1 of 21

Using beneficial ownership data for large-scale risk assessment in public procurement.

The example of 5 European countries

Mihály Fazekas*, Irene Tello Arista*, and Antonina Volkotrub**

​

*Central European University (Austria)

** Anticorruption Action Centre (Ukraine)

Symposium on Systems of Financial Secrecy, 21/02/2024

2 of 21

Outline

  • Motivations

​

  • Conceptual framework and expectations

​

  • Data and methods

​

  • Results

​

  • Open questions, next steps

​

3 of 21

Motivations

  • Systematic evidence is lacking…
    • Virtually no papers which use BO data for systematic risk assessment
  • Instead
    • Lots of papers on why BO is so great
    • A few papers looking at impacts: Szakonyi et al (AC/DC and TI)
  • Adjacent literature
    • Many papers using large-scale ownership network data: e.g. Riccardi et al., Heemskerk et al.

4 of 21

Goals of the paper

  • Broad aspirations:
    • Lay the groundwork for the emerging literature.
  • Specifically 1:
    • Validity testing commonly suggested indicators of money laundering and corruption in beneficial ownership datasets.
  • Specifically 2:
    • Generating hypothesis for the impact of beneficial ownership registers on financial crime.

5 of 21

Institutional framework

�

Denmark

Latvia

Slovakia

Ukraine

United Kingdom

Name

Central Business Register (CVR)

Registry of Enterprises

Public Sector Partners Register (RPVS)

Unified State Registry (USR)

People with significant control register (PSC)

Launch date

May 2017

2017

2017

September 2015

April 2016

Sector

Full-economy

Full-economy

Procurement

Full-economy

Full-economy

Authority

Danish Business Authority

Ministry of Justice

Ministry of Justice

Ministry of Justice

Companies House

Laws involved

Act amending the Companies Act, the Certain Commercial Undertakings Act, the Corporate Funds Act and various other acts

Law On the Enterprise Register of the Republic of Latvia

Act on the Register of Public Sector Partners (ARPSP)

On State Registration of Legal Entities, Individual Entrepreneurs and Public Associations

Small Business Enterprise and Employment Act

Link

Open to public

Yes

Yes

Yes

Yes

Yes

Threshold used to determine beneficial ownership

25%

25%

25%

25%

25%

6 of 21

Conceptual framework

  • Risky ownership features of a company higher likelihood of financial crime, including money laundering, corruption and fraud
  • Validity testing logic: Convergent validity
    • If a company is risky, it is more likely to participate in corrupt procurement tenders
  • However, our main hypothesis is that BO registers carry little directly valuable information for corruption risk assessment (e.g. why would a corrupt politician hiding behind a Cayman Islands corporation ever declare true ownership to the public?)
    • What is more likely is that missing values, errors, unreasonable values and unusual records in the BO register would point at transparency circumvention and the use of nominees

7 of 21

Company indicators and theoretical expectations

Indicator group

Indicator name

Indicator definition

Expected relationship with public procurement corruption risks

BO

Company frequency by BO

Number of companies a BO owns

Exceptionally many companies owned by a BO leading to higher PP risks

BO information change frequency

Number of changes in BO information of a company (total)

Exceptionally many changes to a company’s BO information leading to higher PP risks

BO age

Age of the BO (number of years when contract is won)

Very young or very old BO of a company leading to higher PP risks

BO country: Foreign

Is at least one of the BOs of the company foreign (non-domestic)?

Foreign BOs are more often associated with higher PP risks

BO country: China

Is at least one of the BOs of the company citizen of China?

Chinese BO leading to higher PP risks

BO country: Sanctions (Russia, Belarus and Iran)

Is at least one of the BOs of the company citizen of a sanctioned country?

BO from sanctioned country leading to higher PP risks

BO country: Offshore jurisdictions

Is at least one of the BOs of the company citizen of an offshore jurisdiction?

BO from offshore jurisdiction leading to higher PP risks

BO country: Multinational (2+ countries)

Are the BOs of the company citizens of at least 2 different countries?

Very many different BO nationalities leading to higher PP risks

No BO data

Is mandatory BO data missing for the company?

Failing to properly disclose BO data leading to higher PP risks

General company

Company age

Number of years between company foundation and contract award

Very young companies (e.g. 1 year or younger) leading to higher PP risks

BO PEP

Is at least one other BOs of the company a politically exposed person?

Politically connected companies leading to higher PP risks.

8 of 21

Data

  • Country coverage
    • Denmark, Latvia, Slovakia, UK, and Ukraine
  • Period
    • 2009-2021 (but differs by country)
  • Source datasets
    • Opentender.eu
    • Open Ownership Register + national registers
  • Data merging process
    • ID-based matching
    • Company name-based matching

9 of 21

Data overview

10 of 21

Indicator overview

Corruption Risk Index, Slovakia

For more on these indicators see: Fazekas, Mihály, and Kocsis, Gábor, (2020), Uncovering High-Level Corruption: Cross-National Corruption Proxies Using Public Procurement Data. British Journal of Political Science, 50(1).

For data access: opentender.eu

11 of 21

CRI vs World Governance Indicators’ Control of Corruption

For more on these indicators see: Fazekas, Mihály, and Kocsis, Gábor, (2020), Uncovering High-Level Corruption: Cross-National Corruption Proxies Using Public Procurement Data. British Journal of Political Science, 50(1).

For data access: opentender.eu

12 of 21

Methods

  • Bivariate: simple linear correlations
  • Linear regression – OLS for each country
    • DV: CRI
    • IV: BO risk indicator (individually)
    • Controls: Contract value (deciles+missing), year, region (NUTS), market (main CPV, 2 digits), buyer type (central government, regional/local government, etc), number of beneficial owners per company.

13 of 21

Results II

14 of 21

Result details: BO frequency - UK

Outlier number of companies linked to the same BO is associated with higher PP CRI

15 of 21

Result details: BO age - Latvia

16 of 21

Result details: Company age Denmark

17 of 21

Result details: BO country - UK

…Even more interestingly, the specific country the BO is coming from carries even higher PP risks

18 of 21

Result details: No BO information - Latvia and Ukraine

Where we have reliable missing BO information flag, it works well

19 of 21

Conclusions

  • BO datasets are high enough quality for systematic, large-scale risk flagging, or rather they have the right kinds of errors.

​

  • Many widely discussed BO indicators appear to be valid.

​

  • Risk definitions are country-specific and reflect regulatory and corrupt scheme specificities.

20 of 21

Next steps

  • Hypotheses generated:
    • BO registers induced a move from offshore to strawman evasion techniques.
    • BO registers have no impact in low AC enforcement capacity countries (e.g. Ukraine) as knowing illegitimate owners would not increase the risk of punishment.

​

  • Improvements, extensions planned:
    • Update data of Slovakia
    • Add Estonia as a sixth country
    • Check residence and not nationality to see if it is a risk factor

21 of 21

Looking forward to your feedback!

​

​

Q&A