Blog Post

Machine learning and economics

Machine learning (ML), together with artificial intelligence (AI), is a hot topic. Economists have been looking into machine learning applications not only to obtain better prediction, but also for policy targeting. We review some of the contributions.

By: Date: November 29, 2018 Topic: Innovation & Competition Policy

Writing on PWC blog last year, Hugh Dance and John Hawksworth discussed what machine learning (ML) could do for economics in the future. One aspect is that of prediction vs causal inference. Standard econometric models are well suited to understanding causal relationships between different aspects of the economy, but when it comes to prediction they tend to “over-fit” samples and sometimes generalise poorly to new, unseen data.

By focusing on prediction problems, machine learning models can instead minimise forecasting error by trading off bias and variance. Moreover, while econometric models are best kept relatively simple and easy to interpret, ML methods are capable of handling huge amounts of data, often without sacrificing interpretation.

Susan Athey provides an assessment of the early contributions of ML to economics, as well as predictions about its future contributions. At the outset, the paper highlights that ML does not add much to questions about identification, which are of concern when the object of interest, e.g. a causal effect, can be estimated with infinite data. Rather, ML yields great improvements when the goal is semi-parametric estimation or when there are a large number of covariates relative to the number of observations.

The second theme is that a key advantage of ML is that it views empirical analysis as “algorithms” that estimate and compare many alternative models. This approach contrasts with economics, where in principle the researcher picks a model based on principles and estimates it once. The third theme deals with the  “outsourcing” of model selection to algorithm. While it manages the “simple” problems fairly well, it is not well suited for the problems of greatest interest for empirical researchers in economics, such as causal inference, where there is typically no unbiased estimate of the ground truth available for comparison. Finally, the paper notes that the algorithms also have to be modified to provide valid confidence intervals for estimated effects when the data is used to select the model.

Athey thus thinks that using ML can provide the best of both worlds: the model selection is data-driven, systematic, and considers a wide range of models; all the while, the model selection process is fully documented and confidence intervals take the entire algorithm into account. She also expects the combination of ML and newly available datasets to change economics in fundamental ways, ranging from new questions and new approaches to collaboration (larger teams and interdisciplinary interaction), to a change in how involved economists are in the engineering and implementation of policies.

David McKenzie writes on the World Bank blog that ML can be used for development interventions and impact evaluations, in measuring outcomes and targeting treatments, measuring heterogeneity, and taking care of confounders.

One of the biggest use cases currently seems to be in getting basic measurements in countries with numerous gaps in the basic statistics – and machine learning has been applied to this at both the macro- and micro-level. But scholars are increasingly looking into whether ML could be useful also in targeting interventions, i.e. in deciding when and where/for whom to intervene.

McKenzie, however, points to several unanswered challenges, such as the question of the gold standard for evaluating these methods. Supervised ML requires a labelled training dataset and a metric for evaluating performance, but the very lack of data that these approaches are trying to solve also makes it hard to evaluate. Second, there are concerns about how stable many of the predicted relationships are, and about the behavioural responses that could affect the reliability of ML for treatment selection. Lastly, McKenzie also points to several ethics, privacy and fairness issues that could come into play.

Monica Andini, Emanuele Ciani, Guido de Blasio, Alessio D’Ignazio take up the question of policy targeting with ML in a recent VoxEU article and two papers. They present two examples of how to employ ML to target those groups that could plausibly gain more from the policy.

One example considers a tax rebate scheme introduced in Italy in 2014 with the purpose of boosting household consumption. The Italian government opted for a coarse targeting rule and provided the rebate only to employees with annual income between €8,145 and €26,000. Given the policy objective, an alternative could have been to target consumption-constrained households who are supposed to consume more out of the bonus. Applying ML to two household survey waves, the authors implement the second strategy. An additional econometric analysis suggests that this version of targeting would have been better, because the effect of the rebate is estimated to be positive only for the consumption-constrained households.

In the second application, Andini et al. focus on the “prediction policy problem” of assigning public credit guarantees to firms. In principle, guarantee schemes should target firms that are both creditworthy and rationed in their access to credit. In practice, existing guarantee schemes are usually based on naïve rules that exclude borrowers with low creditworthiness.

Aldini et al. propose an alternative assignment mechanism based on ML predictions, in which both creditworthiness and credit rationing are explicitly addressed. A simple comparison of the growth rate of disbursed bank loans in the years following the provision of the guarantee shows – confirmed by a regression discontinuity design – that the ML-targeted group performed better.


Republishing and referencing

Bruegel considers itself a public good and takes no institutional standpoint. Anyone is free to republish and/or quote this post without prior consent. Please provide a full reference, clearly stating Bruegel and the relevant author as the source, and include a prominent hyperlink to the original post.

Read article More by this author

Opinion

Europe may be the world’s AI referee, but referees don’t win

The EU needs to invest in homegrown technology.

By: Guntram B. Wolff Topic: European Macroeconomics & Governance, Innovation & Competition Policy Date: February 19, 2020
Read article More on this topic More by this author

Podcast

Podcast

The EU's plan to catch up on artificial intelligence

While the US and China have been setting the pace when it comes to Artificial Intelligence, the European Union seems to be lagging behind. What are the Commission's plans to finally catch up? Will AI increase the gap between big and small companies? This week, Nicholas Barrett is joined by Julia Anderson and Guntram Wolff to discuss the EU's plan for AI.

By: The Sound of Economics Topic: Innovation & Competition Policy Date: February 14, 2020
Read article More on this topic More by this author

Blog Post

The dynamics of data accumulation

The bigger you are, the more data you can harvest. But does data accumulation necessarily breed monopolies in AI and related machine learning markets?

By: Julia Anderson Topic: Innovation & Competition Policy Date: February 11, 2020
Read article More on this topic More by this author

Podcast

Podcast

Industrial revolutions might not be as fun as they look

AI promises a new industrial revolution but history warns us that industrial revolutions aren't always that fun for people in the eye of the storm. This week, Nicholas Barrett and Maria Demertzis spoke with Dr. Carl Frey, author of the book "The technology trap: capital, labor, and power in the age of automation", and Robert D. Atkinson, President of Information Technology and Innovation Foundation (ITIF), about how artificial intelligence will affect the job market.

By: The Sound of Economics Topic: Innovation & Competition Policy Date: February 5, 2020
Read about event More on this topic

Past Event

Past Event

The quality and quantity of work in the age of AI

At this event, the panelists will discuss the implications of Artificial Intelligence on the labour market and the future of work in general.

Speakers: Robert Atkinson, Anna Byhovskaya, Maria Demertzis, Carl Frey and Daniel Samaan Topic: Innovation & Competition Policy Location: Bruegel, Rue de la Charité 33, 1210 Brussels Date: February 5, 2020
Read article More on this topic More by this author

Blog Post

Do AI markets create competition policy concerns?

AI markets are young and their structure is yet to crystallise. Is European competition law ready for what happens next?

By: Julia Anderson Topic: Innovation & Competition Policy Date: January 23, 2020
Read about event More on this topic

Past Event

Past Event

The European AI policy: writing the first lines of the code

At this closed-door event, an open discussion with Margrethe Vestager will contribute to her work on artificial intelligence. The aim of these discussions is to set out a policy and regulatory approach in the form of a white paper, including the human and ethical implications of AI.

Speakers: Margrethe Vestager and Guntram B. Wolff Topic: Innovation & Competition Policy Location: Bruegel, Rue de la Charité 33, 1210 Brussels Date: January 21, 2020
Read article More on this topic More by this author

Podcast

Podcast

AI in Europe: a conversation with Google's CEO

It seems almost inevitable that Google will be big part of Europe's future. And Europe will be a huge part of Google's too. This week, Alphabet, Google's parent company, hit $1 trillion market cap for the first time. Can Google's AI be socially beneficial? Are big tech companies intrinsically bad? This week, Guntram Wolff talked to Google and Alphabet's CEO, Sundar Pichai.

By: The Sound of Economics Topic: Innovation & Competition Policy Date: January 20, 2020
Read about event More on this topic

Past Event

Past Event

Partnering with Europe on responsible AI: a conversation with Sundar Pichai, CEO Google and Alphabet

At this event, Google's and Alphabet's CEO Sundar Pichai will elaborate on his views on Artificial Intelligence.

Speakers: Sundar Pichai and Guntram B. Wolff Topic: Innovation & Competition Policy Location: SQUARE, Mont des Arts, 1000 Brussels Date: January 20, 2020
Read article More on this topic More by this author

Opinion

Could the U.S. economy be experiencing a hidden tech-driven productivity revolution?

In the last decade, most advanced economies have grown more slowly than before. Slower growth has frequently been seen as a legacy of financial crises, especially that of 2007–2009.

By: Marek Dabrowski Topic: Innovation & Competition Policy Date: January 6, 2020
Read article More on this topic More by this author

Blog Post

AI and the Productivity Paradox

In this blog post, I review the main explanations for this paradox and I briefly discuss relevant policy options in order to increase the contribution of AI on productivity

By: Georgios Petropoulos Topic: Innovation & Competition Policy Date: December 24, 2019
Read article More on this topic More by this author

Podcast

Podcast

The Sound of Margrethe Vestager

Will AI exacerbate the gap between big companies and small ones? Do ordinary Europeans gain anything from having European tech giants? This week, Nicholas Barrett and Guntram Wolff went to the Berlaymont to interview Margrethe Vestager, the Executive Vice President of the European Commission for a Europe Fit for the Digital Age.

By: The Sound of Economics Topic: European Macroeconomics & Governance Date: December 19, 2019
Load more posts