Abstract
This thesis develops a structural vector autoregression model of the global crude market, motivating and extending the analysis of \cite{kilian2009not} with a novel indicator of real non-OPEC extraction cost as an exogenous conditioning variable. The cointegrating relationship between real Brent prices and non-OPEC extraction costs alleviates the unit-root challenges associated with crude price modelling. Contrary to previous empirical work, I find a significant role for supply side factors in the medium and long run. The evidence from impulse response functions, historical decompositions, and forecast performance suggest that the extraction cost variable is necessary for consistent estimation of structural models of the global crude market. I make two contributions: The proposed model, using freely available and regularly updated data, can be used directly for forecasting and analysis. In addition, I contribute to the theoretical literature on oil price determinants by providing an important set of stylized facts. These form a base from which more refined theories on the crude market can be derived.