Administrative law judges might be unconstitutional under the separation of powers because they should be subject to presidentially controlled removal.
When serving as Department Head a few years back, I noticed administrative items routed for comment and approval through a series of “stakeholders” (jargony word, I know) sometimes still had glaring errors in them at the end of the process. The circulation to a wide group aimed to receive broad input from people with interests and, perhaps, knowledge, related to the item. The actual outcome was sometimes the opposite: Because of the number of reviewers, it seemed everyone assumed someone else would review the proposal carefully. So no one read the proposal carefully.
The irony is that had the proposal been assigned to one person or office for review, the single iteration of review would have produced a better outcome than a review process with multiple points of review.
This ironic outcome is an example of informational free riding: Multiplying steps in a review process aims to create a superior outcome, yet it can sometimes produce the opposite, an outcome worse than if the review process had only one step to it. To be sure, this perverse outcome did not happen all the time, or even all that often. Nonetheless, it happened often enough to get my attention when it did occur (and when I happened to notice it).
The risk of informational free riding extends beyond the byzantine bureaucracy of the academy. It can happen in separation-of-power political systems, given their multiple veto points.
To be sure, separation-of-power political systems first aim to create institutional “checks and balances,” and thereby to deter abuses of power. Beyond that, separation-of-power systems create a policy process to generate input from a broad class of (… wait for it …) political stakeholders: states in the Senate, the demos in the House, a national constituency in the Presidency, and legal experts in the judiciary.
But a significant, if often overlooked, part of the argument for the US separation-of-power system is that multiple actors will generate informationally superior policies – or at least superior ones relative to a simpler system in which legislative power is held by a unicameral body. While Congress is bicameral, that body does not exhaust the national legislative process. The President’s veto is, in fact, legislative power given to the President. While only a partial veto, it nonetheless can be thought effectively of creating a tricameral legislative process. Add judicial review, particularly during the era of substantive due process when courts reviewed the “reasonability” of legislation, and we might be excused for thinking the U.S. had the effective equivalent of a quadcameral legislative system.
Each additional veto player in the separation-of-power policy process, in principle, provides a “second sober look” at what has gone before. Or a third. Or even a fourth. Yet like the academic memo circulating among different administrators, it’s possible all of the institutional veto players wrongly expect the other veto players to provide the requisite “sober second look.” In such a case, a single institution with full responsibility for review and adoption would generate an informationally superior policy relative to the multiplied review process of the separation-of-power system.
Again, though, this doesn’t happen all the time. It doesn’t even happen much of the time. In the jargon of game theory, the players are implementing mixed strategies, that is, they are scrutinizing probabilistically, sometimes paying attention and sometimes not. What that means, though, is that with positive probability sometimes something slips through in which no veto player paid close attention.
The striking implication, however, is not that the separation-of-power system did not produce legislation any better than a unicameral decision maker, because of informational free riding among the multiple players in separation-of-power systems, the legislative outcome is actually worse than would have been made by a unicameral decision maker.
This is not to say that a system of system in which a unicameral legislature makes policy directly, without other veto players, is perforce better overall. But there are trade-offs. As Madison writes of the bicameral Congress in The Federalist 62, “It must be acknowledged that this complicated check on legislation may in some instances be injurious as well as beneficial.”
The institutional players themselves sometimes express awareness of this possibility, if implicitly. The U.S. judiciary adopted its extremely deferential “rationality review” for ordinary (non-suspect) socioeconomic legislation, in part, to avoid creating a moral hazard for U.S. legislatures. They wanted to avoid heightened judicial scrutiny becoming akin to a judicial insurance policy against bad legislation, thereby inviting legislative indolence in the policy process.
At the same time, bicameralism itself, along with the presidential or gubernatorial veto, can create informational free riding. Must judges ignore lousy legislation? The post hoc nature of judicial review of legislation, along with the standing doctrine (which creates informationally rich litigants with actual injuries caused by statutes), positions courts nicely to provide ex post checks on the reasonability of legislation.
The surprise, though, isn’t just that legislative bicameralism (or tricameralism, or quadcameralism) will sometimes not be able to improve on unicameral outcomes, it’s the possibility that, because of informational free riding, the multicameral process will produce worse legislation than a unicameral process. The question for institutional designers is to gauge the size of the loss relative to the gain when adding more veto players to the process.