"Tracking the COVID-19 Crisis with High-Resolution Transaction Data" (wi th Vasco M. Carvalho, Juan R. Garcia, Stephen Hansen, Álvaro Ortiz, Tomasa Rodrigo, and Pep Ruiz)
Abstract:
F inancial and payments systems throughout the world generate a vast amount o f naturally occurring, and digitally recorded, transaction data, but nation al statistical agencies mainly rely on surveys of much smaller scale for co nstructing official economic series. This paper considers billions of trans actions from credit- and debit-card data from BBVA, one of the largest bank s in the world, as an alternative source of information for measuring consu mption, a key component of GDP. We show, through extensive validation exerc ises against official consumption measures, that transaction data can usefu lly complement slow-moving national accounts and consumption surveys. We sh ow that this holds (i) over time, as a high frequency consumption proxy bot h at national and subnational levels; (ii) over consumption categories, ren dering it a naturally occurring consumption survey and (iii) over space, as a covariate-rich mobility dataset. We apply the idea of card spending as a real-time, geographically resolved and covariate-rich consumption proxy to the COVID- 19 crisis, where we present four findings: (1) a strong consump tion reaction to lockdown enactment and a rapid, V-shaped consumption recov ery at the global, national and subnational levels; (2) differential respon ses of expenditure to alternative lockdown policies at a subnational level; (3) an adjustment to the average consumption basket during lockdown toward s the goods basket of low-income house- holds; (4) a divergence in mobility patterns during lockdown according to income in which poorer households tr avel more during the workweek. We conclude that transaction data provides h igh-quality information about household consumption and mobility, rendering it a potentially important input into national statistics and research on household consumption.
DTSTAMP:20260403T173114Z DTSTART:20201005T163000Z DTEND:20201005T173000Z SEQUENCE:0 TRANSP:OPAQUE END:VEVENT END:VCALENDAR