## wpiea2022019-print-pdf

## Source details

**Canonical URL:** [wpiea2022019-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022019-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022019-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022019-print-pdf.pdf.json)

---

### Key questions and scope
- How much did the digitalization of consumption increase during the COVID-19 pandemic?
- Did economies with low e-commerce penetration catch up, or did the digital divide widen?
- Was the increase transient or persistent (levels or trends)?
- Which factors explain differences across economies and sectors?
- Sample: 47 economies and 26 sectors, January 2018 to September 2021, using Mastercard universe transaction data. Main variable: share of transactions online.

### High-level findings (summary)
- Economies with a higher pre-COVID share of online transactions accelerated relative use of e-commerce, exacerbating the digital divide across economies.
- Aggregate spikes in online spending shares are gradually dissipating at the aggregate level.
- The increase in online spending shares is more persistent in the retail sector and restaurants, suggesting e-commerce learning and longer-lasting impacts of social mobility restrictions in those sectors.

### Data & Methodology
- Primary dataset: universe of all credit and debit card transactions cleared through the Mastercard network between January 2018 and September 2021; sample limited to 47 economies where Mastercard has a significant market share.
- Data fields observed: Date & time; Type of card; Merchant location; Industry classification; Channel (online/offline); Transaction amount.
- Main variable: s_{c,t} = Total online spending / Total consumption.
- “Online” spending defined as transactions where neither the cardholder nor the card are physically present (internet, payment by phone, mail-order, etc.).
- Estimation approach: scale Mastercard online share s_{c,t}^{MC} by card-share of consumption to estimate universe-level online shares (assumes representativeness of Mastercard online shares and that online transactions occur primarily through cards).
- Pre-COVID trends estimated via simple regression of monthly online shares on a time trend between 2018–2019.
- Baseline analysis uses raw series without seasonal adjustment (results similar after seasonal adjustment).

### Key data details and assumptions
- Mastercard “switched” universe captures authorizations when payment initiated.
- Total card spending from annual estimates from Mastercard.
- Total consumption uses annual final private consumption expenditures from national statistics, Eurostat, and the IMF.
- Nilson Report issue 1199 (June 2021) global shares cited: Visa 40%, UnionPay 32%, Mastercard 24%, Rest 4%.
- Sample selection: economies where Mastercard has a significant market share; robustness: changing cutoff has little impact; correlation coefficient between high- and low-market-share groups is 0.97.

### Stylized facts — Fact 1: Significant heterogeneity in deviations from economy-wide trends
- The share of online spending differs markedly across economies in levels and trends.
- Reporting approach: levels at three points (2019 average, crisis peak with year/month, latest) and deviations relative to pre-COVID trends (2018–2019).
- Aggregate summary statistics (all data in percentages; differences in percentage points):
  - Online share rose from 10.3 in 2019 to 14.9 at the peak, then fell to 12.2 in 2021.
  - Latest level is 0.6 percentage points above the average level predicted by pre-pandemic trends.
  - Mean 2019 average: 10.3
  - Mean crisis peak: 14.9
  - Mean latest: 12.2
  - Mean Latest minus pre-COVID trend: 0.6
  - Median 2019 average: 5.7
  - Median crisis peak: 9.9
  - Median latest: 6.8
  - Median Latest minus pre-COVID trend: 0.1
  - Standard Deviation 2019 average: 9.6
  - Standard Deviation crisis peak: 13.2
  - Standard Deviation latest: 10.9
  - Standard Deviation Latest minus pre-COVID trend: 2.0
- Top economies by Latest minus pre-COVID trend (first five rows):
  - Bahrain: 2019 average 21.2; Crisis peak 38.7 (2021/05); Latest 34.1; Latest minus pre-COVID trend 10.0
  - Jamaica: 2019 average 33.1; Crisis peak 45.8 (2021/09); Latest 43.1; Latest minus pre-COVID trend 8.3
  - New Zealand: 2019 average 28.3; Crisis peak 39.2 (2020/04); Latest 29.0; Latest minus pre-COVID trend 5.1
  - Norway: 2019 average 18.2; Crisis peak 23.4 (2021/09); Latest 19.7; Latest minus pre-COVID trend 4.2
  - United Kingdom: 2019 average 28.4; Crisis peak 39.8 (2021/01); Latest 31.3; Latest minus pre-COVID trend 3.7
- Distributional note: About half of the economies still have positive deviations relative to pre-pandemic trends; others (including the United States and many developed economies) are at or below predicted pre-COVID trend levels.
- Example dynamics:
  - United States (retail): online share rose from 13.9 to 17.7 at peak, then fell to reach the 15 level predicted by the pre-COVID trend by May 2021.
  - Brazil (retail): peaked in early 2021 and remained significantly above the level predicted by the pre-COVID trend after more than 18 months of the pandemic.

### Stylized facts — Fact 2: Digital divergence across economies increased during COVID-19
- COVID-19 led to divergence in e-commerce adoption across economies: economies with higher pre-COVID online shares experienced the largest increases at the peak.
- Estimated relationship at peak: a one percentage point increase in pre-COVID online shares is associated with an additional increase of 0.33 percentage points during the peak.
- Peak global (weighted) average difference between online shares and predicted pre-crisis trends: 4.3 pp higher.
- By September 2021 this global difference dropped to 0.3 pp.
- On average, less than 7 percent of the deviation during the peak persists in the latest data.
- Slope of difference between latest and pre-COVID trends against pre-COVID trend shares: 0.07 (compared to 0.33 at peak).
- Conclusion: digital divergence increased during the crisis but largely dissipated by September 2021; economies with larger pre-existing online shares saw the biggest temporary increases.

### Correlates of deviations in online spending shares (regression insights)
- Strong correlates:
  - Residential mobility (Google index) strongly correlated with online share deviations, especially in Q2 2020; correlation declined as the pandemic continued.
  - A one percentage point higher residential mobility is associated with a 0.09 pp higher online share gap (Table 2 summary).
  - A one pp higher pre-COVID online share is associated with a 0.08 pp higher gap.
  - A one pp higher fiscal spending during COVID (as % 2019 GDP) is associated with a 0.02 pp higher online gap.
- Interaction effects:
  - Fiscal support interacted with severity of restrictions: interaction term positive, indicating greater fiscal support associated with higher online shares when restrictions were more severe.
  - Computation example: Average mobility in the sample is 7.5%. Average effect of fiscal spending on e-commerce shares (based on Column (3) coefficients) computed as -0.06+7.5*0.01=0.015.
- Other covariates:
  - Richer economies returned faster to trend once pandemic receded (negative coefficient on per capita income).
  - Tourism: tourism as % of GDP has a positive coefficient; interaction of tourism with mobility is negative. Evaluated at average mobility, more tourism-dependent economies are associated with lower online shares.
- Summary statistics for regression variables (Table A1):
  - Online share gap (online share-trend) (in pp), NSA: Obs 899; Mean 0.9; Std. Dev. 2.4; Min -7.0; Max 15.0
  - Residential Mobility, %: Obs 899; Mean 7.5; Std. Dev. 7.3; Min -8.8; Max 40.3
  - Pre-covid trend online share: Obs 899; Mean 12.0; Std. Dev. 10.2; Min 1.2; Max 34.8
  - COVID Fiscal spending as % 2019 GDP: Obs 899; Mean 12.9; Std. Dev. 9.9; Min 0.6; Max 43.4
  - GDP per capita-2019, in '000: Obs 899; Mean 35.6; Std. Dev. 23.7; Min 2.8; Max 115.6
  - Residential mobility * Fiscal spending, 2019: Obs 899; Mean 96.4; Std. Dev. 132.4; Min -149.8; Max 1228.0

### Sectoral patterns and heterogeneity
- Restaurants-bars experienced largest peak increase and notable persistence:
  - Restaurants-Bars: 2019 average 4; Peak crisis 18; Latest 9; Latest vs pre-COVID trend 2.2
- Aggregate sectoral statistics (Table 3):
  - All Categories: 2019 average 18; Peak crisis 23; Latest 21; Latest vs pre-COVID trend 0.3
  - Retail: 2019 average 11; Peak crisis 17; Latest 14; Latest vs pre-COVID trend 0.7
  - Services: 2019 average 34; Peak crisis 41; Latest 36; Latest vs pre-COVID trend -1.0
- Top sectors by Latest vs pre-COVID trend (Table 4):
  1. Department Stores — 2019 average 14; Peak 36; Latest 26; Latest vs pre-COVID trend 6.8
  2. Electric-Appliance — 23; 39; 29; 3.2
  3. Clothing Stores — 12; 38; 16; 3.0
  4. Discount Stores — 8; 17; 13; 2.3
  5. Restaurants-Bars — 4; 18; 9; 2.2
- Observations:
  - Sectors with already high pre-pandemic e-commerce shares (mail order, airline, travel agencies, utilities) experienced little change or convergence to an upper limit.
  - Industries with lower or mid pre-pandemic online shares saw larger peak increases (clothing, electrical appliances, sporting goods-toys).
  - Some sectors show declines below trend in latest data: Professional Services (-1.4), Utilities (-1.5), Recreation (-1.6).
- Seasonality note: evaluating seasonally adjusted series for retail, close to 25% of the peak shift online appears to be persisting by comparing current deviation from trend vs peak deviation from trend.

### Representativeness and limitations of Mastercard data
- Representativeness assumption: Mastercard online spending is representative of all card transactions; other online payment methods negligible or similar in pattern.
- Validation: comparison with national statistics for the United States and the United Kingdom shows similar dynamic patterns:
  - US official data suggests online spending share returned to pre-COVID trend levels.
  - UK remains slightly above the trend-predicted levels.
- Data limitations:
  - Not all segments captured (e.g., vehicle sales often not card-based).
  - Transaction-level fields not observed: cardholder demographics, item-level/SKU details.
  - Estimation assumes MC’s online share representative across cards and industries.

### Conclusions and interpretation
- The share of online spending rose more in economies where e-commerce was already large; increases are reversing as the pandemic recedes.
- Sector-level evidence shows more persistent increases in retail, restaurants, and healthcare — sectors affected by limited social mobility and with lower pre-pandemic digitalization allowing persistent learning.
- Persistence of learning concentrated in sectors with lower pre-pandemic e-commerce shares; not broad-based across all sectors.
- Mobility plays a crucial role: economies with faster recoveries and higher income per capita returned closer to pre-pandemic trends.
- Open questions for future research: whether post-pandemic increases are driven by the extensive margin (new online customers) or the intensive margin (existing customers increasing online purchases), and long-term effects given varying travel and gathering restrictions across economies.

*Source: wpiea2022019-print-pdf*

### 1. Introduction

### 1. Introduction

### Key questions and scope
- How much did the digitalization of consumption increase during the COVID-19 pandemic?
- Did economies with low e-commerce penetration catch up, or did the digital divide widen?
- Was the increase transient or persistent (levels or trends)?
- Which factors explain differences across economies and sectors?
- Sample: 47 economies and 26 sectors, January 2018 to September 2021, using Mastercard universe transaction data. Main variable: share of transactions online.

### High-level findings (summary)
- Economies with a higher pre-COVID share of online transactions accelerated relative use of e-commerce, exacerbating the digital divide across economies.
- Aggregate spikes in online spending shares are gradually dissipating at the aggregate level.
- The increase in online spending shares is more persistent in the retail sector and restaurants, suggesting e-commerce learning and longer-lasting impacts of social mobility restrictions in those sectors.

### Paper organization
- Section 2: Related literature.
- Section 3: Data and assumptions.
- Section 4: Stylized facts.
- Section 5: Conclusions and tentative explanations.

### Related literature (high-level links)
- Consumer benefits from the internet: Goolsbee and Klenow, 2006; Brynjolfsson and Oh, 2012; Varian, 2013.
- Dolfen et al. (2020): Visa U.S. data 2007–2017, e-commerce spending 8% of consumption by 2017; equivalent of 1% permanent boost to consumption.
- Pricing and inflation: Cavallo (2018); Jo, Matsumura and Weinstein (2019); Goolsbee and Klenow (2018); Reinsdorf and Schreyer (2020) (~0.5 pp per year overstatement for OECD).
- Use of private data for real-time economic measurement: Cavallo (2013); Chetty et al. (2020); Carvalho et al. (2020); Aladangady et al. (2019).

### Data & Methodology (summary)
- Primary dataset: universe of all credit and debit card transactions cleared through the Mastercard network between January 2018 and September 2021, covering more than 200 economies and territories; sample limited to 47 economies where Mastercard has a significant market share.
- Data include total dollar amount per transaction and sectoral classification spanning 26 sectors and over 100 industry subcategories; daily frequency.
- “Online” spending: transaction where neither the cardholder nor the card are physically present (internet, payment by phone, mail-order, etc.). All other transactions classified as “offline”.
- Computing online spending share s_{c,t}:
  - Defined as Total online spending / Total consumption.
  - Under assumptions (cash offline; other online payments negligible or ultimately processed through cards), s_{c,t} is computed using Mastercard online share s_{c,t}^{MC} and annual estimates of total card spending and total consumption.
- Baseline analysis uses raw series without seasonal adjustment (results similar after seasonal adjustment).

### Key data details and assumptions
- Mastercard “switched” universe: transactions that authorize merchant and transaction when payment initiated.
- Total card spending used from annual estimates from Mastercard.
- Total consumption uses annual final private consumption expenditures from national statistics, Eurostat, and the IMF.
- Pre-COVID trends estimated via simple regression of monthly online shares on a time trend between 2018–2019.
- Note: Nilson Report issue 1199 (June 2021) global shares: Visa 40%, UnionPay 32%, Mastercard 24%, Rest 4% (cited in Appendix).

### Stylized facts: heterogeneity, divergence, and transiency

- The paper presents five stylized facts on online shares across economies and sectors. The content provided covers Facts 1 and 2.

H3: Fact 1 — Significant heterogeneity in deviations from economy-wide trends
- The share of online spending differs markedly across economies in levels and trends.
- Method: report levels at three points (2019 average, crisis peak with year/month, latest) and deviations relative to pre-COVID trends (pre-COVID trend estimated 2018–2019).
- Examples:
  - United States (retail): online share rose from 13.9% to 17.7% at peak, rose again toward end of 2020, then fell to reach the 15% level predicted by the pre-COVID trend by May 2021.
  - Brazil (retail): initially similar pattern but peaked in early 2021 and remained significantly above the level predicted by the pre-COVID trend after more than 18 months of the pandemic.
- Table 1 summary statistics (all data in percentages; differences in percentage points):
  - Online share rose from 10.3% in 2019 to 14.9% at the peak, then fell to 12.2% in 2021.
  - Latest level is 0.6 percentage points above the average level predicted by pre-pandemic trends.
  - Mean 2019 average: 10.3
  - Mean crisis peak: 14.9
  - Mean latest: 12.2
  - Mean Latest minus pre-COVID trend: 0.6
  - Median 2019 average: 5.7
  - Median crisis peak: 9.9
  - Median latest: 6.8
  - Median Latest minus pre-COVID trend: 0.1
  - Standard Deviation 2019 average: 9.6
  - Standard Deviation crisis peak: 13.2
  - Standard Deviation latest: 10.9
  - Standard Deviation Latest minus pre-COVID trend: 2.0
- Top economies by Latest minus pre-COVID trend (first five rows of Table 1):
  - Bahrain: 2019 average 21.2; Crisis peak 38.7 (2021/05); Latest 34.1; Latest minus pre-COVID trend 10.0
  - Jamaica: 2019 average 33.1; Crisis peak 45.8 (2021/09); Latest 43.1; Latest minus pre-COVID trend 8.3
  - New Zealand: 2019 average 28.3; Crisis peak 39.2 (2020/04); Latest 29.0; Latest minus pre-COVID trend 5.1
  - Norway: 2019 average 18.2; Crisis peak 23.4 (2021/09); Latest 19.7; Latest minus pre-COVID trend 4.2
  - United Kingdom: 2019 average 28.4; Crisis peak 39.8 (2021/01); Latest 31.3; Latest minus pre-COVID trend 3.7
- Distributional note: About half of the economies still have positive deviations relative to pre-pandemic trends; others (including the United States and many developed economies) are at or below predicted pre-COVID trend levels.

H3: Fact 2 — Digital divergence across economies increased during COVID-19
- The COVID-19 crisis led to divergence in e-commerce adoption across economies.
- Peak values relative to contemporaneous trend predictions show almost all economies at or above their pre-COVID trend peaks.
- The fitted regression line is positively sloped: economies with higher pre-COVID online shares experienced the biggest increases during the peak.
- A one percentage point increase in pre-COVID online shares (as predicted by the trends) is associated with an additional increase of [text truncated in source].

*Source: wpiea2022019-print-pdf - 1. Introduction*

### 0.33 percentage points during the peak. Figure 2b presents the same information in a different format, by

### wpiea2022019-print-pdf - 0.33 percentage points during the peak. Figure 2b presents the same information in a different format, by

### Key findings on digital divergence and peak effects
- COVID-19 exacerbated cross-economy digital inequality — termed “digital divergence”.
- Economies with higher pre-COVID shares experienced acceleration in e-commerce shares (positive slope in fitted line comparing peak vs pre-COVID trend).
- At the peak, the global (weighted) average difference between online shares and predicted pre-crisis trends was 4.3 pp higher.
- By September 2021 this global difference dropped to 0.3 pp.
- At a global level, on average, less than 7 percent of the deviation during the peak persists in the latest data.
- The slope of the difference between latest and pre-COVID trends against pre-COVID trend shares is 0.07 (compared to 0.33 at peak).

### Persistence and aggregate dynamics
- Much of the increase in online shares at an economy level appears transitory.
- The average global deviation is calculated by: i) devations of online shares from trend in each economy, ii) weighting each economy-wide deviation by the share of Mastercard spending of that economy over total world spending, iii) multiplying each economy-wide deviation by the weight using the Q3 2021 weight, and iv) computing the weighted sum defined as the average global deviation.
- The digital divergence appears to be dissipating: fitted regression slope for latest data is positive but has declined significantly compared with peak shares and is much closer to the 45-degree line.

### Correlates of deviations in online spending shares
- Residential mobility (Google’s index of residential mobility) strongly correlated with online share deviations, especially in Q2 2020; correlation declined as the pandemic continued.
- Regression results (Table 2) summary:
  - A one percentage point (pp) higher residential mobility is associated with a 0.09 pp higher online share gap.
  - A one pp higher pre-COVID online share is associated with a 0.08 pp higher gap.
  - A one pp higher fiscal spending during COVID (as % 2019 GDP) is associated with a 0.02 pp higher online gap.
- Interaction effects:
  - Fiscal support interacted with severity of restrictions: interaction term positive, indicating greater fiscal support associated with higher online shares when restrictions were more severe.
  - Average mobility in the sample is 7.5%. Average effect of fiscal spending on e-commerce shares (based on Column (3) coefficients) computed as -0.06+7.5*0.01=0.015, similar to average effect in Column [2].
- Other covariates:
  - Richer economies returned faster to trend once pandemic receded (negative coefficient on per capita income).
  - Tourism: tourism as % of GDP has a positive coefficient; interaction of tourism with mobility is negative. Evaluated at average mobility, more tourism-dependent economies are associated with lower online shares.

### Sectoral patterns and heterogeneity
- Restaurants-bars had the largest increase in online spending shares:
  - 2019 average: 4
  - Peak crisis: 18
  - Latest: 9
  - Latest vs pre-COVID trend: 2.2
- Aggregate sectoral statistics (Table 3):
  - All Categories: 2019 average 18, Peak crisis 23, Latest 21, Latest vs pre-COVID trend 0.3
  - Retail: 2019 average 11, Peak crisis 17, Latest 14, Latest vs pre-COVID trend 0.7
  - Services: 2019 average 34, Peak crisis 41, Latest 36, Latest vs pre-COVID trend -1.0
- Sector-level persistence varies; persistent increases observed in retail, restaurants, and healthcare.
- Top sectors by Latest vs pre-COVID trend (Table 4):
  1. Department Stores — 2019 average 14, Peak 36, Latest 26, Latest vs pre-COVID trend 6.8
  2. Electric-Appliance — 23, 39, 29, 3.2
  3. Clothing Stores — 12, 38, 16, 3.0
  4. Discount Stores — 8, 17, 13, 2.3
  5. Restaurants-Bars — 4, 18, 9, 2.2
  - (Complete table lists 26 disaggregated sectors with exact 2019 average, Peak crisis, Latest, and Latest vs pre-COVID trend values.)
- Some sectors with high pre-pandemic e-commerce shares (e.g., mail order, airline, travel agencies, utilities) experienced little change or convergence to an upper limit.
- Industries with lower or mid pre-pandemic online shares saw larger peak increases (e.g., clothing, electrical appliances, sporting goods-toys).
- Several sectors show declines below trend in latest data: Professional Services (-1.4), Utilities (-1.5), Recreation (-1.6) as Latest vs pre-COVID trend.

### Representativeness of Mastercard data
- Assumption: Mastercard online spending is representative of all card transactions and other online payment methods are negligible or similar in pattern.
- Validation: Comparison with national statistics for the United States and the United Kingdom shows similar dynamic patterns:
  - US: official data suggests online spending share returned to pre-COVID trend levels.
  - UK: remains slightly above the trend-predicted levels.

### Conclusions and interpretation
- The share of online spending rose more in economies where e-commerce was already large; increases are reversing as the pandemic recedes.
- Sector-level evidence shows more persistent increases in retail, restaurants, and healthcare — sectors still affected by limited social mobility and with lower pre-pandemic digitalization allowing for persistent learning.
- The persistence of learning is not broad-based; it is concentrated in sectors with lower pre-pandemic e-commerce shares.
- Mobility plays a crucial role: economies with faster recoveries and higher income per capita returned closer to pre-pandemic trends.
- Open questions for future research include whether post-pandemic increases are driven by the extensive margin (new online customers) or the intensive margin (existing customers increasing online purchases), and the long-term effects given varying travel and gathering restrictions across economies.
- Note on retail seasonality: evaluating seasonally adjusted series for the retail sector, close to 25% of the peak shift online appears to be persisting by comparing current deviation from trend vs peak deviation from trend.

*International Monetary Fund.*

### References

### References

### Key literature cited
- Abraham, Katharine G., Ron S. Jarmin, Brian Moyer, and Matthew D. Shapiro. “Big Data for Twenty-First Century Economic Statistics: The Future Is Now.” National Bureau of Economic Research, 2020.
- Aladangady, Aditya, Shifrah Aron-Dine, Wendy Dunn, Laura Feiveson, Paul Lengermann, and Claudia Sahm. “From Transactions Data to Economic Statistics: Constructing Real-Time, High-Frequency, Geographic Measures of Consumer Spending.” National Bureau of Economic Research, 2019.
- Brynjolfsson, Erik and JooHee Oh, “The Attention Economy: Measuring the Value of Free Digital Services on the Internet,” 2012.
- Carvalho, Vasco M., Stephen Hansen, Álvaro Ortiz, Juan Ramon Garcia, Tomasa Rodrigo, Sevi Rodriguez Mora, and Pep Ruiz de Aguirre. “Tracking the Covid-19 Crisis with High-Resolution Transaction Data” (2020).
- Cavallo, Alberto. “Online and Official Price Indexes: Measuring Argentina’s Inflation.” Journal of Monetary Economics 60, no. 2 (2013): 152–65.
- Cavallo, Alberto. “More Amazon Effects: Online Competition and Pricing Behaviors.” Jackson Hole Economic Symposium Conference Proceedings - Federal Reserve Bank of Kansas City, 2018.
- Choi, Hyunyoung, and Hal Varian. “Predicting the Present with Google Trends.” Economic Record 88 (2012): 2–9.
- Couture, Victor, Benjamin Faber, Yizhen Gu, and Lizhi Liu, “Connecting the Countryside via E-Commerce: Evidence from China,” American Economic Review: Insights, 2020.
- Dolfen, Paul, Liran Einav, Peter J. Klenow, Benjamin Klopack, Jonathan D. Levin, and Larry Levin, 2020, “Assessing the Gains from E-Commerce”, mimeo, Stanford University, May 2020.
- Einav, Liran, and Jonathan Levin. “The Data Revolution and Economic Analysis.” Innovation Policy and the Economy 14, no. 1 (2014): 1–24.
- Glaeser, Edward L., Hyunjin Kim, and Michael Luca. “Nowcasting the Local Economy: Using Yelp Data to Measure Economic Activity.” National Bureau of Economic Research, 2017.
- Goolsbee, Austan and Peter J. Klenow, “Valuing Consumer Products by the Time Spent Using Them: An Application to the Internet,” American Economic Review, 2006, 96 (2), 108–113.
- Jo, Yoon J., Misaki Matsumura, and David E. Weinstein, “The Impact of ECommerce on Relative Prices and Consumer Welfare,” 2019. National Bureau of Economic Research Working Paper 26506.
- Kinda, 2019, “E-commerce as a Potential New Engine for Growth in Asia”, IMF Working Paper Working Paper No. 19/135, July 1, 2019.
- Mian, Atif, Kamalesh Rao, and Amir Sufi. “Household Balance Sheets, Consumption, and the Economic Slump.” The Quarterly Journal of Economics 128, no. 4 (2013): 1687–1726.
- Reinsdorf, Marshall and Paul Schreyer (2020), “Measuring consumer inflation in a digital economy”, in Barbara M. Fraumeni eds Measuring Economic Growth and Productivity, Academic Press, 2020, Pages 339-362.
- Syverson, Chad, “Challenges to Mismeasurement Explanations for the US Productivity Slowdown,” Journal of Economic Perspectives, 2017, 31 (2), 165–86.
- Turrell, Arthur, Bradley J. Speigner, Jyldyz Djumalieva, David Copple, and James Thurgood. “Transforming Naturally Occurring Text Data into Economic Statistics: The Case of Online Job Vacancy Postings.” National Bureau of Economic Research, 2019.
- Varian, Hal, “The Value of the Internet Now and in the Future,” The Economist, 2013.

### Appendix A — Summary of Mastercard transaction-level data and sample statistics
- Data scope and limitations:
  - Evaluated aggregated & anonymized card transactions across the Mastercard network.
  - Certain segments (e.g., vehicle sales) are not captured because the primary form of payment for vehicles are deposits as opposed to cards.
  - Online transactions are primarily by card, producing an inherent payment-form bias; the paper adjusts for this bias by estimating online shares in the universe (detailed in Appendix B).
  - A related database from Mastercard was used by Mian et al (2013).
- Mastercard network scale (as reported):
  - Presence in 210 countries and territories.
  - Connected to 20,000 financial institutions.
  - Over 80 million merchant locations.
  - More than 2.9 billion cards in force.
  - Switching roughly 100 billion transactions in a year.
  - In Q3 2021, the total value of purchases made and cash disbursements obtained on Mastercard cards was nearly 2 trillion dollars.
- Transaction-level fields observed:
  - Date & time: the date and time in which the transaction occurred
  - Type of card: whether the card is a credit, debit, or pre-paid card
  - Merchant location: the address on record for the merchant where the transaction took place
  - Industry classification: what industry the merchant is classified as
  - Channel: whether the payment was an online or brick & mortar transaction
  - Transaction amount: the value of the transaction made
- Transaction-level fields not observed:
  - Cardholder information: Mastercard does not see details around the cardholder, such as their location, income, or account balances.
  - Item-level / SKU-level breakout of the purchase being made. Mastercard sees a single payment amount for all goods purchased.
- Use of Mastercard data in the paper:
  - Leveraged aggregated & anonymized transaction level data at an industry & sector level by market, measured as a monthly time series.
  - Using the sum of transactions that occur each month for point of interaction/card present and online sales, the paper estimates share of online spend by economy.

- Table A1. Summary Statistics for Variables in Table 2 (variables, Obs, Mean, Std. Dev., Min, Max):
  - Online share gap (online share-trend) (in pp), NSA: Obs 899; Mean 0.9; Std. Dev. 2.4; Min -7.0; Max 15.0
  - Residential Mobility, %: Obs 899; Mean 7.5; Std. Dev. 7.3; Min -8.8; Max 40.3
  - Pre-covid trend online share: Obs 899; Mean 12.0; Std. Dev. 10.2; Min 1.2; Max 34.8
  - COVID Fiscal spending as % 2019 GDP: Obs 899; Mean 12.9; Std. Dev. 9.9; Min 0.6; Max 43.4
  - GDP per capita-2019, in '000: Obs 899; Mean 35.6; Std. Dev. 23.7; Min 2.8; Max 115.6
  - Residential mobility * Fiscal spending, 2019: Obs 899; Mean 96.4; Std. Dev. 132.4; Min -149.8; Max 1228.0

### Appendix B — Estimation of online shares in the universe (methodology and formulas)
- Overall-economy online share estimation (two-card example: MC and Visa (V)):
  - Aim: compute s_{c,t}^{Universe} = Total online spending / Total consumption
  - Equations as presented:
    - s_{c,t}^{Universe} = OS_C = MC O_C / C + V O_C / C
    - MC O_MC / MC = V O_V / V = s_{c,t}^{MC}
    - s_{c,t}^{Universe} = s_{c,t}^{MC} * Card_C (MC_Card + V_Card) = s_{c,t}^{MC} * Card_C
    - assuming MC_Card + V_Card = 1
  - Note: the derivation assumes MC’s online share is representative and that online transactions occur only through cards (acknowledged simplification; online transactions can also occur through e-wallets, etc.).

- Sectoral-level online share estimation (industry I, two-card example):
  - Aim: compute s_{c,I,t}^{Universe} = Total online spending in industry I / Total consumption in industry I
  - Equations as presented:
    - s_{c,I,t}^{Universe} = OS_I_C = MC O_I_C / C_I + V O_I_C / C_I
    - MC O_I / MC_I * MC_I_Card_I * Card_I_C_I + V O_I / V_I * V_I_Card_I * Card_I_C_I (as in source notation)
    - Assume MC O_I / MC_I = V O_I / V_I = s_{c,I,t}^{MC}, so MC’s online share in industry I is representative
    - s_{c,I,t}^{Universe} = s_{c,I,t}^{MC} * Card_I / C_I (MC_I_Card_I + V_I_Card_I) = s_{c,I,t}^{MC} * Card_I / C_I
    - assuming MC_I_Card_I + V_I_Card_I = 1

- Practical implication:
  - The paper scales Mastercard’s observed online shares by the card-share of consumption to estimate universe-level online shares, under the representativeness assumption.

### Appendix C — Mastercard market-share cutoff and robustness
- Sample selection rule:
  - Data limited to economies where Mastercard has a significant market share to increase representativeness.
- Robustness statement:
  - Changing this cutoff value has little impact on results.
  - Online shares computed for economies where Mastercard has a small share of the card market were compared to results from the 47 economies in the sample.
  - Reported correlation:
    - Online shares are highly correlated between these two groups, with a correlation coefficient of 0.97.
- Figure referenced:
  - Figure C1: Online shares in economies where Mastercard has >=20% and <20% market share.

*E-commerce During Covid: Stylized Facts from 47 Economies Working Paper No. [WP/YYYY/###]*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022019-print-pdf.pdf_