Comment on “Big Data in Asian Central Banks”

Abstract

My first comment is on why central banks have started using big data in the first place. Even before the Covid-19 pandemic, private financial institutions and investors extensively used big data, “alternative data” in their terminology, in predicting the future of stock prices. Central banks, including the Bank of Japan (BOJ), were concerned that with the rapid growth in the use of big data by the private sector, they would fall behind if it looked only at traditional data, and this was the driving force behind central banks' use of big data.

In 2020, as Covid-19 spread, the use of big data by the Fed, the BOJ, and other central banks accelerated even further. Here is what happened in Japan at that time. The Japanese government declared a state of emergency on April 7, 2020. In the Monthly Economic Report released on April 23, the Japanese government stated that “[t]he Japanese economy is getting worse rapidly in an extremely severe situation, due to the Novel Coronavirus” (sic).

What is important to note here is how this assessment was made. Normally, assessments in monthly reports are based on official government statistics. In this case, it is clear that the impact of Covid-19 on consumer spending, especially service spending, was pronounced, so the assessment in April last year should have been based on the “Family Income and Expenditure Survey” (FIES). However, this was not the case. The reason is simple: as of April 23, the latest information was for February 2020, which hardly reflected the impact of Covid-19.

Instead of the FIES data, the government used alternative (big) data. Specifically, in the April Monthly Report, expenditure figures calculated from credit card data (e.g., spending on eating out and spending at supermarkets) were included to show that an unprecedented sharp decline was occurring. Credit card spending figures had never been used in a monthly report before. The crisis thus suddenly provided an opportunity for the use of alternative data.

How will the use of alternative data by governments and central banks develop in the future? Once Covid-19 is under control, the need to rely on alternative data may diminish. Statistics, be they traditional or nontraditional, are a tool for responding to crises: In normal times, no one needs statistics and, in fact, no one pays much attention to them. However, once a crisis occurs, statistics are indispensable to obtain a real picture of the crisis. Since traditional data tend to be available too late to play a role in a crisis, we have no choice but to supplement traditional data with alternative data. Alternative data represent one of the means to deal with crises, and societies should make sure such data are available as part of their preparations for the next crisis.

Given the prospect that the use of alternative data by central banks will expand in the future, who is best placed to produce such data? This brings me to my second comment. There are two possibilities. The first is that central banks do everything themselves: they contact data holders, process the data, and make policies based on it. The second possibility is that central banks are involved only as users of alternative data compiled by the private sector.

Cornelli et al. (2022) seem to prefer the first possibility. Specifically, they state that it is desirable to reduce the dependence on private data service providers as much as possible because of high costs and the “significant legal and operational risks” of relying on them. However, many of the examples of alternative data use in Asian central banks cited by Cornelli et al. (2022) rely on data service providers such as Google. Even in the Japanese example given earlier, the government relied on alternative data provided by private companies.

If it is not practical for central banks to prepare an environment for alternative data on their own, they should seriously consider the second possibility: relying on the private sector. It is the private sector that holds the big data, not central banks or governments. It is also the private sector that is skilled in processing and analyzing big data. Based on this reality, central banks should start discussing how best to work with the private sector.

In doing so, as pointed out by Cornelli et al. (2022), it is necessary to pay sufficient attention to the protection of personal information and data ethics. In addition to these general matters, there are issues specific to central banks. The first of these is ensuring that data are provided on a sustainable basis. It is important to avoid a situation where data become unavailable due to changes in the business policies of private companies. The second is the need to deal appropriately with fraud by private data service providers. Central banks and governments will need to play a role in monitoring for and preventing such fraud. Further, countries will need to develop and coordinate rules for the handling of alternative data.