Sources of employment fluctuations in Taiwan's industries and regions

Abstract

This empirical study explores the sources of employment fluctuations in Taiwan's industries and regions over the period 1978 to 2004. The quarterly growth rates of employment in nine industries and four regions are modelled with a structural vector autoregression (VAR), and the employment shocks are measured by VAR residuals. The covariance matrix of the VAR residuals is decomposed using system estimation method that selects the parameters to make the error model close to the covariance matrix and, in turn, to estimate the relative importance of national as well as industry-specific and region-specific shocks. The empirical results show that industry-specific shocks account for the major fluctuations in industries and regions. On average, about 83.95% of an industry's cyclical variations and 56.28% of the volatility in a region may be attributed to industry-specific shocks. National shocks account for little employment volatility in industries. Only the finance and personal service industries are highly sensitive to national shocks.

Notes

¹Through 2001–2003, Taiwan's average economic growth rate was below 4%, and the unemployment rate was more than 4% higher than the 1978–2000 average.

²The structural VAR model not only contains current real variables, but also adds past variables to capture a shock's propagation among variables. Therefore, the VAR residuals can filter out the feedback effects and actually measure the shocks in business cycles. The same approach is adopted in the research of Long and Plosser (1987), Norrbin and Schlagenhauf (1988), Abowd and Card (1989), Blanchard and Katz (1992), Clark (1998), as well as Chang and Coulson (2001).

³The unemployment rate can also reflect the waste of human capital in the labour market. As far as the discouraged theorem is concerned, an unemployed worker may fail to find work and become a member of the nonlabour force. The unemployment rate excludes the so-called ‘discouraged worker’ as this would underestimate the variations in the labour market. Therefore, employment as opposed to unemployment is adopted to avoid making underestimates based on the unemployment rate.

⁴Several frequencies of data have been adopted in previous empirical research. To cite a few examples, Long and Plosser (1987) used monthly data; Norrbin and Schlagenhauf (1988) as well as Clark (1988) used quarterly data; Altonji and Ham (1985) and Bai (2004) used yearly data. But Long and Plosser (1987) claimed that monthly data may reduce the potential role of common shocks if the shocks only influence some sectors after some delay. Time aggregated to quarterly data is sometimes suggested as a way of mitigating this phenomenon.

⁵The number of employees in the construction industry has steadily declined since the second quarter of 1995.

⁶It was observed in the preceding Figs 1 and 2 that the variations in the pattern of employment appear reasonably stable within each industry and region. Therefore, this article makes the assumptions that the employment structure did not change during the survey data and that the shocks came from the same VAR model.

⁷The estimation restricts the variance of a national shock to equal one (σ _c ² = 1) because, as Clark (1998) put it, one scale restriction is necessary to identify the error model. The relative magnitudes of the responses to a national shock and the shares of variance attributable to sector-specific shocks are invariant to the specific scale restriction imposed.

⁸The definition of a business cycle is economic activity that rules out seasonal fluctuations and long-time trends. The augmented Dickey–Fuller tests reject unit roots for all the growth rates of employment in industries and regions. These results imply that the log quarterly growth rates of employment in each industry and region have no time trend.

⁹With quarterly data for the growth rate of employment, the four lags suffice to eliminate all serial correlations in the residuals. In this article, the VAR model estimated as four lags also has the smallest Akaike information criterion (AIC) and Schwartz criterion (SC) over all the lags.

¹⁰The estimated results of the structural VAR show that the import energy price index has negative effects on the trade industry and the northern region, and the exchange rate has positive effects on the manufacturing and trade industries as well as the northern region. The estimated response of industry employment to exchange rate shocks is also supported by the evidence of Hatemi-J and Irandoust (2006).

¹¹The least squares are applied to the system estimation with the weighting matrix for the system of equations set to an identity matrix. The parameter estimates achieve convergence after 16 iterations.

¹²The amendments to the Labor Standard Law have failed to explain national shocks because the personal service industry is less restricted by the policy.