Empirical implementation of the buffer stock money (DSM) notion tends to concentrate either on the 'shock absorber aspects or the 'spillover' ('disequilibrium money') aspects but rarely combines both. Moreover, a potential buffer role for non-money assets is usually precluded without explicit empirical testing. This paper examines the role of financial buffers in an ex ante sedtoral model of expenditure and portfolio behaviour incorporating both the shock absorber and spillover aspects in terms of cross-equation parameter restrictions. These are tested for a range of different assets and liabilities using quarterly data for the UK personal sector.