Sanctions: Some Simple Analytics

Literature › Works

Jonathan Eaton; Maxim Engers (1999). Sanctions: Some Simple Analytics. The American Economic Review (Papers and Proceedings of the 111th Annual Meeting of the AEA), Vol. 89, No. 2. American Economic Association.

Type	journal article
Scope	Comparative and methods
Method	theoretical or conceptual
Period studied	-
Regions	General/international (no South African content)
Themes	sanctions divestment, comparative theory, methods data sources
Citation key	eaton1999

Argument

The paper builds a simple game-theoretic framework (repeated games and bargaining under incomplete information) to explain when sanctions can succeed. Success is more likely when the threatened measure costs the sender little relative to the gain from changed behaviour and hurts the target a lot relative to his cost of compliance. In a world of perfect information sanctions are never actually imposed, because the threat alone works or is never made. Incomplete information about targets (stubborn types) or senders (pit bull versus paper tiger) produces observed episodes of actual punishment. These episodes are an unrepresentative tip of the iceberg, so studies of imposed sanctions can mislead about overall effectiveness, either understating or overstating it.

Evidence and method

No empirical data of its own. It is a theoretical model with stylised payoffs, and it cites Hufbauer et al. (1990) and trade-law studies for the empirical success rates. It mentions historical examples (Abyssinia, Cuba, Iraq) but does not discuss South Africa.

Findings

• Success is more likely when the sender's cost is low relative to her gain and the target's damage is high relative to his compliance cost; patient senders do better and patient targets do worse. (p. 2)

  We find success more likely when the threatened measure costs the sender little relative to the gain from modifying the target's behavior, while the damage to the target is large relative to his cost of complying

• Under perfect information, sanctions are never actually carried out, so imposition signals that something went awry. (p. 3)

  In a world of perfect information, senders never actually resort to punishing. If the measure works, the target shapes up in anticipation.

• Observed sanctions episodes are an unrepresentative sample of the role of sanctions. (p. 3)

  but these episodes constitute a very unrepresentative tip of an iceberg.

• Sanctions can achieve the sender's aim if reputation value outweighs punishment cost and punishment is worse than compliance for the target; the threat then works without being executed. (p. 4)

  Even though the threat of sanctions works, in equilibrium the measure is never carried out.

• With stubborn targets, punishment falls only on those unmoved by the threat, so observers may wrongly conclude sanctions are futile. (p. 4)

  An observer might conclude that sanctions are futile, when in fact they exact the compliance of targets against whom measures are never taken, possibly the vast majority.

• When a sender's resolve is uncertain, a demonstration of punishment can usher in a hegemonic regime, so observers may overrate effectiveness. (p. 5)

  Since this demonstration triggers an era of hegemony, an observer might conclude that punishment is extremely effective.

• In protracted interaction a complaisant target may balk to get the sender to drop sanctions, so the threat does not necessarily work against him. (p. 6)

  Hence, in the penultimate period the threat of sanctions does not necessarily work against a complaisant target, who may balk in hope of averting sanctions subsequently.

• A sender does better by committing to sanctions than by keeping period-by-period discretion. (p. 6)

  Commitment eliminates the complaisant target's incentive to balk simply to raise the chances that sanctions will be lifted subsequently.

• Quantifying sanctions' effectiveness must account for the circumstances that lead to their use, which requires theory-embedded econometrics. (p. 6)

  any attempt to quantify their effectiveness must consider the circumstances that lead to their use in the first place.