Estonia has 35,000 e-residents, most are Finnish, Russian and Ukrainian

15 May 2018

Estonia is attracting consultants from around the globe with the nation’s digital identity cards for business. Accessing the card means becoming a digital or e-Resident and allows the holder to start a business and pay tax in Estonia through the country’s ultramodern technology infrastructure. The country’s aim is to spread the digital citizenship globally, empowering people from all corners of the globe to connect and innovate.  

Estonia is tucked up in the Baltic States and was part of the former Soviet Union, so when the Iron Curtain came down and the country gained independence, the government had to look for a new ‘identity’. With a relatively small, spread out population of 1.3 million and an ageing Russian infrastructure, government officials in Tallinn knew they had to go big to put Estonia on the map. The country digitalised rapidly and within twenty years became a global leader in digital operations and governance.

The infrastructure put in place is considered by pundits as state-of-the-art and leading worldwide. From voting to municipality services and beyond, the country has transformed almost all functions performed by the government into an online service. And the actual government itself has gone paperless; even the President of Estonia signs national bills on a screen and all legislation is available online for public scrutiny and input.

Where do Estonia’s registered e-residents come from?

Not only is this e-governance system a testament to Estonian innovation, it is also designed to attract innovation itself. Since late 2014, it has been possible for foreign citizens from anywhere in the world to become an e-Resident of Estonia and take advantage of the country’s business environment. E-citizenship does not provide entrepreneurs working-rights in the country or residency in the EU, but it does allow them to administer a business through the country’s infrastructure and benefit from a lower cost of government administration.

The e-Residency programme currently hosts over 35,000 people (e-Residents) from over 150 countries. The concept has attracted people from all over the world to what the Republic of Estonia calls the ‘new digital nation’ – many of the e-residents are tech-savvy entrepreneurs who are excited about being part of something new. The Estonian government believes that with every dollar that they put into further developing the infrastructure, it will bring $100 to the nation.

An analysis on the different demographics of e-citizenship shows that Finland, Russia and the Ukraine are the biggest users of the e-Residency programme, with 88% being male. The data also shows that the main sector utilising the programme for their business activities are management and business consultants.

e-Residency popular among consultants

Consultants are attracted by the digital nation concept due to the international nature of consulting work. Business and marketing consultancy Anywhere Consulting’s founder Peter Benei is an Estonian e-Resident and advocates the programme for other entrepreneurs and consultants citing three main reasons: minimalism, transparency, and long-term benefits.Most of Estonia's e-residents are consultants and IT professionalsAnywhere Consulting helps entrepreneurs to launch, build and grow their remote business. Benei commented on the programme;  “I run my business remotely, which means business without borders. Everything is not just digital – it has to be digital, otherwise I can’t work with it. I have clients from all around the world, I need access to global payment services and tools which help me to automate, run and manage my business, anywhere I go. With the Estonian company, I have full access to everything.”

He continues: “I truly think e-Residency is the future for everyone. All other governments should treat their countries’ business environment like this little European country does: 100% online, transparent and minimalist. After all, it’s a service for those who want to get things done and do business. Governments should not waste their entrepreneurs’ time and should help them whenever they can to provide services that help building fruitful and growing businesses.”

AI can improve operational risk management in banking

17 April 2019

Risk management is an integral part of banking. By taking financial risks, banks are able to generate the profits that are necessary to survive. Risk management aims to control this process by making potential losses more predictable. This makes the bank more robust to external fluctuations.

Whereas profits can be made by accepting certain financial risks, operational risk is intrinsically different. Operational risks only cause losses – financially in terms of bottom-line impact and non-financially in the form of for instance reputational damage. The consequences of operational risk events can have a large impact on an organisation and the financial system as a whole as experienced during the last financial crisis. It is not surprising therefore that operational risk is receiving more attention within the financial sector, with banks trying to minimise the operational risks they take, given the resources available while keeping in mind the strategic goals of the organisation.

In the past decade, there has been major progress in the development of artificial intelligence (AI). AI algorithms excel at data analysis and have evolved to the point where they surpass human performance for a wide variety of tasks. More and more businesses exploit these technological advances to optimise different kinds of processes such as marketing, sales and e-commerce, manufacturing and logistics. In today’s growing data-driven world, this trend is expected to continue on the back of widening opportunities for use cases, including in the area of operational risk management.AI can improve operational risk management in banking

Challenges in operational risk management

In June 2011 the Basel committee published the Principles for the Sound Management of Operational Risk (BCBS), which provides a framework for the development of proper operational risk management. Three years later, a survey was conducted to measure to what extent banks complied with these principles. One of these principles states that banks should write a risk appetite and tolerance statement. Banks reported that this is more challenging for operational risk than for other risk categories and attributed this to the nature and pervasiveness of operational risk. The banks that did comply with this principle often reported the use of backward-looking metrics of operational risk, such as operational losses as a percentage of gross revenue.

The above example underlines the challenges which banks face in the management and measurement of operational risk. Compared to financial risk, operational risk is a more qualitative field of study. Whereas financial risk management has been the main priority of banking for a longer time, operational risk management is much younger resulting in less extensive historical data. Predictive modeling becomes more of a challenge in this situation.

On top of that, the events in operational risk are much more diverse in scope. The Basel committee defines operational risk as the “risk of loss resulting from inadequate or failed internal processes, people and systems or from external events”. Internal fraud, data leakage and reputational damage are very different problems, yet can be very closely related as well. A lot of the data in operational risk consists of textual input which contain qualitative information. The qualitative nature of operational risk is reflected in the Basel framework, which encompasses guidelines for organisational structures, culture and awareness, and qualitative reporting.

The computer as a reader

Artificial intelligence could play a valuable role in upgrading operational risk practices, with in particular machine learning – the field of self-learning computer algorithms – showing promise. Machine learning algorithms can make predictions based on data fed to the algorithm. Recently, major progress was made in the field of natural language processing (NLP). NLP focuses on using textual data for predictions. As an example, an algorithm can learn to rate hotel reviews. By processing large amounts of reviews together with their given ratings, the algorithm can learn to give a rating to a new review it has never seen before.

“There are many opportunities for operational risk management to exploit AI and other related technological advances.”
– Lars de Ruiter and Matthias Geerse, Solid Professionals

In operational risk, many textual reports are written regarding specific risks or risk appetite of the organisation as a whole. These reports are generally written by risk managers – experts in their field. Assuming that risk reports hold information which is absent in historical loss data, this information could be extracted and used for predictive purposes. So why should computers perform this task instead of risk experts? The answer lies within the physical limitations of human being – as well as the natural biases they have to cope with. While humans may struggle to remember a piece of text, a computer algorithm easily processes thousands of books and finds structure within using statistical methods with near-perfect precision.

Towards prediction of operational risk

The performance of NLP algorithms using word embeddings has increased tremendously of late. Using the embedding as a starting point, these algorithms learn the meaning of full sentences and use this information for further predictions. This is more generally called sentiment analysis. A popular application in the financial setting is predicting stock price movement from news articles, or twitter feeds. A recent paper by Denmark’s national bank predicts corporate distress of firms by analysing their annual reports. Taking textual data under scrutiny can make a contribution to quantitative predictive modelling, even in the financial sector where the use of words is very different to day-to-day language.

In summary, there are many opportunities for operational risk management to exploit AI and other related technological advances. One possibility is the classification of risk events. As an example, a computer algorithm can read risk descriptions written by risk managers and classify them according to their impact and frequency. Combining loss data with risk reports, improved prediction of risk events might lead to more accurate prediction of future losses. A different application is the measurement of more abstract concepts such as the financial health or maybe the cultural aspects of a company.

An article by Lars de Ruiter and Matthias Geerse. Both are finance & risk consultants at Solid Professionals, a consultancy from the Netherlands.