Federal agencies involved in budgeting, tax administration and other activities that call for economic modeling can get a boost from artificial intelligence (AI) technologies that have matured over the years to become powerful tools.

For example, multidimensional economic modeling enabled by technological advances allows researchers to rapidly predict and consider the potential effects of a large number of variables on the economy across multiple dimensions. Traditional modeling has just two dimensions available. In addition to government activities, this advanced modeling approach has applications in policy analysis, academic settings and other organizations.

AI is expanding the capabilities of modeling in ways that could make research significantly more effective. In particular, work is underway on detecting patterns in vast volumes of data and interpreting their meaning.

Economic modeling in real time offers an alternative visual approach to observing the behavior of macroeconomic variables. Real-time, multidimensional economic modeling will enable government policymakers to study the impacts and trends of major economic failures in the world economy from a new, big data analytics perspective.

AI techniques tied to behavioral science can leverage big data to discover important, statistically significant details in smaller sample populations. Multidimensional economic modeling can generate a large number of simulations in the same graphical space in real time to study the behavior of complex economic phenomena.

Numerous opportunities afforded by AI

Using AI in these research studies can result in a number of benefits by allowing agencies and other organizations to:

  • Monitor trends and conditions in the U.S. and international markets using current data regarding money, credit, foreign exchange rates and commodities—the AI advanced algorithm can quickly identify emerging situations that could harm the economy
  • Analyze current developments in major industrial sectors of the economy—including housing, real estate, technology, agriculture, energy, communications, transportation and manufacturing—for their potential effects on the federal government’s financial risk
  • Analyze trends in various critical open data sources available on U.S. economic activity, including private consumption and investments, government spending and foreign trade and investment
  • Collect big data for massive economic studies to ensure that all formulation and evaluation policy recommendations and implementation procedures are aligned with government agency policies

Many government agencies cannot predict the impact of changes in regulations. AI enables organizations to use an agent-based simulation of complex systems to understand various issues that influence the tax behavior and interaction of entities in the system. This would be an effective technique if, for example, researchers or policymakers wanted to predict the effect of various tax reform scenarios on the U.S. economy.

Reaping the benefits

By using AI to run these types of analyses, business and government executives can:

  • Execute routing operations and scale fast for research projects
  • Reduce costs, because—once trained—the machine learning model improves over time
  • Improve efficiency by reducing human error

AI also aids in fraud prevention. Machine-learning algorithms help organizations identify suspicious activities based on transaction history and behavioral study of individual populations. Conventional algorithms catch fraudulent transactions only when the transactions violate preset rules.

For example, if a large check is written on a bank account that usually only issues small payments, a traditional system would flag it only if the amount exceeds a preset threshold. An advanced machine learning algorithm, however, can trigger a hold on the transaction until a human verifies it based on how widely it diverges from recent activity, with no need for a designated maximum amount to be on file. AI can perform such analysis in real time, as well as learn from the results of past actions.

The approach is also ideal for risk management. Traditional systems rely on historical data such as transaction history or credit history to understand the risk associated with a given transaction. However, historical data is not always an accurate standard to predict future behavior. Al allows analysis of real-time data of recent transactions to identify potential risks in credit decisions or selecting accounts to audit. The machine learning algorithm can analyze petabytes of data to understand micro activities and assess the behavior of parties to identify possible fraud.

CGI can assist clients with designing, implementing and managing AI for multidimensional economic modeling and many other uses. For more information, download our white paper, Intelligent Automation Opportunities in the Federal Government.

About this author

Picture of Karina Kasztelnik

Karina Kasztelnik

Dr. Karina Kasztelnik is a Director of Consulting in CGI Federal’s Regulatory Agency Program division where she helps to bridge the gap between technical modeling needs and business issues with a strong skillset in data science and big data engineering. She holds a Ph. D. ...

Add new comment

Comment editor

  • No HTML tags allowed.
  • Lines and paragraphs break automatically.
Blog moderation guidelines and term of use