Gerald Prante on Means-Testing
By Bryan Caplan
Saw your post with regards to means-testing Social Security and
Medicare. Such a system would likely be tied to income in retirement. A
lot of retirement income of the “retired rich” is based on how their
investments performed. In other words, it is to some degree a selection
bias of the winners of the stock market (in addition to those with high
labor income and high savings rate as has been pointed out by critics).
So the question then becomes to a large extent the economic effects
of an increased tax on the return of stock market winners. Part of this
retirement winnings comes from inframarginal investment returns that
have no deadweight loss. The rest would be reflected mostly in the risk
portion of the return to investment as opposed to the normal part of
the return to saving (i.e., the time value of money return). Government
reducing the net payout from the risk return has no effect if they
decrease the loss portion of the risk return as well. (For example,
capital gains taxes with no loss limit would simply be equivalent to
government being a “partner” in the investment.) So if government is
largely redistributing from stock market winners to stock market losers,
then it doesn’t really change much economic activity as the expected
return stays the same.
The behavioral assumption you raised is probably the best critique, and I’ve thought about this issue for a while now and from an economic
perspective, I’m convinced that means-testing has a significantly
smaller DWL than present-day marginal tax hikes.
In a follow-up email, Gerald adds:
The behavioral critique you raised is the number one argument against
the implicit means-testing tax argument. Thaler and Sunstein even argue
that government needs to mandate employers make 401k participation yes
by default because workers otherwise don’t save enough (despite the tax
advantages of 401k savings).