Rees, T. (2016). Finding the appropriate Chinese economic indicators to model copper price development [Master Thesis, Technische Universität Wien; Slovak University of Technology in Bratislava]. reposiTUm. https://doi.org/10.34726/hss.2016.37304
Copper is an extremely important metal for use in today's industrial society, but it is also known in some circles to be an indicator for economic activity regionally and globally. Additionally, Chinas commodity strength means that copper prices are largely dependent on the economic activity in China. Purchasing professionals have few resources such as those in large trading or commodity houses, therefore an analysis was run using various Chinese economic indicators against the LME copper price using commercially available Chinese economic data from the Trading Economics website. The research was set up in the following manner: 1. Questionnaire sent by email to 5 copper industry professionals asking them to rank from 1 to 10 the most likely highly correlated Chinese Economic Indicators provided with the LME cash settlement copper price. 2. Multiple regression analysis of the variables against LME cash settlement copper prices and tested for auto-correlation. 3. Time Series analysis whereby auto-regressive (AR), auto-regressive with integrated moving average (ARIMA) and generalized autoregressive conditional heteroskedasticity (GARCH) models were tested 4. Proceed with hypothesis testing to be able to determine the significance of the results A multiple regression was run using the various dependent economic variables in order to determine the best indicators as well as a time series analysis to attempt to find the best indicators as well as the best model fit. It was discovered in the regression analysis that the Chinese Leading Economic Indicator Index (P-Value: 0.0889), Chinese Manufacturing Purchasing Managers Index (P-Value: 0.0912), Chinese Exports (P-Value: 0.1440) and the USD/CNY Exchange Rate (P-Value: 0.1560) demonstrate the lowest P values (which is the probability that the data we have seen are purely based on chance) with the lowest autocorrelation and thus the best dependent variables against the LME copper prices.
