ABSTRACT
Public procurement across the European Union (EU) accounts for approximately 14% of GDP. The EU reports member states’ procurement execution through the Single Market Scoreboard. One of these performance measures quantifies procurement decision speed, which assumes if member states reduce the time it takes to select vendors, there will be a reduction in transaction costs (increased efficiency). This research empirically evaluates procurement decision speed across a variety of government settings and procurement procedures to determine if decision speed, as defined by the Scoreboard, is a valid measure of performance. Findings suggest that decision speed as an efficiency performance measure lacks informational context; masks important underlying variation; and structurally omits the influence of important variables that contribute to variations in decision speed. Consequently, the utility of the Scoreboard measures of performance is called into question.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. This discussion of the Single Market relies heavily on websites such as http://ec.europa.eu/growth/single-market/and https://ec.europa.eu/growth/single-market/public-procurement_en as well as http://ec.europa.eu/internal_market/scoreboard/performance_per_policy_area/public_procurement/index_en.htm (accessed April 21, 2018).
2. Strict reproduction of the SMS approach will further diminish usable N to 696,292 for the years 2013–2018.
3. According to the EU, any business can submit an open tender. While the minimum time limit for submission of tenders is 35 days from the publication date of the contract notice, a published prior information notice can reduce this to 15 days.
4. TED reports ten different CAE categories that have been collapsed to five.
5. Separate ANOVA tests confirm that even when accounting for variance heterogeneity and sample size differences, there are statistical differences in mean decision speeds across the contract types (and CAEs), significant at the p <.000 level.
6. The European Commission (EC) (Citation2019, p. 9) states that “very lengthy procedures are bad because they are expensive and cause uncertainty for both the public buyers and companies.”
7. The period is confined to 2013 because of changes instituted based on 2012 Directives.
8. The 730-day cutoff pool had 473,044 cases compared to 467,931 cases for 365 days.
9. Respectively, Model 1 Cox & Snell R² =.051, Model 2 =.001; Model 1 Nagelkerke R² =.085, and Model 2 =.002. Following Tabachnik and Fidell (Citation2013), Model 1 R² =.06, and Model 2 R² =.001. Further, the Hosmer-Lemeshow tests both show non-significant test results (for both models; p = 1.0). Model 2 results are χ²(4) =.958, p <.916.