Academics in the Uber files
July 23, 2022
The Uber files reveal that some, using data provided by the company, highly respected economists have produced a report for which they were paid a very high fee. The subtext is that they have been corrupted, especially as they also provided comments for the public media, apparently a commitment included in the contract. This raises two main questions concerning, respectively, the content of the report and its ethical appropriateness.
The report, written in French, does not deal with the economic logic of Uber, but with its impact on the drivers’ employment and income. Augustin Landier and David Thesmar are well known scholars for their work on financial issues. They have had brilliant careers, rewarded by professorships at the Toulouse School of Economics for the former, and at MIT for the latter. Co-written with Daniel Szomoru, a Uber employee, the paper has been available on the internet ever since it was completed in 2016. Using a combination of public and private data provided by Uber, it documents who the drivers are in the Paris area. They are young and they often come from poor neighborhoods where young men are massively unemployed, likely to engage in illicit activities leading many of them to jail. Working as Uber drivers has become a full-time activity, which they find financially rewarding. Many of them log long hours to escape poverty. The authors also report that those who succeed tend to stay in the job while those who are less successful soon drop out, which they interpret as successful on-the-job experimentation. My personal, casual experience in many other European cities suggest that these results extend beyond Paris.
This study rests on a fairly simple treatment of the data. More sophisticated analyses are probably possible, and would be needed to check whether the evidence survives. When a new topic emerges, it is often the case that research starts simple and then gradually becomes more elaborate. The simple first investigations are usually praised for attracting the attention of other researchers. I am not a specialist of this new area of study, so I don’t know whether more work has been produced and, of so, whether the initial results are robust (a quick search did not turn out much). Interestingly, the critics do not mention any other research. In fact, few of them actually report the results and, when they do, they argue that the conclusions are unwarranted on the basis of cursory considerations. In doing so, the critics violate the very basic principle of science, that results cannot be dismissed because we don’t like them. Until formally proven wrong, the results of Landier and co-authors currently stand.
The ethical aspects are more delicate. Countless studies use private data to explore new issues. In fact, over the last decade or so, economic empirical research has forcefully evolved toward ‘big data’. This is a way of trying to escape the ‘small sample’ curse, which limits the ability of researchers to use statistical techniques that are more precise and therefore more reliable. Many of the big data banks now in use can only be obtained from private companies. For example, early on during the early phase of Covid-19, when everything was unknown, anonymized data provided by Google made it possible to quickly understand how the pandemic and lockdowns were affecting people’s mobility, going to work and shopping behavior (yes, Google knows that about you). Use of private data, therefore, is not a proof of corruption.
In the pandemic case, the data was provided to any researcher requesting access, and it was exhaustive. Landier and co-authors indicate that they used a poll conducted by a specialized firm (IFOP) under contract for Uber, with further information provided by Uber from ‘internal data’. It is not known whether other researchers could get the same data, nor whether Landier and co-authors were given all the relevant information. This is not original either. Many studies rely on proprietary data that the researchers are not allowed to publish or even share. A number of professional reviews have started to turn down submissions based on confidential information. At the time when the paper was written, in 2016, such restrictions were not yet imposed, I think. The fact that this paper was not published since then may indicate that the authors themselves anticipated some difficulties in this respect, but this is pure conjecture.
Then there is the fee issue. The Uber files indicate that the three authors were jointly paid 100,000 euros. This sum is at the very end of the spectrum for what was, after all, a simple exercise. It may be that such amounts are less exceptional in the field of finance for large studies that require a lot of work for data treatment. It may be that very well-known economists fetch fees of this order of magnitude, but then ethical issues immediately arise. In addition, one author was a Uber employee. It is not known what his share of the work was, nor whether he had received instructions about what the report should say or not say. This is a vast problem. It is possible that many researchers tilt their work to please, or to avoid displeasing their employers, be they academic institutions or public institutions like central banks, government agencies or indeed private corporations, not to mention ideological biases. My reading of the economic literature has long made me feel that the economics profession is not above suspicion. The usual response is that the refereeing process corrects for this risk, but is is not fully convincing. Had the report been published, the authors could invoke this justification.
Surprisingly, the report does not deal with the bigger picture of what has become known as uberization. Back then, in 2016, my perception of initial thinking was that most economists saw Uber as a positive innovation. In most countries, taxi services were – and still often are – a closed shop activity where entry was strictly limited and prices set by an unholy alliance between the authorities and the drivers’ unions. Service was poor, availability of service was often insufficient (remember waiting forever under the cold rain?) and rents were sizeable if often captured by firms. Most economists saw Uber as a welcome contribution to making entry possible. In addition, Uber made it possible to mobilize unused productive capital, namely the private cars that remain idle most of the time. More generally, Uberization is seen as promoting a new form of employment, away from salaried jobs. Self-employed workers can decide when, where and how they provide services. This freedom comes with costs, including the risks inherent to having to seek customers and the need to provide for their welfare needs, including unemployment compensation, health benefits and retirement pensions. Dealing with this new form of employment will require adjustments to current arrangements. Such adjustments are likely to be vigorously opposed by trade unions, structurally attached to traditional forms of employment. If confirmed by subsequent studies, the evidence offered by the report indicates that uberization is attractive to a number of people so far marginalized from the labor markets. Post-Covid, this trend may have accelerated. Many of the attacks on the report seem to reflect this brewing conflict more than a genuine perception of corruption.