Rigobon heteroskedasticity-based estimator

[philipecon]

In their 2004 paper on the effect of monetary policy on stock prices, Rigobon and Sack use an IV approach, exploiting the conditional heteroskedasticity in interest rates present on policy vs non policy dates. If anyone previously worked on this paper, could you please suggest how to derive this estimator and explain the IV procedure in this specific case? Thanks