That's 0.5 percentage points. Would be nice to get that metric, too, alongside percentages sometimes. Here the initial likelihood is relatively high, when the likelihood is low, percentage increases can be very misleading. Particularly if the uncertainty around the increase is high.