## wpiea2022057-print-pdf

## Source details

**Canonical URL:** [wpiea2022057-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022057-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022057-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022057-print-pdf.pdf.json)

---

### Introduction: scope and motivation
- E-invoicing replaces paper invoices with machine-readable billing and payment data that can be imported into accounts payable systems and shared automatically with the tax authority.
- More than 50 countries have implemented e-invoicing for VAT, including ten countries in Latin America and the Caribbean.
- Study objective: investigate adoption spillovers of e-invoicing and whether mandates create tax compliance spillovers within firm networks using administrative- and transactions-level data from Peru.

### Conceptual framework
- Network externalities: increased value to others when more agents adopt the product/service.
- VAT asymmetry:
  - VAT refund credits increase in reported taxable purchases but decline in reported taxable sales.
  - Stronger incentives for firms to fully account for purchase invoices from upstream sellers to lower VAT liabilities.
  - Downstream firms mandated to switch to e-invoicing have stronger incentives to draw upstream suppliers into e-invoicing, producing larger upstream spillovers.
- Peer learning may also contribute to diffusion; channels beyond documenting spillovers and asymmetries cannot be conclusively distinguished.

### The e-invoicing reform in Peru: design and implementation
- E-invoicing available since mid-2000s; SUNAT’s multi-stage plan in 2013 to permanently switch from paper to e-invoicing.
- Implementation options for firms: in-house systems, authorized third-party systems, free SUNAT software, and an online platform for small/micro-enterprises.
- Post-transition: firms expected to remain e-issuers permanently; paper-based invoices allowed only in exceptional circumstances.
- As of 2020:
  - 61 percent of taxpayers used SUNAT platforms,
  - one third used in-house systems,
  - remainder used third-party solutions.
- Staggered implementation in waves to reduce adjustment costs and prioritize larger firms and taxpayers with poor compliance records.
- Six reform waves with original deadlines: October 2014, January 2015, July 2015, July 2016, December 2016, January 2017; deadline extensions ranged between 2 and 5 quarters for many groups.

### Reform waves and targeted groups (2013–2017)
- Waves 1, 3, 4, and 5:
  - 13,343 firms targeted using administrative classifications focusing on relatively larger firms (69 percent were SMEs).
  - Average reported annual sales: $15 million; average employment: 178 workers.
  - These firms represented just over 77 percent of total value added in the database.
  - Adoption rates over 80 percent within 6 months of deadlines for most waves by end-2017.
- Waves 2 and 6:
  - 7,443 firms flagged for Operaciones No Reales (ONR) during audits (high risk).
  - Average reported annual sales: $0.2 million; average employment: 3 workers.
  - These firms accounted for less than 1 percent of total value added in the database.
  - Adoption did not exceed 20 percent by end-2017.
- Firms not yet mandated: adoption slightly above 10 percent on average by end-2017.
- From 2018 onward: four groups totaling 90,347 firms (including 4,550 firms that committed some tax fraud) were mandated for 2018 deadlines; these were typically SMEs (average annual sales $0.8 million; average employment 14 workers).

### Data, sample construction, and scope
- Analysis period: 2013-2017 (earliest adoption deadline 2014; estimation period ends 2017; T = 3 in lag specifications).
- Data sources: three administrative tax databases (partnership-level annual transactions, firm-level VAT declarations, VAT payments) plus e-invoicing requirement/adoption data.
- Coverage:
  - Nearly 200,000 firms covering 53 percent of GDP and 95 percent of domestic VAT collections in 2013.
  - Excludes micro firms with less than 150 UIT (about $175,000) in annual sales and informal firms not registered with the tax authority.
  - Firms with less than 96,000 soles (about $29,000) in annual sales were automatically excluded due to simplified tax regime.
- Partnership analysis:
  - Top-five suppliers and top-five buyers per firm (based on 2014 ranking) used; top five account for about 30 percent of total purchases and sales respectively in a given year.
  - Firm-network frozen in 2014; follow transactions with 2014 top partners over 2013–2017.
- Final estimation samples:
  - Dataset covering about 150 thousand firms every year and 25,000 partners.
  - Balanced sample (positive sales every year 2013–2017): approximately 74,616 firms.
- Descriptive averages:
  - Average annual sales for not-yet-mandated firms: just below $0.5 million.
  - Not-yet-mandated sample: 3 percent large enterprises.
  - Voluntary adoption among not-yet-mandated firms: 1 percent in 2014 and 9 percent in 2017.
  - Average number of partners in 2014: 4.2 suppliers and 2.5 buyers.
  - Partners maintained in 2015: 2.3 suppliers and 1.5 buyers; by 2017: 1.5 suppliers and 0.8 buyers.
  - Average annual transaction between a firm and one of its partners in 2014: around $40,000.
  - Partnership-level links in 2014: 641 thousand links with top-five suppliers and 364 thousand links with top-five buyers.
  - Top-five mandated averages: suppliers 0.9 in 2014 → 2.8 in 2017; buyers 0.3 in 2014 → 1.5 in 2017.
  - Top-five partners skewed toward large enterprises: three fourths of top-five suppliers and two-thirds of top-five buyers are large enterprises; above one fourth of partners are exporters.

### Empirical approach (overview)
- Identification leverages staggered mandate waves; focus on firms not mandated as of 2017 to evaluate spillovers among trading partners.
- Main firm-level specification: dynamic panel difference-in-differences linear probability model where dependent variable = indicator that firm i voluntarily adopted e-invoicing at time t.
  - Key indicators: I(MandatedSupplier_{t-s}=1) and I(MandatedBuyer_{t-s}=1) for s = 0,...,T.
  - Coefficients of interest: β^S_0 and β^B_0 (first-year effects), β^S_s and β^B_s for s = 1,...,T.
  - Controls: δ_i (firm fixed effects) and γ_gt (group-year fixed effects where group g = unique combination of export status, sector, number of suppliers, number of buyers, and sales quartile at sample start).
- Partnership-level analysis unit: firm i — partner p — year t with fixed effects δ_{i,p} (pair) and γ_{i,t} (firm-year).
- Outcomes include indicators of any transactions (extensive margin) and log of transaction values (intensive margin).
- Compliance outcomes include firm-level reported taxable sales, purchases, VAT liabilities, new VAT credits and VAT payments.

### Key empirical findings — adoption spillovers
- First-year voluntary adoption spillovers (balanced sample):
  - Supplier mandated → voluntary adoption among treated firms increases by 7 percent in the first year relative to untreated firms (calculation: 0.00166/0.0242*100 = 6.86%).
  - Buyer mandated → voluntary adoption among treated firms increases by 26 percent in the first year relative to untreated firms (calculation: 0.00639/0.0242*100 = 26.41%).
- Asymmetry: downstream partner (buyer) effects substantially larger than upstream partner (supplier) effects; consistent with VAT incentives for purchase-reporting.
- Dynamics: spillovers increase over time (second-year spillovers more than double for buyers and increase four-fold for sellers from a low baseline); lag coefficients subject to selection bias due to staggered treatment waves.

### Key empirical findings — network structure and transactions
- Extensive margin (partnership survival):
  - Upstream partner mandated → 14 percent reduction in likelihood of partnership survival in the year of the reform.
  - Downstream partner mandated → 11 percent reduction in likelihood of partnership survival in the year of the reform.
  - If the non-mandated firm voluntarily adopts when its partner is mandated → net reduction in partnerships with suppliers and buyers is approximately 7 percent.
  - Likelihood of partnership termination increases between year 1 and year 2 after partner mandate; longer voluntary adoption by the partner improves partnership survival odds relative to non-adopters.
- Intensive margin (transaction volumes among surviving partnerships):
  - Mandating a partner has an insignificant effect on levels of transactions for sellers and a weakly negative effect for buyers.
  - When the non-mandated firm voluntarily adopts e-invoicing concurrent with partner mandate, transaction levels increase symmetrically for buyers and sellers (coefficient for sellers imprecisely estimated).
  - Interpretation: network externalities from belonging to the same VAT filing system raise intra-network transaction levels among surviving links; results warrant caution due to link attrition and partnership instability.

### Key empirical findings — spillovers in tax compliance
- In the year a partner is mandated to adopt e-invoicing, the treated firm reports:
  - 1.5 percent lower taxable sales.
  - 4 percent lower taxable purchases.
- Large reduction in new VAT credits observed among partners of mandated firms.
- No evidence of an aggregate decline in VAT payments; no systematic evidence of positive spillovers into tax compliance among partners in aggregate sales and purchases.
- Alternative interpretation: reductions in reported sales and purchases may reflect a reduction in firms filing false invoices (consistent with a much larger negative coefficient for purchases than for sales and the reduction in new VAT credits).
- Among firms that voluntarily adopt when their partners are mandated, VAT obligations and VAT payments rise but results are not statistically significant due to large standard errors.
- Filing of VAT credits goes down among firms with mandated partners, with a reversal (statistically insignificant) when the firm also voluntarily adopts.

### Heterogeneity by firm size and voluntary adoption
- Size definitions in analysis:
  - Sample median sales in 2013: approximately 750,000 soles (approximately 200 UIT). Larger firms in this analysis = sales above 200 UIT; smaller firms = sales below 200 UIT.
- Smaller firms (below median sales) with a mandated partner:
  - report 11 percent higher taxable sales in the first year of the reform.
  - report 15 percent higher VAT liabilities in the year their partner gets treated.
  - make 17 percent higher VAT payments in the year their partner gets treated.
- Larger firms (above median sales) with a mandated partner:
  - report 10 percent lower taxable sales in the first year of the reform.
  - report 9 percent lower VAT liabilities and make 11 percent lower VAT payments in the year their partner gets treated.
- Voluntary adoption by the firm when a partner is mandated is associated with a large and positive impact on reported sales and purchases, but statistical significance is found only for sales.
- Interpretation: positive compliance spillovers concentrated among smaller firms; negative spillovers for larger partners consistent with transaction costs or network disruption.

### Robustness and limitations
- Robustness checks:
  - Splitting sample between large groups mandated in 2017 and other firms yields qualitatively and quantitatively similar results.
  - Interaction of treatment with treatment intensity (number of mandated partners; share of transactions with mandated partners) produces qualitatively similar main coefficients; greater treatment intensity associated with strengthened likelihood of e-invoicing adoption and larger increase in taxable sales at adopting firms.
- Data limitations:
  - Dataset excludes micro firms (<150 UIT ≈ $175,000) and informal firms; estimated impacts not directly applicable to these groups.
  - Balanced sample constructed to avoid composition bias from zero observations; balanced sample (~74,616 firms) is somewhat larger on average than full sample firms.
  - Analysis-size categories differ from SUNAT administrative classifications; caution in generalizing to the entire universe of firms in Peru.

### Policy-relevant implications and research directions
- Targeted e-invoicing mandates generate sizeable spillovers across trading networks:
  - Positive technology-adoption spillovers, especially from downstream partners (buyers).
  - Positive tax-compliance spillovers for smaller partner firms; amplified if partners voluntarily adopt e-invoicing.
- Mandates can disrupt firm networks and impose transition/transaction costs (loss of partnerships); policy should aim to minimize such disruptions.
- Evaluations of e-invoicing reforms should account for indirect effects (voluntary adoption spillovers and compliance spillovers for non-mandated partners) in addition to direct effects on mandated firms.
- Important area for future research: identify sources of transition costs from adopting new technologies and effective measures to alleviate them.

*Source: wpiea2022057-print-pdf - References; Digitalization and Tax Compliance Spillovers — IMF Working Paper (selected sections).*

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

### wpiea2022057-print-pdf - References .................................................................................................

### Introduction: scope and motivation
- Digital transfer of invoice information (e-invoicing) replaces paper invoices with machine-readable billing and payment data that can be imported into accounts payable systems and shared automatically with the tax authority.
- More than 50 countries have implemented e-invoicing for VAT, including ten countries in Latin America and the Caribbean (Barreix and Zambrano, 2018).
- Study objective: investigate adoption spillovers of e-invoicing and whether mandates create tax compliance spillovers within firm networks using administrative- and transactions-level data from Peru.

### Conceptual framework: network externalities and asymmetries
- Network externalities: increased value to others when more agents adopt the product/service (Katz and Shapiro, 1985).
- Asymmetry in VAT setting:
  - VAT refund credits increase in reported taxable purchases but decline in reported taxable sales.
  - Stronger incentives for firms to fully account for purchase invoices from upstream sellers to lower VAT liabilities.
  - Downstream firms mandated to switch to e-invoicing have stronger incentives to draw upstream suppliers into e-invoicing, producing larger upstream spillovers.
- Peer learning may also contribute to diffusion, but the study cannot conclusively distinguish channels beyond documenting spillovers and asymmetries.

### The e-Invoicing Reform in Peru: design and implementation
- E-invoicing available since mid-2000s; voluntary adoption remained low, prompting SUNAT's multi-stage plan in 2013 to permanently switch from paper to e-invoicing.
- Firms given multiple implementation options: in-house systems, authorized third-party systems, free SUNAT software, and an online platform for small/micro-enterprises.
- Post-transition: firms expected to remain e-issuers permanently; paper-based invoices allowed only in exceptional circumstances (e.g., internet outage). As of 2020:
  - 61 percent of taxpayers used SUNAT platforms,
  - one third used in-house systems,
  - remainder used third-party solutions.
- Staggered implementation in waves to reduce adjustment costs and prioritize larger firms and taxpayers with poor compliance records.

### Reform waves and firm characteristics (2013–2017)
- Six reform waves with original adoption deadlines: October 2014, January 2015, July 2015, July 2016, December 2016, January 2017.
- Deadline extensions ranged between 2 and 5 quarters for many groups.
- Two main groups targeted across waves:
  - Waves 1, 3, 4, and 5: 13,343 firms targeted using administrative classifications focusing on relatively larger firms (69 percent were SMEs). Average reported annual sales: $15 million; average employment: 178 workers. These firms represented just over 77 percent of total value added in the database.
  - Waves 2 and 6: 7,443 firms flagged for Operaciones No Reales (ONR) during audits (high risk). Average reported annual sales: $0.2 million; average employment: 3 workers. These firms accounted for less than 1 percent of total value added in the database.
- Adoption behavior by end-2017:
  - Waves 1, 3, 4, and 5: adoption rates over 80 percent within 6 months of deadlines for most waves.
  - Waves 2 and 6: adoption did not exceed 20 percent by end-2017.
  - Firms not yet mandated: adoption slightly above 10 percent on average by end-2017.
- From 2018 onward: four groups totaling 90,347 firms (including 4,550 firms that committed some tax fraud) were mandated for 2018 deadlines. These were typically SMEs (on average annual sales $0.8 million; average employment 14 workers). After 2017, additional triggers for mandatory adoption included exporting, registration under certain tax regimes, or crossing sales thresholds.

### Empirical approach and sample notes
- Identification leverages staggered mandate waves and focuses on firms not mandated as of 2017 to evaluate spillovers among trading partners.
- Subsample composition: predominantly small and medium sized firms.
  - SUNAT administrative classifications: large firms have sales >2300 UIT; medium firms have sales between 2300 UIT and 1700 UIT; small firms have sales below 1700 UIT.
  - Within analysis subsample, ‘smaller’ firms have sales below roughly 200 UIT and ‘larger’ firms have sales above 200 UIT (200 UIT is the subsample median).
- Caution: do not conflate ‘larger’ firms in the estimation subsample with SUNAT’s ‘large’ classification; generalization to the entire universe of firms in Peru requires care.

### Key empirical findings
- Adoption spillovers
  - A firm with a downstream partner (buyer) mandated to adopt e-invoicing is 26 percent more likely to voluntarily adopt e-invoicing in the year the downstream partner is mandated than a firm with no such partnership.
  - A firm with an upstream partner mandated to adopt e-invoicing is 7 percent more likely to voluntarily adopt e-invoicing in the year the upstream partner is mandated than a firm with no such partnership.
  - Larger adoption spillovers from downstream partners and from exporting firms who receive cash refunds under the Peruvian VAT system.
- Network structure and transaction margins
  - Partnership termination (extensive margin): 14 percent reduction in likelihood of partnership survival in the year a firm’s upstream partner is mandated; 11 percent reduction when a downstream partner is mandated.
  - Intensive margin: among surviving partnerships, transactions increase between firms that voluntarily adopt e-invoicing when their partners are mandated, consistent with network externalities from using the same VAT filing system.
- Sales, purchases, and VAT reporting
  - Firms with partners mandated to adopt e-invoicing report reductions in sales and purchases, consistent with higher partnership termination rates.
  - No evidence of an aggregate decline in VAT payments.
  - Heterogeneity by firm size: smaller firms report higher sales, purchases, VAT liabilities, and make higher VAT payments following the reform relative to larger firms and firms without partners mandated to adopt e-invoicing — indicative of indirect positive tax compliance spillovers concentrated among smaller firms.
- Compliance interpretation
  - Results align with prior literature showing small firms more responsive to increased monitoring or audit probability.
  - Earlier work found direct effects of the reform were strongest for small firms (sales below 1700 UIT).

### Contribution to literature
- Provides evidence on propagation of e-invoicing adoption and tax compliance spillovers in firm supply networks, with asymmetric upstream/downstream effects.
- Connects to studies showing tax regime-induced segmentation of firm networks and the role of networks in technology diffusion.
- Complements literature on digital technologies in tax administration (e-invoicing, e-filing, electronic sales registries) by focusing on cross-firm spillovers.

### Implications and interpretation
- Targeted mandates can generate voluntary adoption spillovers through trading networks, but these spillovers are asymmetric and concentrated where incentives for accurate purchase reporting are strongest.
- Mandates induce both partnership terminations and intensified transactions among surviving or jointly adopting partners, suggesting non-trivial adjustment costs and partial segmentation of networks.
- Small firms are more likely to increase reporting and tax payments in response to increased monitoring via partners’ e-invoicing adoption, implying that network-based effects can amplify compliance outcomes beyond direct targets.

*Source: wpiea2022057-print-pdf - References .............................................................................................................*

### 2019. Subsequently, SUNAT announced plans to continue expanding the coverage of e-invoicing by lowering

### wpiea2022057-print-pdf - 2019. Subsequently, SUNAT announced plans to continue expanding the coverage of e-invoicing by lowering

### Background and context
- SUNAT announced continued expansion of e-invoicing coverage by lowering the threshold.
- By 2019:
  - 775 thousand firms were using e-invoicing.
  - 1.8 million electronic invoices were issued.
  - SUNAT estimated that 84 percent of formal sales were processed with e-invoices.
- The analysis period: 2013-2017. The earliest adoption deadline is in 2014 and the estimation period ends in 2017 (hence T = 3 in lag specifications).

### Empirical specification (overview)
- Primary focus: spillover effects from mandated partner firms on non-mandated firms (treated firms = firms with mandated partners; untreated firms = firms whose partners were not mandated).
- Main firm-level dynamic panel difference-in-differences linear probability model (equation (1)):
  - Dependent variable: indicator that firm i voluntarily adopted e-invoicing at time t.
  - Key indicators:
    - I(MandatedSupplier_{t-s}=1): equals one when, for the first time at time (t-s), at least one supplier is mandated to adopt e-invoicing.
    - I(MandatedBuyer_{t-s}=1): analogous for buyer.
  - Coefficients of interest:
    - β^S_0 and β^B_0: effect on voluntary adoption in the first year a partner is mandated.
    - β^S_s and β^B_s for s=1,...,T (T=3): effects for partners mandated s years before t.
  - Controls:
    - δ_i: firm fixed effects.
    - γ_gt: group-year fixed effects where group g = unique combo of export status, sector, number of suppliers, number of buyers, and sales quartile at sample start.
- Partnership-level analysis (equation (2)):
  - Unit: firm i — partner p — year t.
  - Y_{i,p,t} alternately: indicator of any transactions (extensive margin) and log of transaction values (intensive margin).
  - Fixed effects: δ_{i,p} (pair), γ_{i,t} (firm-year).
- Interaction with voluntary adoption (equation (3)) to test whether voluntary adoption by the non-mandated partner alters transaction outcomes.
- Compliance outcomes (equation (4)): firm-level reported taxable sales, purchases, VAT liabilities, new VAT credits and VAT payments, with analogous controls.

### Data and sample construction
- Data sources: three administrative tax databases (partnership-level annual transactions, firm-level VAT declarations, VAT payments, plus e-invoicing requirement/adoption data).
- Coverage:
  - Nearly 200,000 firms covering 53 percent of GDP and 95 percent of domestic VAT collections in 2013.
  - Excludes micro firms with less than 150 UIT (about $175,000) in annual sales and informal firms not registered with the tax authority.
  - Firms with less than 96,000 soles (about $29,000) in annual sales were automatically excluded due to simplified tax regime.
- For partnership analysis:
  - Top-five suppliers and top-five buyers per firm (based on 2014 ranking) are used; top five account for about 30 percent of total purchases and sales respectively in a given year.
  - Firm-network frozen in 2014; follow transactions with 2014 top partners over 2013–2017 even if rankings change.
- Final estimation samples:
  - Dataset covering about 150 thousand firms every year and 25,000 partners.
  - Balanced sample (positive sales every year 2013–2017): approximately 74,616 firms.
- Descriptive averages and counts:
  - Average annual sales for not-yet-mandated firms: just below $0.5 million.
  - In the not-yet-mandated sample: only 3 percent are large enterprises.
  - Voluntary adoption among not-yet-mandated firms: 1 percent in 2014 and 9 percent in 2017.
  - Average number of partners in 2014: 4.2 suppliers and 2.5 buyers.
  - In 2015: firms maintained positive transactions with 2.3 of their 2014 top-5 suppliers and 1.5 of their 2014 top-5 buyers on average.
  - By 2017: averages were 1.5 for suppliers and 0.8 for buyers.
  - Average number of top-5 partners mandated into e-invoicing: 0.9 suppliers and 0.3 buyers in 2014; 2.8 suppliers and 1.5 buyers in 2017.
  - Partnership-level links in 2014: 641 thousand links with top-five suppliers and 364 thousand links with top-five buyers.
  - Average annual transaction between a firm and one of its partners in 2014: around $40,000.
  - Top-five partners are skewed toward large enterprises: three fourths of top-five suppliers and two-thirds of top-five buyers are large enterprises; above one fourth of partners are exporters.

### Results — Spillovers in technology adoption
- First-year voluntary adoption spillovers (balanced sample):
  - If a firm’s supplier is mandated, voluntary adoption among treated firms increases by 7 percent in the first year relative to untreated firms that do not have suppliers mandated to adopt e-invoicing.
    - Footnote calculation: treatment effect from suppliers = 0.00166/0.0242*100 = 6.86%.
  - If a firm’s buyer is mandated, voluntary adoption among treated firms increases by 26 percent in the first year relative to untreated firms.
    - Footnote calculation: treatment effect from buyers = 0.00639/0.0242*100 = 26.41%.
- Asymmetry and interpretation:
  - Spillovers from downstream partners (buyers) are substantially larger than from upstream partners (suppliers).
  - Mechanism consistent with VAT incentives:
    - Downstream mandated firms benefit from collecting purchase invoices (reducing VAT obligations); they have incentives to induce upstream partners into the same e-filing regime.
    - Upstream partners have weaker incentives because VAT obligations increase with recorded sales.
- Dynamics:
  - Spillovers increase over time: second-year spillovers more than double for buyers and increase four-fold for sellers (from a low baseline), but buyer spillovers remain higher than seller spillovers.
  - Coefficients for lags > 1 year are subject to selection bias due to staggered treatment waves.

### Results — Market segmentation: transactions between firms
- Extensive margin (partnership survival):
  - When an upstream partner (seller) is mandated, probability of partnership surviving in the year of the reform falls by 14 percent.
  - When a downstream partner (buyer) is mandated, the probability of partnership surviving in the year of the reform falls by 11 percent.
  - If the non-mandated firm voluntarily adopts e-invoicing when its partner is mandated, the negative impact on partnership survival is reduced; net reduction in partnerships with suppliers and buyers is approximately 7 percent.
  - Dynamics: likelihood of partnership termination increases between year 1 and year 2 after partner mandate; longer voluntary adoption by the partner improves partnership survival odds relative to non-adopters.
  - Interpretation: evidence consistent with transaction costs from operating in disparate filing systems and possible reduction in fictitious upstream firms.
- Intensive margin (transaction volumes among surviving partnerships):
  - Mandating a partner has an insignificant effect on levels of transactions for sellers and a weakly negative effect for buyers.
  - However, when the non-mandated firm voluntarily adopts e-invoicing concurrent with partner mandate, transaction levels increase symmetrically for buyers and sellers (coefficient for sellers imprecisely estimated).
  - Interpretation: network externalities from belonging to the same VAT filing system raise intra-network transaction levels among surviving links; results warrant caution due to link attrition and partnership instability.

### Results — Spillovers in tax compliance (overview)
- The analysis explores whether partners being mandated to adopt e-invoicing affect a firm’s reported taxable sales, purchases, VAT liabilities, new VAT credits and VAT payments (firm-level, encompassing all trading partners).
- Table A.10 (referenced) indicates the average impact of a partner being mandated into e-invoicing is statistically significant and negative in the first year (detailed coefficients and outcomes in appendix tables not reproduced here).
- Earlier related evidence (Bellon et al., 2019) finds mandated e-invoicing directly increases reported taxable sales and taxable purchases for treated firms, suggesting improved detection and reporting via cross-checking of digital transaction registers.

### Data limitations and sample considerations
- Dataset excludes micro firms (<150 UIT ≈ $175,000) and informal firms; estimated impacts are not directly applicable to these groups.
- To avoid composition bias from many zero observations in small-firm subsamples:
  - A balanced sample of firms with positive sales in every year 2013–2017 is constructed and nominal variables transformed using ln(x + 1).
  - Balanced sample has approximately half as many firms as full sample (~74,616 firms) and firms are somewhat larger on average.
- Heterogeneity by firm size:
  - Size cutoff in analysis defined relative to median sales in balanced sample: larger firms (sales greater than 200 UIT) vs smaller firms (sales below 200 UIT). 200 UIT is the subsample median.
  - Caution: these analysis-size categories differ from SUNAT administrative classifications (e.g., large firms defined administratively as sales above 2300 UIT).

### Key statistics and exact numeric facts (preserved)
- 775 thousand firms using e-invoicing by 2019.
- 1.8 million electronic invoices issued in 2019.
- 84 percent of formal sales processed with e-invoices (SUNAT estimate).
- Analysis period: 2013-2017.
- T = 3 (maximum partner mandate lag because earliest deadline 2014 and estimation ends 2017).
- Dataset initial coverage: nearly 200,000 firms; 53 percent of GDP; 95 percent of domestic VAT collections in 2013.
- Micro firms definition: less than 150 UIT (about $175,000) in annual sales.
- Automatic exclusion threshold: less than 96,000 soles (about $29,000) in annual sales.
- Final merged dataset for analysis: about 150 thousand firms every year and 25,000 partners.
- Average annual sales for not-yet-mandated firms: just below $0.5 million.
- Not-yet-mandated sample: 3 percent large enterprises.
- Voluntary adoption rates among not-yet-mandated firms: 1 percent in 2014; 9 percent in 2017.
- Average partners in 2014: 4.2 suppliers; 2.5 buyers.
- Partners maintained in 2015: 2.3 suppliers; 1.5 buyers. By 2017: 1.5 suppliers; 0.8 buyers.
- Top-five suppliers/buyers account for about 30 percent of purchases and sales respectively.
- Top-five mandated averages: suppliers 0.9 in 2014 → 2.8 in 2017; buyers 0.3 in 2014 → 1.5 in 2017.
- Partnership links in 2014: 641 thousand (suppliers) and 364 thousand (buyers).
- Average annual transaction per firm-partner in 2014: around $40,000.
- Voluntary adoption spillover magnitudes (first year):
  - Suppliers: increase by 7 percent (calculation: 0.00166/0.0242*100 = 6.86%).
  - Buyers: increase by 26 percent (calculation: 0.00639/0.0242*100 = 26.41%).
- Partnership survival effects in year of reform:
  - Upstream mandated partner → 14 percent reduction in partnership survival.
  - Downstream mandated partner → 11 percent reduction in partnership survival.
  - If firm voluntarily adopts when partner mandated → net partnership reduction approximately 7 percent.
- Balanced sample size: approximately 74,616 firms.

*Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022057-print-pdf.pdf*

### Section IV for the details of the database).

### Digitalization and Tax Compliance Spillovers

### Spillover effects on reported transactions and VAT
- In the year a partner is mandated to adopt e-invoicing, the treated firm reports 1.5 percent lower taxable sales.
- Treated firms report 4 percent lower taxable purchases in the first year following their partner being mandated.
- Large reduction in new VAT credits is observed among partners of mandated firms.
- No systematic evidence of positive spillovers into tax compliance among partners in aggregate sales and purchases (an increase in reported transactions would indicate this).
- Alternative interpretation: reductions in reported sales and purchases may reflect a reduction in firms filing false invoices (consistent with a much larger negative coefficient for purchases than for sales and the reduction in new VAT credits).

### Heterogeneity: voluntary adoption by partners
- When a partner is mandated, voluntary adoption by the firm is associated with a large and positive impact on reported sales and purchases, but statistical significance is only found for sales.
- The un-interacted coefficient for voluntary adoption (firms without mandated partners) is statistically insignificant and negative.
- Positive compliance or productivity spillovers are possible when partners voluntarily adopt e-invoicing after a partner is mandated; cannot distinguish whether effects reflect lower transaction costs (productivity) or improved tax compliance.

### Effects on VAT obligations and payments
- No statistically significant spillovers of being associated with a mandated firm on reported VAT obligations or VAT payments in the aggregate.
- Among firms that voluntarily adopt when their partners are mandated, VAT obligations and VAT payments rise but results are not statistically significant due to large standard errors.
- Filing of VAT credits goes down among firms with mandated partners, with a reversal (statistically insignificant) when the firm also voluntarily adopts.

### Firm survival and regime exit
- Firm survival is defined as observing a firm making positive sales in years subsequent to their partners being mandated (implying continued filing in the VAT general regime).
- Firms connected to mandated partners are significantly more likely to survive and report positive sales in the year that their partner is mandated and in subsequent years.
- Being connected to a treated firm increases the likelihood of firms filing under the general regime in all years following the reform.
- Overall, firms do not exit the VAT general regime in larger numbers if they are connected to firms mandated to adopt e-invoicing.

### Heterogeneity by firm size (sales in 2013)
- Sample median sales in 2013: approximately 750,000 soles (approximately 200 UIT).
- Larger firms in this analysis are defined as firms with sales above 200 UIT (majority still small by administrative classification).
- Firms with below median sales who have a mandated partner:
  - report 11 percent higher taxable sales in the first year of the reform.
  - report 15 percent higher VAT liabilities in the year their partner gets treated.
  - make 17 percent higher VAT payments in the year their partner gets treated.
- Firms with above median sales who have a mandated partner:
  - report 10 percent lower taxable sales in the first year of the reform.
  - report 9 percent lower VAT liabilities and make 11 percent lower VAT payments in the year their partner gets treated.
- Interpretation:
  - Negative spillovers for larger partners are consistent with transaction costs or network disruption (loss of partnerships).
  - Positive spillovers for smaller firms are consistent with improved monitoring via e-invoicing leading to more comprehensive reporting and higher VAT collections for small and medium firms.
  - Smaller firms are more responsive to increased likelihood of detection, consistent with prior literature.

### Robustness checks
- Splitting sample between large groups mandated in 2017 and other firms yields qualitatively and quantitatively similar results; findings are not driven by expectations about future e-invoicing requirements.
- Interaction of treatment with treatment intensity (number of mandated partners; share of transactions with mandated partners) produces qualitatively similar main coefficients.
  - Greater treatment intensity is associated with:
    - a strengthened likelihood of e-invoicing adoption.
    - a larger increase in taxable sales at firms that adopt e-invoicing.

### Policy-relevant implications and research directions
- E-invoicing mandates generate sizeable spillovers across trading networks, notably:
  - Positive technology-adoption spillovers (propensity to adopt e-invoicing), especially with downstream partners (buyers).
  - Positive tax-compliance spillovers for smaller partner firms; these effects are amplified if partners voluntarily adopt e-invoicing.
- E-invoicing adoption can disrupt firm networks and impose transition/transaction costs (loss of partnerships); policy should aim to minimize such disruptions.
- Evaluations of e-invoicing reforms should account for indirect effects (voluntary adoption spillovers and compliance spillovers for non-mandated partners) in addition to direct effects on mandated firms.
- Important area for future research: identify sources of transition costs from adopting new technologies and effective measures to alleviate them.

*Source: Digitalization and Tax Compliance Spillovers — IMF Working Paper (selected sections).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022057-print-pdf.pdf_