Applied Time Series Using Stata

I will briefly introduce Stata. First of all, you can use an older version of Stata – it does not affect the functionality. At the moment, Stata 17 is available – but I will work with Stata 13 without any issues.

This session discusses various ways to get access to Stata. You might have access to an older version of Stata, which will be perfectly fine for this introductory course. All the code has been tested on Stata 13, which I have used for many years without any problems.

What is our goal? In fact, there are different answers to this question. In time series analysis, we observe different variables, say X and Y, over time t. We can see how the values (or realisations) of X and Y change over time. These observations could lead to an understanding of causal order – put differently if X changes today – this change is followed by higher values of Y tomorrow. Establishing these sequences can help to derive causal relationships. We will explore these issues in our unit on vector autoregressions. Testing theories is another valid objective – but most practitioners are not worried about it. Hence, this course will not focus on these issues. Simulating time series is yet another area of interest – but simulations go beyond the scope of this course. Finally, forecasting is the main purpose for most practitioners. Obviously, being able to forecast time series has substantial economic benefits. Hence, this course will focus on our ability to forecast time series.

This session introduces basic concepts in time series analysis.

We explore data on food prices and try some basic forecasting methods.

This session discusses the concept of stationarity, which is essential in understanding time series models.

We simulate autoregressive processes in Stata. This is a useful exercise to understand the ideas behind data-generating processes.

We explore the Dickey-Fuller test to detect non-stationary time series.

ARIMA models are quite useful for forecasting. These models tend to be simple and stable. If you work with an older version of Stata, please use the Stata 11 dta dataset provided.

This session demonstrates the use of the autocorrelation and partial autocorrelation function to identify ARIMA processes.

We discuss adjustments for seasonality. If you use an older version of Stata, please download the Stata 11 dta dataset.

This session demonstrates the use of intervention analysis to measure the impact of policy changes on time series.

We apply an intervention study to explore the impact of lockdowns on retail sales. The results are a surprise.

This session introduces vector autoregressions (VAR), which can be used to model short-term dynamics between time series.

We apply a VAR model to capture the dynamics of the UK property market. Can we forecast house prices?

This session demonstrates how to estimate VAR models in Stata.

This session explains the stability condition for VAR models. We implement stability testing in Stata.

To illustrate the dynamics of a system, impulse response functions are useful. They can be powerful tools to summarize your findings.

We obtain stock market data from Yahoo Finance and merge datasets.

This session outlines the main ideas without going too deep into the underlying theory.

This session explains the implementation of VECM models in Stata.

This session introduces the conditional variance equation. We explore ARCH and GARCH models.

This session explains how ARCH and GARCH models can be conducted in Stata.

Structural breaks make it much harder to forecast time series. This session introduces tools that can be used to detect structural breaks.

This session conducts structural break tests in Stata.

This session provides a brief introduction to panel VARs and cointegration. These are active areas of current research.

We explore panel VAR models in Stata.

Well done! We will discuss your next steps to enjoy the Joy of Data Analysis.