Un modelo basado en agentes para explicar tendencias de largo plazo en precios de commodities

Abstract

Historically commodity prices have shown a tendency to move cyclically. However, these trends have only been identified a posteriori and have not been considered to try to obtain future trends, with the hypothesis that future prices are unpredictable. During the first years of this century, metals of industrial use went through a phase of sustained increase in their prices, reaching prices levels unprecedented in recent history. This boom generated significant changes both in the global mining industry and in the economies of the main exporting countries, including Chile. Given the relevance of the mining industry in the Chilean economy, it is vital to have methodologies that allow an estimation of the future price. Several methodologies have been used for these purposes; Being the most relevant econometric, financial and structural. None of them is capable of generating price trajectories like those historically observed in commodities. It is therefore necessary to look for new methodologies that account for these cycles. This thesis proposes to use an agent-based model to model the copper market. In this class of models individual economic agents are constructed, which have their own characteristics and autonomy to make decisions as the system advances. They are also characterized by being able to capture properties that emerge from the interaction of all elements of the system. The proposed model considers a system populated by deposits, mines, plaintiffs and a coordinating market, the price variable being an emergent property resulting from the interaction of the mentioned agents. The model also considers two types of decision, the decision of production and the decision of investment, which have consequences in the short and long term respectively. The price trajectories generated by the model indicate that the price cycles emerge as a consequence of the dynamics of the industry, mainly the delay between investment and the start of production. Another notable aspect is that price cycles emerge independently of the properties of mines and deposits; this fact is consistent with what is observed where the prices of different commodities tend to move together following a cyclical behaviour despite the differences in their extraction and concentration processes. Finally, there is a need to carry out research in this area, emphasizing aspects such as the mechanism by which prices are formed, developing variables that can adjust the model to different commodities and improving the method through which agents make decisions.