Opinion | Corona has put new forms of economic data into play
4 min read . Updated: 23 Jun 2020, 08:20 PM ISTMicro data is useful but it can’t replace the basic analytical framework we need for policy responses
How do you measure economic activity? This was an irrelevant question through most of human history. Economic stagnation was the rule rather than the exception. Little was to be gained by measuring something that did not change for centuries. The business of measuring economic activity began in earnest only when economic growth became the norm after the Industrial Revolution. The measurement of economic activity truly came into its own only in the twentieth century. Some of it was for war purposes. Some of it was to satisfy the needs of national economic planning.
Modern governments depend heavily on large surveys as well as administrative data to track their respective economies. Such data is the bedrock of estimates on gross domestic product, industrial production, farm output, inflation, etc. There are well established procedures for such estimates of economic activity. However, these procedures can sometimes crumble in extraordinary times, such as the one we are living through now.
The covid-19 shock has disrupted the usual business of national statistics. For example, the Reserve Bank of India (RBI) has temporarily suspended its economic forecasts for the quarters ahead. The lockdown has meant that government surveys of prices are now based on telephone interviews, rather than field visits to various markets, thus threatening their credibility. But there are more profound challenges as well.
Consider consumption patterns during the lockdown. The relative importance of items such as food or telecom services or entertainment have perhaps increased in most family budgets. And the relative importance of items such as petrol or eating out or consumer durables has decreased. A standard inflation index is weighted by consumer spending patterns. These change when an economy undergoes a large shock such as this.
Harvard University economist Alberto Cavallo has questioned the official consumer price inflation estimates for the US for the same reason. He has build an alternative covid consumption basket based on what the average US family is spending on during the pandemic. Two items dominate the shift—an increase in the importance of groceries and a decline in the importance of fuel because of less transport. The new price index that Cavallo has constructed based on the covid consumption basket shows that consumer inflation in the US is higher than the official government estimate based on the country’s traditional consumption basket. (In another context, economists working with the Billion Prices Project at the Massachusetts Institute of Technology, or MIT, had constructed an inflation index to show that Argentina was manipulating its inflation statistics earlier this decade.)
Analysts and economists in India have also been using new data sources such as the Google mobility report, Uber movement data, electricity consumption trends, digital payments, customs duty collections and E-Way bills to understand what is happening to the Indian economy, especially how it is recovering as India’s lockdown restrictions are eased. These high-frequency data points have been used for “nowcasting" economic activity, or estimating it in real time rather than predicting its future value based on past trends, as most forecasting models do. For example, some studies show that Google mobility data correlates well with industrial production in select countries where the link has been studied well.
Economists such as Roberto Rigobon at MIT and Raj Chetty at Harvard are in the vanguard of the new data revolution in economics. I would highly recommend the second Suresh Tendulkar Lecture given by Rigobon at RBI in 2018 and a recent Princeton University webinar that had Chetty as a speaker. The use of such new age data is here to stay. However, there are certain challenges that lie along the way, other than the obvious privacy concerns about online data even when it is anonymized.
First, there is the question of how well such micro data reflects changes in standard macroeconomic aggregates. This is especially true in a country such as India with a large informal sector. So data collected from the organized private sector may not reflect what is happening in the unorganized sector, especially during extreme episodes. We saw this during demonetization.
Second, a lot of such new-age data comes at different frequencies—daily, weekly, monthly. Crafting all this into one number is a tricky job, though not impossible. The fact that assorted components can have different seasonal patterns also complicates matters.
Third, it is not always clear whether a change in a particular data series is because of temporary substitution. For example, either the inability to go out of the house or the fear of using contaminated currency notes could have led to a spike in digital payments. This spike may reflect a change in the mode of payment, rather than signal an increase in underlying consumer demand.
Fourth, policymakers will continue to need strong analytical frameworks when thinking about how to respond to a shock such as covid-19. Even the richest data set cannot answer a key analytical question such as whether the world is facing a demand or supply shock. This is a key puzzle that would determine what the optimal policy response is. New data can complement analytical work; it cannot replace it.
Niranjan Rajadhyaksha is a member of the academic board of the Meghnad Desai Academy of Economics
Click here to read the Mint ePaperLivemint.com is now on Telegram. Join Livemint channel in your Telegram and stay updated
×![]()