewhat involved and there’s never been a serious analysis showing large, clear-cut impacts in a single direction. “[T]here is far more demand for Uber than there ever was for traditional taxis. It is bizarre — and perhaps telling — that Horan flatly denies this market expansion in Part 3.” Like the claim of massive efficiencies from deregulation, Thompson imagines that Uber operates totally independently of the basic economic constraints every other company faces. Other companies cannot permanently expand the overall market unless the industry can (or will soon be able to) earn profits on all the added capacity, but Thompson thinks evidence of billions in losses could not possibly have any bearing on his claim the Uber has massively expanded demand. To claim that an industry that currently loses billions every year has permanently increased demand is akin to claiming that volume stimulated by a ruinous price war better is a better reflection of underlying demand than the lower volumes observed under stable, profitable market conditions.
Thompson is also ignoring Uber’s explicit goal of driving traditional operators out of business; it is trying to displace existing capacity, not to add to the capacity traditional operators already provide.
Likewise, you cannot say that Uber’s evasion of longstanding capacity regulations improved economic welfare without showing how a mature open entry industry can profitably provide more capacity and lower prices. Thompson’s claim that “Uber’s full potential is to replace personal car ownership” is similarly untethered to economic reality. This claim doesn’t just require demonstrating that Uber is massively more efficient than Yellow Cab; it requires showing that Uber can achieve such astounding levels of productive efficiency that it could drive taxi prices down to a tiny fraction of what they are today.
Lee’s counterarguments: Uber is much more efficient, Uber’s scale economies will drive profits, and Uber’s real strategy is driverless cars
In parts two and three of the Naked Capitalism series I lay out the reasons why Uber has higher costs than traditional operators, why nothing in its business model constitutes a powerful competitive advantage, and why it lacks the scale or network economies that digitally based companies could use to grow into profitability. Although Lee reasonably summarizes other parts of my argument, he simply ignores the evidence I presented on these points, asserts that Uber does have powerful competitive advantages and scale/network economies, but presents no actual economic evidence showing why he might be right and I might be wrong.
Lee correctly notes that I did not think that Uber’s investors were too stupid to understand industry costs, but (as discussed in detail in part four) that I thought those investors thought they could convert urban transport into a “winner-take-all” game and earn returns from exploiting monopoly power. But he rejects my “Uber returns require industry dominance” argument. “The other possibility — one that seems more likely to me — is that Uber really has figured out ways to make the taxicab market more efficient… that Uber’s model is superior to conventional taxicabs.” I had addressed those issues in detail in parts two and three; what points does Lee see that I might have missed? “The most obvious one is that smartphone hailing is inherently more customer-friendly than having to call an old-fashioned taxi dispatcher.” I had explicitly noted in part three that a lot of people seem to like Uber’s app, but there was no evidence that whatever marginal advantages it might have had any material impact on overall cost efficiency (dispatching accounts for roughly 7% of the cost of a traditional taxi operation) or driver revenue utilization. “The Uber app gives customers a realistic estimate of how long it will take for a car to pick them up. And once a customer hails a cab, it allows him to track a car’s progress.”
The software in Uber’s app can be readily copied (my local cab company in Phoenix has an app with the same functionality) so this can’t be a source of powerful, sustainable advantage. More importantly, there are dozens of other industries with ordering apps even more sophisticated and customer friendly than Uber’s, but none of them have any impact on market competition, and none helped create $68 billion in corporate value.
“And Uber’s pickup times have gotten even shorter as the company’s fleet has grown.” Here Lee is simply ignoring all of the data clearly showing that the prices and service quality Uber customers like is due to massive multi-billion dollar subsidies, not superior efficiency. Customers like the Uber app because it shows them multiple empty cabs nearby who can quickly pick them up. My Phoenix taxi app often shows no cars nearby or long wait times before pickup. That’s not because Uber’s software represents an efficiency breakthrough, it is because my fare in Phoenix needs to cover the entire cost of the cars and drivers on the street, while Uber subsidies make it possible to offer greater (but unsustainable) capacity.
Lee then claims that Uber somehow achieves much higher driver revenue productivity than traditional operators.
As Uber gets more customers, it reduces the average distance a driver must drive to reach a customer — and hence the amount of time they waste driving without getting paid….And this suggests one hole in Horan’s argument: If Uber can use driver time more efficiently, that really could lead to a sustainable cost advantage…. In economics jargon, this is known as a network effect. It’s the reason that eBay has a stranglehold over the collectibles market and Craigslist dominates online classified ads.
Needless to say Lee has no actual evidence of this huge productivity advantage, but he is at least trying to come up with a hypothesis that has some logical relationship to competitive efficiency.
The first problem here is that driver revenue productivity is fundamentally driven by major demand differences by time-of-day and geographic location, not the problem of driving twenty blocks to find a new customer when there was another customer only ten blocks away. A dispatcher at a traditional taxi company has the same information on the location of people who want cabs as Uber’s app, and as noted, traditional operators could adapt software with the same functionality as Uber’s if the additional automation would have a big impact on productivity.
But drivers who have just dropped off a passenger at the airport at 6:30 am (or anywhere at 8 am on a Sunday morning), or in a distant low density residential neighborhood, are likely to have a long wait before their next fare and there is absolutely nothing in Uber’s business model that solves this problem.
The second problem is that the customer perception of shorter waits for Uber cars actually means Uber has a driver productivity disadvantage. I wait longer for my Phoenix cabs than Uber customers because fewer of them are sitting around empty.
And the third problem is that this is neither a scale or network economy, and any appearance of “efficiency” is an artifact of the unsustainable excess capacity Uber has been subsidizing. There is absolutely nothing in the business model of Uber that produces Ebay/Craigslist type network economies.
As noted earlier, if urban car services had powerful growth economics here you would have seen strong tendencies towards market competition in cities dominated by dispatch cab services long before Uber. Lee is arguing that the dispatch company with the most cabs would gain a huge cost advantage due much better driver productivity; not a 1-2% type improvement, but as much as a 30% advantage. If so we would have lots of historical evidence of bigger companies capturing most of the traffic from smaller companies with higher costs and longer customer wait times. This never happened; taxi companies large and small needed to ensure a close balance between their driver/vehicle capacity and their revenue base. If the biggest company added capacity wait times would fall but profits would fall faster because the efficiencies Lee thinks are powerful don’t actually exist.
As with Thompson, Lee thinks that Uber can rapidly eliminate multi-billion dollar losses because he imagines they have the exact same scale economies as Amazon.
“Amazon’s unprofitability during the 1990s was an illusion created by Amazon’s aggressive investment in growth. Amazon was spending a lot of money on things like warehouses and new software that were going to take a few years to pay off….thanks to economies of scale in software and warehousing, Amazon’s costs per book kept going down as the company grew…[Uber is] just pursuing an even more ambitious version of Amazon’s strategy — that Uber’s core business is sound, but Uber is investing heavily in things like expanding to India and developing self-driving cars.”
Needless to say Lee provides absolutely no economic evidence here, and does not explain why my analysis showing that urban car service costs are overwhelmingly variable might be wrong, he just baldly asserts that if this worked for Amazon, it must fully explain what Uber is doing.
Amazon’s worst losses were $1.4 billion in its fifth year of operations, but shrank rapidly thereafter, while Uber losses have been steadily growing and will be over $3 billion in its seventh year. If Uber had huge scale economies, its profit margin would have been rapidly improving; in fact Uber had no margin improvement whatsoever between 2012 and 2015, and its 2016 improvement was totally driven by unilaterally imposed driver pay cuts, not growth economies.
As with Thompson, Lee does not seem to have a clear idea of what scale economies are, or how they work, makes no effort to explain how his imagined future scale economies could quickly achieve over $3 billion in savings, and does not seem to care that his argument is totally inconsistent with all of the available data.
Perhaps, Lee argues, none of this matters because the big payoff for Uber investors will come from driverless cars.
“The real prize in the ride-hailing wars, then, isn’t the conventional ride-hailing market itself. Rather, it’s the opportunity to become the customer’s gateway to the emerging self-driving car market. There are good reasons to think that most self-driving cars will be hailed on demand rather than owned by individual drivers. If that’s true, having the most popular ride-hailing app will be a huge strategic advantage in the self-driving car market.”
That’s Lee’s entire driverless car argument, and Thompson had made a similarly unelaborated and unsubstantiated assertion. There’s no explanation of how (or when) he thinks driverless technology and markets will develop, no explanation of what “winning” this yet-to-be-created market will actually involve, and no evidence as to why Uber is better positioned to earn billions in profits than the other major companies investing here.
Uber had raised $10 billion of its $13 billion cash base before it made driverless cars a strategic priority, so there is no basis for claiming its investors saw this as “the real prize.” Rationally, companies should not expand into riskier, more capital-intense markets until a strongly profitable position in its core market has been completely secured. At this point the main reason Uber and its supporters seem to be emphasizing them is to distract attention from Uber’s inability to make money in its core business.
Tech press coverage of Uber: ignore economics, keep repeating the Uber narrative
The focus here has been on the Thompson and Lee articles, in the interest of seeing whether they had raised any issues that the Naked Capitalism series had missed, or whether they had presented new economic evidence that could constructively contribute to the public discussion about Uber’s business model and its potential impact on economic welfare. They had not, but I want to reemphasize that Lee and Thompson remain “best case” examples of tech press coverage of Uber. There are scores of articles in the tech and mainstream media much, much worse than these, but these “best-case” examples help illustrate the power of Uber’s narrative.
Unlike many other reporters, they are free of direct Uber influence, make serious effort (as they have done here) to consider different viewpoints and I have always found their work to be worth reading. But neither is yet able to present analysis or engage in debate centered on economic evidence.
Their willingness to acknowledge that hard evidence about P&L results and industry economics contradicts the narrative is limited and grudging. Thompson is sure there’s evidence out there showing that Uber is actually profitable in most markets and that its costs aren’t actually higher than traditional competitors, even though he can’t produce any of it. Lee is sure there’s something in Uber’s business model that is powerfully efficient, but he can’t point to anything that would actually create major cost advantages or help explain $68 billion in corporate value. Both insist that since they consider Uber to be a “tech” company it must magically possess Amazon caliber scale economies and/or Ebay type network economies, but see no need to demonstrate that these magical economies actually exist.
A key issue here is that “tech” has morphed from one element of a larger business toolkit into a quasi-ideological/religious “movement” based on inherent virtues that the “non-tech” world lacks. As the broader Silicon Valley world has convinced itself that it is the avatar of progress and economic growth, and that “industry disruption” is virtuous regardless of actual impacts, most of the journalists and industry analysts embedded in that world have devolved from critical, objective thinkers to advocates and cheerleadiers for “our side.”
Uber’s narrative was designed to evoke these emotive/tribal loyalties, by portraying itself as the heroic defender of progress, innovation and disruption, engaged in mortal battle with the “evil taxi cartel” and the corrupt government regulators fighting to prevent the entrepreneurs who were using advanced technology to provide the world with better taxi service at lower prices. Despite hard evidence ($3 billion losses in year seven, massive failure of Uber China, no efficiency or scale driven margin improvements) many find it difficult to abandon the framing they bought into years ago.
Tom Slee (I would highly recommend his book “What’s Yours is Mine: Against The Sharing Economy”) has been following Thompson’s arguments and the overall Uber discussion over the years. When Brad DeLong posted links to the Naked Capitalism series, to Izabella Kaminska’s Financial Times article and to Thompson’s Stratechery piece at his website, Slee posted the following comment, which captures the ongoing refusal to reconsider the Uber narrative in light of growing economic evidence nicely:
It sounds like Ben Thompson is falling for the Uber bait and switch. Stages of which:
– Uber has a nice business as a status product (Uber Black Car ~ 2010)
– Uber Black may not be profitable, but Uber will displace taxis and be hugely profitable because of technology-driven efficiencies (UberX: 2014-2015)
– UberX may not be profitable, but UberPool will lead to new efficiencies in mass transit (2015-2016)
– UberX may not be profitable, but Uber is a logistics company and will rewrite the rules of delivery (UberEats, various speculative stories, 2013-2015)
– UberPool may not be profitable, but when Uber displaces car ownership the scale of the market will make it profitable (2016)
– Uber with drivers may not be profitable, but driverless cars will make Uber profitable (2014-)
– Driverless cars may not be profitable, but Uber is looking into flying vehicles (2016)
The Uber makes losses while maintaining credibility for bringing “the future” in some form or other.
 Uber executive Emil Michaels “suggested that the company should consider hiring a team of opposition researchers to dig up dirt on its critics in the media — and specifically to spread details of the personal life of a female journalist who has criticized the company.” Smith, Ben, Uber Executive Suggests Digging Up Dirt On Journalists, Buzzfeed, 17 Nov 2014. Lacy, Sarah, The moment I learned just how far Uber will go to silence journalists and attack women, Pando Daily, 17 Nov 2014.
 Lee, Timothy, Why Uber lost $2.2 billion in 9 months, Vox, 9 Jan 2017
 Thompson, Ben, Uber Losses (but China Gains?), Uber and unit Economics, Reconsidering Uber, Stratechery, 20 Dec 2016. Stratechery is a subscription-only tech industry publication, similar to Pando (which has published numerous articles strongly critical of Uber) and The Information (which has said nothing critical of Uber but was the primary source of the Uber financial data presented in part one.
 Newcomer, Eric, Uber’s Loss Exceeds $800 Million in Third Quarter on $1.7 Billion in Net Revenue, Bloomberg, 19 Dec 2016. 2nd and 3rd quarter results were actuals; the 4th quarter number was Uber’s mid-December expectation of what the actual results would be.
 There is published evidence that Uber can easily manipulate customer displays to show many more cabs than are actually available. Hwang, Tim & Elish, Madeleine, Uber’s Algorithms and The Mirage of the Marketplace, SLATE, 27 Jul 2015, Rosenblat, Alex, Uber’s Phantom Cabs, VICE, 27 Jul 2015. The larger issue is that Uber’s app was never designed as a passive conduit between independent drivers and customers; it was always designed so that Uber (particularly if it achieves industry dominance) could totally control all of the supply, demand and pricing information taxicab markets require in order to function.
 Although he did not claim that Ebay/Craigslist type network efficiencies existed, Thompson has laid out similar arguments about superior Uber driver productivity in the past (Thompson, Ben, Why Uber Fights, Stratechery, 24 Nov 2014) and repeated the general claim in the article discussed above.
 DeLong, Bradford, There is a serious debate about “Uber, floor wax or desert topping?”–excuse me: “Uber: grift or technological and organizational breakthrough?, Grasping Reality Blog, 20 December 2106. Kaminska, Izabella, The Taxi Unicorn’s New Clothes, Financial Times, 1 Dec 2016.