Is the Phillips Curve A Curve? Some Evidence and Implications for Australia

Abstract

The Phillips curve has generally been estimated in a linear framework. This paper investigates the possibility that the Phillips curve is indeed a curve, and shows that a convex short‐run Phillips curve may be a more accurate representation of reality than the traditionally used linear specification. The paper also discusses the policy implications of convexity in the Phillips curve. These include the need for policy to be forward looking and to act pre‐emptively. Convexity provides a strong rationale for stabilization policy, and it reinforces the need for policy makers to proceed cautiously. It also implies that deep recessions may have only a marginally greater disinflationary impact than shallower ones, unless they induce large credibility bonuses.

An Estimate of the Range of Equilibrium Rates of Unemployment for Australia

“…Crosby and Olekalns (1998) ¢nd the NAIRU increasing from 2.3 per cent in the period 1959^73 to 9.2 per cent in the period 1984^97. Debelle and Vickery (1998) also ¢nd the NAIRU was about 2 per cent in the 1960s and then increased after 1973, but by a smaller amount. According to their estimates, the NAIRU was about 6 per cent in the 1990s.…”

“…Such a pattern, called`speed limit e¡ects', is evident for Australia, in Debelle and Vickery (1998), Ng (1998), and Gruen et al (1999). The standard argument is that, as activity expands, employers will be induced to increase wages to attract new labour or to induce existing labour to work longer hours.…”

“…Those estimates are 7.6 per cent (Powell and Murphy 1997), 9.2 per cent (Crosby and Olekalns 1998), 7 per cent (Debelle and Vickery 1998) and 7 per cent (Gruen et al 1999). 23 Second, the estimates of NAIRU by Debelle and Vickery (1998) and Gruen et al (1999) can be interpreted as being strongly 21 As Debelle and Vickery (1998) point out, an implication of a convex short-run Phillips curve is that for a non-increasing rate of in£ation over a period of time when the rate of unemployment has been hit by a sequence of shocks distributed evenly around zero the average rate of unemployment has to be greater than the NAIRU. 22 Groenewold and Hagger (1998) use a di¡erent approach based on labour market £ows, due to Lilien (1982).…”

“…These two concepts, U min and NAIRU, share the property of being the lowest rate of unemployment consistent with non-increasing in£ation. 21 The estimates of the NAIRU for AUSTRALIA are those of Powell and Murphy (1997), Crosby and Olekalns (1998), Debelle and Vickery (1998), Groenewold and Hagger (1998) and Gruen et al(1999). For the Murphy Model 2, Powell and Murphy (1997) estimate a ¢xed NAIRU as 7.6 per cent (their earlier estimate based on data from 1976:1 to 1991:4 was lower, at 7.1 per cent).…”

Inflation and Capacity Constraints in Australian Manufacturing Industry

“…For example, v t = 0 implies that is time invariant. Alternatively, if v t is modelled as say a white noise series, we have a time-varying parameter model and can be estimated within a state space formulation, using the Kalman Filter procedure (Debelle & Vickery 1997;Gruen et al 1999). A potential rationale for the time-varying parameter formulation would be hysteresis in the labour market, such that any shock affecting the system has a long-lasting or permanent impact on the equilibrium position of the economy (Blanchard & Summers 1986;Laubach 2001).…”