Chowdhury Dixon 2023 Cardiff Wp E2023 17 Econstor

Summary

In Cardiff Economics Working Paper No. E2023/17, Aftab Chowdhury and Huw Dixon use actual household expenditure data from the national accounts to build a true inflation rate with the Fisher index for the first wave of the COVID-19 pandemic. They find that the official inflation rate in the 33 OECD countries overestimated true inflation in 22 countries and underestimated it in 11. Their results for countries where true inflation exceeded the official rate match those of Cavallo (2020) and Reinsdorf (2020), but differ significantly for countries where official inflation was higher.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

Text recognized from a scanned document; may contain errors. Page 18 of 31 is not shown because the text could not be recognized reliably.

A Service of

a | oe | Leibniz-Informationszentrum
e Wirtschaft
bd Leibniz Information Centre
L. for Economics

Make Your Publications Visible.

Chowdhury, Aftab; Dixon, Huw

Working Paper

Measuring inflation during the Pandemic with the benefit

of hindsight

Cardiff Economics Working Papers, No. E2023/17

Provided in Cooperation with:
Cardiff Business School, Cardiff University

Suggested Citation: Chowdhury, Aftab; Dixon, Huw (2023) : Measuring inflation during the Pandemic
with the benefit of hindsight, Cardiff Economics Working Papers, No. E2023/17, Cardiff University,

Cardiff Business School, Cardiff

This Version is available at:
https://hdl.handle.net/10419/284179

Standard-Nutzungsbedingungen:

Die Dokumente auf EconStor durfen zu eigenen wissenschaftlichen
Zwecken und zum Privatgebrauch gespeichert und kopiert werden.

Sie diirfen die Dokumente nicht fiir 6ffentliche oder kommerzielle
Zwecke vervielfaltigen, offentlich ausstellen, 6ffentlich zuganglich
machen, vertreiben oder anderweitig nutzen.

Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen
(insbesondere CC-Lizenzen) zur Verfiigung gestellt haben sollten,
gelten abweichend von diesen Nutzungsbedingungen die in der dort
genannten Lizenz gewdhrten Nutzungsrechte.

WWW.ECONSTOR.EU

Terms of use:

Documents in EconStor may be saved and copied for your personal
and scholarly purposes.

You are not to copy documents for public or commercial purposes, to
exhibit the documents publicly, to make them publicly available on the
internet, or to distribute or otherwise use the documents in public.

If the documents have been made available under an Open Content
Licence (especially Creative Commons Licences), you may exercise
further usage rights as specified in the indicated licence.

Mitglied der

Leibniz-Gemeinschaft 4
Cardiff Economics Working Papers

Cardiff Ysgol

Business Fusnes
School Caerdydd

Cardiff University Prifysgol Caerdydd

Working Paper No. E2023/17

Measuring inflation during the Pandemic with the
benefit of hindsight

Aftab Chowdhury and Huw Dixon

May 2023

ISSN 1749-6010

Cardiff Business School
Cardiff University
Colum Drive
Cardiff CF10 3EU
United Kingdom
t: +44 (0)29 2087 4000
f: +44 (0)29 2087 4419
business.cardiff.ac.uk

This working paper is produced for discussion purpose only. These working papers are expected to be published in
due course, in revised form, and should not be quoted or cited without the author’s written permission.

Cardiff Economics Working Papers are available online from:

http://econpapers.repec.org/paper/cdfwpaper/ and

Enquiries: EconWP@cardiff.ac.uk
Measuring inflation during the Pandemic with the

benefit of hindsight.‘

Aftab Chowdhury* and Huw Dixon’

Abstract

This study has adopted the actual household expenditure data from the national accounts to
construct a true inflation rate (using the Fisher index) and found that the official inflation rate in
the 33 OECD countries was an overestimate of true inflation for 22 and underestimate in 11
countries in the first wave of the COVID-19 pandemic. The result obtained for the countries
where true inflation was higher than the official rate in this study matches the results obtained by
Cavallo (2020) and Reinsdorf (2020). However, a significant difference has been detected for the
countries where the official inflation exceeds the true measure in this study. The core reason
behind the discrepancies is in the use of appropriate expenditure weights. This suggests caution

in using credit-card based expenditure data when spending behaviour has changed dramatically.

JEL codes: C43, E01, E31, 110
Keywords: Coronavirus (COVID-19), Pandemic, Inflation measurement, Household Final

Consumption Expenditure (HHFCE), Consumer Price Index (CPI)

* We would like to thank participants at the Cardiff ONS workshop in 2021 for their comments. In addition, we are
grateful to the participants at the 8th International Conference on Economics of the Turkish Economic Association
(ICE-TEA 2022) and the 2nd Welsh Postgraduate Research Conference in Business, Management & Economics
(WPGRC 2022) organized by the Cardiff Business School for their valuable comments. Faults remain our own.

* Cardiff Business School, ChowdhuryAM@cardiff.ac.uk (corresponding author).

8 Cardiff Business School.
1. Introduction

The first documented case of Covid-19 was found in December 2019, and within three months
the World Health Organization (WHO) declared Covid-19 as a Pandemic. Economies around the
world experienced significant losses due to the disruption in global supply chains, temporary and
permanent closures of businesses, temporary closures/limits of the production plants, restriction

on international travel etc. It had a massive effect on the spending patterns within economies.

As was realized early on (see for example Dixon 2020b and Tenreyro 2020), this raised specific
challenges for the measurement of CPI inflation, since the standard methods of constructing
inflation statistics were not designed for periods when there were rapid and large changes in
expenditure patterns. There were several attempts to use real time data such as credit card
expenditures to try to understand exactly how expenditure patterns were evolving and how this
might distort the official inflation figures. See for example Cavallo (2020) Chronopoulos et al.
(2020), Dixon (2020a), Jaravel and O'Connell (2020b) in the UK; Chetty et al. (2020), Bachas et
al. (2020), Dunn et al. (2020), Baker et al. (2020) in the US; Andersen et al. (2020) in the
Denmark; Carvalho et al. (2021) in Spain; Seiler (2020) in Switzerland. We review some of the

main papers published using real time data in Appendix B.

The official headline inflation of most countries uses the “Lowe” price index which is a
Laspeyres-type or fixed based weight index. Moreover, the advice offered by the international
agencies! was to keep the pre-pandemic weights. Where goods and services were unavailable
due to lockdowns the advice given was to impute the missing prices on the basis of the prices

that were available.”

As is well known, using a base weighted Laspeyres index will tend to overstate inflation (since
consumers are likely to consume less of items that become more expensive), whilst a current

weighted Paasche index will tend to understate inflation. The true inflation will be between these

' Advice from Eurostat to European Union Countries, UNCE, IMF, US Bureau of Labor Statistics.

* Diewert and Fox (2020) argued that the implicit price of unavailable goods was in fact very high (the price that
would have driven demand to zero in normal times), so that the imputation method recommended would likely
understate the real inflation.
two, and an alternative “ideal index” is the Fisher Index, which is the geometric average of the
Laspeyres and Paasche indices (Fisher 1921). In this paper, we construct a Fisher index to
evaluate the true inflation during the pandemic and compare to the official inflation figures with
this measure. Most statistical agencies have now published the actual expenditure shares for the
period 2020-2021 which we can use to answer the question of how the official measures
published in 2020-21 compare with the ex-post true inflation. Real time indicators used at the
time were mostly based on a limited range of consumer expenditure (mostly using credit card
data’, as in Cavallo 2020, Carvalho et al 2021, Reinsdorf 2020, and Sieler 2020) and the national
statistical agencies have since published figures that are much more comprehensive and cover
the whole range of consumer expenditure. We will also compare the actual data with some
examples of the contemporary real time data. Hindsight is indeed an advantage when it comes to

tracking what was going on in 2020-2021.

When we use the actual household expenditure data from the national accounts of the 33 OECD
countries, that the official inflation figures were too high in 22 countries (including Belgium,

Italy, Norway, and the UK) and too low in 11 (including the US, South Korea, and Iceland).*

This paper is structured as follows. Section 2 is about the measurement issues of the official
inflation rate during the pandemic. Section 3 explains the methodology and the data. Section 4
has measured the impact on the UK and the US inflation rates. Section 5 illustrates the impact on
the other OECD countries’ inflation rates and the differences between this study’s outcome with
some earlier studies. Finally, we conclude the paper by mentioning the importance of appropriate

data set in measuring inflation.

2. Measurement issues.

During COVID-19 Pandemic, the nationwide lockdown has produced significant challenges to

the inflation measurement by the national statistical offices (NSOs) as there were large and

3 The main exception being Jaravel and O'Connell (2020b), which used store scanner data. This is best for capturing
grocery expenditure and in particular food and beverage prices.

“ However, the results were a little mixed for 19 out of 33 countries: the bias during the 2020 Q2 was the opposite to
the whole period 2020-2021. See Table 3 below.
sudden changes in the consumers’ expenditure patterns. The standard method of measuring
inflation measure does not consider the possibility of sudden large changes in the expenditure
patterns, but rather a gradual adjustment with most NSOs updating the expenditure weights used
in the official CPI annually, with lagged expenditure data from previous calendar years (Jaravel
and O'Connell 2020b). Therefore, sudden large changes in the expenditure shares across different

sectors, products and outlets may all create biases in the official headline inflation rate.

There are of course other issues that made measurement more challenging during the pandemic.
The physical collection of prices by agents was not possible, so collection moved online. Many
goods and services became unavailable during the lockdown, and hence there were no prices to
be collected. Some items were completely unavailable to buy, and even when items were

available, consumers were less interested in purchasing the items during lockdown.

How did the UK calculate the official price index during the pandemic? If the item was
unavailable, or with limited availability (with the sample of price quotes being less than 20% of
its usual size), the Office for National Statistics (ONS) decided to impute the prices, based on the
method that best reflects the price behavior of the missing item. For unavailable items*, the ONS
followed the Eurostat guidelines to impute using one of three methods: firstly, imputing from the
index immediately above it in the classification structure; secondly, imputing based in the price
movement of a similar item; thirdly, carrying forward prices from earlier months. If the
unavailable item is seasonal, the ONS will use aggregate annual growth of all available non-
imputed items. However, if the unavailable item is not seasonal, the ONS will use aggregate

monthly growth of all available non-imputed items.

Dixon (2020b) argued that the imputation of unavailable prices is equivalent to altering
expenditure shares, since if we use the available prices to impute the unavailable prices, we are

effectively increasing the weight of the available prices in the price index.

> ONS has identified 92 items as unavailable for April Index, where 90 items for CPIH (around 16.3% of CPIH
basket), 90 items for CPI (around 20.2% of CPI basket) and 89 items for RPI (around 17.7% of RPI basket).

3
3. Data and Methodology.

3.1 Methodology

Generally, CPI uses “Lowe” price index which is a Laspeyres-type or fixed based weight index.
The Lowe price index measures the proportional change between 0 and t periods in the total
value of a specific basket of goods and services. The basket does not necessarily have to consist

of the actual quantities in same period. The formula for calculating Lowe Price Index is:

wih pra yi i i (i)

Where, /°’= Index value for period t based on period 0
p; = Price level of item i at period t
p’ = Price level of item i at period 0 (base period)
qi = Quantity on item I at period r (weighted reference period)

wor = Peat

is the weight or expenditure share of item i at period r (at 0 period price)

In the UK and most NSOs, the expenditure weights come from previous years and are used to
weight the price relatives across months within the current calendar year. For example, in 2020
the expenditure weights used by the ONS were based on expenditure data from 2018, whilst the
price relatives were between each month and January of 2020. Considering this fact, Diewert
and Fox (2020) suggest using the Fisher index because the substitution bias in Laspeyres (too
high) or Paasche (too low) or fixed basket index (too high) will be very large during the period of
exceptional changes in the consumer expenditure pattern. More studies are also advocating the
use of various superlative price indices such as the Fisher and Térnqvist indices during pandemic
period, including Fox et al. (2022); Jaravel and O'Connell (2020a); Kantur et al. (2021); Alvarez
and Lein (2020). Therefore, this study will adopt the Fisher price index as a pandemic/alternative
price index to measure the true inflation rate during the pandemic. So, how has the

pandemic/alternative price index been calculated?

4
Laspeyres price index: I,” = 1 0 = vet Ri".w? (i1)
i=1 Fi Ti
n itt —1
os Ot _ Liar PiVi n 1 ves
Paasche price index: J," = ==> = bares wi] (iii)
Yie1 Pi Wi i
. . . 0,t 0,t i
Fisher price index: J; = 1,0" x Ip )*0.5 (iv)

Where, /°*= Index value for period t based on period 0
pi = Price level of item i at period t
p’ = Price level of item i at period 0 (base period)
q; = Quantity on item i at period t

q? = Quantity on item i at period 0 (base period)

Re" = , 1S the price relative
Wi = F qe is the weight or expenditure share of item 1 at period 0 (at 0 period price)
‘= ae is the weight or expenditure share of item 1 at period t (at t period price)

The Laspeyres price index is the weighted arithmetic mean of price relatives. The basic
difference between Laspeyres and Lowe price index is the calculation of the expenditure share.
The Lowe price index uses a past reference period quantity, whereas Laspeyres uses base period
quantity. If the reference and the base periods are the same, then both the indexes provide
identical results. On the other hand, the Paasche price index is a weighted harmonic mean of
price relative which is based on the current period’s quantity. The Fisher price index is the
geometric average of the Laspeyres and the Paasche Price Indexes, which is also known as a

superlative or ideal price index.

To calculate the Laspeyres price index, the study will adopt the previous year’s consumer
expenditure data as the expenditure weight for the current year. For example, the expenditure
weight for 2020 is the expenditure share data from 2019. We will use expenditure share data of
2020 for the expenditure weights in constructing our Paasche price index in 2020. Therefore,
how frequently expenditure weights have been updated will depend on the availability of

expenditure share data in each country we consider.

5
3.2 Data

In this study, we use (depending on availability) the annual and quarterly/monthly household
final consumption expenditure (HHFCE) data from the national accounts of 33 OECD countries
to directly measure the changes in spending pattern during the pandemic. In the UK, the HHFCE
data is contained in Consumer Trends published by the Office for National Statistics (ONS). For
the US, we have collected personal consumption expenditure (PCE) data from the U.S. Bureau
of Economic Analysis (BEA), which is proxied by the consumer expenditure survey (CEX) from
the U.S. Bureau of Labor Statistics (BLS) due to the significant error in collecting data directly
from consumers during the pandemic period (Curtin 2022). The HHFCE data for the rest of the
countries in the list are collected mostly from Eurostat, OECD Stat, and other statistical agencies,
including Statistics Canada, Australian Bureau of Statistics, Statistics of Japan, Statistics Korea.
In Appendix-A, we will find the full list of countries along with the frequency of the data and

collecting institution.

4. Impact on the UK and the US Inflation

In this section we will look in detail at the UK and US experience to illustrate the main issues
involved in measuring the effect of the pandemic on inflation. Each country has a slightly

different set up for its CPI and indeed how it dealt with the pandemic.

4.1 Impact on the UK inflation

The first nationwide lockdown was enforced on the 23" of March 2020, and it continued until
the relaxing of restrictions and social distancing rules on the 23" of June 2020. The subsequent
lockdowns were the Second Lockdown (5" November 2020 to 2" December 2020) and Third
Lockdown (6" January 2021 to 12" April 2021). ° From the literature, we have already identified
that the consumers’ expenditure pattern has changed significantly during the lockdown period.

Table 1 has well demonstrated this concern with the help of actual household final consumption

® For details see Institute for Government Analysis, UK
expenditure during the COVID-19 pandemic period. Major changes took place during the first
lockdown. Among the 12 divisions in the COICOP category, significant changes happened in the
Food and non-alcoholic beverages; Alcoholic beverages and tobacco; Housing, water, electricity,
gas, and other fuels; Transport; Education; Restaurants and hotels division. However, the
expenditure patterns were moving to their previous track after the end of the third lockdown in
the UK. Although we have several references on the inflation measurement biases, it is time
worthy to cross-check the literature with help of actual national account information rather than

card-based data.

Table 1: Official expenditure weight and HHFCE in the UK

.. Official Expenditure Household Final Consumption Expenditure (HHFCE)

COICOP Divisions Weight
2020 2021 |2020Q1 |2020Q2|2020Q3|2020Q4 |2021Q1] 2021Q2 |2021Q3} 202104

01 Food and non-
alcoholic beverages 79 89 87 113 91 93 98 87 82 81
02 Alcoholic
beverages and 32 35 34 48 39 39 42 Al 38 37
tobacco
03 Clothing and 51 59 49 51 52 53 52 53 49 50
footwear
04 Housing, water,
electricity, gas and 296 328 265 333 277 283 297 270 259 258
other fuels
05 Furniture,
household 50 49 46 54 51 52 54 63 56 56
equipment and
maintenance
06 Health 22 20 20 18 20 22 22 18 17 17
07 Transport 120 107 128 66 108 101 91 110 122 126
08 Communication 17 19 17 21 18 18 19 20 19 19
09 Recreation and 136 112 115 | 120 | 115 120 | 124 107 104 104
culture
10 Education 24 30 25 32 27 28 29 26 25 25
11 Restaurants and 96 69 92 13 75 63 41 79 103 102
hotels
12 Miscellaneous 77 83 121 | 1231 | 127 | 129 | 133 | 128 | 126 | 125
goods and services

Source: ONS
Figure 1: UK Official and COVID Inflation

109.00 6.00

108.00
1 5.00

107.00

106.00

4.00
105.00

103.00
102.00
| 2.00
101.00
100.00 1.00
99.00
98.00 “0.00
E== Official CPIH Index Fisher Price Index
=== Official inflation rate (annual) «© Pandemic inflation rate (annual)

Source: Author calculation

Figure 1 shows how far the official headline inflation rate is from the pandemic inflation rate (the
Fisher index). The figure depicts both the price-index level (in bars) and the Pandemic inflation
rate. The two measures diverge in two periods: the official inflation rate is slightly higher than
the pandemic inflation in the first lockdown period (1.e., second quarter of 2020) and after the
third lockdown period (i.e., second quarter of 2021). The major contributors accounting for the
differences in the second quarter of 2020 inflation were restaurant and hotels, recreation and
culture, housing and household services, food and non-alcoholic beverage, clothing and footwear,
and transport. As the official inflation used the earlier expenditure weight, the higher inflation in
the restaurant and hotels, recreation and culture divisions have positively contributed to the
overall inflation. However, the increased expenditure in the food and non-alcoholic, housing and

household services divisions has also positively contributed to higher pandemic inflation but
remaining below the official headline inflation rate because of the sharp decline in the

expenditure on transport where inflation rate was largely negative.

In 2021, the ONS has updated its expenditure weight significantly considering the dramatic
change in the expenditure behaviour in 2020. The major contributor to the inflation in the second
quarter of 2021 are housing and household services, transport, recreation and culture, other
goods, and services divisions. As the actual expenditure pattern has started to return to the earlier
track, the increase in housing and household services expenditure weight in the official inflation
calculation is the main factor behind the difference with the pandemic inflation rate. However,
the official inflation rate and the pandemic inflation rate look quite similar after the second
quarter of 2021 because the increased inflation rate in the housing and household services and
the transport division were cancelled out due to the expenditure weight in these two different
inflation rate calculations. The official expenditure weight of the housing and household services
is higher than the actual expenditure and the vice-versa for the transport after the second quarter
of 2021. These are the main divisions where the change took place in the form of the inflation

rate as well as the expenditure weight in the two methods of calculating the inflation rate.

The UK experience shows how the comparison of the official inflation figures with the pandemic
measure depends on a detailed comparison of often offsetting effects across the expenditure

categories with offsetting trends.
4.2 Impact on the US inflation

10

The first COVID-19 case was found in the US on the 20" of January 2020. Since then, more than

one-million people have died and the US remains the top country in the world in terms of total

cases and deaths. Table 2 gives us information regarding the change in the expenditure pattern by

consumers during the period of the Pandemic. The expenditure pattern changed dramatically in

Personal Consumption Expenditure (PCE)

2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 2020

M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 Mill M12
Food and
beverages
purchased for 73 73 98 100 93 86 85 84 83 82 83 82
off-premises
consumption
Clothing,
footwear, and 29 29 23 19 24 29 29 29 31 30 29 28
related services
Housing,
utilities, and 183 185 202 236 214 200 198 196 193 193 194 197
fuels
Furnishings,
household
equipment, and | 43; | 43 | 46 | 47 | so | so | 49 | so | 49 49 49 47
routine
household
maintenance
Health 222 223 214 199 212 218 219 218 220 221 224 225
Communication | 19 19 21 23 21 20 20 20 20 20 20 20
Recreation 91 90 85 79 83 89 88 88 88 89 88 85
Education 22 22 24 24 22 21 21 21 21 21 21 20
Food services
and 72 72 56 42 49 56 58 61 61 61 59 58
accommodations
Financial
services and 83 83 91 103 92 88 87 87 86 86 87 89
insurance
Other goods and} 6, 67 | 64 | 59 | 60 | 62 | 62 | 62 | 62 63 63 63
services

the first wave, 1.e., from March to May 2020.

Table 2: Personal Consumption Expenditure in the US

10
11

Source: U.S. Bureau of Economic Analysis (BEA)

Figure 2: US Official and COVID Inflation

112.00 8.00
110.00 _ 7.00
108.00 : 6.00
106.00 e 5.00
104.00 : 4.00
100.00 H 2.00
98.00 : 1.00
96.00 | 9.00
E55 Official CP! Index Fisher Price Index
Official inflation rate (annual) =< © Pandemic inflation rate (annual)

Source: Author calculation

Figure 2 depicts the official inflation rate and the Pandemic inflation rate, as well as the
corresponding indices. There is a much greater divergence between the two inflation measures
than in the UK. In 2020 the Pandemic inflation rate was higher, reflecting large price changes
and the increased consumer expenditure in food and non-alcoholic beverage, alcoholic beverage,
housing, and household services. In 2021, the situation was reversed. This is partly a result of the
base effect (higher pandemic prices in 2020 tend to reduce inflation implied by prices in 2021),
but also the differences in expenditure shares (the Fisher price index price levels are below the

official levels).

The U.S. Bureau of Economic Analysis also calculates an alternative price index known as the

Personal Consumption Expenditure (PCE) Price Index, which is calculated using the Fisher-

11
12

Index.’ This is the measure primarily used by the FED in formulating its policy. In a report by
Curtin (2022) has also confirmed that during pandemic the data quality in the CE Survey by the
BLS is poor in comparison to the PCE by BE. In Figure 4 we show the PCE inflation measure in
addition to the official and Pandemic rates. The PCE and Pandemic are both Fisher indices but
use different expenditure data: our pandemic figure uses the final national accounts HHFCE data,
whilst the PCE uses the most recent data available when it is published. Figure 3 also supports
this proposition as the minimum bias is observed in the PCE Price Index. In 2020, the PCE
followed the Official figure more closely (both below the pandemic measure), whilst in 2021 the

PCE follows the Pandemic measure more closely (both below the official measure).

Figure 3: US Official, PCE and COVID Inflation

8.00
7.00
6.00
5.00
4.00
3.00
2.00

1.00

0.00

=== Official inflation rate (annual) = © Pandemic inflation rate (annual) = ° PCE Inflation rate (annual)

Source: Author calculation
5. Impact on the other OECD countries
The UK and US provide contrasting experiences of how the pandemic inflation and official

figures can differ. In the UK there is a light difference, with the pandemic being slightly lower,

whilst there is a more substantial difference in the US with pandemic inflation being higher in

’ The relative weights used in the PCE index are derived from business surveys—for example, the Census Bureau’s
annual and monthly retail trade surveys, the Service Annual Survey, and the Quarterly Services Survey.

12
13

2020 and lower in 2021. We now go on to explore the differences across the remaining OECD
countries using the same methodology, as shown in Table 3. The first two columns give the
difference between the official and pandemic inflation: in column | we see the difference in

2020 Q2 and in column 2 we see the difference over the two years 2020-2021.

Among the rest of the OECD countries, the official inflation rate is mostly higher than the
pandemic measure in the second quarter of 2020 and less over the whole COVID period of 2020-
21. A common pattern in those countries is that the bias is greater in the second quarter of 2020
compared to the whole year as a result of the large change in the expenditure patterns in that
quarter. An identical pattern observed in almost all the countries with a higher official rate: (1)
pandemic weight is higher where the price level is decreasing; (11) official weight is higher where

the price level is increasing.

In three countries out of the sample (Iceland, Poland and South Korea), the official inflation rate
was lower than the pandemic inflation in both the second quarter of 2020 and the whole COVID
period of 2020-21. Among these three, the changes in the expenditure pattern were quite similar
except in South Korea where the pattern changed earlier 1.e., in the first quarter of 2020. In these
countries, pandemic weights have increased in food and non-alcoholic beverage, alcoholic
beverage, housing and household services, furniture, health, communication, and education. In
addition, the price level has increased in food and non-alcoholic beverage, alcoholic beverage,
housing and household services, furniture, health, recreation and culture, education, restaurant,
and hotels; and the price level has decreased in clothing and footwear, transport, and
communication. Again, a similar pattern has also been observed in those countries where the
official inflation rate is higher, they are: (i) pandemic weight is higher where the price level is

increasing; (ii) official weight is higher where the price level is decreasing.

For most countries, there are mixed results, meaning the bias during 2020 Q2 was the opposite to
the whole period 2020-2021. In Norway, the official inflation rate was significantly higher than
the pandemic in 2020 and lower over the whole period. The opposite scenario was seen the
Czech Republic where the official inflation rate was lower in the second quarter of 2020 and

higher for the whole. Clearly, the differences can be very small (smaller than 0.1 pp in absolute

13
14

terms) ranging to over | pp in absolute terms in 2020 Q2 (Norway and the US). There is great
heterogeneity in how the pandemic impacted om the inflation experience in different countries,
reflecting different expenditure patterns, government policies and the behavior of households and

firms.

In columns 3 and 4, we show the earlier results obtained by Cavallo (2020) and Reinsdorf (2020)
using real time debit and credit card-based expenditure data. Whereas study is based on the
actual data from the national accounts with the benefit of hindsight, these studies were attempts
to measure the effects as they were happening. Both Cavallo and Reinsdorf used credit card data
from one country and applied it to several other countries: Cavallo used US card data, Reinsdorf
Canadian and US data. Even when applied to the US or Canadian data, the credit card data can
be misleading. For example, Reinsdorf (2020) and Cavallo (2020) use 37.12 and 35.80
respectively for the COICOP category 04 housing and household services in Canada, but the
actual data from Statistics Canada gives us 32.59 for this COICOP category. In the case of the
UK, the official expenditure share for the COICOP category 01 food and non-alcoholic beverage
in the second quarter of 2020 is 11.3 whereas the expenditure share for the similar category in
Cavallo (2020) is estimated as 29.5 in April 2020.* The difference is quite high and the potential
source of bias in the inflation measurement. Whilst using card data provides real time
information, it can be significantly different to the real expenditure behavior since it only

captures a part of expenditure.

8 Cavallo (2020) used Spanish card-based expenditure data to estimate the weight for the European countries in his
sample.

14
16

7. Conclusion

During the coronavirus pandemic, specifically the first wave (second quarter of 2020) of the
pandemic, the official Consumer Price Index (CPI) inflation figure wasn’t as informative as
usual about the balance of supply and demand in the economy. The reason behind that is the
conceptual challenges that have affected price measurement during the lockdown period. The
major change was that the large shifts in spending patterns due to the stringent lockdown policies
adopted by the government across the world, which changed the representative household

consumption basket.

The aim of this study is to identify whether the official inflation rate was higher or lower than the
true inflation rate (using the Fisher index) during the COVID-19 pandemic. Adopting the actual
household expenditure data from the national accounts, this study found that the official inflation
rate in the 33 OECD countries was an overestimate of true inflation for 22 and underestimate in
11 countries in the first wave of the COVID-19 pandemic. The result obtained for the
underestimating countries in this study has matched the result obtained by Cavallo (2020) and
Reinsdorf (2020). However, a significant difference has been detected for the countries where
the official inflation rate was higher than the true inflation rate in this study. The core reason
behind the discrepancies in the use of appropriate expenditure weight. Finally, this study
suggests caution while adopting card-based expenditure data in calculating any superlative price

indexes for measuring general price levels when spending behaviour has changed dramatically.

16
17

References

Alvarez, S. E. and Lein, S. M. 2020. Tracking Inflation on a Daily Basis.

Andersen, A., Hansen, E. T., Johannesen, N. and Sheridan, A. 2020. Consumer Responses to the COVID-
19 Crisis: Evidence from Bank ~~ Account Transaction Data. Available at:
https://ideas.repec.org/p/cpr/ceprdp/14809.html\

Bachas, N., Ganong, P., Noel, P. J., Vavra, J. S., Wong, A., Farrell, D. and Greig, F. E. 2020. Initial
Impacts of the Pandemic on Consumer Behavior: Evidence from Linked Income, Spending, and Savings
Data. National Bureau of Economic Research Working Paper Series No. 27617, doi: 10.3386/w27617

Baker, S. R., Farrokhnia, R. A., Meyer, S., Pagel, M. and Yannelis, C. 2020. How Does Household
Spending Respond to an Epidemic? Consumption During the 2020 COVID-19 Pandemic. National
Bureau of Economic Research Working Paper Series No. 26949, doi: 10.3386/w26949

Balleer, A., Link, S., Menkhoff, M. and Zorn, P. 2020. Demand or supply? Price adjustment during the
Covid-19 pandemic.

Barclays. 2020. UK Consumer Spending Report-August 2020. . PLC., B.B.U.

Benchimol, J., Caspi, I., Levin, Y., Benchimol, J., Caspi, I. and Levin, Y. 2021. The COVID-19 Inflation
Weighting in Israel. The Economists’ Voice 0(0), doi: 10.1515/ev-2021-0023

Bonam, D., Smadu, A., Bonam, D. and Smadu, A. 2021. The long-run effects of pandemics on inflation:
Will this time be different? Economics letters 208, p. 110065. doi: 10.1016/j.econlet.2021.110065

Carvalho, V. M., Garcia, J. R., Hansen, S., Ortiz, A., Rodrigo, T., Rodriguez Mora, J. V. and Ruiz, P.
2021. Tracking the COVID-19 crisis with high-resolution transaction data. Royal Society Open Science
8(8), p. 210218. doi: doi:10.1098/rsos.210218

Cavallo, A. 2020. Inflation with Covid Consumption Baskets. National Bureau of Economic Research
Working Paper Series No. 27352, doi: 10.3386/w27352

Chetty, R., Friedman, J. N., Hendren, N., Stepner, M. and Team, T. O. I. 2020. How Did COVID-19 and
Stabilization Policies Affect Spending and Employment? A New Real-Time Economic Tracker Based on
Private Sector Data. National Bureau of Economic Research Working Paper Series No. 27431, doi:
10.3386/w2743 1

Chronopoulos, D. K., Lukas, M. and Wilson, J. O. 2020. Consumer Spending Responses to the COVID-
19 Pandemic: An Assessment of Great Britain. Available at SSRN 3586723,

17
18

Curtin, S. 2022. Consumer Expenditure Surveys Program Report Series U.S. Bureau of Labor Statistics
(BLS)

Diewert, W. E. and Fox, K. J. 2020. Measuring Real Consumption and CPI Bias under Lockdown
Conditions. National Bureau of Economic Research Working Paper Series No. 27144, — dot:
10.3386/w27144

Dixon, H. 2020a. The measurement of inflation during the lockdown: a trial calculation.

Dixon, H. 2020b. Will CPIH inflation understate inflation for April 2020? ,

Dunn, A., Hood, K. and Driessen, A. 2020. Measuring the effects of the COVID-19 pandemic on
consumer spending using card transaction data. US Bureau of Economic Analysis Working Paper

WP2020-5

Fisher, I., 1921. The best form of index number. Quarterly Publications of the American Statistical
Association, 17(133), pp.533-551.

Fox, K., Levell, P. and O’Connell, M. 2022. New index number methods in Consumer Price
Statistics Economic Statistics Centre of Excellence (ESCoE).

Jaravel, X. and O'Connell, M. 2020a. High-Frequency Changes in Shopping Behaviours, Promotions and
the Measurement of Inflation: Evidence from the Great Lockdown. Fiscal studies 41(3), pp. 733-755.

Jaravel, X. and O'Connell, M. 2020b. Inflation spike and falling product variety during the Great
Lockdown.

Kantur, Z., Ozcan, G., Kantur, Z. and Ozcan, G. 2021. What pandemic inflation tells: Old habits die hard.
Economics letters 204, p. 109907. doi: 10.1016/j.econlet.2021.109907

Reinsdorf, M. 2020. COVID-19 and the CPI: Is inflation underestimated? Available at SSRN 3758057,
Seiler, P. 2020. Weighting bias and inflation in the time of COVID-19: evidence from Swiss transaction

data. Swiss Journal of Economics and Statistics 156(1), p. 13. doi: 10.1186/s41937-020-00057-7
Tenreyro, S. 2020. Monetary policy during pandemics: inflation before, during and after Covid-19.

18
Appendix-A: Selected countries, Frequency and Sources of HHFCE data

Frequency of

OECD Countries HHECE Data Source of HHFCE
Australia Quarterly Australian Bureau of Statistics
Austria Annually Eurostat

Belgium Annually Eurostat

Canada Quarterly Statistics Canada/OECD
Czech Republic Annually Eurostat

Denmark Annually Eurostat

Estonia Annually Eurostat

Finland Annually Eurostat

France Annually Eurostat

Germany Annually Eurostat

Greece Annually Eurostat

Hungary Annually Eurostat

Iceland Annually Eurostat

Ireland Annually Eurostat

Israel Annually OECD Stat

Italy Annually Eurostat

Japan Annually Statistics of Japan
Latvia Annually Eurostat

Lithuania Annually Eurostat

Luxembourg Annually Eurostat

Mexico Annually OECD Stat

Netherlands Annually Eurostat

Norway Annually Eurostat

Poland Annually Eurostat

Portugal Annually Eurostat

Slovakia Annually Eurostat

Slovenia Annually Eurostat

South Korea Quarterly Statistics Korea/Eurostat
Spain Annually Eurostat

Sweden Annually Eurostat

Turkey Annually Eurostat

United Kingdom Monthly Office For National Statistics

19

19
20

United States Monthly Bureau of Economic Analysis

Appendix B: Review of real time indicator papers published in 2020.

CPI measures inflation with reference to a fixed basket of goods and services, which is
constructed based on the weighted average of prices using the budget shares of a reference
“base” period. These weights are based on the expenditure patterns taken from well before the
index calculation period (one or two years old). During the first wave of coronavirus,
expenditure share has been changed significantly around the world that might made official
inflation rate biased. Over the last two year, several studies were conducted across the world for
identifying the expenditure pattern and measurement error in the headline inflation rate during
the COVID-19 Pandemic. Therefore, the continuous development in this topic has been
categorized as (1) the expenditure pattern and (2) the measurement issues of the inflation rate

during the Pandemic.

The expenditure pattern during COVID-19 Pandemic

The Covid-19 Pandemic has brought a dramatical change in consumer expenditure pattern in
many countries due to the lockdowns, social-distancing rules, and mobility restrictions.’ Cavallo
(2020) mentioned that consumers were spending less on transportation, hotels, restaurants,

recreation, and spending more on food and groceries.

Figure 4: Consumer Spending and CPI Basket Weights During the Pandemic

° Andersen et al (2020), Bachas et al (2020), Baker et al. (2020), Carvalho et al (2020), Chetty et al (2020),
Chronopoulos et al (2020), Dunn et al (2020).

20
21

84
8 S|
8
5
a
8 a
o
=] 4
Jan-2020 Apr-2020 Jul-2020 octo21 «February March April May Jime July’ AugustSepiember
—— Grocery ———“‘—sSC Apparel & Gen. Merch. [HM Food athome — J Alcoholic beverages HM pparel «= Housing
itis! nett cane: —-—-- Transportation EB Medical care 1 Transportation HB) Recreation Educ. and comm.
——— Entertainment and Recreation —-—-—-- Restaurants and Hotels [BN Food away home [iJ Other goods and services
(a) Consumer Spending (Opportunity Insights) (b) Covid Basket Weights

Source: Cavallo (2020)

Chronopoulos et al. (2020) found that discretionary spending remains relatively stable
throughout the incubation, outbreak, and most of the fever phases. However, the discretionary
spending has declined significantly during the lockdown phase as the government imposed strict
lockdown restrictions. '° After the announcement of Covid-19 as a pandemic by WHO, there was
a strong increase in the groceries spending which is consistent with the panic-buying and
stockpiling behaviour reported by the UK media. Further, grocery spending has declined
noticeably at the beginning of the Lockdown period. Moreover, spending on dining and drinking
has increased during the outbreak and the early weeks of fever period before declining. The
study also identified heterogeneity in consumer spending based on demographically and
geographically across the nation. For example, consumer based in Scotland appears to spend
more in the early stage of the outbreak period. Moreover, those consumers also appear to reduce
spending in dining and drinking before counterpart located in England and Wales. Furthermore,
the males spend significantly more than females, younger individual spends more than older and
high-income individual spend more than low-income counterpart. Overall, the study suggests

that consumer spending has declined significantly since the beginning of the Covid-19 outbreak.

‘0 Discretionary spending is the sum of spending in groceries, dining and drinking, alcohol, gambling, games and
gaming and other related items. The study categorized different phases as: Incubation (Jan 1 to Jan 17), Outbreak
(Jan18 to Feb 21), Fever (Feb 22 to March 22), Lockdown (March 23 to May 10) and Stay Alert (Mayl1 to
onwards).

21
22

Moreover, Barclays (2020) identified that the non-essential spending in the UK has declined
significantly after the government announcement of initial lockdown on 23" March 2020.!! The
study also identified an initial rise of spending in essential items after the WHO announcement
of Covid-19 as global pandemic. In overall, the consumer spending has declined significantly
after the announcement of lockdown. However, the spending growth is showing signs of
recovery since relaxation of government lockdown policy, which is basically led by the online
retail and home deliveries. The study also found a dramatic change in online shopping in UK
after the spread of virus. The online spending has grown by around 10.6%, whereas a decline in

in-store spending by around 6.7%.

Carvalho et al. (2021) has studied the evolution of aggregate expenditure across the provinces in
Spain and the postal codes within Madrid during the Covid-19 crisis using the high frequency
transaction data as a proxy to traditional data collected by the national statistical office.!* The
expenditure has declined dramatically in Spain after the implementation of strong lockdown and
social distancing policy and afterwards the expenditure seems fast recovery (more V than L-like)
due to the relaxation of public health measures. The expenditure share in Spain has experienced a
clear re-allocation due to the national lockdown. The spending on food (small stores), tobacco
stores, mobile phone credit, supermarket, hypermarkets (large surface supermarket), pharmacy
and para pharmacy has grown significantly. On the contrary, the spending on fashion, pubs and
disco clubs, furniture, restaurant, leather shops, shoe shops, toys, massage, and personal care has
collapsed dramatically. In fact, the expenditure categories that suffered most mainly come from

the lockdown / the state of alert.

Figure 5: COVID-19 and evolution of market share in Spain

‘| The study is based on the Barclays debit card and Barclays credit card transitions in the UK

Credit and debit card transaction data collected from Banco Bilbao Vizcaya Argentina, S.A. (BBVA), the second-
largest bank in Spain. The dataset contains 6 billion transactions collected from BBVA cardholders and the study
used 2.1 billion card transactions. The timeline of data collection is 1‘ January 2019 to 29" June 2020.

22
23

food
' bars and restaurants
fashion
transportation
health
MQ" other services
home and furniture
automobile
MEE technology and computers
hypermarkets, malls and dept. stores
=
=a

personal care and beauty
sports and toys
leisure and entertainment
hotels and lodging
travel
" books and newspapers

real estate

market share (7-day MA)

14 Feb
5 Mar 4
25 Mar
4 May
24 May +
14 June

& 14 Apr J

Source: Carvalho et al. (2021)

Similar outcome has also observed in the Danish economy in a study by Andersen et al. (2020).
The study was based on the transaction data of about 760,000 individuals at Danske Bank, the
largest retail bank in Denmark with a customer base which is a good approximation of the
Danish economy. The study addressed the strong cyclicality of spending over the week, the
month, and the year by comparing consumer spending on each day in 2020 to consumer
spending on a reference day 364 days earlier. The study first aimed to calculate the excess
spending!, then compute the crisis-induced change in spending as the difference between excess
spending in the post-lockdown period (March 11 to May 3) and excess spending in the pre-
lockdown period (January 2 to February 15). The study has come up with two findings. First,
aggregate spending has dropped by 27% relative to the counterfactual trajectory. This reveals
that excess spending was slightly positive in the pre-lockdown period and strongly negative in
the post-lockdown period. Second, the changes in consumer expenditure across categories have
been strongly correlated with the extent of supply restrictions. In overall, the spending increased
by more than 10% in the open sector (around half of the economy) and dropped by almost 70%

in the constrained sector (around one quarter of the economy).

‘3 Excess spending is the difference between spending on a given day in 2020 and spending on the reference day in

2019.

23
24

In USA, government statistics reveal that the Covid-19 pandemic led to a sharp reduction in
GDP and an extraordinary surge in unemployment. However, the national accounts data
published that most of the reduction in GDP came from a reduction in consumer spending rather
than business investment, government purchases, or exports. Chetty et al. (2020) has studied the
nature consumer spending during pandemic using the credit and debit card spending of
consumer.!* The study found that the major reduction in consumer spending in the U.S. has come
from the reduces spending by the high-income households. Moreover, most of the reduction in
consumer spending has been accounted for by reduced spending on goods and services that
requires in-person physical interaction, such as hotels, transportations, and food services, which
is completely consistent with the findings of other studies. The consumption pattern has
noticeably difference from the previous financial crisis. In the current pandemic, a large
reduction in services sector has been more apparent, whereas spending in service sector has
remain unchanged and spending in durable goods has fall sharply during previous financial crisis.
The study further identified that spending on luxury goods (that do not require physical contact,
such as landscaping services or home swimming pools) did not fall, while spending at salons and
restaurant plunged. Moreover, Financial, and professional services firms (that offer fewer in
person services) has faced less losses compare to other businesses. The income of the rich has
fallen less compared to the poor as the high-income individual can self-isolated more easily (i.e.,

work from home).

In line with the other literature, Baker et al. (2020) has also identified the sharp increase in the
retail, credit card spending and food items in the early stage of the pandemic, later it has
followed by a dramatic decline in overall spending as the lockdown has stringent in the U.S. The
study further explored the heterogeneity among partisan affiliations and demographics, which are
closely related to stated beliefs about the influence of the Covid-19 pandemic. The study found
that Republican were less concerned about the new virus. An Axios Poll between March 5 and 9
found that 62% of Republican were less concern (thought that the virus was largely exaggerated),

while 31% of Democrats and 35% of Independents thought the same. Moreover, a Quinnnipac

'4 Consumer spending data were collected from Affinity Solutions Inc., a company that aggregates consumer credit
and debit card spending information. The raw data were collected at the country-by-ZIP code income quartile-by-
industry-by-day level starting from January 1, 2019. Small business transactions and revenues were collected from
Womply, a company that aggregates data from several credit card processors to provide insights to small business
and other clients. Employment and earning data for low-income workers were collected from Earnin and Homebase.

24
25

Poll between March 5 and 8 also found that 68% of Democrats were concerned, while only 35%
of Republican were concerned. Despite lower concern about the new virus, Republican spent
more than Democrats in the early days of the pandemic. Additionally, the more spending by
Republicans at the restaurant and in retail shops also reveals the lower concern about the

infection of the virus.

The measurement issues of the inflation rate during COVID-19 Pandemic

The COVID-19 pandemic has forced almost all the nations across the world to implement
lockdowns, which brought a worldwide economic downturn. To formulate a national policy to
confront the economic crisis, it is essential to measure the changes in the price level accurately
during the crisis period. However, the nationwide lockdown has produced significant challenges
to the inflation measurement by the national statistical offices (NSOs) as there were large and
sudden changes in the consumers’ expenditure patterns. The standard method of inflation
measure has not considered the sudden changes in the expenditure patterns because most NSOs
update the expenditure weights used in the official CPI annually, usually with lagged expenditure
data Jaravel and O'Connell (2020b). Therefore, changes in the expenditure share across different

sectors, products and outlets may all create biases in calculating the headline inflation rate.

Most of the NSOs use the “Lowe” price index which is a Laspeyres-type or fixed based weight
index to calculate official CPI. Diewert and Fox (2020) states that fixed basket indexes, such as
Lowe Index to construct the CPI, are inadequate when there are dramatic changes in consumer
expenditure. Moreover, they also stated that the advice from the international agencies to
calculate the price index during lockdown leads to a Downward Bias in estimating changes in the
cost of living and an Upward Bias in estimating changes in real consumption. However, the
study suggests using the Fisher index when the consumption pattern has changed a lot during the
lockdown, as the amount of substitution bias in a Laspeyres (too high) or Paasche (too low) or

fixed basket index (too high) will be very large.

Dixon (2020a) has proposed guesstimate, a trial calculation of inflation during lockdown using
new (lockdown) expenditure weight and identified that CPIH would trend to overstate the fall in

inflation because it gives excessive weight to items in the basket which may have falling prices

25
26

(petrol, clothing, and footwear) and very little demand in lockdown. For example, during
lockdown the prices of clothing and footwear were falling as well as the consumption of those
items. But the official CPIH has applied 2019 expenditure weights to calculate the inflation
figure and hence understates the inflation. Similar result also obtained by the Jaravel and
O'Connell (2020b) using UK scanner monthly data of 30,000 household (purchase record)!>.
They measured that the month-on-month (MoM) inflation rate was 2.4% in the first month of
lockdown. The study identified that the distribution of inflation rates across product categories
has shifted rightwards compared with previous year. In overall, they have identified that the
lockdown coincides with unusually high inflation, which was experienced by almost all

households and in almost all product categories.

Similar evidence has been found in Switzerland by Seiler (2020). Using public data from debit
card transaction, the study constructs an alternative price index to measure the effect of the
Covid-induced weighting bias on the Swiss consumer price index. Therefore, the study found
that inflation was higher during the lockdown than the official CPI inflation. Basically, this is a
consequence of the relative increase in consumption of “Food & Alcoholic Beverage”, which are
more inflationary than other spending categories. The annual inflation rate of the Covid price
index was -0.4% by April 2020 compared to -1.1%of the equivalent CPI. However, the monthly
inflation rate has been recovered by the end of April due to the government policy changes.'® In
addition, the study also identified a persistent change in consumer behavior which is mainly
driven by “Low Touch” considerations due to new working habits, prolong uncertainty and the
lifestyle adopted during the lockdown period. The study predicts that this low touch
consideration keeps underestimating short to medium-term inflation through-out the year by

more than a quarter of a percentage point.

In absence of the national account information, most of the studies in measuring inflation bias
use card-based expenditure data. One of the pioneer study in this area was conducted by Cavallo

(2020) based on the card-based U.S. expenditure data'’. The study was conducted on the impact

'S The data has been collected by the market research firm Kantar FMCG Purchase Panel. The periods start from
December 30, through to May 17.

'6 In Switzerland, the initial Lockdown measure has been started on 16" March 2020. The first relaxation of
lockdown happened on 27" April 2020 and further relaxation took place on 11" May 2020.

'7 Data are publicly available as part of the Opportunity Insights Economic Tracker at Harvard and Brown
University.

26
27

of changes in expenditure patterns on the measurement of CPI inflation in 17 countries. The
study found that out of 17 countries 10 countries have higher inflation than that of the official
CPI. In case of US, Covid-CPI is greater than official CPI in both monthly and annual inflation
rate. In April, the annual inflation rate in US is 1.06% for Covid-CPI and 0.35% for official CPI.
The study also mentioned that Covid basket bias presents in the core index that excludes food
and energy. The Covid core deflation in April was only half of the core CPI, while the annual

inflation rate is at 1.73% compared to the 1.43% in the official core index.

Figure 6: US Covid Inflation (All-items, 12-month changes)

— CPI +++ Covid CPI +++ Covid- High Income --* Covid- Low Income

Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan
2019 2020 2021 2022

Source: Cavallo (2020)

Diewert and Fox (2020) has mentioned that the out-of-stock products are likely to have higher
market clearing prices than those of continuing goods, which potentially introduces an additional
downward bias in the CPI measurement. Moreover, Cavallo (2020) has also identified that some
retailer’s online delivery has soared during pandemic in US, where most of the retailer in this
platform tend to have higher prices than in their physical stores. Therefore, if this is not
accounted for in the collection methodology used by the national statistical office, the change in

spending outlets could cause another downward bias in the official CPI. The study found greater

27
28

Covid-CPI than Official CPI annual inflation in the following countries: Argentina, Brazil,
Canada, Chili, France, Japan, Korea, Spain, Turkey, Uruguay, and US. whereas annual inflation
by Covid-CPI is smaller than official CPI for: Germany, Greece, Ireland, Italy, Netherlands, and
UK. The study identified that most of the countries experienced higher Covid inflation rate is
driven by an increase in spending in “Food and Beverages” and a decrease in the weight of
“Transportations”. In contrast, closures of firms due to lockdown policy of government has

induced deflationary pressure rather than inflation Balleer et al. (2020).!8

Reinsdorf (2020) has done a similar exercise based in 83 countries in 09 different regions. The
study proxied the card-based expenditure data on US and Canada as expenditure weights for the
rest of the countries. The study found that out of 83 economies where official CPI is
underestimated in 65 countries during Feb-May 2020. Similarly, Kantur et al. (2021) conducted a
study in Turkey using the credit and debit card total expenditure data from January 2020 to
February 2021. They found that pandemic inflation is higher than the official inflation rate
during the first lockdown, suggesting a behavioral change in consumption.!’ In contrast to the
above-mentioned studies, Benchimol et al. (2021) following Cavallo (2020) methodology and
using credit card spending data found that the bias in the overall inflation rate is relatively

marginal in Israel.

In terms of methodology, almost all the studies emphasized the importance of using superlative
indexes in the time of dramatic changes in the spending behaviour of the consumer. However, it
is a matter of debate whether to use a superlative index or a simple Laspeyres type price index as
the availability of current expenditure data timely is quite challenging. Therefore, the core
motivation of this study is to identify whether card-based/home scanner/online data can be used

as a substitution for the actual household expenditure data from national accounts.

'8 The reason behind deflationary pressure is that the strongly negatively affected firms in manufacturing that
experience closures are 11% more likely to decrease prices than comparable firms that do not experience closure.
However, similar, and significant effects are exhibited for very strongly negatively affected firms in wholesale/retail.
Positively exposed firms that experience closures in service are less likely to increase prices.

'S Pandemic inflation as a geometric average of these two inflation figures Chained and Paasche.

28


File and source

File

chowdhury-dixon-2023-cardiff-wp-e2023-17-econstor.pdf

Size

2,556,915 bytes

SHA-256

b3ee633f2402707120cc010693e81234d39db736a8a473db5b753e704f64177e

Our copy

chowdhury-dixon-2023-cardiff-wp-e2023-17-econstor.pdf

Original

No public link identified.