## wpiea2020138-print-pdf

## Source details

**Canonical URL:** [wpiea2020138-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020138-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020138-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020138-print-pdf.pdf.json)

---

### I. Introduction — access to finance and mobile money in India
- Only 20 percent of adults in India save with a financial institution (World Bank Global Findex Database (2017)).
- Within the population possessing a bank account, 48.5 percent of the accounts remain inactive (2017 survey).
- About 39 percent of survey respondents reported sending or receiving domestic remittances using a financial institution; the bulk of remittance transfers are conducted with cash.
- Paytm: largest mobile money payments firm in India since 2010, serving over 400 million users and 14 million businesses as of 2019 (BusinessWorld, 2019).
- Two research questions:
  - Does mobile money improve resilience to economic shocks by enabling cheaper and more efficient saving and transfers?
  - Can adoption of low-cost payment technology increase sales of micro and small enterprises by reducing payment frictions and costs?

### Mobile money adoption surge during demonetization
- Demonetization enacted in November 2016 unexpectedly withdrew major banknotes and constrained cash use.
- Paytm transaction volumes show a large spike immediately following the policy announcement.
- Growth of debit card transactions around October-December 2016 was 129 percent (RBI (2017 Feb)).
- Paytm average growth rate of monthly transaction volume in the six months prior to demonetization was 52 percent.
- Transaction-volume level attained during demonetization was broadly sustained through 2017 and 2018.
- Surge used as experimental setting to compare periods with and without mobile money in close proximity.

### Mechanisms: risk-sharing and remittance channels
- Theoretical background:
  - Under complete markets, outcomes respond only to aggregate shocks (Cochrane, 1991; Mace, 1991; Townsend, 1994); developing-economy frictions prevent complete risk-sharing.
- Mobile money functions:
  - Reduce transaction costs for transferring resources and serve as storage (Jack and Suri, 2014; Morawczynski and Pickens, 2009).
- Prior empirical evidence:
  - Jack and Suri (2014): welfare benefits of on average 3 to 4 percent of income via improved risk-sharing.
  - Mbiti and Weil (2015) and Wieser and others (2019): mobile money lowers use of informal savings and increases remittance transactions.
  - Aron (2018): cautions micro-studies may misjudge economy-wide effects due to spillovers and measurement error.
- Paper contribution: use large-scale Paytm transaction data (nearly half a billion users) and district-wise P2P split to explicitly test remittance channel for insurance against shocks.

### Main empirical risk-sharing result
- Rainfall shocks have a significant negative impact on economic activity proxied by nighttime lights, reducing it by 23 percent on average.
- A 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3 percent.

### Data on transactions and firm survey — scope and key features
- Paytm transaction data:
  - Monthly users and transactions at the district level for 643 districts, May 2016 – April 2019.
  - Transactions include offline payments and peer-to-peer transfers, disaggregated by value and volume.
  - Payments data disaggregated by payments to formal and informal sector firms.
  - Peer-to-peer transfers disaggregated by within-district and across-district transfers.
  - Note: a total of 654 districts used, largely consistent with the 2011 Census of India.
  - Paytm data exclude online transactions through websites or online retail platforms; all other mobile-phone conducted transactions in physical shops are included as offline transactions.
  - Peer-to-peer transactions data exclude a subset made through the unified payments interface which Paytm joined beginning November 2017; P2P analysis run on data before November 2017 to ensure comparability.
- Firm survey around Paytm intervention:
  - Total surveyed firms: 3,046.
    - Treatment group (targeted January 2019): 1,417 firms.
    - Control group (targeted July 2019): 1,629 firms.
  - Firms reporting sales for both reference months: 925 firms.
  - Timing anchoring:
    - Treatment recall/current: February (recall)/August (current) 2019.
    - Control recall/current: January (recall)/July (current) 2019.
  - Survey elicited current and six-month-ago total monthly sales (recall), number of employees, business category, bank account access, loan amounts, and subjective expectations (triangular distribution fitted to derive mean and standard deviation).

### Auxiliary datasets and measurement choices
- Nighttime lights:
  - Data from Earth Observations Group (EOG) using VIIRS Day/Night Band (DNB); satellites collect images twice a day at 15 arc-second resolution (1km grid), covering 75° North to 65° South latitude.
  - DNB data filtered to exclude stray light, lightning, lunar illumination, cloud-cover (VIIRS Cloud Mask product (VCM)); edge-of-swath zones 29-32 excluded.
  - Each 1 sq. km grid assigned pixel radiance measured in W/cm^2; district luminosity obtained by summing lights over gridded area using administrative boundaries.
  - District-wise luminosity time-series detrended to account for monthly seasonality.
- PLFS 2017-18 used to corroborate luminosity measure for labor supply hours and per-capita expenditure on NSS region level for 4 quarters (July 2017 - June 2018).
- Rainfall shocks:
  - Meteorological data from Indian Meteorological Department (IMD) based on about 3500 stations compiled into district-level rainfall statistics.
  - IMD provides long-term rainfall averages for each district using rainfall records for the period from 1951-.

### Data and measurement (additional)
- Rainfall shock indicator:
  - Binary = 1 if a district’s monthly rainfall deviation from its long term average is at least 1.5 standard deviations of the cross-sectional distribution (roughly 100mm).
  - Alternative continuous measure: district month-wise deviations from long-term average (robustness check).
- Bank availability: quarterly district-level number of reporting/functioning offices from the Reserve Bank of India (RBI).
- District-level covariates: Census 2011 (share of rural households, literacy rate, share of unemployed/casual workers, mobile users, under-30 age), DHS 2015 (share of households with bank accounts, average wealth index), Economic Census 2013 (firm informality index).
- Night-time lights variable Y_it: demeaned log of the sum of night-time lights within each district.
- Peer-to-peer transfers: Paytm P2P volumes used to capture within- and across-district remittance intensity.

### Empirical strategy — risk-sharing specification and identification
- Main specification:
  - Y_it = β S_it + γ M_it + δ S_it·M_it + χ X_it + α_i + η_t + e_it
    - S_it = binary rainfall shock; M_it = intensity of Paytm use (total users or peer-to-peer transfers).
    - Hypothesis: β < 0; δ > 0 (mobile money mitigates negative shock effect).
- Controls:
  - District fixed effects α_i and time effects η_t (year-by-month or separate year and month).
  - Time-varying bank availability included in X_it.
- Regression discontinuity (RD) around demonetization augments with district-specific polynomial time trend P(t)·α_i and post-demonetization dummy.
- Placebo test: impute post-demonetization district averages of mobile money use into pre-demonetization period (δ_Pre).
- Identification concerns:
  - Rainfall shocks assumed fully exogenous.
  - Potential endogeneity addressed by exploiting sudden demonetization adoption spike, RD design with narrow window, placebo pre-period imputation, and instrumental-variable (fuzzy RDD) approach.

### Effects on firms — difference-in-differences and IV
- DiD specification:
  - Sales_it = α·Treat_i + γ·Post_t + β·Treat_i·Post_t + e_it
  - Treat_i·Post_t interaction gives relative six-month sales differential.
- Identification checks:
  - Add firm location fixed effects τ_i and business-type fixed effects b_i.
  - Subset check: firms that never adopt despite targeting show no sales differential.
- Outcomes on expectations:
  - Triangular distribution fitted to subjective probabilities used to derive mean (subjective expectation) and standard deviation (subjective uncertainty).

### Results — who uses mobile money?
- Correlates of adoption (end-2018 standardized effects):
  - Bank availability: one standard deviation increase in number of bank branches associated with a 0.2 to 0.3 standard deviation increase in consumer and firm adoption.
  - District mobile users strongly predict adoption.
  - Higher adoption in districts with younger population, higher wealth, and more urbanization.
  - Higher adoption in areas with larger proportion of unemployed and casual workers.
  - Firm informality: one standard deviation increase in district informality increases take-up by 0.1 standard deviation.

### Results — effects of mobile money on risk-sharing (aggregate economic activity)
- Baseline (Table 1) exact coefficients reported:
  - Rainfall Shock: -0.169 ∗∗∗ (column 1), -0.169 ∗∗∗ (column 2), -0.184 ∗∗∗ (column 3).
  - Mobile Money: 0.014 ∗∗∗ (column 1), 0.017 ∗∗∗ (column 2), -0.006 (column 3).
  - Mobile Money × Rainfall Shock: 0.030 ∗∗∗ (column 1), 0.030 ∗∗∗ (column 2), 0.031 ∗∗∗ (column 3).
  - Observations: 188271 (col 1), 860218 (col 2), 6028602 (col 3).
  - r2: 0.011 (col 1), 0.012 (col 2), 0.071 (col 3).
- Interpretation from text:
  - Rainfall shock reduces economic activity by 17 percent on average.
  - A 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3 percent.
- Marginal effects (Figure 4, described):
  - Lower tenth percentile district: reduces negative effect of rainfall shock from 18 percent to 16 percent.
  - Median-value district: reduces negative effect from 18 percent to 1 percent.

### Robustness and placebo tests (exact reported figures)
- Placebo and RD (Table 2) exact coefficients:
  - Pre-period Rainfall Shock: -0.209 ∗∗∗ (col 1).
  - Post-period Rainfall Shock: -0.146 ∗∗∗ (col 2).
  - RDD Rainfall Shock: -0.204 ∗∗∗ (col 3).
  - RDD + controls Rainfall Shock: -0.116 ∗∗ (col 4).
  - Mobile Money × Rainfall Shock in placebo: 0.008 (col 1).
  - Mobile Money × Rainfall Shock in other specs: 0.025 ∗∗∗ (col 2), 0.032 ∗∗∗ (col 3), 0.026 ∗∗∗ (col 4).
  - Observations: 362614 (col 1), 976723 (col 2), 972397239? (col 3), 72397239 (col 4).
  - r2: 0.043 (col 1), 0.070 (col 2), 0.366 (col 3), 0.366 (col 4).
- Notes:
  - District-specific polynomial time trends and control for post-demonetization period included.
  - Including bank availability × shock interaction (col 4) does not overturn results.

### Channels — peer-to-peer transfers and across-district remittances
- Table 3 exact coefficients:
  - Column (1) Users: Rainfall Shock -0.204 ∗∗∗; Mobile Money × Rainfall Shock 0.032 ∗∗∗.
  - Column (2) P2P (all transfers): Rainfall Shock -0.283 ∗∗∗; Mobile Money × Rainfall Shock 0.017 ∗∗∗.
  - Column (3) Cross-district P2P: Rainfall Shock -0.237 ∗∗∗; Mobile Money × Rainfall Shock 0.014 ∗∗∗.
  - Observations (cols 1–3): 72397239, 72397239, 72397239.
  - r2: 0.368 (col 1), 0.366 (col 2), 0.366 (col 3).
- Interpretation:
  - Across-district P2P transfers mitigate negative effects of rainfall shocks, supporting a remittance channel.
  - Magnitude of P2P-based mitigation is smaller (roughly half) compared to total-user based effect, suggesting additional channels (e.g., mobile wallet savings) also contribute.
  - Across-district money transfers increase following a rainfall shock.

### Household consumption and regional results
- Sample period: 2016 May - 2019 July; mobile money transactions accounted for 30.4 percent of the total value for this sample period.
- Aggregation note: The average number of districts within a given region is 7.3.
- Effect on HH Per-Capita Consumption (Table 4 exact coefficients):
  - Column (1) Average:
    - Rainfall Shock: -0.090 ∗∗∗ (0.032)
    - Mobile Money: -0.025 ∗ (0.014)
    - Mobile Money × Rainfall Shock: 0.009 ∗∗∗ (0.003)
    - Bank Availability: 0.072 (0.045)
    - Observations: 325
    - r2: 0.029
  - Column (2) Rural:
    - Rainfall Shock: -0.095 ∗ (0.051)
    - Mobile Money: -0.025 (0.018)
    - Mobile Money × Rainfall Shock: 0.011 ∗∗ (0.004)
    - Bank Availability: -0.098 (0.063)
    - Observations: 325
    - r2: 0.025
  - Column (3) Urban:
    - Rainfall Shock: -0.033 (0.036)
    - Mobile Money: -0.023 (0.015)
    - Mobile Money × Rainfall Shock: 0.003 (0.003)
    - Bank Availability: 0.098 ∗∗ (0.045)
    - Observations: 325
    - r2: 0.015
- Interpretation:
  - Mitigation effects concentrated in rural areas where bank access is low.
  - Robustness: restricting to last quarter of 2018 still shows a mitigation effect: a 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3.8 percent.

### Effects of Mobile Payment Technology on Firm Sales — key estimates
- Baseline balance (Table 5) sample means:
  - # Employees: Control 4.92; Treatment 3.05; Difference 1.86; P-value 0.30.
  - Bank Account: Control 0.72; Treatment 0.70; Difference 0.02; P-value 0.48.
  - Firm Age: Control 6.72; Treatment 6.28; Difference 0.43; P-value 0.38.
  - Business Loss (thousands of Rs.): Control 22.96; Treatment 19.31; Difference 3.27; P-value 0.71.
- Intent-to-treat sales effects (Table 6: Change in Sales (log), 6 month sales growth differential):
  - Mobile Technology Firms:
    - (1) 0.331 ∗∗∗ (0.028)
    - (2) 0.308 ∗∗∗ (0.030)
    - (3) 0.234 ∗∗∗ (0.053)
    - (4) 0.226 ∗∗∗ (0.063)
    - (5) 0.276 ∗∗∗ (0.081)
  - Observations: 804, 791, 791, 791, 791 respectively.
  - r2: 0.035, 0.050, 0.211, 0.232, 0.485 respectively.
- Interpretation:
  - Six-month sales improvement for treatment relative to control: approximately 33 percent in baseline specification; most robust specification finds 28 percent.
- Propensity-score matched DID (Table 7 Mobile Technology × 6 months Post):
  - Column (1) match by covariates: 0.312 ∗∗∗ (0.077); Observations: 1172.
  - Column (2) match by location: 0.257 ∗∗∗ (0.095); Observations: 923.
  - Column (3) match by business type by location: 0.282 ∗∗ (0.118); Observations: 73738 (as reported).
- Common trends check (never adopters, Table 8):
  - Mobile Technology × 6 months Post: 0.131 (0.180); Observations: 106.
- Firm expectations, uncertainty, borrowing (Table 9):
  - Future sales growth: -0.120 ∗∗∗ (0.036); Observations: 712.
  - Uncertainty (log s.d. future sales): -0.226 ∗∗ (0.102); Observations: 699.
  - Bank loan (Yes/No): 0.190 ∗∗ (0.096); Observations: 603.
  - Interpretation: Paytm-using firms are more optimistic and less uncertain about future sales and more likely to take a loan during the six-month treatment period.
- IV average treatment effects (Table 10 — Mobile Technology Use (% of sales) on 6 month sales growth differential):
  - (1) 0.041 ∗∗∗ (0.009); First-Stage F-Statistic 48.64; Observations: 780; r2: 0.039.
  - (2) 0.040 ∗∗∗ (0.011); First-Stage F-Statistic 6.93; Observations: 780; r2: 0.067.
  - (3) 0.043 ∗∗∗ (0.014); First-Stage F-Statistic 8.9; Observations: 780; r2: 0.337.
  - Interpretation: Increasing the intensity of mobile money use by 1 percent (higher Paytm sales as a percentage of total sales) is associated with a 4.2 percent sales growth differential over six months (reported text).

### Displacement, externalities, and usage intensity
- Average proportion of treated firms in each location: 25 percent with a standard deviation of 38 percent.
- Intensity of Paytm use averages 5 percent of treatment firms’ total sales, range 0 to 71 percent.
- Intent-to-treat estimates understate local average treatment intensities unless IV used.

### Robustness, identification, and limitations (selected)
- Identification strategies:
  - Difference-in-differences exploiting phased targeting.
  - Propensity-score matching with kernel weights following Heckman, Ichimura, and Todd (1997, 1998).
  - Instrumental-variable two-stage least squares using intervention as instrument for actual adoption intensity.
- Limitations:
  - Survey did not collect pre-intervention monthly sales series; common trends tested using never-adopter subset.
  - Some baseline differences may be underpowered given sample size (925 firms).
  - Intensity of Paytm use varies (average 5 percent; range 0–71 percent), affecting interpretation of ITT estimates.

### Conclusion and policy-relevant implications
- Two distinct use cases analyzed:
  - Remittances/payments improving household resilience to shocks via cheaper and more efficient saving and transfers.
  - Mobile-based payment technology increasing sales of micro and small enterprises by reducing frictions and costs of payments.
- Summative findings:
  - Mobile money use meaningfully reduces negative impacts of rainfall shocks on consumption, especially in rural areas.
  - Firms adopting payment technology see robust six-month sales improvements of approximately 26 percent (paper summary) to 28–33 percent in various specifications; IV estimates imply a 4.2 percent sales growth differential per 1 percentage point increase in mobile payment share of sales.
- Open policy questions:
  - Sustainability of impacts in the longer run as cash dependence and informality potentially ease and regulatory/competitive dynamics evolve.
  - Expansion of mobile money use-cases beyond payments and P2P transfers may enhance benefits.
  - Broader fintech impacts include alleviation of credit frictions, improved financial decision making, and reduced barriers to financial intermediation.

### Key numeric findings and parameters (preserved)
- Rainfall shock binary threshold: 1.5 standard deviations (≈ 100mm).
- Main estimated impacts (selected exact coefficients):
  - Rainfall Shock baseline: -0.169 ∗∗∗; alternative -0.184 ∗∗∗.
  - Mobile Money × Rainfall Shock: 0.030 ∗∗∗; alternative estimates 0.031 ∗∗∗, 0.032 ∗∗∗, 0.014 ∗∗∗, 0.017 ∗∗∗, 0.026 ∗∗∗.
  - A 10 percent increase in mobile money use reduces shock effect by 3 percent (text interpretation).
  - Aggregate shock reduces economic activity by 17 percent on average.
- Sample sizes and observations preserved as reported:
  - Paytm district panel: 643 districts, May 2016 – April 2019 (note: total of 654 districts used in a district list).
  - Firm survey: 3,046 firms (1,417 treatment; 1,629 control; 925 reporting both sales months).
  - Paytm transactions accounted for 30.4 percent of the total value for the sample period 2016 May - 2019 July.

*Source: wpiea2020138-print-pdf - References and selected empirical sections.*

### References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### References

### I. Introduction — access to finance and mobile money in India
- Only 20 percent of adults in India save with a financial institution (World Bank Global Findex Database (2017)).
- Within the population possessing a bank account, 48.5 percent of the accounts remain inactive, making India the country with the highest inactivity rate in the world in the 2017 survey.
- About 39 percent of survey respondents reported sending or receiving domestic remittances using a financial institution; the bulk of remittance transfers are conducted with cash.
- Paytm: largest mobile money payments firm in India since 2010, serving over 400 million users and 14 million businesses as of 2019 (BusinessWorld, 2019).
- Two research questions examined using large-scale monthly mobile money transaction data (nearly half a billion users):
  - Does mobile money improve resilience to economic shocks by enabling cheaper and more efficient saving and transfers?
  - Can adoption of low-cost payment technology increase sales of micro and small enterprises by reducing payment frictions and costs?

### Mobile money adoption surge during demonetization
- Demonetization policy episode: enacted in November 2016, unexpectedly withdrew major banknotes and constrained cash use.
- Paytm transaction volumes show a large spike immediately following the policy announcement (Figure 1).
- Growth of debit card transactions around October-December 2016 was 129 percent (RBI (2017 Feb)).
- Paytm average growth rate of monthly transaction volume in the six months prior to demonetization was 52 percent.
- Transaction-volume level attained during demonetization was broadly sustained through 2017 and 2018.
- The surge provides an experimental setting comparing periods with and without mobile money in close proximity.

### Mechanisms: risk-sharing and remittance channels
- Risk-sharing implication: under complete markets, individual or regional outcomes should respond only to aggregate economy-wide shocks (Cochrane, 1991; Mace, 1991; Townsend, 1994); developing-economy frictions often prevent complete risk-sharing.
- Mobile money can reduce transaction costs for transferring resources and serve as storage (Jack and Suri, 2014; Morawczynski and Pickens, 2009).
- Evidence on welfare: Jack and Suri (2014) find mobile money provides welfare benefits of on average 3 to 4 percent of income via improved risk-sharing.
- Other supporting findings:
  - Mbiti and Weil (2015) and Wieser and others (2019) show mobile money lowers use of informal savings and increases remittance transactions.
  - Aron (2018) reviews empirical evidence and cautions that micro-studies may misjudge economy-wide effects due to spillovers and measurement error.
- Contribution of the paper: use large-scale Paytm transaction data to measure both number of users and transactions precisely, and district-wise peer-to-peer transactions split within and across districts to explicitly test the remittance channel for insurance against shocks.

### Main empirical risk-sharing result
- Rainfall shocks have a significant negative impact on economic activity proxied by nighttime lights, reducing it by 23 percent on average.
- A 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3 percent.

### Firm-level payments adoption, channels, and expected effects
- Channels by which electronic payments can increase firm sales:
  - Avoiding missed sales when customers face high inconvenience of cash or card refusal/surcharge (Bourguignon, Gomes, and Tirole (2019); Bolt, Jonker, and Van Renselaar (2010); Chakravorti and To (2007)).
  - Improved transaction efficiency (QR-code based payments) promoting demand (Aggarwal, Brailovskaya, and Robinson (2019)).
  - Reduced risky cash-holdings and increased access to credit improving firm performance (Beck and others (2018)).
- China example: mobile payments for consumption reached RMB 14.5 trillion (or 16 percent of GDP) in 2017; Alipay and WeChat Pay have over 500 million and 900 million monthly active users, respectively (Frost and others, 2019).
- Paytm’s QR-code payment function: enables immediate payments by generating/scanning QR codes on the mobile app.

### Empirical firm-level design and results
- Identification exploits a phased targeting intervention by Paytm incentivizing firms to adopt the mobile app; sequencing used to define treatment and control firms.
- Difference-in-difference strategy comparing firms with six months of Paytm experience to non-Paytm firms.
- Main finding: Paytm-using firms improve their sales by approximately 26 percent relative to non-Paytm firms after six months of use.
- Survey evidence: Paytm-using firms report lower subjective uncertainty around future sales.
- Results robust to matching on a large vector of location and other characteristics and consistent with volatility-reducing effects at both firm and household levels.
- Related evidence: Dalton and others (2019) find adopting firms have better access to finance and reduced sales volatility; Aggarwal, Brailovskaya, and Robinson (2019) find mobile-payment adoption raises saving behavior and likelihood of extending credit to consumers.

### II. Data on transactions and firm survey — scope and key features
- Paytm transaction data: monthly users and transactions at the district level for 643 districts, May 2016 – April 2019.
- Transactions include offline payments and peer-to-peer transfers, disaggregated by value and volume.
- Payments data further disaggregated by payments to formal and informal sector firms.
- Peer-to-peer transfers further disaggregated by within-district and across-district transfers, enabling identification of remittance transfers from other districts.
- Note on district list: a total of 654 districts used, largely consistent with the 2011 Census of India.
- Paytm data exclude online transactions through websites or online retail platforms; all other mobile-phone conducted transactions in physical shops are included as offline transactions.
- Peer-to-peer transactions data exclude a subset made through the unified payments interface which Paytm joined beginning November 2017; P2P analysis run on data before November 2017 to ensure comparability.

### Auxiliary datasets and measurement choices
- Nighttime lights:
  - Used as proxy for district-level monthly economic activity (Henderson, Storeygard, and Weil (2011); Chen and Nordhaus (2011); Kulkarni and others (2011); Alesina, Michalopoulos, and Papaioannou (2016)).
  - Data from Earth Observations Group (EOG) using VIIRS Day/Night Band (DNB); satellites collect complete earth images twice a day at 15 arc-second resolution (1km grid interval), covering 75° North to 65° South latitude.
  - DNB data filtered to exclude stray light, lightning, lunar illumination, cloud-cover (VIIRS Cloud Mask product (VCM)); edge-of-swath data (aggregation zones 29-32) excluded.
  - Each 1 sq. km grid assigned a pixel radiance value measured in W/cm^2; district luminosity obtained by summing lights over gridded area using administrative boundaries.
  - District-wise luminosity time-series detrended to account for monthly seasonality.
  - Rationale: monthly district GDP panel does not exist; luminosity correlates strongly with standard economic outcomes; identification focuses on changes rather than levels.
- Periodic Labor Force Survey (PLFS) 2017-18 used to corroborate luminosity measure:
  - Provides labor market measures in urban and rural areas.
  - Rotational panel sampling design allowed calculation of average labor supply hours and per-capita expenditure on NSS region level for 4 quarters (July 2017 - June 2018).
- Rainfall shocks:
  - Meteorological data from Indian Meteorological Department (IMD) based on about 3500 stations compiled into district-level rainfall statistics.
  - IMD provides long-term rainfall averages for each district using rainfall records for the period from 1951- [text ends].

*Source: wpiea2020138-print-pdf - References (https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020138-print-pdf.pdf)*

### 2000. We take the amount of deviation of each district’s current recorded level of rainfall in

### wpiea2020138-print-pdf - 2000

### Data and measurement
- Rainfall shock indicator: binary variable = 1 if a district’s monthly rainfall deviation from its long term average is at least 1.5 standard deviations of the cross-sectional distribution (roughly 100mm of rainfall); captures conditions of drought or flood within a district.
- Alternative continuous measure: district month-wise deviations from long-term average (robustness check).
- Bank availability: quarterly district-level number of reporting/ functioning offices from the Reserve Bank of India (RBI).
- District-level covariates: from Census 2011 (share of rural households, literacy rate, share of unemployed/casual workers, mobile users, under-30 age), DHS 2015 (share of households with bank accounts, average wealth index), Economic Census 2013 (firm informality index).
- Night-time lights: Y_it, demeaned log of the sum of night-time lights within each district used as a proxy for economic activity.
- Peer-to-peer transfers: Paytm peer-to-peer transaction volumes used to capture within- and across-district remittance intensity.

### Firm survey around Paytm intervention
- Total surveyed firms: 3,046.
  - Treatment group (targeted January 2019): 1,417 firms.
  - Control group (targeted July 2019): 1,629 firms.
- Firms reporting sales for both reference months: 925 firms.
- Timing anchoring for recall/current sales:
  - Treatment group anchoring months: February (recall)/August (current) 2019.
  - Control group anchoring months: January (recall)/July (current) 2019.
- Survey elicited:
  - Current and six-month-ago total monthly sales (recall).
  - Number of employees, business category, bank account access, loan amounts.
  - Subjective expectations: minimum and maximum monthly sales one year ahead and probability (0-10 scale) that sales will be at least the midpoint; triangular distribution fitted to derive mean (subjective expectation) and standard deviation (subjective uncertainty).

### Empirical strategy — risk-sharing specification
- Main specification (equation (1)):
  - Y_it = β S_it + γ M_it + δ S_it·M_it + χ X_it + α_i + η_t + e_it
  - S_it = binary rainfall shock; M_it = intensity of Paytm use (total users or peer-to-peer transfers).
  - Hypothesis: β < 0; δ > 0 (mobile money mitigates negative shock effect).
- Controls and fixed effects:
  - District fixed effects α_i and time effects η_t (year-by-month or separate year and month).
  - Time-varying bank availability included in X_it.
- Regression discontinuity (RD) around demonetization (equation (2)):
  - Augment with district-specific polynomial time trend P(t)·α_i and include post-demonetization dummy in η_t to exploit abrupt Paytm adoption surge.
- Placebo test (equation (3)):
  - Impute post-demonetization district averages of mobile money use (M̄_Post_i) into the pre-demonetization period to test for pre-existing differences (δ_Pre).

### Identification concerns and strategies
- Main identification: rainfall shocks assumed fully exogenous (timing and spatial occurrence orthogonal to mobile money use).
- Potential endogeneity: districts with existing strong risk-sharing may have higher mobile money adoption; addressed by:
  - Exploiting sudden adoption spike around demonetization as quasi-experiment.
  - RD design with narrow window around policy to compare observations just before and after announcement.
  - Placebo (pre-period imputation) to test for pre-existing correlations.
  - Instrumental variable (fuzzy RDD) approach discussed: instrument S_it·M_it with post-demonetization interaction identifying rainfall shocks in the post period.

### Effects on firms (difference-in-differences)
- Treatment design:
  - Firms targeted January 2019 (≈ six months experience with Paytm) vs firms targeted July 2019 (no Paytm at survey).
- DiD specification (equation (4)):
  - Sales_it = α·Treat_i + γ·Post_t + β·Treat_i·Post_t + e_it
  - Treat_i·Post_t interaction gives relative six-month sales differential.
- Identifying assumption: conditional common trends; validated by:
  - Adding firm location fixed effects τ_i and business-type fixed effects b_i (equation (5)).
  - Checking subset of firms that never adopt despite targeting: expect no sales differential for never-adopters.
- Outcomes on expectations:
  - Use fitted bi-triangular distributions from subjective probability elicitation to derive first two moments; test impact of adoption on expected sales growth and subjective uncertainty.

### Results — who uses mobile money?
- Major correlates of adoption (end-2018 standardized effects):
  - Bank availability: one standard deviation increase in number of bank branches associated with a 0.2 to 0.3 standard deviation increase in consumer and firm adoption.
  - District mobile users strongly predict adoption.
  - Higher adoption in districts with younger population, higher wealth, and more urbanization.
  - Higher adoption in areas with larger proportion of unemployed and casual workers (appeal among unbanked).
  - Firm informality: one standard deviation increase in district informality increases take-up by 0.1 standard deviation.

### Results — effects of mobile money on risk-sharing (aggregate economic activity)
- Baseline (Table 1, equation (1) results):
  - Rainfall Shock: -0.169 ∗∗∗ (column 1), -0.169 ∗∗∗ (column 2), -0.184 ∗∗∗ (column 3). (Standard errors in parentheses.)
  - Mobile Money: 0.014 ∗∗∗ (column 1), 0.017 ∗∗∗ (column 2), -0.006 (column 3).
  - Mobile Money × Rainfall Shock: 0.030 ∗∗∗ (column 1), 0.030 ∗∗∗ (column 2), 0.031 ∗∗∗ (column 3).
  - Observations: 188271 (col 1), 860218 (col 2), 6028602 (col 3) — preserve reported values exactly as in table.
  - r2: 0.011 (col 1), 0.012 (col 2), 0.071 (col 3).
- Interpretation:
  - Rainfall shock reduces economic activity by 17 percent on average.
  - A 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3 percent.
- Marginal effects (Figure 4):
  - Lower tenth percentile district: reduces negative effect of rainfall shock from 18 percent to 16 percent.
  - Median-value district: reduces negative effect from 18 percent to 1 percent.

### Robustness and placebo tests
- Placebo and RD (Table 2):
  - Pre-period Rainfall Shock: -0.209 ∗∗∗ (col 1).
  - Post-period Rainfall Shock: -0.146 ∗∗∗ (col 2).
  - RDD Rainfall Shock: -0.204 ∗∗∗ (col 3).
  - RDD + controls Rainfall Shock: -0.116 ∗∗ (col 4).
  - Mobile Money × Rainfall Shock in placebo: 0.008 (col 1) — insignificant and reduced in magnitude.
  - Mobile Money × Rainfall Shock in other specs: 0.025 ∗∗∗ (col 2), 0.032 ∗∗∗ (col 3), 0.026 ∗∗∗ (col 4).
  - Observations: 362614 (col 1), 976723 (col 2), 972397239? (col 3), 72397239 (col 4) — preserve reported observation figures exactly as printed.
  - r2: 0.043 (col 1), 0.070 (col 2), 0.366 (col 3), 0.366 (col 4).
- Notes on RD specification:
  - District-specific polynomial time trends and control for post-demonetization period included.
  - Including bank availability × shock interaction (col 4) does not overturn results.

### Channels — peer-to-peer transfers and across-district remittances (Table 3)
- Column (1) Users result (reproduced): Rainfall Shock -0.204 ∗∗∗; Mobile Money × Rainfall Shock 0.032 ∗∗∗.
- Column (2) P2P (all transfers):
  - Rainfall Shock: -0.283 ∗∗∗.
  - Mobile Money × Rainfall Shock: 0.017 ∗∗∗.
- Column (3) Cross-district P2P:
  - Rainfall Shock: -0.237 ∗∗∗.
  - Mobile Money × Rainfall Shock: 0.014 ∗∗∗.
- Observations (columns 1–3): 72397239 (col 1), 72397239 (col 2), 72397239 (col 3) — preserve exact reported values.
- r2: 0.368 (col 1), 0.366 (col 2), 0.366 (col 3).
- Interpretation:
  - Across-district peer-to-peer transfers mitigate negative effects of rainfall shocks, supporting a remittance channel.
  - Magnitude of P2P-based mitigation is smaller (roughly half) compared to total-user based effect, suggesting additional channels (e.g., mobile wallet savings) also contribute.
- Empirical evidence of channel: across-district money transfers increase following a rainfall shock.

### Household consumption and regional results
- Linked PLFS 2017-2018 quarter-wise average per-capita consumption to district rainfall shocks and mobile money use.
- Finding: negative effect of rainfall shock is mitigated by region’s intensity of mobile money use; effects concentrated on consumption measures (details and exact coefficients not provided in supplied excerpt).

### Key numeric findings and parameters (preserved)
- Rainfall shock binary threshold: 1.5 standard deviations (≈ 100mm).
- Main estimated impacts (selected exact coefficients):
  - Rainfall Shock baseline: -0.169 ∗∗∗; alternative -0.184 ∗∗∗.
  - Mobile Money × Rainfall Shock: 0.030 ∗∗∗; alternative estimates 0.031 ∗∗∗, 0.032 ∗∗∗, 0.014 ∗∗∗, 0.017 ∗∗∗, 0.026 ∗∗∗.
  - A 10 percent increase in mobile money use reduces shock effect by 3 percent (interpretation from text).
  - Aggregate shock reduces economic activity by 17 percent on average.
- Survey sample sizes: 3,046 firms total; 1,417 treatment; 1,629 control; 925 firms reported both sales months.

*Source: wpiea2020138-print-pdf - 2000 (sections on data, firm survey, empirical strategy, and results).*

### 30.4 percent of the total value for our sample period of 2016 May - 2019 July.

### wpiea2020138-print-pdf - 30.4 percent of the total value for our sample period of 2016 May - 2019 July.

### Risk-sharing and Household Consumption
- Sample period: 2016 May - 2019 July; mobile money transactions accounted for 30.4 percent of the total value for this sample period.
- Aggregation note: A region is at a higher level of aggregation compared to a district. The average number of districts within a given region is 7.3.
- Main finding: Mobile money adoption strengthens risk sharing, with effects concentrated in rural areas where bank access is low.

Key regression estimates (Table 4: Effect of Mobile Money on Risk Sharing: Labor Supply and Consumption; dependent variable: HH Per-Capita Consumption):
- Column (1) Average:
  - Rainfall Shock: -0.090 ∗∗∗ (0.032)
  - Mobile Money: -0.025 ∗ (0.014)
  - Mobile Money × Rainfall Shock: 0.009 ∗∗∗ (0.003)
  - Bank Availability: 0.072 (0.045)
  - Observations: 325
  - r2: 0.029
- Column (2) Rural:
  - Rainfall Shock: -0.095 ∗ (0.051)
  - Mobile Money: -0.025 (0.018)
  - Mobile Money × Rainfall Shock: 0.011 ∗∗ (0.004)
  - Bank Availability: -0.098 (0.063)
  - Observations: 325
  - r2: 0.025
- Column (3) Urban:
  - Rainfall Shock: -0.033 (0.036)
  - Mobile Money: -0.023 (0.015)
  - Mobile Money × Rainfall Shock: 0.003 (0.003)
  - Bank Availability: 0.098 ∗∗ (0.045)
  - Observations: 325
  - r2: 0.015

Additional points on risk-sharing:
- Results indicate that neither rainfall shocks nor their interaction with mobile money use have significant effects on urban consumption, implying stronger risk-sharing benefits in rural settings.
- Robustness across time: Restricting baseline results to the last quarter of 2018 (post-demonetization dissipation) still shows a mitigation effect: a 10 percent increase in mobile money use in districts hit by a rainfall shock reduces the negative effect of the shock by 3.8 percent.

### Effects of Mobile Payment Technology on Firm Sales
- Experimental design: Targeted Paytm intervention with Treatment firms (targeted January 2019) and Control firms (targeted July 2019); treatment firms had approximately six months of experience accepting mobile payments at survey time.
- Baseline balance (Table 5: Balance of Firm Covariates; means for Control, Treatment, Difference, P-value):
  - # Employees: Control 4.92, Treatment 3.05, Difference 1.86, P-value 0.30
  - Bank Account: Control 0.72, Treatment 0.70, Difference 0.02, P-value 0.48
  - Firm Age: Control 6.72, Treatment 6.28, Difference 0.43, P-value 0.38
  - Business Loss (thousands of Rs.): Control 22.96, Treatment 19.31, Difference 3.27, P-value 0.71
- Sample composition: firms typically have 3-4 employees, an unbanked proportion of 30 percent, average firm age 6-7 years, experience business losses equivalent to approximately 20 percent of baseline sales.

Intent-to-treat sales effects (Table 6: Effect of Payment Technology on Firm Sales):
- Column (1) to (5) (Dep Var: Change in Sales (log), 6 month sales growth differential):
  - Mobile Technology Firms:
    - (1) 0.331 ∗∗∗ (0.028)
    - (2) 0.308 ∗∗∗ (0.030)
    - (3) 0.234 ∗∗∗ (0.053)
    - (4) 0.226 ∗∗∗ (0.063)
    - (5) 0.276 ∗∗∗ (0.081)
  - Observations: 804, 791, 791, 791, 791 respectively
  - r2: 0.035, 0.050, 0.211, 0.232, 0.485 respectively
- Interpretation:
  - Six-month sales improvement for treatment relative to control: approximately 33 percent in baseline specification; most robust specification finds 28 percent.

Displacement and externalities:
- Average proportion of treated firms in each location is 25 percent with a standard deviation of 38 percent.
- Coefficients remain broadly similar when including location and business-type fixed effects, suggesting small displacement effects and a positive net effect on firm sales.

Propensity-score matched DID estimates (Table 7):
- Mobile Technology × 6 months Post:
  - Column (1) match by covariates: 0.312 ∗∗∗ (0.077)
  - Column (2) match by location: 0.257 ∗∗∗ (0.095)
  - Column (3) match by business type by location: 0.282 ∗∗ (0.118)
- Observations: 1172, 923, 73738 (as reported)

Common trends check (Table 8: Evidence from Never Adopters):
- Never adopters (no experience with mobile payments):
  - Mobile Technology × 6 months Post: 0.131 (0.180)
  - Observations: 106
- Result: No statistically significant differential, consistent with common trends assumption.

Firm expectations, uncertainty, and borrowing (Table 9):
- Dependent variables: Future sales growth (growth rate), Uncertainty (log s.d. future sales), Bank loan (Yes/No)
- Mobile Technology Firms:
  - Future sales growth: -0.120 ∗∗∗ (0.036)
  - Uncertainty: -0.226 ∗∗ (0.102)
  - Bank loan (Yes/No): 0.190 ∗∗ (0.096)
- Observations: 712, 699, 603 respectively
- Interpretation: Paytm-using firms are more optimistic and less uncertain about future sales and show a higher likelihood of taking a loan during the six-month treatment period.

Average treatment effects via IV (Table 10):
- Mobile Technology Use (% of sales) — 6 month sales growth differential:
  - (1) 0.041 ∗∗∗ (0.009)
  - (2) 0.040 ∗∗∗ (0.011)
  - (3) 0.043 ∗∗∗ (0.014)
- First-Stage F-Statistic: 48.64, 6.93, 8.9 (reported across specifications)
- Observations: 780, 780, 780
- r2: 0.039, 0.067, 0.337
- Interpretation: Increasing the intensity of mobile money use by 1 percent (higher Paytm sales as a percentage of total sales) is associated with a 4.2 percent sales growth differential over six months (reported text: "4.2 percent sales growth differential over six-months").

### Robustness, Identification, and Limitations
- Identification strategies deployed:
  - Difference-in-differences exploiting phased targeting (intent-to-treat).
  - Propensity-score matching with kernel weights following Heckman, Ichimura, and Todd (1997, 1998).
  - Instrumental-variable two-stage least squares using intervention as instrument for actual adoption intensity.
- Limitations and caveats:
  - Survey did not collect pre-intervention monthly sales series; common trends tested using never-adopter subset.
  - Some baseline differences, while statistically insignificant, could be underpowered given sample size (925 firms); t-tests may be underpowered.
  - Intensity of Paytm use averages 5 percent of treatment firms’ total sales, range 0 to 71 percent, so intent-to-treat estimates understate local average treatment intensities unless IV used.

### Conclusion and Policy-relevant Implications
- Two distinct use cases of mobile money analyzed:
  - Remittances/payments improving household resilience to shocks via cheaper and more efficient saving and transfers.
  - Mobile-based payment technology increasing sales of micro and small enterprises by reducing frictions and costs of payments.
- Summative findings:
  - Mobile money use meaningfully reduces negative impacts of rainfall shocks on consumption, especially in rural areas.
  - Firms adopting payment technology see robust six-month sales improvements of approximately 26 percent (paper summary) to 28-33 percent in various specifications; IV estimates imply a 4.2 percent sales growth differential per 1 percentage point increase in mobile payment share of sales.
- Open questions for policy and future research:
  - Sustainability of impacts in the longer run as cash dependence and informality potentially ease and regulatory/competitive dynamics evolve.
  - Expansion of mobile money use-cases beyond payments and peer-to-peer transfers may enhance benefits.
  - Broader fintech impacts under study include alleviation of credit frictions, improved financial decision making, and reduced barriers to financial intermediation.

*Source: wpiea2020138-print-pdf - 30.4 percent of the total value for our sample period of 2016 May - 2019 July.*

### REFERENCES

### REFERENCES

### Citations
- Agarwal, Sumit, Shashwat Alok, Pulak Ghosh, Soumya Ghosh, Tomasz Piskorski, and Amit Seru, 2017, “Banking the unbanked: What do 255 million new bank accounts reveal about financial access?”Columbia Business School Research Paper, , No. 17-12.
- Aggarwal, Shilpa, Valentina Brailovskaya, and Jonathan Robinson, 2019, “Cashing In (and Out): Experimental Evidence on the Effects of Mobile Money in Malawi,”mimeo.
- Alesina, Alberto, Stelios Michalopoulos, and Elias Papaioannou, 2016, “Ethnic Inequality,” Journal of Political Economy, Vol. 124, No. 2, pp. 428–488.
- Altig, David, Jose Maria Barrero, Nicholas Bloom, Steven J Davis, Brent Meyer, and Nicholas Parker, 2019, “Surveying Business Uncertainty,”Federal Reserve Bank of Atlanta, Working Paper Series, Vol. 2019, No. 13.
- Aron, Janine, 2018, “Mobile money and the economy: a review of the evidence,”The World Bank Research Observer, Vol. 33, No. 2, pp. 135–188.
- Auffhammer, Maximilian, and Ryan Kellogg, 2011, “Clearing the air? The effects of gasoline content regulation on air quality,”American Economic Review, Vol. 101, No. 6, pp. 2687–2722.
- Beck, Thorsten, Haki Pamuk, Ravindra Ramrattan, and Burak R Uras, 2018, “Payment instruments, finance and development,”Journal of Development Economics, Vol. 133, pp. 162–186.
- Berg, Tobias, Valentin Burg, Ana Gombovi ́c, and Manju Puri, 2018, “On the rise of fintechs–credit scoring using digital footprints,” Techn. rep., National Bureau of Economic Re-search.
- Beyer, Robert, Esha Chhabra, Virgilio Galdo, and Martin Rama, 2018, “Measuring Districts’ Monthly Economic Activity from Outer Space,”World Bank Policy Research Working Paper, , No. 8523.
- Bhalotra, Sonia, 2010, “Fatal fluctuations? Cyclicality in infant mortality in India,”Journal of Development Economics, Vol. 93, No. 1, pp. 7–19.
- Bhandari, Laveesh, and Koel Roychowdhury, 2011, “Night lights and economic activity in India: A study using DMSP-OLS night time images,”Proceedings of the Asia-Pacific Advanced Network, Vol. 32, pp. 218–236.
- Bharadwaj, Prashant, William Jack, and Tavneet Suri, 2019, “Fintech and household resilience to shocks: Evidence from digital loans in Kenya,” Techn. rep., National Bureau of Economic Research.
- Bolt, Wilko, Nicole Jonker, and Corry Van Renselaar, 2010, “Incentives at the counter: An empirical analysis of surcharging card payments and payment behaviour in the Netherlands,”Journal of Banking & Finance, Vol. 34, No. 8, pp. 1738–1744.
- Bourguignon, Hélène, Renato Gomes, and Jean Tirole, 2019, “Shrouded transaction costs: must-take cards, discounts and surcharges,”International Journal of Industrial Organization, Vol. 63, pp. 99–144.
- Brodersen, Kay H, Fabian Gallusser, Jim Koehler, Nicolas Remy, Steven L Scott, and others, 2015, “Inferring causal impact using Bayesian structural time-series models,”The Annals of Applied Statistics, Vol. 9, No. 1, pp. 247–274.
- Burgess, Robin, and Rohini Pande, 2005, “Do rural banks matter? Evidence from the Indian social banking experiment,”American Economic Review, Vol. 95, No. 3, pp. 780–795.
- BusinessWorld, 2019, “Paytm Targets 1.5 Billion Merchant Payments During Festive Season,” .
- Carlin, Bruce, Arna Olafsson, and Michaela Pagel, 2019, “FinTech and Consumer Financial Well-Being in the Information Age,” .
- Chakravorti, Sujit, and Ted To, 2007, “A theory of credit cards,”international Journal of industrial organization, Vol. 25, No. 3, pp. 583–595.
- Chaturvedi, Mayuri, Tilottama Ghosh, and Laveesh Bhandari, 2011, “Assessing income distribution at the district level for India using nighttime satellite imagery,”Proceedings of the Asia-Pacific Advanced Network, Vol. 32, pp. 192–217.
- Chen, Xi, and William D Nordhaus, 2011, “Using luminosity data as a proxy for economic statistics,”Proceedings of the National Academy of Sciences, Vol. 108, No. 21, pp. 8589–8594.
- Chodorow-Reich, Gabriel, Gita Gopinath, Prachi Mishra, and Abhinav Narayanan, 2020, “Cash and the economy: Evidence from India’s demonetization,”The Quarterly Journal of Economics, Vol. 135, No. 1, pp. 57–103.
- Cochrane, John H, 1991, “A simple test of consumption insurance,”Journal of political economy, Vol. 99, No. 5, p. 957.
- Crépon, Bruno, Esther Duflo, Marc Gurgand, Roland Rathelot, and Philippe Zamora, 2012, “Do Labor Market Policies Have Displacement Effects? Evidence from a Clustered Ran-domized Experiment,” Techn. rep., National Bureau of Economic Research.
- ———, 2013, “Do labor market policies have displacement effects? Evidence from a clus-tered randomized experiment,”The quarterly journal of economics, Vol. 128, No. 2, pp. 531–580.
- Crouzet, Nicolas, Apoorv Gupta, and Filippo Mezzanotti, 2019, “Shocks and Technology Adoption: Evidence from Electronic Payment Systems,” Techn. rep., Northwestern Uni-versity Working Paper.
- Dalton, Patricio, H Pamuk, R Ramrattan, Daan van Soest, and Burak Uras, 2019, “Transpar-ency and Financial Inclusion: Experimental Evidence from Mobile Money (revision of CentER DP 2018-042),”CentER Discussion Paper, Vol. 2019.
- Davis, Lucas W, 2008, “The effect of driving restrictions on air quality in Mexico City,”Jour-nal of Political Economy, Vol. 116, No. 1, pp. 38–81.
- Demirguc-Kunt, Asli, Leora Klapper, Dorothe Singer, Saniya Ansar, and Jake Hess, 2018, The Global Findex Database 2017: Measuring financial inclusion and the fintech revolu-tion(The World Bank).
- Demyanyk, Yuliya, Charlotte Ostergaard, and Bent E Sørensen, 2007, “US banking dereg-ulation, small businesses, and interstate insurance of personal income,”The Journal of Finance, Vol. 62, No. 6, pp. 2763–2801.
- Frost, Jon, Leonardo Gambacorta, Yi Huang, Hyun Song Shin, and Pablo Zbinden, 2019, “BigTech and the changing structure of financial intermediation,” .
- GSMA, 2018, “State of the industry report: Mobile money,”London: Groupe Speciale Mobile Association.
- Guiso, Luigi, Tullio Jappelli, and Luigi Pistaferri, 2002, “An empirical analysis of earnings and employment risk,”Journal of Business & Economic Statistics, Vol. 20, No. 2, pp. 241–253.
- Handley, Kyle, and J Frank Li, 2018, “Measuring the effects of firm uncertainty on economic activity: New evidence from one million documents,”University of Michigan.
- Hau, Harald, Yi Huang, Hongzhe Shan, and Zixia Sheng, 2018, “Fintech credit, financial inclusion and entrepreneurial growth,”Unpublished working paper.
- Henderson, J Vernon, Adam Storeygard, and David N Weil, 2011, “A bright idea for measur-ing economic growth,”American Economic Review, Vol. 101, No. 3, pp. 194–199.
- Hoffmann, Mathias, and Iryna Shcherbakova-Stewen, 2011, “Consumption risk sharing over the business cycle: the role of small firms’ access to credit markets,”Review of Economics and Statistics, Vol. 93, No. 4, pp. 1403–1416.
- Jack, William, and Tavneet Suri, 2014, “Risk sharing and transactions costs: Evidence from Kenya’s mobile money revolution,”American Economic Review, Vol. 104, No. 1, pp. 183–223.
- Kinnan, Cynthia, and Robert Townsend, 2012, “Kinship and financial networks, formal finan-cial access, and risk reduction,”American Economic Review, Vol. 102, No. 3, pp. 289–93.
- Kulkarni, Rajendra, Kingsley E Haynes, Roger R Stough, and James D Riggle, 2011, “Re-visiting Night Lights as Proxy for Economic Growth: A Multi-Year Light Based Growth Indicator (LBGI) for China, India and the US,”GMU School of Public Policy Research Paper, , No. 2011-12.
- Lalive, Rafael, 2008, “How do extended benefits affect unemployment duration? A regression discontinuity approach,”Journal of econometrics, Vol. 142, No. 2, pp. 785–806.
- Mace, Barbara J, 1991, “Full insurance in the presence of aggregate uncertainty,”Journal of Political Economy, pp. 928–956.
- Manski, Charles F, 2004, “Measuring expectations,”Econometrica, Vol. 72, No. 5, pp. 1329–1376.
- Mbiti, Isaac, and David N Weil, 2015, “Mobile banking: The impact of M-Pesa in Kenya,” in African Successes, Volume III: Modernization and Development, pp. 247–293 (University of Chicago Press).
- Morawczynski, Olga, and Mark Pickens, 2009, “Poor people using mobile financial services: observations on customer usage and impact from M-PESA,” .
- Obstfeld, Maurice, 1995, “Risk-taking, global diversification, and growth,” Techn. rep., National Bureau of Economic Research.
- Philippon, Thomas, 2019, “On Fintech and Financial Inclusion,” Techn. rep., National Bureau of Economic Research.
- RBI, 2017 Feb, “RBI Bulletin,” .
- Suri, Tavneet, 2017, “Mobile money,”Annual Review of Economics, Vol. 9, pp. 497–520.
- Tanaka, Mari, Nicholas Bloom, Joel M David, and Maiko Koga, 2019, “Firm performance and macro forecast accuracy,”Journal of Monetary Economics.
- Townsend, Robert M, 1994, “Risk and insurance in village India,”Econometrica: Journal of the Econometric Society, pp. 539–591.
- Wieser, Christina, Miriam Bruhn, Johannes Philipp Kinzinger, Christian Simon Ruckteschler, and Soren Heitmann, 2019, “The Impact of Mobile Money on Poor Rural Households: Experimental Evidence from Uganda,”World Bank Policy Research Working Paper, , No. 8913.

*Source: wpiea2020138-print-pdf - REFERENCES*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020138-print-pdf.pdf_