Monica Schuster (joint with Miet Maertens)

Size: px
Start display at page:

Download "Monica Schuster (joint with Miet Maertens)"

Transcription

1 Division of Bioeconomics Do Private Standards Create Exclusive Food Supply Chains? New Evidence from the Peruvian Asparagus Export Sector Monica Schuster (joint with Miet Maertens)

2 Background Modernization of food chains in developing countries Increasing importance of private food standards

3 Background (cont.) Effects of private standards? Two views in the literature: The smallest and poorest farmers are unable to adapt to new requirements and are excluded from the value chain (Reardon et al., 2003; Minot et al., 2004; Blandon et al., 2009; Maertens et al., 2009; Okello et al., 2011 etc.) Farm characteristics are more important than size: modern chains lead to more vertical coordination, i.e., formal contracts, larger access to inputs/ technology and increased investments and productivity (Minten et al., 2006; Gulati et al., 2007; Hernandez et al., 2012 etc.)

4 Background (cont.) Empirical evidence is weak: Little focus on firm/ exporter perspective: existing evidence mainly focuses on farmers/ producers Little control for farmers / households selection effects: comparison of included vs excluded farmers only Lack of micro-level panel data: failure to account for dynamic effects & unobserved heterogeneities

5 Our contribution Peruvian fresh asparagus export sector: 567 export firms, Certification Production standards: Global Gap, Tesco, Leaf, SQF1000 etc. Processing standards: HACCP, BRC, IFS, SQF2000 etc. Sourcing from external producers

6 Exported volume in tons Why Peru & fresh asparagus? Growing high value export chain OTROS 42% CAFÉ VERDE 25% Number of companies M ANGOS FRESCOS 2% PALTAS FRESCAS 3% ALCACHOFAS PREPARADAS 3% UVAS FRESCAS 3% PÁPRIKA ENTERA 3% LECHE EVAPORADA 3% ESPARRAGOS FRESCOS 9% ESPARRAGOS PREPARADOS 7% Source: Sunat Customs Peru, Year Number of Companies Exported Volume Note: Population of 567 export companies over Source: Author's calculation based on SUNAT Custom data, Peru All companies in 2000 All companies in 2011 Mean Std. Dev. min max Mean Std. Dev. min max Certification Number of certificates Production certification Processing certification Source: own survey, July - Sept 2011

7 % Sourced Why Peru & asparagus? Involvement of external producers Volume in Tons % of total exported volume sourced from producers Export volumes produced and sourced - in tons Year Year % Sourced from all external producers % Sourced from small producers(<10ha) Volumes Produced Volumes Sourced - all producers Volumes Sourced - small producers Note: Population of 567 export companies over Source: Author's calculation based on survey data Note: Population of 567 export companies over Source: Author's calculation based on survey data Asparagus Census, 2005: external producers, of which small producers (< 10 ha of land) Distribution

8 Data Fresh asparagus export firm level data from: Secondary sources: Custom database at export transaction level (567 fresh asparagus export firms from ): export volume, value, destinations, export by boat/ plane Tax administration data: foundation date, core activities, general managers, location, branches, fiscal benefits or activity status Primary sources: Company survey (July-Sept 2011) on 87 companies: recall data on certification, land, sourcing strategies, processing, ownership etc.

9 Main company characteristics by certification in 2011 Variables Certified companies in 2011 Non certified companies in 2011 Mean Std. Dev. Mean Std. Dev. General sourcing - dummy (2003) General sourcing - dummy (2011) Small producer sourcing - dummy (2003) Small producer sourcing - dummy (2011) Export Volume Asparagus Land - ha Processing Plant Foreign Capital

10 Evolution of sourcing from first certification Time span: Years from 1st certification % Sourced from external producers % Sourced from small producers (<10ha) Note: Sample of 45 export companies that ever became certified between Source: Author's calculation based on survey data

11 Model specification Analysis at the Yearly level ( ) PercSourced Cert X D + u it 0 1 it 2 it t i it PercSourced it Percentage sourced by company i in year t, either from: - all types of producers - small producers only (<10ha land) Cert it -1 dummy variable if company is certified to at least one certification scheme -2 dummy variables if certified to production/ processing certification -4 dummy variables if certified to low/ high level production/ processing certification - dummy variables for certification to specific type of certification, i.e., HACCP/Global Gap etc. X it Vector of company specific covariates: export volume, geographical location, foreign capital, origin of initial capital, number of establishments, processing plant, hectars of land, changes in management/ staff, changes in company organization, type of company (private/limited liability etc.), enrollment in specific tax regime, multiple tax identification numbers, export experience, green vs white asparagus D t : Year dummies; ε i : Company specific unobservable characteristics; u it : residual

12 Identification 3 sources of endogeneity: 1. Unobserved firm characteristics: time constant unobserved company characteristics evtl. both correlated with the company s sourcing and certification preferences 2. Feed back or anticipation - past sourcing shocks/ behaviors affect the adoption of certification; - anticipation of certification leads companies to change sourcing behavior 3. Simultaneity: time and company specific unobservable shocks simultaneously affecting sourcing and certification decisions. Moreover: Non- linear nature of the dependent variable

13 Identification (cont...) Dep Var: Fraction sourced from all external producers or small producers Controlling for? Model Unobserved heterogeneity? Feed- back/ Anticipation or Simultaneity? Fractional dependent variable? OLS No No No GLM No No Yes Fixed effects Yes No No Difference GMM - Xtabond Yes Yes No

14 Results on general certification dummy Percentage sourced from ALL producers Percentage sourced from SMALL Producers OLS GLM Fixed Difference Fixed Difference OLS GLM Effects GMM Effects GMM (1a) (2a) (3a) (4a) (1b) (2b) (3b) (4b) Certification *** *** * * ** ** ** ** (0.072) (0.053) (0.036) (0.114) (0.048) (0.045) (0.049) (0.048) Processing plant * 0.101** (0.062) (0.058) (0.040) (0.041) (0.064) (0.050) (0.047) (0.052) Lag (total asparagus land) *** *** ** ^ * ^ (0.001) (0.001) (0.001) (0.001) Total asparagus land * ^ (0.001) (0.001) Foreign capital *** * ^ ^ *** ** (0.067) (0.054) (0.042) (0.079) (0.041) (0.040) (0.058) (0.069) [.] Year dummies yes yes yes yes yes yes yes yes Location dummies yes yes - - yes yes - - Other firm covariates yes yes yes yes yes yes yes yes R Obs Number of collapsed IV's nd order autocorr Hansen Difference test Company cluster robust standard errors in parenthesis; *** p<0.01, ** p<0.05, * p<0.1, ^ p<0.15; Average marginal effects (APE) are reported in column 2.

15 Results Production/Processing certifications: Low & high Dep Var: Sourcing from all producers Dep Var: Sourcing from small producers Fixed Effect Diff- GMM Fixed Effect Diff- GMM (1 a) (1 b) (1 c) (1 d) (2 a) (2 b) (2 c) (2 d) Production certification *** ** ** ** (0.057) (0.111) (0.036) (0.044) Processing certification 0.087* 0.141** * (0.051) (0.069) (0.043) (0.067) Production certification: baseline (0.044) (0.456) (0.056) (0.502) Production certification: high level *** ** ** ** (0.054) (0.104) (0.035) (0.055) Processing certification: baseline 0.097^ 0.274* ** (0.064) (0.198) (0.062) (0.115) Production certification: high level * (0.037) (0.088) (0.024) (0.079) Company covariates yes yes yes yes yes yes yes yes Year dummies yes yes yes yes yes yes yes yes R Obs Company cluster robust standard errors in parenthesis; *** p<0.01, ** p<0.05, * p<0.1, ^ p<0.15; Average marginal effects (APE) are reported in columns 2.

16 Results - specific certification schemes Dep Var: Sourcing from all producers Dep Var: Sourcing from small producers Fixed Effects Difference GMM Fixed Effects Difference GMM (1c) (1d) (2c) (2d) Global Gap (high level prod) *** ** *** ** (0.050) (0.084) (0.029) (0.056) HACCP (low level proc) 0.028^ 0.144^ * (0.018) (0.099) (0.058) (0.098) BRC (high level proc) (0.035) (0.087) (0.037) (0.070) BASC (other certif) ^ * (0.036) (0.107) (0.062) (0.062) GMP (low level proc) ^ (0.109) (0.194) (0.104) (0.117) SQF 2000 (high level proc) 0.081** (0.031) (0.221) (0.076) (0.323) Company covariates yes yes yes yes Year dummies yes yes yes yes R Obs Company cluster robust standard errors in parenthesis; *** p<0.01, ** p<0.05, * p<0.1, ^ p<0.15; Average marginal effects (APE) are reported in column 2.

17 Conclusions General certification: sourcing for all types of producers & small producers - Production certification: results driven by HIGH level certifications sourcing for all types of producers & small producers - Processing certification: results driven by LOW level certifications sourcing for all types of producers, but (but results only slightly significant) sourcing from small producers Everything else equal, certification leads to more vertical integration BUT: effect on poverty/ wellbeing? Labour market & socio-economic spill-over effects are crucial!

18 Division of Bioeconomics Thank you for your attention! Questions?

19 Difference GMM - xtabond First difference transformation get rid of unobservable firm characteristics: ΔSource it = β 1 Δ Cert it + Δ X it β 2 + Δ v it IV estimation strategy get rid of endogeneity (anticipation & feed-back) All valid IV s for t=3 for t=4 W i i1 i1 i1 i1 y, C, L, P y, y, C, C L, L, P, P 0 i1 i2 i1 i2 i1 i2 i1 i2 0 0 y... y, C... C, L... L, P... P i1 it 2 i1 it 2 i1 it 2 i1 it 2 back

20 0 0 Fraction Fraction Distribution of farm size At a national level: 70% of all asparagus farmers cultivate < 10ha 90% of all farmers cultivate < 50ha 95 % of all farmers cultivate < 100ha [Source: Agencia agraria Ica & Trujillo, 2011] Land distribution; farmers with <= 100ha Land distribution; farmers with <= 10ha Hectares of asparagus Hectares of asparagus Back