The network connectedness of oil shocks, currencies and commodities: The adaptive t-lasso approach

Absztrakt (kivonat)

Fluctuations in the price of oil have been shown to be major movers of exchange rates and other commodity prices. It is therefore crucial to central banks to use models that account for those shocks to oil that can strongly in uence the exchange rate of a country. It is equally important for a developing country that is dependent on commodities to experience relatively stable and predictable prices. Motivated by such issues, in this study I explore the directed connectedness between different types of oil shocks and a wide variety of currency volatilities, while also controlling for the effects of commodity price movements and changes in economic policy uncertainty. I do so by constructing connectedness tables based on VAR forecast error variance decompositions. I propose an extension of the adaptive lasso that is useful in a number of modelling situations, as it is able to shrink and select the important coeffcients of the VAR and simultaneously account for possible fat tails in the distribution of the VAR model errors. Simulation results highlight the excellent properties of the proposed adaptive t-lasso estimator. Risk and demand shocks to oil dramatically affect exchange rate and commodity volatility connectedness, while supply shocks have virtually no effect. Economic policy uncertainty affects the connectedness dynamics only to a minor degree. US headline in ation seems to be a major driver of oil-related connectedness, raising evidence that the monetary cycle in uences connectedness dynamics.