| dc.description.abstract | Reserves are the technically-recoverable hydrocarbon volumes that can be economically produced given the current economic condition. Uncertainty about future oil prices is experienced in the present, not the future, and is a property of the current economic condition. Average or risked oil price scenarios are deterministic and may not correctly capture the impacts of oil price volatility on field-level reserves or other economic outcomes such as net present value. Stochastic price-volatility models tend to operate on the scale of days to a couple weeks, which is short compared to the life of a hydrocarbon well. The lack of a long-term stochastic model for price volatility motivates a look at how current stochastic models are made, and the creation of a model consistent with the life of hydrocarbon wells. With such a model, the dependence of reserves volume on price volatility can be assessed.
Random-walk models have been used to simulate the behavior of market prices and the uncertainty of future price changes over time, but usefulness is limited when the distribution of observable price changes is not well defined. A new density function is proposed to model returns on oil price. This density function, having a shape that depends on the coefficient of variation of the returns, is formed by the product of two Laplace distributed random variables. Although the new distribution was developed in context of West Texas Intermediate (WTI) spot price, the model was capable of modeling other markets such as Johnson and Johnson stock price.
Traditional methods of calibrating mean and variance behavior of returns for use in random-walk models has been inadequate. Models often err by failing to explicitly include or exclude intra-week behavior. In this thesis, mean and variance metrics are determined in a novel way that defines then removes day-of-the-week effects that may have cumulative bias when estimating project value measured in years or decades.
Using probabilistic decline-curve parameters from the Eagle Ford shale, reserves and profitability were estimated for a synthetic project. The average reserves volume determined under a stochastic scenario was less than the average reserves volume determined using average price as a deterministic input. This means that using average prices when estimating reserves volume does not obtain the true average reserves volume in practice. The magnitude of the difference between stochastic and deterministic price scenarios depends on the variable cost per barrel of oil. For $10/bbl variable cost, average reserves did not differ significantly between deterministic and stochastic price modeling; at $30/bbl variable cost, average reserves were 17% lower using stochastic modeling. Accounting for volatile oil prices is paramount to obtain true average reserves volume, and average-price-in does not always result in true average-reserves-volume out. | en |
