## wpiea2025192-source-pdf

## Source details

**Canonical URL:** [wpiea2025192-source-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025192-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025192-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025192-source-pdf.pdf.json)

---

### Key findings on credit and product innovation
- Firms introduce a new product in 47% of years on average.
- Firms introduce a more complex product than prior offerings in 12% of years on average.
- MSMEs account for about 50% of all new product introductions during the sample period.
- Newly eligible (treated) firms borrow 12–22% more per year, depending on the credit measure used.
- On average, increased access to bank credit has little impact on product innovation:
  - The point estimate for product scope is close to zero; with 95% confidence the study rejects that product scope grew by more than 0.05 products (0.04 SD).
  - With 95% confidence the study rejects that product innovation increased by more than 2 percentage points (0.04 SD).

### Mechanisms and firm responses to increased credit
- Newly eligible firms’ outcomes:
  - Sales increase by 28%.
  - Net income increases by 24%.
- Input and labor responses:
  - Spend 26% more on materials.
  - Use 28% more worker days.
  - Pay workers 18% higher wages.
  - Do not increase physical investment (capital investment point estimates positive but statistically insignificant).
- Interpretation: in the developing-country context studied, firms often face tighter credit constraints and indivisibilities in inputs; additional credit is frequently used to better exploit existing capacity rather than to innovate.

### Non-financial barriers: measurement and prevalence
- Ten non-financial barriers constructed as binary pre-treatment measures:
  1. Market Concentration: Herfindahl-Hirschman Index for each state×year×product market.
  2. Market Size: total sales across all firms, defined at state×year×product level.
  3. Import Reliant: indicator for whether the firm imports inputs.
  4. Electricity: measure of electricity shortage at state×industry level.
  5. Road Infrastructure: fraction of villages in state with paved roads, from last pre-reform census.
  6. Labor Regulation: measure of the strictness of each state’s labor regulation.
  7. Land Access: proxy for efficiency of land markets based on prior land reforms.
  8. Legal Contract Enforcement: proxy based on court congestion.
  9. Education: share of children meeting reading and math proficiency standards.
  10. Rural: whether a firm is located in a rural area.
- The median firm faces five non-financial barriers (Barrier Count: Mean 5.17, P25 4.00, P50 5.00, P75 6.00, SD 1.70, N 33,094).

### Heterogeneous effects: when credit increases innovation
- Among firms that face no barriers (Barrier Count = 0):
  - Product scope increases by 0.17 products (0.14 SD).
  - Firms are 12.7 percentage points (0.25 SD) more likely to introduce a new product.
  - Firms are 7.9 percentage points (0.23 SD) more likely to introduce a more complex product.
- These effects are slightly larger in magnitude than those seen in advanced economies (Granja & Moreira, 2023).

### Dilution of credit effects by market barriers
- For each additional barrier faced by a firm:
  - Credit access has a −0.03 product weaker effect on scope (−0.03 SD).
  - Credit access has a −2.5 percentage points (−0.05 SD) weaker effect on product innovation rates.
  - Credit access has a −1.8 percentage point (−0.05 SD) weaker effect on the odds of introducing a more complex product.
- No single barrier drives the patterns; multiple overlapping non-financial barriers collectively reduce firms’ ability to translate credit into new and more complex products.
- For the median firm (five barriers), credit has a net null effect on innovation, consistent with baseline results.

### Effects of barriers on firm growth, profits, and investment
- Among firms that face no barriers:
  - Sales increase by 98%.
  - Profits increase by 72%.
- Each additional barrier reduces the impact of credit by:
  - 13% on sales.
  - 8% on profits.
- Dilution effects on investment:
  - Newly eligible firms facing no barriers raise investment by 62%.
  - Each additional barrier reduces the impact of credit on investment by 10%.
- For the median firm (five barriers), credit increases sales and profits by roughly 20–30% but does not increase investment or innovation.

### Institutional background and empirical setting
- Context: India’s MSMEs accounted for nearly 35% of manufacturing employment in 2005.
- 2006 MSME Development Act change: raised the size cutoff for MSME classification from under 10 million rupees to 50 million rupees, making firms with assets between 10–50 million rupees newly eligible for Priority Sector Lending (PSL) and other credit access programs.
- As a result of the reform, 10% of all Indian manufacturing firms, which collectively account for 15% of manufacturing output, became newly eligible for credit access programs.

### Data, measures of innovation, and sample
- Data: Annual Survey of Industries (ASI), nationally representative survey of registered manufacturing establishments in India.
- Product-level granularity: 9 sections, 64 product divisions, and 5,400 unique products.
- Primary innovation measures:
  - Product Scope: number of unique products a firm sells in a given year.
  - Product Innovation: indicator for whether a firm introduces a product it has not previously sold.
  - Product Complexity: indicator for whether a firm sells a more complex product than its previous offerings, using an index based on the country-product export network (Hidalgo and Hausmann, 2009).
- Continuous sales-based measures:
  - Innovation Sales Share: fraction of total sales from new products.
  - Complex-Innovation Sales Share: fraction of sales from new and more complex products.
- Comparison to scanner data: scanner products account for only 34% of products and 52% of sales in the ASI-mapped sample; ASI includes intermediate goods not captured in typical scanner datasets.
- Sample restrictions:
  - Analysis limited to years 2001–2010 due to ASI product classification changes.
  - Sample restricted to firms observed at least twice prior to and once after the 2006 reform.
  - Firms with more than 9 products are dropped because ASI only lists a firm’s top 10 products.
  - Final sample: Firms 10,942; Firm-Years 48,835 (Table A.2).

### Stylised facts on innovation (selected statistics)
- Average firm sells 1.8 products in a given year (Product Scope: Mean 1.64 in Table 1, Panel A).
- Probability of introducing a new product in any year: 47% (Product Innovation: Mean 47%).
- Over 2001–10, more than 66,000 instances of firms introducing new products were observed.
- Over 2002–10, distinct products grew from 3,499 to 3,813 (8.9% increase).
- New products account for 33% of a firm’s annual sales on average (Innovation Sales Share: Mean 33%).
- Approximately 11% of innovations raise product complexity (Product Complexity: Mean 12% in Panel A; note measure definitions vary in table).
- Indian MSMEs average annual revenue: INR 96 million (USD 2.2 million in 2005).
- Indian firms sell on average 1.8 products per year versus 25.7 in Granja and Moreira (2023).

### Main estimation results (preserved point estimates and standard errors)
- Impact on borrowing (Table 2; Post × Newly Eligible):
  - Outstanding Loans (Log): 0.12 (0.07) *
  - Overdraft (Log): 0.21 (0.07) ***
  - Credit (Log): 0.22 (0.07) ***
- Impact on innovation (Table 3; Post × Newly Eligible):
  - Product Scope: 0.003 (0.024)
  - Product Innovation: −0.005 (0.013)
  - Product Complexity: −0.015 (0.011)
  - Innovation Sales Share: 0.017 (0.012)
  - Complex-Innovation Sales Share: 0.003 (0.008)
- Impact on sales, profits, and wages (Table 4; Post × Newly Eligible):
  - Sales (Log): 0.28 (0.08) ***
  - Gross Value Added (Log): 0.24 (0.07) ***
  - Net Income (Log): 0.24 (0.07) ***
  - Managerial Wages (Log): 0.14 (0.05) ***
- Impact on inputs and investment (Table 5; Post × Newly Eligible):
  - Materials Consumed (Log): 0.26 (0.08) ***
  - Days Worked (Log): 0.28 (0.07) ***
  - Workers’ Wages (Log): 0.18 (0.05) ***
  - Investment (Log): 0.08 (0.06)
- Heterogeneous impacts by barrier count (Table 6; triple difference):
  - Post × Newly Eligible:
    - Product Scope: 0.174 (0.085) **
    - Product Innovation: 0.127 (0.047) ***
    - Product Complexity: 0.079 (0.038) **
    - Innovation Sales Share: 0.132 (0.042) ***
    - Complex-Innovation Sales Share: 0.058 (0.029) **
  - Post × Newly Eligible × Barrier Count:
    - Product Scope: −0.032 (0.017) *
    - Product Innovation: −0.025 (0.009) ***
    - Product Complexity: −0.018 (0.007) **
    - Innovation Sales Share: −0.020 (0.008) **
    - Complex-Innovation Sales Share: −0.010 (0.005) *
  - Median Effect reported: Product Innovation −0.073 (and other median effects).

### Robustness, appendix, and additional evidence
- Robustness checks: firm-size restricted subsamples, exclusion of de-reserved products, narrow sample checks, granular barrier exclusions, and individual barrier interactions (Tables B.1–B.6).
- Appendix highlights:
  - Table C.5: firms that face no barriers make investments in both buildings and plant & equipment.
  - Index-specification example (Table C.1): Post × Newly Eligible on Credit (Index): 15.95 (4.72) ***; Sales (Index): 12.94 (3.65) ***; Investment (Index): 6.63 (3.76) *; Net Income (Index): 28.36 (8.67) ***.
  - Figure 4: dynamic DID plots; Figure 5: impacts by barrier count evaluated at 0, 3, and 5 barriers.

### Interpretation and policy implications
- Heterogeneous effects: credit broadly expands production and profits but enables product innovation primarily for firms facing few or no non-financial barriers.
- Mechanism: many firms operate below efficient scale and use credit to intensify variable inputs and exploit existing capacity rather than to introduce new or more complex products.
- Policy recommendation (high-level):
  - Unlocking product innovation requires addressing multiple, overlapping non-financial constraints in addition to expanding credit.
  - A coordinated big-push strategy that combines credit support with interventions reducing non-financial barriers (input and output market frictions, electricity reliability, infrastructure, legal congestion, market competition) can increase the effectiveness of credit programs in fostering product innovation in emerging markets.

*Source: wpiea2025192-source-pdf - introduction of new products (Granja & Moreira, 2023); and (iii)product complex-*

### introduction of new products (Granja & Moreira, 2023); and (iii)product complex-

### introduction of new products (Granja & Moreira, 2023); and (iii)product complexity

### Key findings on credit and product innovation
- On average, firms introduce a new product in 47% of years and a more complex product than prior offerings in 12% of years.
- MSMEs account for about 50% of all new product introductions during the sample period.
- Treated (newly eligible) firms borrow 12-22% more per year, depending on the measure of credit used.
- On average, increased access to bank credit has little impact on product innovation:
  - The point estimate is close to zero and with 95% confidence, the authors can reject that product scope grew by more than 0.05 products (0.04 standard deviations, SD).
  - With 95% confidence, they can reject that product innovation increased by more than 2 percentage points (0.04 SD).

### Mechanisms and firm responses to increased credit
- Newly eligible firms increase sales by 28% and net income by 24% despite not changing product mix or introducing new or more complex products.
- Newly eligible firms’ input and labor responses:
  - Spend 26% more on materials.
  - Use 28% more worker days.
  - Pay workers 18% higher wages.
  - Do not increase physical investment.
- Interpretation: in developing country contexts, firms often face tighter credit constraints and indivisibilities in inputs, leading constrained firms to operate with slack; additional credit is often used to better exploit existing capacity rather than to innovate.

### Non-financial barriers to innovation: measurement and prevalence
- Ten non-financial barriers are identified and measured using pre-treatment data; all ten barriers are constructed as binary measures:
  1. Market Concentration: Herfindahl-Hirschman Index for each state×year×product market.
  2. Market Size: total sales across all firms, defined at state×year×product level.
  3. Import Reliant: an indicator for whether the firm imports inputs.
  4. Electricity: a measure of electricity shortage at state×industry level.
  5. Road Infrastructure: fraction of villages in state with paved roads, from last pre-reform census.
  6. Labor Regulation: measure of the strictness of each state’s labor regulation.
  7. Land Access: a proxy for the efficiency of land markets based on prior land reforms.
  8. Legal Contract Enforcement: proxy based on court congestion.
  9. Education: share of children meeting reading and math proficiency standards.
  10. Rural: whether a firm is located in a rural area.
- The median firm faces five non-financial barriers, though there is significant variation across firms.

### Heterogeneous effects: when credit increases innovation
- Among firms that face no barriers, credit access significantly increases product innovation:
  - Product scope increases by 0.17 products (0.14 SD).
  - Firms are 12.7 percentage points (0.25 SD) more likely to introduce a new product.
  - Firms are almost 8 percentage points (0.24 SD) more likely to introduce a more complex product.
- These effects are slightly larger in magnitude than those seen in advanced economies (Granja & Moreira, 2023).

### Dilution of credit effects by market barriers
- For each additional barrier faced by a firm:
  - Credit access has a 0.03 product weaker effect on scope (0.03 SD).
  - Credit access has a 2.5 percentage points (0.05 SD) weaker effect on product innovation rates.
  - Credit access has a 1.8 percentage point (0.05 SD) weaker effect on the odds that firms introduce a more complex product.
- No single barrier drives the patterns; multiple overlapping non-financial barriers collectively reduce the ability of firms to translate credit into new and more complex products.
- For the median firm (facing five barriers), credit has a net null effect on innovation, consistent with baseline results.

### Effects of barriers on firm growth, profits, and investment
- Among firms that face no barriers, credit access increases:
  - Sales by 98%.
  - Profits by 72%.
- Each additional barrier reduces the impact of credit by:
  - 13% on sales.
  - 8% on profits.
- Dilution effects are stronger for investment:
  - Newly eligible firms facing no barriers raise investment by 62%.
  - Each additional barrier reduces the impact of credit on investment by 10%.
- For the median firm, credit increases sales and profits (by roughly 20-30%) but not investment.

### Institutional background and empirical setting
- Context: India’s MSMEs played a crucial role in manufacturing employment; in 2005, MSMEs accounted for nearly 35% of manufacturing employment.
- In 2006, the MSME Development Act raised the size cutoff for MSME classification from under 10 million rupees to 50 million rupees, making firms with assets between 10-50 million rupees newly eligible for Priority Sector Lending (PSL) and other credit access programs.
- As a result of the reform, 10% of all Indian manufacturing firms, which collectively account for 15% of manufacturing output, became newly eligible for credit access programs.

### Data, measures of innovation, and sample
- Data source: Annual Survey of Industries (ASI), nationally representative survey of registered manufacturing establishments in India.
- Product-level granularity: 9 sections, 64 product divisions, and 5,400 unique products.
- Three primary innovation measures:
  - Product Scope: number of unique products a firm sells in a given year.
  - Product Innovation: indicator for whether a firm introduces a product it has not previously sold.
  - Product Complexity: indicator for whether a firm sells a more complex product than its previous offerings, using an index based on the country-product export network following Hidalgo and Hausmann (2009).
- Continuous sales-based measures:
  - Innovation Sales Share: fraction of total sales from new products.
  - Complex-Innovation Sales Share: fraction of sales from new and more complex products.
- Comparison to scanner data: scanner products account for only 34% of products and 52% of sales in the ASI-mapped sample; ASI includes intermediate goods not captured in typical scanner datasets.
- Sample restrictions:
  - Analysis limited to years 2001-2010 due to ASI product classification changes.
  - Sample restricted to firms observed at least twice prior to and once after the 2006 reform.
  - Firms with more than 9 products are dropped because ASI only lists a firm’s top 10 products.

### Contribution to the literature
- Adds causal evidence on how access to bank credit affects product innovation in a developing country context using an exogenous policy reform in India.
- Shows that easing financial constraints increases innovation and sales for firms that face few non-financial barriers, but that non-financial barriers significantly dampen the effects of credit on innovation for the average firm.
- Complements findings from advanced-economy studies (e.g., Granja & Moreira, 2023) by highlighting the role of non-financial market frictions and firm size differences in shaping the credit–innovation relationship.
- Suggests that big-push policies that address multiple frictions simultaneously may be required to foster product innovation in emerging markets.

*Source: wpiea2025192-source-pdf - introduction of new products (Granja & Moreira, 2023); and (iii)product complex-*

### 3.1   Stylised Facts on Innovation

### 3.1   Stylised Facts on Innovation

### Key empirical facts on innovation
- #1: Firms regularly introduce new products.
  - The average firm sells 1.8 products in a given year.
  - The probability of introducing a new product in any given year is 47%.
  - Over the 2001-10 period, more than 66,000 instances of firms introducing new products were observed.
  - Over the 2002-10 period, the number of distinct products in the Indian economy grew from 3,499 to 3,813, an 8.9% increase.
  - On average, new products account for 33% of a firm’s annual sales.

- #2: MSMEs account for a significant fraction of product innovation.
  - Nearly 50% of new product introductions come from MSMEs.
  - 13% of all new products are sold exclusively by MSMEs and never by larger firms.

- #3: A meaningful share of innovations raise product complexity.
  - Approximately 11% of innovations involve firms introducing more complex products than they have previously sold.
  - Over the sample period, the sales-weighted complexity of Indian manufacturing products rose, with the magnitude comparable to moving from Cuba’s export basket in 2002 to Russia’s export basket in 2010.

### Comparison with advanced-economy evidence
- Product innovation rates are similar across contexts:
  - 47% annual probability of introducing a new product in the Indian sample versus 34% in Granja and Moreira (2023).
- Scale and scope differences:
  - Indian MSMEs average annual revenue: INR 96 million (USD 2.2 million in 2005).
  - Indian firms sell an average of 1.8 products per year versus 25.7 in Granja and Moreira (2023).
- Interpretation:
  - Much smaller scale and narrower product scope among Indian MSMEs may affect how additional capital is deployed (e.g., expanding existing lines versus introducing new products).

### How the study evaluates credit’s role in innovation (empirical framework highlights)
- Primary identification: difference-in-difference specification exploiting a 2006 policy change.
  - Baseline model: y_i,t = β1 Post_t × NewlyEligible_i + a_s,t + b_j,t + c_i + ε_i,t
  - Extended model introduces Barriers_i (count 0–10) and triple interaction to capture how pre-treatment barriers modify the effect of credit.
- Fixed effects control:
  - firm FEs (c_i), state×year FEs (a_s,t), industry×year FEs (b_j,t).
- Outcomes include innovation measures as well as sales, profits, investment, inputs, and wages.
- Standard errors clustered at the firm level; observations weighted by inverse sampling probability.

### Main findings on credit access and innovation
- Impact on credit access (Table 2):
  - Long-term loan balances increased by 12% for newly-eligible firms relative to firms with no shift in eligibility.
  - Short-term and total borrowing expanded by more than 20%.
  - On average, these loans represented 51% of borrowing for newly-eligible firms in the post-reform period.

- Average impact on product innovation (Table 3):
  - Post × Newly Eligible coefficients for product scope, probability of introducing a new product, and probability of introducing more complex products are small and not statistically different from zero.
  - At the 95% confidence level, the study rules out that newly eligible firms:
    - expand product scope by more than 0.05 products (0.04 SD),
    - increase product innovation rates by more than 2 percentage points (0.04 SD),
    - raise odds of introducing more complex products by more than 0.7 percentage points (0.02 SD).
  - Share of sales from new or more complex products: no average effect; estimates small, precisely estimated, and statistically insignificant.
  - Robustness: results persist across firm-size restricted subsamples (0–100mm, 0–50mm, 0–30mm, 0–20mm) and when excluding firms producing de-reserved products.

### Why credit did not raise innovation on average
- Two main explanations:
  1. Firms expanded existing production rather than innovating (profitably scaling current lines).
  2. Non-financial market barriers dilute the translation of credit into innovation.

### Evidence that firms expanded existing production (Tables 4–5)
- Newly eligible firms increased scale and profitability:
  - Sales increased by 28%.
  - Gross value added (GVA) grew by 24%.
  - Net income increased by 24%.
  - Managerial wages rose by 14%.
- Input use rises concentrated in variable inputs:
  - Raw materials increased by 26%.
  - Worker days increased by 28%.
  - Wage expenditure increased by 18%.
  - Effects on capital investment: point estimate positive but statistically insignificant and smaller in magnitude than variable input responses; disaggregated capital investment estimates for machinery, buildings, and transport equipment are economically small and statistically indistinguishable from zero (Appendix evidence).

### Non-financial barriers dilute credit’s effect on innovation (Table 6)
- Barrier measure: count of up to ten market barriers faced pre-reform; median firm faces five barriers.
- Among firms that face no barriers:
  - Credit access increases product scope by 0.17 products (0.14 SD).
  - Product innovation rates increase by 12.7 percentage points (0.25 SD).
  - Odds of introducing more complex products increase by 7.9 percentage points (0.23 SD).
- Each additional barrier attenuates innovation effects:
  - For product innovation rates, each additional barrier reduces the effect by 2.5 percentage points (0.05 SD).
  - For the typical firm facing five barriers, the net effect on product innovation is effectively null.
- Robustness:
  - Results hold across pre-treatment size-restricted samples and when excluding firms that ever produced de-reserved products.
  - No single barrier drives results; excluding any individual barrier from the count does not change conclusions.

### Impacts on firm growth and how barriers modify them (Table 7)
- For firms that face no barriers, credit substantially boosts growth:
  - Sales increase by 98%.
  - Profits increase by 72%.
- Each additional barrier reduces the growth impact:
  - Each additional barrier reduces the impact of credit on sales by 13% and on profits by 8%.
- For the median newly eligible firm (five barriers):
  - Still experiences substantial increases in sales and profits of 20–30% and increases use of raw materials and labor.
  - However, due to stronger dilution effects on innovation and investment, credit has no treatment effect on innovation and investment for the median firm.

### Interpretation and policy implications
- Heterogeneous effects: credit expands production and profits broadly, but only enables product innovation for firms facing few or no non-financial barriers.
- Mechanism:
  - Many firms operate below efficient scale and respond to additional credit by intensifying variable inputs and utilizing existing capacity rather than launching new or more complex products.
  - Overlapping non-financial barriers (input and output market frictions, unreliable electricity, poor infrastructure, legal congestion, small/uncompetitive markets) weaken incentives or ability to innovate.
- Policy recommendation (high-level):
  - Unlocking innovation requires addressing multiple, overlapping constraints beyond finance.
  - A coordinated big-push strategy that combines credit support with interventions reducing non-financial barriers can increase the effectiveness of credit programs in fostering product innovation.

*Source: 3.1   Stylised Facts on Innovation (wpiea2025192-source-pdf).*

### Appendix Table C.5 shows that firms that face no barriers make investments in both buildings

### wpiea2025192-source-pdf - Appendix Table C.5 shows that firms that face no barriers make investments in both buildings

### Findings
- Appendix Table C.5 shows that firms that face no barriers make investments in both buildings and plant & equipment.

### Key references in source
- Appendix Table C.5
- 21

*Source: wpiea2025192-source-pdf - Appendix Table C.5 shows that firms that face no barriers make investments in both buildings*

### References

### wpiea2025192-source-pdf - References

### References cited
- Extensive bibliography covering finance, innovation, development, and firm-level studies. Representative authors and works include:
  - Acharya, V., & Xu, Z. (2017). Financial dependence and innovation: The case of public versus private firms. Journal of Financial Economics, 124(2), 223–243.
  - Aghion, P., & Howitt, P. (1992). A model of growth through creative destruction. Econometrica, 60(2), 323–351.
  - Akcigit, U., & Kerr, W. R. (2018). Growth through heterogeneous innovations. Journal of Political Economy, 126(4), 1339–1783.
  - Allcott, H., Collard-Wexler, A., & O’Connell, S. D. (2016). How do electricity shortages affect industry? Evidence from India. American Economic Review, 106(3), 587–624.
  - Banerjee, A. V., & Duflo, E. (2014). Do firms want to borrow more? Testing credit constraints using a directed lending program. Review of Economic Studies, 81(2), 572–607.
  - Bloom, N., Eifert, B., Mahajan, A., McKenzie, D., & Roberts, J. (2013). Does management matter? Evidence from India. The Quarterly Journal of Economics, 128(1), 1–51.
  - Duval, R., Hong, G. H., & Timmer, Y. (2020). Financial frictions and the great productivity slowdown. The Review of Financial Studies, 33(2), 475–503.
  - Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences, 106(26).
  - Khan, and many additional field- and macro-level studies, historical analyses, NBER working papers, and methodological contributions (complete list of citations appears in the source).

### Figures and illustrative examples (selected notes)
- Figure 1: Product Innovation — Example
  - Illustrative firm switching from production of Ingot, Iron/Steel to Bars Rods & Rounds, Iron/Steel.
- Figure 2: Product Complexity — Example
  - Panel A: A firm originally manufacturing Unwrought aluminum introduces Plates, sheets, and strip of aluminum...; the new product assigned a Product Complexity Index (PCI) of 0.08 and classified as a Product Complexity Innovation.
  - Panel B: Lists five products with highest PCI and five with lowest PCI observed. Examples include:
    - Top Product Complexity: 38942 Photographic plates and film, exposed and developed, other than cinematographic film; 48315 Liquid crystal devices n.e.c.; lasers, except laser diodes; 43123 Compression-ignition internal combustion piston engines...
    - Bottom Product Complexity: 26170 Jute and other textile bast fibres...; 01922 Jute, kenaf, and other textile bast fibres, raw or retted...; 01141 Sorghum/Jowar, seed.
- Figure 3: Distribution of Market Barriers
  - Ten non-financial barriers considered. The median firm faces five barriers.
- Figure 4: Difference in Difference coefficient plot
  - Dynamic estimates of equation (1) on natural log of sales, materials, credit, and innovation measures. Year 2005 omitted; vertical line between 2005 and 2006 marks policy change. Outcomes winsorized at 1st and 99th percentiles; 90 percent confidence intervals; SE clustered at firm level.
- Figure 5: Impact of eligibility reform on product innovation and sales by barrier count
  - Estimates evaluated at barrier counts of 0, 3, and the sample median (5). Barrier counts range from 0–10. Specification: linear regressions with state×year, industry×year, and firm fixed effects; inverse sampling-probability weights; clustered SE at firm level; outcomes winsorized at 1st and 99th percentiles.

### Key summary statistics (Table 1, Panel A & B; values preserved exactly)
- Panel A: Dependent variables (real 2005 Rupees; winsorized at 1st and 99th percentiles; weighted)
  - Product Scope: Mean 1.64, P25 1.00, P50 1.00, P75 2.00, SD 1.18, N 48,829
  - Product Innovation (in %): Mean 47%, P25 0%, P50 0%, P75 100%, SD 50%, N 48,829
  - Innovation Sales Share (in %): Mean 33%, P25 0%, P50 0%, P75 100%, SD 45%, N 48,829
  - Product Complexity (in %): Mean 12%, P25 0%, P50 0%, P75 0%, SD 33%, N 44,577
  - Complex-Innovation Sales Share (in %): Mean 8%, P25 0%, P50 0%, P75 0%, SD 33%, N 44,577
  - Outstanding Loans (in millions): Mean 13.02, P25 0.75, P50 3.27, P75 11.83, SD 28.79, N 48,829
  - Overdraft (in millions): Mean 12.48, P25 0.58, P50 2.70, P75 10.64, SD 28.49, N 48,829
  - Credit (in millions): Mean 26.01, P25 12.13, P50 7.64, P75 25.54, SD 52.15, N 48,829
  - Materials Consumed (in millions): Mean 66.67, P25 4.41, P50 18.57, P75 66.42, SD 131.26, N 48,677
  - Days Worked (in 000s): Mean 27.78, P25 4.01, P50 10.86, P75 32.99, SD 45.30, N 48,783
  - Workers’ Wages (in millions): Mean 2.82, P25 0.27, P50 0.84, P75 2.87, SD 5.46, N 48,608
  - Investment (in millions): Mean 2.35, P25 0.16, P50 0.66, P75 2.59, SD 4.15, N 48,801
  - Sales (in millions): Mean 96.89, P25 7.01, P50 27.66, P75 100.62, SD 181.05, N 48,739
  - Gross Value Added (in millions): Mean 14.49, P25 1.06, P50 3.86, P75 14.26, SD 30.47, N 48,749
  - Net Income (in millions): Mean 9.97, P25 0.43, P50 2.04, P75 8.78, SD 25.50, N 48,724
  - Managerial Wages (in millions): Mean 1.50, P25 0.12, P50 0.39, P75 1.33, SD 3.30, N 44,516
  - Survey Weight: Mean 4.45, P25 1.00, P50 4.20, P75 5.44, SD 4.66, N 48,829
  - Newly Eligible: 24%, N 48,829
- Panel B: Market barriers (selected)
  - Education (in %): Mean 66%, P25 57%, P50 66%, P75 76%, SD 15%, N 43,230
  - Electricity Shortage (in %): Mean 10%, P25 3%, P50 9%, P75 19%, SD 7%, N 48,829
  - Import Reliance (in %): Mean 18%, P25 0%, P50 0%, P75 0%, SD 39%, N 48,829
  - Infrastructure (in %): Mean 75%, P25 55%, P50 82%, P75 94%, SD 22%, N 48,829
  - Labor Regulation: Mean 1.40, P25 0.00, P50 0.00, P75 2.00, SD 4.18, N 43,242
  - Land Access: Mean 3.26, P25 2.00, P50 3.00, P75 4.00, SD 2.33, N 43,242
  - Legal Congestion (Days): Mean 779.44, P25 567.97, P50 791.78, P75 833.71, SD 230.57, N 36,988
  - Market Concentration (HHI): Mean 2420.61, P25 712.10, P50 1663.99, P75 3525.64, SD 2230.34, N 48,829
  - Market Size (Percentile): Mean 73.45, P25 59.09, P50 76.26, P75 91.38, SD 19.89, N 48,829
  - Rural (in %): Mean 66%, P25 0%, P50 100%, P75 100%, SD 47%, N 48,829
  - Barrier Count: Mean 5.17, P25 4.00, P50 5.00, P75 6.00, SD 1.70, N 33,094

### Selected estimation results (preserved point estimates and standard errors)
- Table 2: Impact of Credit Access Reform on Firm Borrowing (difference-in-difference; state×year, industry×year, firm FE)
  - Post x Newly Eligible:
    - Outstanding Loans (Log): 0.12 (0.07) *
    - Overdraft (Log): 0.21 (0.07) ***
    - Credit (Log): 0.22 (0.07) ***
  - R2: 0.69 / 0.70 / 0.73; Firm-years 48,799; Firms 10,939
- Table 3: Impact of Credit Access Reform on Innovation (difference-in-difference)
  - Post x Newly Eligible:
    - Product Scope: 0.003 (0.024)
    - Product Innovation: -0.005 (0.013)
    - Product Complexity: -0.015 (0.011)
    - Innovation Sales Share: 0.017 (0.012)
    - Complex-Innovation Sales Share: 0.003 (0.008)
  - Means: Product Scope 1.71; Product Innovation 0.44; Product Complexity 0.11; Innovation Sales Share 0.30; Complex-Innovation Sales Share 0.07
  - R2 range: 0.43–0.77; Firm-years and Firms reported per outcome.
- Table 4: Impact of Credit Access Reform on Sales and Profits
  - Post x Newly Eligible:
    - Sales (Log): 0.28 (0.08) ***
    - Gross Value Added (Log): 0.24 (0.07) ***
    - Net Income (Log): 0.24 (0.07) ***
    - Managerial Wages (Log): 0.14 (0.05) ***
  - R2 range: 0.74–0.76; Firm-years and Firms reported per outcome.
- Table 5: Impact on Input Use and Investment
  - Post x Newly Eligible:
    - Materials Consumed (Log): 0.26 (0.08) ***
    - Days Worked (Log): 0.28 (0.07) ***
    - Workers’ Wages (Log): 0.18 (0.05) ***
    - Investment (Log): 0.08 (0.06)
- Table 6: Heterogeneous Impact on Product Innovation by Barrier Count (triple difference)
  - Post x Newly Eligible:
    - Product Scope: 0.174 (0.085) **
    - Product Innovation: 0.127 (0.047) ***
    - Product Complexity: 0.079 (0.038) **
    - Innovation Sales Share: 0.132 (0.042) ***
    - Complex-Innovation Sales Share: 0.058 (0.029) **
  - Post x Newly Eligible x Barrier Count:
    - Product Scope: -0.032 (0.017) *
    - Product Innovation: -0.025 (0.009) ***
    - Product Complexity: -0.018 (0.007) **
    - Innovation Sales Share: -0.020 (0.008) **
    - Complex-Innovation Sales Share: -0.010 (0.005) *
  - Median Effect (reported): Product Innovation -0.073 (and other median effects for outcomes)
  - Firm-years and Firms reported per outcome.

### Appendix and robustness highlights
- Online Appendix includes:
  - Data Appendix with Figure A.1 (evolution of priority sector lending, 2000–2010; amounts deflated to 2005 USD) and Figure A.2 (product data coverage comparison to scanner data).
  - Table A.1: Illustrative innovation examples (Panel A: Product Innovation; Panel B: Product Complexity) with specific product code examples and years (2002–2010).
  - Table A.2: Sample filters; final sample: Firms 10,942, Firm-Years 48,835.
  - Table A.3: Definitions of market barriers (Education, Electricity Shortage, Import Reliant, Infrastructure, Labor Regulation, Land Access, Legal Congestion, Market Concentration, Market Size, Rural).
  - Table A.5: Definitions of dependent variables (Outstanding Loans, Overdraft, Credit, Sales, Gross Value Added, Net Income, Managerial Wages, Materials, Days Worked, Workers’ Wages, Investment, Product Scope, Product Innovation, Innovation Sales Share, Product Complexity, Complex-Innovation Sales Share).
- Robustness and heterogeneous analyses:
  - Narrow sample checks (Table B.1), exclusion of de-reserved products (Table B.2), heterogeneous effects by market frictions (Tables B.3–B.6), granular barrier exclusions and individual barrier interactions (Tables B.5–B.6).
  - Additional results on TFPQ and index specifications (Tables C.1–C.2) and investment by type and heterogeneity by barriers (Tables C.3–C.5). Example index-specification result (Table C.1):
    - Post x Newly Eligible on Credit (Index): 15.95 (4.72) ***
    - Sales (Index): 12.94 (3.65) ***
    - Materials Consumed (Index): 11.12 (3.74) ***
    - Days Worked (Index): 11.30 (1.90) ***
    - Investment (Index): 6.63 (3.76) *
    - Net Income (Index): 28.36 (8.67) ***

*Content compiled from the source section titled "References" and accompanying figures, tables, and appendix notes in the provided PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025192-source-pdf.pdf_