Data disruption: The impact of Covid-19 on inflation measurement

Lockdowns, working from home, and social distancing caused people to spend larger shares of their household budgets on food and housing, while fewer people bought nonessentials, like airline tickets and clothing. And with incomes down as millions have lost their jobs, spending on nonessential items will likely remain depressed. 

The consumer price index (CPI) does not reflect these abrupt changes in spending patterns because the CPI weights are not continuously updated. For example, the CPI could be pulled down by a decline in the prices of nonessentials that are no longer purchased.

A new IMF staff paper uses spending estimates derived from credit and debit card data to adjust the CPI weights to match spending patterns during the pandemic. The study finds that inflation during the first three months of the pandemic was actually higher than we thought.  

The chart of the week looks at the difference over the February–May timeframe between a COVID-19 price index that adjusts the CPI weights based on the impacts of COVID-19 on spending in Canada and an index with unchanged CPI weights. The diamonds in the chart show the difference between the two indexes by region. In seven of the eight regions shown, the CPI is below the COVID-19 index. Looking at the average for all regions combined, the gap is 0.23 percentage points.

The main positive contributors to the gap between the COVID-19 index and the CPI are food and transport, each contributing 0.16 percentage points to the world gap. Rising food prices contribute to the faster growth of the COVID-19 index in all eight regions. Falling transport prices, which have a larger weight in the CPI than in the COVID-19 index, also contribute to the faster growth of the COVID-19 index in all regions except sub-Saharan Africa.

The main negative contributors to the world gap are housing, which contributes –0.03 percentage points, and clothing, which contributes –0.08 percentage points. Housing has a higher weight in the COVID-19 index than in the CPI, but its price index is so close to the overall CPI that increasing its weight does little to move the COVID-19 index away from the CPI. The downward effect of clothing is due to seasonal price increases having a smaller weight in the COVID-19 basket.

Despite the finding that CPI weights underestimated inflation in the early months of the pandemic, a quick update of the CPI weights to reflect the spending patterns during the pandemic would be impractical. Furthermore, introducing weights that are based on a short timeframe can reduce an index’s accuracy over the longer run. A better approach would be for statistical agencies to develop a supplementary index whose weights reflect spending patterns during the pandemic. This would give policymakers a better picture of the effect of inflation on the prices that consumers are actually paying. 

Next week’s 8th IMF Statistical Forum will delve deeper into the data disruptions and challenges arising from the pandemic.

Marshall Reinsdorf is a senior economist in the IMF’s Statistics Department.

The article first appeared on IMFBlog.