Hedonic regression model for coffee futures: an analysis of effect of weather, exchange rate, past price and spot price

Abstract

Objectives

The first objective of the thesis is determining the form of time series data that would be suitable for conducting empirical research on coffee futures. The second objective is to conduct a hedonic regression model and to test the observed data for multiple linear regression. Finally, this paper would like to understand the relative importance of economic factors affecting coffee futures price.

Summary

Unit root test is conducted for 228 samples of Coffee “C” Futures and the independent variables: precipitation in Brazil and in Colombia, exchange rates USD/BRL and USD/COD, spot coffee price and coffee futures price of the previous month from January 1994 to December 2012. The Augmented Dicker-Fuller test suggests the first differencing method to remove unit root in time series data. Several tests for ensuring multiple regression assumptions are performed. The input variables could be said to be valid for conducting hedonic regression model.

Conclusions

The hedonic regression model reveals that 96% of coffee futures price could be explained by precipitation in Brazil and in Colombia, exchange rates USD/BRL and USD/COD, spot coffee price, coffee futures price of the previous month and dummy variable shock. Only spot price is of great significance at 95% confidence, while the others are not. Spot price also has the highest coefficient, while coefficients of other independent variables are relatively low.