## wpiea2022064-print-pdf

## Source details

**Canonical URL:** [wpiea2022064-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022064-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022064-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022064-print-pdf.pdf.json)

---

### Overview and methodology
- Purpose: assess household vulnerability and resilience to the income shock inflicted by the Covid-19 pandemic using household-level survey data from Cambodia, Nepal and Vietnam.
- Country thematic focus:
  - Cambodia: labor markets and cash transfer policy.
  - Vietnam: financial access and financial health.
  - Nepal: availability of income-smoothing mechanisms and historical shock responses.
- Key methodological caveats:
  - Analysis relies on data collected before the pandemic; actual post-pandemic balance sheet effects are not observed directly.
  - Simulations use ad-hoc assumptions on shock distribution across households and abstract from general equilibrium effects.
  - Results should be interpreted with caution given strong assumptions.

### Cambodia — labor market vulnerability and cash transfer simulation
- Macro and survey indicators:
  - Aggregate output declined by 3.1 percent in 2020.
  - 83 percent of Cambodian households reported a decline in income in the first half of 2020.
  - About 50 percent witnessed their incomes declining further since then.
  - Retail and hospitality employ around 20 percent of Cambodian workers.
- 2020 Cash Transfer Program for the Poor and Vulnerable Households:
  - Initial budget: USD 300 mln for the first year (1 percent of 2020 GDP).
  - About 710,000 households (2.8 million people) received the cash transfer in 2020.
  - Monthly subsidy: 45 USD on average, corresponding to about 33 percent of monthly household earnings at the 25th percentile of the distribution in 2019/2020.
  - Transfers delivered electronically via Wing agents after SMS notification.
  - Many recipients received transfers until end-2020 (6-7 months on average); parts of the scheme expanded into 2021.
- Simulation assumptions:
  - Households’ income declines in proportion to the share of household working members employed in retail, restaurants, logistics or hospitality; workers in other sectors assumed unchanged.
  - Transfer eligibility approximated by households below the 12th percentile of the baseline income distribution and/or in possession of an IDPoor card.
  - Transfer size set to 45USD per month (US$ 1.50 per day).
- Simulated outcomes and distributional patterns:
  - Baseline pre-pandemic poverty: 10 percent of Cambodian households below US$ 1.90 per worker/day (PPP).
  - Without the cash transfer, 17.3 percent of households would have seen income fall below US$ 1.90 per worker/day (PPP).
  - With the cash transfer, share below the international poverty line estimated to be 12.6 percent, or a 2.6. p.p. increase from the pre-pandemic baseline.
  - The cash transfer may have lifted 164,000 households out of income poverty, compared to the pre-pandemic baseline.
  - Estimated 260,000 households may have fallen below the US$ 1.90 threshold in 2020 but were outside the cash transfer coverage.
  - Total estimated households living below poverty line after the pandemic: 459,000.
  - “New poor” are typically urban and work in services; cash transfer recipients are less likely to be urban and less likely to be employed in hospitality.
- Private coping mechanisms:
  - Urban households increased selling durable goods and propensity to reduce food spending in early 2020; reliance on savings drawdown and additional borrowing remained stable.
  - Rural households reduced their tendency to rely on most income-smoothing mechanisms in the same period.
- Targeting trade-offs and gaps:
  - Transfers "did very well in tackling the forms of extreme poverty which pre-dated the pandemic" by targeting bottom deciles.
  - Cash transfer tended to miss many urban, service sector workers (a “missing middle”) because recent income losses are unlikely to be fully recorded in real time and urban households often hold illiquid wealth or are ineligible for IDPoor.
  - Urban service-sector workers may constitute a new pocket of poverty, more likely to draw down on savings and borrow extensively.

### Vietnam — financial vulnerability, measures, and simulated shocks
- Data:
  - Source: 2018 Vietnam Household Living Standards Survey (VHLSS).
  - Subsample: approximately 9,400 households (9,399 answered extended questionnaire).
- Three measures of financial vulnerability:
  - Savings rate (net income – total expenditure) / net income; does not include debt services.
  - Debt-to-income ratio (DTI): total debt / net income.
  - Financial margin (FMR): FMR_h = (net income_h − total expenditure_h) / (net income_h − DSTI_h), if net income_h > 0; DSTI_h computed from loan values, interest rates and imputed durations.
- Household characteristics and averages:
  - Median household: family of four and an annual income of about USD 5,500 (using the exchange rate from Dec. 2018).
  - Approximately 2 out of 3 households located in rural areas.
  - Over half have at least one member working directly in agriculture.
  - Over 45 percent operate as small businesses.
  - 53 percent have at least one member that works informally.
- Savings and expenditure patterns:
  - Average household saves 23 percent of their income.
  - Over 16 percent of households report negative savings (calculated as a flow).
  - Food and daily items comprise 68 percent of total expenses for the lower quintile versus 61 percent for the top quintile.
- Debt distribution (percentile debt values in USD):
  - 1: 107.78
  - 25: 862.20
  - 50: 1810.62
  - 75: 4311.00
  - 99: 64665.00
  - MEAN: 9223.45
  - SD: 117977
- Debt and FMR statistics:
  - Household debt-to-GDP grew rapidly after 2014, reaching levels similar to China’s.
  - Calculated DTI at household level: average ratio around 23 percent.
  - Three-quarters of households in the data do not have any outstanding loans.
  - DTI for households at the bottom quintile is almost twice as large as for households at the top quintile.
  - In VHLSS, 19 percent of all households have a negative FMR; increases to 37 percent when considering only households that hold outstanding debt.
  - Around 3 percent of households have FMR < -1 (very likely to be in financial distress).
- Access to financial services:
  - Most households do not report use of financial services (ATM card, bank account, savings account, credit card, life/other insurance, securities).
  - Financing source categories: (1) commercial banks; (2) other banks (Social Policy Bank, Bank of Agriculture and Rural Development, People’s Credit Funds, local government, other credit institutions); (3) group associations; (4) friends and relatives; (5) private or informal.
  - “Other banks” account for almost two-thirds of all loans but only about one-third of the loan value (smaller loans with purpose constraints).
- Regression evidence:
  - A 1 percent increase in household income decreases the probability of financial distress by 0.26 percentage points.
  - Financial access decreases financial vulnerability, with larger effects for poorer households (positive interaction between financial access and log(income)).
  - Financial access reduces DTI and increases FMR; coefficient on savings rate positive but not significant.
- Simulated COVID-19 shock model:
  - Income under shock ε: y_h(ε) = y_h(0) × (1 − s_hi ε − 0.4 s_hf ε), ε ∈ {0.25, 0.50}; formal workers who lost jobs eligible for benefit equal to 60 percent of their income.
- Predicted share of households with FMR < 0 (%):
  - ε = 0 (baseline): 12.23
  - ε = 0.25: 15.11 (difference relative to baseline: 2.88)
  - ε = 0.5: 22.56 (difference relative to baseline: 10.33)
- Alternative (conservative) recomputation holding debt and expenditures constant (Appendix B.3) — share of households with FMR < 0 (%):
  - ε = 0 (baseline): 19.00
  - ε = 0.25: 30.67 (Difference = 11.67)
  - ε = 0.5: 46.21 (Difference = 27.21)
- Interpretation notes:
  - Savings rate measure may understate risk for indebted households; DTI misses households without financial access; FMR requires imputed loan durations and excludes liquid assets.
  - Conclusions based on common trends across all three measures indicate substantial financial vulnerability concentrated among informal service workers.

### Nepal — shocks, coping strategies, regressions, and simulated Covid-19 impacts
- Data:
  - Nepal Household Risk and Vulnerability Survey (HRVS): three-year panel (2016-2018) of 6,000 households and 400 communities in non-metropolitan Nepal.
  - Retention: approximately 94% of households surveyed in all three waves.
  - Geographic coverage targeted rural and semi-urban areas; Kathmandu Valley excluded.
- Baseline exposure and response:
  - In any given year, 45 percent of households face at least one predefined shock.
  - 9 percent of households reduce consumption in response to shocks.
  - Average financial loss per shock: 64,844 NRs.
  - Disease and death constitute over a quarter of observed shocks; other natural disasters comprise another quarter.
- Coping strategies (ex-post, past 12 months):
  - Over three quarters of households relied on savings following the shock.
  - 15 percent borrowed from friends and families.
  - Borrowing from money lenders was the next most common source.
  - Very few households accessed formal finance.
  - Over 80 percent of households used only one coping strategy.
- Anticipated ex-ante coping strategies for hypothetical shock of NRs25,000 (approx. US$210):
  - Most anticipated relying on savings and borrowing from friends/family.
  - Very few anticipated using formal financial loans.
- Regression framework (selected results):
  - Y_it = β0 + β1 NumberShocks_it + β2 X_it + β3 Rem_it + β4 Gov_it + β5 Credit_it + d_i + d_t + ε_it
  - Each additional shock associated with:
    - Financial loss: 30,000 rupees per additional shock.
    - Likelihood of reduced consumption: ~13 percentage points per additional shock.
  - Remittance receipt associated with a 2.3 percentage point reduction in the probability of reducing consumption (stronger for non-food consumption).
  - Government assistance has no statistical impact on probability of reducing overall consumption in this dataset (programs minimal at survey time).
  - For every 100,000 Nepali Rupees in savings, households are 0.3 percentage points less likely to reduce consumption.
  - Bank account savings (not cash holdings) provide consumption insurance.
  - Access to loans associated with a 2 percentage point reduction in probability of reducing consumption; most loans come from friends or family.
  - Loans from financial corporations provide slightly better consumption smoothing; money lenders provide no consumption smoothing benefits.
  - Net assets (savings minus loans) are not associated with weakened consumption impacts, implying loans and savings effects are not simply a wealth channel.
  - Results robust to probit/logit specifications and additional controls.
- Simulation of Covid-19 employment shock:
  - Identification: households working in services (dominated by tourism) and construction classified as directly at risk.
  - 43 percent of households employed or self-employed in these industries.
  - Average expected loss for an additional shock for these households: 53,671 NRs (~US$450).
  - Simulated additional 6 percent of households expected to reduce consumption in direct response to Covid-19 job disruption (a total of 15 percent of households).
  - Assumptions: full employment effects for exposed occupations and no indirect effects for other households.
  - External validation: Egger et al (2021) survey found 40 percent lost income, 20 percent lost jobs, and 11 percent missed or reduced meals — broadly consistent with simulated magnitudes.

### Cross-country findings and policy implications
- Common findings:
  - Occupation and ex-ante income levels are main drivers of vulnerability to Covid-19.
  - Low-wage workers, more likely informal and in services, were most affected and have limited means to smooth consumption due to limited savings and lack of access to financial services or self-insurance mechanisms.
  - Government policy and access to self-insurance mechanisms have key roles in dampening the effects of aggregate shocks.
- Policy implications:
  - Social protection: Creating appropriate fiscal room for well-targeted social spending is crucial for building resilience to shocks in low-income and developing economies.
    - Cambodia example: cash transfer scheme may have kept 5 percent of the population out of income poverty compared to a counterfactual without government support.
  - Financial sector: Foster consumer access to a formal and adequately regulated financial services sector to improve self-insurance for idiosyncratic shocks and reduce reliance on informal lending.
    - Ensuring access to traditional banking services may reduce reliance on informal lending and mitigate rising household leverage.
  - Targeting principle: Understanding heterogeneous responses to aggregate events is crucial to ensure adequate and equitable policy responses, inform appropriate targeting, and maximize impact of scarce fiscal resources in LICs and EMEs.

*IMF WORKING PAPERS Household Vulnerability to Income Shocks in Emerging and Developing Asia — INTERNATIONAL MONETARY FUND*

### Introduction............................................................................................................

### Introduction

### Overview
- Document sections listed:
  - Introduction ......................................................................................................................................................................... 4
  - Data ....................................................................................................................................................................................... 5
  - Labor market vulnerability in Cambodia ..................................................................................................................... 6
  - Financial vulnerability in Vietnam ............................................................................................................................... 12
  - Coping strategies and resilience in Nepal ................................................................................................................ 17
  - Summary and policy implications............................................................................................................................... 23
  - References ........................................................................................................................................................................ 24

### Figures included
- Figure 1: Distribution of Cambodian Workforce by Sector ................................................................7
- Figure 2: Simulating the Effects of the Pandemic on the Distribution of Income .....................................9
- Figure 3: Simulating the Effects of the Cash Transfer .........................................................................9
- Figure 4: Demographics of Recipients vs New Poor .........................................................................10
- Figure 5: Percentage of Households Resorting to Income-Smoothing Mechanisms .............................11
- Figure 6: Average Household Savings Rate ....................................................................................12
- Figure 7: Debt Value Distribution and Purpose of Loans ...................................................................13
- Figure 8: Household Debt and Income ...........................................................................................13
- Figure 9: Financial Margin Distribution ..........................................................................................14
- Figure 10: Financial Services Currently Used .................................................................................14
- Figure 11: Total Loan Value and Interest Rate by Source of Finance ...................................................15
- Figure 12: Predicted and Counterfactual FMR .................................................................................16
- Figure B.1: Loan Age and Predicted Duration .................................................................................27
- Figure B.2: FMR Density After COVID-19 Income Shock ...................................................................31

### Tables included
- Table 1: Percentage of HHs Using Each Coping  Strategy Following  a Shock in the Preceding 12 Months .................................................................................................................................................17
- Table 2: Baseline Indicators for Vulnerability and Resilience ............................................................20
- Table 3: Impacts of Financial Access on Shock Outcomes ...............................................................21
- Table A.1: Descriptive Statistics; Income Distribution Under Different Scenarios ................................25
- Table B.1: Probability of Debt and Financial Distress .......................................................................28
- Table B.2: Determinants of Household Vulnerability ........................................................................30
- Table C.1: Household Descriptive Statistics for Nepal’s Household Risk and Vulnerability Survey (HRVS) .................................................................................................................................................32
- Table C.2: Types of Shocks Faced by Households in Nepal’s HRVS ...................................................33
- Table C.3: Household  Ex-ante Coping  Strategies: “How would you  respond to a shock of NRs25,000 (approx. US$210)?” ....................................................................................................................34

### Appendices
- Appendix A: Details for Cambodia.............................................................................................................................. 25
- Appendix B: Details for Vietnam ................................................................................................................................. 26
- Appendix C: Details for Nepal ...................................................................................................................................... 32

*IMF WORKING PAPERS Household Vulnerability to Income Shocks in Emerging and Developing Asia — INTERNATIONAL MONETARY FUND*

### Introduction

### Introduction

### Overview
- Purpose: assess household vulnerability and resilience to the income shock inflicted by the Covid-19 pandemic using household-level survey data from three countries in emerging and developing Asia: Cambodia, Nepal and Vietnam.
- Country focus by theme:
  - Cambodia: labor markets.
  - Vietnam: financial access and financial health.
  - Nepal: availability of income-smoothing mechanisms and historical shock responses.
- Common findings:
  - Occupation and ex-ante income levels are the main drivers of vulnerability to Covid-19.
  - Government policy and access to self-insurance mechanisms have key roles in dampening the effects of this aggregate shock.

### Cambodia — labor market vulnerability and policy response
- Macro and survey indicators:
  - Aggregate output is estimated to have declined by 3.1 percent in 2020.
  - As many as 83 percent of Cambodian households reported suffering a decline in income in the first half of 2020.
  - About 50 percent witnessed their incomes declining further since then.
  - Retail and hospitality employ around 20 percent of Cambodian workers.
- Social protection before and during the pandemic:
  - Pre-June 2020: “Identification of Poor Households” (IDPoor) used mainly for maternal/disability support and scholarships; unemployment insurance is absent; public health and retirement pension schemes limited to civil servants.
  - 2020 Cash Transfer Program for the Poor and Vulnerable Households:
    - Initial budget: USD 300 mln for the first year (1 percent of 2020 GDP).
    - About 710,000 households (2.8 million people) received the cash transfer in 2020.
    - Monthly subsidy: 45 USD on average, corresponding to about 33 percent of monthly household earnings at the 25th percentile of the distribution in 2019/2020.
    - Transfers delivered electronically via Wing agents after SMS notification.
    - Many recipients received the cash transfer until end-2020 (6-7 months on average); parts of the scheme expanded into 2021.
- Simulation of income shock and policy effectiveness:
  - Simulation assumptions:
    - Households’ income declines in proportion to the share of household working members employed in retail, restaurants, logistics or hospitality; workers in other sectors assumed unchanged.
    - Transfer eligibility approximated by households below the 12th percentile of the baseline income distribution and/or in possession of an IDPoor card; transfer size set to 45USD per month (US$ 1.50 per day).
  - Key simulated outcomes:
    - Without the cash transfer, 17.3 percent of households would have seen income fall below 1.9 USD per worker/day (PPP).
    - Baseline pre-pandemic poverty: 10 percent of Cambodian households below that threshold.
    - With the cash transfer, the share of households below the international poverty line is estimated to be 12.6 percent, or a 2.6. p.p. increase from the pre-pandemic baseline.
    - The cash transfer may have lifted 164,000 households out of income poverty, compared to the pre-pandemic baseline.
    - Survey evidence: percentage of households reporting some degree of food insecurity declined from 67 percent to 36 percent between August and October 2020.
  - Coverage gaps and distributional patterns:
    - An estimated 260,000 households may have fallen below the US$ 1.90 threshold in 2020 but were outside the cash transfer coverage.
    - Total estimated households living below poverty line after the pandemic: 459,000.
    - “New poor” are typically urban and work in services; cash transfer recipients are less likely to be urban and less likely to be employed in hospitality.
  - Private coping mechanisms:
    - Urban households increased selling durable goods and propensity to reduce food spending in early 2020; reliance on savings drawdown and additional borrowing remained stable.
    - Rural households reduced their tendency to rely on most income-smoothing mechanisms in the same period.

### Vietnam — financial vulnerability and access
- Data and measures:
  - Analysis uses the 2018 vintage of the Vietnam Household Living Standards Survey (VHLSS).
  - VHLSS subset: approximately 9,400 households (9,399 households answered extended questionnaire).
  - Three measures of financial vulnerability constructed: savings rate, debt-to-income ratio (DTI), and financial margin (FMR).
- Findings and concerns:
  - Cash transfers were significantly smaller and did not reach a large proportion of affected households.
  - Much of the population coped by tapping savings or taking debt.
  - Access to financial services is strongly related to better financial health but remains a constraint for a large share of households.
  - Estimated that the pandemic could have placed over 10 percent of the population in a financially vulnerable position, with impact concentrated on informal workers in the service sector.
- Methodological notes on measures:
  - Savings rate does not include debt services and could underestimate financial risk for indebted households.
  - DTI offers limited insight into households with no financial access.
  - FMR includes both savings rate and debt services but requires variables not observed in the data.
  - Conclusions are based on common trends across all three measures.

### Nepal — shocks, coping strategies, and simulation
- Data source:
  - Nepal Household Risk and Vulnerability Survey (HRVS): three-year panel (2016-2018) of 6,000 households and 400 communities in non-metropolitan Nepal.
  - Retention: approximately 94% of households surveyed in all three waves.
  - Geographic coverage targeted rural and semi-urban areas; Kathmandu Valley excluded.
- Baseline exposure and response:
  - In any given year, 45 percent of households face at least one predefined shock.
  - 9 percent of households reduce consumption in response to shocks, indicating imperfect consumption insurance.
  - ‘Shock’ is defined per the HRVS shock list (e.g., natural disasters, loss of health/life, financial shocks such as job loss, bankruptcy).
- Mitigating strategies and simulated Covid-19 impact:
  - Remittances and access to finance are effective mitigating strategies, reducing the probability that households reduce consumption.
  - Simulated results estimate an additional 6 percent of households (15 percent in total) are forced to reduce consumption in response to Covid-19 employment effects.

### Contributions, methods, and caveats
- Main contributions:
  - Descriptive analysis of household vulnerability to income shocks in developing Asia, focusing on institutional context differences (informality, limited fiscal space, financial access constraints).
  - Quantitative estimates of potential distributional implications of the Covid-19 pandemic via household-level simulations highlighting channels: coping mechanisms, financial inclusion, and labor markets.
- Methodological caveats:
  - Analysis relies on data collected before the pandemic; actual effects on household balance sheets and relative responses cannot be observed directly.
  - The paper simulates potential implications under ad-hoc assumptions on the distribution of the shock across households; results should be interpreted with caution due to strong assumptions and abstracting from general equilibrium effects.
- Structure of the paper:
  - Section 2: data description.
  - Section 3: Cambodia analysis and cash transfer impact.
  - Section 4: Vietnam financial vulnerability.
  - Section 5: Nepal coping strategies.
  - Section 6: conclusions and policy implications.

*IMF Working Paper — Introduction*

### 2020. This evidence is line with the predictions of the simulation, which indicates that rural households may

### wpiea2022064-print-pdf - 2020

### Cambodia — distributional effects and cash transfers
- Rural households may have been less negatively affected by the 2020 shock, explained in part by lower exposure to the services sector and a higher relative incidence of government support in rural regions.
- The simulation assumes that transfers are targeted to the bottom decile of the income distribution.
- Many urban, service sector workers are above these income thresholds; the poorest households tend to be rural and living off subsistence agriculture.
- The cash transfer scheme:
  - "did very well in tackling the forms of extreme poverty which pre-dated the pandemic" by targeting households in the bottom deciles.
  - tended to miss many urban, service sector workers (a “missing middle”) because:
    - recent income losses are unlikely to be fully recorded in real time;
    - urban households tend to hold higher levels of illiquid wealth and may have been ineligible for IDPoor cash transfers;
    - most service sector workers tend to be self-employed and are not eligible for furlough schemes.
- Urban service-sector workers may constitute a new pocket of poverty, more likely to draw down on savings and borrow extensively in the aftermath of the pandemic.

*Source: Cambodia Socio-Economic Survey (2019/2020) and authors’ calculations.*

### Vietnam — financial vulnerability and access to finance
- Background household characteristics (VHLSS 2018):
  - Median household: family of four and an annual income of about USD 5,500 (using the exchange rate from Dec. 2018).
  - Approximately 2 out of 3 households surveyed are located in rural areas.
  - Over half have at least one member working directly in agriculture.
  - Over 45 percent of households operate as small businesses.
  - 53 percent have at least one member that works informally.
- Savings and expenditure:
  - On average, households save 23 percent of their income.
  - Over 16 percent of households report negative savings (calculated as a flow, not a stock).
  - Food and other daily items comprise 68 percent of total expenses for households in the lower quintile, versus 61 percent for households in the top quintile.
- Debt distribution (Figure 7):
  - Percentile debt values (USD):
    - 1: 107.78
    - 25: 862.20
    - 50: 1810.62
    - 75: 4311.00
    - 99: 64665.00
  - MEAN: 9223.45
  - SD: 117977
- Household debt-to-GDP and DTI:
  - Household debt-to-GDP grew rapidly after 2014, reaching levels similar to China’s.
  - Calculated DTI at household level from VHLSS data: average ratio around 23 percent.
  - Three-quarters of households in the data do not have any outstanding loans.
  - DTI ratio for households at the bottom quintile is almost twice as large as for households at the top quintile.
- Financial margin (FMR):
  - Approximately 1 out of 5 households in the VHLSS sample are in a financially unsustainable position, measured by FMR below zero.
  - In the VHLSS, 19 percent of all households have a negative FMR, increasing to 37 percent if only considering households that hold outstanding debt (loans).
  - Around 3 percent of households have FMR < -1 (very likely to be in financial distress).
- Access to financial services:
  - Most households do not report use of financial services (ATM card, bank account, savings account, credit card, life/other insurance, securities).
  - Classification of financing sources: (1) commercial banks; (2) other banks (Social Policy Bank, Bank of Agriculture and Rural Development, People’s Credit Funds, local government, other credit institutions); (3) group associations; (4) friends and relatives; (5) private or informal.
  - “Other banks” own almost two-thirds of all loans but only about one-third of the loan value, indicating smaller loans with purpose constraints.
- Regression evidence on vulnerability:
  - A 1 percent increase in household income decreases the probability of financial distress by 0.26 percentage points.
  - Financial access decreases financial vulnerability, with larger effects for poorer households (interaction between financial access and log(income) is positive).
  - Financial access reduces DTI and increases FMR; coefficient on savings rate is positive but not significant.

### Vietnam — simulated COVID-19 shock impacts
- Shock modelling:
  - Household h income under shock ε: y_h(ε) = y_h(0) × (1 − s_hi ε − 0.4 s_hf ε)
    - y_h(0): household h’s observed income (pre-COVID-19)
    - s_hi: share of adults working informally in the service sector
    - s_hf: share of adults with formal jobs in the service sector
    - ε ∈ {0.25, 0.50} (size of income shock; ε=0.25 indicates crisis lasts one quarter; ε=0.50 indicates longer duration)
  - Formal workers who lost jobs were eligible for a benefit equal to 60 percent of their income (World Bank, 2021b).
- Predicted share of households with FMR < 0 (%):
  - Value of income shock, ε:
    - 0 (baseline): 12.23
    - 0.25: 15.11
    - 0.5: 22.56
  - Difference relative to baseline:
    - 0.25 shock: 2.88
    - 0.5 shock: 10.33
- Notes:
  - Predicted FMR fails to capture the relevant mass of households with FMR < -1; an alternative method in appendix C yields even stronger results.

### Nepal — shocks, coping strategies, and resilience
- Background (HRVS):
  - In any given year, 9 percent of households had to reduce consumption in response to a shock.
  - 45 percent of households faced at least one shock.
  - The average financial loss was 64,844 NRs.
  - Disease and death of a family member make up over a quarter of observed shocks; other natural disasters comprise another quarter.
- Coping strategies (responses to shocks in prior 12 months):
  - Over three quarters of households relied on savings following the shock.
  - 15 percent borrowed from friends and families.
  - Borrowing from money lenders was the next most common.
  - Very few households accessed formal finance.
  - Over 80 percent of households used only one coping strategy.
- Anticipated coping strategies (ex ante, for a shock of NRs25,000 ≈ USD210):
  - Very few indicated they would access formal financial channels.
  - More indicated they would borrow from friends and family.
- Regression framework (Equation 1):
  - Y_it = β0 + β1 NumberShocks_it + β2 X_it + β3 Rem_it + β4 Gov_it + β5 Credit_it + d_i + d_t + ε_it
  - Outcomes: financial loss (NRs) and binary indicator for reduced consumption; household fixed effects and time fixed effects included.
- Baseline regression results (Table 2 summary):
  - Each additional shock associated with:
    - Financial loss: 30,000 rupees per additional shock.
    - Likelihood of reduced consumption: ~13 percentage points per additional shock.
  - Number of children, number of elderly, and vulnerable employment status associated with smaller financial loss (possibly due to lower pre-shock affluence).
  - Households self-employed in agriculture are 1.3 percentage points more likely to reduce consumption.
  - Remittance receipt associated with a 2.3 percentage point reduction in the probability of reducing consumption (stronger for non-food consumption).
  - Government assistance has no statistical impact on the probability of reducing overall consumption in this dataset (caution: government assistance programs were very minimal at survey time).
- Savings and credit effects (Table 3 summary):
  - For every 100,000 Nepali Rupees in savings, households are 0.3 percentage points less likely to reduce consumption.
  - Bank account savings (not cash holdings) provide consumption insurance.
  - Access to loans associated with a 2 percentage point reduction in the probability of losing consumption; most loans come from friends or family.
  - Loans from financial corporations provide slightly better consumption smoothing.
  - Money lenders (loan sharks) provide no consumption smoothing benefits.
  - Net assets (savings minus loans) are not associated with weakened consumption impacts, implying loans and savings effects are not simply a wealth channel.
- Robustness: results hold under probit/logit specifications and additional controls (value of government assistance and remittances, type of shock).

### Nepal — simulated COVID-19 shock impacts
- Identification of directly exposed households:
  - Households working in services (dominated by tourism) and construction are identified as directly at risk.
  - 43 percent of households are employed or self-employed in these industries.
  - Average expected loss for an additional shock for these households: 53,671 NRs (~US$450).
- Simulation result:
  - Estimated additional 6 percent of households expected to reduce consumption in direct response to COVID-19 job disruption (a total of 15 percent of households).
  - Assumptions: full employment effects for exposed occupations and no indirect effects for other households.
- External validation:
  - Egger et al (2021) surveyed Nepali households in 2020 and found 40 percent lost income, 20 percent lost jobs, and 11 percent missed or reduced meals — broadly consistent with simulated magnitudes.

### Cross-country summary and policy implications
- Aggregate pandemic shock in emerging Asia was large but effects were heterogeneous.
- Low-wage workers, more likely informal and in services, were most affected and have limited means to smooth consumption due to limited savings and lack of access to financial services or self-insurance mechanisms.
- Public intervention findings:
  - Cambodia example: newly enacted cash transfer scheme may have kept 5 percent of the population out of income poverty compared to a counterfactual without government support.
  - Creating appropriate fiscal room for well-targeted social spending is crucial for building resilience to shocks in low-income and developing economies.
- Financial sector policy implications:
  - Experiences of Nepal and Vietnam highlight fostering consumer access to a formal and adequately regulated financial services sector.
  - Ensuring access to traditional banking services may improve self-insurance for idiosyncratic shocks and reduce reliance on informal lending, which may underlie rising household leverage.
- General principle:
  - Understanding heterogeneous responses to aggregate events is crucial to ensure adequate and equitable policy responses, inform appropriate targeting, and maximize impact of scarce fiscal resources in LICs and EMEs.

*Italic: IMF WORKING PAPERS Household Vulnerability to Income Shocks in Emerging and Developing Asia — INTERNATIONAL MONETARY FUND*

### References

### wpiea2022064-print-pdf - References

### References cited
- Bellon, Matthieu; Pizzinelli, Carlo; Perrelli, Roberto (2020). Household Consumption Volatility and Poverty Risk: Case Studies from South Africa and Tanzania IMF Working Paper No. 20/51.
- Carlson, Stacy; Dabla-Norris, Era; Saito, Mika; Shi, Yu (2015). Household financial access and risk sharing in Nigeria, IMF Working Paper No. 15/169.
- Coady, David; Grosh, Margaret; Hoddinott, John (2004). Targeting of Transfers in Developing Countries, World Bank Publishing.
- Cobion, Olivier; Gorodnichenko, Yuriy; Weber, Michael (2021). How Did U.S. Consumers Use Their Stimulus Payments? NBER Working Paper Series, 27693/2020.
- Dabla-Norris, Era; Gradstein, Mark; Inchauste, Gabriela (2005). What Causes Firms to Hide Output? The Determinants of Informality, IMF Working Paper No. 05/160.
- Dabla-Norris, Era; Ganelli, Giovanni; Nguyen, Anh Thi Ngoc Nguyen; Nguyen, Mai Thi Thanh; Vu, Thuy Thi Thu (2020). Role of Individual Characteristics and Policies in Driving Labor Informality in Vietnam, IMF Working Paper 2020/273.
- Dercon, Stefan (2002). Income Risk, Coping Strategies, and Safety Nets. The World Bank Research Observer, Volume 17, Issue 2, September 2002, Pages 141–166.
- Gerard, François; Imbert, Clément; Orkin, Kate (2020). Social protection response to the COVID-19 crisis: options for developing countries, Oxford Review of Economic Policy, Volume 36, Issue Supplement_1, 2020, Pages S281–S296.
- Egger, Dennis et al. (2021). Falling living standards during Covid-19 crisis: Quantitative evidence from nine developing countries. Science Advances 7(6).
- Kinnan, Cynthia; Townsend, Robert (2012). Kinship and Financial Networks, Formal Financial Access, and Risk Reduction, American Economic Review 102(3), pp. 289-93.
- Townsend, Robert (1995). Consumption Insurance: An Evaluation of Risk-Bearing Systems in Low-Income Economies. Journal of Economic Perspectives 9(3).
- Morgan, Peter; Trinh, Long (2021). Impacts of COVID-19 on Households in ASEAN Countries and Their Implications for Human Capital Development, ADBI Working Paper Series 1226/2021.
- World Bank (2020a). Cambodia Economic Update, November.
- World Bank (2020b). The Socioeconomic impact of Covid-19 on Households in Cambodia, n.2 November.
- World Bank (2020c). Reversals of Fortune: Poverty and Shared Prosperity, World Bank Group, 2020.
- World Bank (2021a). The Socioeconomic impact of Covid-19 on Households in Cambodia, n.3 February.
- World Bank (2021b). The Labor Market and the COVID-19 Outbreak in Vietnam: Impacts and Lessons Learned for Social Protection, World Bank Publishing.

### Appendix B: Details for Vietnam
- B.1 Measures of financial vulnerability:
  - Savings Rate:
    - Defined as the ratio of savings (net income – total expenditure) to net income.
    - Net income computed as the sum of:
      - All revenue from employment, including wages, bonuses, subsidies, and other revenues;
      - Revenue from education and healthcare aid, and from rental of properties;
      - Sales of crops, and revenues from land, animal husbandry, hunting, agriculture, aquaculture, forestry and other non-farm activities, net of their respective costs of production.
    - Total expenditure includes:
      - Expenditures of education, healthcare and festive occasions;
      - Recurrent expenditures such as food and daily-use items (measured monthly and multiplied by 12 for yearly expenditures);
      - Expenditure on non-food items (yearly), durable goods and housing.
    - Durable goods assumed lifespan: 10 years.
    - Housing expenditures assumed lifespan: 30 years.
    - Expenditure values divided by lifespan to construct yearly expenses.
  - Debt-to-Income Ratio:
    - Defined as the ratio of a household’s total debt to its net income.
    - Total debt computed by summing outstanding amount of each loan (total loan amount minus principal payments already made).
    - Note: Loans and amounts are self-reported in the VHLSS; other forms of debt are not observed—both can lead to underestimation of total obligations.
  - Financial Margin (FMR):
    - Defined for each household h as:
      - FMR_h = (net income_h − total expenditure_h) / (net income_h − DSTI_h), if net income_h > 0.
    - DSTI_h (debt service-to-income) assumed equal amortization payments each year and defined as:
      - DSTI_h = (1 / net income_h) * sum_{j=1 to N_h} (L_j d_j + L_j r_j),
        where N_h is number of loans for household h, L_j is loan j value, r_j is annual interest rate, d_j is duration in years.
    - Liquid assets excluded from FMR due to data constraints (difference from Leika and Marchettini, 2017).
    - Loan duration not observed; loan age available and used to impute maturity:
      - Regress log(loan age) on log(interest rate), dummies for lending institution, and dummies for loan purpose.
      - Use predicted loan age and rescale so average mortgage loan maturity = 10 years (see figure B.1).
      - Predicted duration plugged into DSTI formula to compute FMR for each household.
- B.2 Regression analysis:
  - Modelled probabilities:
    - Probability household has debt.
    - Probability household is in unsustainable financial position (FMR < 0).
  - Explanatory variables:
    - Household net income, education attainment and gender of household head, family size, share of working-age members employed, share working informally, household region (6 macro-regions of Vietnam: North Midlands and Mountains, Red River Delta, North and Coastal Central, Central Highlands, Southwest, and Mekong River Delta), urban/rural dummy, internet access, financial access.
    - Interaction term between household income and financial access.
  - Estimation approach:
    - Linear probability model (OLS); results qualitatively consistent with non-linear models (e.g., logistic regression).
    - Coefficients shown in table B.1. Column (1): probability FMR < 0; column (2): same excluding households with no outstanding debt; column (3): probability of having debt.
  - Key regression results and interpretations:
    - A 1 percent increase in a household’s income, all else equal, decreases the probability that a household has a negative FMR by 0.26 percentage points.
    - Stronger impact when analysis restricted to households that hold debt.
    - Higher income associated with lower probability of outstanding loans.
    - Financial access drastically decreases probability household is in unsustainable financial position (causality caveat due to possible unobserved variables).
    - Interaction coefficient between financial access and log(income) positive: low income households benefit more from having financial access.
    - Financial access does not have a statistically significant impact on probability of debt (possibly because social banks and other credit institutions provide loans to households without financial access).
    - Larger families have higher probability of taking loans and being in more vulnerable financial position.
    - Higher share of working-age individuals employed associated with similar pattern but with less precise estimates.
    - Households with larger share working informally have lower probability of negative FMR (possible explanations: informal workers build more savings or have less access to debt).
    - More educated households and those with internet access are more likely to have debt and a negative FMR (effect of education not monotonic). Possible reasons: access to sophisticated financial services or ability to obtain larger loans; unobserved assets may bias FMR downward.
  - Table references:
    - Table B.1: Probability of Debt and Financial Distress (table referenced).
    - Table B.2: Determinants of Household Vulnerability (table referenced). Columns (1)–(3): full sample; columns (4)–(6): households with some amount of debt.
  - Continuous measures analysis:
    - Dependent variables: savings rate, DTI, FMR.
    - Income is a strong determinant: higher income → higher savings rate, lower debt relative to income, higher FMR (holds for full sample and for debt-holding subsample).
    - Financial access reduces DTI and increases FMR; impact on savings rate not statistically significant.
    - Interaction: financial access effects larger for low-income households (coefficient on interaction: positive for DTI, negative for FMR).
    - Larger households save less, reducing FMR.
    - Households with higher share employed save less and have lower FMR; households with more informal workers behave oppositely (standard errors larger; many estimates not statistically significant).
    - Urban, more educated, and internet-access households tend to be more financially vulnerable, possibly due to access to larger loans or unobserved wealth/assets biasing FMR.
- B.3 Alternative computation of the impact of COVID-19:
  - Approach:
    - Re-compute financial vulnerability measures using post-shock income y_h(ε), assuming debt and other expenditures remain constant.
  - Results (table of shock scenarios):
    - Value of income shock, ε: 0 (baseline) → Share of households with FMR < 0 (%) = 19.00
    - Value of income shock, ε: 0.25 → Share of households with FMR < 0 (%) = 30.67
    - Value of income shock, ε: 0.5 → Share of households with FMR < 0 (%) = 46.21
    - Differences relative to baseline:
      - For ε = 0.25: Difference = 11.67
      - For ε = 0.5: Difference = 27.21
  - Footnote on expenditure elasticity:
    - While households likely adjust expenditures after an income shock, elasticity of expenditures with respect to income for most households likely small: 66 percent of total yearly expenditure of the median household in sample consists of food and other daily essential items.
  - Figure references:
    - Figure B.1: Loan Age and Predicted Duration (figure referenced).
    - Figure B.2: FMR Density After COVID-19 Income Shock — note: FMR is truncated at -1 for plotting; series shown include Predicted FMR, CF FMR 25% Shock, CF FMR 50% Shock.

### Appendix C: Details for Nepal
- Household-level descriptive statistics and shock breakdowns:
  - Table C.1: Household Descriptive Statistics for Nepal’s Household Risk and Vulnerability Survey (HRVS) (table referenced).
  - Table C.2: Types of Shocks Faced by Households in Nepal’s HRVS (table referenced).
  - Table C.3: Household ex-ante coping strategies responding to hypothetical shock of NRs25,000 (approx. US$210); households could select multiple responses:
    - Most anticipated coping strategies: relying on savings and borrowing from friends/family.
    - Very few households anticipated using formal financial loans, indicating minimal financial access.

*Household Vulnerability to Income Shocks in Emerging and Developing Asia: The Case of Cambodia, Nepal and Vietnam — Working Paper No. WP/2022/064*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022064-print-pdf.pdf_