Testing for cointegration with threshold effect between unemployment and stock prices

ABSTRACT

Under the dominant role of a belief function, Farmer argues that the stock market is the Granger cause of the unemployment rate, which implies that the natural rate hypothesis is an outdated idea. This article provides some new empirical evidence supporting this view using threshold cointegration and asymmetric error correction models. The results show that these models can assess asymmetric dynamics between unemployment and the stock market. Moreover, regime switches of the momentum threshold autoregressive adjustment specification are highly consistent with recessions in the US economy during the last 60 years.

KEYWORDS:

• Unemployment
• the stock market
• animal spirits

JEL CLASSIFICATION:

• E30
• E32

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ By applying several techniques for German data set, Fritsche and Pierdzioch (2017) conclude that there is a causal relationship running from stock market to unemployment in the medium to long run. To the best of my knowledge, there is no other formal attempt to investigate the validity of this methodology with a different data set other than Farmer’s.

² I am grateful to Professor Roger E. A. Farmer for making the data set publicly available, which can be downloaded from http://www.rogerfarmer.com/NewWeb/Spreadsheets/Data%20for%20JEDC%20Paper.xlsx.

³ I also examined the existence of threshold cointegration by employing the Seo and Hansen (2002) test, which is not subjected to noise but does not allow M-TAR specification. The null of linear cointegration against the threshold alternative is rejected with a p-value of 0.037 using the parametric bootstrap method with 50 000 replications, for the value of 0.055.

⁴ Indicates problems in the housing sector ended by the Kennedy tax cuts.

⁵ Related to the soft landing action of the Federal Reserve (FED) to cool down the economy after reaching certain growth limits.

⁶ Indicates problems related with the 2001 recession ended by the leap in defence expenditures, which were the largest in a single quarter since the Korean war.

⁷ The equation for the stock market has no statistically significant coefficients on error correction terms; therefore, it is not reported in this article.