Dividend tax holiday?

The White House and Eric Cantor are sparring over the idea of enacting a dividend repatriation tax holiday, Part Deux. Cantor proposes that U.S. companies be given a temporary low rate for bringing foreign earnings home. We tried this back in 2004, and academic research predominantly found that it did not (as promised by the proponents) lead to job creation. Instead, it led almost exclusively to increased share buybacks and dividend payouts.

In general, temporary tax holidays of this sort are a horrendously bad idea, because they send the message that one shouldn't repatriate any earnings, once the holiday ends, until the next one is declared - surely just a matter of time if we do it for the second time in only eight years. The reason this is bubbling up again is that it has powerful backing from narrow economic interests that know how to make friends in Washington, in particular the companies (e.g., Cisco) that have lots of earnings trapped out there (in large part because they have tax-planned effectively to treat a disproportionate share of their global earnings as arising in tax havens abroad).

In defense of the tax holiday, James Pethokoukis of Reuters argues:

"[E]ven if all the cash returning to the United States went to companies’ shareholders, that could still generate more consumption, growth and jobs, a knock-on effect Treasury ignores. Yet this so-called wealth effect is explicitly part of the rationale behind the Fed’s second round of quantitative easing."

OK, fair enough, and the general rule that one wants to avoid creating tax-motivated timing distortions in economic decision-making can rightly be called off during a recession (as when Congress enacts temporary expensing rules for business investment).

But I see two problems. The first is that this is really poorly-directed stimulus, although no more so than extending the Bush tax cuts for high-bracket taxpayers. The second is that U.S. multinationals' decisions over where to report their earnings and when to repatriate their funds play out more time-sensitively than, say, when to buy new machines for a business. A temporary expensing rule doesn't significantly induce companies to wait for the next time such a rule is enacted during the next recession. But multinationals' decisions about where to report earnings and when to move their funds around can very well involve waiting for the next holiday.

Pethokoukis may be arguing, not entirely without reason, that even a really badly designed stimulus with bad behavioral effects on the side is worth considering if all better-designed stimulus measures are ruled out on political grounds - though I must say, I still don't like the terms of trade here. But he is further off-base when he argues that "Treasury is stretching a point in assuming the government would somehow lose revenue by taxing repatriated income at a sharply lower rate. In reality, without the reduction most of the money will remain offshore."

The revenue estimates that show the holiday as costly are based on actual repatriation rates, which are expected to be low in any event but not zero. They decline, as independent studies (e.g., by Thomas Brennan) have also suggested, when granting a holiday affects expectations.

Our policy process has grown amazingly debased when this sort of proposal keeps coming up even though it was done so recently, so decisively failed to have the promised effects, and is so transparently a bad idea. The irony, of course, is that this very debasement offers the best argument for doing it, on the ground that the entire vast universe of better-designed stimulus options is politically off-limits.