Structural Change in the Austrian Inflation Process

Abstract

More than forty years ago Phillips has triggered a long lasting discussion about the relationship between money wage changes, price inflation and unemployment. Although the first investigation into these relationships was only empirical in nature, the striking findings have given rise to numerous attempts at theoretical foundations for the seeming empirical regularity. However, with the advent of the seventies such empirical regularities vanished accompanied with explanations why it had to do so. The Phillips curve was found to be unstable and, in view of a renewed emphasis on the role of inflation expectations, a distinction was made between a short run and a long run relationship, with the latter actually implying the non-existence of such a curve. A new concept had to be brought into discussion, the natural rate of unemployment. To make it a relevant tool in policy discussions about trade-offs between inflation and unemployment as well as pointing out ineffectiveness of policies, this concept was related to the change in inflation, but not to the change in the price level, and termed the NAIRU. The rate of unemployment consistent with an unchanging inflation rate is an analytical concept that can help to understand the causes of inflation. As an empirical basis for predicting changes in the inflation rate as well as a policy guideline its usefulness is questionable as this concept suffers from being unobservable. It is therefore subject to considerable uncertainty and controversy and several attempts have been and are still made to render plausible and well researched empirical estimates of this rate. A prominent example is given by the „triangle model“ of inflation (Gordon 1997) which makes inflation to depend on three basic sets of factors (demand, supply, and inertia). The importance given to supply factors which can create positive correlation between inflation and unemployment distinguishes this model from the traditional Phillips curve specification. Apart from the search for determinants explaining the NAIRU the issue of its variability has recently become a major research objective (e.g. Gordon 1997, Staiger — Stock — Watson 1997, Clark and Laxton 1997). Variable parameter models estimated by a variety of techniques have been applied. Almost all empirical estimates are obtained under the maintained assumption of the existence of a long run rate, i.e. have imposed a long run vertical Phillips curve by assuming the sum of lagged price effects to be unity. This assumption, being at the core of the expectations-augmented Phillips curve of Friedman and Phelps, has also been used in other, New Keynesian, models. A common specification of several of the New Keynesian models (Taylor 1979, Calvo 1983, Rotemberg 1982) has been presented by Roberts (1995) under limited information assumptions. The crucial assumption for this exercise related to the estimate to be used for inflation expectations i.e. actual future values or survey based ones. This contrasts with the full information approach of rational expectations which requires a fully fledged correctly specified model, with the associated risk of misspecification of even only a part of it. A sufficiently robust Phillips curve could be useful for economic policy by exploiting the short run trade off and targeting the NAIRU. For risk neutral policy makers it will then depend on the curvature of the Phillips curve whether they should experiment with the unemployment rate. With a convex Phillips curve they will be averse to such experiments, but they would engage in them in the case of concavity. Thus, the question of linearity of the Phillips relation (in which case policy makers would be indifferent) as opposed to a non-linear curve and, in addition, the type of non-linearity, have been issues addressed in recent empirical work. Those non-linearities implying asymmetries in the relation between unemployment and inflation may follow from several economic theories. In the capacity constraints model firms are assumed to have difficulties increasing their production capacities in the short run which has stronger effects on price increases the more firms are affected by this inflexibility. The Phillips curve will be convex in this case as the sacrifice ratio will be higher in periods of excess supply than in periods of excess demand. The misperception model exploits the alleged inability to distinguish aggregate from relative price changes. Higher volatility of aggregate inflation induces people to believe that price changes are less due to relative price changes and will therefore respond with smaller output and employment changes. The slope of the Phillips curve will thus rise with volatility. Considerations of price adjustment cost may also imply a possibly convex Phillips curve as price adjustments increase in frequency and size during accelerated inflation, leading to stronger effects on inflation than on output. Downward nominal wage rigidity too may account for asymmetry because allocation inefficiencies are easier eradicated when inflation rates are high (with hardly a need for cuts in nominal wages) as compared to low inflation (when the rigidity prevents appropriate real wage adjustments). A concave short run Phillips curve may result in economies with dominant oligopolistic or monopolistically competitive markets. Price rises may be carried out slower and price reductions faster in order to keep out potential new competitors.