Favouritism and corruption in procurement auctions (Q6173735)

The authors model favoritism and corruption in a procurement auction where two firms compete to get a contract. The firms are characterized by their inefficiency parameter \(\theta_{i} \in [\underline\theta,\overline\theta]\) \((i=1,2)\) -- that is, lower \(\theta_{i}\) corresponds to more efficiency -- which are independent and identically distributed over \([\underline\theta,\overline\theta]\) with strictly positive density. The quality of the service delivered by the winning firm is described by \(q\in [\underline q,+\infty)\). The cost function \(c(q,\theta)\) of the firms is assumed to satisfy \(\partial_{1} c > 0\), \(\partial_{2} c > 0\), \(\partial_{11} c > 0\), \(\partial_{12} c > 0\) everywhere on \( (\underline q,+\infty)\times [\underline\theta,\overline\theta]\); for example, \(c(q,\theta) = q^{2}/2+\theta\) satisfies these requirements. The minimum required quality is \(k > \underline q\), an exogenous parameter. The reported quality is obtained as \(q + h_{i}\cdot b\) where \(q\) is the true realized quality, \(b\) is the bribe paid, and \(h_{i}\) is the known quality manipulation index. It is assumed that \(1 = h_{2} \leq h_{1} \leq \overline h\) -- that is, firm 1 is favored -- and \(0 < \partial_{1}c(\underline q,\overline \theta) < 1/\overline h\) and \(1 < \partial_{1}c(k,\underline \theta)\); that is, underperformance and bribing is incentivized at the minimum required quality level. The authors argue this is representative for emerging economies where the marginal cost of supplying good quality is high. The procurement auction is modeled as a three-stage game: \begin{itemize} \item[1.] The firms quote their bids according to \(p_{i}\colon [\underline\theta,\overline\theta]\rightarrow [0,+\infty)\) \((i=1,2)\) to win the contract, the lower bidder being the winner. \item[2.] The winning firm \(i\) chooses the quality of execution \(q\) and delivers the service. \item[3.] If \(q < k\) the bribe \(b\) is chosen, else no bribe is paid. The firm gets paid if and only if \(q + h_{i}\cdot b\geq k\), in which case its profit is \(p_{i}(\theta_{i})-c(q,\theta_{i})-b\); else it does not get paid hence its payoff is \(-c(q,\theta_{i})-b\). \end{itemize} The authors calculate the ex-post equilibrium bribe at stage three; show that there exists a unique optimal quality \(q\) in stage two, which satisfies \( \partial_{1} c(q,\theta_{i}) = 1/h_{i}\); and compute the Bayesian Nash equilibrium bids \(p_{i}\) under no favoritism (\(h_{1}=1\)) and with favoritism (\(h_{1}>1\)). It is shown that in this model, less efficient firms supply lower quality; firm 2 cannot win the auction if \(\theta_{2}\) is sufficiently high, hence favoritism is discriminatory and reduces competition; \(p_{1}(\theta) < p_{2}(\theta)\) holds for all \(h_{1} \in (1, \overline h]\) and \(\theta \in (\underline\theta,\overline\theta)\), that is favoritism leads to more aggressive bidding and inefficiency. The model is solved explicitly for \(c(q,\theta) = q^{2}/2+\theta\) and uniformly distributed \(\theta_{i}\). It is observed that the expected welfare can increase or decrease as a function of \(h_{1}\), the intuition behind the former being that in case the decline in quality does not result in material decline in utility the cost reduction may increase overall welfare.