Discount Factors and Sustainable Cooperation

Why the Discount Factor Matters

A firm considering collusion compares an immediate gain from undercutting its rivals with the future profits it may lose if the other firms respond with punishment. The central question is not simply whether deviation pays more today. It is whether that short-run gain is worth sacrificing the stream of future cooperative payoffs.

This comparison is summarized by a critical discount factor. If firms value future payoffs sufficiently highly, the prospect of punishment can make continued cooperation preferable. If they place too little weight on future payoffs, a one-time deviation is attractive even when it triggers a lasting response.

Definition: Let $\beta$, with $0 \leq \beta < 1$, be the per-period strategic discount factor. A payoff received $t$ periods from now is valued today at $\beta^t$ times that payoff. A larger $\beta$ means greater effective patience or a greater likelihood that future periods will occur. In a strategic model, $\beta$ can incorporate time preference, continuation risk, or both.

The symbol $\beta$ here is a game-theory discount factor. It is distinct from the actuarial notation $\delta$ for the force of interest. If the only reason to discount is a constant effective annual interest rate $r$, the financial discount factor for one year is $1/(1+r)$. If there is also a probability $q$ that the interaction continues to the next year, a common model is $\beta=q/(1+r)$. The appropriate interpretation depends on the model.

Payoffs and the Trigger-Punishment Model

Consider an indefinitely repeated interaction with the following assumptions. Each period, a firm can follow a cooperative agreement or deviate. While all firms cooperate, the firm receives a constant per-period payoff $\pi_C$. A one-period deviation gives it $\pi_D$ in that period. The deviation is detected, and from the next period onward a credible punishment is applied, giving the firm $\pi_P$ each period. Assume $\pi_D>\pi_C>\pi_P$, and assume the firm’s best available one-period deviation payoff is $\pi_D$.

The timing matters: a firm that deviates receives $\pi_D$ now, not the punishment payoff. Punishment begins next period. Under these assumptions, the firm compares the value of cooperating forever with the value of deviating once and then receiving the punishment payoff forever.

V_C = \pi_C + \beta\pi_C + \beta^2\pi_C + \cdots = \frac{\pi_C}{1-\beta}

V_D = \pi_D + \beta\pi_P + \beta^2\pi_P + \cdots = \pi_D + \frac{\beta\pi_P}{1-\beta}

These expressions value the payoff stream starting in the current period. The first term in $V_D$ is the immediate deviation payoff; the punishment stream is delayed by one period and is therefore multiplied by $\beta$.

Critical discount-factor condition: Under the assumptions above, cooperation is at least as good as a one-period deviation if $$ \beta \geq \beta^* = \frac{\pi_D-\pi_C}{\pi_D-\pi_P}. $$ The threshold $\beta^*$ is the critical discount factor. The result assumes a credible punishment that follows deviation, constant per-period payoffs, and an indefinitely repeated interaction. It does not establish that cooperation is sustainable if the punishment itself is not credible or if some other deviation is more profitable.

Why the Threshold Has This Form

The condition follows by setting the value of cooperation at least as high as the value of deviation. Since $1-\beta>0$, multiplying the comparison by $1-\beta$ does not reverse the inequality.

Write the incentive comparison.
Require $\pi_C/(1-\beta) \geq \pi_D+\beta\pi_P/(1-\beta)$.

Clear the common denominator.
Multiplying by $1-\beta$ gives $\pi_C \geq \pi_D(1-\beta)+\beta\pi_P$.

Collect the terms involving $\beta$.
Rearranging gives $\beta(\pi_D-\pi_P) \geq \pi_D-\pi_C$.

Divide by the positive payoff gap.
Because $\pi_D>\pi_P$, division by $\pi_D-\pi_P$ preserves the inequality and yields $\beta \geq (\pi_D-\pi_C)/(\pi_D-\pi_P)$.

The numerator, $\pi_D-\pi_C$, is the immediate gain from deviating rather than cooperating. The denominator, $\pi_D-\pi_P$, is the gap between the deviation payoff and the payoff during punishment. A larger immediate temptation raises the threshold. A more severe punishment, meaning a lower $\pi_P$, makes the denominator larger and lowers the threshold.

When $\beta=\beta^*$, the firm is exactly indifferent between the two choices. If $\beta>\beta^*$, cooperation gives a strictly higher value; if $\beta<\beta^*$, deviation gives a strictly higher value. The weak inequality is appropriate when asking whether cooperation is at least as good as deviation. If strict preference for cooperation is required, use $\beta>\beta^*$.

Worked Example: Finding the Threshold and Comparing Values

In a hypothetical repeated market, a firm earns 40 payoff units per period while all firms comply with an agreement. A one-period deviation earns 70 units, after which permanent punishment gives 10 units per period. The strategic discount factor is 0.60. Determine whether the agreement is sustainable against this deviation. State the method and check the result by comparing values.

Method. Use the incentive condition for a one-period deviation followed by permanent punishment. Here $\pi_C=40$, $\pi_D=70$, $\pi_P=10$, and $\beta=0.60$. The payoff ordering $70>40>10$ meets the model’s assumed ordering.

Step 1: Calculate the threshold.

\beta^*=\frac{70-40}{70-10} =\frac{30}{60} =0.50

Step 2: Compare the actual discount factor with the threshold. Since $0.60>0.50$, the condition predicts that cooperation is better than deviation.

Step 3: Verify using present values. The value of cooperating is $40/(1-0.60)=40/0.40=100$ payoff units. The value of deviating is $70+(0.60)(10)/(1-0.60)=70+6/0.40=70+15=85$ payoff units. Cooperation is worth 100 units, compared with 85 units from deviation, so the agreement is sustainable against the specified deviation.

Common mistake avoided: Comparing only the current payoffs, 40 and 70, would suggest deviation. That comparison omits the future punishment stream. The relevant comparison is between the complete payoff paths.

Worked Examples with Different Discount Factors

Worked Example: Checking a Discount Factor Below the Threshold

Suppose a hypothetical firm receives 24 units per period under cooperation, 42 units from a one-period deviation, and 6 units per period under permanent punishment. Its strategic discount factor is 0.40. Determine whether cooperation is sustainable against that deviation, and verify the conclusion by calculating both values.

Method. First calculate the critical discount factor, then compare the two infinite payoff streams. The payoff order is $42>24>6$, so the standard threshold formula applies.

Step 1: Calculate the threshold.

\beta^*=\frac{42-24}{42-6} =\frac{18}{36} =0.50

Step 2: Compare factors. The actual factor is $0.40<0.50$, so the firm should prefer deviation under the stated model.

Step 3: Check by valuing each path. Cooperation is worth $24/(1-0.40)=24/0.60=40$ units. Deviation is worth $42+(0.40)(6)/(1-0.40)=42+2.4/0.60=42+4=46$ units. The deviation value exceeds the cooperation value by 6 units, confirming the conclusion.

Common mistake avoided: A punishment payoff below zero is not required for punishment to deter deviation. But a low punishment payoff cannot ensure cooperation when the firm discounts the future enough that the immediate gain dominates, as this calculation shows.

Worked Example: Incorporating Continuation Risk and Interest

In a hypothetical market, a firm’s cooperation, deviation, and punishment payoffs are 30, 50, and 5 units per year, respectively. The market has a 0.60 probability of continuing to the next year, conditional on reaching the current year. Payoffs are also discounted at an effective annual interest rate of 8%. Determine whether cooperation is sustainable using a strategic discount factor that incorporates both effects.

Method. Use $\beta=q/(1+r)$, where $q=0.60$ is the continuation probability and $r=0.08$ is the effective annual interest rate. This model assumes continuation risk and financial discounting apply once per year.

Step 1: Calculate the strategic discount factor.

\beta=\frac{0.60}{1+0.08} =\frac{0.60}{1.08} =\frac{5}{9} \approx 0.5556

Step 2: Calculate the threshold.

\beta^*=\frac{50-30}{50-5} =\frac{20}{45} =\frac{4}{9} \approx 0.4444

Step 3: Compare the factors. Since $5/9>4/9$, the firm prefers cooperation to the specified deviation.

Step 4: Verify by comparing values. The cooperation value is $30/(1-5/9)=30/(4/9)=67.5$ units. The deviation value is $50+(5/9)(5)/(4/9)=50+25/4=56.25$ units. Cooperation is worth 67.5 units, exceeding the deviation value of 56.25 units.

Common mistake avoided: Using $\beta=1/1.08$ would account for interest but ignore the 0.60 continuation probability. Using $\beta=0.60$ would account for continuation risk but ignore interest. Under the assumptions given, both factors apply, so they are combined as $0.60/1.08$.

Interpreting the Result and Its Limits

The threshold is a compact way to compare the temptation to deviate with the cost of losing future cooperation. It is not a claim that firms always collude when $\beta$ is high. It says that, under the stated payoff and punishment assumptions, a firm does not gain by making the specified one-period deviation when the threshold is met.

The test also depends on the details of the model. If punishment lasts only a fixed number of periods, its value must be calculated over that finite period rather than as a perpetual stream. If firms have different payoff levels, each firm’s own incentive constraint must be checked. If several possible deviations exist, the most profitable one must be tested. And if the punishment action would not actually be carried out, its promised payoff cannot be used as a credible deterrent.

A further boundary case is useful. Under the assumed ordering $\pi_D>\pi_C>\pi_P$, the threshold lies strictly between zero and one: the numerator is positive, and it is smaller than the denominator because $\pi_C>\pi_P$. If the best deviation instead gives no more than cooperation, a one-period deviation of that type creates no immediate temptation. The usual positive-threshold interpretation is then unnecessary; the relevant decision is still to compare the complete payoff paths.

Key takeaway: Under an indefinitely repeated interaction with a credible permanent punishment, cooperation deters a one-period deviation when the per-period strategic discount factor is at least $(\pi_D-\pi_C)/(\pi_D-\pi_P)$. Identify the timing of punishment and define the discount factor before applying the formula.

This reasoning is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus. In a calculation, make the payoff timing explicit, distinguish the strategic discount factor from an interest-rate notation, and verify a threshold result by comparing the two payoff values when practical.

Check Your Understanding

For each problem, treat punishment as credible and permanent, beginning in the period after a one-time deviation. Use the stated per-period payoffs and discount-factor assumptions.

A firm receives 18 units per period under cooperation, 30 from a one-period deviation, and 6 per period under punishment. If $\beta=0.55$, find the critical discount factor and decide whether cooperation is sustainable against this deviation. (Answer: $\beta^*=0.50$; cooperation is sustainable.)
A firm has $\pi_C=32$, $\pi_D=56$, and $\pi_P=8$, with $\beta=0.40$. Calculate the threshold and identify which path has the higher value. (Answer: $\beta^*=0.50$; deviation has the higher value.)
A hypothetical annual interaction has $q=0.66$, an effective annual interest rate of 10%, and payoffs $\pi_C=15$, $\pi_D=27$, and $\pi_P=3$. Find $\beta$, find $\beta^*$, and determine whether cooperation is sustainable. (Answer: $\beta=0.60$, $\beta^*=0.50$; cooperation is sustainable.)
A firm has $\pi_C=20$, $\pi_D=35$, and $\pi_P=5$. Find the threshold and decide whether cooperation is at least as good as deviation when $\beta=0.50$. State whether cooperation is strictly better. (Answer: $\beta^*=0.50$; the firm is indifferent, so cooperation is at least as good but not strictly better.)
A firm has $\pi_C=24$, $\pi_P=0$, and a strategic discount factor of $0.75$. What is the largest one-period deviation payoff $\pi_D$ for which the incentive condition still holds? (Answer: $\pi_D=96$ units; at this payoff the firm is indifferent, and lower deviation payoffs satisfy the condition.)

What You'll Learn