**3. Public Works in Italy and the Veneto Region—Sample Data and Descriptive Statistics**

The present analysis considers the time elapsing between the conclusion of a project's design phase and the beginning of the works, that is, all the time it takes to complete the contract awarding procedure.

This enables us to isolate the phase in the timeline that is most affected by the norms and regulations governing the contract awarding process. Project design may be affected by technical and managerial issues, while the construction phase may come up against various problems, from design errors to a workforce's low productivity or lack of specialization.

We collected an original database from the Italian State's General Accounting Department covering 19,117 public works, planned and implemented between 1999 and 2018 in the Veneto Region in north-east Italy. As in similar Italian analyses [40], we chose to include unfinished works too. Otherwise, the sample could be biased by an over-representation of the more efficient (or smaller and/or more straightforward) projects. The database includes the following information:


The database is part of the Open Data on Italian public economy and finance collected by the State's General Accounting Department. Data on the single works are provided by the work's managers of each public contractor. So, often data are entered incomplete. We excluded projects for which we had incomplete data concerning the costs and each phase of the timeline, thus reducing the sample to 4781 public works. As shown in Figure 1 and Table 1, 55% of the sample (i.e., 2625 projects) are extraordinary maintenance works, while new builds account for only 22% (1047 projects). Renovations of existing buildings involving a change of use amount to less than 9% and restoration work on historical buildings about 6.6%.


**Table 1.** Proportions of different types of project.

As shown in Table 2, the incomplete data we excluded from the sample has substantially the same proportion of the selected sample for the attribute "Type of project," with the clear predominance of extraordinary maintenance and new builds among other.

**Table 2.** Proportions of different types of projects of the incomplete and excluded data.


We identified 33 categories of works (see Table 3 and Figure 2) but seven of them accounted for more than 70% of the sample. Road infrastructure is the largest category with 1365 works, followed by school buildings and buildings for social activities. Social infrastructure (urban and neighborhood theaters, libraries and cultural centers) make up 11.15% of the sample. These proportions are consistent with the trend for public works in northern Italy. Countrywide, the availability of public infrastructure varies enormously between north and south [41–43], so drawing comparisons on the whole of Italy would be misleading.



As for the previous feature, the category of works also has the same structure in the selected and the incomplete, excluded data (see Table 4). The latter also has the same seven prevalent categories, with mostly the same proportion of the selected data.


**Table 4.** Proportions of categories of works of the incomplete and excluded data.

The contract awarding procedure takes 271 days on average. In other words, about nine months go by between the completion of the project design phase and the start of building work on site (Table 3). This time varies enormously, from just one day to about ten years. According to the Agency for Territorial Cohesion (Agenzia per la Coesione Territoriale) [40], the average time to assign works worth less than €1 million is about 5 months, so the Veneto Region seems to perform poorly compared with the national average, despite the public authority managing the procedure being reasonably well organized [44].

**Figure 2.** Categories of works (frequency 1 ×

1000).

We used "population" as a proxy of the dimension of the public authority awarding the contract, assuming that it would be more structured and specialized, the larger the population it serves [4,37]. This may indicate the skills available for managing the contract awarding process and consequently of the time needed to complete the procedure. As shown in Table 5, the average population is 426,650, which is higher than that of any of the provincial capitals in the region. Although most of the contracts were awarded by local authorities (85.30%, see Table 6), many of them (about 9%) were promoted by territorial entities, such as Energy Service Companies (ESCOs) and utility consortia that cover larger territories than provinces. The Veneto Regional Authority and other regional bodies account for just 3% of the contracts awarded and national bodies (Ministries or the Italian Central Government) for just 9 contracts awarded in the Veneto Region during the period considered. There is marked variability in the 'local authorities' category, which reflects the different organizational capabilities of local government bodies.

### **Table 5.** Descriptive statistics of cardinal variables.



**Table 6.** Contracts awarded by public authorities in the sample data.

### *The Threshold for Abnormal O*ff*ers—Clustering the Sample Data*

Given the variability of the sample and the different requirements for awarding contracts worth more or less than €1 million, we clustered the sample into two groups based on this €1 million cutoff.

The clusters confirm that small-scale works make up most of the sample, while the projects and works costing more than €1 million accounts for just 11% (see Table 7).

**Table 7.** Descriptive statistics for clusters of contracts worth more or less than €1 million.


Larger-scale works refer to a broader territory, although smaller works are also awarded on average by public bodies larger than the local authorities. The relevance of the €1 million contract value threshold emerges from the time taken to award the contracts, which is 244 days (about 8 months) for works worth less than €1 million but 480 days (16 months) for those worth more, that is, it takes twice as long to complete the contract awarding procedure for the latter.

Looking at the categories of works, nine categories account for more than 70% of the works in both contract value clusters but the incidence of the various categories differs significantly between the two. As shown in Table 8, "Road infrastructure" makes up 30% of the works worth less than €1 million but only 16% of the costlier contracts. "Schools and buildings for social activities" and "social infrastructure" account for about 20% and 12% respectively of the cluster. For the works costing more than €1 million, the "Road infrastructure" and "Soil protection" are the largest categories (both making up more than 16% of the sample), followed by "Schools and buildings for social activities" and "Housing." This latter category accounts for less than 11% of the costlier projects and just under 6% of the works costing less than €1 million. The interpretation of the different incidence of each category in the two clusters is contentious. For instance, projects relating to soil protection and water resources tend to involve a broad territory and expensive activities but we might have expected an incidence of small-scale maintenance works in the "Housing" category higher than a mere 5.7%. Projects relating to "Social infrastructure" and "Sports, entertainment and free time activities" are usually on a neighborhood scale, so they are unsurprisingly more relevant in the cluster of projects worth less than €1 million.


**Table 8.** Categories of works by cluster (works costing more vs less than €1 million).
