« Expenditure Baselines: A Powerful Tool for Budget Practitioners | Main | Unexpected Help in Fiscal Distress »

June 06, 2022

How Digital Technology will Transform Public Financial Management

 

IStock-1318623693- June 6

iStock.com/iambuff

 

Posted by Khaled Eltokhy[1]

 

The additional investment needed through 2030 to reach the SDGs for roads, electricity, water, and sanitation has been estimated at 2.7 percent of GDP per year (for emerging markets) and 9.8 percent of GDP per year (for low-income countries). These huge demands present both challenges and opportunities for government policymakers, finance officials, and PFM practitioners. On the one hand, there is a need to ensure that adequate resources are available to finance infrastructure development. On the other hand, there is an opportunity to use data and new technology to improve the efficiency and quality of infrastructure projects, as well as many other areas of PFM.

Machine learning (ML) and natural language processing (NLP) are two emerging technologies (see below) that are already having a big impact on many areas of economics and public policy. For example, in finance, ML is being used to detect fraud, automate financial analysis (e.g., loan applications and credit card transactions), and analyze large data sets (e.g., satellite images). But the technologies are also beginning to play a role in PFM, particularly in the areas of forecasting, risk management, and project monitoring. To cite the example of investment in public infrastructure, how can ML and NLP help forecast future investment needs, assess the impact of policy on infrastructure development, and automate the analysis of large volumes of data? And what are the potential risks of using these technologies?

Machine Learning and Natural Language Processing: Terms and Definitions

Machine learning (ML) is an area of computer science that gives computers the ability to learn without being explicitly programmed. The process involves using algorithms, usually based on statistical modeling, to automatically improve the performance of a task over time through experience (i.e., learning by example).

For example, the goal of a supervised learning algorithm might be to correctly classify objects into categories (e.g., animals, plants, or specific types of infrastructure) by analyzing a set of training data that includes the known classifications for each object. An unsupervised learning algorithm, on the other hand, might be used to cluster data points into groups without knowing the groups beforehand.

Natural language processing (NLP) falls under the broad category of ML. In NLP, computer programs are used to analyze and interpret human language (both written and spoken). NLP can be used for tasks such as speech recognition, text classification, sentiment analysis, machine translation, question answering, and chat bots.

Forecasting Future Infrastructure Needs

Making accurate forecasts of future infrastructure needs is often difficult because it requires understanding how economic activity will evolve and how this will impact the demand for infrastructure. ML can help address this challenge by providing a tool for forecasting future patterns of economic activity.

For example, ML can be used to develop models that predict infrastructure needs based on historical data. These models can consider a variety of factors, including population growth, GDP growth, and changes in sectoral composition. They can also include non-traditional data sources such as Google Search trends, and newspaper and social media sentiment analysis. The models can also be updated regularly to reflect changes in the underlying conditions. This would allow finance officials to have a better understanding of future infrastructure needs and the budget implications.

Assessing the Impact of Policy on Infrastructure Development

Another area where ML can help finance officials and PFM practitioners is in assessing the impact of policy on infrastructure development. For example, satellite imagery can be used to track the construction of new roads and assess their impact on economic activity. Additionally, ML can be used to improve the accuracy of predictions made by infrastructure planning models. This is important because the right policies can lead to a more efficient and effective use of resources, while the wrong policies can result in poor investment decisions.

ML can simulate how different policy choices would impact infrastructure development. For example, a simulation could be used to compare the impact of two different taxes on the development of a highway system. This would allow policymakers to compare how each tax would affect traffic flows, construction costs, and revenue generation. The simulations could also be used to assess the impact of other policy levers, such as changes in interest rates or energy prices.

Automating the Analysis of Large Volumes of Data

Another area where ML and NLP can help finance officials is in automating the analysis of large volumes of data. This would allow the officials to spend less time on data entry and more time on analysis and decision-making.

NLP can be used to develop algorithms that automatically extract information from text documents. This would allow finance officials to quickly analyze large volumes of data, such as tender documents or project reports. The algorithms could be programmed to look for specific patterns, such as cost overruns or delays in project delivery. NLP could also be used to develop chatbots[2] that provide information about infrastructure projects to the public, as well as assist with customer service enquiries.

Conclusion

As with any new technology, there may be risks associated with using ML and NLP. First, there is a risk that the technologies will be used to replace human judgment completely. This could lead to errors and bad decision-making if the algorithms are not properly calibrated or if they do not consider all the relevant factors. Second, there is a risk that the technologies will be used to reinforce existing biases and inequalities. For example, if data from the past is used to train ML algorithms, then existing patterns of inequality could be perpetuated. Finally, there is a risk that the use of these technologies will lead to a loss of privacy and confidentiality. This could happen if data is mishandled or if there are breaches in security. At the same time, the technologies offer many advantages that should be considered by policymakers and finance officials. The potential payoffs could be huge.

 

[1] Research Assistant, Fiscal Affairs Department, International Monetary Fund.

[2] A chatbot or digital assistant is a computer program designed to simulate human conversation. Chatbots are used in a variety of settings, including customer service and marketing. In policymaking, chatbots can help automate the process of gathering public input and feedback - for example, to collect data from citizens about their experiences with a particular government program, or to provide information about a specific policy proposal.

 

Note: The posts on the IMF PFM Blog should not be reported as representing the views of the IMF. The views expressed are those of the authors and do not necessarily represent those of the IMF or IMF policy.

Comments

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Comments are moderated, and will not appear until the author has approved them.

Back to top of page
©2007 IMF. All Rights Reserved. About Us | Terms of Use
/************* DO NOT ALTER ANYTHING BELOW THIS LINE ! **************/ var s_code=s.t();if(s_code)document.write(s_code)//-->