How properly apply the bekk model and test for spillover effect?

Hi everybody,

I'm trying to replicate the paper "Investigating the Co-Volatility Spillover Effects between Cryptocurrencies and Currencies at Different Natures of Risk Events" by Shu-Han Hsu in Matlab. The paper main aim is to find if there is a spillover effect between cryptocurrency market and the currency market by applying the diagonal BEKK Model and so to determine if and where the crypto-market is a safe heven.

I used bekk function in the Kevin Sheppard's ToolBox but the results seemeds inconsistent to me even though I used the same time horizon of the paper.

Could you please help me try to figure it out?

My main questions are:

1. Do I have to apply the model separately for each cryptocurrency wrt the different foreign exchange rates?
2. How can I test and establish if there is spillover or not?

Thank you in advance!

Answer 1

[Vijeta]

Hi Jacopo,

In terms of your specific questions, applying the diagonal BEKK model separately for each cryptocurrency with respect to each foreign exchange rate is consistent with the methodology used in the paper. To test for spillover effects, you can use a combination of statistical tests and graphical analysis to identify any patterns in the data that suggest the presence of spillover effects. I believe it is important to carefully follow the methodology described in the paper and conduct a thorough analysis to ensure reliability and accuracy.

Testing for spillover effects in MATLAB typically involves performing statistical tests on the time series data.

Here are some steps you could take to test for spillover effects:

1. Collect and preprocess the data: Collect the time series data for the cryptocurrency and currency markets, and preprocess the data to ensure it is stationary (i.e., has constant mean and variance over time). You may also want to consider normalizing the data to facilitate comparisons between the different markets.
2. Estimate the diagonal BEKK model: Use the BEKK model or its diagonal variant to estimate the conditional volatility of the cryptocurrency and currency returns. The diagonal BEKK model is particularly useful when dealing with large numbers of assets, such as multiple cryptocurrencies and currency pairs. MATLAB provides several toolboxes that can be used for this, such as the Econometrics Toolbox or the Financial Toolbox.
3. Perform statistical tests: Use the appropriate statistical tests to test for spillover effects between the cryptocurrency and currency markets. Some common tests include the Granger causality test, vector autoregression (VAR) analysis, variance decomposition analysis, and co-integration analysis (as mentioned in my previous response). MATLAB provides several built-in functions for performing these tests, such as the “granger_causality_test” function, the “varm” function, and the “vgxvarx” function.
4. Interpret the results: After performing the statistical tests, interpret the results in the context of the research question. Look for evidence of spillover effects and identify the direction and magnitude of the spillovers, if any. It is important to carefully interpret the results and consider the limitations and assumptions of the statistical tests used.

Overall, testing for spillover effects in MATLAB involves a combination of data preprocessing, model estimation, and statistical testing. The specific approach and tools used will depend on the research question and the nature of the data.