Connectedness of Financial Markets and Real Economies: Analysis of Yield Curve Factors and Commodity Markets

Absztrakt (kivonat)

Commodity markets have a special attention within the topic of financial connectedness due to their direct links to the real economy and central role among markets. Many countries are dependent on goods that are important element of their production or export activity. If the natural operation of these markets is changed abruptly then the stability of economies can be in danger. My paper investigated the connectedness between commodities and yield curve factors as level, slope and curvature by U.S., United Kingdom, France, Germany, Norway, Japan and Australia from 2000 to 2020. The commodity group consisted of crude oil, heating oil, wheat, corn, copper,aluminum, silver and gold. An emphasis was put on the nature of connections between financial variables whether those linear or non-linear. My finding is that both types of relations react intensively to the shocks. Commodities work as a catalyst in the system and behave sensitively to the raised information flow in the system. Those exploited more connection during shocks than the other three groups. Furthermore, the nature of their outgoing links is mainly non-linear but those internal connections mainly are dominated by linear edges. Additionally, during hard times the magnitude of the number of non-linear links increase. By different different factors the different crises can have a different effect due to the fact that the Great Financial Crisis caused the increase in the proportion of non-linear links until the European Sovereign Debt Crisis favored the number of linear links. It means that one has to consider that different shocks can cause the increase in the proportion various types of links by different financial assets. To obtain these results, the three factors yield curve decomposition was applied. The pairwise linear and non-linear connections are identified by Granger causality and transfer entropy, respectively. Based on these results, networks are built up where one edge is equal to one significant relations between two time series. Static analysis is done for the whole, pre-crisis, crisis and post-crisis periods and it is extended by dynamic analysis using rolling window technique.