Public procurement is big business. Various estimates place its average value between 15 and 20% of GDP, and for some countries the number is much bigger. Optimising public procurement procedures is therefore a first order policy issue. The benefits of open, competitive auctions have been widely documented by economists with respect to a number of different markets. When referring to the effects of government procurement, however, the praise for open, transparent auctions goes well beyond that of competitive outcomes. Many scholars formerly regarded open competition as a crucial ‘preventative tool’ that, when combined with transparent rules, could limit government discretion and its abuse. In addition, the objective of a single European market has pushed the European Commission to enact Directives that limit public buyer discretion and push for open auctions in the hope that they could foster cross-border trade in public procurement markets. Typically, discretion in public procurement is limited through the use of thresholds, where works with a value above a given threshold have to be rigidly awarded through an open auction and works below this threshold can be awarded through a restricted auction, where the buyer has some discretion in terms of who (not) to invite to bid.
Alternatively, current research on transaction costs, contract theory and procurement has identified numerous drawbacks to using open auction in public procurement, particularly in complex transactions. In open auctions, limits to contracting and enforcement have been shown to have negative effects on procurement outcomes, particularly if important quality dimensions are not sufficiently protected by the credible threat of a contractual remedy (Spulber 1990, Manelli and Vincent 1995). Discretion may then help, rather than harm, as it allows incomplete contracts to be complemented by dynamic informal governance mechanisms typical of the private sector, such as long-term relationships and reputation (Bannerjee and Duflo 2000, Malcomson 2013).
Edward Banfield and Steven Kelman recognised these conflicting forces early on in their academic work (Banfield 1975, Kelman 1990) and the latter played a key role in reforming US procurement rules when serving as Administrator of the Office of Federal Procurement Policy during the first Clinton administration. The Federal Acquisition Streamlining Act of 1994 and the Federal Acquisition Reform Act of 1995 substantially increased flexibility and discretion in US procurement. However, some legal experts are now arguing that both accountability and performance have fallen in the US in recent years (e.g. Yukins 2008), claiming there is little evidence in favour of the increase in government discretion and in flexibility in public procurement brought about by these reforms. The EU has been going in the opposite direction.
Since administrative rules that try to prevent corruption by limiting ex ante discretion also make it difficult for honest and capable public managers to use these important mechanisms, the positive and negative effects of discretion will always co-exist to at least some degree (Banfield 1975). Whether discretion should be reduced or increased in a given institutional environment is therefore a purely empirical question. It is hard to be in favour of or against a change in these rules and thresholds without some robust empirical evidence on their effects in the current setting. Surprisingly, these debates are instead typically based on no serious empirical evidence at all. At most, back of the envelope calculations of the purely administrative costs of adopting different procurement mechanisms are mentioned, which even if correct, are misleading because they are likely to be negligible relative to the effects on overall procurement outcomes and on the accountability of the public sector.1
Our study and results
In a new paper, we measure the effects of increased buyer discretion on a set of public procurement outcomes (Coviello et al. 2016). We exploit a threshold determined by the Italian procurement regulation, such that works with a value above the threshold have to be awarded through an open auction in almost all cases. Works below the threshold can be run more easily through a restricted auction, where the buyer has discretion in terms of who (not) to invite to bid. To do so, we run a regression discontinuity design analysis to document the causal effect of increasing buyers’ discretion on procurement outcomes in a large database for public works in Italy.
Our identification strategy relies on the assumption that the value of the project (i.e. the auction starting value representing the reserve price for the public buyer running the auction) is not perfectly manipulated around the discontinuity threshold. We test this assumption using graphical and statistical tests discussed by McCrary (2008) and Lee (2008), and focus on the sample of projects for construction works that do not show sorting around the threshold. In contrast, we drop from our sample road works where bunching around the threshold appears to be a problem. We further select our sample using the procedure suggested by Imbens and Kalyanaraman (2012).
Our main finding is that increased discretion (i.e. our treatment) causes a significant increase in the probability that the same firm is awarded a project repeatedly by the same public buyer. In an attempt to understand how to interpret this finding better, we analyse the impact of discretion on observable auction outcomes. In our main sample, we find that discretion has no effect on ex ante auction outcomes (number of bidders, rebates, size of the winners, distance of the winner from the public buyer) and on most of our ex post performance measures (i.e. duration of the works, monetary renegotiations). We find some evidence that discretion may increase delays in the delivery of the works. However, this evidence was not robust. In a closer neighbourhood of the discontinuity threshold (i.e. a smaller sample), we found instead evidence that in the environment we study the positive effects of discretion may slightly dominate the negative ones. Discretion appears to reduce the total duration of works – leading to the selection of larger (incorporated) firms, which typically have better quality control systems, reducing the number of firms submitting bids, saving administrative costs associated to bid screening, and not affecting other outcome measures, like the winning rebate, cost overrun and the probability that the project is awarded to a local firm.
To better interpret this set of findings, we also explore the relationship between projects' past and future delays in delivery, and winners' past and future incumbency. We run a propensity score matching analysis and find that contractors who have won in the past and are chosen again systematically deliver current works faster. In addition, we find that contractors characterised by better past performance are more likely to win other auctions. These estimates are sizeable and statistically significant for contracts with a value below the 300,000 euro threshold, where the law allows public buyers to use more discretion, in particular auctions restricted to a minimum number of invited bidders. These additional findings reinforce the case for our interpretation that positive productive relationships may dominate negative corrupt relationships in our sample.
Taken together, these results can be interpreted as coherent with the theoretical framework of Calzolari and Spagnolo (2009), explaining the interaction between competitive auctions and long-term relationships, and with the evidence in Bandiera et al. (2009) that suggests, for public procurement of goods and services in Italy, that corruption is not higher for more autonomous public buyers or those with higher discretion, while the prices they pay are significantly lower than average. The results are also consistent with Kelman (1990) and Banfield (1975) who argued earlier that a certain dose of ex ante discretion, coupled with ex post performance checks, is essential to good public management, even if it comes at the cost of a small loss in accountability. The specific empirical results, however, need not generalise to other, less hyper-regulated countries than Italy, where it may well be the case that the negative effect of discretion currently dominate the positive ones, suggesting that more rigidity or better performance checks are needed. Good data and decent econometric evidence are necessary to assess these important trade offs in every single country. Too bad that with few exceptions (among which are Italy, the US and Brazil) countries - even the richest ones in the EU - do not let researchers access, or do not even collect the basic data needed for ex-post performance checks or to assess the important trade off we have discussed in this piece.
Bandiera, O, A Prat, and T Valletti (2009), “Active and Passive Waste in Government Spending: Evidence from a Policy Experiment”, American Economic Review, 99 (4), 1278- 1308
Banerjee, A V, and E Duflo (2000), “Reputation Effects And The Limits Of Contracting: A Study Of The Indian Software Industry”, The Quarterly Journal of Economics, 115 (3), 989-1017
Banfield, E C (1975), “Corruption as a Feature of Governmental Organization”, Journal of Law and Economics, 18 (3), 587-605
Calzolari, G and G Spagnolo (2009), “Relational Contracts and Competitive Screening”, CEPR Discussion Paper 7434
Coviello, D, A Guglielmo, and G Spagnolo (2016), “The Effect of Discretion on Procurement Performance”, CEPR Discussion Paper 11286
Imbens, G W, and K Kalyanaraman (2012), “Optimal Bandwidth Choice for the Regression Discontinuity Estimator”, The Review of Economics Studies, 79 (3), 933-959
Kelman, S (1990), Procurement and Public Management, AEI Press, Washington
Lee, D S (2008), “Randomized Experiment from Non-Random Selection in U.S. House Elections”, Journal of Econometrics, 142 (2), 675-697
Malcomson, J M (2013), “Relational Incentive Contracts”, in R Gibbon and J Roberts (eds.) Handbook of Organizational Economics, Princeton University Press, Princeton, NJ
Manelli, A M, and D R Vincent (1995), “Optimal Procurement Mechanisms”, Econometrica, 63 (3), 591-620
McCrary, J (2008), “Manipulation of the Running Variable in the Regression Discontinuity Design: A Density Test”, Journal of Econometrics, 142 (2), 698-714
Molander, P (2014), “Public Procurement in the European Union: The Case for National Threshold Values”, Journal of Public Procurement, 14 (2), 181-214
Spulber, D F (1990), “Auctions and Contract Enforcement”, Journal of Law, Economics and Organization, 6 (2), 325-44
Yukins, C R (2008), “Are IDIQs Inefficient? Sharing Lessons with the European Framework Contracting”, Public Contracts Law Journal, 37, 545-568
 In Sweden, for example, a procurement inquiry by the government suggested increasing the threshold for direct (non-competitive) contract awards from about 300,000 SEK to 600,000 SEK. The Director of the Swedish Social Insurance Inspectorate presented a report arguing that this would reduce competition and increase costs for the public sector (see Molander 2014). The procurement inquiry contended that these effects would be overshadowed by reduced transaction costs and more flexibility, and that the report used a limited data set and was mostly based on rough approximations.