3.2 Recent empirical studies
This section considers international and New Zealand specific studies that attempt to estimate the NAIRU. The methodology of this study is similar to that used in cross country studies by Laubach (2001) and Richardson et al (Richardson et al 2000). This section briefly outlines the methodology used in these studies and key differences with this study, and then discusses a number of previous New Zealand specific studies that investigated the NAIRU.
The Richardson et al (2000) paper includes a detailed review of NAIRU estimation methods and concludes that the reduced-form Phillips curve approach, utilising the Kalman filter, is the most promising framework for providing up to date estimates of the NAIRU. This general approach is applied across 21 OECD member countries, including New Zealand, for the period 1980 to the start of 1999.
The two reduced form equations used by Richardson et al (2000) are similar to those set out in section 3.1.4. As the OECD were attempting to measure a medium-term NAIRU concept they controlled for short-term supply shocks, with short-term supply shocks defined as those that would be expected to revert to zero over one or two years. The shocks the OECD selected were real detrended import prices and the change in real oil prices, as these were found to be statistically significant for a large number of countries.
Overall the methodology in this study is very similar to that used by the Richardson et al (2000). A key difference is that they somewhat arbitrarily chose their signal-to-noise ratio, whereas this paper uses Stock and Watson’s (1998) procedure to select the signal-to-noise ratio. Other key differences include our use of several model specifications and measures of inflation, the use of a survey measure of inflation expectations, and our use of quarterly rather than six monthly data.
Laubach (2001) follows a basic methodology that is similar to that used in the Richardson et al (2000) study but considers the impact of allowing drift in the specification of the NAIRU and the impact of assuming a stationary process for the unemployment gap. The study estimates the NAIRU for the G7, excluding Japan and Australia over the period 1971 to 1998. In the study Laubach found that including the unemployment gap as a stationary process considerably improved the precision of the estimates. The short-term shocks Laubach (2001) controlled for were the nominal exchange rate and commodity prices, and the two measures of inflation considered are the all-items CPI and the GDP deflator.
Laubach (2001) found evidence supporting the presence of a NAIRU and reasonably precise error bands for the United States and to a lesser extent Canada. However, Laubach (2001) felt that the European data were too imprecise to be informative about the existence, let alone the level of the NAIRU.
The most similar New Zealand specific study reviewed is Eaqub and Ward (2001). They estimate constant and time-varying NAIRU series for the period 1977-1999 using ordinary least squares, recursive least squares, and the Kalman filter. They found the Kalman filter reduced-form estimates of a time varying NAIRU to be the most satisfactory.
Eaqub and Ward (2001) considered one measure of inflation, the private consumption deflator, and controlled for one supply shock, real non-oil import prices. It is surprising that oil prices are not included as a shock given the significant impact on the New Zealand economy over that time period. They found that lagged inflation, real non-oil import prices, and the unemployment gap to all be significant at explaining inflation.
Chapple (1995) used a number of techniques to examine the drivers of structural unemployment and concluded that rising unemployment from the mid-1980s to the mid-1990s was due to a failure of aggregate demand to expand at a sufficient rate to absorb increasing potential output rather than a rising NAIRU.
