Futures prices, trade and domestic supply of agricultural commodities

Commodity markets display substantial volatility both in prices and in the quantities traded. This has led to the development of different instruments designed to address this volatility. Processors and traders, who are actively involved in the international market, participate in these commodity markets using cross-hedging strategies by their export and domestic supply decisions. Spot and future prices, as well as the cross-hedging strategies, affect export and the domestic supply decisions. Understanding this complex interaction calls for further and newer insights and this research contributes to this. The primary objective of Chapter 1 of this thesis is to develop a model which explains the export and domestic supply decisions when traders, producers and speculators participate in a futures market for a primary commodity, which can be stored and for which future markets operate. As a result, exports and domestic supply are affected by the prices of the primary product, and jointly by the prices in the external and domestic market. Chapter 2 provides the historical, political and economic context of the Argentine economy and the agricultural sector, specifically on the three agricultural commodities used in the empirical part of this research. In Chapter 3, we perform a comprehensive analysis of the seasonal unit roots of monthly series of exports and domestic supply, using time series that include zero values. In the past, this technique has mostly been applied to quarterly data but never to monthly series that display periods of inactivity. The results indicate that, in general, the seasonality observed in the series analysed can be sufficiently explained by a deterministic approach. The estimation and further analysis of the supply equations derived in Chapter 1 are undertaken in Chapter 4. A comprehensive analysis of seasonal cointegration using monthly data was conducted but, in light of the results obtained in Chapter 3, only the Engle-Granger cointegration is applied. The results indicate weak cointegration relationships. This may indicate the need for improved data and/or alternative econometric techniques.