Governments in developed countries spend roughly one-quarter to one-third of their total expenditures through the public procurement market, purchasing construction works, goods, and services from the private sector. The scope of public procurement ranges from simple items such as office paper supplies to highly complex projects, including software systems and large-scale construction works such as highways. Many of these complex projects – particularly in transportation infrastructure – face a high risk of unexpected circumstances that may lead to cost overruns.
Such cost overruns and project delays are common worldwide. The list of examples is long. In the summer of 2025, for instance, the rising costs of the Federal Reserve’s renovation became a major media story after the US president publicly criticised the chair of the Federal Reserve (Rugaber et al. 2025). The Sydney Opera House and the Channel Tunnel – the longest underwater rail tunnel in Europe, connecting the UK and France – are other well-known examples of public projects that suffered severe cost overruns. For further discussion of cost overruns in public projects, see Flyvbjerg et al. (2003, 2004) and Engerman and Sokoloff (2004).
The inherent incompleteness of public procurement contracts and the frequency of post-award modifications raise the question of how much scope there should be for contract modifications. Excessive discretion granted to procuring authorities in post-award renegotiation can open channels for corruption, while overly restrictive rules may render the procurement process too rigid, ultimately reducing efficiency. This mirrors the broader debate over the extent of discretion granted to procuring authorities during the contract award stage (Baranek and Titl 2024b, Bosio 2021, Shleifer et al. 2020, Spagnolo et al. 2022, Titl and Geys 2019).
In our recent paper (De Jaegher et al. 2025), we study the impact of eased renegotiation rules on procurement outcomes. We focus on the transposition of an EU directive into Czech law, which introduced greater flexibility to modify already awarded contracts. Prior to this reform, Czech law prohibited ex post contract modifications unless a natural disaster, such as a flood, occurred. When cost overruns occurred, procurers faced two options: (i) launch a new procurement procedure, which was administratively costly and could create logistical issues with multiple firms working on the same construction site, or (ii) proceed with a restricted, non-published negotiation procedure, permitted only in rare circumstances such as emergencies. This rigid pre-reform environment, where renegotiation was virtually impossible, followed by a major easing of renegotiation rules provided a natural experiment for studying the effects of more flexible renegotiation rules.
Unsurprisingly, the reform led to a sharp increase in renegotiations. Among construction contracts awarded after the reform, the share of renegotiated contracts stabilised at around 40%, compared to virtually zero previously. More interestingly, the value of winning bids decreased significantly, as firms – anticipating gains from renegotiation in the event of cost overruns – strategically adjusted their bids and bid more aggressively. The decline in winning bids is shown in Figure 1, which plots coefficients from a dynamic difference-in-differences regression of winning bids on the interaction between the post-reform dummy and quarterly dummies (for details of the specification, see De Jaegher et al. 2025). By contrast, the same analysis of average final prices shows no change relative to the pre-reform period. This suggests that the gains from renegotiation were competed away by lower bids.
Figure 1 Quarterly effects on winning bid
Notes: The figure shows estimated quarterly differences in winning-bid ratio between construction and control contracts over time. The effect is normalised to 0 in Q1 2016. In the dynamic difference-in-differences approach, we control for 4-level CPV codes industry fixed effects, selection methods, and public procurement procedure.
To assess the fiscal impact of the reform on the public sector, we weight each contract by its estimated cost so that every euro of procurement carries equal weight in the analysis. Contrary to the null effect on the average contract, we find that the reform increased overall public procurement expenditures by just over 2%. This increase is driven almost entirely by a handful of very large contracts, most of which were renegotiated.
Thus far, our analysis suggests that the reform was neutral in terms of costs for the sector, except for very large contracts. However, the final prices of initial contracts are hardly the only measure of costs for the public sector. Consider two contracts of the same type and with the same final costs (for example, two four-lane highways, each 10 kilometres long and delivered for €10 million): one requires two repairs worth €200,000 five years after delivery, while the other requires five repairs worth €500,000 in the same time span. All else equal, it seems reasonable to assume that the first highway is of higher quality.
Based on this idea (and following Baranek and Titl 2024a), we construct a continuous proxy for quality: the value of follow-up repairs in the first five years after the project was finished, expressed as a percentage of the engineering cost estimate, multiplied by minus one. This measure of quality is applicable only in the construction sector, which means we cannot conduct difference-in-differences analysis. To obtain at least suggestive evidence of changes in the quality of delivered procurement projects, we plot the development of quality over time as a time series in Figure 2. The figure shows a steady increase in the quality of delivered procurement projects in Czechia, with no observable change in the trend around the time of the reform.
Figure 2 Quality of public procurement projects over time
Notes: The figure shows the change in the quality of public procurement projects in the construction sector over time. The coefficients come from a regression of the quality measure on quarterly dummy variables. The effect is normalised to 0 in Q1 2016. We control for 4-level CPV codes industry fixed effects, selection methods, and public procurement procedures.
Overall, our findings suggest that easing renegotiation rules does not substantially change the outcomes of typical procurement contracts. Nevertheless, we are cautious about drawing strong policy recommendations. Allowing post-award modification of contracts likely imposes a higher burden for the parties who become more involved in renegotiation, thereby increasing transaction costs. Unfortunately, we were not able to quantify the extent to which the reform affected transaction costs connected to this greater flexibility. Stronger policy recommendations require a better understanding of the transaction costs that different policy regimes impose on both contracting parties.
One element of the reform that clearly worsened fiscal outcomes for the public sector was its retroactive clause. Contracts awarded in the three years before October 2016 (when the law came into force) were also made eligible for renegotiation under the new rules. These contracts had been awarded on the assumption that renegotiation was virtually impossible and therefore carried higher winning bids. Once renegotiation became permissible, these already inflated bids were pushed up further through price renegotiations. The retroactive application of the reform thus increased public expenditure. A straightforward way to limit these costs of the reform, while preserving the benefits of more flexible renegotiation rules going forward, would have been to apply the reform only prospectively.
References
Baranek, B, and V Titl (2024a), “The cost of political connections in public procurement”, The Journal of Law and Economics 67(2): 445–77.
Baranek, B, and V Titl (2024b), “Politicians on corporate boards and the cost of public procurement”, VoxEU.org, 30 November.
Bosio, E, G Spagnolo, and O Bandiera (2021), “Discretion, efficiency, and abuse in public procurement: A new eBook”, VoxEU.org, 30 November.
De Jaegher, K, M Šoltés, and V Titl (2025), “Easing renegotiation rules in public procurement: Evidence from a policy reform”, Journal of Public Economics 248: 105445.
Engerman, S L, and K L Sokoloff (2004), “Digging the dirt at public expense: Governance in the building of the Erie Canal and other public works”.
Flyvbjerg, B, M K Skamris Holm, and S L Buhl (2003), “How common and how large are cost overruns in transport infrastructure projects?”, Transport Reviews 23(1): 71–88.
Rugaber, C, J Boak and C Megerian (2025), “Trump and Powell bicker over Fed building renovations as president ratchets up pressure campaign”, AP News, 25 July.
Shleifer, A, S Djankov, E Glaeser, and E Bosio (2020), “Corruption in public procurement”, VoxEU.org, 19 November.
Spagnolo, G, A Guglielmo, C Lotti, and D Coviello (2022), “Procurement with manipulation”, VoxEU.org, 24 March.
Titl, V, and B Geys (2019), “Strategic behaviour in procurement allocations towards corporate political donors”, VoxEU.org, 19 January.