Standard double auctions perfectly mimic the neoclassical idea about the functioning of markets. The efficiency of the market outcome and speed of adjustment towards equilibrium have been studied in the literature to validate economic expectations. However, only a few real world examples outside the financial sector exist. In 2000, Germany implemented a sealed‐bid double auction mechanism for trading milk quota. The two main features of this mechanism are: (1) a sealed‐bid double auction that produces excess demands that are covered by state reserves free of charge, and (2) a variable price band that is used to exclude price bids above a certain range. To study these regulations a sealed‐bid double auction experiment is simulated and run with students. It is shown that the regulations lead to significant losses in welfare that are caused by direct effects and by an imperfect adjustment of individual bidding behaviour. Further, learning effects throughout the experiments appear to be very limited. Thus, complex auction mechanisms may need to be thoroughly tested before being introduced in the real world.