以主要成份分析法探討貿易流量對匯率之影響

Abstract

This work presents a study of the relation between Guatemala's major trading goods and the exchange rate of Guatemalan quetzal (GTQ) against the US dollar (USD). Theoretical and empirical literature suggests that the exchange rate is closely linked to the import and export activity, especially in developing countries like Guatemala. An analysis of the relation between Guatemala's imports and exports and the exchange rate can indicate which goods are the most relevant to the exchange rate and Guatemala's economy in general, as it heavily depends on it. A mathematical model that predicts the values of the exchange rate based on the imports and exports of these goods could help in not only stocks investments but also determine which production sector should be the focus of attention of Guatemala's government. To achieve these goals, Principal Component Analysis (PCA) is performed on the data provided by both the National Bank of Guatemala (Banguat) and Guatemala's Stock Exchange (BVNSA). This analysis allows us to identify the desired major trading goods and also to identify the degree of importance of each one of them. This information also allows reducing the number of goods from hundreds to around twenty-five, while conserving around 90\% of the data. As a consequence, is possible to generate a linear modeling based on the principal components, to predict the exchange rate based on these goods. Further, is presented how the linear modeling can predict the exchange rate with several examples, and finalize this work by indicating how this study presents the theoretical foundations for a future research intending to create a neuronal network (machine learning) that is capable of performing supervised learning to predict the exchange rate dynamically.