Business cycle asymmetry: evidence from Japan’s decade-long deflation

ABSTRACT

This paper empirically examines the asymmetric behaviour of macroeconomic time series under different economic regimes, focusing especially on whether the economy is in a deflationary or non-deflationary regime, using a threshold vector autoregression model (TVAR). The results provide evidence in favour of asymmetric spillover effects from demand shocks on economic variables such as wages and consumption. In contrast to the non-deflationary period, during the deflationary period demand shocks are not clearly transmitted to the economy overall due mainly to the rigidity of nominal wages. The findings highlight that the state of the economy also matters when we consider the nature of the business cycle.

KEYWORDS:

• Business cycle asymmetry
• deflation
• Japanese economy
• threshold vector
• autoregression model

JEL CLASSIFICATION:

• E32
• C22

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ The business cycle characteristics in the non-deflationary period here are essentially in line with those reported in Urasawa (2018), which examined the stylized facts of Japanese business cycles. The stylized facts observed in Urasawa (2018) can be summarized as follows: Consumption and investment are contemporaneously procyclical, while the unemployment rate is countercyclical and strongly correlated with output; furthermore, it lags the economy. Employment is procyclical and lags fluctuations in output, reflecting the traditional pattern of labour input adjustment in Japan whereby in the early stages of an upswing or downswing, it is not the number of employees but hours worked that are adjusted. Wages and inflation are basically procyclical and lag fluctuations in output. Finally, stock prices are procyclical with a lead, while the effective exchange rate is counter-cyclical with a lag, that is, there is a negative correlation between an appreciation of the yen and output.

² To examine the propagation mechanisms on the supply side of the economy through wages and the number of workers, the unemployment rate is used an alternative variable. However, no difference can be observed between two regimes.

³ To avoid an implausible number of regime switches over time, the average of four subsequent quarters of the series in log differences is used.

⁴ To examine the propagation mechanisms on the real side of the economy, real wages and real private consumption are used as alternative variables. However, the results remain essentially unchanged.