San Francisco 2014
 
Blog:
blog home Home
 
       Follow:
RSS
 

OR: We’re better than Economists!

by David Hutton on November 10th, 2014

I admit: I often don’t attend the plenary sessions. I usually like to go to the sessions focused on areas close to my research. And I also value my time talking with colleagues I haven’t seen in a while. But, if you missed Jonathan Caulkins talk today on OR in services of Drug & Addiction Policy, you missed a good one. Here are some of the key take-aways:
1. Jonathan Caulkins has lots of cool stories of how he’s used simple OR methods to get some great insights into drug policy. He has several stories where economists were using regression models to look at historical changes in drug use. Unfortunately, the regression models could not to model the big swings and drop-offs observed in actual drug use. As many of us OR-types know, swings up and down might be indicative of more complex systems dynamics. Jonathan was able to use some simple models like the Bass model of product diffusion and a simple two-state Markov model to both dramatically replicate observed historical patterns of drug use and also show how the type of drug users (light and heavy) could have a dramatic impact on what types of intervention policies might work best.  The Office of National Drug Control Policy felt these analyses were some of the most insightful they have ever seen.
2. A little data can go a long way: Jonathan was able to combine small datasets on drug prices and drug purity over time to get some great new insights into how supply-side shocks affected drug use and health outcomes (e.g., accurately explaining an 80% decrease in heroin deaths in 2001 Australia). As he said, “you learn more going from 1 to 100 data points than from 100 million to 1 billion data points.”
3. From his experience influencing policy, he thinks we could make a few slight updates to our OR curriculum to make our students the next generation of policy leaders, leading interdisciplinary teams. He thinks OR grads would be better suited to this than the amalgam of economists and lawyers currently leading policy teams. That’s probably a bigger discussion in need of its own blog post. I’m sure many of you have thoughts on that as well.

1 Comment
  1. Ken Sanford permalink

    David,

    I really enjoyed your post! While I disagree with almost all of it, it did make me smile. You must have a little “economist” in you as you clearly understand how to sell your blog through a catchy title.

    As to the meat of the article, I am not at informs but I am willing to speculate on a few things.

    To your first point I would expect the field of marketing science to object to calling the Bass Diffusion Model an “OR tool.” A social scientist created it and it widely used in spatial econometrics. Next, I’m fairly sure financial economists and applied statisticians have been applying Markov models for a long time.

    I suspect the reason economists “weren’t able to explain the large swings in drug usage,” is because that was not the objective of their analysis. The modern day economist is primarily concerned with estimating treatment effects of certain policies and uses a number of methods to extract causal inference. Among these are a variety of panel data methods, regression discontinuity, propensity score matching and instrumental variables.

    As to the third point, I will put my economist hat on and remind you that it is supply and demand that leads to so many economists being placed into these policy jobs. In this case, it is supply and prevailing wages.

    The NSF study on conferred Ph.D.’s estimates that each year, more than 1,200 Ph.D.’s are granted in Economics, strictly defined. The closest field to “OR” in the study is Industrial Engineering. The report lists 227 Ph.D.’s conferred in 2012. (The subfield report lists 90 Ph.D.’s in OR from Engineering. There are another 80 OR Ph.D.s from the B-schools. I could find more economist-like Ph.D.’s but I have made my point. 5 econ Ph.D.’s to each OR graduate.)
    The reason OR Ph.D.’s are not more prevalent as policy leaders is that there aren’t many of them. And, I would gather, the average wages of policy leaders don’t warrant those with OR skills to even glance in that direction.

    The point about “changing the curriculum” might do more to bring economics students toward OR than create more well-rounded OR professionals. I think this would be a great thing. Economists always thing in terms of objective functions so it would be nice to have more formalized training.

    All in all a fun post that made me do a little digging to support my assertions. Enjoy the intimate conference of 4,000 attendees. The economists will bring 12,000 to ASSA in January. 🙂

    Ken
    http://www.nsf.gov/statistics/sed/2012/pdf/tab12.pdf
    http://www.nsf.gov/statistics/sed/2012/pdf/tab13.pdf

Leave a Reply

Note: XHTML is allowed. Your email address will never be published.

Subscribe to this comment feed via RSS