Most households in the US do not own interest bearing assets, a direct contradiction of the common representative agent assumption of New Keynesian models. Using the Survey of Consumer Finances, I document precautionary saving in the economy with a new measure, checking account balance to income ratio. I find that (a) an augmented medium-scale NK model with a precautionary saving motive can match this ratio well; (b) precautionary saving lowers the relative importance of the direct effect of monetary policy; (c) a NK model with precautionary saving relies less on nominal and real frictions; (d) the precautionary saving mechanism leads to lower inflation during economic recoveries; and (e) an extension with downward rigid wages is able to produce an asymmetric response of the economy to monetary policy in line with the recent literature. Given the central role of the mechanism linking saving to income risk for these results and the lack of clear empirical evidence for this relationship, using data on consumption, income and employment growth across 28 MSA from the Consumer Expenditure Survey, I document with an instrumental variable strategy that the consumption-income ratio is positively correlated with employment growth and increases by 0.4 percentage points in response to a one percentage point increase in employment growth. Based on this estimate, a sizable fraction of 42% of the increase in saving between 2006 and 2010 can be attributed to negative employment growth.