Both institutional economists preparing public policies and academic economists evaluating them keep the premise of perfect competition markets as reference, despite market failures are broadly acknowledged. Actually, a large bunch of economic research has explored deviations from the ideal-type (imperfect competition or information, positive or negative exter- nalities, poverty traps, bounded rationality...). Yet, this large literature keeps the ideal-type of perfect competition general equilibrium as the reference to which are compared more realistic models and toward which are supposed to tend actual economic organizations, thanks to the proposed public policies. Mainly, the virtual reference of perfect competition markets is considered as efficient, but not attained because of market failures: the aim of public policies is then to intervene to close the gap between actual market allocation and this supposed efficiency. The present article aims at understanding the characteristics of this dominant ideal-type, and specifically what is called efficiency. It analyzes: i. the way the concept of Pareto optimum is interpreted as an efficiency criterium; ii. the mechanisms through which the market process is supposed to lead to a Pareto optimal situation; iii. which one is selected amid the multiple potentially Pareto optimal situations. It allows to conclude to a false interpretation of efficiency of perfect competition, which is essentially a mechanism of gathering information by weighting individual preferences in proportion to their purchasing power. The use of perfect competition as reference and the limit of the concept of efficiency are illustrated through examples of public policies, notably green taxes.