📈 economics

On Rent Dissipation in Dynamic Multi-battle Contests

This paper investigates how contest structures, particularly the discouragement effect arising from exchangeability in dynamic multi-battle settings, influence rent dissipation, ultimately establishing a necessary and sufficient condition for almost-full dissipation and demonstrating how environmental volatility can sustain incentives to achieve this outcome.

Original authors: Shanglyu Deng, Qiang Fu, Junchi Li, Zenan Wu

Published 2026-04-23

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Shanglyu Deng, Qiang Fu, Junchi Li, Zenan Wu

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a long, grueling sports tournament. It's not just one game; it's a series of battles, like a best-of-seven playoff or a tug-of-war that could go on forever. The prize is huge—let's say a million dollars.

The big question this paper asks is: How much effort will the players put in? Will they fight so hard that they spend almost the entire million dollars just trying to win it, leaving the winner with almost nothing? Or will they give up early, saving their energy?

The authors, Deng, Fu, Li, and Wu, discovered that the rules of the game determine the answer. They found that in many common types of contests, players actually "give up" too early, meaning the prize isn't fully "dissipated" (spent) on effort. But they also found a special kind of contest where players fight until the very last drop of energy, spending the whole prize value in the process.

Here is the breakdown of their findings using simple analogies.

1. The "Snowball" Effect (The Problem)

In many contests, like a Tug-of-War or a Best-of-Seven Series, the rules have a "memory."

The Analogy: Imagine a snowball rolling down a hill. If it gets a little bigger early on, it picks up more snow, gets heavier, and becomes harder to stop.
The Reality: If Player A wins the first few rounds, they get a "lead." Because the rules say "you need to be N steps ahead to win," that lead feels permanent. Player B (the laggard) looks at the scoreboard, sees they are far behind, and thinks, "I can never catch up." So, Player B stops trying hard. Player A, seeing no resistance, also relaxes because winning is easy.
The Result: The contest ends up being a "lopsided" race. They don't spend all their energy fighting. The "rent" (the prize value) is not fully dissipated. The winner gets the prize, but the total cost of the fight was much lower than the prize was worth. This is called the Discouragement Effect.

2. The "Reset Button" (The Solution)

The authors asked: Is there a way to make the players fight until the bitter end, even if the contest is very long?

They found that you need to break the "Snowball Effect." You need a rule where a lead doesn't matter as much as it used to. They call this "Transient Dominance."

The Analogy: Imagine a game of "Hot Potato" where, every time someone wins a round, there is a 10% chance the game resets to zero. Or imagine a video game where the map changes every few levels.
The Reality: In this scenario, even if Player A is winning, they can't get comfortable. Why? Because the "environment" might change tomorrow. The laggard (Player B) knows that even if they are behind now, a "reset" or a "shock" could wipe the scoreboard clean, giving them a fresh start.
The Result: Because the lead is temporary (transient), Player B never gives up. They keep fighting hard. Player A also keeps fighting hard because they are terrified of losing their lead. They spend all their energy. The total effort equals the value of the prize. This is Full Rent Dissipation.

3. Two Types of Contests

The paper compares two specific types of games to prove this point:

A. The "Tug-of-War" (Bad for Effort)

How it works: You need to pull the rope 5 meters past the center line to win.
The Flaw: If I pull you 4 meters, I am almost there. You are almost defeated. You stop pulling. I stop pulling hard. We waste the potential of the game.
The Math: Even if you make the rope 1,000 meters long, the "Snowball Effect" still happens. The person in the lead eventually becomes so strong that the other person quits.

B. The "Consecutive Wins" Game (Good for Effort)

How it works: To win, you must win 5 games in a row.
The Magic: If I win 4 games, I am one step from victory. But if I lose the next game, my streak is broken, and I am back to zero. My lead is gone!
The Result: The person behind (who lost 4 times) knows that if they just win one game, the leader's streak resets. The leader is terrified of losing that streak. The laggard is hopeful. They both fight with 100% intensity until the very end.

4. Real World Examples: Why This Matters

The authors apply this to real life to explain why some competitions are fierce and others are boring.

Technology Wars (The "Red Queen" Effect):
Think of companies like Nokia vs. Apple. Nokia was the "incumbent" (the leader). In a normal "Tug-of-War" world, Nokia would have just coasted to victory. But the tech world has random shocks (new inventions, changing consumer tastes).
- A new technology (like the smartphone) acts like a Reset Button. It wipes out Nokia's lead.
- Because the lead can be wiped out at any moment, companies keep spending billions on R&D. They never stop running. This is why we see such intense competition in tech.
Evolution (Nature):
Think of a predator (lion) and prey (zebra). If the lion gets faster, it wins. But if the environment changes (drought, new grass), the zebra might evolve a new defense. The lion's advantage is transient. This forces both animals to constantly evolve and expend energy just to survive. They never stop running.

Summary

The Problem: In many contests, an early lead makes the loser give up, so the total effort is low.
The Cause: Rules that let you "save" your wins (like a Tug-of-War) create a "long memory" that discourages the loser.
The Fix: Rules that wipe out leads (like needing "consecutive wins" or having random "resets") create "short memory."
The Outcome: When leads are temporary, everyone fights to the death, spending the entire value of the prize in the process.

The paper essentially tells us: If you want a fierce, high-effort competition, design rules where no one can ever feel safe in the lead.

1. Problem Statement

The paper investigates rent dissipation in dynamic multi-battle contests. In these settings, two players compete over a sequence of component battles to win a final prize. The central question is: Under what contest structures does the total effort exerted by players approach the full value of the prize (full rent dissipation), and when does it fall short?

Existing literature highlights the discouragement effect: if a player gains an early lead, the trailing player may reduce effort due to bleak winning prospects, leading to inefficiently low total effort (under-dissipation). The authors ask whether this effect persists even as the contest length becomes arbitrarily large (infinite horizon) and identify the structural properties that determine the outcome.

2. Methodology and Model Framework

The Model

Players: Two risk-neutral players ( $A$ and $B$ ) competing for a prize $v$ .
Structure: A sequence of component battles. In each battle, players exert effort $x_\ell$ .
Success Function: The probability of winning a battle is determined by a success function $p(x_A, x_B)$ . The baseline analysis assumes a homogeneous success function (e.g., Tullock contest success function), where $p(x, x') = \gamma(x/x')$ .
Contest Rule: A rule maps the history of battle outcomes to a winner. The model is general, accommodating various rules (e.g., Best-of- $K$ , Tug-of-War).
Equilibrium Concept: Symmetric Markov Perfect Equilibrium (MPE), where strategies depend only on the current state (history summary).

Key Definitions

Nontrivial Contest: A contest where the minimum number of battles required to decide a winner is finite ( $L(M) < \infty$ ).
Exchangeability: A contest is exchangeable if the order of past wins/losses does not matter, only the net count. (e.g., In a Tug-of-War or Best-of- $K$ , the state is defined by the score difference or total wins, regardless of the sequence).
Transient Dominance Property: A condition where a player's lead is temporary and can be reversed with high probability, preventing permanent accumulation of advantage.

3. Key Contributions and Results

A. Upper Bound on Rent Dissipation (Theorem 1)

The authors establish that in any nontrivial contest, rent cannot be fully dissipated.

Result: The expected total effort is bounded above by $[1 - (\phi(1))^{L(M)}]v$ , where $L(M)$ is the minimum length of the contest and $\phi(1)$ is a constant derived from the success function.
Implication: Even if a contest can theoretically last forever, as long as there exists some finite path to victory, the equilibrium total effort is strictly less than the prize value.

B. The Discouragement Effect in Tug-of-War and Exchangeable Contests

The paper analyzes the limiting case where the contest length approaches infinity ( $L(M) \to \infty$ ).

Tug-of-War (Theorem 2): In a Tug-of-War contest (where a player must win $N$ $N$ more battles than the opponent), full rent dissipation does not occur even as $N \to \infty$ $N \to \infty$ .
- Mechanism: Winning is self-reinforcing. As a player's lead grows, their incentive to win the next battle increases relative to the opponent's. Eventually, the laggard becomes fully discouraged (probability of winning $\to 0$ ), and the frontrunner wins subsequent battles with near certainty.
- Result: There exists a uniform upper bound on rent dissipation strictly less than 1, independent of $N$ .
Exchangeability (Proposition 3): The authors identify exchangeability as the structural root of this under-dissipation.
- Exchangeable contests possess "long memory." Early wins accumulate into a persistent advantage that cannot be erased by a single loss. This cements the discouragement effect.
- Conclusion: Any sequence of exchangeable contests (e.g., Best-of- $2K+1$ as $K \to \infty$ ) fails to achieve full rent dissipation in the limit.

C. The Role of Non-Exchangeability: Consecutive-Win Contests

To contrast, the authors analyze $K$ -consecutive-win contests (the first to win $K$ battles in a row wins).

Mechanism: This contest is non-exchangeable. A single loss resets the "streak" counter to zero (or negative), effectively wiping out the leader's accumulated advantage. This creates "short memory."
Result (Theorem 3): As $K \to \infty$ $K \to \infty$ , the equilibrium total effort converges to the prize value $v$ $v$ (almost-full rent dissipation).
- The laggard is never fully discouraged because a single win can reset the game state, keeping the competition fluid and the lead insecure.

D. Necessary and Sufficient Condition: Transient Dominance

The paper proposes the Transient Dominance Property (Definition 4) as the general condition for almost-full rent dissipation.

Definition: A contest exhibits transient dominance if:
1. There exist states where a player's continuation value is negligible (weak position).
2. The probability of the contest path visiting both players' weak states is high (leadership is reversible).
Theorem 4: A contest achieves almost-full rent dissipation if and only if it satisfies the transient dominance property. This property ensures that no player can accumulate an unbounded advantage, sustaining incentives for both.

E. Application: Iterated Incumbency Contests

The authors introduce the Iterated Incumbency Contest to model real-world dynamic competition (e.g., Schumpeterian competition, biological evolution).

Structure: Rounds of competition occur. In each round, there is a probability $q$ that the "fundamentals" reset (exogenous shock), forcing a new sub-contest for incumbency. The incumbent has a head start (biased sub-contest).
Result (Theorem 6): If the sub-contests are sufficiently biased but the number of rounds $N$ $N$ is large, the contest satisfies transient dominance.
- Logic: While the incumbent has an advantage in any single round, the probability of an exogenous shock resetting the contest ensures that leadership is temporary. The laggard remains hopeful of overtaking the incumbent before the final round.
- Outcome: As $N \to \infty$ , rent is almost fully dissipated. This provides a theoretical basis for the Red Queen effect (species must constantly run to stay in the same place) and sustained innovation in markets.

4. Significance and Implications

Structural Determinants of Efficiency: The paper shifts the focus from specific contest rules to the structural property of exchangeability. It demonstrates that "long memory" (exchangeability) inherently limits efficiency by allowing early luck to compound into permanent dominance, whereas "short memory" (non-exchangeability) sustains competition.
Resolution of the Infinite Horizon Paradox: Contrary to intuition, making a contest longer (increasing $N$ in Tug-of-War) does not necessarily increase effort to the prize value. The discouragement effect persists asymptotically in exchangeable settings.
Policy and Design Insights:
- To maximize effort (rent dissipation) in dynamic competitions (e.g., R&D races, political primaries), designers should introduce mechanisms that break exchangeability or reset the state (e.g., random shocks, intermediate resets, or rules where a single loss negates previous gains).
- Pure "race" models (first to $K$ wins) are inherently less efficient than models where leadership is fragile.
Interdisciplinary Application: The "Iterated Incumbency" model offers a rigorous game-theoretic explanation for phenomena in economics (technological disruption) and biology (evolutionary arms races), showing how environmental volatility sustains competitive effort.

5. Extensions

The authors verify that their main results hold for a broader class of ratio-form success functions (not just homogeneous ones) and discuss the robustness of the findings when noise is introduced into the battle success functions.

In summary, the paper provides a unified framework showing that dynamic incentives and rent dissipation are fundamentally shaped by how a contest treats past history. Exchangeability leads to discouragement and under-dissipation, while transient dominance (via non-exchangeability or random resets) ensures sustained competition and near-full rent dissipation.