Abstract. This paper empirically investigates the nature of the underlying stochastic processes characterizing real oil prices using a Structural Time Series Model for the period 1960 to 2016. Based on the state space framework the STM decomposes the data into separate stochastic components by the maximum likelihood via the Kalman Filter. In contrast to the extant literature, this approach obviates the need to attain stationarity. Instead, it explicitly represents the nonstationarity properties of real oil prices through time varying structures and incorporation of structural breaks. The results establish that real oil prices are a composite of a long term trend, effect of shocks and short term fluctuations. The trend exhibits stochastic evolution and is punctuated by distinctive breaks triggered by unpredictable and significant events. Short term fluctuations are driven by transitory market influence and result in mean reverting patterns. Overall, the model captures, in a nonstructural framework, oil price movements and allows generation of forecasts based on the concrete implications of data characteristics.

Keywords: cycle, oil resources, real prices, structural time series model, trend.