Predicting gold and silver spot prices in Turkey

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, İktisadi ve İdari Bilimler Fakültesi, İktisat Bölümü, Türkiye

Tezin Onay Tarihi: 2013

Öğrenci: DUYGU DEVECİ

Danışman: DİLEM YILDIRIM KASAP

Özet:

The major aim of this study is to determine the best models describing the price movements of gold and silver and generate further reliable forecasts using appropriate macroeconomic and financial indicators. In that sense, modeling and forecasting analysis is conducted under the framework of a benchmark linear model-ARIMAX and a probit model using monthly data for the period between January 2002 and November 2012. To determine the best models, a variable selection procedure is implemented through a search algorithm aimed at minimizing Akaike Information Criteria for the ARIMAX and probit models. Then, the selected ARIMAX model and the probit models are adopted to predict the directional change in gold and silver prices. The predictive power of the models is evaluated based on two goodness-of-fit measures, namely direction-of-change error and root mean square error. Model performances are further assessed by a validation exercise through a recursive one step ahead forecasting for the twelve months starting from October 2008, financial turmoil period. The results for gold suggest that linear model outperforms the probit models in predicting the negative growth of gold prices whereas the probit models provide timely predictions for the positive growth of gold prices. As regards to silver prices, it is pointed out that the best probit model is superior to the ARIMAX model considering the overall predictive performance.