Teams can coordinate successfully and still invest too little in working together

Cooperation is not the whole problem

Shared code structures, testing tools and ideas that combine complementary expertise are examples of what David Gill, Victoria Prowse and J. Lucas Reddinger call team-specific skill or capital. These investments can make a particular partnership more productive over time. But someone must bear the initial cost, and the benefits depend on teammates choosing to work together.

In their preprint, Dynamic investment and coordination in teams: Theory and experimental evidence, the researchers develop a mathematical model and test it in a laboratory. Their central finding is that teams can coordinate successfully while still investing less than would maximize their combined earnings.

A partnership in three rounds

The experiment involved 606 students at Purdue University across 30 sessions. Participants played one unpaid training match followed by eight paid matches, with random pairing at the start of each match. Each partnership lasted three rounds, producing 2,424 paid matches for analysis.

Every round had two stages. First, each participant independently decided whether to pay 15 points to invest in team skill. Each investment increased the payoff available from successful teamwork for the current round and every remaining round of that match. The size of this increase was randomly selected at the beginning of each paid match, ranging from 2 to 20 points in steps of two.

After seeing both investment decisions, participants chose between a team task and an individual task. The individual task guaranteed 73 points. If both chose the team task, each received the current team-skill payoff, which started at 87 points before investment. Someone choosing the team task alone received only 17 points. Investment costs were deducted separately.

This design made teamwork potentially more rewarding but risky: a participant could pay to improve the partnership and then fail to benefit because the other person chose the safe alternative.

Why investment can stop too soon

The model separates what is best for an individual from what maximizes the pair’s combined payoff. If both teammates will always choose the team task, an individual has reason to invest when their own remaining benefits cover the cost. For the pair, however, an investment benefits two people, so it can be worthwhile even when the investor’s personal return is insufficient.

As the partnership approaches its end, fewer rounds remain to recover the cost. One equilibrium therefore predicts that teammates stop investing earlier than would be jointly efficient, even while continuing to coordinate on the team task. The model also permits a fully efficient equilibrium supported by conditional task choices, but this requires costly coordination on the inferior individual task to punish failures to invest.

Incentives changed behavior

Participants invested in 65% of decisions and selected the team task in 86%. Full social efficiency—both efficient investment decisions and successful team-task coordination throughout a match—occurred in 43% of matches.

Another 32% achieved team-task coordination throughout but had inefficient investment patterns. These were predominantly cases of underinvestment. The remaining 25% failed to coordinate fully on the team task.

Investment rose strongly with the randomly varied return to investment and declined as rounds progressed. Changes were larger around thresholds where the model predicted a switch in investment behavior. These patterns support the theory’s account of how investment returns and a shortening horizon shape decisions.

Social understanding predicted outcomes

The researchers also examined three pre-registered individual predictors. Theory of mind—the ability to understand others’ mental states—was measured using a test requiring participants to interpret photographs of eyes. A Coordination Attraction Score summarized earlier choices in seven separate coordination games. Cognitive ability was assessed with a matrix-reasoning test.

Higher theory-of-mind scores and coordination scores predicted investment and team-task choice. A one-standard-deviation increase in the pair’s average theory-of-mind measure was associated with roughly a five-percentage-point higher probability of full efficiency; the corresponding association for the coordination score was about seven points. Evidence for cognitive ability was weaker. These are predictive associations, not evidence that changing a trait would cause better teamwork.

What the experiment establishes

The authors also report a roughly ten-percentage-point increase in team-task choice when a partner invested, using detailed controls for prior play and the decision setting. They interpret this as causal evidence consistent with investment signaling an intention to cooperate; signaling itself is not directly established.

The study is a preprint, and its evidence comes from anonymous student pairs making monetary choices over three-round partnerships, not workplace teams developing actual skills. Team skill here is a payoff variable increased by spending points. The authors’ proposed applications—such as team bonuses and selecting teammates using predictive traits—remain implications rather than directly tested organizational policies. The experiment’s sharper lesson is that coordinating production and financing the partnership’s future are distinct problems.