In a free market, the creation of hospitals, schools, sports and public residential facilities, requires the expertise-and possibly the capital-of the private sector. The traditional contract, in which the public administration pays private operators to make or maintain buildings and services, is flanked by public private partnership, in which the private operator is usually delegated to carry out the entire process receiving a fixed fee. For years, governments and administrations have been incentivized to use this kind of contract, assuming that it would increase the building qualities and reduce the risk of higher expenses. Empirical evidence refutes this assumption, and this can be caused by to the so-called moral hazard of the private operator. One of the main problem in public private partnership is the difficulty to define an optimal risk allocation, as there no formulas exist to simulate the performance of the contract in advance. In this paper, Evolutionary Algorithms are used to compute an optimal specifications document, while, at the same time, foreseeing an optimal effort in work. Experimental results clearly demonstrate the feasibility of this approach, also helping the public administration to check if their knowledge is sufficient to structure an efficient specifications document.