The study looked at how stock market volatility tends to cluster together over time, even on a monthly basis. They used a special statistical model to analyze this pattern and found that volatility clustering is not just a thing in short-term data, but also in longer-term monthly data. This means that periods of high volatility are likely to be followed by more high volatility, and the same goes for low volatility periods. By using this model, they were able to make more accurate predictions about future volatility levels.