Cabello et al., Consumer Behavior in the Face of the Coronavirus Pandemic: An Analysis of Credit Card Spending (2022)
Summary
A 2022 Revista de Economia article by Andrea Felippe Cabello, Lorena Silva Brandão and Deise Maria Bourscheidt analyzing credit and debit card spending in Brazilian states during the 2020 coronavirus pandemic. The authors correlate changes in card spending across economic sectors with the Social Distancing Index (IDS) proposed by IPEA, using weekly spending for the 27 Brazilian states from customers of a large Brazilian multiple bank. They report that essential sectors without clear mobility problems saw increased consumption, while non-essential goods sectors fell sharply and then gradually recovered. Sectors associated with mobility fell sharply with no clear recovery in the period.
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
Cite this document
“Cabello et al., Consumer Behavior in the Face of the Coronavirus Pandemic: An Analysis of Credit Card Spending (2022).”
Original at web.archive.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/cabello-et-al-consumer-behavior-in-the-face-of-the-coronavirus-pandemi-ed12765591/
, SHA-256 3444c5fcfef2b675d0634754ad1e301fb93f22972c3f431fa9535daad76ae402.Full text
Revista de Economia
Consumer Behavior in the Face of the Coronavirus Pandemic: An
Analysis of Credit and Debit Card Spending in Brazilian States*
Comportamento do consumidor diante da pandemia do coronavírus: uma análise
dos gastos com cartão de crédito e débito nos estados brasileiros
Andrea Felippe Cabello, Lorena Silva Brandão e Deise Maria Bourscheidt** 1
Resumo: o artigo analisa os gastos dos consumidores efetuados com o uso de seus
cartões de crédito e débito e os correlaciona com as medidas de isolamento social
decretadas em decorrência do combate à pandemia do coronavírus de 2020. Foram
calculadas correlações entre a variação dos gastos com cartões nos diferentes setores
econômicos e o Índice de Distanciamento Social (IDS) elaborado pelo IPEA. Os
resultados mostram a importância de questões como a essencialidade dos bens, a
mobilidade de consumidores durante a pandemia e sua quarentena e a forma como
medidas de isolamento social incidem sobre cada setor de consumo. Setores essenciais,
sem problemas claros de mobilidade, tiveram aumento de consumo e comportamento
oscilatório de gastos de consumo durante o período. Setores de bens não essenciais
tiveram forte redução nos gastos de consumo seguida de gradual recuperação. Setores
associados à mobilidade (sejam essenciais ou não) tiveram forte redução e não tiveram
recuperação clara no período. Além disso, há fortes indícios de que variações
possivelmente institucionais entre os estados tenham efeitos relevantes.
Palavras-chave: Coronavírus. Covid-19. Gastos de consumo. Gastos em cartão.
Medidas de isolamento social.
Abstract: the paper analyzes consumer spending with the use of their credit and debit
card and correlates it with the social isolation measures enacted as a result of the 2020
coronavirus pandemic. Correlations were estimated between the variation in card
spending in the different economic sectors and the Social Distancing Index (IDS)
proposed by IPEA. The results show the importance of issues such asthe essentiality of
goods, the mobility of consumers during the pandemic and quarantine, and how social
isolation measures affect each consumer sector. Essential sectors without clear mobility
problems, had an increase in consumption and oscillatory behavior of consumption
spending during the period. Non-essential goods sectors had a strong reduction in
consumption spending followed by a gradual recovery. Sectors associated with mobility
*
Submissão: 09/09/2020 | Aprovação: 07/02/2021 | DOI: 10.5380/re.v43i81.76420
**
Respectivamente: (1) Professora do Departamento de Economia da Universidade de Brasília (UnB),
Brasil | ORCID: 0000-0003-1489-0676 | E-mail: andreafc@unb.br | (2) Departamento de Economia da
Universidade de Brasília (UnB), Brasil | ORCID: 0000-0002-0797-2852 | E-mail:
lo.lorenabrandao@gmail.com | (3) Departamento de Economia da Universidade Federal da Fronteira Sul
(UFFS) | ORCID: 0000-0001-6782-8628 | E-mail: deise.bourscheidt@uffs.edu.br
Esta publicação está licenciada sob os termos de
Creative Commons Atribuição-Não Comercial 4.0 Internacional
Revista de Economia
(whether essential or not) had a sharp reduction and did not have a clear recovery in the
period. In addition, there are strong indications that possibly institutional variations
between states have relevant effects.
Keywords: Coronavirus. Covid-19. Consumer spending. Credit/Debit Card spending.
Social isolation measures.
JEL: E20.
Cabello, Brandão e Bourscheidt 315
1. Introduction
Epidemics have not been uncommon events throughout history. In the 20th
century, besides the Spanish flu in 1918, the world was also affected by the Asian
flu (H2N2) in 1957 and Hong Kong flu (H3N2) in 1968. Other diseases, with
global economic and health consequences, also marked the 21st century. The first
of these was the Severe Acute Respiratory Syndrome (SARS), which appeared in
late 2002 and mainly affected China and Hong Kong. In 2009, the World Health
Organization (WHO) announced avian influenza (H1N1) as a pandemic and, in
2012, the Middle East Respiratory Syndrome (MERS), the epidemic associated
with the Ebola virus, which has two recent episodes between 2014 and 2016 and
between 2018 and 2019.
As common features, these epidemics have an impact on the real economy.
The magnitude of these impacts varies according to the capacity of spread of each
disease and the health interventions carried out, such as, for example, measures of
social distance to flatten its epidemiological curve (Chronopoulos; Lukas; Wilson,
2020). With a pandemic, the economic impacts are global and the first aspect
observed are the shocks to supply chains and their commercial implications and
world production.
From an economic perspective, the effects may differ between different
countries, depending on their characteristics and the strategies adopted. In the case
of the 2020 coronavirus pandemic, despite the relatively low lethality, the virus
has a high degree of contagion and the need for social isolation is the main measure
against the spread of the pandemic. Therefore, besides the consequences for the
health system, the coronavirus crisis involves several economic aspects such as:
consumption reduction, interruption of some supply chains, stock market crisis,
company bankruptcy, unemployment, income reduction, among others. Fernandes
(2020) suggests that the economic impacts are not determined by the mortality of
the virus, because the strategies adopted by countries and their governments can
be a source of shock.
As it occurs worldwide, the spread of the virus in Brazil also happens in a
heterogeneous way, as well as its consequences for the health system, the political
environment, and economic impacts. Thus, our goal is to investigate the
consequences of the pandemic on consumer behavior, based on its relationship
with legal measures of social distance, during the first twenty-four weeks of 2020.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 316
As a proxy for family consumption, we use changes in the volume and
spending categories of purchases made with debit and credit cards. The data covers
ten types of merchants and the period from 01/06/2019 to 06/22/2019 and
05/01/2020 to 06/20/2020. Weekly spending for the 27 Brazilian states were
analyzed. Legal measures of social distance were measured using the index of legal
measures of social distance (IDS) proposed by IPEA (2020).
The analysis carried out in this article points to changes in consumer
behavior, which vary regionally according to the essentiality of the goods,
consumers mobility and the length of the period of social distance. In addition to
this introduction, the article presents a brief literature review with a synthesis of
some recent work. Then, the methodology highlights how the empirical analysis is
developed in third section. The fourth section covers the results and discussions,
ending with the considerations in the fifth section.
2. Consumer behavior and spending during the Covid-19 pandemic
Despite the short time since the start of the 2020 coronavirus pandemic,
concern about the impacts of the pandemic on the economy have already led to a
fruitful literature. Analyses have been carried out on the length of the isolation
period, seeking to minimize adverse effects (Tellis et al., 2020; Bjornskov, 2020;
Acemoglu et al., 2020; Crokidakis, 2020; Morato et al., 2020), the effects on the
global economy (Fernandes, 2020; Mckibbin; Fernando, 2020; Ozili; Arun, 2020;
Baldwin; Mauro, 2020) or on the Brazilian economy (Dweck et al., 2020;
McKubbin; Fernando, 2020; World Bank, 2020; Cereda et al., 2020).
In relation to the latter, some focus on the impact of the pandemic and the
measures adopted to combat it on consumption. It is within this literature that this
work is inserted. Several studies have been done in several countries around the
world with congruent results.
The study developed by Binder (2020) in the USA shows that the main
concerns facing a scenarioof coronavirus spread refer to the economic effects of
the pandemic in the country on family health and personal finances. These
concerns materialize in the expectations of a higher level of inflation and
unemployment. Regarding consumption habits, it appears that concerned
consumers have opted to cancel trips and buy more food.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 317
Dietrich et al. (2020) seek to understand how the expectations of North
American families adjust to the face of the pandemic. The results point to
anticipated uncertainty and news shocks causing families to implement changes in
financial planning, reduce spending and have more fear of unemployment. The
uncertainty shock also correlates with more spending on food and medicine.
Baldwin and Mauro (2020) analyze how exposure to news about Covid-19
can impact individual consumption. According to the authors, two aspects
influence demand shock: the practical and the psychological. Social isolation
“makes it impossible” for consumers to go shopping (or use services) in physical
establishments. That is, uncertainty and its psychological consequences make
economic agents more likely to adopt a “wait and see” posture.
Some work has made use of financial data to analyze the impact of the
pandemic on consumer spending habits in various countries. Chronopoulos,
Lukas, and Wilson (2020) used current account transactions and the use of credit
and debit cards to investigate the impact of the coronavirus on consumer spending
in Great Britain. Their timeline comprises the initial and spread period of the
pandemic, specifically from January 1 to April 7, 2020. The authors also divided
the analysis into four periods, considering that the way people deal with the
pandemic differed over time. Thus, the analysis is divided into four phases:
incubation (January 1st to 17th), outbreak (January 20th to February 21st), fever
(February 24th to March 22nd), and lockdown (March 23th to March 7th) April).
The results point to a trend toward a continuous reduction in total spending in the
phase called fever and an even greater decline in the first two weeks of lockdown.
The effects vary according to demographic characteristics and location.
Grocery spending (the stocking up of food) increased at the beginning of
the fever period and in the two weeks after the WHO characterized Covid-19 as a
pandemic, but after the lockdown phase, grocery spending decreased considerably.
Regarding the consumption of meals and beverages, there was an increase in
spending until the second week of the fever period, then the numbers related to
these items decreased by 47.1% over four weeks, with similarities across the
surveyed locations.
Joyce and Xu (2020), also in a study for the United Kingdom, consider that
retail trade (non-food and non-pharmaceutical), passenger transport,
accommodation, restaurants, tourism, arts, leisure, personal care and domestic
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 318
services are most impacted by social isolation measures, as a result of
Chronopoulos, Lukas and Wilson (2020). The analysis concludes that the impacts
are not uniform, with young people under 25 and low-income people more likely
to work in these fields of activity, drawing attention to the effects on the falling
incomes of those employed in these sectors.
The study by Chen, Qian, and Ven (2020), on the other hand, analyzes the
impact of Covid-19 on consumption in China using data from banking transactions
conducted with offline credit cards. Their database comprises the period between
January 1 and April 14, 2020, and as a counterfactual period, they use the same
days in the year 2019. The intention is to check the change in total spending and
also in consumption by type, differentiating goods from services and separating
categories, which are: daily necessities, durable goods, discretionary goods,
restaurants, entertainment, and others. Among the key findings, the researchers
show that the median city recorded a 33% drop, with more than 90% of the cities
analyzed recording a drop in consumption of more than 20%. The study finds a
1.2% drop in GDP through offline consumption in the twelve weeks after the
outbreak.
Carvalho et al. (2020) conducted a survey to check consumption patterns in
Spain using 1.4 billion transactions with individual cards from Banco Bilbao
Vizcaya Argentina, S.A (BBVA). Nominal spending, in the period immediately
after the announcement of a national blockade by the Spanish government on
March 14, is up to 49% lower than in the same period in 2019. On days prior to
lockdown, there is an increase in purchases for stock, as well as a decrease in
offline spending, in physical establishments, compared to online purchases, which
also decreased. Furthermore, they observe a substantially heterogeneous behavior,
when analyzing expenditures by category, during confinement. Spending on goods
related to basic needs, or products with very low elasticity of demand (such as
tobacco) more than doubles during the lockout period, compared to the same
period last year. Spending on goods and services with higher elasticity of demand
(related to food and out-of-home entertainment, fashion, or personal services)
almost disappears.
A study by Alexander and Karger (2020) in the US assesses the
heterogeneity of consumption spending in response to requests for isolation
between March 1 and April 17, 2020. The financial data used to measure
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 319
consumption was small business revenue data from Womply and large business
spending from Second Measure, which is complemented with information about
the characteristics of different municipalities. Mobility data is extracted from cell
data locations. The main results show, in addition to a significant reduction in
mobility, a drop in revenue from restaurants and personal purchases.
In relation to studies that analyzed the Brazilian case, we can mention
Bretas and Alon (2020), who used data from webinars to analyze the impact of the
pandemic on the franchise sector in Brazil. Non-essential expenses have been cut,
such as consulting services, communication and maintenance. The retail sector
would have been more affected and there would be great concern with the post-
pandemic return, especially with the adoption of new technologies to mitigate the
effects on the business.
Thus, despite an incipient literature, there are already some stylized facts -
a gradual change in behavior throughout the quarantine period possibly motivated
by the public's reaction to news about the pandemic and the strategies adopted as
the social isolation measures. In countries where studies have been carried out,
there is an increase in the consumption of essential goods at the beginning of the
pandemic and the duration of social isolation measures, followed by a reduction in
this consumption as the pandemic and the isolation period lengthen. These
findings, therefore, will guide our analysis.
3. Methodological aspects
The analysis of consumer behavior in the face of the coronavirus pandemic
will be carried out based on the volume of credit and debit card spending by
individual customers of a large Brazilian multiple bank1. The information is daily
and detailed in ten categories of commercial activities (or services). The economic
1
It is a financial institution with 5,316 service units distributed in 3,154 municipalities in the country. In
19 states, branches and/or advanced service posts (PAA) cover more than 50% of all municipalities. In
Goiás, Minas Gerais and Maranhão, between 40 and 50% of the municipalities have a bank presence at this
institution. In Paraíba, Amazonas, Rio Grande do Norte, Tocantins and in Piauí less than 40% of the
municipalities have branches or PAA of this bank. Regarding the banking system of the Brazilian
population aged 15 or over, according to information from the Central Bank and the IBGE in 2019, 89.3%
of the Brazilian population has a banking relationship with a financial institution. In the South region this
percentage is 95.1%, in the Southeast 94.1%, in the Midwest 86.5%, in the Northeast 83.7% and in the
North 74.7%.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 320
impact, in percentage, will be measured by the variation between purchases made
in the first 24 weeks of 2020, compared to the same period in the previous year.
Expenses with debit and credit cards are generally considered to be variable
expenses of individuals and, for this reason, capture more exogenous variations
such as social isolation, since fixed expenses such as rent, taxes, etc. Are generally
not paid with debit or credit cards and should not have such a strong impact, even
in a situation of falling individual income. Furthermore, we draw attention to the
fact that, in a country like Brazil, the use of debit and credit cards assumes that the
individuals analyzed have a minimum level of income, thus excluding a significant
portion of the population.
Ipea (2020) measures the IDS, which is correlated with the variation in
consumption. Regarding the spread of the pandemic, the variables of interest in the
study are the numbers of new cases (infected) and deaths per 100,000 inhabitants,
also accumulated per week and per federation unit. Table 1 shows the description
of the variables.
Table 1 – Description of the variables used:
Variable Description
Credit card and debit card Variation in credit card and debit card spending in the
spending by individual first 24 weeks of 2020 compared to the same period in
consumers the previous year (2019) in each state. The 2019 figures
were deflated by the IPCA.
Legal Measures of Social Index weekly average (originally daily).
Distancing Index (IDS)
Number of Covid-19 Number of newly infected per week by Covid-19, per
Infected 100 thousand inhabitants, in each Brazilian state.
Number of Covid-19 Number of deaths, in the reference week, by Covid-19,
Deaths per 100 thousand inhabitants, in each Brazilian state.
According to Ipea (2020), the IDS measures the degree of rigidity of legal
measures of social distancing adopted due to the Covid-19 pandemic, based on the
policies adopted by state governments and city halls in the capitals of Brazil. The
information used comes from news published by the press and legal documents
from governments, such as decrees, ordinances, provisional measures, rules,
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 321
resolutions, etc. The study considers that, although all states have adopted
measures to contain the epidemic, they took place in different periods and formats.
The IDS considers the following variables: i) whether the holding of events
and the opening of cultural, sports or religious establishments was suspended; ii)
whether the activities of bars, restaurants and similar establishments were
suspended; iii) whether the activities of commercial establishments and services in
general (except essential ones) were suspended; iv) whether the activities of
industrial establishments (except essential ones) were suspended; v) whether
classes were suspended and vi) whether there were restrictions on passenger
transport.
According to the study carried out by Ipea (2020), each variable was
assigned a value of 2, 1 or 0, indicating, respectively, total or partial suspension or
restriction. The IDS is calculated by the adding these values, which can vary from
0 to 12, subsequently, the index was adjusted to be between 0 and 10, with 10
being the states and/or capitals with the greatest restrictions.
The Ipea study (2020) clarifies that, in general, the value 2 is attributed to
the total suspension of activities considered non-essential, as well as operation
through on-site delivery or removal. Value 1 applies to cases in which some
establishments, in addition to the essential ones, can operate. Beside these, the
value 1 also characterized activities conditioned to certain criteria or rules. In
situations where there are no restrictions, the assigned value was 0.
We correlated the variation in spending with credit and debit cards and IDS,
considering ten commercial categories in the 27 states. They are: supermarkets;
drugstores and pharmacies; gas stations; furniture and department stores;
Construction Materials; clothing and sporting goods; other durable goods sectors;
tourism and transportation; other services; and bars and restaurants.
From March to April, according to the IPEA indicator, the weighted average
IDS of the states was above 6.0. In May and June, this average was between 5.7
and 5.2. As of June 29, there was a reduction in the legal measures of social
distance and, since then, the IDS has remained below 5.0. As of the second week
of September, the IDS is below 4.0. These indicators suggest a greater flexibility
in social isolation starting in June, which justifies the timing of this research.
According to the Ministry of Health (2020), in Brazil, the first record of the
disease occurred on February 26, the number of confirmed cases until June 20 was
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 322
1.1 million and deaths totaled 50.0 thousand. The Northeast region has the highest
number of cases, but from a relative perspective, the number of cases and deaths
per 100,000 inhabitants in the North is more than double the Brazilian average.
Thus, and following Chronopoulos, Lukas and Wilson (2020), three time frames
were considered, according to table 2:
Table 2 – Description of the time frame considered:
Period Description
01/05/2020 to Period corresponding to the first 24 weeks of the year 2020.
06/20/2020 The variation in expenses was compared with the period
from 01/06/2019 to 06/22/2019.
02/23/2020 to The analysis is made from the eighth week of the year,
06/20/2020 when the first case of coronavirus in Brazil was reported.
The variation in expenses was compared with the period
from 02/24/2019 to 06/22/2019.
03/15/2020 to The analysis is made from the eleventh week of the year,
06/20/2020 when the first death from coronavirus occurred and the first
legal measures of social isolation were implemented in
Brazil. The variation in expenses was compared with the
period from 03/17/2019 to 06/22/2019.
Regarding social distancing, two ambiguous effects should be considered.
On the one hand, it is also the result of a possible effect of expectations associated
with pandemic-related fear, i.e., as much as the focus is not on the number of
infected and deaths due to Covid-19, they also have an influence on the volume of
expenses analyzed. On the other hand, as this pandemic and the isolation itself
continue, population adherence no longer occurs rigidly, either because of
economic or other issues, including tolerance to the isolation itself. This, in fact,
reinforces the limit of the analysis used here, as this is not a causality analysis, only
a correlation analysis.
As for the results, as mentioned in the previous section and in line with the
conclusions of the international literature already developed on the topic, it is
expected that essential goods will not have their consumption reduced, even under
strong isolation measures - that is, the expected correlation for these types of goods
should be weak or even positive, except for those goods that require displacement
of residence. Assets that are both superfluous and require a commute should show
strong drops in the volume of resources - therefore, stronger levels of negative
correlation with the distance index. Other goods with intermediate situations, both
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 323
from the point of view of essentiality and travel restriction can fall in intermediate
situations.
4. Results
Considering the information until June 20, 2020, there are two major
"corridors" of infection in Brazil: in the North, Northeast and Southeast regions of
the country and locations with a relatively smaller number of cases in the interior
and south of the country (Figure 1)2.
Figure 1 – Covid-19 cases and deaths (per 100 thousand people) and deaths
and cases ratio
Amapá, Amazonas, Roraima and Acre are the four states with the highest
number of cases per 100 thousand people. The state of Amazonas has the highest
number of deaths per 100,000 people in the country (63.9), followed by Ceará and
Pará. The ratio between the number of deaths and those contaminated is higher in
the state of Rio de Janeiro and Pernambuco. Regarding household spending, the
analysis is separated into three periods and represented in Figures 2 to 12. Table 4
shows the representation of the periods used in the analysis.
2
The three-corridor pattern does seem to, in general, replicate in some the spending patterns as well, that
is, one in the North/Northeast of the country, one in the Southeast and a third in the interior.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 324
Table 4 – Representation of the periods analyzed in the figures
Period Weeks Figures 1 to 12
01/05/2020 to 1 to 24 Map “a” e table “i”
06/20/2020
02/23/2020 to 8 to 24 Map “b” e table “ii”
06/20/2020
03/15/2020 to 11 to 24 Map “c” e table “iiii”
06/20/2020
Figure 2 – Correlation between the total spending variation and IDS
In terms of time, there is an apparent weakening of the effect of the isolation
measures, something that may be associated with the loss of effectiveness of the
measures, as they extend over long periods of time. The first correlation (in this
case, 2a and 2i), covers the first few weeks of the year, when there are no cases of
Covid-19 in Brazil, nor any legal social distancing measures. In contrast, in the
third correlation (2c and 2iii), the series starts in the eleventh week, when social
distancing begins. Thus, the series is more static, and the correlation tends to be
smoothed out.
The first sectors to be analyzed will be those most impacted by legal
measures of social isolation, that is, bars and restaurants, gas stations and tourism
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 325
and transportation. These sectors are linked to a degree of mobility, which, in a
pandemic moment, explains the strong reduction in spending they had, as shown
in graphs 1, 2 and 3.
Graph 1 – Consumer spending behavior in the Bars and Restaurants sector
during the first 24 weeks of 2020
Graph 2 – Consumer spending behavior in the Gas Station sector during the
first 24 weeks of 2020
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 326
Graph 3 – Consumer spending behavior in the Tourism and Transport
sector during the first 24 weeks of 2020
The three sectors show similar behavior, with a sharp drop in consumer
spending as of the eleventh week, when the first social isolation measures were
enacted without a recovery since then. Figures 3, 4 and 5 show the correlation
between the variation in spending in the bars and restaurants, gas stations and
tourism and transport sectors, respectively:
Figure 3 – Correlation between the total spending on bars and restaurants
and IDS
Spending in bars and restaurants proved to be highly correlated with IDS.
In most large cities, decrees were issued prohibiting the opening of these
establishments, or restricting operation to the delivery format, reinforcing the
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 327
importance of the mobility issue for the impact of the isolation measures on
consumption. In the week following the first measures to close these
establishments, the average reduction in spending in Brazil was more than 70%.
In April and May, the drop in consumption remained around 60% and in June,
53%. Figure 3c shows a change in behavior, especially in some states of the
Northeast and North, as well as in Santa Catarina, in line with the behavior
observed in figure 1, which shows a timid recovery in spending in the sector at the
end of the period considered, related to a process of easing of social isolation
measures.
Figure 4 – Correlation between the variation of expenses at gas stations and
IDS
As for gas stations, it can be said that the legal measures of social distance
had a more intense effect, since the correlation was high in all regions in the first
two periods, showing a certain reduction only in the third period and in the states
of São Paulo. Goiás, Maranhão, Roraima, Santa Catarina and Paraíba. This shows
that the essentiality associated with the mobility issue enhances the negative
effects on consumption.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 328
Figure 5 – Correlation between the variation in spending on tourism and
transport and the IDS
A similar case is the tourism and transportation sector. Both the first and the
second periods analyzed, the negative correlations are greater than 0.7 in all
locations. With the recommendation of social isolation, spending on tourism,
including business, hotels, airline tickets, and car rentals, became lower. The
reduction in spending in this category was more than 85% from the 12th week of
the year.
The furniture and department stores; construction materials; clothing and
sporting goods; other durable goods sectors; and other service sectors are also hard
hit after the social isolation measures begin to be experienced. However, in several
locations such measures have been attenuated over time in various ways, and
therefore, their effects are gradually mitigated after the sharp drop around the
twelfth week. Graphs 4, 5, 6, 7 and 8 show the variation in consumer spending that
occurred in the furniture and department stores; construction materials; clothing
and sporting goods; other durable goods sectors; and other sectors in services.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 329
Graph 4 – Consumer spending behavior in the Furniture and Department
Stores sector during the first 24 weeks of 2020
Graph 5 – Consumer spending behavior in the Construction Material sector
during the first 24 weeks of 2020
Graph 6 – Consumer spending behavior in the Clothing and Sporting Goods
sector during the first 24 weeks of 2020
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 330
Graph 7 – Consumer spending behavior in the sector of other sectors of
durable goods during the first 24 weeks of 2020
Graph 8 – Consumer spending behavior in the other Sectors in Services
sector during the first 24 weeks of 2020
Figure 6 shows the correlation between the variation in spending in the
furniture and department store sector and the IDS.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 331
Figure 6 – Correlation between the variation in spending on furniture and
department stores and the IDS
Figure 6 shows a negative correlation in the first period, but which eased
more quickly when analyzing the correlation between spending on furniture
purchases and in department stores and IDS. Anyway, it is clear that a negative
correlation equal to or greater than 0.30 (6a and 6i) prevailed. Figure 7 shows the
correlation between the variation in spending in the construction materials sector
and the IDS.
Figure 7 – Correlation between the variation in spending on construction
materials and the IDS
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 332
Like supermarkets, pharmacies, and gas stations, building material stores
are part of the segment that was considered an essential service and remained open
even after the trade decrees were closed. Although there was a drop in consumer
spending immediately after the enactment of the social isolation measures,
compared to the other categories, the building materials segment is among those
that showed the least retraction, indicating that, from the consumer's point of view,
its essentiality was different, not always of an urgent nature. Once pandemic and
isolation measures were in place, the impact on this sector was greater, possibly
due to uncertainty about the duration of the crisis.
In the eighth week and in the five weeks following the implementation of
the first social distance measures (12th to 16th), there was a reduction in the
consumption of these items, a standard behavior among the federation units. Once
it was realized that it would be a long period, the consumer's view of this sector
changed, and the data suggest that work and reforms gradually resumed. Figure 8
shows the correlation between the variation in spending in the clothing and sports
equipment sector and the IDS.
Figure 8 – Correlation between the variation in spending on clothing and
sports equipment and the IDS
Regarding the expenses with clothing and sports equipment related to the
IDS, they show a negative correlation greater than 0.5 in all states and in the DF,
in the period covering the 24 weeks (8th and 8th). In the period between the 8th
and 24th week, Mato Grosso do Sul and Paraná, showed a different behavior (8b
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 333
and 8ii), and between 11th and 24th week there was a reduction in the correlation
in most states (8c and 8iii). Figure 9 shows the correlation between the variation
in spending in the other durable goods sectors and the IDS.
Figure 9 – Correlation between variation in spending in other durable goods
sectors and IDS
Even if not uniform, Figure 9 shows a significant negative correlation
between social distance and spending on other durable goods. In Distrito Federal,
Santa Catarina, Alagoas and Minas Gerais, the correlation remained negative and
high in the three periods analyzed. In other locations, there was variation over the
periods, despite the data showing that there was a retraction in sales of these
categories from the 11th week on, in all states. As of the 16th week, Mato Grosso
do Sul, Roraima and Tocantins show predominantly positive variations in the
expenditure with this type of asset. Moreover, in these states, in the three periods,
the correlations with the IDS are positive and/or weaker (Figure 9i, 9ii and 9iii).
Figure 10 shows the correlation between the variation in spending in the other
service sectors and the IDS.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 334
Figure 10 – Correlation between the variation in spending in other service
sectors and the IDS
The services that were not included in the above sectors were aggregated as
other sectors and showed a relatively different behavior from those previously
addressed with respect to the correlation with the social isolation measures, as
shown in Figure 10. The lack of uniformity can also be attributed to the variety
activities included in this category, such as essential services like hospitals, clinics
and transportation, and others that are non-essential and difficult to maintain even
partially with social distance, such as the beauty and aesthetics segment.
Unlike the sectors associated with mobility issues, it is observed that these
intermediate sectors, besides enjoying a recovery in consumer spending over the
period, also register a more heterogeneous behavior regarding the correlation with
social isolation measures - although with negative values and tending to decrease
when considering only the narrower period when the measures were already
underway, but with large differences between the states .
Finally, the sectors related to essential goods and not associated with
mobility issues, that is, those related to food (supermarkets) and medicines
(drugstores). Graphs 9 and 10 show the variation in consumer spending for these
two sectors over the period considered.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 335
Graph 9 – Consumer spending behavior in the Supermarket sector during
the first 24 weeks of 2020
Graph 10: Behavior of consumer spending in the Drugstores and
Pharmacies sector during the first 24 weeks of 2020:
In essential sectors, differently from others, there is an increase in
consumption spending in line with the stylized facts determined by the literature,
around the eleventh week. This may be associated with a stockpiling behavior of
food and medicine. After this period, the behavior across sectors is different. The
medication sector has seen a strong reduction in spending, returning to its level
prior to social isolation measures, probably due to the population's recognition that
Covid-19's treatment was more complex than simply going to the pharmacy.
Spending in the supermarket sector, however, are quite oscillating in the
period, suggesting that isolation may have distanced trips to these establishments,
as people sought to avoid contamination. When the correlation of spending in
sectors considered essential with the IDS was evaluated, according to the forecast,
goods with a higher degree of essentiality, such as food (supermarket segment) and
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 336
medicines (drugstore segment) showed weaker correlations. These results are
shown in Figures 11 and 12.
Figure 11 – Correlation between the variation in spending in supermarkets
and the IDS
Figure 11, unlike what happened in the previous sectors, shows positive
correlations, consistent with this increase in expenses related to a possible
stockpiling effect previously described here and in the literature. Thus, the
smoothing process that occurred in the previous sectors here implies a change in
the correlation, that is, a negative correlation in spending on supermarkets / IDS,
possibly associated with the relaxation of social isolation measures. Figure 12
shows the correlation between the variation in spending in the drugstore and
pharmacy sector and the IDS.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 337
Figure 12 – Correlation between variation in spending on
drugstores/pharmacies and IDS
Figures 12a and 12b, particularly in the North, point to a positive correlation
in spending on medicines/IDS. The correlation becomes negative between
spending on drugstores and the IDS over the weeks, with emphasis on the third
period of the analysis (12c and 12iii), in which 16 states have a correlation greater
than -0.5. Most of the states that exhibited this behavior were those that took a little
longer to register a greater number of infected people, located in the Midwest and
South of the country (Figure 12c).
Summarizing the analysis, it is observed that, as noted in the literature, the
degree of essentiality has been an important factor in maintaining “normal” levels
(at least on average terms) of consumption expenditures during the 2020
coronavirus pandemic.
However, issues such as mobility associated with the consumption sector
and how social isolation measures affect the sector are also important, such that
three primary types of behavior were observed: (i) sectors that experienced an
abrupt negative shock and did not recover, associated with mobility issues; (ii)
sectors that experienced a negative shock and recovered gradually, generally of
non-essential goods and more sensitive to social isolation measures; (iii) finally,
sectors that experienced a sudden positive shock and have since had a more erratic
behavior, although they did not actually show such a considerable drop in the level
of consumer spending. These are goods characterized by their essentiality .
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 338
It is also observed that the heterogeneity of social isolation measures (both
in the form in which they were adopted, the timing of adoption, and the
population's adherence to these measures) differs between states, which possibly
explains the different correlations observed in this study.
5. Final considerations
This paper aimed to correlate the social isolation measures adopted as a
result of the coronavirus pandemic with variations in consumption spending in the
various Brazilian UFs. Following stylized facts from an incipient literature, there
was a greater inertia in consumption expenditure in essential goods.
However, the article contributes to the literature by observing that this
inertia is more complex, at least in the Brazilian case, than is suggested at the
outset. Issues such as consumer mobility during the pandemic and its quarantine
and how social isolation measures affect each consumer sector are relevant.
Thus, three types of primary behavior were classified among consumer
expenditures in the various sectors considered: i) spending in sectors that were
sharply reduced and did not recover, associated with consumer mobility problems
during quarantine; ii) expenditures in sectors that declined sharply and that
gradually increased, generally, on non-essential goods and more sensitive to social
isolation measures; iii) finally, spending in sectors that increased when social
isolation measures were enacted, probably associated with a stock effect, and since
then have been more erratic, although they did not, in fact, fall during the period,
even as they were considered essential goods.
As expected, the first correlations present the full shock of the social
distance measures and family expenses, that is, the beginning of the series
contemplates weeks when there were no impacts of the pandemic. The second and
third correlations, which started in the eighth and 11th week, are calculated from
more static series, that is, with negative variations in consumption and with non-
zero IDS. Thus, as expected, the correlation tends to be less negative or smooth.
In addition, heterogeneity was also observed in how the social isolation
measures were adopted may have had different effects on the different measures
in each UF. The moment of adoption and flexibility and the population's adherence
to these measures also seems to be relevant. As a future study, it is suggested to
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 339
analyze the influence of institutional differences in the form and timing of the
measures on their compliance and their economic and demographic consequences.
Furthermore, it should be noted that the loosening of legal social distance
measures are not necessarily related to controlling the pandemic and reducing the
fear of infection or the collapse of the health care system. The weakening of the
severity of isolation is also a political and economic decision, both in terms of
public finances and personal / family finances.
As mentioned before, according to information from the Central Bank of
Brazil (BCB) and the IBGE in 2019, 89.3% of the Brazilian population has a
banking relationship with a financial institution, and while credit/debit card data
may be income biased, we believe it sheds light on consumer behavior during the
isolation period for the 2020 covid pandemic.
References
ACEMOGLU, D.; CHERNOZHUKOV, V.; WERNING, I.; WHINSTON, M.
Optimal Targeted Lockdowns in a Multi-Group Sir Model. NBER Working Paper,
No. w27102, 2020.
ALEXANDER, D.; KARGER, E. Do stay at-home orders cause people to stay at
home? Effects of stay-at-home orders on consumer behaviour. 2020. Disponível
em: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3583625.
BCB, Sistema Gerenciador de Séries Temporais, 2020. Disponível em:
https://www3.bcb.gov.br/sgspub/localizarseries/localizarSeries.domethod=prepar
arTelaLocalizarSeries. Acesso em: 01 de Novembro de 2020.
BALDWIN, R.; MAURO, B. Economics in the time of COVID-19. CEPR Press,
2020.
BINDER, C. Coronavirus Fears and Macroeconomic Expectations. Review of
Economics and Statistics, v. 102, n. 4, p. 721–730, 2020.
BJORNSKOV, C. Did Lockdown Work? An Economist’s Cross-Country
Comparison, 2020. Disponível em: SSRN: https://ssrn.com/abstract=3665588 or
http://dx.doi.org/10.2139/ssrn.3665588
BRETAS, V.; ALON, I. The impact of COVID‐19 on franchising in emerging
markets: An example from Brazil. Global Business and Organization Excellence,
vol. 39, no. 6, p. 6–16, 2020.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 340
CARVALHO, V.; HANSEN, S.; ORTIZ, A.; GARCIA, J.; TOMASA, R.; MORA,
S.; AGUIRRE, P. Tracking the COVID-19 Crisis with High-Resolution
Transaction Data. INET Working Paper Series, Number 2009, 2020.
CEREDA, F.; RUBIAO, R.; SOUSA, L. COVID-19, Labor Market Shocks,
Poverty in Brazil: A Microsimulation Analysis. World Bank Background note,
2020.
CHEN, H.; QIAN, W.; WEN, Q. The Impact of the COVID-19 Pandemic on
Consumption: Learning from High Frequency Transaction Data. SSRN Papers,
2020.
CHRONOPOULOS, D. K.; LUKAS, M.; WILSON, J. O. S. Consumer Spending
Responses to the COVID-19 Pandemic: An Assessment of Great Britain. The RBF
Working Papers, Number 20-012, 2020.
CROKIDAKIS, N. COVID-19 Spreading in Rio de Janeiro, Brazil: Do the policies
of social isolation really work? Chaos Solitons Fractals, v. 136, 2020.
DIETRICH, A. M.; KUESTER, K.; MULLER, G. J.; SCHOENLE, R. S. News
and Uncertainty about COVID-19: Survey Evidence and Short-run Economic
Impact. Working Papers of the Federal Reserve of Cleveland, Number 20-12,
2020.
DWECK, E.; ROCHA, C.; FREITAS, F.; FERRAZ, J.; TORRACCA, J.;
FERREIRA, K.; VILAR, M.; BASSETTI, M.; CASTILHO, M.; MIGUEZ, T.
Impactos Macroeconômicos e Setoriais da Covid-19 no Brasil: Resultados
Selecionados. Grupo de Indústria e Competitividade, IE-UFRJ, Abril, 2020.
FERNANDES, N. Economics effects of coronavirus (COVID-19) on the world
economy. SSRN Paper, 2020.
IBGE, Pesquisa Nacional por Amostra de Domicílios Contínua - PNAD Contínua,
2020. Disponível em:
https://www.ibge.gov.br/estatisticas/sociais/habitacao/9171-pesquisa-nacional-
por-amostra-de-domicilios-continua-mensal.html?edicao=28095&t=destaques.
Acesso em 12 de julho de 2020.
IPEA, Instituto de Pesquisa Econômica Aplicada. Medidas legais de incentivo ao
distanciamento social: comparação das políticas de governos estaduais e
prefeituras das capitais do Brasil. Nota Técnica, n.16, 2020.
JOYCE, R.; XU, X. Sector Shutdows during the Coronavirus Crisis: Which
Workers are most Exposed?. IFS Briefing Note, Number BN278, 2020.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
Cabello, Brandão e Bourscheidt 341
MCKIBBIN, W.; FERNANDO, R. The economic impact of COVID-19. In:
BALDWIN, R.; MAURO, B. (Orgs.) Economics in the Time of COVID-19.
Londres: CEPR Press, 2020.
MINISTÉRIO DA SAÚDE. Disponível em: http://covid.saude.gov.br/. Acesso em
12 de julho de 2020.
MORATO, M.; BASTOS, S.; CAJUEIRO, D.; NORMEY-RICO, J. An Optimal
Predictive Control Strategy for COVID-19 (SARS-CoV-2) Social Distancing
Policies in Brazil. arXiv preprint, 2020.
OZILI, P.; ARUN, T. Spillover of COVID-19: Impact on the Global Economy.
SSEN Papers, 2020.
TELLIS, G.; SOOD, A.; SOOD, N. How Long Should Social Distancing Last?
Predicting Time to Moderation, Control and Containment of Covid-19. USC
Marshall School of Business Research Paper, 2020.
WORLD BANK. COVID 19 in Brazil: Impacts and Policy Responses. 2020.
Revista de Economia v. 43 | n. 81 | p. 313-341 | 2022
File and source
- File
- cabello-brandao-bourscheidt-2022-revista-de-economia-43-81-consumer-behavior-pandemic.pdf
- Size
- 1,549,473 bytes
- SHA-256
- 3444c5fcfef2b675d0634754ad1e301fb93f22972c3f431fa9535daad76ae402
- Original
- web.archive.org