Persistent unilateral action

Persistent unilateral action (DOI: https://doi.org/10.1017/psrm.2024.37) was written by David Foster in 2025. It was published in Political Science Research and Methods (vol. 13, no. 3).

The author explores why some executive orders issued by U.S. presidents have lasting impact on policy, while others are reverted immediately.

• 'Mexico City policy' first issued by Reagan, barring an organization which offers abortion care from receiving federal funding; while successive Republican administrations resurrect it predictably, Democratic administrations always immediately repeal it.

• Obama's memos to the DOJ directing nonintervention in the emergent state markets for legalized cannabis. Repealed by Trump and Sessions, but no actual enforcement has occurred.

The author analyses this problem in terms of policy-seeking groups. If a group has enough political power, they can compel the government to either enact an executive order or pass legislation. The core assumption is that legislation is more lasting, so is preferrable, but also requires greater power.

The hypothesis is that the persistence of executive orders is predicted by how the relevant policy-seeking group's power changed over time. The author is especially interested on formulating a policy feedback effect whereby an executive order empowers the group, or enables the group to empower itself, leading to a situation where the next administration has no effective option to revert.

The author uses a game theory model formulated as:

• Game takes two rounds, and actors are myopic: they do not anticipate the next 'round' (i.e. the next election of a new Congress and/or president).

• All actors are primarily policy motivated; they derive utility from proximity of the outcome policy position (x) to their preferred position.

• The set of actors are...

  • two presidents (P₁ and P₂) (one for each round), with preferred policy points p₁ and p₂.

  • the median member of Congress (C), with preferred policy point c.

  • a left-leaning policy-seeking group (L), with preferred policy point l

  • a right-leaning policy-seeking group (R), with preferred policy point r

• The actions that can be taken, in order, are:

  • L and R select a level of expenditures each to influence C and P_t. More on the budget later...

  • C can pass legislation moving x_t, but there is a small cost to legislate. As such, when indifferent, Congress prefers not to act.

  • P_t can (1) veto legislation, and (2) enact an executive order moving x_t. Presidents are assumed to have limited power, and can only move x_t by ±d.

As for the budget...

• The utility function for C is fully specified as: -(x_t - (c + k_R,t - k_L,t) where k_i,t is the level of expenditures selected by policy-seeking group i in round t. This demonstrates that this 'budget' has 1:1 equivalence to the policy preferences of C. (The same is true for P_t.)

• The budget constraint in round t is given by α + βF(x_t-1) where α is an exogenous 'starting point' constant, β is a scaling factor, and F is a monotonic function converting the previous policy position into a budget term. This demonstrates that there is a 1:β * dF/dx equivalence between policy and a group's budget, as long as there is ideological congruence between the policy and group.

  • Note furthermore that each group i has a distinct α_i, whereas β is global.

• The budget constraint function is characterized by F_R(0) = F_L(0) = 0.

The game is then simplified by assumptions:

• status quo is 0.
  • The set of policies feasible by executive order is now given by [-d,d].

  • The budget constraint in round 1 is now given simply by α_i, as the βF(.) component is constant at 0.

• no group has sufficient budget to 'win' outright.

  • The author stresses that an equivalently effective assumption is that l are r are infinitely extreme. All that matters is that the ordering of preferences is characterized by l < {p_t, c} < r.

• the full ordering of preferences in round 1 is characterized by l < p₁ < -d < c < r.

  • In other words, the assumption is that P₁ is left-leaning and bound by -d, while C is right-leaning.

• the full ordering of preferences in round 2 is characterized by l < c = p₂ < r.

  • In other words, the assumption is that P₂ and C are right-leaning and perfectly aligned.

Altogether, the interesting interaction in this game is the level of expenditures k_t,i selected by group i in round t. The executive order route will give a more preferential policy to L, as compared to the legislative route. A powerful L (i.e., high α_L relative to α_R) will vary between aiming for either of the two, depending on how much feedback that can be extracted from the preferential policy. A weak L (i.e., relatively low α_L) can only pursue an executive order.

Varying the exogenous β parameter is critical to the author's area of interest. As expected, the higher this term's value, the more likely an executive order will persist. Also, the higher this term's value, the more likely L will pursue the executive order route.

Reading Notes

First of all, the budget constraint is shoddy. One of these must be true:

1. groups can have negative budgets

2. the slope β * dF/dx is shallow enough that Congress could never move x_t far enough to bankrupt a group's budget

   • presidents are already constrained by ±d, so Congress is the risky actor here

I don't think (1) makes sense under any interpretation of what the 'budget' actually is. (2) means that the starting point dominates feedback. I think that the author has hand-waved this away by focusing so much (too much?) on the left-leaning group's incentives, and the game is designed that the left-leaning group is always ascendant in round 1.

The whole 'budget' thing is written so vaguely. It could mean campaign contributions, it could mean mobilizing electoral resources, it could mean trading votes in elections. I assume this was done to sidestep criticisms from any singular direction. But the paper would be so much easier to reason about if the author picked one.

The biggest issue is the equivalence between policy, political power, and politicians' utility. If the 'budget' were trading votes, for example, the implication becomes that politicians gain utility from both policy preferences and security in elections. But while feasible, there must be diminishing returns to the 'security in elections' component there.

If the 'budget' were campaign contributions, the equivalence between policy and political power is very questionable. There is considerable informal evidence that (the appearance of) moving policy in one direction mobilizes opponents, whereas the feedback theory relies upon there being a demobilizing effect following such an outcome.

CategoryRicottone CategoryReadingNotes