The remarkable transformation of the UV curve in economic theory

Abstract

The purpose of this paper is to provide an analysis of the impact the Unemployment–Vacancy (UV) curve (or Beveridge curve) had on economic theory and to provide an account of the subsequent radical changes in its place and role over the decades since its first appearance in 1958. The paper traces the historical development of the UV curve and argues that the role of the UV curve has changed from that of a measuring device, to a graphical representation of full employment, to an axiom necessary for matching models of unemployment to a diagnostic tool. This changing role is best understood in the light of the paradigmatic change from Keynesianism to neoclassical search theory.

Keywords:

• UV curve
• Beveridge curve
• UV analysis
• structural relations
• matching function
• search theory
• full employment
• labour economics

JEL classification:

• B
• E
• J

Acknowledgements

The author would like to thank Mary Morgan, Walter Eltis, John Davis, Atsushi Komine, Marcel Boumans, Harro Maas and two anonymous referees for their helpful comments on earlier drafts and for helpful discussions. The usual disclaimer applies.

Notes

 1 See, for example, Leeson (2000).

 2 At that time, DDM were associated with the National Institute of Economic and Social Research (NIESR) in London, an independent, non-profit organization founded in 1938.

 3 A slightly rewritten summary of this article can be found in Dow (1970: 337–43).

 4 In addition, DDM employed two tests of unemployment and vacancy data, and found indications that the measurement error of vacancies is fairly stable. First, they compared the effect of seasonal changes on vacancies with the effect of seasonal changes on unemployment. The amplitude of the seasonal variations turned out to be of the same order, suggesting a stable measurement error (DDM 1958: 26). Second, DDM considered changes in notification of vacancies when the statutory regulations for labour exchanges changed. For the period 1946 to 1956, there were two periods when notification of vacancies to labour exchanges was compulsory in the UK: October 1947 to March 1950 and February 1952 to April 1956. However, the change from voluntary to compulsory notification had only a modest effect on the rate of notified vacancies. DDM again interpret this as evidence that the measurement error in measuring vacancies is fairly stable.

 5 Maladjustment is then defined as the amount of unemployment at the point where u = v/s (DDM 1958: 20) and is thus measured in terms of the corresponding unemployment rate.

 6 Members of NIESR in the late 1960s and early 1970s were, among others, Arthur Brown, David Worswick, John Bowers, Paul Cheshire, Edward Webb and Robert Weeden.

 7 The results were presented in Brown (1972), Cheshire (1973), Weeden (1973, 1974), and Webb (1974), while the framework is explained in close detail in Cheshire (1973).

 8 See, for example, Cheshire (1973), Webb (1974) and Armstrong and Taylor (1980).

 9 The relation between the Phillips curve and the UV curve was formally derived as follows. The UV curve can be written as:

• The increase in wages can be written as a function of excess demand for labour: Δw = f(XD). With the difference between vacancies and unemployment taken as excess demand, this relation becomes:

• Substituting Equation (n1) into Equation (n2) then gives:

which is the original 1958 Phillips-curve relation; that is, the relation between wage increases and unemployment.

10 All unemployment is thus structural unemployment since frictional unemployment is absent. Hansen's assumption of markets is akin to the Phelps–Lucas island paradigm, which was invented at about the same time and became a popular tool of analysis in search theory.

11 Bowden (1980) stresses that the real wage rate w / p would be more appropriate.

12 Translated from Japanese into English by Atsushi Komine in e-mail correspondence.

13 Solow's empirical study was motivated by the debate in the early 1960s that became known as the ‘structuralist-deficient demand’ debate, or, as it was sometimes referred to, the ‘structuralist–antistructuralist’ debate, on the nature of unemployment in the US. The claim of the structuralists was that the high unemployment in the US in the 1960s (around a 5% level) was caused by an increase in structural unemployment.

14 The Help-Wanted Index of the National Industrial Conference Board is a weighted average of indexes of the number of help-wanted advertisements posted in leading newspapers of 52 cities in the US.

15 Cohen and Solow (1967, 1970), Burch and Fabricant (1968, 1971) and Gujarati (1969), all in The Review of Economics and Statistics.

16 Bowers et al. (1970), Gujarati (1972a, 1972b, 1973), Taylor (1972), Foster (1973), Knight and Wilson (1974), Evans (1975, 1977), Holden and Peel (1975, 1977), Warren (1977), Parikh (1977), and Bewley (1979), almost all in The Economic Journal and Applied Economics.

17 Burch and Fabricant (1968: 279–80) had already stressed this point in 1968.

18 By now, the amount of literature on search and matching theory is enormous. A number of books and survey articles provide excellent overviews, such as Pissarides (2000), Mortensen and Pissarides (1999), Petrongolo and Pissarides (2001) and Rogerson et al. (2005).

19 Increasing returns to matching, for example, make analysis much more complicated since multiple equilibria could occur (Petrongolo and Pissarides 2001: 4).

20 Petrongolo and Pissarides (2001) show that the Cobb–Douglas specification of the matching function commands a fair amount of empirical support. There are, however, no compelling theoretical reasons why the matching function should be of the Cobb–Douglas form.

21 Other important contributions are Abraham and Katz (1986, 1987) and Brainard and Cutler (1993). For a survey of the aggregate shocks versus reallocation shocks debate, see Gallipoli and Pelloni (2008).