## 5. Markups Before/After 2017m1

## Source details

**Canonical URL:** [5. Markups Before/After 2017m1](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020115-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020115-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020115-print-pdf.pdf.json)

---

### I. Introduction — purpose and context
- Market integration matters for equity and efficiency; food is a large share of consumption in India (Food and beverages account for 45.9 percent of the consumer price index (CPI) basket in India).
- Recent policy initiatives aimed at promoting agricultural market integration:
  - electronic National Agriculture Market (eNAM) platform (promote uniformity, real time price discovery, pan-India trade).
  - suggested revisions to APMC laws (model Agriculture Produce and Livestock Marketing Act) to loosen restrictions on sale/trade.
  - infrastructure programs including Bharatmala Pariyojana.
  - May 2020 economic relief package reforms: amendments to the Essential Commodities Act and agriculture-marketing liberalization measures.
- Paper goals:
  - Use monthly prices for 21 commodities across up to 60 mandis to measure cross-market price differentials (markups) and their evolution over time.
  - Document proximate determinants of agricultural market integration for five staples: onions; potatoes; rice; tomatoes; wheat.

### II. Stylized facts from the price dataset
- Data: monthly prices of 21 agricultural goods and 60 mandis (Indian Department of Consumer Affairs, via CEIC).
- Key stylized facts:
  - Price-dispersion heterogeneity across goods:
    - Tomatoes and onions: periodic/seasonal common spikes across markets, low cross-market dispersion.
    - Loose tea and various oils: much greater cross-market dispersion.
    - Less perishable goods (rice, packed salt, wheat): India-wide mean price comparatively stable with market-specific fluctuations.
  - Some mandis have systematically higher prices irrespective of commodity:
    - High markups: Port Blair (Andaman and Nicobar Islands) and Aizawl (Mizoram).
    - Low markups: Jaipur (most goods priced at a discount vs India-wide averages).
  - Bivariate lowess relationships (tomato example; similar for other staples):
    - Positive correlation between cross-market markup and distance (in km).
    - Negative relationship between markups and average road density (road km per 100 square km) for low-to-moderate densities; relationship fades at higher densities (diminishing returns).
    - Negative relationship between markups and average rail density (rail km per 1,000 square km).
  - Road and rail densities are highly correlated across market-pair observations.
  - Markups across the five staple goods are generally highly correlated, indicating market-specific factors.

### III. Empirical analysis
- Sample period for staple-commodity regressions: 2009m1 through 2018m12.
- Dependent variable: MKP_{ij,c} = cross-market price difference (markup) for commodity c between markets i and j; markups computed as (price_{i,t} − price_{j,t})/((price_{i,t}+price_{j,t})/2).
- Baseline cross-section specification (estimated separately for each commodity):
  MKP_{ij,c} = α_c + β1 DIS_{ij} + β2 TRANS_{ij} + β3 URBAN_{ij} + β4 SGDP_{ij} + γ_{c,t} + ε_{i,t,c}
  - DIS_{ij}: distance between markets (log kms).
  - TRANS_{ij}: average of road and rail density measures for the two markets (log units).
  - URBAN_{ij}: percentage of population classified as urban.
  - SGDP_{ij}: log level of state real GDP (average across both markets).
- Cross-section regression results (Table 2):
  - DIS (i,j) coefficients (coefficient and robust standard error):
    - Onion: 0.038*** (0.004)
    - Potato: 0.070*** (0.004)
    - Rice: 0.028*** (0.003)
    - Tomato: 0.085*** (0.004)
    - Wheat: 0.116*** (0.005)
  - TRANS (i,j) coefficients:
    - Onion: -0.068*** (0.008)
    - Potato: -0.015** (0.007)
    - Rice: -0.010** (0.004)
    - Tomato: -0.064*** (0.009)
    - Wheat: -0.009 (0.007)
  - URBAN (i,j) coefficients:
    - Onion: 0.109*** (0.012)
    - Potato: 0.104*** (0.011)
    - Rice: 0.029*** (0.008)
    - Tomato: 0.125*** (0.014)
    - Wheat: 0.089*** (0.012)
  - SGDP (i,j) coefficients:
    - Onion: -0.031*** (0.005)
    - Potato: -0.002 (0.005)
    - Rice: -0.000 (0.004)
    - Tomato: -0.043*** (0.006)
    - Wheat: 0.027*** (0.006)
  - Observations:
    - Onion 1,540; Potato 1,540; Rice 1,540; Tomato 1,540; Wheat 1,403.
  - R-squared:
    - Onion 0.171; Potato 0.180; Rice 0.058; Tomato 0.325; Wheat 0.342.
- Interpretation and illustrative magnitudes:
  - Distance: a one-standard-deviation increase in distance between markets (0.73 log units) is roughly equivalent to a 500km increase (assessed at the average distance in sample). Example: for tomatoes this implies an increase in the markup of about 0.06, ~15 percent of the average tomato markup (~0.4).
  - Transportation infrastructure: a one-standard-deviation increase in transport density (0.64 unit) would reduce the tomato-price markup by about 0.04, ~10 percent of the average tomato markup, ceteris paribus.
  - Urbanization: higher urbanization is associated with higher markups in regressions; included as a control but coefficient positive and significant across staples.
  - State GDP: coefficients mixed (negative for onion and tomato, positive for wheat), primarily used as control.
- Panel regressions (random effects; Table 4) corroborate cross-section findings:
  - DIS coefficients remain positive and significant (examples with coefficient and standard error):
    - Onion 0.050*** (0.005)
    - Tomato 0.085*** (0.004)
    - Wheat 0.114*** (0.005)
  - TRANS coefficients remain negative and significant for several commodities (example: Tomato -0.078*** (0.010)).
  - URBAN coefficients remain positive and significant in panel as well.
  - Observations in panel: large sample (example: Tomato 62,463 observations; number of pairs 1,475).
  - Hausman testing shows no systematic difference between random- and fixed-effects specifications.

### IV. Evolution of market integration over time (before/after 2017m1)
- Finding: Price dispersion between mandis has generally increased since early 2017 for 4 of the 5 staples considered.
- Regressions of markups vis-à-vis the nationwide average with a post-2017m1 dummy (Table 5) — post-2017m1 coefficients (coefficient and robust standard error):
  - Tomatoes: 0.006** (0.002)
  - Onions: 0.002 (0.003)
  - Potatoes: 0.054*** (0.003)
  - Rice: 0.006* (0.004)
  - Wheat: 0.007** (0.003)
- Sample sizes for pre/post analysis:
  - Tomatoes 133,697 observations; Onions 146,014; Potatoes 145,824; Rice 146,069; Wheat 104,314.
- Distributional comparison:
  - Kernel/empirical distributions of markups before vs after 2017m1 show considerable similarity; no clear narrowing of distributions after reforms.
  - Two-sample Kolmogorov-Smirnov test does not provide strong evidence of distributional change (test has low power and sensitivity to outliers per Gail and Green (1976)).
- Caveats and alternative explanations:
  - Choice of 2017m1 as cutoff is somewhat arbitrary; results robust to shifting cutoff by up to a year.
  - Post-2017 increase in dispersion could reflect short sample period, effects of demonetization, or record production years in 2016-17 and 2017-18 for some commodities.
  - Identification of reform effects may strengthen with a longer post-reform sample.

### V. Key findings and policy implications
- Empirical findings:
  - Cross-market price markups are positively associated with distance (DIS) and urbanization (URBAN).
  - Cross-market price markups are negatively associated with transportation infrastructure density (TRANS), with evidence of diminishing returns in road density.
  - Infrastructure and geographical proximity are primary determinants; market-specific effects are important (many mandis consistently high or low).
  - Agricultural market integration has not increased in recent years (no robust evidence of reduced cross-market markups after 2017m1).
- Policy implications:
  - Prioritize infrastructure investments that facilitate cross-market trade (roads, rail) to reduce markups.
  - Complement infrastructure with policies that reduce barriers to inter-state movement of goods and improve information sharing (consistent with literature and Reserve Bank of India (2019)).
  - Reforms such as eNAM and APMC-law flexibility are in the right direction but may require time, broader adoption, or complementary measures to reduce persistently high markups.
  - Further evaluation should use longer post-reform samples to assess the effectiveness of recent liberalization measures.

*Source: Authors’ calculations; Indian Department of Consumer Affairs, CEIC; content unit: 5. Markups Before/After 2017m1.*

### 1. Standardized Prices...................................................................................7

### 1. Standardized Prices...................................................................................7

### Standardized Prices (section)
- Section title: "Standardized Prices".
- Located on page 7.

### Distribution of Markups with Respect to Nation-Wide Average (section)
- Section title: "Distribution of Markups with Respect to Nation-Wide Average".
- Located on page 8.

### Selected LOWESS Smoother Plots (section)
- Section title: "Selected LOWESS Smoother Plots".
- Located on page 10.

### Distribution of Cross-Market Price Markups Before/After 2017m1 (section)
- Section title: "Distribution of Cross-Market Price Markups Before/After 2017m1".
- Located on page 15.

### Tables (list of tables included)
- Table 1. Correlation Matrix....................................................................................9
- Table 2. Cross-Section Regressions..........................................................................12
- Table 3. Variable Mean and Standard Deviations...........................................................12
- Table 4. Panel Regressions.....................................................................................13

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020115-print-pdf.pdf*

### 5. Markups Before/After 2017m1.....................................................................14

### 5. Markups Before/After 2017m1.....................................................................14

### I. Introduction — purpose and context
- Market integration matters for equity and efficiency; food is a large share of consumption in India (Food and beverages account for 45.9 percent of the consumer price index (CPI) basket in India).
- Recent policy initiatives aimed at promoting agricultural market integration:
  - electronic National Agriculture Market (eNAM) platform (promote uniformity, real time price discovery, pan-India trade).
  - suggested revisions to APMC laws (model Agriculture Produce and Livestock Marketing Act) to loosen restrictions on sale/trade.
  - infrastructure programs including Bharatmala Pariyojana.
  - May 2020 economic relief package reforms: amendments to the Essential Commodities Act and agriculture-marketing liberalization measures.
- Paper goals:
  - Use monthly prices for 21 commodities across up to 60 mandis to measure cross-market price differentials (markups) and their evolution over time.
  - Document proximate determinants of agricultural market integration for five staples: onions; potatoes; rice; tomatoes; wheat.

### II. Stylized facts from the price dataset
- Data: monthly prices of 21 agricultural goods and 60 mandis (Indian Department of Consumer Affairs, via CEIC).
- Key stylized facts:
  - Price-dispersion heterogeneity across goods:
    - Tomatoes and onions: periodic/seasonal common spikes across markets, low cross-market dispersion.
    - Loose tea and various oils: much greater cross-market dispersion.
    - Less perishable goods (rice, packed salt, wheat): India-wide mean price comparatively stable with market-specific fluctuations.
  - Some mandis have systematically higher prices irrespective of commodity:
    - High markups: Port Blair (Andaman and Nicobar Islands) and Aizawl (Mizoram).
    - Low markups: Jaipur (most goods priced at a discount vs India-wide averages).
  - Bivariate lowess relationships (tomato example; similar for other staples):
    - Positive correlation between cross-market markup and distance (in km).
    - Negative relationship between markups and average road density (road km per 100 square km) for low-to-moderate densities; relationship fades at higher densities (diminishing returns).
    - Negative relationship between markups and average rail density (rail km per 1,000 square km).
  - Road and rail densities are highly correlated across market-pair observations.
  - Markups across the five staple goods are generally highly correlated, indicating market-specific factors.

### III. Empirical analysis
- Sample period for staple-commodity regressions: 2009m1 through 2018m12.
- Dependent variable: MKP_{ij,c} = cross-market price difference (markup) for commodity c between markets i and j; markups computed as (price_{i,t} − price_{j,t})/((price_{i,t}+price_{j,t})/2).
- Baseline cross-section specification (estimated separately for each commodity):
  MKP_{ij,c} = α_c + β1 DIS_{ij} + β2 TRANS_{ij} + β3 URBAN_{ij} + β4 SGDP_{ij} + γ_{c,t} + ε_{i,t,c}
  - DIS_{ij}: distance between markets (log kms).
  - TRANS_{ij}: average of road and rail density measures for the two markets (log units).
  - URBAN_{ij}: percentage of population classified as urban.
  - SGDP_{ij}: log level of state real GDP (average across both markets).
- Cross-section regression results (Table 2): estimated coefficients (robust standard errors in parentheses; significance levels preserved)
  - DIS (i,j):
    - Onion: 0.038*** (0.004)
    - Potato: 0.070*** (0.004)
    - Rice: 0.028*** (0.003)
    - Tomato: 0.085*** (0.004)
    - Wheat: 0.116*** (0.005)
  - TRANS (i,j):
    - Onion: -0.068*** (0.008)
    - Potato: -0.015** (0.007)
    - Rice: -0.010** (0.004)
    - Tomato: -0.064*** (0.009)
    - Wheat: -0.009 (0.007)
  - URBAN (i,j):
    - Onion: 0.109*** (0.012)
    - Potato: 0.104*** (0.011)
    - Rice: 0.029*** (0.008)
    - Tomato: 0.125*** (0.014)
    - Wheat: 0.089*** (0.012)
  - SGDP (i,j):
    - Onion: -0.031*** (0.005)
    - Potato: -0.002 (0.005)
    - Rice: -0.000 (0.004)
    - Tomato: -0.043*** (0.006)
    - Wheat: 0.027*** (0.006)
  - Observations: Onion 1,540; Potato 1,540; Rice 1,540; Tomato 1,540; Wheat 1,403.
  - R-squared: Onion 0.171; Potato 0.180; Rice 0.058; Tomato 0.325; Wheat 0.342.
- Interpretation and illustrative magnitudes:
  - Distance: a one-standard-deviation increase in distance between markets (0.73 log units) is roughly equivalent to a 500km increase (assessed at the average distance in sample). Example: for tomatoes this implies an increase in the markup of about 0.06, ~15 percent of the average tomato markup (~0.4).
  - Transportation infrastructure: a one-standard-deviation increase in transport density (0.64 unit) would reduce the tomato-price markup by about 0.04, ~10 percent of the average tomato markup, ceteris paribus.
  - Urbanization: higher urbanization is associated with higher markups in regressions; included as a control but coefficient positive and significant across staples.
  - State GDP: coefficients mixed (negative for onion and tomato, positive for wheat), primarily used as control.
- Panel regressions (random effects; Table 4) corroborate cross-section findings:
  - DIS coefficients remain positive and significant (e.g., Onion 0.050*** (0.005); Tomato 0.085*** (0.004); Wheat 0.114*** (0.005)).
  - TRANS coefficients remain negative and significant for several commodities (e.g., Tomato -0.078*** (0.010)).
  - URBAN coefficients remain positive and significant in panel as well.
  - Observations in panel: large sample (e.g., Tomato 62,463 observations; number of pairs 1,475).
  - Hausman testing shows no systematic difference between random- and fixed-effects specifications.

### IV. Evolution of market integration over time (before/after 2017m1)
- Finding: Price dispersion between mandis has generally increased since early 2017 for 4 of the 5 staples considered.
- Regressions of markups vis-à-vis the nationwide average with a post-2017m1 dummy (Table 5):
  - Post-2017m1 coefficients (markup pre/post 2017m1):
    - Tomatoes: 0.006** (0.002)
    - Onions: 0.002 (0.003)
    - Potatoes: 0.054*** (0.003)
    - Rice: 0.006* (0.004)
    - Wheat: 0.007** (0.003)
  - Sample sizes: Tomatoes 133,697 observations; Onions 146,014; Potatoes 145,824; Rice 146,069; Wheat 104,314.
- Distributional comparison:
  - Kernel/empirical distributions of markups before vs after 2017m1 show considerable similarity; no clear narrowing of distributions after reforms.
  - Two-sample Kolmogorov-Smirnov test does not provide strong evidence of distributional change (test has low power and sensitivity to outliers per Gail and Green (1976)).
- Caveats and alternative explanations:
  - Choice of 2017m1 as cutoff is somewhat arbitrary; results robust to shifting cutoff by up to a year.
  - Post-2017 increase in dispersion could reflect short sample period, effects of demonetization, or record production years in 2016-17 and 2017-18 for some commodities.
  - Identification of reform effects may strengthen with a longer post-reform sample.

### V. Key findings and policy implications
- Empirical findings:
  - Cross-market price markups are positively associated with distance (DIS) and urbanization (URBAN).
  - Cross-market price markups are negatively associated with transportation infrastructure density (TRANS), with evidence of diminishing returns in road density.
  - Infrastructure and geographical proximity are primary determinants; market-specific effects are important (many mandis consistently high or low).
  - Agricultural market integration has not increased in recent years (no robust evidence of reduced cross-market markups after 2017m1).
- Policy implications (directly implied by results and discussion):
  - Prioritize infrastructure investments that facilitate cross-market trade (roads, rail) to reduce markups.
  - Complement infrastructure with policies that reduce barriers to inter-state movement of goods and improve information sharing (consistent with literature and Reserve Bank of India (2019)).
  - Reforms such as eNAM and APMC-law flexibility are in the right direction but may require time, broader adoption, or complementary measures to reduce persistently high markups.
  - Further evaluation should use longer post-reform samples to assess the effectiveness of recent liberalization measures.

*Source: Authors’ calculations; Indian Department of Consumer Affairs, CEIC; content unit: 5. Markups Before/After 2017m1.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020115-print-pdf.pdf_