We use wavelets to decompose the volatility (standard deviation) of in
traday (S&P500) return data across scales. We show that when investiga
ting two-point correlation functions of the volatility logarithms acro
ss different time scales, one reveals the existence of a causal inform
ation cascade from large scales (i.e. small frequencies) to fine scale
s. We quantify and visualize the information Aux across scales. We pro
vide a possible interpretation of our findings in terms of market dyna
mics.