To analyze the effectiveness of stabilization policies which includes effects on households future income it is central to account for anticipation effects on consumption. We investigate this using high-frequency spending and balance sheet data from a major Danish bank. We examine the behavior of borrowers with adjustable rate mortgages, and exploit that the bank sends a letter before the annual reset containing advance information on the expected change in mortgage payments. We find that unconstrained households respond immediately, while liquidity constrained households instead wait and respond around the time the cash-flow-arrives. The cumulative response is similar across the liquidity distribution. This is in line with a standard buffer-stock consumption model, and implies that it is less effective to target stimulus to low liquidity households when the effect on household income is partly in the future.