Hysteresis in East Asian unemployment

Abstract

High-performing Asian economies are quite distinctive with efficient, flexible and responsive labour markets. Comparing the persistence of unemployment in East Asian economies to that in Western countries is difficult due to the data's short time spans and possible structural breaks. This article employs advanced unit root tests to deal with these problems and fails to reject ‘hysteresis’ in the unemployment rates of the high-performing Asian countries, even after taking into account structural change. An alternative explanation of different productivity growth for the hysteresis phenomenon of the Asian countries is proposed herein.

Notes

¹ See, for instance, Brunello (1990), Neudorfer et al. (1990), Mitchell (1993), Jaeger and Parkinson (1994), Røed (1996), and so on.

² There are no strong unions in these countries except for Japan. The governments of Singapore and Taiwan have suppressed independent unions, while Hong Kong and Japan do not intervene in labour relations. However, most of Japan's unions are company-based and do not extract economic benefits for their members at the expense of economic growth.

³ Related works can be seen in Kim (1990), Ceglowski (1996), Ito et al. (1999) and Lee et al. (2002). However, some authors think that the effect of different productivity growths on the relative price can be modelled by time trend and thus interpret that Balassa–Samuelson effect implies the real exchange rate is stationary with a deterministic trend. See, for example, Obstfeld (1993), Drine and Rault (2003) and Peel and Venetis (2003).

⁴ Perron (1989) argues that the unit root hypothesis is hard to reject by the conventional ADF test if a series is comprised of a stationary fluctuation around a trend function, which contains a one-time break.

⁵ Wu (2003) finds persistence in China's unemployment rate during the period of 1978 to 1998. However, the data series crosses the period of China being a Communist nation and China's openness to markets. The definition of unemployment rate has, therefore, fundamentally changed during the sample period.

⁶ Some detailed descriptions about labour markets in Hong Kong, Indonesia, South Korea, Malaysia, the Philippines, Singapore, Taiwan and Thailand can be found in Manning and Fong (1990) and Manning (1999).

⁷ The lag order of the ADF regression model is selected based on the recursive t-statistic procedure suggested by Ng and Perron (1995) with an upper bound of on p _n .

⁸ We begin with the two-break LM test. If both breaks are not significant at the same time, we then employ the one-break LM unit root test for the investigated unemployment rate. If no break is significant in the one-break LM unit root test, then, we further employ the no-break LM unit root test of Schmidt and Phillips (1992).

⁹ Other discussions about cross-correlation in panel data unit root tests can be referred to Chang (2004), Smith et al. (2004) and Pesaran (2007).

¹⁰ Likewise, Baddeley et al. (1998) suggest that the apparent stochastic nonstationarity in British unemployment over 1965 to 1995 is, in fact, due to an upward structural shift in the mean rate, and Kratena (2000) concludes that the persistence of Austrian unemployment in 1964–1996 may arise from a stepwise stationary unemployment rate.