post-autistic economics review
Issue no. 36, 24 February 2006

issue 36 contents PAE Review index home page

Comment on "Some Primitive Robust Tests of Some Primitive Generalizations"

I am writing to comment favorably on "Some Primitive Robust Tests of Some Primitive Generalizations" (Kurt Rothschild, post-autistic economics review, issue no. 35, 5 December 2005, article 1, pp. 2-10, http://www.paecon.net/PAEReview/issue35/Rothschild35.htm)

Rothschild argues that not one of the central points of the neo-liberal paradigm {viz.
(1) Low inflation promotes economic growth (2) Low inflation promotes low unemployment (3) Low government expenditure (slim states) promotes GDP growth (4) Low Trade Union Density (TUD) is favourable for GDP growth (5) Low TUD promotes low unemployment (6) Low corporatism favours high GDP growth } is supported by the last 35 years of Western European experience.

Rothschild examines the EU15 less Luxembourg and adds Switzerland and Norway.  For each neoliberal claim, which pairs a "good behavior" with a good outcome, Rothschild compares the list of countries in the top half and bottom half in each part of each pair.  He finds little correspondence between the lists.

I am writing with an alternative exploratory data analysis that largely confirms Rothschild's results and may have some advantages with respect to elegant and parsimonious display of the data and the ability to identify outliers and special cases.

The attached graphic, a matrix of scatterplots of the data in Rothschild's Country Table, addresses all six neoliberal claims.  In five of the six cases, the scatterplots provide no support or outright contradiction of the neoliberal claims. GDP growth versus inflation and unemployment versus union density provide particularly noteworthy counterevidence against the neoliberal claims.

In one case, claim (3), government expenditure appears to have a slight negative correlation with GDP growth.  However, the graphs also illustrate and support Rothschild's observation about the catch-up countries, Greece, Ireland, Portugal and Spain.  In the case of claim (3), if the catch-up countries are excluded, the relationship between government expenditure and GDP growth becomes tenuous indeed.  I, for one, would not cashier my welfare state on the dubious promise of 0.3 percentage points of GDP growth from converting Sweden into the U.K.

Furthermore, graphical analysis flags the exceptional Irish case as a clear outlier in the GDP growth; the exceptionalism is most likely a mixture of catch-up from a particularly disadvantaged position and the effectiveness of a developmental state (see, for example, O Riain 2000).  In any case, the simple graphical approach makes the exceptionalism obvious.

In conclusion, the scatterplot matrix provides confirming evidence for Rothschild's findings, and the method is more accessible than the comparison of top halves with bottom halves in the original paper.

Sincerely,

Michael Ash

University of Massachusetts Amherst

Notes

1. The data for the plots come directly from the Country Table in Rothschild's paper,

http://www.paecon.net/PAEReview/issue35/Rothschild35.htm)

2. The plots were produced with the R statistical language and application, a free/libre open-source application available from http://www.r-project.org

3. The short script used to produce the plots is available from Michael Ash,

mash@econs.umass.edu

Reference

O Riain, Sean (2000) The flexible developmental state: globalization,information technology and the "Celtic Tiger". Politics and Society 28(2):157-193.

___________________________

SUGGESTED CITATION:
Michael Ash, “Comment on ‘Some Primitive Robust Tests of Some Primitive Generalizations’"

post-autistic economics review, issue no. 36, 24 February 2006, pp. 36-38, http://www.paecon.net/PAEReview/issue36/Ash36.htm