The Nuts And Bolts For Setting Up Shop Across The Globe: RegTech And Machine Learning

Manfred Esser, Vivek Gollapudi and Gert Eichberger

The revenue potential of globalization is too huge to ignore. Not to mention the added opportunities for foreign investments and joint ventures, access to diverse talent, and the expansion of your brand in new countries. It is all too clear that global markets form the backbone of our economy. This calls for global market leadership, which is typically determined by how successfully you have expanded into new countries.

Before entering a new country, international companies analyze the economical ecosystem of a country on parameters such as market size, strength, purchasing power, import-export capacity, and legislation complexity. Most of this data is freely available in databases of the World Bank, at least on a high level. Among these key criteria is the Economic Complexity Index (ECI), which is defined as a measure of a country’s production properties and serves as a measurable quantity to classify a country’s complexities and economic strength. In addition, local tax advisers and market experts provide a first glimpse of the required efforts to enter a new market. Dimensions of such an analysis can map revenue potential of the intended market segment vs business complexity.Market Potential

In an ideal scenario, huge market returns and the ease of doing business, in any given country, go hand in hand. It might be tempting to conquer deregulated markets. However, competitors might already be ready to dominate such attractive markets. As a result, companies capable of running businesses in complex environments often become overall world market leaders. Their operational efficiency is optimized under tough conditions while conquering large markets, which helps them withstand competition (the upper-right quadrant in the graph above). And this is exactly where RegTech IT can help companies attempting to enter such complex yet profitable markets, ensuring a win in the magic quadrant in the upper-right.

The most important factors driving complexity in setting up businesses in new countries are local laws and regulations. They significantly influence the agility with which (global) companies can offer their products to local markets. The frequency of legal changes per year can become a key driver of complexity. In addition, branch-specific regulations may have to be followed. As part of digital transformation, several tax authorities have started asking for real-time tax reporting per transaction from taxpayers. Such legal requirements can impose significant upfront IT investments before an organization even enters a local market.

Country localization complexity

The decision of a global player to start doing business in a country is often influenced by a gut feeling rather than objective criteria and measurable figures. As part of an informed market-entry decision, an increasingly important parameter for global companies is the country localization complexity (CLC), which is defined by legislation complexity, regulation diversity, noncompliance penalties, and reporting obligations – all of which can change from year to year. For this reason, we need flexible tools, instead of static reports, to dynamically analyze these factors.

Harnessing machine learning to analyze country localization complexity

Machine learning (ML) can support us in analyzing CLC by learning from historical data like official publications and documentation. Typically, the sources of such data are local authority websites.

Supervised ML automatically keeps cross-country comparisons and detailed country-specific complexity rankings up to date, based on new data provided by ratings of expert users. In addition, ML algorithms can indicate which parts of an IT enterprise software landscape will most likely require enhancement due to new (or upcoming) legal requirements. Furthermore, ML can help determine the efforts required to introduce or maintain legally compliant IT infrastructure for a given country.

Today we have reached the tipping point of having large amounts of legal data (for example, from government websites) and of boosting IT performance to process this data. Machine learning is the ideal tool of choice to exploit these data mines.

Intelligent tools of the trade

Here is a small illustration of the techniques used to analyze CLC:

  • Algorithms like Naïve Bayes and Deep Learning can determine, for example, if new legal documents (such as decrees, software vendor notes, or consulting digests) require the attention of a company in a specific industry or not.
  • Going one level deeper, in technical terms, neural networks allow multiclassification or voting, as they handle nonlinearities by varying the number of hidden layers.
  • Latent Semantic Indexing (LSI) is another important procedure of (legal) information retrieval, which is especially important for searching and structuring Big Data. The goal of LSI is to identify the principal components of (legal) documents. The principal components are general terms or classifications.

Alternatively, Word2Vec is a context-based method to keep semantically similar terms together. For example, the word “payment” comprises all terms like “postdated check,” “bank transfer,” or “bill of exchange.” This is useful to semantically group all words that are related to “payment,” even if the word “payment” does not appear in the text. LSI can also help distinguish between documents that, for example, talk about taxes from those in which the word “payment” is related to a lottery payout.

Machine learning: Guiding global customers in local decisions

In today’s global economy, local, country-specific regulations are still determining the rules of how business must be run. ML can analyze the legislation complexity for each country and compare legal differences across countries of interest. While tax advisory services still play a role in consulting for market investment decisions, we see ML as a rapidly expanding toolset to take over key parts of such advisory and alerting services.

ML algorithms learn from the diversity of country regulations and their changes, their frequency, and in comparison to changes in other countries. For example, Brazil would be rated “very complex” with respect to tax categories like ICMS, sales tax, and IPI taxes, while other countries have a simpler tax structure based only on VAT. ML can, for example, analyze tax reporting and alert companies when and what to consider for legal compliance and to avoid fines. Unbiased data of country localization complexity, analyzed by ML, enables executives to make effective investment decisions rather than relying on high-level presentations.

CLC not only helps reduce the cost of compliance for companies operating in established markets but also supports organizations during their market entry decisions.

Learn more about “Simplifying Complicated Tax Reporting And Compliance With Automation.”


Manfred Esser

About Manfred Esser

Dr. Manfred Esser is a Localization Product Manager at SAP Globalization Services, supporting the Legal Change Maintenance and New Business processes in South East Europe. He started as a IT developer for Commercial Software in DCW, owned by Dr. Claus Wellenreuther, one of the co-founders of SAP. In 2004, he moved to SAP in the role of Country Advocate for Eastern Europe. Since 2017, he offers lectures in Machine Learning and Quantum Computing for innovation and new development at SAP.

Vivek Gollapudi

About Vivek Gollapudi

Vivek Gollapudi is a Machine Learning Developer for SAP Globalization Services, working on SAP Law to Action, a solution that determines the relevance of incoming legal changes. He has a degree in Electrical Engineering from the University of Warwick, England. His current research interests include recommender systems and Bayesian statistical methods in Natural Language Processing.

Gert Eichberger

About Gert Eichberger

Gert Eichberger is the Director of SAP Localization Product Management for Eastern Europe and Africa. This group supports SAP customers in legal compliance and offers new innovations such as SAP Cloud solutions and other cutting-edge compliance platform offerings. In addition to this, he is a core member of the strategic Machine Learning offerings at SAP. Gert started his career at SAP 17 years ago. Prior to that he worked as a development engineer for management consulting at PwC.