Imbens Fires Back at Deaton

A few months ago, Princeton economist Angus Deaton offered his vision for development economics.

In his piece, he rails against the movement toward relatively atheoretical, randomized experiments, calling for closer ties between theory and empirics.

“The great economists should be trying to do something that is harder.”

Now, in an excellent new paper, Harvard economist Guido Imbens fires back.

Imbens argues that Deaton is too dismissive of the special value that randomized experiments have in assessing causality, and that natural experiments, while not as good as randomized experiments, are far better than Deaton gives them credit for.

While it might seem difficult to mostly agree with both Deaton and Imbens, given that they espouse polar opposite views, strangely enough that is where I stand in this debate.

On the points Imbens makes, I think he is exactly right. However, what Imbens minimizes, in my opinion, is the importance of models and theory for motivating empirical research, including randomized experiments.

Given a question, of course you want a randomized experiment to give you the best answer possible. What I think has happened too much in economics recently is that the availability of experiments has trumped the asking of good questions. Or, put another way, anyone can do program evaluations based on true randomization, so why should some of the world’s best economists be devoting so much of their time to such exercises?

The great economists should be trying to do something that is harder.

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.

 

COMMENTS: 9

View All Comments »
  1. charles says:

    I love the last line.

    Thumb up 0 Thumb down 0
  2. Nitin says:

    Disagree with the last line.

    Why should great economists try to do something that’s harder rather than do useful stuff better? Is there a rational reason to get higher brag value by cracking the harder problem? Other than ego massaging, of course.

    Thumb up 0 Thumb down 0
  3. Morley says:

    Nitin, you’re missing the point of what he’s saying. He’s not saying economists should tackle hard problems for the bragging rights. He’s saying economists should tackle hard problems because it’s a waste of talent to focus on easy problems. It’s the same reason you don’t use your SWAT team to hand out traffic tickets.

    Thumb up 0 Thumb down 0
  4. Caliphilosopher says:

    So what exactly are the harder issues/questions that economists be trying to answer?

    Thumb up 0 Thumb down 0
  5. FC says:

    I agree with your last point for the most part. However, the great economists have more success than relative unknowns in convincing policy makers to implement large scale experiments or to try innovative programs, and in being the “heads” behind publicizing successful policies so that they’re implemented more widely.

    Economists in general care about truth and making a difference. Sometimes the pursuit of truth takes a backseat to making a difference.

    Thumb up 0 Thumb down 0
  6. James BonTempo says:

    My interpretation of that last line is that economists should focus more attention on discovering what the good questions are & asking them. Sounds like a pretty good use of their time in an environment awash in experimentation.

    Thumb up 0 Thumb down 0
  7. intricatenick says:

    As someone awash in poorly understood empirical data (metagenomics) it is my experience that often times theories influence the interpretation of empirical results in a way that leads to their confirmation. Hypothesis driven research often results in dialectical wars of causation vs correlation that miss the larger point – which is that to understand any model or system everything (all variables)must be accounted for. A lot of time, prior hypotheses force the interpreter of empirical data to make more assumptions than he/she is comfortable with in order to fit the data into that framework.

    Randomized data is often useful for generating hypotheses by itself and testing these back on the data set by random partitioning in a feedback mechanism. Neural net modeling is an example of this.

    Both methods are still dependent on statistical distributions and to me this is a larger problem in economics thatn this debate.

    Thumb up 0 Thumb down 0
  8. Princess Leia says:

    I can’t understand this! Too academic for me. Personally, I think great economists should lighten up and make plans to connect with good friends at an upcoming college reunion. :)

    Thumb up 0 Thumb down 1