Open Access

Valuation of traits of indigenous sheep using hedonic pricing in CentralEthiopia

  • Zelalem G Terfa1Email author,
  • Aynalem Haile2,
  • Derek Baker3 and
  • Girma T Kassie4
Agricultural and Food Economics20131:6

https://doi.org/10.1186/2193-7532-1-6

Received: 19 December 2012

Accepted: 8 July 2013

Published: 20 August 2013

Abstract

This study estimates the implicit prices of indigenous sheep traits based onrevealed preferences. A hedonic pricing model is fitted to examine thedeterminants of observed sheep prices. Transaction data were generated fromrural markets of Horro-Guduru Wollega Zone of Ethiopia. Both OLS andheteroscedasticity consistent estimations were made. The empirical resultsconsistently indicate that phenotypic traits of traded indigenous sheep (age,color, body size, and tail condition) are major determinants of price implyingthe importance of trait preferences in determining the price of sheep in localmarkets. Season and market locations are also very important price determinantssuggesting the need to target season and market place in sheep improvementprogrammes. Therefore, the development of a comprehensive breeding program thathas marketing element is crucial to make sheep improvement sustainable and sheepkeepers benefit from the intervention.

Keywords

Hedonic pricing Heteroscedasticity consistent Phenotypic Indigenous Trait preference

Background

Small ruminants are a key component of the rural livelihood systems in ruralEthiopia. It is estimated that in 2010 Ethiopia owned about 48 million smallruminants (FAOSTAT, 2010) and this is one of the largestpopulations in sub Saharan Africa (SSA). Small ruminants contribute substantially toincome, food (meat and milk), and non-food products like manure, skins and wool.They also serve as part of the crop failure risk coping portfolio of enterprises,for asset wealth security as form of money saving and investment as well as manyother cultural functions (Tibbo, 2006). At farm householdlevel, sheep contribute up to 63% of the net cash income derived from livestockproduction in the crop-livestock production systems in Ethiopia. In the dry lowlandsof the country, sheep play a key role in sustaining the livestock-based pastoral andagro-pastoral livelihoods (Negassa and Jabbar, 2008). Despitethe pronounced importance of small ruminants in general and sheep in particular, theproductivity of the animals per head is considerably low. FAO (2009) estimated the average annual off-take rate and carcass weight perslaughtered animal for the period 2000 to 2007 to be 32.5% and 10.1 kg,respectively, the lowest even among the sub Saharan African countries. In fact, inthe highlands of the country, sheep off-take was found to be even lower at 7%(Negassa and Jabbar, 2008).

In the study area (Horro-Guduru Wollega Zone of Ethiopia) sheep are equally importantin the rural economy. The sheep production systems in the area, however, aretraditional and semi-subsistence oriented. So far, only very limited efforts havebeen exerted to introduce and promote market-oriented sheep production and hence thecurrent income generating capacity of the sector is not at all justifiable.Re-orientation of the production system, which involves designing an effective andinformed breeding programme, is a necessity to bring about improvements inproductivity and in the production system of the sector. This re-orientation entailsproper valuation of both traded and non-traded products and services generated fromthe system. Information on the economic value of populations, traits and processeswould ease the management of animal genetic resources that requires many decisions(Scarpa et al. 2003). Proper identification and valuation ofthe different characteristics would make resource allocation decisions among thedifferent livestock improvement interventions for commercialization of the systemquite fast and smooth (Drucker et al. 2001; Kassie, 2007). This will also enable identification of sheep marketopportunities by identifying preferred traits of sheep. This is crucial asconsumers’ demand and preference is continuously changing over time.

Researchers have applied different economic valuation methods to understand thepreference for and the value of animal traits in different contexts. Revealedpreference and stated preference based models are the two most commonly usedapproaches. Revealed preferences based valuation methods record and analyze actualpayments on observable transactions for the commodities/services of interest whilestated preference based valuation methods make use of data on hypothetical choicesand implicit payments (Hensher et al. 2005). Richards andJeffrey (1996) employed a hedonic pricing model to establishindices of genetic worth of a dairy bull in Alberta, Canada. Their study indicatedthat the most important factors used by dairy farmers in valuing dairy bulls aremilk volume, protein and fat content, general conformation, body capacity, andpopularity of the bull. Barret et al. (2003) used astructural-heteroskedasticity-in-mean estimation method to identify the determinantsof livestock producer prices in the dry lands of northern Kenya. Their result showsthe importance of animal characteristics, periodic events that shift local demand orsupply, and rainfall in determining prices producers receive. Williams et al. (2006), similarly used a hedonic model using weekly salestransactions to analyze cattle prices in West Africa and reported that location,season, and cattle attributes influence sheep price.

In their study that aimed at investigating determinants of inter-annual pricevariation of small ruminants’ price in the eastern highlands of Ethiopia,Gezahegn et al. (2006) employed hedonic price modeling andreported significant differences in prices between seasons and markets, controllingfor attributes of animals. Kassie et al. (2011a) similarlyapplied heteroscedasticity consistent hedonic price modeling to examine factors thatinfluence cattle prices in the rural markets of central Ethiopia. The results ofthis study showed that season, market location, age, sex and body size are veryimportant determinants of cattle price. Chang et al. (2010)employed hedonic price modeling to study price differentials of retailed eggs andreported significant premiums attributed to production method, variation ingeographic locations and egg color. Similarly, Satimanon and Weatherspoon (2010) employed the same approach to determine price premiums oftraits of fresh eggs using sustainable attribute data from retail markets in theUnited States. Their study indicates that welfare-managed eggs have a significantprice premium while the sustainable packaging attributes are insignificant.

Other studies used a combination of revealed and stated preference data (e.g., Scarpaet al. 2003; Kassie et al. 2011b).Stated preference based valuation of animal genetic resources has also been widelyused (e.g., Omondi et al. 2008; Kassie et al. 2009 and Faustin et al. 2010). In recent yearsthere is a growing interest in using stated preference approaches which specificallyemploy choice experiments as real choice data in actual market are hardly available.Whenever available, however, revealed preference data have obvious advantages overstated preference data. Real world representation, embodiment of real constraints,reliability and validity are advantages of revealed preference data (Haab andMcConnel, 2002; Hensher et al. 2005).

This brief review has shown that there is an enormous body of knowledge on therelevance and application of hedonic price models. Although the focus of most of thestudies is market oriented production systems, the importance of the attributes oflivestock in determining prices observed in the market is a key lesson to learn.Interestingly though, there are hardly any publications done on sheep price modelingin subsistence and/or semi-subsistence crop livestock mixed farming systems. Thisresearch employs the well-established hedonic price modeling in a context wheremarkets are yet to develop and sheep have a more complex role than serving simply assources of meat or in some cases wool.

Methods

The study area and the rural markets

The study was conducted in the Horro-Guduru Wollega zone of Ethiopia. Theadministrative capital of the zone is called Shambu and is located at about310 km west of Addis Ababa. The 2007 population and housing census of theCentral Statistical Agency showed that in 2007 the total population of the zonewas about 580,000 out of which 50.1% were male and 49.9% were female (CSA, 2007). About 89% of the population in the zone lived inrural areas. The total area of the zone is about 710,000 hectares. According tothe agency’s national agricultural survey, the livestock population of thezone encompassed 127,000 heads of cattle, 25,000 sheep, and 12,000 goats (CSA,2009).

The study covered four sheep markets, namely, the markets of Shambu, GabaSanbata, Harato, and Fincha. All the markets, exceptShambu, are weekly markets that set on once in a week on adesignated day. Shambu operates throughout the week, except on Sundays.The market infrastructure in the zone is very poor and there is no fence orshed, information provision, and feed provision in these four markets.Fincha is the only fenced sheep market where livestock are tradedin a relatively organized manner. Livestock are trekked to and from all themarkets. The study markets are among the remote rural markets dominated by (cropand livestock) farmers, farmer-traders, peri-urban butchers and small restaurantowners.

In these markets, grading and standardization do not exist and transactions takeplace after a long one-to-one bargaining between sellers and buyers on aper-head basis. The price paid by the buyer and received by the seller,therefore, depends, among others, on how well he or she can bargain. Under suchcircumstances, prices paid will reflect buyers' preference for various sheeptraits, the type of buyer and seller and characteristics of the market place.The identification and analysis of the preferred traits and householdcharacteristics that influence the prices actually paid in the marketaccordingly forms the basis for effective market development interventions. Thisstudy generated primary data and analyzed the factors that determine sheepprices in rural parts of the zone for this particular purpose.

The data

Data on 195 traded sheep and on sheep marketers’ attributes were collectedin the four rural sheep markets mentioned above. The main traits of traded sheepwe focused on were coat color, body size, tail condition, age, and sex. Marketsin developing countries in general and in such rural setups in particular arehardly competitive due to the sources of inefficiency mentioned above and allother generic sources of market imperfections. This entails the inclusion offactors apart from the attributes of the goods and services – in this casethe sheep – in the model specification (Abdulai 2000; Kassie et al. 2011a). Therefore, we havegenerated and analyzed data on other factors that are expected to affect sheepprice. These factors include the attributes of buyers and sellers, such asoccupation and education level to serve as proxies for bargaining power.Seasonality of demand and supply was also captured. Description of variablesused in this study is presented in Table 1 below.
Table 1

Summary of variables and coding method used in hedonic pricemodel

Attribute

Code

Attribute

Code

Color

 

Market place

 

White

1 = white

Gaba sanbata

1 = Gaba sanbata

 

−1 = red

 

−1 = Shambu

 

0 = otherwise

 

0 = otherwise

Black

1 = black

Fincha

1 = Fincha

 

−1 = red

 

−1 = Shambu

 

0 = otherwise

 

0 = otherwise

Brown

1 = brown

Harato

1 = Harato

 

−1 = red

 

−1 = Shambu

 

0 = otherwise

 

0 = otherwise

Creamy white

1 = creamy white

Seller type

 
 

−1 = red

Farmer

1 = farmer

 

0 = otherwise

 

−1 = trader

White mixed

1 = white mixed

 

0 = otherwise

 

−1 = red

Farmer trader

1 = farmer trader

 

0 = otherwise

 

−1 = trader

Sex of sheep

0 - female

 

0 = otherwise

 

1 - male

Buyer type

 

Body size

 

Trader

1 = buyer is trader

Medium size

1 = medium

 

−1 = other buyers

 

−1 = small

 

0 = otherwise

 

0 = otherwise

Farmer

1 = buyer is farmer

Big

1 = big

 

−1 = other buyers

 

−1 = small

 

0 = otherwise

 

0 = otherwise

Farmer trader

1 - buyer is farmer trader

Tail type

  

−1 = other buyers

Medium and thin

1 = medium and thin

 

0 = otherwise

 

−1 = long and fat

Season

 
 

0 = otherwise

Christmas (season 1)

1 = Christmas

Medium and fat

1 = medium and fat

 

−1 = season 2

 

−1 = long and fat

 

0 = otherwise

 

0 = otherwise

Fasting (season 3)

1 = fasting season

   

−1 = season 2

   

0 = otherwise

Three survey rounds of individual sheep level transactions were conducted over aninterval of one month. The first round was conducted during the beginning ofJanuary 2009; i.e., the Ethiopian Christmas season. This round was targeted tocapture the price change that occurs during holidays. The second round was donein February 2009. This is a period with no important festival or planned socialoccasion and it overlaps with the time when farmers have completed cropharvesting. By this time farmers are expected to be less forced to sell theirlivestock for generating liquid capital (Kassie et al. 2011a). The third round was undertaken in March 2009. This periodcorresponds to the Ethiopian lent. From each market, 15 buyers were consideredin each round except in Harato where 20 buyers were interviewed, takinginto account the relative size of the market. That means 65 buyers wereinterviewed in each round.

The descriptive statistics of the variables used in the econometric model showconsiderably variation across respondents (Table 2).The average age of marketed sheep in the surveyed rural markets was one year(st.dev. = 11.4 months). In the observed sheep transactions in the fourmarkets, 59% of traded sheep were male implying that female sheep are lessfrequently marketed as they are usually kept for reproduction (herd replacement)and less for generating cash. The average sheep price during this study was morethan Ethiopian Birr 238.00. Sheep in those markets were observed to havedifferent patterns of fur color although red was the dominant color (44%)followed by creamy-white (29%) over the survey period and observed transactions.More than 10% of the marketed sheep were black while the remaining were brown,white and mixed colored sheep. Data on body condition of the marketed sheep,which indicates relative fatness and appearance, were also observed and it wasfound that 48% of the sheep in the markets were in good condition and another48% had a medium body condition. The remaining 3.6% were sheep with bad bodycondition. Related is the body size of the sheep marketed and 43% of them weremedium sized, 33% small and the rest large size. The dominant tail condition ofthe traded sheep were long and thin tail type (48%) followed by long and fattail type (24%). Most actors in the local sheep markets were farmers and sheeptraders. Though data on the qualitative attributes were entirely based onbuyers’ perception of traded sheep, it is important to understandbuyers’ preference for these attributes. Generally, the typical sheeptraded in these markets is 12 months old, red coated, of good or mediumbody condition, medium body size, and long thin tailed.
Table 2

Descriptive statistics for sheep in the local market and marketparticipants

Description

Mean(SD)/percentage

Price per head of sheep (ETB*)

238.36(83.92)

Age of sheep in months

12(11.431)

Male sheep (%)

58.5

Sheep coat color (%)

 

Red

44.1

Creamy white

28.7

Black

10.3

Brown

3.1

White

4.6

White mixed with other colors

9.2

Body condition (%)

 

Good

48.2

Medium

48.2

Bad

3.6

Body size (%)

 

Small

32.8

Medium

42.6

Large

24.6

Tail condition (%)

 

Long fat tail

24.1

Long thin tail

47.7

Medium length thin tail

14.9

Medium length fat tail

13.3

Buyers’ occupation (%)

 

Trader

41

Farmer

10.8

Farmer-trader

6.2

Others

42.1

Sellers’ occupation (%)

 

Trader

12.8

Farmer

61

Farmer-trader

13.8

Others

12.4

*ETB stands for Ethiopian Birr which is the Ethiopian currency.

Analytical framework

Revealed preference is manifested through the actual prices paid for goods andservices with expected utility. Hence, the prices sheep sellers receive arereflections of the utility anticipated by the buyers and this utility is derivedfrom the attributes of the product as sheep can be considered as quality(attribute) differentiated goods (Lancaster, 1966; Rosen,1974; Ekeland et al. 2004;Nesheim, 2006). This research focuses on the mainphenotypic attributes that buyers inspect when buying a sheep. The externalfeatures farmers look at and attach value to are age, fur color, body size, andtail type. The different levels of the homogenous attributes that differentiatesheep are known to both buyers and sellers. The levels considered in thisanalysis are those perceived by the buyers, despite the possibility of imperfectknowledge and differences in measurement. The buyers and sellers in the marketsconsidered are mainly farmers who raise the sheep. In line with the householdmodeling literature, where goods are produced, consumed and sold by households,a hedonic model can be employed to value the attributes of the qualitydifferentiated indivisible goods. Therefore, estimation of the relationshipbetween the characteristics of the sheep and their prices can be made throughhedonic price modeling.

Following Rosen (1974) and Palmquist (2006), let x 0j be the total amount of the jth product characteristic provided to the consumer by consumption ofall products, x ij be the quantity of the jth characteristic provided by one unit of product i, andq i be quantity of the ith product consumed. Then, the total consumption of eachcharacteristic can be given as
x 0 j = f i q 1 , . . . , q n ; x 1 j , . . . , x nj
(1)
and the consumer’s utility function is expressed as
U = q 1 , . . q n ; x 11 , x 12 , . . , x 1 m , x 21 , x 22 , . , x nm
(2)

where n is the number of products and m is the number of characteristics.

The consumer is assumed to maximize this utility function subject to a budgetconstraint that can be specified as
Y = i p i q i
(3)
where Y is fixed money income, and p i is fixed price paid for the ith product. The consumer’s utility maximizing level quantity ofeach product can then be estimated by maximizing the Lagrangian:
L = U x 01 , . . , x 0 n λ i p i q i Y
(4)

where λ is the Lagrangian multiplier.

Assuming an interior solution, the first-order condition of the Lagrangian forq i is given as
L q i = 0 = U x 0 j x 0 j q i λ p i .
(5)
It can easily be shown that λ is equal to the marginal utility of income (∂U/∂Y). Substituting ∂U/∂Y for λ and solving for p i , equation (5) can be rewritten in order to express the demand forattributes as a function of the marginal utility of the attribute and themarginal utility of income.
p i = x 0 j q i U x 0 j U Y .
(6)

As income is defined to be equal to expenditure (equation 4), the term in thesquare bracket is the marginal rate of substitution between expenditure and thej th product characteristics.

Under competitive market conditions, implicit prices will normally be related toproduct attributes alone, without accounting for producer or supplierattributes. However, as widely documented in the literature, rural markets indeveloping countries, particularly in sub-Saharan Africa, are rarely competitive(Barret and Mutambatsere 2007). This is essentially due topoor communication and transport infrastructure, limited rule of law, andrestricted access to commercial finance, all of which make markets function muchless effectively. Several empirical studies have shown that prices are alsorelated to the attributes of buyers, season and market location (e.g.,Oczkowski, 1994; Abdulai, 2000;Jabbar and Diedhiou, 2003). Hence, essentialcharacteristics of the buyer and sellers were included in the models estimatedin this research.

Another important issue in estimating hedonic functions is the identification ofthe appropriate functional form and estimation procedure (Ekeland et al. 2004; Nesheim, 2006). In general,the functional form of the hedonic price equation is unknown (Haab and McConnel,2002). Parametric, semi-parametric and non-parametricestimations procedures have all been suggested and used in differentapplications (e.g., Anglin and Gencay, 1996; Parmeter etal. 2007). This research focuses on the estimation of therelative weights of sheep attributes (first step hedonic analysis) and hence thetechnical details of these alternative approaches are not discussed.

The estimation strategy adopted in this study is a simple linear model basedfollowing the suggestion by Cropper et al. (1988) as wellas Haab and McConnel (2002). Cropper et al. (1988) employed Monte-Carlo simulation analysis to show thatthe linear and linear-quadratic functions give the smallest mean square error ofthe true marginal value of attributes. However, when some of the regressors aremeasured with error or if a proxy variable is used, then the linear functiongives the most accurate estimate of the marginal attribute prices. Haab andMcConnel (2002) also argued that when choosing afunctional form and the set of explanatory variables, the researcher must bearin mind the almost inevitable conflict with collinearity. High collinearitymakes the choice of a flexible functional form less attractive, since theinteractive terms of a flexible functional form result in greater collinearity.Given these considerations, we begin with semi-log model given by
ln price = + ϵ .
(7)
Following Champ et al. (2003), the market premium (Γ)for an attribute j is computed as:
Γ j = 100 * e β 1 .
(8)

In equation (7) the error term is assumed to have a constant variance,σ2; hence, homoscedastic. However, if and when the errors areheteroscedastic, the OLS estimator remains unbiased, but becomes inefficient.More importantly, the usual procedures for hypothesis testing are no longerappropriate. Given that heteroscedasticity is common in small samplecross-sectional data, methods that correct for heteroscedasticity are essentialfor prudent data analysis (Long and Ervin, 2000).

Using heteroscedasticity consistent (HC) standard errors is the recommendedapproach (MacKinnon and White, 1985; Long and Ervin,2000) to correct for heteroscedaticity of unknownform. The suggested alternative ways of correction using HC includeHC0, HC1, HC2, and HC3. Thesealternatives are not equally powerful and perform differently under differentconditions depending mainly on sample size. Based on Monte Carlo simulation,MacKinnon and White (1985), for example, recommended thatin small samples one should use HC3. However, Davidson and MacKinnon(1993) later recommended strongly that HC2or HC3 should be used. Long and Ervin (2000),similarly, recommended for N ≤ 250, tests based onHC2 and HC3 than those based on other HC. This MonteCarlo simulation result also shows HC3 is superior for tests ofcoefficients that are most affected by heteroscedasticity and HC2 isbetter for tests of coefficients that are least affected by heteroscedasticity.Accordingly, we have employed HC2 and HC3 in this study.OLS was also applied for comparison.

Following Davidson and MacKinnon (1993), the alternativecovariance matrix estimators of the error term for HC2 andHC3, including the OLS, are specified as:
OLS = e i 2 n k X X 1
(9)
H C 2 = X X 1 X diag e i 2 1 h ii X X X 1
(10)
H C 3 = X X 1 X diag e i 2 1 h ii 2 X X X 1
(11)

where n is number of observations, k number of parameters estimated, and h ii is x i ' X ' X 1 x i .

Results and discussions

General model results

The results of the hedonic price model from both OLS and HC regressions are givenin Table 3. The table summarizes the coefficients ofthe variables used in the model, and the standard errors of OLS andheteroscedasticity consistent (HC2 and HC3) estimations.HC estimation was used as an adjustment to the OLS model since cross-sectionaland small sample price data are intrinsically heteroscedastic. As expected, theOLS standard errors were found to be generally lower than the standard errors ofHC2 and HC3 for all variables except for somevariables in HC2. However, the standard errors of all explanatoryvariables in HC3 were increased and greater than both OLS andHC2 except for three variables brown, thin long tail condition,and sex. Hence, the t-values of the OLS coefficients are inflated and could notbe reliable for inferences. Between HC2 and HC3, thestandard errors in HC2 were found to be lower than that ofHC3. Therefore, the t-values based on standard errors generatedby HC3 estimation were used for inferences.
Table 3

Estimation results of OLS and Heterosecdasticity consistent hedonicmodel

ln(price)

Coefficient

OLS SE

HC2 SE

HC3 SE

 Constant

5.2350

0.0729

0.0725

0.0812

Age and sex

    

 Age

0.0208

0.0068

0.0067

0.0076

 Age square

−0.0003*

0.0001

0.0001

0.0002

 Sex

−0.0457

0.0386

0.0346

0.0372

Coat color

    

 White

−0.1003

0.0664

0.0965

0.1098

 Brown

0.0158

0.0829

0.0615

0.0696

 Black

−0.1614

0.047

0.0488

0.0535

 White mixed

0.1305

0.0502

0.0515

0.0563

 Creamy white

0.0611*

0.0339

0.033

0.0362

Body Size

    

 Medium

0.034

0.0247

0.0247

0.0266

 Large

0.1467

0.0414

0.0396

0.0429

Tail condition

    

 Long thin

−0.0831

0.026

0.0236

0.0257

 Medium & thin

−0.1225

0.0337

0.0342

0.0373

 Medium & fat

0.0319

0.0496

0.0406

0.0454

Season

    

 Season 1

0.0769

0.0226

0.0223

0.0239

 Season 3

−0.0275

0.0236

0.025

0.0268

Market/place

    

 Finchaa

−0.0068

0.0309

0.0314

0.0336

 G/sanbata

0.0152

0.0306

0.0324

0.0344

 Harato

−0.0663

0.0286

0.028

0.0303

Type of seller

    

 Farmer

−0.0041

0.0262

0.0267

0.0293

 Farmer trader

0.0116

0.0375

0.0356

0.0391

 Others

−0.001

0.0417

0.0472

0.0518

Type of buyer

    

 Trader

0.0478

0.028

0.0276

0.0302

 Farmer

0.021

0.041

0.039

0.0423

 Farmer trader

−0.1042*

0.0506

0.0563

0.0631

‡, † and * significant at α = 0.01,α = 0.05 and α = 0.1 respectively,based on HC3 standard errors. SE = standarderror.

Number of observations = 195,R2 = 0.6887.

Due to the changes in standard errors in the three regression results,significant variables in OLS become insignificant and the significance levels ofthe variables have also been changed in HC2 and HC3. Agesquare was significant at 5% in OLS but the significance level in HC2and HC3 changed to 10%. Similarly, farmer trader (one of the buyertypes) was significant at 5% in OLS, but in HC2 and HC3the variable was significant only at 10%. White mixed coat color, wassignificant at 1% in OLS but it became significant only at 5% in HC2and HC3. Further, the variable representing trader buyer wassignificant in OLS and HC2 but became insignificant in HC3estimation.

A model specification test was carried out for the OLS regression model using theRamsey RESET test. The test with the (null) hypothesis that the model has noomitted variables generated an F (3, 161) value of 0.64 which is extremely belowthe critical value of 2.65 at α = 0.05 implying non-rejection ofthe hypothesis that there are no omitted relevant explanatory variables. TheR-square value of the models is 0.6887 implying that the model explained about69% of change in price of sheep in the local markets of Horro-Guduru Wollegazone of Ethiopia.

Determinants of price and premium for indigenous sheep traits

The results of the three estimations (OLS, HC2 and HC3)show that sheep price is determined by sheep traits (such as age, color, bodysize, and tail condition), season, market places, and buyer type. Agesignificantly and positively influenced price of sheep in the study area. Thisis in line with the basic feature of the low input sheep production system inthe area. That is, under a low input production system sheep need a longerperiod of time to attain the required body condition and size to command a goodprice. Age square, however, influenced sheep price negatively implying sheepcommand a higher price up to a very old age and then the price will fall down asage goes up. Given the weight of the coefficient and the average age of sheepbeing marketed, it seems however that old is less of an issue.

From the color dummies, black coat color was found to affect sheep pricenegatively and significantly. Hence, black colored sheep received a pricediscount of about 15% as compared to red coat colored sheep. The negativepremium the black coated sheep received emanates from the fact that people tendto avoid black coated sheep or any other livestock (see e.g., Kassie et al,2009) for two reasons. First, the tsetse fly –the vector for sleeping sickness (trypanosomiasis) – attacksblack coated livestock more? than others. Second, red or other lighter coloredsheep are preferred to others for purposes of ceremonial slaughtering in thestudy area. Whitish and creamy white (locally called dallecha) coatcolors of traded sheep attracted a 14% and 6.3% price premium respectively,compared to red coat color, ceteris paribus. Body size was anothertrait of sheep that significantly affected price of sheep. Intuitively, sheepwith a large body size receive higher prices and hence sheep with a large bodysize were found to fetch about 15.8% higher price premium compared with smallsized sheep. Sheep with thin and long tail and thin and medium length tailreceived 8% and 11.53% less price, respectively, compared to long and fat tailedsheep.

The determinants of sheep price other than traits of sheep were market locationand seasonal factors. Sheep command a significantly higher price in season one(Christmas season) compared with season two (normal season). The Christmasseason is the period of high demand for sheep (or livestock in general) thatoverlaps the crop produce harvesting season that might increase farmers’(as sheep sellers) bargaining power as they can postpone selling when prices arenot right. In season one sheep will attract an 8% higher price premium comparedto selling in season two. This a general tendency in Ethiopian livestock -particularly small ruminants - marketing as sheep or goat slaughtering is anindispensable part of big festivities such as Christmas provided they areaffordable.

Among the market location dummies, sheep in Harato attracted a lowerprice compared with Shambu. This is likely due to the relatively highpotential for sheep population in the Harato area and hence highsupply. These results imply that smallholder sheep keepers would benefit if theycarefully choose the selling time and the market.

The type of buyer was also an important determinant of price paid for sheep inthe study area. Farmer-traders (farmers who do par-time trading) paid a lowerprice as compared to other groups of buyers. This is possibly because thesebuyers are well informed both about the production and the marketing of sheepsuch that they would be in a better position to bargain for a lower price.

In summary, the estimation results show that traits of sheep are much moreimportant determinants of actual price observed than types of buyers and sellersor purposes of buying and selling. Among the attributes considered, age, blackcoat, large body size, and tail condition were found to be most influential indetermining the price paid for sheep in the study area.

Conclusions

This study generated primary data on actual transactions accomplished in four ruralmarkets over three months in central Ethiopia. Using the revealed preferenceanalysis framework and hedonic price modeling, the study determined the level ofinfluence of attributes of sheep and features of buyers and sellers in sheep marketson actual prices paid per head of sheep. As heteroscedasticity is common incross-sectional data and small sample data, alternative estimations, mainlyheteroscedasticity consistent formulations, were employed in addition to OLSestimation.

This estimations have shown how intricate are the relationships between price andtraits of sheep and trait and trait level identification of sheep consumers in therural areas of the study areas. Traits such as age, black coat, large body size, andtail condition were found to be most influential in determining the revealedpreferences and hence the prices paid in these rural markets.

Factors such as season of marketing, market location, and type of buyer were alsofound to be highly important in influencing the prices paid for sheep. Thesignificance of season and market place in influencing price paid for sheep as welljustifies the need of targeting season and market places so that smallholder sheepkeepers could benefit from the required transformation in the sheep productionsystem. Alternatively or even additionally, linking producers to urban markets wherethere is high demand for sheep would be an important step to improve farmers’return from the system.

Two important implications can be drawn from the results of this study. First, theconsumption of sheep in these areas seem to be very sophisticated such that anintervention that focuses on a single attribute would hardly be successful toimprove both supply and demand sides of sheep marketing in the study area. Second,the sheep genetic resources in areas similar to the study areas need to becomprehensively profiled for their attributes. This would be essential inidentifying the important attributes of the existing stock and hence prioritizingthose traits that need to be improved both for biological and economicefficiency.

Declarations

Acknowledgements

The authors are grateful to ILRI-ICARDA-BOKU project for funding this research.This paper has benefited immensely from insightful comments of an anonymousreviewers and editor of the Agricultural and Food Economics and the authors arevery grateful for comments and suggestions received. Yet, the authors alone areresponsible for the contents of the paper.

Authors’ Affiliations

(1)
The University of Liverpool
(2)
ICARDA
(3)
ILRI
(4)
CIMMYT

References

  1. Abdulai A: Spatial Price Transmission and Asymmetry in the Ghanaian Maize Market. Journal of Development Economics 2000, 63: 327–349. 10.1016/S0304-3878(00)00115-2View ArticleGoogle Scholar
  2. Anglin PM, Gencay R: Semiparametric Estimation of a Hedonic Price Function. Journal of Applied Econometrics 1996, 11: 633–648. 10.1002/(SICI)1099-1255(199611)11:6<633::AID-JAE414>3.0.CO;2-TView ArticleGoogle Scholar
  3. Barret CB, Chabari F, Bailey D, Little P, Coppock D: Livestock Pricing in the Northern Kenyan Rangelands. Journal of African Economies 2003, 12(2):127–155. 10.1093/jae/12.2.127View ArticleGoogle Scholar
  4. Barret C, Mutambatsere E: Agricultural Markets in Developing Countries. Entry. In The New Palgrave Dictionary of Economics. 2nd edition. Edited by: Durlauf SN, Blume LE. London: Palgrave McMillan; 2007.Google Scholar
  5. Champ PA, Boyle KJ, Brown TC (Eds): A Primer on Nonmarket Valuation. Dordrecht: Kluwer Academic Press; 2003.Google Scholar
  6. Chang J, Lusk J, Norwood F: The Price of Happy Hens: A Hedonic Analysis of Retail Egg Prices. Journal of Agricultural and Resource Economics 2010, 35(3):406–423.Google Scholar
  7. Cropper ML, Deck LB, McConnel KE: On the Choice of Functional Form for Hedonic Price Functions. Rev Econ Stat 1988, 70(4):668–675. 10.2307/1935831View ArticleGoogle Scholar
  8. CSA (Central Statistical Agency): Statistical Abstract. Ethiopia: Addis Ababa; 2009.Google Scholar
  9. CSA (Central Statistical Agency),: Statistical Abstract. Ethiopia: Addis Ababa; 2007.Google Scholar
  10. Davidson R, MacKinnon JG: Estimation and Inference in Econometrics. Cambridge: Oxford University Press; 1993.Google Scholar
  11. Drucker AG, Gomez V, Anderson S: The economic valuation of farm animal genetic resources: A survey ofavailable methods. Ecol Econ 2001, 36: 1–18. 10.1016/S0921-8009(00)00242-1View ArticleGoogle Scholar
  12. Ekeland I, Heckman JJ, Nesheim L: Identification and Estimation of Hedonic Models. Journal of political economy 2004, 112(1):s60-s109.View ArticleGoogle Scholar
  13. FAOSTAT: FAO (Food and Agricultural Organization of the United Nations),Rome. 2009. . Accessed 19 October http://faostat.fao.org.Google Scholar
  14. FAOSTAT: FAO (Food and Agricultural Organization of the United Nations),Rome. 2010. . Accessed 23April http://faostat.fao.org.Google Scholar
  15. Faustin V, Adegbidi AA, Garnett ST, Koudande DO, Agbo V, Zander KK: Peace, health or fortune?:Preferences for chicken traits in rural Benin. Ecol Econ 2010, 69(9):1848–1857. 10.1016/j.ecolecon.2010.04.027View ArticleGoogle Scholar
  16. Gezahegn A, Mohammad AJ, Hailemariom T, Elias M, Getahun K: Seasonal and Inter-Market Differences in Prices of Small Ruminants inEthiopia. Journal of Food Products Marketing 2006, 12(4):59–77. 10.1300/J038v12n04_05View ArticleGoogle Scholar
  17. Haab TC, McConnel KE: Valuing Environmental and Natural Resources: The Econometrics of Non-MarketValuation. Cheltenham, UK, Northampton, MA, USA: Edward Elgar; 2002.View ArticleGoogle Scholar
  18. Hensher D, Rose J, Greene W: Applied Choice Analysis: A Primer. Cambridge: Cambridge University Press; 2005.View ArticleGoogle Scholar
  19. Jabbar MA, Diedhiou ML: Does Breed Matter to Cattle Farmers and Buyers? Evidence from WestAfrica? Ecol Econ 2003, 45: 461–472. 10.1016/S0921-8009(03)00097-1View ArticleGoogle Scholar
  20. Kassie GT: Economic Valuation of the Preferred Traits of Indigenous Cattle inEthiopia. PhD Dissertation. pp: Christian Albrechts University of Kiel, Kiel,Germany; 2007. 177 177Google Scholar
  21. Kassie GT, Abdulai A, Wollny C: Valuing Traits of Indigenous Cows in Central Ethiopia. Journal of Agricultural Economics 2009, 60(2):386–401. 10.1111/j.1477-9552.2008.00191.xView ArticleGoogle Scholar
  22. Kassie GT, Abdulai A, Wollny C: Heteroscedastic Hedonic Price Model for Cattle in the Rural Markets ofCentral Ethiopia. Applied Economics 2011, 43(24):3459–3464. 10.1080/00036841003670614View ArticleGoogle Scholar
  23. Kassie GT, Abdulai A, Wollny C, Ayalew W, Dessie T, Tibbo M, Haile A, Mwai O: Implicit prices of indigenous cattle traits in central Ethiopia: Applicationof revealed and stated preference approaches. ILRI Research Report 26. University of Kiel: CIMMYT and ILRI; 2011b.Google Scholar
  24. Lancaster K: A new approach to consumer theory. Journal of Political Economy 1966, 74: 132–157. 10.1086/259131View ArticleGoogle Scholar
  25. Long JS, Ervin LH: Using Heteroscedasticity Consistent Standard Errors in the Linear RegressionModel. The American Statistician 2000, 54(3):217–224.Google Scholar
  26. MacKinnon JG, White H: Some Heteroscedasticity Consistent Covariance Matrix Estimators with ImprovedFinite Sample Properties. Journal of Econometrics 1985, 29(3):305–325. 10.1016/0304-4076(85)90158-7View ArticleGoogle Scholar
  27. Negassa A, Jabbar M: Livestock ownership, commercial off-take rates and their determinants inEthiopia. Research Report 9. Nairobi, Kenya: ILRI (International Livestock ResearchInstitute); 2008:52.Google Scholar
  28. Nesheim L: Hedonic price functions. Centre for microdata methods and practice workingpaper CWP18/06. 2006.Google Scholar
  29. Oczkowski E: A Hedonic Price Function for Australian Premium Table Wine. Australian Journal of Agricultural Economics 1994, 38: 93–110. 10.1111/j.1467-8489.1994.tb00721.xView ArticleGoogle Scholar
  30. Omondi I, Baltenweck I, Drucker AG, Obare G, Zander KK: Economic valuation of sheep genetic resources: implications for sustainableutilization in the Kenyan semi-arid tropics. Trop Anim Health Prod 2008, 40(8):615–626. 10.1007/s11250-008-9140-7View ArticleGoogle Scholar
  31. Palmquist RB: Property Value Model. In Handbook of Environmental Economics Edition 1, 2 edition. Edited by: Maler KG, Vincent J. 2006, 763–819.Google Scholar
  32. Parmeter CF, Henderson DJ, Kumbhakar SC: Nonparametric Estimation of a Hedonic Price Function. Journal of Applied Econometrics 2007, 22: 695–699. 10.1002/jae.929View ArticleGoogle Scholar
  33. Richards TJ, Jeffrey SR: Establishing indices of genetic merit using hedonic pricing: an applicationto dairy bulls in Alberta. Canadian journal of Agricultural Economics 1996, 44: 251–264. 10.1111/j.1744-7976.1996.tb00149.xView ArticleGoogle Scholar
  34. Rosen S: Hedonic Prices and Implicit Markets: Product Differentiation in PureCompetition. Journal of Political Economy 1974, 82(1):34–35. 10.1086/260169View ArticleGoogle Scholar
  35. Satimanon T, Weatherspoon D: Hedonic Analysis of Sustainable Food Products. International Food and Agribusiness Management Review 2010, 13(4):57–74.Google Scholar
  36. Scarpa R, Kristjanson P, Ruto E, Radeny M, Rege JEO: Valuing indigenous cattle breeds in Kenya: an empirical comparison of statedand revealed preference value estimates. Ecol Econ 2003, 45: 409–426. 10.1016/S0921-8009(03)00094-6View ArticleGoogle Scholar
  37. Tibbo M: Productivity and health of indigenous sheep breeds and crossbreds in theCentral Ethiopian highlands, PhD Thesis, Swedish University of AgriculturalSciences. Uppsala: Sweden University Press; 2006.Google Scholar
  38. Williams OT, Okike I, Spycher B: A Hedonic Analysis of Cattle Prices in the Central Corridor of West Africa:Implications for Production and Marketing Decisions. In Contributed paper prepared for presentation at the International Associationof Agricultural Economists Conference. Australia: Gold Coast; 2006.Google Scholar

Copyright

© Terfa et al.; licensee Springer. 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permitsunrestricted use, distribution, and reproduction in any medium, provided theoriginal work is properly cited.