UNEMPLOYMENT DISPERSION AND PHILLIPS LOOPS: A DIRECT TEST OF THE LIPSEY HYPOTHESIS
| DOI | http://doi.org/10.1111/j.1468-0084.1979.mp41003004.x |
| Published date | 01 August 1979 |
| Author | DAVID J. SMYTH |
| Date | 01 August 1979 |
UNEMPLOYMENT DISPERSION AND PHILLIPS LOOPS:
A DIRECT TEST OF THE LIPSEY HYPOTHESIS
By DAVID J. SMYTH
In his study of the relationship between money wage rates and unemployment
for the United Kingdom over the period 1861 to 1957, Phillips (1958) found clearly
observable ioops, the observed change in money wage rates being above the fitted
curve when unemployment was falling and below the curve when unemployment
was rising. Lipsey (1960) advanced the hypothesis that the ioops were the result
of aggregation over micro-markets with the dispersion of unemployment rates
between markets varying cyclically. With a Phillips curve convex to the origin
and unemployment rates differing between markets, observed macro-points will lie
above the true relationship, the extent of this divergence increasing with the degree
of dispersion. Lipsey then hypothesized that dispersion was greater when
unemployment was falling than when it was rising, the result being anti-clockwise
loops.A number of empirical studies of the Phillips curve (Archibald (1969), Thomas
and Stoney (1971), Brechling (1973), and Thirwall (1969), (1970)) have included
dispersion measures, mostly with little success, but they have not attempted to test
Lipsey's hypothesis on the cause of the loops. Indeed, they use data sets yielding
little indication of loops anyway. Moreover, the emphasis in the studies is on
regional dispersion rather than on the role of industry dispersion, as was
hypothesized by Lipsey. It is surprising that researchers have not made direct tests
of the Lipsey hypothesis where it might be expected to be strongestthe original
PhillipsLipsey time period and most particularly the pre-World War I period,
when the loops are most apparent. The present paper undertakes such an analysis.
Section i outlines the Lipsey hypothesis and presents a testable form of it.
Section 2 presents empirical evidence for the period 1861 to 1913 and the evidence
for the years 1923 to 1939 and 1948 to 1957 is given in Section 3. The paper's
conclusions are given in Section 4.
1. THE LIPSEY UNEMPLOYMENT DISPERSION HYPOTHESIS
Suppose that the relationship between change in money wage rates and
unemployment in individual micro-markets is given by
thf(u1) f'(u)<0,f"(u)>0 (I)
where w and u are respectively the change in the money wage rate and
unemployment in the jth market. The aggregation over micro-markets to get a
macro-relationship will yield points that lie above the micro-curve if unemployment
rates differ between markets. This is illustrated in Fig. I where there are two
micro-markets with unemployment rates u1 and u2 and the observed macro-point
will lie on the straight line joining points I and 2, the precise point on the line
achieved being dependent on the relative weights assigned to market 1 and market 2.
227
To continue reading
Request your trial