This thesis investigates the behaviour of employees captured by centralised wage-negotiations. It is empirically shown that centralised institutions are correlated with compressed wages, which in turn accumulates several types of conflicts. A centralised unions’ association face similar conflicts of interest as within a plant union. The employees of different skills and crafts have distinct demands towards the employer, representing their importance for the final product. Within this bargaining, no person or group can raise specific demands without creating some sort of externality onto the other employees. Compressed wages lead to higher wages for the least productive employees and lower wages for the most productive employees. This in turn makes the employers demanding the latter better off, as the wage bill becomes lower with compressed wages. With centralisation a potential conflict arises within the employers’ confederation; between the skilled labour employers and the low-skilled labour employers, demanding the distribution of productivity among the labour.By letting the wages in a society be based on norms, I implement the described conflicts into a game-theoretic approach, similar to the Battle of the Sexes. In a game between a more than average productive employee and an employer demanding this specific labour, both players need to coordinate on a wage-contract. As the employee receives a lower wage with a centralised wage-contract, the employee would prefer if they coordinated on a decentralised contract. However, the employer has opposite preferences. He/she has a lower wage-bill in the centralised case and prefers on coordinating on a centralised wage-contract. If they cannot coordinate, they receive nothing. One may argue that the productive employee carries an opportunity cost. This cost is similar to what he/she would have made in a decentralised system, with negligible compression of wages. The game is played many times. The probability of the other player playing a specific contract is based on information the players gather through previously played stage-games, games with other players, word-of-mouth or other sources of information. By using this information both players calculate their best response strategy. If all employers are expected to play the strategy of centralised wage-contract, the employee’s best response strategy becomes to coordinate. When the players are expected to coordinate on this specific contract, and the expectations become self-enforcing, the strategy becomes a convention. This suggests that the most productive employees voluntarily participate in the centralised wage-bargaining, due to their best-response strategy. If the opportunity-cost change, the expectations of the players change. This could be caused by movements in the economic environment, such as economic depression, booms, high level of industrial conflict etc. This may modify the players’ best responses and the system may flip towards a decentralised wage-contract bargaining. Considering individuals being able to make random behavioural mistakes, some players will play non-best-responses. This theory suggests that only a few players not able to coordinate on the expected strategy may switch expectations from one convention to the other. An increase in the opportunity cost associated will impose enhanced pressure on the “satisfied” groups. In the centralised system, this implies the high-technological employers and the least productive employees. The convention becomes more accessible and tolerates fewer non-best responses. These non-best responses can be either stochastic or intentional. When the model is stochastic, the convention flips by the satisfied players playing “wrong”. This may seem unlikely. When the satisfied groups are able to coordinate, the system can only change through intentional non-best responses by the most dissatisfied groups. Equipped with this theory and the historical background of the Nordic countries, this thesis provides an explanation of the implementation of the centralised system in these countries. Further, this thesis discusses forthcoming difficulties in sustaining these institutions. First is the role of migration. Because of free trade-agreements of services within Europe, migrants are faced with fewer problems of moving across the borders. This may alter the contextual best responses by either the most productive employees or the least productive employers. Second is a change of equality preferences within the unions’ association. When unions accelerate their sense of actuarial fairness the system may be altered through intentional collective actions. However, by using theory of collective action, I find that the centralised system may persist, rather than imposing a conventional flip of best-response play. This is because the preference shock may not lead to collective action, but rather alter the distribution of power among the unions participating in the unions’ association. It has been argued by labour economists that the Nordic or Scandinavian model has been evolutionary rather than coerced politically. This thesis adds weight to the argument using models of evolutionary change through game-theory.