Money Exchange

What you see

At the top of the board is a trust bar, Partner 1's trust in you, with a value such as 55 / 100. Below it are two cards, You and Partner 1, each showing that side's share of the pot, and between them a split bar. Two pills above the board show the round (Round 3 / 10) and the points you have earned so far (Points 140).

Every round is one of two kinds.

• You respond. The prompt reads, for example, Partner 1 offers you 41 of 93 points. The cards show 41 for you and 52 for Partner 1. You choose Accept (key A) or Reject (key R). Accept, and you receive the offer. Reject, and nobody gets anything.
• You propose. The prompt reads, for example, How many of the 129 points will you offer Partner 1? The split bar becomes a slider. You move it (or use the − and + buttons, or the Left and Right arrow keys, one point at a time) until the cards read, say, 89 for you and 40 for Partner 1, then choose Send offer (Space). The slider starts each round at half the pot. If the partner accepts, you keep the rest; if they reject, nobody gets anything.

After each round the line under the buttons states the outcome, for example Partner 1 accepted: you keep 89 and Partner 1 receives 40. You then choose Next round (Space), or Finish after the last round. The cards then read You earned and Partner 1 earned (or Partners earned when there was more than one partner).

One run of the game is one trial of 8 to 21 rounds. You deal with each partner in a single block of rounds, and within a block the rounds alternate, starting with you responding. There is no time limit, and speed has no effect on the score.

How trust works

Each partner starts every run with the same trust in you, and every choice you make moves it.

• Accepting their offer raises their trust by 4 or 5 points.
• Rejecting their offer lowers it by 5 points at level 1, rising to 20 at level 10.
• Your offer is accepted if it reaches a minimum they will take. You never see that minimum. An accepted offer of at least 45 per cent of the pot counts as generous and raises trust by 4 or 5; an accepted offer below that leaves trust unchanged.
• Your offer is rejected if it falls short of their minimum. Nobody is paid, and trust falls by 5 points at level 1, rising to 18 at level 10.

Trust matters because it moves both numbers you care about. The higher their trust, the bigger the share they offer you, and the smaller the share they insist on from you. A partner's offer and minimum are fixed percentages of the pot at any given trust level, so you can think in percentages even though the pot size changes from round to round.

What changes with level

Every partner in a run starts with the same trust. As levels rise, partners start less trusting, trust swings further with each point (at level 10 a partner at full trust offers about 65 per cent and accepts about 10, while at zero trust they offer about 15 and demand about 50), and mistakes cost more trust.

What it measures

Money Exchange feeds the Negotiation and trust facet and carries 15 per cent of your Decision Making score.

It combines two classic economics experiments. In the ultimatum game, introduced by Werner Güth, Rolf Schmittberger and Bernd Schwarze in 1982, a proposer offers a split of a sum and a responder accepts or rejects; rejection means both get nothing. Pure self-interest predicts tiny offers that are always accepted. Real people behave very differently.

The dictator game removes the responder's veto. Christoph Engel's meta-study of more than a hundred dictator experiments found dictators give 28.35 per cent of the pie on average, with 36 per cent of people giving nothing and about 17 per cent choosing an equal split. The gap between the 28 per cent people give when they cannot be refused and the 40 per cent they offer when they can suggests that part of ultimatum generosity is strategic: fear of rejection as well as kindness.

Camerer and Thaler's review adds a detail that matters here: responders reject offers mainly when those offers are proportionally stingy and were chosen deliberately by the proposer. Rejection is punishment, and people will pay to punish.

Our version turns these one-shot games into a relationship. Rejection still punishes, but it also teaches your partner to distrust you, and you pay for that in every later round with them. The task measures whether you can see that trade-off and act on the numbers rather than the sting.

How it is scored here

The benchmark for your run is the most any sequence of choices could have earned from the same rounds, found by an exact search over every accept, reject and offer decision for each partner, given the trust rules and the partner's starting trust. Your score is your earnings divided by that benchmark, capped at 100 per cent.

That best-possible path has a clear shape, and knowing it tells you what the game rewards:

1. It always accepts. In our rules, rejecting never pays: you lose the offer now, and the lost trust shrinks later offers and raises later minimums.
2. It mostly offers exactly the partner's minimum. Occasionally, early in a long block at high levels, it makes a generous offer to buy trust for bigger pots later. It never does so on the last proposal to a partner.

In our simulations of the scoring, typical strategies landed here:

The fourth row is the warning. A fixed low offer works while partners' trust is high, then collapses when their minimum rises above it, because each rejection lowers trust, which raises the minimum further.

Level changes follow the usual rule: 98 per cent or more moves you up a level, zero moves you down, anything else holds. Because the partner's minimum is hidden, even the best learning strategies in our simulations rarely reached 98 per cent, so expect to hold your level; the score already adjusts for difficulty. New sessions start at level 3.

Your report and review show descriptive measures that never change the score: Offers refused (the share of offers you rejected), how many offers you accepted and rejected, each partner's final trust, and the mean final trust across partners. Across sessions your Money Exchange score sits on our 0–100 scale, where 50 means what the difficulty you work at predicts; longer runs count as harder.

Where you will meet it

Candidate-preparation sites describe the second pymetrics Money Exchange game as a sharing game close to the dictator design: you and a computer partner each hold a small sum, one of you receives an extra amount, and you decide whether to share some of it, then rate the fairness of the transaction. That description is candidate-reported. In vendor versions like this, how much you give or accept is treated as a trait with no right answer.

The same questions appear in words in MBA entrance testing. XLRI's official XAT pattern includes, within its Decision Making section, situations involving fairness, distribution and prioritisation. The game trains the numerical side of those judgements: what split will be accepted, and what it costs to push too hard.

The method

1. When responding, accept

   Any offer beats nothing, and in this game every rejection also costs trust, which lowers every later offer from that partner and raises the price of your own proposals. Accepting raises trust. There is no round in which rejecting earns more.

2. When proposing, open at about 40 per cent

   Think in percentages of the pot. Forty per cent is below the generous line, so it keeps more for you, and it clears the partner's minimum at every level at the starting trust.

3. Walk down slowly, jump up fast

   After an accepted offer, try about 3 points lower (as a percentage of the pot) next time. After a rejection, go back up by about 6. Rejections cost the whole pot and a chunk of trust, so err on the safe side of the minimum.

4. Read the trust bar

   When trust rises, the minimum falls, so you can shave a little more. When trust has dropped, give yourself extra margin: the minimum has moved up.

5. Restart with each new partner

   A new partner starts at the run's starting trust, whatever happened with the last one. Begin your offers near 40 per cent again.

Worked example 1: the cost of the default

Level 1. Partner 1 starts at 75 trust. Every pot is 100. On each of your four proposing turns you leave the slider where it starts, at half, and press Send offer. What did that cost?

Show the answer

At trust 75, this partner's minimum is about 26 per cent, and it only falls as their trust rises with each acceptance. Offering 50 each time keeps 50. Offering about 30 (a comfortable margin above the minimum) keeps about 70. Over four proposals that is roughly 80 points left on the table.

The half-offer is not wrong in spirit; it is generous, never rejected, and even nudges trust up. But in our simulations of level 1, always offering half captured about 80 per cent of the benchmark. The slider's starting point is a convenience, not a recommendation.

Worked example 2: the satisfying rejection

Level 5. Partner 1 starts at trust 55. They offer you 41 of 93, keeping 52. It feels unfair. Next you will propose on a pot of 129, then respond on a pot of 139. What does rejecting cost across these three rounds, assuming you offer their minimum when you propose?

Show the answer

Round	If you accept	If you reject
Respond, pot 93	+41, trust rises to 60	0, trust falls to 43
Propose, pot 129	their minimum is 33, you keep 96	their minimum is 38, you keep 91
Respond, pot 139	they offer 64	they offer 56
Total	201	147

One rejection cost about 54 points in three rounds: the 41 you refused, 5 more on your proposal, and 8 less on their next offer. And the damage continues for every remaining round with this partner.

Worked example 3: the lowball cliff

Level 10. Partner 1 starts at trust 30. You accepted their first offer, so trust is 34. You then proposed 24 of a 60 pot (40 per cent), which was accepted, and accepted their next offer, lifting trust to 38. Now the pot is 188. You are tempted to try 30 per cent.

Show the answer

At trust 38, this partner's minimum is about 34.8 per cent of the pot: 65 of 188. Following the method, you shave 3 points from your last 40 per cent and offer 37 per cent, 70. It is accepted and you keep 118.

Offer 30 per cent (56) and it is rejected. You get nothing from a 188 pot, and trust drops by 18, to 20. At trust 20 their minimum jumps to 42 per cent and their offers to you fall from about 34 per cent of each pot to 25 per cent. The next lowball is now even more likely to fail. This is why a fixed low offer that works at level 1 scores about 25 per cent at level 10.

Tips and traps

Levelling up

Higher levels bring more rounds and partners (three from level 7), bigger and more variable pots, lower starting trust, steeper trust effects and heavier penalties for rejections and lowballs. Adapt in three ways:

• Widen your margin early. At level 10, partners start with a minimum close to 40 per cent. Open at 40 and walk down only after acceptances.
• Protect trust in long blocks. With seven rounds per partner at level 10, every early point of trust affects several later rounds. Never reject.
• Expect to hold levels. The hidden minimum makes 98 per cent very hard; judge progress by value captured and by keeping Offers refused at zero.

1. Week 1
   • Two runs; accept every offer without exception
   • Open each proposal at 40% and note which offers were accepted
   • Check Offers refused is 0% and Value captured is above 85%
2. Week 2
   • Two runs; walk offers down 3 points after acceptances, up 6 after rejections
   • Say the trust value aloud before every proposal
   • Compare final trust per partner with the starting trust
3. Week 3
   • Two runs at your current level; target no rejections of your offers
   • Rework worked example 3 with a pot size you met
   • Compare Value captured with week 1

Check yourself

5 practice questions

1. In our Money Exchange, what happens when you reject a partner's offer?
2. When is your proposal accepted?
3. What does an accepted offer of 45 per cent or more of the pot do?

and 2 more

Create a free account to answerSign in

Sources

1. Güth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental analysis of ultimatum bargaining. Journal of Economic Behavior & Organization, 3(4), 367–388. doi.org/10.1016/0167-2681(82)90011-7
2. Oosterbeek, H., Sloof, R., & van de Kuilen, G. (2004). Cultural differences in ultimatum game experiments: Evidence from a meta-analysis. Experimental Economics, 7(2), 171–188. doi.org/10.1023/B:EXEC.0000026978.14316.74
3. Engel, C. (2011). Dictator games: A meta study. Experimental Economics, 14(4), 583–610. doi.org/10.1007/s10683-011-9283-7
4. Camerer, C. F., & Thaler, R. H. (1995). Anomalies: Ultimatums, dictators and manners. Journal of Economic Perspectives, 9(2), 209–219. doi.org/10.1257/jep.9.2.209
5. JobTestPrep. Free pymetrics games online practice and guide (candidate-preparation site; descriptions are unofficial). jobtestprep.com
6. XLRI. XAT 2026 syllabus and exam pattern. xatonline.in

Put it into practice: Money ExchangeSign in to try it