The Post-Autistic Economics Network, publishers of the Post-Autistic Economics Review, recently held their "Greatest Twentieth-Century Economists Poll":

Who were the greatest economists of the 20th-Century?

"Greatest" here means not who most influenced the economics profession or ideology, but rather who most added to our understanding of economic phenomena. Vote for your top five. The economist who is your first choice will be credited with five votes, your second choice with four, your third with three, your fourth with two and your fifth with one. [emphasis added]

The PAEN is essentially a group of economists, mainly a lot of European (and some American) graduate students, who don't like the mathematical/modelling approach that modern economics uses. Rather than the current paradigm, these economists use descriptive and normative arguments against free markets and neo-classical/econometrics. There is nothing wrong with a different paradigm, in and of itself, but it has to withstand the test of actual data.

Do you need math to be a good economist? No. Do you need math to publish in the AER? Yes. Can you be bitter about that? Yes. Does the math requirement to publish in the AER suggest that economics is less valuable? No.

With this in mind, here are the top ten vote receivers in the poll in which 1,249 subscribers voted:

1. John Maynard Keynes 3,253
2. Joseph Alois Schumpeter 1,080
3. John Kenneth Galbraith 904
4. Amartya Sen 708
5. Joan Robinson 607
6. Thorsten Veblen 591
7. Michal Kalecki 481
8. Friedrich Hayek 469
9. Karl Polanyi 456
10. Piero Sraffa 383

JMK is the economist "who most added to our understanding of economic phenomena"? To me, this doesn't say "heterodox" as much as big-government liberal. Adding JKG, Sen, and Robinson to that list almost seals the deal in my mind. I am surprised (and pleased - I don't know yet) that Hayek made the top ten. It seems that the heterodox movement has a pretty strong opinion on the economists who helped "make economic phenomena understandable," at least for them.

Economists don't put a lot of faith in surveys, nevertheless I grabbed all of the data that were made available and threw it into STATA. Here's a Lorenz Curve (a concentration curve) of vote totals:

The Lorenz curve suggests a highly concentrated opinion about "who is the most influential" economist, as can be somewhat inferred by the top ten vote totals. I would have thought the heterodox movement would be considerably more "democratic" in its opinion of who is considered "great." It must be admitted, however, that without a comparison survey of non-heterodox economists, we can't say much about the concentration curve. The heterodox group may be considerably more democratic than the rest of us.

There were a total of 587 "greatest economists" nominated. I created a dummy variable for whether the economist had won the Nobel Prize, whether the Nobel was shared with another economist and ran a few gimmick "regressions" to see how orthodox the heterodox actually is.

Here's vote total on NOBEL and SHARED:

     Source |       SS       df       MS              Number of obs =     587

-------------+------------------------------           F(  2,   584) =    3.72

       Model |  191029.174     2  95514.5871           Prob > F      =  0.0248

    Residual |  14997819.4   584  25681.1976           R-squared     =  0.0126

-------------+------------------------------           Adj R-squared =  0.0092

       Total |  15188848.6   586  25919.5368           Root MSE      =  160.25


------------------------------------------------------------------------------

       votes |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]

-------------+----------------------------------------------------------------

       nobel |   70.97212   38.39421     1.85   0.065    -4.435432    146.3797

      shared |   -5.37037    48.7636    -0.11   0.912    -101.1438    90.40302

       _cons |   24.47232   6.883479     3.56   0.000     10.95294    37.99171

A few things come to mind here. The nobel prize explains very little of the overall variation in votes received, which might suggest that the heterodox movement is truly heterodox. On the other hand, the nobel dummy variable is reasonably significant (at the 6.5% level) and positive. Those economists who have won a nobel received approximately 71 more votes than non-nobel winners. Does this suggests that the readers of PAER and who selected into the survey might be more "orthodox" than claimed/believed?

Here's a Rank-Size regression to test the Pareto type distribution (and Zipf's special case):

     Source |       SS       df       MS              Number of obs =     586

-------------+------------------------------           F(  1,   584) =25057.98

       Model |  520.006302     1  520.006302           Prob > F      =  0.0000

    Residual |  12.1192396   584  .020752123           R-squared     =  0.9772

-------------+------------------------------           Adj R-squared =  0.9772

       Total |  532.125542   585  .909616311           Root MSE      =  .14406


------------------------------------------------------------------------------

      lnrank |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]

-------------+----------------------------------------------------------------

     lnvotes |  -.6780374   .0042833  -158.30   0.000      -.68645   -.6696248

       _cons |   6.492468   .0091549   709.18   0.000     6.474488    6.510449

The coefficient on LNVOTES is significantly different from -1, refuting the Zipf rank-size distribution commonly found in city size, etc. This suggests that the distribution of vote totals is not random. If votes reflect true opinions about who is "great," then the heterodoxy has an orthodoxy (or a semblence thereof). Moreover, the fact that the parameter on LNVOTES is less than one in absolute value suggest that the Pareto distribution of votes is concentrated, which simply confirms the Lorenz curve above.

What I plan to do in my "spare" time is gather the number of Google hits for each name in the database (this might take a while). My gut tells me that the vote total will be correlated with the presence on the web, however I have gathered data on 65 random names from the survey and the Google hits-to-votes correlation is only 0.037 (hopefully this will improve, or it's going to be an even bigger waste of time).