Efficient Breach and Ex-post Disgorgement of the Defendant’s Gain: A Comparative Contract Law and Economics Perspective

Appendix: An Economic Model of Contracts

Our model follows Givati and Kaplan with some important distinctions.^[94] First, we present a discrete rather than continuous model for sake of exposition. Second, the likelihood of breach is endogenous in our model (by virtue of seeking outside options) and exogenous (purely random) in theirs. Table 1 summarises the notation used in the model.

The sequence of the game is the following:

1. Time 0 – the contract is formed, and the buyer pays the seller P ₀ (deposit).

2. Time 1 – A) the buyer has an opportunity to make a relation-specific investment I which yields a return R, with R > I; B) the seller has the opportunity to make an investment E that generates outside options. These outside options are reflected by P ₁.^[95] They impose an additional loss L on the buyer (technological externality).

3. Time 2 – in the absence of an investment by the seller that generates outside options, the contract is performed. However, if the seller decided to generate outside options in time 1, she considers the possibility of breaching the original contract.

4. Time 3 – if the seller opted for breaching the original contract in time 2, damages D are paid to the buyer.

All technical assumptions by Givati and Kaplan are satisfied. More specifically, there is symmetric information. Both investments, I and E, are observable but cannot be part of a contract because they are not verifiable to outsiders. All parties are risk neutral, there is no time discounting, and no wealth constraints.

As conventional in the literature, we start with an efficiency analysis. Discussion of appropriate damages follows. Backwards induction is applied to establish the equilibrium of the game.

1. Efficiency analysis

If the contract is formed and the seller performs, the seller receives P ₀ − C net benefit, whereas K + P ₀ is the total benefit to the buyer, who pays P ₀, thus leaving K, the consumer surplus, as buyer’s net benefit (expectancy). Additionally, the buyer also benefits from the relation-specific investment. Therefore, the joint contract surplus, i.e. social net benefit, is K + R + P ₀ − C − I (aggregating seller’s P ₀ − C and buyer’s K + R + P ₀ − I − P ₀):

If the contract is formed but seller does not perform, the seller keeps P ₀, thereby receiving P ₀ + P ₁ − C − E, whereas the buyer simply receives −L − P ₀. In this case, the joint surplus is P ₁ − C − E − L, with P ₁ reflecting outside options generated by an investment E that, in turn, induced an additional loss L to the buyer:

Notice that if the contract is not performed, the relation-specific investment by the buyer should not take place because it generates a cost I with no return R.

A contract should be performed if and only if SWP > SWB (net gain from performance is greater than net gain from breaching). Rearranging, we have P ₁ − P ₀ < K + R − I + L + E: we should require the original performance when the additional revenue generated by the new contract (additional surplus), P ₁ − P ₀, does not justify the lost consumer surplus, K + L + R − I, plus the costs of generating breach, E (assuming, for simplicity, that production costs do not change across contracts).

Note that if seller does move to the alternative contract, the buyer paid P ₀ and suffered a loss L, and received zero benefits of any kind (due to the assumption of no alternative supplier).

1. Private decision

Consider the private decision by the seller when we apply a rule awarding damages for breach. In time 2, the seller will opt for performance if P ₀ − C > P ₀ − C − D + max{P ₁ − E, 0}, that is, the seller’s net gain from performing a contract is more than the seller’s net gain from performing an alternative contract (after damages are paid to the buyer). Consequently, when D> max{P ₁ − E, 0}, the initial contract will be performed. Otherwise, the seller prefers to breach because the gains offered by an alternative contract more than justify the loss to damages.

Clearly, from the seller’s viewpoint the critical relationships to establish are between damages (D) and the alternative price itself (P ₁), given an investment in seeking outside options (E). Furthermore, if in time 1, E does not justify outside options, damages are no longer relevant since there is no possible breach of contract.

We have shown that performance is efficient when SWP > SWB, that is, when K + P ₀ + R − I + L > P ₁ − E. Therefore, by straightforward comparison, damages D should equal K + P ₀ + R − I + L. Let us denote this amount as D*:

If damages are more than D*, then the seller is being over-penalised for breach, hence, we could have inefficient performance. If damages are less than D*, then the seller is being under-penalised for breach, therefore, we could have inefficient breach. In this setting, damages equal to D* guarantee efficient breach.

Efficient damages pay the buyer the lost surplus when the contract does not take place, K + L, plus restitution of the deposit, P ₀ and the lost net benefit from the relation-specific investment, R − I.

1. Damage rules

Under traditional applications of a restitutionary measure of damages, the seller simply returns the deposit, money spent or value of work completed. In our example, the seller owes D _re under a traditional restitutionary rule of returning the deposit:

By inspecting (3) and (4), we can easily observe that D _re < D*. It is straightforward that restitutionary remedies induce inefficient breach because they under-penalise the seller and do not force internalisation of the costs imposed on the buyer.

In comparison, the expectation measure of damages requires the seller to compensate the buyer for lost consumer surplus, K + L + R, plus money spent before the breach, P ₀, leaving the consumer in the same position as if the contract took place and generated K + L + R + P ₀ benefits. The reason is that the relation-specific investment I is a sunk cost for the buyer. Therefore,

Under (5), by comparison with (3), we can easily observe that D _e  > D*. Expectation damages induce inefficient performance because they over-penalise the seller. These remedies offer the buyer full insurance for relation-specific investments. Therefore, under expectation damages, the buyer makes an investment that should not be made when the relation is going to be breached. Consequently, expectation damages are also not efficient.

In terms of conventional economic literature, paying expectation damages is optimal in the traditional lost-deposit case of seller’s breach, absent relation-specific investments. When relation-specific investments exist, expectation damages are excessive.

Consider now a restitutionary requirement to pay over some of a breaching seller’s net benefits from the breach at a rate, r, where seller’s net benefits are the gain in revenue, P ₁ − E − P ₀, plus any buyer’s deposit, P ₀, left with seller (assuming for simplicity that costs do not change for the seller). This requirement to pay over all or some of the seller’s net gain from breach captures the idea of moving to a gain-based measure of damages and incorporates the range of disgorgement measures possible. The restitution measure of damages can be calculated as any amount up to the seller’s entire gain in profit from the breach, reflecting different legal theories about valuing unjust enrichment. Therefore,

is generalised restitution.^[96] Note that the deposit must still be repaid because it is initially held by the seller and is part of the unjust enrichment subject to the weighting by r. Damages are zero (uncompensated breach) when r = 0; they simplify to the traditional ‘money spent/lost’ restitution rule when r = P ₀/(P ₁ − E) and will fully disgorge all gain from breach when r = 1.

The relevant threshold is when r = r*, defined by comparing (3) and (6):

Under this threshold, there will be efficient breach. Therefore, when r = r*, the range of disgorgement is appropriate and produces a better result than both restitutionary measures and expectation damages.

There are two important situations to consider in more detail. One situation is when r* > 1. In this case, performance is always efficient by construction (notice that it means that K + P ₀ + R − I + L is always more than P ₁ − E). Therefore, full disgorgement is efficient (under lexicographic preferences) by fully removing any incentive for breach.

A different situation takes place when 0 < r*<1. A multitude of possibilities must be considered. First, as we have seen, when r = r*, there is efficient breach and disgorgement is better than either restitutionary damages or expectation damages.

Second, when r < r*, then disgorgement rate is insufficient and under-penalises the seller. In this case, the choice between expectation damages and disgorgement depends on the cost of inefficient performance (imposed by the former) versus the cost of inefficient breach (imposed by the latter). We can imagine that if r is in some arbitrarily small ε-neighbourhood of r*, disgorgement is still better. Once r is significantly lower than r*, for example, when r approaches zero, the result flips and expectation damages are better.

Finally, consider, in alternative, the case when r > r* – disgorgement is too severe and over-penalises the seller. In this situation, both expectation damages and disgorgement induce inefficient performance. Again, if r is in some arbitrarily small ε-neighbourhood of r*, disgorgement is necessarily better. However, once r is significantly higher than r*, for example, when r approaches one, expectation damages are less detrimental and become more appropriate.

Disgorgement-based remedies are better than expectation damages in chilling the seller from seeking outside options (in contrast with Givati and Kaplan, who present an exogenous model of breach).^[97] At the same time, the appropriate disgorgement rate, unlike expectation damages, disciplines the buyer’s relation-specific investment which constitutes a negative external effect when breach of contract takes place.