How has inflation behaved during the pandemic?

As the pandemic started in early 2020, consumers’ spending patterns changed rapidly. This was partly due to the stringent lockdown policies implemented by governments across the world. Economists quickly raised the question of whether the standard official measures of inflation would capture price changes accurately (Dixon, 2020). 

The earliest studies found that the official measures of inflation tended to understate its value (Cavallo, 2020; Jaravel et al, 2020; Dixon, 2020; Sieler, 2020). This was because official indices gave excessive weight to items in the basket that had falling prices (such as petrol, clothing and footwear) and for which there was very little demand during the first lockdown. Since most of these early estimates were based on partial data, how has the story changed now that we have more comprehensive and complete data?

How did spending change during the pandemic?

Consumer price inflation (CPI) measures inflation with reference to a fixed basket of goods and services. This basket is constructed based on the weighted average of prices using the budget shares of a reference (or base) period. In the UK, as in many countries, these weights were based on pre-pandemic expenditure patterns taken from well before the outbreak of the pandemic (one or two years old). How did the expenditure shares change during 2020 compared to their pre-pandemic levels?

Table 1: Expenditure share in the UK

Source: ONS

Table 2: Expenditure Share in the United States

Source: BEA

The expenditure shares for 2020Q2 were very different from their pre-pandemic values, while the subsequent quarters were different but less so. Consumption patterns in the United States were also noticeably different from the global financial crisis of 2007-09 (Chetty et al, 2020). In the pandemic, there has been a large reduction in the services sector. During previous financial crises, spending on services remained unchanged while spending on durable goods fell sharply. 

Spending on luxury goods that do not require physical contact (such as landscaping services or home swimming pools) did not fall, while spending at hair salons and restaurants plunged at the start of the pandemic. Financial and professional services firms (which offer fewer in-person services) faced smaller losses compared with other businesses.

These sudden changes in spending patterns has introduced significant biases in the measurement of official CPI by different national statistical offices around the world. Measurement errors may also vary from region to region, and the accuracy of the estimations depend solely on the availability of the recent national account information. 

Figure 1: Spending changes by sector in the United States: Great Recession versus Covid-19

Source: Chetty et al (2020)

How have different lockdowns affected inflation?

National lockdowns have raised several major issues in the measurement of inflation statistics:

• First, the physical collection of prices by a member of the statistical agency has often not been possible during lockdowns.
• Second, many goods and services have become unavailable during lockdowns, and hence no prices have been collected.
• Third, spending patterns have changed a lot during lockdowns. Some items were completely unavailable to buy and, in some cases, items were available but consumers were less interested in consuming those items during, for example, petrol, clothing and footwear. (Dixon, 2020).

During the UK’s first lockdown, the Office for National Statistics (ONS) identified 92 and 76 items, respectively, as unavailable out of the usual 700 items for the April and May index (see Figure 2). The ONS has followed the Eurostat guideline of the Harmonised Index of Consumer Prices (HICP) to fix the unavailable items issue in the index calculation. The ONS method of imputation might have successfully reduced the expenditure share of unavailable items, but it does not capture the effect of lockdown and social distancing on the available items that are less in demand compared with the pre-pandemic period (such as clothing, footwear, petrol and diesel). 

In our own research, we compare three different measures of inflation during the pandemic (from its start to the most recent data). We evaluate: 

• The official CPIH measure published by the ONS
• The ‘lockdown weighted’ CPILW version of CPIH (developed by Huw Dixon and published by NIESR since April 2020)
• The CPIHES measure using the national income account data for expenditure shares

The main difference between the second and third measures is that the NIESR measure was in real time and based on estimates of the expenditure shares. In contrast, the third measure is based on the actual measures from national income accounts.

The official CPIH in the UK was quite low compared with CPILW (but less so than compared with CPIHES) in the first lockdown and over the next few months. The reason is fairly straightforward: stringent lockdown and social distancing policies have significantly changed consumer spending patterns and made some available items less in demand compared with the pre-pandemic period. 

The official CPIH measures for April and May were 0.90% and 0.70%, respectively, compared with 1.14% and 1.06% according to CPILW, and 0.91% and 0.76% according to CPIHES. The reason behind this low annual inflation rate was the negative annual inflation rate in several Classification of Individual Consumption According to Purpose(COICOP) divisions, such as clothing and footwear; furniture, household equipment and maintenance; and transport. These negative annual inflation rates contributed negatively to the headline annual inflation calculation. For example, the usual CPIH weight (expenditure share) in the transport division is around 12% (at pre-pandemic levels). But the HHFCE data for 2020Q2 had shown that the actual expenditure share in the transport sector was around 6.71%. This meant that the large change in the expenditure share caused the under-biased headline CPIH in the first lockdown, as well as during subsequent months.

Figure 2: Comparing official CPIH, CPILW, CPIHES – UK

Source: Author’s calculation based on ONS, Dixon (2020), Institute for Government (IFG)

In between the first and second lockdowns, the official CPIH understated inflation. The exception was during July and October, when the annual inflation rates were almost identical for all three measurements. The consistent negative (low) annual inflation rate in the clothing and footwear and transport sectors was a major reason for the under-biased inflation rate for the whole period. 

More specifically, the Eat Out to Help Out scheme in August 2020 caused a massive decline in the annual inflation rate for August in both the official CPIH and the CPIHES. The CPILW measure did not fall as much in August due to the low pandemic expenditure weight in the restaurant and hotel division. But the October CPILW measure was close to the other measurements because of the low annual inflation rate in the food and non-alcoholic beverages sector. 

In the second and third lockdowns, the official CPIH is overstated compared with the CPIHES and CPILW. The exception is during February 2021. The key point is that the ONS has changed its CPIH weight (expenditure share) significantly based on the previous year Household Final Consumption Expenditure (HHFCE).

The reasons behind over-biased official CPIH in the second and third lockdowns are as follows. First, there was a negative annual inflation rate in the food and non-alcoholic beverage sector along with the higher expenditure weight in the HHFCE data for 2020Q4 and 2021Q1. Second, there was increasing annual inflation rates in housing; water; electricity, gas and other fuels; transport; and the restaurant and hotels sectors with the higher CPIH weight in the ONS headline CPIH calculation.

But decreasing the annual inflation rate in the clothing and footwear sector with the higher CPIH weight in the ONS headline CPIH calculation has reduced some gap between the official CPIH and the CPIHES. Even so, it is not sufficient to equalise these two measurements. The CPILW measure was significantly higher in February 2021 because of the high negative annual inflation rate in the clothing and footwear sector.

How have different income groups been affected by changing inflation levels?

Official CPI was fairly low during the last year. Most countries found their headline inflation rate understated due to the sudden change in the consumption pattern. But the official CPI started rising from early 2021. In the UK, the headline annual inflation rate was 0.9% in January and 2.1% in May 2021. The situation is more severe for the United States, where the headline annual inflation rate was 1.43% in January and 5.06% in May 2021. 

An important question is how inflation affects different income groups. In the case of the UK, it is not possible to identify the effect of inflation on different income groups during the pandemic because we only have data until March 2020. Usually, the data come from Family Spending in the UK dataset which is collected by the Living Cost and Food Survey (LCF). We can get the US scenario from Cavallo (2020). From Figure 3, we have identified that US Covid-19 inflation is high for low-income groups compared with higher-income groups. The gap was greater during the early stage of the pandemic and started to shrink as stringent lockdown measures were relaxed.

Figure 3: US Covid inflation (All-times, 12-month changes)

Source: Cavallo (2020)

Conclusion

Lockdowns affected consumer spending considerably. The change in spending patterns has produced a significant bias in calculating the inflation rate by national statistical offices across the world. This bias has emerged from the expenditure weight, which changed substantially during lockdown periods, as consumption patterns shifted depending on what was available, what was allowed and what felt safe.

Who can I find out more?

• Consumer Responses to the COVID-19 Crisis. This VoxEU column by Andersen et al uses card spending data from Denmark to study the impact of the pandemic
• How Does Household Spending Respond to an Epidemic? Consumption During the 2020 COVID-19 Pandemic. This article by Baker et al uses household-level transaction data to study consumption patterns during the crisis
• Tracking the COVID-19 Crisis with High-Resolution Transaction Data. This working paper by Carvalho et al presents consumption data from Spain during the pandemic
• Inflation with Covid Consumption Baskets. This working paper by Alberto Cavallo presents consumption data from the United States during the pandemic

Who are experts on this question?

• Huw Dixon, Cardiff University
• Jagjit Chadha, NIESR
• Michael McMahon, University of Oxford
• Richard Davies, Economics Observatory & University of Bristol

Author: Aftab Chowdhury

Image Credit: Pickawood on Unsplash