Can agricultural traders be trusted? Evidence from Ethiopia

ETHIOPIAN DEVELOPMENT RESEARCH INSTITUTE

Can agricultural traders be trusted? Evidence from Ethiopia

Authors Thomas W. Assefa and Bart MintenIFPRI ESSP

Ethiopian Economics Association 13th International Conference on the Ethiopian EconomyJuly 23-25, 2015Addis Ababa

1

1. Introduction• Long-standing debate on the appropriate role of the state in the

governance of markets• Markets in developing and developed countries often not trusted and

therefore often argued that there is need for regulation • However, no good empirical evidence whether these trades can or

can’t be trusted and to what extent? • Lack of “Trust” important topic: 1/ public health issues related to

adulteration; 2/ lack of incentives to stick to proper practices; 3/ high search and transaction costs

• Moreover, modern marketing practices [branding, packaging and supermarkets] are emerging to deal with trust issues

1. Introduction• We look at this issue for the case of coffee in Addis • Interesting case because: 1/ large price and quality differentiation2/ government controls; By law, all marketed coffee has to be divided in export and local quality. Only coffee that is of lower quality is supposed to stay in the country. 3/ Emerging presence of modern retail and modern market practices (branding and packing)

1. Introduction• Focus on three main research questions: Question 1: Can we trust traders? Do traders cheat with quality? Do traders cheat with weights?Question 2: Is regulation effective? Question 3: Are modern market practices different and more trust-worthy?

3. Data and methodology• Sample semi-wholesalers: 100 randomly selected from the 240 in

Merkato• Sample retailers:

- 10 sub-cities in Addis: half of them randomly selected (after geographical stratification)

• All coffee traders in all open markets in the 5 sub cities were visited [104]• All supermarkets and minimarkets in the 5 sub cities [97 minimarkets and 53

supermarkets]- 4 kebeles in each sub city from an average of 10 in a sub city are selected randomly

• 10 regular shops from each kebele [200 regular shops]

• 543 coffee traders were surveyed in October 2013

3. Data• Weight assessment:

- Purchase of 1 kg from all semi-wholesalers; half of the open market traders, supermarkets, minimarkets; 25% of regular shops

- 262 obs.: weighted with 2 different electronic scales; average used in analysis

• Quality assessment:- All samples sent to Coffee Liquoring Unit (CLU) for analysis (tasting/raw bean

inspection)

3. Data - Descriptive statisticsRegion of origin

Wholesalers RetailersDon’t know 3% 46%Wollega/Nekempt 32% 6%Djimma 36% 13%Harar 1% 1%Others 28% 6%Not raw coffee 0% 28%

Washing Wholesalers Retailers

Don’t know 1% 10%Washed 23% 13%Unwashed 77% 46%Not raw coffee 0% 30%

FormWholesalers Retailers

Raw 100% 62%Roasted 0% 2%Grounded 0% 35%

PackagingWholesalers Retailers

Packed 0% 58%Loose 100% 42%Branded 0% 40%

4. Traditional markets 4.1. What is valued in coffee markets?

• Definition traditional: 1/ loose formats; 2/ no cash registers and no self-service

• Stated origins of coffee little influence on prices• Washed coffee valued at a premium of 9% compared to natural-

sundried coffee• The more un-pure the coffee, the lower the price• The lower the stated grade, the lower the price• When we use measured grades of CLU, no impact on prices

4. Traditional markets 4.1. What is valued in coffee markets?

• Overall, rewards to easily observable quality measures; few to not easily observables

• Comparison with formal export markets: - Large differences in premiums for origins and measured grades of CLU • Seemingly a dissipation of the not easily observable quality premiums

in local markets• Why? Lack of trust? Lack of knowledge?

4. Traditional markets 4.2. Assessing trust

• Cheating with weights? Very little.

Semi-

wholesalers Traditional retail markets Loose Unit Regular shop Open market Total products

Number of observations 100 44 51 95 202Mean grams 992.6 1004.2 1002.8 1,003.4 998.2Median grams 991.5 1004.8 1001.5 1,002.5 996.5Underweight % 75 36 43 40 57Overweight % 25 64 57 60 43

T-test if weight is t-value* -5.42 2.09 1.48 2.51 -1.69diff. than 1 kg Pr(|T| > |t|) 0.00 0.04 0.15 0.01 0.09


• Cheating with quality? Yes with not easily observables; No with easily observables

Semi- Traditional retail markets Loose

wholesalers Regular shopOpen

market Total productsStatements originUnderstated % 21 24 15 20 14Match % 13 5 10 7 8Overstated % 66 71 75 73 78Total % 100 100 100 100 100WashingUnderstated % 3 2 6 4 3Match % 91 89 90 90 89Overstated % 6 9 4 6 8Total % 100 100 100 100 100Number of observations 100 44 51 95 202


• Cheating with export quality coffee? Yes Semi- Traditional retail markets Loose

wholesalers Regular shopOpen

market Total productsOverall quality assessmentFit for grade 2 % 16 0 4 2 9Fit for grade 3 % 1 2 0 1 1Fit for grade 4 % 0 2 2 2 1Fit for grade 5 % 4 2 8 5 4Fit for Peaberry coffee % 2 0 0 0 1Rejected for grades (but > under-grade) % 41 52 39 45 42Fit at under-grade level % 33 32 31 32 33Unfit for export % 3 9 16 13 8Total % 100 100 100 100 100

5. Modern marketing practices 5.1. Typology

• Two modern marketing practices emerging:1/ Modern retail: - Becoming very important in developing countries- Despite prohibition of FDI in retail in Ethiopia, domestic modern

sector emerging2/ Packaging and branding:- Unpacked and unbranded products indistinguishable from

competitors- Brandings adds “brand value”


• Significant quality premiums for modern retail

0.0

1.0

2.0

3D

ensi

ty

0 50 100 150 200Birr/kg

modern retail traditional retailwholesale


• Significant quality premiums for packing and branding

0.0

1.0

2.0

3D

ensi

ty

0 50 100 150 200Birr/kg

branded bags retail unbranded bags retailloose


• In hedonic pricing model; impact modern practices:- Supermarkets 33% more expensive- Mini-markets 8% more expensive- Branded bean bags 23% more expensive than “unpure” loose coffee- Unbranded bean bags 16% more expensive than “unpure” loose

coffee- Grounded coffee (all branded) 34% more expensive


• In hedonic pricing model of modern markets regression: - No quality premiums for origin- Quality premium for washing (23%)- No premiums for measured grades, except the worst ones that are

valued lower

5. Modern marketing practices 5.2. Assessing trust

• Cheating with weights? Yes, but very little.

Modern retail Packed

Unit Supermarkets Mini-markets Total Branded Non-branded Total

Number of observations 26 41 67 13 47 60

Mean grams 989.1 994.2 992.2 990.9 990.4 990.5

Median grams 994.5 997.5 995.5 993.5 996.5 995.5

Underweight % 73 63 67 85 66 70

Overweight % 27 37 33 15 34 30T-test if weight is t-value** -1.48 -2.02 -2.33 -2.35 -2.24 -2.75

diff. than 1 kg Pr(|T| > |

t|) 0.15 0.05 0.02 0.04 0.03 0.01

• Cheating with quality? Yes with not easily observables; No with easily observables

Modern retail Packed

Super-

markets Mini-markets Total BrandedNon-

branded TotalStatements originUnderstated % 20 13 15 50 11 14Match % 0 0 0 0 0 0Overstated % 80 87 85 50 89 86Total % 100 100 100 100 100 100WashingUnderstated % 6 0 2 0 3 2Match % 89 91 91 100 95 96Overstated % 6 9 7 0 2 2Total % 100 100 100 100 100 100Number of observations 26 41 67 13 47 60


• Cheating with export quality coffee? Yes. But also high quality. Modern retail Packed

Super-

marketsMini-

markets Total BrandedNon-

branded TotalOverall quality assessmentFit for grade 2 % 50 15 28 46 28 32Fit for grade 3 % 0 2 1 0 0 0Fit for grade 4 % 0 10 6 0 6 5Fit for grade 5 % 8 2 4 0 6 5Fit for Peaberry coffee % 0 0 0 0 0 0Rejected for grades (but >UG) % 23 22 22 38 21 25Fit at under-grade level % 15 39 30 15 32 28Unfit for export % 4 10 7 0 6 5Total % 100 100 100 100 100 100


6. ConclusionsMajor findings from the research:Q1: Can we trust traditional traders? Answer: Depends. Can be relatively trusted with weights; On quality: 1/ Quality indicators that are not easily observable not rewarded (origins of coffee); 2/ Indicators that easily observable rewarded (ECX reject cheaper than others; washed and pure coffee higher prices; packed and branded coffee higher prices) Q2: Is regulation effective? Answer: No. There is a flourishing informal market Q3: What is different with modern markets? Answer: Deliver high quality and more processed products at a high price but no more trust-worthy than traditional markets

7. Implications 1. Markets not to be trusted in non-observables. More adapted

market institutions needed.

2. Flourishing illegal market. Liberalize? Yes, but maybe not completely (especially on regional indicators)

3. Modern markets. Strong heterogeneity. Special situation? Informal markets and early roll-out of modern retail. Possible that FDI and reduction of informality would solve some of the issues.

Thank You

Government & Nonprofit

Can agricultural traders be trusted? Evidence from Ethiopia