National culture and international business cycle co-movements

ABSTRACT

We investigate the relationship between national culture and international business cycle co-movements. We first characterize international business cycle co-movements with a dynamic latent factor model that decomposes 101 countries’ real GDP growth rates into world, regional and country-specific factors and estimate the proportion of real GDP growth rate variance explained by the world, regional and country factors for each country. We then regress these proportions on each of the five dimensions of Hofstede’s national culture with a set of control variables. We find that four dimensions (except for power distance) are significantly related to business cycle co-movements, and they have different impacts. In countries with masculinity, uncertainty avoidance and long-term orientation preferences, the global factor explains a larger real GDP growth rate variance. Individualism weakens a country’s business cycle co-movement with the global business cycle. Countries with similar cultural values have similar economic structures and similar exposures to global shocks, which lead to business cycle co-movements. Our results are robust after addressing the issue of endogeneity and subsample analyses.

KEYWORDS:

• Business cycle co-movement
• national culture
• dynamic latent factor model
• real GDP growth rate

JEL CLASSIFICATION:

• C32
• E32
• F43

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Geert Hofstede proposed the sixth dimension of national culture: restrained and indulgent. However, the sixth dimension has not received much attention. In this way, we also consider the five dimensions in Hofstede national culture following the most related studies.

² We further include the interactive fixed effects (Year*Region) in Equation (12)(12) ${\theta }_{i,t}^w=\alpha +\beta n{c}_i+\gamma control{s}_{i,t}+yea{r}_t+regio{n}_i+{\in }_{i,t},$(12) -(14(14) ${\theta }_{i,t}^c=\alpha +\beta n{c}_i+\gamma control{s}_{i,t}+yea{r}_t+regio{n}_i+{\in }_{i,t},$(14) ) and find similar results.(Bai 2009) The results are available upon request.

³ The reasons why these instrument variables are utilized can be found in (Nash and Patel 2019; Boubakri et al. 2021) and.(Berger et al. 2021).

Additional information

Funding

This paper is financially supported by the National Natural Science Foundation of China (grant number 71871234), the Natural Science Foundation of Shandong Province, China (grant number ZR2020QG034), the Social Science Planning Project of Shandong Province, China (grant number 22DJJJ14) and the Young Scholar Future Project of Shandong University