Comparative statics effects independent of the utility function. When do we act the same way under risk?