We test the key implication of the buffer stock model, namely that any revision in permanent income leads to a proportionate revision in target wealth. We use panel data on the amount of wealth held for precautionary purposes available in the 2002-2016 SHIW. Using an instrumental variable approach to overcome measurement error issues and direct estimates of the permanent component of income, we find that households indeed revise approximately one-for-one their target wealth in response to permanent income shocks. We explore heterogeneity of the response across the cash-on-hand distribution, for positive and negative shocks, and for shocks of different size. We also find that the change in the ratio of cash-on-hand to permanent income is negatively correlated with the “wealth gap”, particularly for individuals whose wealth is substantially above target.