Skip to content

Another stylized fact bites the dust

According to economist Henry Farber (link from Dan Goldstein):

In a seminal paper, Camerer, Babcock, Loewenstein, and Thaler (1997) find that the wage elasticity of daily hours of work New York City (NYC) taxi drivers is negative and conclude that their labor supply behavior is consistent with target earning (having reference dependent preferences). I replicate and extend the CBLT analysis using data from all trips taken in all taxi cabs in NYC for the five years from 2009-2013. The overall pattern in my data is clear: drivers tend to respond positively to unanticipated as well as anticipated increases in earnings opportunities. This is consistent with the neoclassical optimizing model of labor supply and does not support the reference dependent preferences model.

I explore heterogeneity across drivers in their labor supply elasticities and consider whether new drivers differ from more experienced drivers in their behavior. I find substantial heterogeneity across drivers in their elasticities, but the estimated elasticities are generally positive and only rarely substantially negative. I also find that new drivers with smaller elasticities are more likely to exit the industry while drivers who remain learn quickly to be better optimizers (have positive labor supply elasticities that grow with experience).

It’s good to get that one out of the way.


  1. numeric says:

    Don’t know that this kills the “stylized” fact, as the periods studied are different–there simply could have been a behavioral shift between the two periods. As a counter-example to your conclusion, potatoes displayed a Giffen good property during the Irish potato famine but it you replicate in a different time period they would not.

  2. Tom says:

    With all due respect to Farber, analyzing average daily cab driver wages while controlling for precipitation, driver experience, etc. is incorrect. The analysis should be done at a much more disaggregate and local level, i.e., hour by hour within a day, specific to GPS location within the city. The reasons for this should seem obvious: precipitation can occur throughout the day and will vary in intensity by location, impacting demand. Precipitation is just as likely to occur when a driver is starting a shift as it is when he is making a decision to end a shift. Moreover, “precipitation” isn’t just precipitation. Snow day behavior should be different from rain day behavior. Not to mention that the intensity of the precipitation (e.g., light snow that doesn’t stick vs days long blizzards vs heavy rain with flooding) can have significant and widely differing impacts on driver decision-making. Driver strategy should play a role here as well insofar as some drivers prefer driving days vs nights (when traffic can be lighter), while some only work Manhattan where others prefer to sit in the queues at the airports. Finally, driving conditions can be significant in shaping decisions. E.g., traffic speed is a consequence of many more factors than just weather. The impact of traffic speed may be the single most important consideration with respect to shift termination, especially late in the shift.

  3. dmk38 says:

    Don’t we need to know more before we conclude that the “reference-dependent preferences” hypothesis has “bitten the dust”?

    If there is some reason for thinking CBLT’s study was *not valid,* then fine– treat their finding as entitled to zero weight & go w/ our priors that “of course” markets work the way microeconomics say, times the weight of the Farber corroboration, etc.

    But if CBLT found one thing in a valid study and Farber another, then why would we say “ha, CBLT has been refuted!”

    Of course, that’s not the way empirical proof works. Valid studies don’t “prove” things; they just give us more reason to believe one thing or another than we would have had otherwise, and it is not at all unusual for their to be multiple valid studies that put evidentiary weight on opposing sides of the scale at once. One then aggregates all the proof, makes a provisional assessment, and waits for more evidence– which then in turn gets added to the scale, resulting in a revised provisional assessment pending new evidence, and so forth and so ad infinitum.

    You properly make this point all the time, e.g., in connection with the thoughtless stance some thoughtless researchers & others take about matters being “proven” by “significant” results published in “peer-reviewed” journals.

    So I guess I’m just wondering why you are expressing an attitude toward CBLT that doesn’t seem to fit. Is there more to the story?

    • Andrew says:


      You’re right, “bites the dust” is too strong. I’ve always been skeptical of the CBLT claim, and I’ve always felt that people have tried to push a counterintuitive and not so plausible story based on that study. I guess what I’m saying is that this new claim is more consistent with my priors. So, at the very least, I’d say that we no longer “have no choice but to accept that the major conclusions of these studies are true,” as Kahneman might say.

  4. dmk38 says:

    (I see I’m replicating Rahul’s comment– but much less economically)

    • Rahul says:

      As an aside, why the glee when a “stylized fact bites the dust”?

      We do need stylized facts to function & reason in this complex world, right? No one thinks that stylized facts are universal nor without exception.

      But so long as, on average, they hold, stylized facts seem to serve a useful purpose. Without generalization it’s difficult to progress.

      • Andrew says:


        If you think the statement, “It’s good to get that one out of the way,” expresses glee, you don’t have enough glee in your life!

        More to the point, true stylized facts are good, false stylized facts are not so good. To take a couple of extreme examples, the stylized fact “viruses cause the flu” serves a useful purpose, while the stylized facts “beautiful people have more daughters” or “Daryl Bem discovered evidence of ESP” are unhelpful (indeed I believe they have a negative effect in that they impart a misleading view of science to various audiences who are exposed to these stylized facts, audiences such as the readership of the Freakonomics blog).

        I have produced many stylized facts myself; I think it’s good when I do this, and it’s also good when someone points out problems with one of my stylized facts. Conditional on the fact having a problem, I’d like to know about it.

  5. dmk38 says:

    I aspire to *be* a stylized fact

Leave a Reply