Writing in The Wall Street Journal on April 27–making another last-ditch pitch for a 20% border tax on business imports–Martin Feldstein asserts that unless corporate tax rate cuts are “offset” by tax increases on imports or payrolls then larger projected deficits would crash the stock market by raising long-term interest rates. “The markets’ current fragility,” he writes, “reflects overpriced assets–the S&P 500 price/earnings ratio is now 70% above its historical average–after a decade of excessively low long-term interest rates engineered by the Federal Reserve.”
The odd notion that the Fed could somehow depress bond yields for a decade is an irrelevant ambiguity, since the whole point of Feldstein’s story is to claim budget deficits raise bond yields and higher bond yields threaten “overpriced” stocks.
In a recent blog, I found no evidence to support the dogma that bond yields rise and fall with rising or falling budget deficits (actual or projected). Wall Street Journal columnist Greg Ip opines that “interest rates haven’t responded to deficits lately because private investment has been so lackluster.” But that excuse makes interest rates dependent on private investment, not deficits, and leaves us tangled in circular illogic. If interest rates depend on private investment and deficits “crowd out private investment,” then interest rates could never respond to deficits because private investment would always be lackluster when deficits were large (which would also make deficits the opposite of a “fiscal stimulus”).
Switching from bonds to stocks in this blog, I find no evidence that the S&P 500 stock index is “overpriced” relative to long-term interest rates (which is the only meaning of “overpriced” that relates to Feldstein’s argument about deficits and bond yields).
Feldstein claims stocks are “overpriced” because “the S&P 500 price/earnings ratio is now 70% above its historical average.” But there is no reason to expect the p/e ratio to revert to its long-term average unless bonds yields revert to their long-term average.