BSP to use machine learning apps – Manila Bulletin

The Bangko Sentral ng Pilipinas’ (BSP) big data use will increasingly apply machine learning (ML) applications or techniques such as natural language processing, nowcasting, and banking supervision, according to BSP Governor Benjamin E. Diokno.

ML, while often interchanged with artificial intelligence (AI) is not AI, but rather a subfield of AI involving algorithms to deliver output based on patterns learned from data.

Artificial Intelligence/photo

“The BSP continues to explore ML applications that can be useful in the conduct of its key functions while carefully taking note of the associated challenges,” said Diokno during an online press briefing on Thursday, May 5.

Diokno noted the continued advances in computing power and increases in digitized information that could provide further impetus for ML. “Central banks’ interest in ML has been increasing over the years, mainly due to its potential to enhance the existing tools used for regular monitoring as well as its ability to uncover underlying relationships between data to better understand the economy and the financial system,” he said.

The BSP, currently, while exploring use cases for ML, have noted a number of challenges and limitations with its use.

Diokno identified one of these challenges as the so-called black-box approach to ML.

“While empirical studies provide evidence that ML techniques can outperform traditional regression analysis in forecasting, there is difficulty in interpreting the causal relationships in ML models. By contrast, traditional econometric models allow users to make inferences on causation and identify how an explanatory or independent variable influences the dependent variable. This is critical to economic analysis and policy formulation,” he said.

The BSP is also challenged by ML algorithms and its accuracy in predicting tail risk events. “Institutions, including central banks, that wish to dabble in ML need to have sufficient computing power and a sizable and secure data storage capability,” said Diokno.

The BSP chief said the use of big data in ML also have data privacy, ethical data and modeling practice issues. For one, the BSP must handle data mined from the internet “delicately” and two, they must treat as confidential the large transactional databases such as payment data or bank reports since the law requires it.

“To address these challenges, the BSP is modernizing its IT infrastructure and enhancing the skills of its human resources. It has also put in place a data governance policy to ensure robust data management within the BSP,” said Diokno.

Meantime, BSP use cases for natural language processing is used to convert text into data to produce a quantitative summary, such as the news sentiment index and economic policy uncertainty index that are currently being developed.

The BSP also employs ML approaches to generate nowcasts of regional inflation and domestic liquidity. These models supplement the BSP’s existing suite of models for macroeconomic forecasting, said the BSP.

As for banking supervision, the BSP plans to use ML techniques to enhance its data validation processes and better identify atypical data.

“Different ML methods are being explored to determine which combination of methods would better identify atypical data based on an account’s historical behavior, relationship with other accounts, and peer-based movements,” said Diokno.

Last January, the BSP announced that it will create a new high-frequency monitoring index or the “News Sentiment Index” for economic and financial-related news from online sources. The index is part of its big data indicator.

The BSP has been using big data since it was set up by BSP with the help of the University of the Philippines’ School of Economics in 2019.

Big data, as defined by the BSP, is characterized by high volume, velocity or variety of data that cannot be processed using conventional tools and software, and that require specific technology and analytical algorithms for its transformation into value for mission critical processes.

Using big data, the BSP is assessing existing policies’ applicability and responsiveness to the current needs of the economy, including possible recalibration of policies as necessary such as when it amended its foreign exchange regulatory framework last year.



New Technology Era

Leave a Reply

Your email address will not be published. Required fields are marked *