I was thinking of using Geometric Brownian Motion to forecast future prices of timber (say one variable, the stumpage price of sawtimber).

I tested the time series with Augmented Dickey-Fuller test and found the data series as non-stationary which means the series follows a random walk. Then, I went on to use it for price forecast.

However, my professor comes and says we cannot use GBM to forecast future prices that has long horizon.

In my case, I wanted to use quarterly price into 10 years or 15 years into the future. I know GBM is a good model for stock prices for short periods, but is 10 or 15 years too far considering the confidence limits and probable volatility?

  • $\begingroup$ I'd read up on Schiller's CAPE ratio. $\endgroup$ – Chan-Ho Suh Feb 11 '15 at 5:38
  • $\begingroup$ The reason why GBM would be better short-term than not long-term is not obvious to me. Did your professor justify his claim by something? Besides, GBM models asset prices (not returns), were you expecting these to be stationary? $\endgroup$ – SRKX Feb 11 '15 at 6:00
  • $\begingroup$ I suspect the typical price evolution of commodities to be different from stock prices. I do know nothing about timber, but do prices have some seasonality? What might be main drivers of the unit root? Demand growth, inflation etc.? That said, I agree with SRKX. Maybe your professor meant that there is some specific mid to long term structure in timber prices which can be exploited by forecasting? $\endgroup$ – Marco Breitig Feb 11 '15 at 7:54
  • $\begingroup$ In how far does GBM help you in forecasting? As a stochastic model - yes. As a forecast: no. $\endgroup$ – Ric Feb 11 '15 at 8:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.