Trend stationary of inflation rates: evidence from LM unit root testing with a long span of historical data

Abstract

The article applies the LM univariate unit root test recently developed by Lee and Strazicich (2003, 2004) to re-examine the validity of trend stationary in the inflation rates of 11 OECD and Asian countries using a longer span of historical data. Our empirical findings are favourable to the trend stationary of the inflation rates when we control the structural breaks in series, and therefore they point to the absence of hyperinflation in the majority of the countries. The results indicate that shocks to inflation rates are temporary and soon converge, with the inflation rates being trend stationary. Hence, most structural breaks in the inflation rate occur around the Great Depression, World War I, World War II, and energy shock periods. For the convergence effect, we repeat the unit root tests utilized above for smaller sub-samples so as to provide a robust analysis. The outcomes show that by selecting a longer data span, we can catch more powerful convergent evidence. Overall, some policy implications are obtained in this article.

Acknowledgements

We thank an anonymous referee for helpful comments on an earlier version of the paper. Research for this article has been financially supported by the National Policy Research Center, National Sun Yat-Sen University. We would like to thank Professor Junsoo Lee for making available his GAUSS codes used to generate the results reported in this article.

Notes

¹ They also have different results, while Rose (1988) rejected the null of unit root in US inflation rates.

² The other one is the panel unit root test. For example, Drinea and Rault (2006) applied the panel LM unit root test to re-examine the empirical validity of PPP using quarterly data for nine Central and East European countries by combining panel data with structural breaks. The evidence is strongly confirmed, which means the existence of a convergence process of prices between these countries and Europe.

³ There are three structural break models considered in Perron (1989): the first model is called the ‘crash’ model A which allows for a one-time change in level; the ‘changing growth’ model B allows for a change in trend slope; and model C allows for a change in both the level and trend. We also see a detailed discussion in Perron (1989), Lee and Strazicich (2003), Fève et al. (2003) and Smyth and Inder (2004).

⁴ Hakkio and Rush (1991) point out that when using monthly or quarterly data in an empirical analysis, increasing the number of observations does not add any robustness to the results.

⁵ The ZA test selects the break point where the t-statistic that tests the unit root null is minimized. A similar test is developed by Perron (1997), who selected the break point where the absolute value of the t-statistic on the break term is maximized (Strazicich et al., 2004).

⁶ This causes size distortions leading to frequent spurious rejections in a conventional test like the ZA test; see Nunes et al., 1997 and Lee and Strazicich, 2001, 2003, 2004.

⁷ The example is a break. Ben-David and Papell (2000) found evidence of multiple breaks in per capita real GDP of the G7 countries over the past 120 years. Papell et al. (2000) showed more than one structural break in macroeconomic time series data in 16 OECD countries.

⁸ The hypotheses implied in the above endogenous break unit root tests differ from those in Perron's (1989) exogenous break unit root test, which allows for the possibility of a break under both the null and alternative hypotheses.

⁹ As summarized by Green (1996), some developed countries adopted official inflation targets in the 1990s, including Australia, Canada, New Zealand and Spain. After the European exchange rate mechanism (ERM) crisis in September 1992, the UK, Sweden and Finland announced an inflation target in the same period. Countries that have been relatively successful at achieving and maintaining low inflation include Germany, Japan and the US.

¹⁰ The Asian financial crisis in 1997 caused serious currency devaluations in almost all countries in Asia, such as Thailand, South Korea, Japan, the Philippines and Taiwan.

¹¹ It is called the Jiawu War in Chinese, because it occurred in the Chinese year of that same name. The Japanese refer to this conflict as the Japan-Qing War (Nisshin Sensou).

¹² Camarero et al. (2000) defined a temporary shock as two cases: One is when the price variable is I(1), a one-time disturbance that leads to a temporary deviation of the inflation rate from its equilibrium value. On the other hand, if the price series is stationarity, then interestingly, though the same initial deviation of the inflation rate will occur, it will be followed in future periods by an adjustment with the opposite sign and then it will converge.