Of course, a cost-minimizing government may be tempted to abandon the policy ex post, i.e. to select the lowest bidder - even if it is a foreign firm whose bid is not sufficiently low to satisfy the requirements of the preference policy. In the numerical example given above, the preference margin of 20 per cent was tailored to maximise rent extraction given the cost differences between firms, while ensuring that procurement was still from the low cost source. This may not be possible if a uniform preference margin is applied to several sectors where cost differences are not the same, or the preference margin has to be chosen for a particular sector before the government can observe cost differences. Say the preference margin is again announced to be 20 per cent, foreign costs in a particular sector are still 100 but domestic costs are 115. In this case, a cost-minimizing government would prefer to abandon its preference policy, which would force it to buy from the more expensive domestic firm, and buy from abroad. But for the policy to be credible, the government must in certain cases award the contract to a domestic firm even when it is not the lowest bidder. Though this may raise the costs to the government in specific instances, it will have the effect of inducing lower bids in general by firms with a cost advantage.11
Perhaps the most significant contribution of this literature is to make us aware that attempts to estimate the cost of procurement preferences without taking into account their effects on bidding behaviour may produce biased estimates. Thus a zero preference should not be used as a benchmark with which to evaluate the welfare effects of procurement policies in the same way that a zero tariff is used to evaluate the welfare effects of trade policies.
(In the McAfee and McMillan model, the foreign industry is assumed to have a stochastic comparative advantage over the domestic industry. In the simplest case, the extent of cost variation among domestic firms and foreign firms is the same, but foreign firms have on average lower costs than domestic firms. Favouring high cost firms raises the probability that a high cost firm will win. But it also increases the competitive pressure on the low cost firms, forcing them to bid lower. The former effect tends to higher procurement costs and the latter tends to lower it. The resolution of this tradeoff always involves some preference being given to the high-cost firms.