Marginal Notes on Pierre-Olivier Gourinchas's Presentation of Golosov, Tsyviniski, and Werning

Golosov, Tsyvinski, and Werning, "New Dynamic Public Finance: A User's Guide"

• This is a very useful paper.
• Emmanuel Saez is even more of a genius than I had thought.
  • The Mirrlees (1971) don't-tax-the-richest result is a very local result

• The Mirrlees optimal-tax formula trades off three consierations:

  • Desired redistribution--social marginal weights...
  • Higher taxes lead to substitution effects away from effort somewhat counterbalanced by income effects towards effort. Hicksian elasticities are what matter.
  • The shape of the income distribution--weighing the distortion caused by the marginal rate at income level y against the concentration of income above y from which you raise revenue

• Golosov, Tsyviniski, and Werning

  • Two periods
  • Unobservable skill shocks
  • Aggregate shocks as well
  • Government observes earnings but not skills
• Distortions
  • Consumption-labor wedge at t=1
  • Consumption-labor wedge at t=2
  • The intertemporal wedge
• No aggregate uncertainty, no skill shocks, then...
  • Perfect smoothing over time--i.e., no taxes on capital income
  • Labor tax wedge that solves static Mirrlees problem
• With skill shocks...
  • Planner wants to encourage effort in period 2 by those who get positive skill shocks--which means it would want to find some way to reduce their savings, which means there will be taxes on capital in period 2...
  • Time consistency problems...
  • Double deviations; once you deviate from the optimal labor supply, you may well want to deviate from the optimal savings plan as well...

It is very nice to have, thanks to Golosov, Tsyviniski, Werning, and company, a model that says that you want to tax capital because you don't want the highly-skilled and productive programmers of Google retiring in their 30s because they are wealth-satiated...