Back Matter
Author:
Aly Abdou
Search for other papers by Aly Abdou in
Current site
Google Scholar
PubMedClose
,
Olivier Basdevant
Search for other papers by Olivier Basdevant in
Current site
Google Scholar
PubMedClose
,
Elizabeth David-Barrett
Search for other papers by Elizabeth David-Barrett in
Current site
Google Scholar
PubMedClose
, and
Mihaly Fazekas
Search for other papers by Mihaly Fazekas in
Current site
Google Scholar
PubMedClose
Publication Date:
20 May 2022
eISBN:
9798400207884
ISBN:
9798400207884
Language:
English
Keywords:
public procurement; corruption; corruption risks; procurement cost; price differential; corruption risk; price impact; procurement contract; data collection methodology; Global
Download PDF (581.2 KB)

References

  • Auriol, E., Flochel, T., and, S., Straub, 2016, “Public Procurement and Rent-Seeking: The Case of Paraguay,” World Development Vol. 77, pp: 395–407. https://doi.org/10.1016/j.worlddev.2015.09.001.

    • Search Google Scholar
    • Export Citation

  • Bandiera, O., Prat, A., and, T., Valletti, 2009, “Active and Passive Waste in Government Spending: Evidence from a Policy Experiment,” American Economic Review Vol. 99(4), pp: 1278–1308. https://www.aeaweb.org/articles/pdf/doi/10.1257/aer.99.4.1278.

    • Search Google Scholar
    • Export Citation

  • Banerjee, A., Duflo, E., Imbert, C., Mathew, S., and, R., Pande, 2016, “E-governance, Accountability, and Leakage in Public Programs: Experimental Evidence from a Financial Management Reform in India,” NBER Working Paper #22803. http://www.nber.org/papers/w22803.

    • Search Google Scholar
    • Export Citation

  • Basdevant, O. and M., Fazekas, 2022, “How to Assess the Impact of Corruption Risks on Public Procurement: The Corruption Cost Tracker (CCT),” IMF How-To-Notes (to be published).

    • Search Google Scholar
    • Export Citation

  • Björkman, M., and, J., Svensson, 2009, “Power to the People: Evidence from a Randomized Field Experiment on Community-Based Monitoring in Uganda,” The Quarterly Journal of Economics, Vol. 124(2), pp. 735–69. https://doi.org/10.1162/qjec.2009.124.2.735

    • Search Google Scholar
    • Export Citation

  • Bosio, E., Djankov, S., Glaeser, E., and, A., Shleifer, 2020, “Public Procurement in Law and Practice,” NBER Working Paper #27188. https://www.nber.org/papers/w27188.

    • Search Google Scholar
    • Export Citation

  • Charron, N., Dahlström, C., Fazekas, M., and, V., Lapuente, 2017, “Careers, Connections, and Corruption Risks: Investigating the impact of bureaucratic meritocracy on public procurement processes,” Journal of Politics, Vol. 79(1), pp: 89–103. https://www.journals.uchicago.edu/doi/10.1086/687209.

    • Search Google Scholar
    • Export Citation

  • Chong, E., Klien, M., and, S., Saussier. 2015, “The Quality of Governance and the Use of Negotiated Procurement Procedures: Evidence from the European Union,” Institut d’Administration des Entreprises Discussion Paper #2015–3. http://www.chaire-eppp.org/files_chaire/chong-klien-saussier-2015.pdf.

    • Search Google Scholar
    • Export Citation

  • Coviello, D., and, S., Gagliarducci, 2017, “Tenure in Office and Public Procurement,” American Economic Journal: Economic Policy, 2017, Vol. 9 (3), pp. 59–105.

    • Search Google Scholar
    • Export Citation

  • Coviello, D., and, M., Mariniello, 2014, “Publicity Requirements in Public Procurement: Evidence From a Regression Discontinuity Design,” Journal of Public Economics, Vol. 109, pp. 76–100. https://doi.org/10.1016/j.jpubeco.2013.10.008.

    • Search Google Scholar
    • Export Citation

  • Coviello, D., Guglielmo, A., and, G., Spagnolo, 2018, “The Effect of Discretion on Procurement Performance.,” Management Science, Vol. 64(2): pp. 715–738. https://doi.org/10.1287/mnsc.2016.2628.

    • Search Google Scholar
    • Export Citation

  • Di Tella, R., and, E., Schargrodsky, 2003, “The Role of Wages and Auditing during a Crackdown on Corruption in the City of Buenos Aires,” Journal of Law and Economics Vol. 46(1), pp: 269–92. https://doi.org/10.1086/345578.

    • Search Google Scholar
    • Export Citation

  • Fazekas, M., and B., Tóth, 2017, “Infrastructure for whom? Corruption risks in infrastructure provision across Europe,” In G. Hammerschmid, G. Kostka, & K. Wegrich (Eds.), Governance Report 2016, Chapter 11 -- Governance of Infrastructure, (Oxford University Press, Oxford).

    • Search Google Scholar
    • Export Citation

  • Fazekas, M., Tóth, I., and, L., King, 2016, “An Objective Corruption Risk Index Using Public Procurement Data,” European Journal of Criminal Policy and Research, Vol. 22(3), pp: 369–97. https://doi.org/10.1007/s10610-016-9308-z.

    • Search Google Scholar
    • Export Citation

  • Fazekas, M., and, G., Kocsis, 2020, “Uncovering High-Level Corruption: Cross-National Objective Corruption Risk Indicators Using Public Procurement Data,” British Journal of Political Science, Vol. 50(1), pp: 155–64. https://doi.org/10.1017/S0007123417000461.

    • Search Google Scholar
    • Export Citation

  • Fazekas, M., and, J., Blum, 2021, “Improving Public Procurement Outcomes: Review of Tools and the State of the Evidence Base” World Bank Policy Research Working Paper # WPS 9690. http://hdl.handle.net/10986/35727

    • Search Google Scholar
    • Export Citation

  • Ferwerda, J., Deleanu, I., and, B., Unger, 2016, “Corruption in Public Procurement: Finding the Right Indicators,” European Journal on Criminal Policy and Research, Vol. 23, pp: 245–267. https://doi.org/10.1007/s10610-016-9312-3.

    • Search Google Scholar
    • Export Citation

  • Golden, M., and, L., Picci, 2005, “Proposal for a New Measure of Corruption, Illustrated with Italian Data,” Economics and Politics, Vol. 17(1), pp: 37–75. https://doi.org/10.1111/j.1468-0343.2005.00146.x.

    • Search Google Scholar
    • Export Citation

  • Hafner, M., Taylor, J., Disley, E., Thebes, S., Barberi, M., Stepanek, M., and, M., Levi, 2016, “The Cost of Non-Europe in the area of Organised Crime and Corruption: Annex II – Corruption. European Union,” RAND Corporation Research Report # RR-1483-EP. https://www.rand.org/pubs/research_reports/RR1483.html.

    • Search Google Scholar
    • Export Citation

  • Hyytinen, A., Lundberg, S., and, O., Toivanen, 2009, “Politics and Procurement: Evidence from Cleaning Contracts,” Helsinki Center of Economic Research Discussion paper #277. https://core.ac.uk/download/pdf/14912943.pdf.

    • Search Google Scholar
    • Export Citation

  • IMF, 2019, “Curbing Corruption,” IMF Fiscal Monitor.

  • Klasnja, M., 2016, “Corruption and the Incumbency Disadvantage: Theory and Evidence,” Journal of Politics, Vol. 77(4), pp: 928–42. https://www.journals.uchicago.edu/doi/10.1086/682913.

    • Search Google Scholar
    • Export Citation

  • Lewis-Faupel, S., Neggers, Y., Olken, B. A., and, R., Pande, 2016, “Can Electronic Procurement Improve Infrastructure Provision? Evidence from Public Works in India and Indonesia,” American Economic Journal: Economic Policy, Vol. 8(3), pp. 258—83. https://pubs.aeaweb.org/doi/pdfplus/10.1257/pol.20140258

    • Search Google Scholar
    • Export Citation

  • OECD, 2016, Preventing Corruption in Public Procurement, OECD Publishing, Paris. http://www.oecd.org/gov/public-procurement/publications/Corruption-Public-Procurement-Brochure.pdf.

    • Search Google Scholar
    • Export Citation

  • Olken, B., 2006, “Corruption and the Costs of Redistribution: Micro Evidence from Indonesia,” Journal of Public Economics, Vol. 90(4–5), pp: 853–70. https://doi.org/10.1016/j.jpubeco.2005.05.004.

    • Search Google Scholar
    • Export Citation

  • Olken, B., 2007, “Monitoring Corruption: Evidence from a Field Experiment in Indonesia,” Journal of Political Economy, Vol. 115(2), pp: 200–249. HYPERLINK https://www.journals.uchicago.edu/doi/full/10.1086/517935.

    • Search Google Scholar
    • Export Citation

  • Palguta, J., and, F., Pertold, 2017, “Manipulation of Procurement Contracts: Evidence from the Introduction of Discretionary Thresholds,” American Economic Journal: Economic Policy, Vol. 9(2), pp. 293–315. https://doi.org/10.1257/pol.20150511.

    • Search Google Scholar
    • Export Citation

  • Saussier, S., and, J., Tirole, 2015, “Strengthening the Efficiency of Public Procurement,” Notes du Conseil d’Analyse Economique, Vol. 22(3), pp. 1–12. https://doi.org/10.3917/ncae.022.0001.

    • Search Google Scholar
    • Export Citation

  • Schwartz, G., Fouad, M., Hansen, T., and, G., Verdier, eds, 2020, Well Spent, International Monetary Fund (Washington DC). https://doi.org/10.5089/9781513511818.071

    • Search Google Scholar
    • Export Citation

  • Zamboni, Y., and, S., Litschig, 2018, “Audit Risk and Rent Extraction: Evidence from a Randomized Evaluation in Brazil,” Journal of Development Economics, Vol. 134, pp. 133—149. https://doi.org/10.1016/j.jdeveco.2018.03.008

    • Search Google Scholar
    • Export Citation

Appendix I. Data Collection Methodology and Descriptive Statistics

The data is entered by the procuring organizations into standard reporting forms through government-run electronic procurement platforms. For every observed tender, we have information from contract award announcements as publication is always mandatory, while information from calls for tenders may not be published under specific circumstances.

We developed an automated web crawler to scrape data from each of the official sources. The methodology is composed of the following steps. We use Python (together with other programs such as Java) to collect HTML, XML, and CSV outputs from the sources. As noted above, the collection of these data requires from countries having an open-data practice of disclosing procurement contracts on a web portal. All countries of this pilot are already advanced in their public disclosure of procurement data in rather standardized formats. We then transpose each publication from its original format into a uniformly structured data template, including converting structured text to standard data types (numbers, dates, enumeration values), and cleaning the database from nonsensical values and/or ballast information.

We then link all the information which describes a tender, where a tender ideally begins with one Call for Tenders (or more) followed by one Contract Award (or more) and completed by a series of payments (or contract completion announcement). We also take into account if any modifications or cancellations occur to the tender at any point during the process. After successfully linking related publications, we reconcile all linked data records to create a single best image of a public tender covering its whole tendering cycle (importantly, this is the step where we reconcile conflicting information or fill in empty fields if available in a related notice).

The data is then cross-checked manually with the publications’ sources. Once checked manually, we standardize buyer s’ and suppliers’ names. For Indonesia and Uganda, we also implemented a multi-step token-based string-matching algorithm for observations with missing tender product codes. We used a combination of tender title, lot title, and/or product description to match them with relevant product codes. For full technical documentation and codes see: https://github.com/digiwhist/backend.

Table 6.

Descriptive Statistics of Red Flags and the CRI

article image
Sources: Authors computations.
Figure 2.
Figure 2.

Observed Relative Price Distributions by Country

(Histogram, thousands of contracts, left axis, relative price on the x-basis)

Citation: IMF Working Papers 2022, 094; 10.5089/9798400207884.001.A999

Sources: countries authorities and authors computations.

Appendix II. Regressions Underpinning the Validity of the CRI

As our corruption definition implies lack of competition favoring a connected bidder, we consider two key outcomes of corrupt contracting being single bidding (lack of competition) and supplier contract share (repeatedly favoring the same company). Other indicators of corruption risks are related to these two outcomes in order to establish that they can serve as tools for corrupt contracting.

Assessing the relevance of red flags through indicators of uncompetitive bids. We first tested the validity of the red flags, by running a logit model where single bidding is the dependent variable and the red flags are explanatory variables plus including a host of economic controls, following the methodology of Fazekas and Kocsis, (2020).

Zi=α+Σj=17βjXj,t+Σj=1nβjCj,t+εi(1)

Where Zi is the Log of the probability of the ith contract to be awarded by single bidding, X is the vector of red flags, and C control variables (contract values in real terms, types of market—based on assigned product codes, buyer types, and tender year). The results showed positive and significant coefficient for each of the seven indicators, thus suggesting that taken altogether they do explain how tendering process become less competitive, thus opening the door to corruption.

Assessing the relevance of red flags through indicators of supplier contract share. We also tested the validity of the indicators using suppliers share (S) in total procurement spending by procuring body. In a similar equation as the one above (although not in a logit model) we tested if the red flags were positive and significant in explaining a higher share of supplier contract. The regressions were limited to suppliers with more than 4 contracts per year. We have robustness test with larger (10+ contracts) as well as smaller (3+ contracts) entities, results don’t change much. The reason excluding very small entities is that they trivially have high concentration, e.g., if you have a 1 contract entity it will trivially have 100 percent concentration.

Si=α+Σj=17βjXj,t+Σj=1nβjCj,t+εi(2)

We test the validity of the five remaining red flags (on top of single bidding and winner contract share) in these two sets of validity regressions. The precise definition of each red flag and the evidence for their validity is discussed one by one below.

For the two indicators related to time periods (submission of bids and decision), we split the data into deciles and mark as risky the deciles that significantly increase the probability of single bidding. One difficult aspect with these two indicators is that because of country-specific market and institutional conditions, the notion of an inadequate period for either submitting bids or deciding on the winning bidder varies. Yet, we want to avoid that our indicators are based on expert judgments, and instead we rely on hard data.

  • To “let the data speak,” we ran two types of regressions (see above), where the dependent variable is either single-bidding contracts (in a logit model), or the share of procurement contract value awarded to a given bidder (in fixed-effects OLS regressions). Both models are run with two alternative explanatory variables (as well as a host of controls): the advertisement period (in the case of tends submissions), and the period for making the decision.

  • In both cases, we refer to the period as the deviation from a norm. To identify the norm, we split periods (for both submission and decision) into deciles. We then identify in each case a decile that will serve as a norm, i.e., associated with more likely open and competitive bidding process, either because it has the longest time period of all deciles (in the case of submission) or because it has the value closest to the average decision period.

  • Using the two regression models mentioned above, we search for significant and positive coefficients of the difference between the norm and the time period of each decile, as such a coefficient would indicate a higher probability of single bidding (in the first model) and/or higher spending concentration (in the second model).

  • When using these two regression models, we are not looking for the “best fit” or causal identification but instead an indication that these indicators are correlated across a range of market contexts, as corruption is expected to manifest itself through a wide range of indicators and/or change over time, as discussed earlier in the paper.

Results confirm that in most cases short durations are associated with a significant impact on single bidding contracts and higher share of procurement spending on specific bidders (Table 9 for single bidding, and Table 10 for spending concentrated on specific bidders). The results clarify what categories of time periods can be considered as corruption risks (Table 7 and Table 8).

Table 7.

Submission Period Threshold Red Flags by Country

article image
Sources: Authors’ computations.

Not including medium risk for Romania was a choice coming from the validity regressions shown in Table 9 and 10. In essence, there was no improvement to be made from adding an intermediate risk category (the regression models did not improve).

Table 8.

Decision Period Threshold Red Flag by Country

article image
Source: Authors’ computations.

While 64 days could be seen, in absolute terms, as an excessive period for decision making in simple procurement processes. However, these periods are derived from regressions in Table 9 and 10. In both tables, medium risks (23–64 days) are associated with higher single bidding rate and higher spending concentration, so they could indicate risky behavior. Nevertheless, it could still be beneficial to shorten decision periods in general in Paraguay to lessen the administrative burden and make purchasing timelier.

Table 9.

Validation Using Single Bidding

article image
Source: Authors estimates. Regression includes controls for contract values, buyer type, market, and tender year. *** p<0.01, ** p<0.05, * p<0.1

See Table 7 for definition for each country.

See Table 3 for definition for each country.

See Table 4 for definition for each country.

Table 10.

Validation of Red Flags Using Supplier Contract Share

article image
Source: Authors estimates. Regression includes controls for contract values, buyer type, market, and tender year. *** p<0.01, ** p<0.05, * p<0.1

More than 4 contracts per year.

See Table 7 for definition for each country.

See Table 3 for definition for each country.

See Table 4 for definition for each country.

Administrative procedures can also be a red flag for corruption risks when they lead to the lack of competition. Similarly, to the issue we faced with time periods, assessing the non-openness of procedures without using experts’ judgments (and thus some degree of perception) is difficult. To do so we followed a methodology similar to the one described above for the duration indicators. We first identify all administrative procedure types related to public procurement (e.g., if there is a minimum number of bidders required), through text search in laws, secondary legislations, and tender documents. Following the logic of identifying risky categories introduced above for period length indicators, we run the same set of regressions with single bidding and concentration of procurement spending by specific bidders as dependent variables. We employ the same set of control variables. We denote those procedure types as risky, hence marked as red flags, which have a significant and positive link between the procedure type and the probability of single-bidding and/or concentration risk. For further granularity, we decompose the red flags into high and medium corruption risks, depending on the size of the coefficients with direct awards without any expectation of competition representing the highest risk and invitation tenders where at least the invited bidders are expected to compete as medium risk. Moreover, those procedure types which are statistically indistinguishable from the reference category of open procedure type are classified as non-risky (Table 11, “not a red flag” column). Full regression results are reported in Table 9 and Table 10.

Table 11.

Non-Open Procedure Red Flags

article image
Source: Authors Assessment.

The last two indicators, the lack of published tenders and the residency in tax heavens of suppliers14 are also tested using our two regression models, and also contribute to informing transnational aspects of corruption. Using the two models we found again positive15 and significant coefficients, thus reinforcing the choice of all these seven red flags. For the suppliers registered in jurisdictions to avoid disclosing details on their ownership, we relied on an independent, objective metrics of company and financial secrecy in countries and territories developed by the Tax Justice Network.16 This type of indicator is not only helpful for the direct purpose of identifying corruption risks in the procurement system or a given country, it can also inform how transnational corruption can occur.

Arguably, both single bidding and concentration of procurement spending on specific bidders are proxy indicators of corruption. This means that they can arise due to non-corrupt conditions as well as corruption may happen without their presence. Our methodology tries to minimize these measurement errors. First, we looked for association and co-occurrence between single bidding and high spending concentration outcomes on the one hand and known methods and signals for favoring connected bidders such as direct awards or short advertisement periods on the other hand. Such co-occurrence of risky tendering processes and outcomes should lower our measurement error. Second, while we indeed expect a positive correlation between our corruption risk indicators, we also expect the fit to be far from perfect which suggests that risky tendering process can give rise to corruption even in the absence of our simple indicators of restricted competition (for example when bidders collude with each other while bribing public officials at the same time).17

While empirical estimations broadly support the selection of red flags, counter-intuitive results serve as guidance to further improve data quality and reinforce the need to rely on a broad composite indicator and not just a limited set of indicators. For example, in the Table 10 below, which presents the results of the validation through supplier contract share, some results are not congruent with that of the validation using single bidding (Table 9). Even when a risk indicator has a significant positive impact on one of the outcomes, say single bidding, it can have insignificant or negative impact on the other one, say supplier contract share. In the case of Romania, for example, some of the risk indicators which behave as expected for the single bidder regression, they are negative significant in the supplier contract share regression. This could be caused, inter alia, by reliability issues in the identification of businesses, making the corresponding regression noisier. Overall, the single bidding regression are more reliable hence offer sufficient evidence for indicator validity (Fazekas and Kocsis, 2020). Even the counter-intuitive results can suggest relying on a broad composite indicator, as corruption risks may not always be adequately capture by one specific indicator.

Appendix III. Additional Price Regressions

Paraguay

Table 12.

Paraguay – Main Results

article image
Source: authors estimates. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Table 13.

Paraguay – Alternative Specification

article image
Source: authors estimates. Regression includes controls for contract values, buyer type, buyer location, market, and tender year. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Uganda

Table 14.

Uganda – Main Results

article image
Source: authors estimates. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Table 15.

Uganda – Alternative Specification

article image
Source: authors estimates. Regression includes controls for contract values, buyer type, buyer location, market, and tender year. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Romania

Table 16.

Romania – Main Results

article image
Source: authors estimates. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Table 17.

Romania – Alternative Specification

article image
Source: authors estimates. Regression includes controls for contract values, buyer type, buyer location, market, and tender year. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Indonesia

Table 18.

Indonesia – Main Results

article image
Source: authors estimates. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.

Table 19.

Indonesia – Alternative Specification

article image
Source: authors estimates. Regression includes controls for contract values, contract type, buyer type, buyer location, market, and tender year. Model 2 shows an alternative specification to the bidding structure in Indonesia, instead of single bidding we define cut-offs based on the distribution of the bidding behaviour such as 1.corr_bid corresponds to 12 to 22 bidders and 2.corr_bid corresponds to 1 to 11 bidders. Robust standard errors in parentheses. Clustered over buyers. *** p<0.01, ** p<0.05, * p<0.1

Defined as the ratio of actual contract value and normal value.
1

Government Transparency Institute, aly.abdou@govtransparency.eu

2

International Monetary Fund, obasdevant@imf.org.

3

University of Sussex, E.David-Barrett@sussex.ac.uk.

4

Central European University and Government Transparency Institute, misi.fazekas@gmail.com.

5

The authors would like to thank colleagues form the European Union, the IMF, UKAID, the World Bank, for the very helpful comments.

6

Since this study, the project was expanded to cover more countries https://public.tableau.com/app/profile/gti1940/viz/CorruptionCostTracker/Overviewofcountries?publish=yes

7

Note that while cross-country comparisons are possible, they would not necessarily fully reflect idiosyncrasies in the development of each indicator (as, for example, procurement procedures differ across countries), and as such would be subject to caution.

8

There is, however, a trend in procurement systems to make an increasing use of negotiations which may result in longer periods for selecting the winning bid. For example, the EU introduced a “competitive dialogue” in its legal framework after the 2014 directives on procurement (see Saussier and Tirole, 2015).

9

Red flags are defined using a cardinal order, low-medium-high risk. In effect they get assigned values, 0, 0.5, and 1, respectively, so that they can be turned into a composite score.

10

For example, individual instances of single bidding may be explained by a number of non-corrupt reasons (e.g., known most productive bidder, limited number of potential bidders for a specific market and/or country). While this feature could be seen as a limitation in the use of the CRI, it’s also, in some ways, a strength. Indeed, as noted at the beginning of the paper, what we aim for is to provide an indicator of corruption risks and not corruption instances. As such, the use of the CRI (especially in informing relative price differential, as described in the next section), should be seen as a first step to gauge corruption risks in public procurement in a given country or sector. To further refine and assess the pertinence of the CRI, it should naturally be complemented by qualitative information, for example on specificities of certain markets. However, because qualitative information would be critically country-dependent, and hard to come-by, a broader framework to analyze corruption risks should be developed.

11

Contract values are estimated by the procuring entity before the launch of the tender. It is needed in most countries for budgetary purposes (practically, the amount the public sector allocated for a contract depends on this estimate). Such estimations are highly regulated, requiring, inter alia, consulting past similar tenders and market analysis.

12

Please note that in spite of our best efforts to standardize datasets and indicator definitions, non-negligible differences remain in terms of data quality, regulatory prescriptions, and data scope. By implication, the curves can be compared across countries only to a limited degree.

13

For an up-to-date overview of e-procurement adoptions around the world see the World Bank’s tally: https://www.globalpublicprocurementdata.org/gppd/

14

While the information on tax heaven residency of suppliers is helpful, as information on beneficial owners become more widely available it would be helpful in subsequent iterations to include information on beneficial owners residency.

15

For the case of suppliers registered in jurisdictions no favoring transparency on ownership, the reference category can either be foreign suppliers not registered in such jurisdictions or domestic firms. Similarly, foreign suppliers registered in those have a negative coefficient compared to other foreign suppliers (stronger test), however their coefficient is always larger than national suppliers (weaker test). While this evidence is not as strong as we would like it, it is nevertheless confirmatory and given the large literature on the subject we included it in the CRI.

17

Please also note that the two regressions for each country may not yield fully congruent results, that is while a risk indicator has a significant positive impact on one of the outcomes, say single bidding, it can have insignificant or negative impact on the other one, say supplier contract share. For example, for Romania some of the risk indicators which behave as expected for the single bidder regression, they are negative significant in the supplier contract share regression. This is likely due to the lack of reliable organization IDs in Romania making the latter regression noisier. In such situations, the single bidding regression are more reliable hence offer sufficient evidence for indicator validity.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Working Papers
Assessing Vulnerabilities to Corruption in Public Procurement and Their Price Impact
Author:
Aly Abdou
,
Olivier Basdevant
,
Elizabeth David-Barrett
, and
Mihaly Fazekas
Volume/Issue:
Volume 2022: Issue 094
Publisher:
International Monetary Fund
ISBN:
9798400207884
ISSN:
1018-5941
Pages:
31
DOI:
https://doi.org/10.5089/9798400207884.001
  • <sup><ext-link xmlns:xlink="http://www.w3.org/1999/xlink" ext-link-type="uri" xlink:href="https://elibrary.imf.org/view/journals/001/2022/094/article-A001-en.xml#RA001fig02">Figure 2.</ext-link></sup>
    View in gallery
    Figure 2.

    Observed Relative Price Distributions by Country

    (Histogram, thousands of contracts, left axis, relative price on the x-basis)

Figure 2.

Observed Relative Price Distributions by Country

(Histogram, thousands of contracts, left axis, relative price on the x-basis)