1. ETHIOPIAN DEVELOPMENT
RESEARCH INSTITUTE
Determinants of Cattle Prices in Ethiopia
Fantu Nisrane Bachewe
International Food Policy Research Institute (IFPRI)
(Ethiopia Strategy Support Program, ESSP-II)
Workshop theme: Taking the Stock of the Economy of the Livestock Sector in Ethiopia
4 November 2011
Addis Ababa, Ethiopia
The views expressed in this paper are those of the author and do not represent the official position of his institution.
2. Presentation Outline
1. General background ,
2. Model: Hedonic price formation system of equations,
3. Data description
4. Results, and
5. Summary of findings.
3. General background
• Food prices have been increasing rapidly in Ethiopia starting from
the middle of the last decade.
• Cattle and beef prices have risen more quickly than other meat
items, and constitute an important source of protein in Ethiopia
• Beef price increased so much that the government included it
amongst its list of prices capped items in 2011
• Many factors could be driving beef/cattle price inflation:
o Increased livestock and beef exports driven by international prices
movements, depreciation and greater openness, and improved
commercialization/quality
o Food/cattle prices follow trends in general price levels, including fuel prices
o Production shocks: loss of grazing land, recent drought
• This work tests some of these hypotheses using a Hedonic price
formation analysis (HPFA).
4. General background
• Previous works using HPFA include Teklewold et. al. (2009) and
Ayele et. al. (2006) while the recent work by Kassie et. al. (2011)
uses heteroscedasticity efficient random parameters logit,
• This work, in addition to explaining factors that caused increase in
cattle prices, has two more contributions
o Applies Heckman’s sample selection approach to account for cases not
observed in each market.
o Accounts for endogeneity of body mass condition of cattle on other
attributes.
• Coatney et. al. (1996) account for endogeneity but use linear HPFA
while others use HPFA without accounting for sample selection.
5. Model: Hedonic price formation system of equations
• HPFA postulates price of qualitatively different good is a function of
the sum total of consumers’ valuation of cattle attributes, as well as
other variables affecting the market environment
• Suppose cattle prices are a function of factors affecting the market
environment and cattle attributes and this function be given as
J K
Pit
( )
0 j X jit ( )
k Dkit eit
j 1 k J 1
• where
• P is real price of cattle i at week t,
it
• X jit is continuous variable j associated with cattle i at time t,
• Dkit is dummy variable k associated with cattle i at time t, and
• eit the error term.
• Note that body mass is included in X but is a function of other factors,
6. Model: Hedonic price formation system of equations
• 0 , 1,..., J , J 1,..., K are coefficients to be estimated, while
Pit 1 and X ( ) X 1
jit
( )
Pit jit
• If 1 the equation is linear and as 0 partially log-linear,
• With body mass condition assumed endogenous, the body mass
L
condition equation is specified as Git 0 l X lit it
• where l L
• Git is the body mass index of cattle i and X lit are variables that
affect body mass condition and 's are parameters to be estimated.
7. Data description
• We use powerful data set from ILRI: a panel cover time (01/2005-
03/2011), space (32 markets, 8 regions), and animal characteristics
• There appear to be several distinct periods of price trends, During:
o The January 2005-March 2007 period, average real and nominal prices grew at
monthly rates of 0.6 and 1.9 percent,
o The April 2007-January 2009 period, when real and nominal prices grew at 1.8
and 4.1 percent (a period of high food inflation in general)
o The January-November 2009 period: real and nominal price declines in cattle
and other food types
o December 2009-September 2010: strong recovery (high international prices?)
o October 2010 onwards: decline in cattle prices (drought in pastoralist areas?
government price controls?)
• This growth in prices was followed by a relatively faster growth in the
number of cattle sold in an average market of 23 percent.
8. Price/Real price (December 2006 prices)
500
2000
2500
3000
3500
4000
4500
1000
1500
January-05
April-05
July-05
October-05
January-06
April-06
July-06
October-06
January-07
April-07
Period of rapid
overall inflation
July-07
October-07
Nominal prices
January-08
April-08
July-08
October-08
January-09
Real prices
April-09
July-09
October-09
Overall
deflation
January-10
April-10
July-10
Figure 1. Mean monthly nominal and real prices, January 2005-March 2011.
October-10
January-11
high int. prices?
Strong recovery;
lowlands?
Drought in
9. Price/Real price (December 2006 prices)
500
1000
1500
2000
2500
3000
3500
4500
4000
January-05
April-05
July-05
October-05
January-06
cattle sold
April-06
Real prices
July-06
Nominal prices
October-06
January-07
Average number of
April-07
July-07
October-07
January-08
April-08
July-08
October-08
January-09
April-09
July-09
October-09
January-10
April-10
July-10
October-10
January-11
Figure 2. Mean monthly nominal and real prices and volumes of sales.
0
500
9
1000
1500
2000
Market level average number of cattle sold
10. Data description
• Male cattle had 34 percent larger prices
• Better body mass grade entails a 36 percent increase real prices,
o Increase in age bracket entails 48 percent increase real prices,
• Volumes of official live animal and meat exports grew at average
monthly rates of 5.1 and 10.9 percent, respectively (Customs Auth.).
o Significant increase in official volume of live animal exports during the
January 2010-March 2011 period, with growth averaging 24 percent
relative to the 7.6 percent between January 2005 and December 2009,
o The difference in growth rates of during 1/2010-3/2011 and 1/2005-
12/2009 of volume of meat and meat products exports was only 1 %
• Other variables from CSA, MoFED, and the World Bank used in the
econometric analysis had faster growth rates during the latter period.
11. Price/Real price (December 2006 prices)
exports.
500
1000
1500
2000
2500
3000
0
January-05
May-05
September-05
January-06
Real prices
May-06
September-06
January-07
May-07
September-07
January-08
May-08
September-08
January-09
May-09
September-09
January-10
May-10
September-10
January-11
Real value of live animal and meat exports
-
50,000
Figure 3. Trends in real prices of cattle and value of live animals and meat
100,000
150,000
200,000
250,000
12. Data description
Table 1. Growth rates in live animal and meat exports, animal
transportation fares, local cooking oil, and international beef price.
January 2005- January 2005- January 2010-
March 2011 December 2009 March 2011
Variable Real Nominal Real Nominal Real Nominal
Value of live animal and
meat exportsa 8.1 9.9 6.8 9.1 10.2 11.3
Animal transportation fareb -0.5 0.9 -1.0 0.8 0.3 1.2
Locally produced cooking
oil priceb 0.2 1.6 -0.2 1.6 0.8 1.7
Australian beefc 0.3 0.8 -0.2 0.1 1.1 2.0
Source: Data with superscripts a, b, and c are from MoFED, CSA, and the World Bank,
respectively.
13. Price/Real price (December 2006 prices)
500
1000
1500
2000
2500
3000
3500
4500
4000
0
January-05
May-05
September…
January-06
May-06
Real prices
September…
Nominal prices
animal transportation fares.
January-07
May-07
September…
Animal transportation fare
January-08
May-08
Locally produced cooking oil price
September…
January-09
May-09
September…
January-10
May-10
September…
January-11
Figure 4. Trends in prices of cattle and domestic produced cooking oil and
5
0
10
15
20
25
30
35
40
45
50
Nom. prices of cooking oil and animal transport
14. Price/Real price (December 2006 prices)
500
1000
1500
2000
3000
3500
4000
2500
4500
0
January-05
May-05
Nominal prices
Swine real price
September-05
January-06
May-06
September-06
January-07
May-07
September-07
Real prices
January-08
May-08
Poultry real price
September-08
January-09
May-09
September-09
January-10
May-10
September-10
January-11
20
30
40
50
60
70
80
90
100
110
120
14
Australian beef real price
Figure 5. Trends in real and nominal cattle prices and real global prices of meat.
US Cents per KG (April 2001 prices)
15. Results
• Elasticity of prices of cattle with respect to (WRT) live animal and
meat exports was 0.056,
o Cattle prices increase by 5.6 cents for a 1 birr increase in value of exports.
• Elasticity of cattle prices WRT real international beef price was 0.9 –
so international price transmission has been very important in recent years!
• Cattle prices increase with swine prices, although its effect was small
• Elasticity WRT cattle transportation fares is 0.2;
o A bull worth 1,000 birr gets priced 1,200 when real price of transportation grow
by 1 birr, but its overall effect was small
• Elasticity WRT body mass index was 0.45, a cattle that is
fat/moderate sold at 45 percent larger price than moderate/thin
• Real prices of cattle in a given age category were 54 percent larger
relative to ones in immediately younger category.
• Male cattle sold at prices 36 percent larger than female cattle.
16. Table 2. HPFA equation estimates of international prices and cattle attributes.
variable Coefficient Elasticity variable Coefficient Elasticity
International prices and factors Cattle characteristics
Value of live animal and
meat exports 0.026 0.056 Grade index 2.224 0.449
Locally produced cooking oil 0.257 0.099 Boran 0.291 0.062
Animal transportation fare 0.444 0.200 Danakil -0.593 -0.116
Total monthly rainfall 0.040 0.020 Harar 0.913 0.208
International price of beef 1.581 0.904 Mixed 0.446 0.097
International lamb price 0.163b 0.121 Raya Azebo -0.555 -0.109
International swine price 0.349 0.174 Zebu -0.09b -0.018
International poultry price 0.191b 0.098 Young 1.930 0.492
Lambda 0.207 Mature 3.513 1.072
Constant -8.695 Male 1.500 0.365
Inverse Mill's ratio -0.034a -0.004
Note: All estimates are significant at 1 percent except those with superscript a and b, which
are significant at 5 percent and not significant, respectively.
16
17. Results
• Cattle prices increase with price of locally produced cooking oil,
• Estimated coefficient of 6-month lagged average monthly rainfall
implies with sufficient rains prices increase slightly,
• Relative to Tigray, real prices were lower in Dire Dawa and larger in
all regions except Somali, which is not different from Tigray,
• Real prices in urban center markets were 25 percent larger,
o May explain markets that appear to have integrated prices due to other
reasons than through actual trade flows.
• The Harar, Boran, and mixed breed sold at larger prices than Arussi;
Danakil and Raya Azebo breeds sold lower, and Zebu and Arussi
were not different.
• Relative to 2005, real prices of cattle were larger during the 2006-
2008 period, lower in 2011, and were no different in 2009 & 2010.
18. Table 3. HPFA estimates of period, festival, and region dummy variables.
Variable Coefficient Elasticity Variable Coefficient Elasticity
Region and population size Year dummy
Afar 0.770 0.173 2006 1.198 0.282
Amhara 0.337 0.072 2007 0.883 0.201
Oromia 1.357 0.325 2008 0.485 0.106
Somali -0.030a -0.006 2009 0.137a 0.029
SNNP 0.338 0.073 2010 0.214a 0.045
Addis Ababa 0.654 0.145 2011 -0.707 -0.136
Dire Dawa -0.438 -0.087
Urban center 1.081 0.251
Festival/season New Year 0.003a 0.001
Eid Alfetir -0.044a -0.009 Christian fasting 0.108a 0.023
Fasika 0.088a 0.018 Muslim Fasting 0.076a 0.016
Note: All coefficient estimates are significant at 1 percent except those with superscript a,
which are not significant.
18
19. Results
• Orthodox Christian and Muslim fasting seasons were no different
from the rest of the year,
o In all years, except 2006, the largest number of fasting days were in March,
during which prices were lower.
• Prices of cattle during the one week period of Eid Alfetir, Fasika, and
New Year were no different from the rest of the year.
• Relative to January, prices were larger during June, July, and
September, and lower in March, November, and December
o This may have to do with the meher agricultural season, when farmers and
their oxen are engaged in crop production,
o November and December, months just before the crops are harvested, are the
most likely period in which farmers are strapped for cash and cattle feed.
• Indirect effects of cattle attributes, region, and period dummies via
cattle grade were mostly significant, although small in magnitude.
• Results of analysis were robust under system and single HPFA
specifications, among which the system performed superior.
20. Summary of Findings
• We can use the changes in explanatory variables and the
estimates elasticities to explain why prices have change
• This decomposition shows that the model’s variables account
for 2.36 of the 2.5 percent growth in real prices (i.e. model
explains 94% of actual price changes)
• Basically, two big factors account for the price change:
1. Rising international prices and the related boom in formal
exports (one might conjecture that informal exports have also
gone up)
2. The changing composition of animals sold; i.e. switch to higher
value animals. This could also be related to higher prices as
well as general commercialization processes.
21. Figure 5. Explaining changes in cattle prices (%)
Body mass
6% Other factors
6%
Transport prices
8%
Exports & int.
prices
36% Male
10%
Age
34%
21
22. Summary of Findings
• Our analysis does not say much about welfare effects, though urban
consumers are obviously losers when cattle prices grow
• Rural effects are more complex, with winners and losers.
The results therefore suggest that:
• Ethiopian cattle market is increasingly integrated into international
markets-with livestock earning much more foreign exchange,
• Some indication that producers and traders have been changing the
composition of herds so as to increase revenue.
• The concern: A limited supply response, because of grazing
constraints, as well as the drought in lowland areas, and the lagged
negative impact of animal sales on stock accumulation,
• The challenge: Ensure a genuine supply response through increased
productivity, especially given increase feed constraints.
Previous works that used HPFA include Teklewold et. al. (2009) for cattle and Ayele et. al. (2006) for shoats and the recent work by Kassie et. al. (2011) uses heteroscedasticity efficient estimation and random parameters logistic models to estimate the implicit prices of indigenous cattle traits.
Note that body mass is included in D but is a function of other factors (i.e. we account for reverse causality problem between body grade and prices
Data collected from 35 markets, 3 of which had 5 or less observations. Each market in which 1 breed is sold should have over 7,500 observation, we included all of those markets with 24 or more observations.There were remarkable monthly growth rates in the December-September 2010 period in which real prices of cattle grew at 47 % b/n Dec-Jan 2010, 25 % b/n March-April 2010, and at19 % b/n August-September 2010This growth in prices was followed by a relatively faster growth in the number of cattle sold in an average market of 23 percent. This growth was heavily influenced by the 594 % growth in Nov -Dec 05, by the 208 % growth in Sept-Oct 06, by the 171 % growth in April-may 07, by the615 % growth in August-Sep 07, by the150 % growth inNov-Dec 08, and by the107 % growth in March-April 2010. Excluding the six fastest growths and decline rates, average monthly volume of sales grew at only 1.4 percent.
There appear to be several distinct periods of price trends, During:The January 2005-March 2007 period, average real and nominal prices grew at monthly rates of 0.6 and 1.9 percent,The April 2007-January 2009 period, when real and nominal prices grew at 1.8 and 4.1 percent (a period of high food inflation in general)The January-November 2009 period: real and nominal price declines in cattle at 3.7 % and other food typesDecember 2009-September 2010: strong recovery (high international prices?) October 2010 onwards: decline in cattle prices (drought in pastoralist areas? government price controls?)This growth in prices was followed by a relatively faster growth in the number of cattle sold in an average market of 23 percent. However, this growth was heavily influenced by the 594 % growth in Nov -Dec 05, by the 208 % growth in Sept-Oct 06, by the 171 % growth in April-may 07, by the615 % growth in August-Sep 07, by the150 % growth in Nov-Dec 08, and by the107 % growth in March-April 2010. Excluding the six fastest growths and decline rates, average monthly volume of sales grew at only 1.4 percent.
This growth in prices was followed by a relatively faster growth in the number of cattle sold in an average market of 23 percent. However, this growth was heavily influenced by the 594 % growth in Nov -Dec 05, by the 208 % growth in Sept-Oct 06, by the 171 % growth in April-may 07, by the615 % growth in August-Sep 07, by the150 % growth in Nov-Dec 08, and by the107 % growth in March-April 2010. Excluding the six fastest growths and decline rates, average monthly volume of sales grew at only 1.4 percent.As we can see from the graph, for most of the period increase in demand for cattle seem to drive prices up. That is, following an increase in demand, the market equilibrates at larger volume of sales which is then followed by increased cattle prices in the subsequent months. The reverse holds when demand declines. An exception to this seem the January-November 2009 period in which increased cattle supply seem to drive prices down. During this period nominal and real prices fell at 3.7 percent and volume of sells increased at 6.1 %,
Other features that describe this data set include that ……… and on average.Data obtained from Ethiopian Customs Authority implies that volumes of
The difference between the two periods was 3.4, 1.3, 1, and 1.3 percent for real value of exports of live animal and meat, real prices of animal transportation, local cooking oil, and Australian beef, respectively.
Elasticity of cattle prices WRT real international beef price was 0.9 – so international price transmission has been very important in recent years!
Elasticity of prices of cattle with respect to (WRT) live animal and meat exports was 0.056, Cattle prices increase by 5.6 cents for a 1 birr increase in value of exports.Elasticity of cattle prices WRT real international beef price was 0.9 – so international price transmission has been very important in recent years!Cattle prices increase with swine prices, although its effect was small Elasticity WRT cattle transportation fares is 0.2; A bull worth 1,000 birr gets priced 1,200 when real price of transportation grow by 1 birr, but its overall effect was small Cattle prices increase with price of locally produced cooking oil,Estimated coefficient of 6-month lagged average monthly rainfall implies with sufficient rains prices increase slightly,Elasticity WRT body mass index was 0.45, a cattle that is fat/moderate sold at 45 percent larger price than moderate/thinReal prices of cattle in a given age category were 54 percent larger relative to ones in immediately younger category. Male cattle sold at prices 36 percent larger than female cattle. The Harar, Boran, and mixed breed sold at larger prices than Arussi; Danakil and Raya Azebo breeds sold lower, and Zebu and Arussi were not different.
Estimated coefficient of 6-month lagged average monthly rainfall implies that if there were sufficient rains that do not force farmers distress sell their cattle, prices increase slightly,
Relative to Tigray, real prices lower in Dire Dawa and larger in all regions except Somali, which is not different from Tigray,Real prices in urban center markets were 25 percent larger, May explain urban center markets that appear to have integrated prices due to other reasons than through actual trade flows.The Harar, Boran, and mixed breed sold at larger prices than Arussi; Danakil and Raya Azebo breeds sold lower, and Zebu and Arussi were not different. Relative to 2005, real prices of cattle were larger during the 2006-2008 period, lower in 2011, and were no different in 2009 & 2010.
November and December, months just before the crops are harvested, are the most likely period in which farmers are strapped for cash.Additional results include:Indirect effects on prices of … only one of the 23 variables was insignificant.
External prices and factors accounted for 1.21 percent,Increases in value of live animal and meat exports accounted for 0.80 percent, increase in prices of local cooking oil (0.016),Increase in transport cost (0.2), global beef (0.2), & swine prices (-0.08)Improvements in cattle attributes accounted for 1.24 percentImprovements in the body mass condition (0.15), increase in the number of young (0.12), mature (0.72), and male cattle (0.24) soldWhile the 0.6 percent average annual growth in prices implied by year dummies can be an upper bound for increase in real prices of cattle not captured by factors included in the analysis, The 0.14 percent that is unexplained by the factors included in the analysis can serve as a lower bound.
Overwhelmingly, trade factors are the basic explanation of price increasesThis is good for livestock producers, many of whom are very poor, but effect on consumers is not clear. In general, consumers will be worse off, but other meat prices have not risen as rapidly, so substitution may have occurred. Also, the poor consume little meat.One remaining concern is whether a supply side (productivity) response is possible given feed constraints, drought in lowlands, and the lagged negative impact of animal sales on stock accumulationGiven high international prices and this limited supply response, high domestic prices could persist for several years to come