Higher retirement savings might not translate into net wealth accumulation if, rather than cutting spending, individuals reduce their non-retirement savings and take on more debt. We use newly merged deposit-, credit-, and pension-account data from a large UK financial institution to examine a policy that gradually increased retirement contributions from 2% to 8% of salary between March 2018 and April 2019. For every£ 1 reduction in take-home pay due to higher employee contributions, employees cut their spending by only£ 0.34 (especially in the restaurant and leisure categories) and financed the remaining with lower deposit balances and higher credit card debt. Those with lower initial deposit balances cut their spending the most, while those with significant liquid savings draw down their deposits. Since employees cannot draw down their balances indefinitely, we use a lifecycle model–calibrated to match the observed short-term responses–to predict that long-run spending responses are larger but feature similar heterogeneity. A social planner concerned about undersaving for retirement due to heterogeneous present bias should, therefore, avoid targeting retirement interventions at high-liquidity individuals who are both less likely to cut their spending and less likely to be present-biased.