5.6.4. Other Types of Affected Clients


### *5.7. The Unbanked, Vulnerable Communities, Minorities and Gender Issues*

Unbanked populations in developing countries often struggle to obtain access to financial services because of barriers that constrain their financial inclusion, such as financial illiteracy, a low level of income, inconsistent cash flows and significant costs related to transportation (Durner and Shetret 2015). The rural poor living in developing countries have historically been financially excluded by banks due to the high costs incurred to operate from remote locations (Durner and Shetret 2015).

The communities that are most significantly affected by a lack of access to financial services are "rural, low-income, and minority communities, such as women and youth" (Durner and Shetret 2015, p. 22). Women are sometimes victims of procedures that are inherently discriminatory in nature. In fact, women sometimes face financial exclusion as finance is usually much harder to obtain when compared to men, inevitably hindering women's empowerment (Durner and Shetret 2015).

### *5.8. Technology-Based Solutions to the Derisking Problem*

#### 5.8.1. Blockchain and Distributed Ledger Technologies (DLTs)

These help to mitigate the negative effects associated with derisking as their implementation helps improve the data gathering and identification processes. This ultimately leads to reduced burdens associated with compliance and enhanced transparency (Babe 2017). Moreover, blockchain and DLTs result in lower regulatory and compliance costs such as KYC requirements (IFC 2017), reduced costs associated with the verification of customers, and increased transparency of transactions.

These technology-based solutions offer numerous benefits: they facilitate the sending and receiving of remittances, make trade finance easier for businesses to obtain and assist charities operating in certain areas of conflict (Neocapita 2017).

While blockchain and DLTs are able to adequately protect confidential information, the level of anonymity provided by such technologies enables bad actors to conceal their identities. Therefore, this means that individual payments may end up being difficult to track (IFC 2017).

#### 5.8.2. Big Data, Machine Learning and Biometrics


#### *5.9. Derisking in the Context of Malta*

#### 5.9.1. The MONEYVAL Report of 2019

In 2019, MONEYVAL, a monitoring body of the Council of Europe (CoE) specialising in antimoney laundering, published a report that provided a brief overview of the AML/CFT measures Malta has implemented and commented on the robustness of the AML/CFT system the country has in place. Moreover, this report analysed the extent to which Malta is compliant with the FATF recommendations (CoE 2019a).

To obtain a more comprehensive understanding of money laundering and financing of terrorism risks, Malta conducted a national risk assessment (NRA) exercise in 2013/2014. Following this NRA exercise, Maltese authorities were able to more extensively comprehend the risks and vulnerabilities in Malta's AML/CFT system, especially in the case of regulated sectors (CoE 2019a). However, MONEYVAL noted that there were factors, including predicate offences, legal persons and arrangements and financing of terrorism, which appeared to be insufficiently understood (CoE 2019b).

#### 5.9.2. Derisking in Malta: A Local Case

The banking sector in Malta has been experiencing a downward trend for these past few years. Late in the year 2018, Pilatus Bank, which was considered Malta's most controversial bank, had to close down its operations after allegations of money laundering breaches emerged. In July 2019, Satabank was fined a sum of €3 million for having weak structures in place, which facilitated the carrying out of criminal activities such as money laundering (Vento 2019).

Lately, Malta's banking sector has increasingly suffered as foreign banks are withdrawing their correspondent bank services from Malta (Costa 2019, Costa 2019). This has mainly been the result of failures vis-à-vis antimoney laundering in Europe (Vento 2019).

In the year 2019, a core domestic bank, Bank of Valletta (BOV), embarked on a derisking exercise (Grech 2019). This resulted in BOV ceasing the provision of certain banking services to some of its customers and closing their bank accounts (Martin 2019). To ensure that it followed international best practices, BOV decided to undertake this derisking exercise alongside a KYC retail client remediation exercise (Grech 2019).

The decision to initiate such derisking was taken following discussions BOV had with the Malta Financial Services Authority (MFSA) and the European Central Bank (ECB) (Grech 2019). It was brought to light that the ECB had given BOV clear instructions to engage in derisking in order to establish a new risk profile that involves decreased risk (Martin 2019). Derisking was also carried out so as to allow BOV to strengthen its frameworks, systems, policies and procedures regarding risk, compliance, antifinancial crime and antimoney laundering (Costa 2019).

The customers affected by this derisking exercise mainly comprised of businesses deemed as being high risk, such as iGaming companies. These iGaming companies, including Malta Gaming Authority (MGA) licensees, had their bank accounts closed (Martin 2019). Other affected customers that also had their accounts closed consisted of gaming affiliate companies (Martin 2019), as well as individual investor programme (IIP) buyers who had opened a bank account in Malta to obtain a Maltese passport (Grech 2019).

Derisking was carried out with respect to both international corporate customers (ICC) and international personal banking (IPB) customers. ICCs typically consist of companies that are registered in Malta with the Malta Business Registry (MBR), where the majority of the company's business activity occurs outside of Malta, and the beneficial owner is not Maltese (Grech 2019). When it comes to IPB customers, there is a mix of clients, with some located outside EU jurisdictions. The accounts of ICCs and IPB customers that BOV determined as not having an economic nexus in Malta were closed (Grech 2019).

The implementation of this derisking strategy has made it difficult and cumbersome for foreign companies to set up their operations in Malta and open a bank account (Martin 2019). Additionally, given limited local options, a number of foreign clients had no other choice but to transfer their operations overseas or start making use of e-money institutions (Vento 2019).

### **6. Methodology**

#### *6.1. Tool Used to Collect Data*

For the purpose of this study, a purpose-built semistructured survey was designed and administered to personnel in banks, financial institutions, accounting firms, regulatory authorities, as well as any other relevant parties and people dealing with these institutions.

Some of these surveys, which were built using a software application (Qualtrics XM) in order to provide an online link, were administered using social media and email; the rest were carried out over the phone or using other communication applications such as Skype.

The survey consisted of three sections. The first section consisted of five demographic questions asking for the "age", "gender", "locality", "capacity" and "level of expertise" of the participant. The next section consisted of six themes with three statements each (i.e., 18 statements in total), to which participants were asked to rate each statement using a 1 to 5 Likert scale, "1" being "Strongly Disagree" and "5" being "Strongly Agree". The themes were associated with certain drivers, factors and implications of derisking identified from existing literature (derived using the thematic approach, as suggested by Braun and Clarke 2006). These themes were "Financial", "Socioeconomic", "Financial Exclusion", "Antimoney Laundering/Combatting the Financing of Terrorism (AML/CFT) and Risk", "Compliance and Regulation" and "Technology-Based Solutions". The third section consisted of a set of seven open-ended questions related to the topic of derisking and its main drivers, factors, implications and effects, particularly with regards to the Maltese scenario (Saunders et al. 2009; Yin 2002; Yazan 2015; Stake 1995). Other questions dealt with


#### *6.2. Sample Population*

Since the population of possible candidates was not known, we used a mix of nonprobability purposive sampling by contacting persons on our social media and email, and they connected us with further possible participants (nonprobability snowballing sampling; Mack et al. 2005; Neuman 2005).

The initial sample of potential study participants consisted of 60 organisations employing individuals who possess a relatively high level of knowledge regarding the derisking practice, money laundering and terrorism financing. The organisations that were targeted included banks operating in Malta, the Central Bank of Malta (CBM), local regulatory authorities (the MFSA and the FIAU), the large accounting firms and other accounting and risk management firms.

To determine the participants' level of expertise, a 1 to 3 Likert scale was utilised, where 1 corresponded to "a fair level of expertise", 2 corresponded to "a good level of expertise" and 3 corresponded to "an excellent level of expertise". To ensure that the data obtained were reflective of expert knowledge, we filtered participants by level of expertise and only used data from participants who expressed a good or excellent level of expertise.

Between 15 November 2019 and 21 April 2020, we carried out a total of 32 interviews with individuals who have a good or excellent level of expertise in derisking. Although it seemed that additional interviews beyond this point did not provide new information or added value, we maintained an open mind and supplemented our data with the 296 completed online surveys received from participants during the same period in which we carried out the interviews. This ensured that our selection of participants was, as much as possible, free from any selection bias since the surveys were answered anonymously. These surveys were then filtered to ensure that the level of expertise criteria was met, and these resulted in 285 valid participants. In total, we had 317 participant surveys.

#### *6.3. Analysis*

Following the collection of data, such data was first transferred into Microsoft Excel and Microsoft Word, and, later, the quantitative data were loaded onto Statistical Package for the Social Sciences (SPSS version 21) software. Subsequently, the data were analysed using frequencies and by subjecting the data responses to the statements to exploratory factor analysis, which is a method of testing the theoretical understanding of the factors for the main drivers of derisking posed in the first research question (RQ1), and, as determined from the literature, no a-priori fixed number of factors was set. The final number of factors was determined by the data and the authors' interpretation of them (Hair et al. 1998).

Since the items used the ordinal scale of measurement, we used the median (Md) as a measure of central tendency and the interquartile range (IQR) as a measure of spread. Where a group of items could be grouped into a construct (or theme), we assessed the internal consistency reliability of the measures via Cronbach's alpha. After the items were combined into a single Likert scale, we computed the mean (M) as a measure of central tendency and the standard deviation (SD) as a measure of spread.

We then carried out multiple linear regression to determine how this measure varies with the demographic factors; we also used STATA application software applying White robust standard errors to account for any heteroskedasticity and analysed the answers to the last section using the thematic analysis approach, as suggested by Braun and Clarke (2006, p. 6), to answer RQ2.

#### **7. Results and Discussion**

#### *7.1. Summary Statistics*

As can be noted from the tables in Appendix B, the largest group of participants were aged between 30 and 39 (39.7%), most were men (68.5%), coming from a central location (54.9%), and users of financial institutions (67.5%), with a good level of expertise (59.6%). This shows that most of the participants were exposed to a similar culture and environment, and this could be the reason why the results of the multiple linear regression, noted and discussed below, are not significant.

Additionally, as noted in Appendix A, all participants are in agreement (mean between 3.4 and 4.19) with the statements posed in the survey/questionnaire, with the exception of Statement 13 (S13) wherein the participants strongly agreed. This further confirms our findings from the multiple linear regression carried out and explained below.

#### *7.2. Exploratory Factor Analysis*

For exploratory factor analysis, we used direct oblimin via principal components extraction with Kaiser normalization. The Kaiser–Meyer–Olkin (KMO) statistic, which is a measure of sampling adequacy for the appropriateness of applying factor analysis, fell within the acceptable range (above 0.6) with a value of 0.93. This further supported the continuance of factor analysis, and so the analysis proceeded.

Exploratory factor analysis loaded best as three factors and 22 statements, which, in combination, explained 60.68% of the variance of the perceived implications of derisking in Malta. Table 2 shows which statements are grouped under each of the three factors. Factor 1, which has now been termed "Research and Development Solutions", explained 46.687% of the variance and comprised five items that include the following statements:

S3 Derisking leads to an increase in costs incurred to transfer money. Therefore, export earnings decrease, resulting in a lower level of investment. Moreover, given that domestic banks will not be able to effectively service their clients, this leads to a lower level of foreign direct investment (FDI). As a result, a country's productivity and growth in the long-term may be negatively affected, rendering such a country unable to meet growth and development targets.



**Table 2.** Exploratory factor analysis a.

Extraction method: principal component analysis. Rotation method: oblimin with Kaiser normalization. <sup>a</sup> Rotation converged in 12 iterations. Source: authors' computations.

Factor 2, which has now been termed "Compliance and Regulatory", explained 7.621% of the total variance and comprised seven items that include the following statements:


Factor 3, which has now been termed "Direct and Indirect Effects", explained 6.373% of the total variance and comprised six items that include the following statements:


The 18 statements can be used as an inventory checklist grouped under the 3 main themes "Research and Development Solutions", "Compliance and Regulatory" and "Direct and Indirect Effects" to understand the implications that a derisking exercise can bring to a financial firm. Moreover, one can use this checklist/inventory to develop a risk exposure matrix to arrive at a single qualitative rating and/or quantitative measure of the implication of derisking.

This is of specific interest to risk managers working in the financial services sector, who, as per regulatory requirements, would need to summarise and classify the risk rating of a firm, project and strategy to the board of directors, shareholders and the regulators. It is also of interest to policymakers who can develop policies based on such results.

#### *7.3. Reliability Test*

The Cronbach's alpha coefficients of this scale were between 0.83 and 0.86. Therefore, we can conclude that this scale is reliable as part of our statistical analysis (Table 3).


Source: authors' computations.

**Table 3.** Cronbach's alpha values (*n* = 317).

The computed "Derisking Model" for Malta shows a mean of 4.03 (SD = 0.67). All the factors (Factors 1, 2 and 3) produced means that were close to the computed model. This shows that participants

from Malta, overall, believe that these are the drivers of derisking in Malta.
