Monday, April 28, 2014

Minnesota Mafia Challenges Piketty

Let me begin by noting that I have as yet been unable to obtain a copy and so have not read Piketty's smash hit book yet.  However, I think that I know enough about what is in it to post on this particular matter.  Anyway, Tyler Cowen at Marginal Revolution has posted a challenge put to him by Tony Smith regarding Piketty's book that can be labeled as coming from the Minnesota Mafia.  The econ department at the University of Minnesota, along with the closely allied research department at the Minneapolis Fed, which have long had people going back and forth or simultaneously in both, has long been the real fountainhead of new classical real business cycle DSGE macro, even if some of those who initiated that movement there are either not there anymore (Prescott, and many former students) or no longer a follower of it (Kotcherlakota, now Minneapolis Fed President, although for its failure to predict the crash or say much useful about Fed policy) or both (Sargent).  Nevertheless, a major contingent remains in one or both of those places, including Chari, Kehoe, McGrattan, Rios-Bull, and several others, and their former students continue to identify strongly with the place while they are now all over the world.

The centerpiece of  post is a paper by Castaneda, Dias-Gimenez, and Rios-Bull ( entitled, "Accounting for the U.S. Earnings and Wealth Inequality," published in the Journal of Political Economy in 2003.  Several other papers are linked, with the most impressive by Heathcote, Storeslettin, and Violante
, an overview paper of the broader approach from 2009.

These models are variations of DSGE models, except that they involve incomplete markets, with Heathcote, et al. calling this approach the "standard incomplete markets"(SIM) approach.  The difference from usual DSGE models is what markets are incomplete, in this case idiosyncratic uninsured risks.  So, the missing markets are insurance ones, and these are what eventually explains the development of inequality.  The other part needed is that there are heterogenous agents done in the Minnesota way, an interval on some variable, in this case discount factors.  The main paper also adds a social security mechanism.  With proper calibration, they claim to reproduce the pattern of income and wealth inequality in the US economy up to 2003.  Most controversially they claim that introducing an estate tax will make little change in wealth distribution, only raising the wealth Gini from .79 to .80.  Cowen is impressed.

So, what is going on here?  Heathcote et al lay out the various mechanisms and review the broader related literature.  Various models have thrown in as shocks family and human capital ones, with the missing insurance markets including finance, public, and some others.  There is even a policy claim that being able to separate initial condition effects from later shock effects suggest policies focused on early education to improve human capital or others focused on later "insurance."  Obviously this is very different from the reputed story that Piketty ultimately develops regarding dynastic families emerging in a "patrimonial capitalism" and his focus on the overall return to financial capital compared to the overall rate of growth.  The stories are extremely different.

In the end, what is really driving Minnesota models is this distribution of different discount factors.  So, it is at the bottom line that those on top are patient and willing to abstain while those at the bottom are short-sighted and impatient, tsk tsk.  In this view, obviously the losers deserve what they (don't) get, while the virtuous rich deserve theirs.  This is the Protestant Ethic triumphant!

Barkley Rosser


Sandwichman said...

Piketty appears to have anticipated this objection. pp. 358-9:

"The problem with this theory [time preference] is that it is too simplistic and systematic; it is impossible to encapsulate all savings behavior and all attitudes toward the future in a single inexorable psychological parameter. If we take the most extreme version of the model ["infinite horizons"]... it follows that the net rate of return on capital cannot vary by even as little as a tenth of a percent: any attempt to alter the net return (for example, by changing tax policy) will trigger an infinitely powerful reaction in one sense or another (saving or dissaving) in order to force the net return back to its unique equilibrium." said...

I am not fully clear why Castaneda et al get the result they do on the ineffectiveness of estate taxes to affect wealth distributions, but it may be something like what Piketty says. The other part may have to do with them not really redistributing that much wealth, although it strikes me that they are a way of stopping, or limiting at least, the emergence of feudal style dynasties.

Tony said...

Just to be clear, Castaneda et al do not assume heterogeneity in discount factors, but instead (amongst other things) calibrate the income process so that at any point in time there is a small group of very high earners who have a relatively large probability of dropping to a lower level of earnings. Krusell and Smith (1998) have a simpler income process (employed or unemployed) but do assume heterogeneity in discount factors. In any event, both papers can generate realistic wealth inequality using mechanisms that appear to be very different from the Piketty "r-g" mechanism.

Kevin Donoghue said...

As Tyler Cowen notes in his update, Piketty commented on exercises of the kind done by Castaneda et al.:

" key driving force in these models is naturally the macroeconomic importance of inheritance flows: other things equal, larger inheritance flows tend to lead to more persistent inequalities and higher steady-state levels of wealth concentration. However this key parameter tends to be imprecisely calibrated in this literature, and is generally underestimated: it is often based upon relatively ancient data (typically dating back to the KSM controversy and using data from the 1960s-1970s) and frequently ignores inter vivos gifts. We hope that our findings can contribute to offer a stronger empirical basis for these calibrations."

So the calibrated models "predict" outcomes matching suspect data. I'm no expert on calibration, but that doesn't sound like a scientific triumph to me.

JW Mason said...

allied research department at the Minneapolis Fed

Not so much anymore: said...

Tony (Smith?) is right that it is Krusell and Smith rather than Castaneda et al who have varying discount factors, thus fulfilling the Protestant Ethic model.


Calibration is a key method associated with the Minnesota approach, and is highly controversial, generally viewed as in competition with structural econometric testing oro time-series VAR-type models, the latter inspired heavily by Sims, who was at Minnesota for a period of time.

Anonymous said...

The reason why the authors get the result they do about the ineffectiveness of estate taxes is because all of the earnings in their model are earnings from labor and there are no earning from capital!

Tony said...

Dan, that's not right. Look at equation (5), a household's budget constraint: a is a household's stock of wealth and r is the return on wealth. (Both are endogenous variables.)