Taxation and Liquidity: Evidence from Retirement Savings

Elizabeth Ann Chorvat

Research output: Working paper


This paper tests the response of a cross section of U.S. households to a reduction in the tax cost of holding liquid assets by the 2003 Jobs and Growth Tax Relief Reconciliation Act, whether that response was tax efficient, and the distribution of that response. Empirical results based on regression analyses of Survey of Consumer Finances data between 1998 and 2010 suggest that lower, middle, and high-income households responded to the enactment of the dividend preference in a tax-efficient manner, increasing allocations to liquid accounts and away from tax-preferred retirement accounts. Notwithstanding the conventional wisdom that behavioral responses to changes in the taxation of investments occur predominantly among the wealthy, the paper finds that the largest behavioral response to the 2003 dividend preference appears to have been among those households in the highest and lowest income groups, with the largest elasticity of response among the lowest income households. If household income is an important determinant to the value of liquidity, we might well understand that those with the highest need for liquidity might have the largest response to a reduction in the cost of that liquidity. Curiously, while middle-income households responded to the lower cost of liquidity in a tax-efficient manner, theirs was a distinctly smaller response, suggesting the possibility of a nonlinear relationship between taxation and liquidity response.
Original languageEnglish (US)
PublisherUniversity of Chicago
StatePublished - 2013

Publication series

NamePublic Law Working Paper


  • dividends
  • liquidity
  • retirement accounts
  • investment income
  • capital gains
  • savings behavior
  • portfolio allocation
  • efficiency
  • behavioral response
  • welfare
  • retirement
  • elasticity


Dive into the research topics of 'Taxation and Liquidity: Evidence from Retirement Savings'. Together they form a unique fingerprint.

Cite this