Freedom, Justice, and Economics

I was impressed to discover today that a decade-old paper by Richard Blundell and Steve Bond, “Initial conditions and moment restrictions in dynamic panel data models,” has 1503 cites on Google Scholar. The paper discusses the validity of panel data estimation approaches using particular selections of GMM identification restrictions. Techniques validated by the paper have become ubiquitous for their generality.

Saltwater empiricists in the US, however, tend to expect clean experimental variation. And freshwater empiricists in the US also generally pursue a different approach, relying on identifying restrictions that come from models with more structure. The Blundell-Bond paper and the related literature suggest a generically valid approach conditional on certain lags and lagged changes of error terms and dependent variables being uncorrelated.  I am led to the question: which real-world settings reliably satisfy the required conditions?

Every now and then someone picks a fight with me about the epistemology of science. As a former physicist and current economist, I might be particularly touchy on this topic. But I’ve found myself comfortable with a simple position that efficiently resolves most debate.

Often at issue is that many scientists demand that we are searching for capital-T-Truth. Logic and mathematics are indeed about truth– or at least conditional truth– in the sense that very specific rules tell us what conclusions can be drawn from what premises. To the extent theorists (in physics or economics, say) just do math, that research is also about Truth. However, if the premises– the assumptions of the model– are wrong, that Truth may have no bearing on reality.

For all applied work– work that uses real-world data, sometimes to test various theories– my satisfying criterion is whether we’ve come up with a way to make reliable predictions.  Mixing hydrogen and oxygen gives you water and a bundle of energy:  that’s a reliable prediction.  The next solar eclipse will occur on August 1, 2008.  If a central bank prints a huge amount of money and pours it into an economy, inflation will result.
I care little about whether these are everlasting Truths.  (Sometimes predictions are possible because we’ve observed the same phenomenon repeatedly and reliably: under ordinary circumstances, putting a pot of water on a hot enough fire will cause the water to boil.  Sometimes predictions are possible because we have an encompassing underlying theory:  gravity assists can be used to send probes like Cassini to their destinations.  I guess I would say that to me those underlying theories represent something like Truth.)  Mostly, I just appreciate that science and scientists have learned enough to make these and other predictions about the world with very high levels of confidence.

At the University of Illinois at Urbana-Champaign, construction plans for a library were influenced by an unusual consideration: the shadow the building might cast. Established in 1876, the Morrow Plots are the longest continuous agricultural demonstration plots in the world. Since it would have been unfortunate to interfere with their sunlight, the UIUC library was built adjacent to them but out of the way–underground.

We all make thousands of small decisions each day–whether to snooze a few extra minutes in the morning, how long a workout to have, whether to snack on carrots or cookies, whether to have decaf or regular coffee–that may affect our immediate and future well-being. We get some immediate feedback on some of those decisions (geez, those carrots were tasty), but detailed quantitative feedback on immediate and delayed effects is challenging.

Many such decisions have immediate, identifiable, bioelectrical and biochemical signatures. Imagine a device that automatically tracked and uploaded this information, standard metrics of body function (e.g., pulse, breathing rate, temperature, bp), and manually-inputted subjective measures of well-being (e.g., headache, euphoria, anxiety, zone) and productivity. Imagine using all this information and a decent stats package to make inferences about the effects–specific to oneself–of many of life’s small decisions. Many of the inferences would be obvious and well-known: sleeping very little makes you sluggish; eating carrots makes you feel virtuous; talking with dear old friends makes you elegiac, reflective, and happy.

For a device that would track lots of bio-indicators automatically and make it easy to track food intake, exercise info, and subjective variables on the fly, I doubt I’d blink about paying $10,000. Such a device would give me far better tools for enhancing my own productivity and well-being. Maybe my dear old friends also profoundly believe in me, motivating me to do more good; those carrots can give me spates of indigestion, making them less virtuous; and blogging occasionally loosens the chains and accelerates my other writing. I’d like to run the stats, controlling for daylight hours, age, the weather, the number of seminars I’ve been attending, and my overall workload, see the results, and adjust accordingly. Explicit experimentation could come soon after. Just a 1% increase in productivity would make the gadget pay well within my lifetime.

Roy Baumeister’s talk at the American Psychological Association’s annual meeting provocatively argues the following. Historically, cultures that have matched men’s higher variance in reproductive outcomes with higher-variance occupations for men have been more likely to succeed.

The article is loosely argued, poorly referenced,… and probably, in its central claims, with more than a grain of truth. It certainly has no shortage of interesting fact and anecdote. Three observations:

1. Heterogeneity among men and among women may be at least as large as the differences between the genders. To the extent that’s true, it remains a puzzle why cultures would have so sharply delineated gender roles.
2. One clear implication of Baumeister’s theory is that strong cultural advocacy of lifelong monogamy– which promotes gender equality in reproductive outcomes– ought to be paired with cultural advocacy of occupational equality.
3. Things have changed. In the last century or so, particularly in the West, women have had historic achievements in every sphere. Baumeister doesn’t attempt to explain why.