Back
Year
2024
Tech & Technique
Python, pandas, statsmodels, ARIMA, ADF Testing, OLS, matplotlib, Jupyter Notebook
Description
A rigorous time series analysis project testing whether Purchasing Power Parity holds between Switzerland and
Singapore, using 26 years of monthly macroeconomic data.
After stationarity testing, OLS-based PPP testing, and Box-Jenkins ARIMA model selection, found that neither absolute nor relative PPP holds — and that ARIMA(0,1,1) best describes the real exchange rate path.
What makes it stand out:
After stationarity testing, OLS-based PPP testing, and Box-Jenkins ARIMA model selection, found that neither absolute nor relative PPP holds — and that ARIMA(0,1,1) best describes the real exchange rate path.
What makes it stand out:
- Every modelling decision grounded in statistical testing — ADF, Ljung-Box, AIC/BIC comparison
- Demonstrates comfort with economic theory, time series mechanics, and the limits of models
- The finding that PPP does not hold is analytically interesting in itself
My Role
Sole Analyst
- Collected and processed 26 years of monthly macroeconomic data
- Conducted ADF stationarity tests and OLS-based PPP hypothesis testing
- Applied Box-Jenkins methodology for ARIMA model selection (AIC/BIC comparison)
- Validated residuals with Ljung-Box diagnostics; produced forecast with confidence intervals
scroll for screenshots
