Is the financial sector Luxembourg’s engine of growth?

This paper measures the links between financial services and other production sectors in the Luxembourg economy. The focus is on propagation within the country, without considering growth abroad. Among the 29 sectors in the annual input-output tables, financial intermediation is one of only four “key sectors” that the Leontief inverse identifies as featuring above-average forward and backward linkages to other production sectors. The links from financial services to other sectors became even stronger from 1995 to 2009. At quarterly frequency, Granger causality tests find that no sector leads financial services, although they also fail to find evidence that financial services lead other sectors. Lack of evidence may be attributed to quarter-on-quarter growth rates deviating from the normal distribution, with “fat tails” possibly reflecting volatility clustering. We therefore estimate univariate ARIMAGARCH models to identify normally distributed innovations in each sector. Comparing these growth innovations at different leads or lags, Cheung-Ng tests find little evidence of crosssector causality-in-mean or causality-in-variance. This time, lack of evidence could reflect time variation in cross-sector correlations, which is confirmed by the Engle-Sheppard test.
Estimated dynamic conditional correlations vary significantly through time, with cross-sector correlations surging during the financial crisis. Dynamic correlations are used to decompose the overall volatility of Luxembourg’s macroeconomic “portfolio” of different production sectors, which responds mostly to changes in cross-sector correlations. In conclusion, the financial sector appears to be strongly linked with the rest of the economy, benefiting from growth in other sectors, which it amplifies and propagates to the whole economy.
JEL classifications: C22; C51; C52; E0
Keywords: Output growth, GARCH models, Dynamic conditional correlations