Household Demand for Meat in Nigeria

Olumide Aborisade
Department of Agriculture and Applied Economics
318 Agricultural Sciences Building, Texas Tech University
Email: [email protected]

Carlos E. Carpio
Department of Agriculture and Applied Economics
301 Agricultural Sciences Building, Texas Tech University
Email: [email protected]

Selected Paper prepared for presentation at the Southern Agricultural Economics

Association’s 2017 Annual Meeting, Mobile, Alabama, February, 4-7 2017

Copyright 2017 by Olumide Aborisade and Carlos E. Carpio. All rights reserved. Readers may
make verbatim copies of this document for non‐commercial purposes by any means, provided that
this copyright notice appears on all such copies.
 Abstract

This study was an application of the Linear Approximate Almost Ideal Demand System (LA-

AIDS) model on household aggregate meat demand in Nigeria. The data used was obtained from

the World Bank’s Living Standards Measurement Study (LSMS) on households in Nigeria.

Previous research had studied demand only at a regional level but this study estimates meat

demand at a national level. The results showed that beef was a necessity while goat, chicken and

mutton were luxuries. The results further revealed that all the meat products considered were

normal goods with own-prices that were negative and consistent with demand theory except

mutton. Goat meat and mutton were price elastic and as such, price changes for these products

will affect their consumers more than consumers of other meat products that were less elastic.

Key words: Linear Approximate Almost Ideal Demand System (LA-AIDS), meat, demand,

elasticity, own-price, cross-price.

2
 1. Introduction
Several studies on household meat demand have been carried out around the world but

relatively few studies have been carried out on household demand in Nigeria and none in recent

times. This is not indicative of a general lack of interest on the part of scholars in the region.

Partly responsible for this dearth in research would be inadequacies in available data in terms of

breadth, depth and scope. Previous studies on household food demand in Nigeria have been at

the state (regional) level. Examples include demand for cassava products in Lagos State (Jumah

et al., 2008), food demand in Kogi and Kwara states (Obayelu et al., 2009), meat demand in Abia

state (Igwe and Onyekwere, 2007), rural household food demand in Ondo state (Fashogbon and

Oni, 2013), etc.

Food is necessary for human existence and meat plays a key role in that existence. Meat

products are rich sources of nutrients that enable human growth and development. Enriched with

high value biological protein and vitamins, meat facilitates the development of the

gastrointestinal tract, cranio-dental features (teeth, jaw, etc.) and posture (Pereira and Vicente,

2013). Its consumption in adequate quantities ensures normal functioning of the immune system,

mucous membranes and metabolic processes (Biesalski, 2005).

In Nigeria, meat, fish and animal products are the fourth most commonly consumed food

group (88.9%) by households. Its consumption lags behind grains and flours (97.2%), oils and

fats (96.8%) and vegetables (96.7%). Compared to other food groups, average weekly household

expenditure was highest for meat, fish and animal products (N1, 359 per week) (National Bureau

of Statistics, 2016). A variety of meat products are purchased and consumed across the country.

The acceptance and popularity of each meat product varies by region. For example, pork is not as

widely accepted in the northern part of the country as it is in the south. This is mostly due to

3
religious reasons. The northern population are predominantly Muslims while the southern

populace are predominantly Christians.

This study uses recent household survey data to examine the response of aggregate meat

demand to variations in price and incomes. Household surveys provide the opportunity to test

theories about household behavior and their response to changes in their economic environment

(Deaton, 1997). We analyze meat demand using the linear approximate almost ideal demand

system (LA-AIDS) model. Prior research on this subject have focused on particular regions and

the use of recent country-level data distinguishes this study from others.

The study is organized as follows: section 2 reviews extant literature on food demand;

section 3 explains the AIDS model; section 4 provides information on the data and the

econometric model; section 5 interprets and discusses the results while section 6 concludes.

2. Literature Review

Investigating the demand for various meat products amongst rural and urban households

in Kenya, Bett et al. (2012) surveyed 930 households across six counties and estimated demand

elasticities using the LA-AIDS model. The authors were particularly interested in comparing the

consumption pattern of the indigenous chicken against the exotic chicken, beef, mutton and

“other meat” products. Their keen interest in the indigenous chicken was based on the belief that

it is easily accessible and readily available in rural and urban areas relative to other meat

products. Their results revealed that indigenous chicken, beef and mutton were necessities as

shown by their expenditure elasticities. Furthermore, the indigenous chicken and beef were

substitutes while the indigenous chicken, exotic chicken and goat meat were complements as

shown by the signs of the cross-price elasticities. The socio-demographic factors suggested to

4
have a significant effect on meat consumption were household location, family size and

proportion of household members.

Studying consumption and expenditure at a disaggregated product level provides an

understanding of consumer demographic factors affecting expenditures on the product.

Attempting to achieve this level of understanding, Nayga (1995) studied US household

consumption and expenditures on fourteen meat products at a disaggregate level. Information on

price and quantity were not available in their data and as such they could not impute price

variables for their empirical model but assumed constancy of relative prices. Using a two-step

censored regression technique, they observed various demographic factors affecting meat

expenditures. Their results also revealed that variables such as household size (number of

persons in the household), age, race (black, white or other race), education and region

(Northeast, Midwest, South, and West) had a statistically significant effect on meat expenditures.

Examining consumer demand for cassava food products in Lagos, Nigeria, Jumah et al.

(2008) analyzed the effect of household socio-demographic characteristics on the consumption

of gari, lafun and fufu (all of which are cassava by-products). They attempted to capture

consumer behavior that might lead to informed policy recommendations and hence potentially

boost productivity, improve food security and reduce poverty through job creation. In order to

accomplish their objective of estimating a separable demand system, they surveyed 300

households residing in Lagos State. Employing a LA-AIDS model, and assuming weak

separability, they estimated a separable demand system for these food products. Their results

provides evidence that characteristics such as religion and residential area (that is, whether or not

the household resided in a low or high income area) explained the variations in the consumption

of the three food products.

5
 Estimating household meat demand at both the aggregate (beef, lamb, pork, chicken) and

disaggregate levels (beef, lamb, fresh pork, ham, bacon, chicken) in Australia, Cashin (1991),

employed the LA-AIDS model. As commonly observed in household studies were incomplete

data often leads to exclusion of food items, fish and mutton were excluded from their analysis

due to data unavailability and reliability. Although all the elasticities were of the appropriate

sign, their results revealed large differences in numerical values of cross-price elasticities as well

as a lot of negative values in the uncompensated cross-prices. This was suggested to be

indicative of a strong income effect or data inadequacies. Regarding elasticities, at the

aggregated product level, beef and lamb were price elastic (absolute value greater than one)

while pork and chicken were inelastic and gross substitutes. At the disaggregated level, fresh

pork and ham were price elastic while beef, lamb, bacon and chicken were inelastic.

Furthermore, fresh pork and chicken were gross complements. Such that if demand increases for

fresh pork, quantity of chicken demanded would increase as well.

Household surveys that contain information on expenditures and quantities of food items

purchased but lacking price information have resulted in a number of approaches to compensate

for the lack of price data in the estimation of a demand system. (Castellón et al., 2015) used

budget shares and Consumer Price Index (CPI) to construct Stone-Lewbel (SL) price indices that

could be used to estimate a demand system where prices were absent. They further evaluated the

likelihood of the Consumer Price Index (CPI) influencing estimated marginal effects and

elasticities. They found that the observed variations in the Stone-Lewbel prices were as a result

of household heterogeneity and not the CPIs.

6
 3. Conceptual Framework

The model used here is the linear approximate almost ideal demand system (LA-AIDS)

which is more frequently used due to its ease of estimation than the almost ideal demand system

(AIDS). The AIDS model is a flexible system of demand equations developed by (Deaton and

Muellbauer, 1980). These demand equations exhibit a linear relationship between the budget

shares of commodities, real total expenditure and relative prices. The AIDS model itself belongs

to the PIGLOG family (Price-Independent Generalized Logarithmic preferences) and thus

possess important desirable properties required in conventional demand analysis. However,

difficulty in estimating the AIDS model due to the non-linear price index gave rise to the LA-

AIDS model (see e.g. Blanciforti and Green 1983 and Alston et al., 1994).

The AIDS demand functions in terms of budget share is given by

(1) 𝑤𝑖 = 𝛼𝑖 + ∑𝑗 𝛾𝑖𝑗 ln𝑝𝑗 + 𝛽𝑖 ln (𝑥⁄𝑃 ) + 𝜀𝑖 i = 1, …,n

where 𝑤𝑖 is the budget share of good 𝑖, 𝑝𝑗 is the price of good j, 𝑥 is the total expenditure, 𝜀𝑖 ’s

are random disturbances and 𝑃 is a price index defined by

1
(2) ln 𝑃 = 𝛼0 + ∑𝑘 𝛼𝑘 ln 𝑝𝑘 + 2 ∑𝑗 ∑𝑘 𝛾𝑘𝑗 ln𝑝𝑘 ln𝑝𝑗 k = 1,…,n

where 𝑝𝑘 is the price of good k and 𝛾𝑘𝑗 are the parameters to be estimated.

Replacing the price index in (2) with Stone’s price index ((𝑃∗ ) gives rise to the linear

approximate almost ideal demand system (LA-AIDS) which is simpler and more frequently used

than the non-linear AIDS as we mentioned earlier. Stone’s Price Index is given by

(3) ln 𝑃∗ = ∑𝑤𝑘 ln 𝑝𝑘

7
The following restrictions are imposed on the parameters of the AIDS model:

(4) ∑𝑛𝑖=1 𝛼𝑖 = 1 ∑𝑛𝑖=1 𝛾𝑖𝑗 = 0 ∑𝑛𝑖=1 𝛽𝑖 = 0 Adding-up

(5) ∑𝑗 𝛾𝑖𝑗 = 0 Homogeneity

(6) 𝛾𝑖𝑗 = 𝛾𝑗𝑖 Symmetry

Provided equations (4), (5) and (6) hold, the AIDS model given in equation (1) represents a

system of demand functions that add up to total expenditure (∑ 𝑤𝑖 = 1), are homogenous of

degree zero in prices and total expenditure, and satisfy Slutsky’s symmetry.

Studies on household demand assume a multistage budgeting process within the household and

weak separability. That is, a household first allocates its income across broad food and non-food

categories then further allocates its income amongst items in each category We assume that the

meat group is weakly separable from other food and commodity groups. This assumption allows

us to limit the number of prices appearing in each equation to the four products in the meat group

(see e.g. Cashin, 1991)

4. Empirical Analysis

Data

The data used in this study is cross-sectional data obtained from the Nigerian General

Household Survey (GHS) which was implemented between 2015 -2016. The data collection

process was a team effort between the Nigerian National Bureau of Statistics (NBS), the Federal

Ministry of Agriculture and Rural Development and the World Bank’s Living Standards

Measurement Study (LSMS). The survey captures information on household characteristics

including age, marital status, religion, education, illnesses, disability, child anthropometrics,

consumption, food security and shocks and several other aspects of household living.
8
 Respondents were drawn from each of the thirty-seven states in Nigeria. There were ten

households in each enumeration area (EA or primary sampling unit) and each state had sixty

enumeration areas. The total number of enumeration areas for all thirty-seven states in the

country was twenty two thousand, two hundred (22, 200).

Regarding food unit measures, respondents were allowed to report purchase and

consumption in units that they were most familiar with. We estimated kilogram-equivalent

measures for responses on meat products reported in non-standard units using the national

conversion factor for these non-standard measurements. However, conversion factors for meat

products was unavailable. Rather than dropping these observations, we adopted the conversion

factors for fresh fish since those were available and were the closest proxies we could use for

meat. In some instances, households had reported conflicting measures for purchase and

consumption. For example, if a household reported the quantity of beef purchased in terms of

heap small and reported the quantity consumed in kilograms, we converted heap small to

kilograms.

Commonly observed in micro-data such as household surveys is the issue of missing

values. According to Nayga (1995) and Deaton (1997), the proportion of the sample with zero

consumption or expenditure makes household-level data sometimes problematic. Constrained by

the need to have sufficient information on both quantity and expenditures that could be

estimated, we selected four meat products (beef, goat, chicken and mutton) for demand

estimation. These products had more observations than other meat products - some of which had

less than five observations. We excluded pork, bush meat (wild game meat), duck and canned

beef. Households that reported no information on quantity purchased and expenditure for all four

9
meat products were dropped. We retained those that had information on at least one of the four

meat products.

Price information was not included in the survey. We estimated unit values for each meat

product by dividing the expenditures on that product by its purchased quantity. We regressed

these unit values on variables region and sector (rural or urban) in order to impute the prices. The

imputed prices account for household price variability according to region and sector. Since

information on the time (day, week or month) the survey was conducted was unavailable, we

could not account for price variations according to day, week or month.

The sample size for this study was 4, 580 households. Controlling for households with no

information on purchase, consumption and expenditure reduced our sample size to 2, 412.

Dropping households that failed to report expenditure information may cause self-selection bias.

To assess this, we compared the means of the variables used in the analyses with and without the

dropped observations and it was not significantly different (see Nayga, 1995). The variables used

in this study include region, sector, prices and quantities purchased of beef, chicken, goat meat,

mutton and their total expenditures.

Econometric Model

The econometric model used is the LA-AIDS model with Stone’s price index given by

(7) 𝑤𝑖 = 𝛼𝑖∗ + ∑𝑗 𝛾𝑖𝑗 ln𝑝𝑗 + 𝛽𝑖 ln (𝑥⁄𝑃∗ ) + 𝜀𝑖∗

where 𝛼𝑖∗ = 𝛼𝑖 − 𝛽𝑖 𝛼𝑖 𝑎𝑛𝑑 ln 𝑃∗ = ∑𝑤𝑘 ln 𝑝𝑘 . Imposing homogeneity in equation (5) by

substitution, we estimated equation (8) rather than equation (7)

(8) 𝑤𝑖 = 𝛼𝑖∗ + ∑𝑛−1

̅̅̅ 𝑗 𝛾𝑖𝑗 ln𝑝𝑗 + 𝛽𝑖 ln(𝑥̅⁄𝑃∗ )

10
where ̅̅̅
𝑤𝑖 indicates the sample mean of the budget share of good i, 𝑥̅ is the sample mean of the

log of real total expenditures. According to Green and Alston (1990), the expenditure,

uncompensated and compensated elasticities for the LA-AIDS model are given by

Expenditure Elasticity:

𝛽
(9) 𝑒𝑖 = 1 + 𝑤𝑖
𝑖

Uncompensated (Marshallian) price elasticity:

𝛾𝑖𝑗 𝛽𝑖 𝛼𝑗 𝛽
(10) 𝜂𝑖𝑗 = −𝛿𝑖𝑗 + − − 𝑤𝑖 ∑𝑘 𝛾𝑘𝑗 𝑙𝑛 𝑃𝑘
𝑤𝑖 𝑤𝑖 𝑖

where 𝛿𝑖𝑗 is the Kronecker delta that is unity if i=j and zero otherwise

Substituting for 𝜂𝑖𝑗 in the Slutsky equation we get the compensated elasticities(𝜂𝑖𝑗 ∗ ). The

Slutksy equation is given by

𝛽
(11) 𝜂𝑖𝑗 ∗ = 𝜂𝑖𝑗 + 𝑤𝑗 (1 + 𝑤𝑖 )
𝑖

Hence, the compensated elasticity is given by,

𝛾𝑖𝑗 𝛽𝑖 𝛼𝑗 𝛽 𝛽𝑖
(12) 𝜂𝑖𝑗 ∗ = −𝛿𝑖𝑗 + − − 𝑤𝑖 ∑𝑘 𝛾𝑘𝑗 ln 𝑃𝑘 + 𝑤𝑗 (1 + 𝑤 )
𝑤𝑖 𝑤𝑖 𝑖 𝑖

Parameter estimates and elasticities were estimated using the iterated seemingly unrelated

regression (SUR) procedure in SAS.

5. Results and Discussion

Table 1 presents the descriptive statistics of the variables used in this study. Regarding the

quantity purchased of each meat product, beef was the highest (1935kg) followed by goat

11
(554kg) then chicken (404kg) and mutton (361). In terms of price, beef was the most expensive

(N1253.89/kg) followed by chicken (N980.81/kg) and mutton was the least expensive

(N712.02/kg).

Table 2 presents the parameter estimates of the LA-AIDS model. The estimates were

estimated with restrictions of homogeneity and symmetry imposed. These restrictions were not

satisfied. Ten of the fourteen parameter estimates were statistically significant. The own-price

coefficient of beef (0.0867) in the beef equation means that a percent increase in the price of beef

would increase the household’s budget share for beef by approximately N0.09 (Nine kobo). The

own-price coefficient of goat meat (-0.2560) in the goat equation indicates that a percent increase

in the price of goat meat would decrease the household’s budget share for goat meat by

approximately N0.26 (Twenty-six kobo).

Table 3 presents the uncompensated (Marshallian) own and cross-price elasticities. Three of

the four own-price elasticities had the expected negative sign and is consistent with demand

theory. Mutton has a positive own-price elasticity. This suggests that an increase in its price

increases its demand. This result was consistent with a similar study on meat demand in Nigeria.

Adetunji and Rauf (2012) reported a positive own-price elasticity for mutton in their study on

household demand for meat in Southwest Nigeria. It had a positive sign and it was greater than

one. Mutton is a good source of lean meat but not as regularly consumed like beef, goat and

chicken. In addition to that, Muslims in Nigeria use it to celebrate the festival Id-el-Kabir.

During this festive period, price for mutton increases. As such, the price increase is seasonal.

The own-price elasticity for goat and mutton were greater than one in absolute value

indicating price elastic demands while beef and chicken were price inelastic. Variations in the

price of goat or mutton would affect its consumers more than consumers of beef and chicken.
12
Three of the four expenditure elasticities were greater in absolute value than their corresponding

own-price elasticities. The positive expenditure elasticities for all the meat types indicate they

were normal goods, these elasticities were also significant at the 1% level. The expenditure

elasticity also reveals that with the exception of beef which is a necessity ( 𝜂𝑖 = −0.8191), all

the other meat types were luxuries1 ( 𝜂𝑖 > 1). Among the luxury meats, chicken had the highest

expenditure elasticity (𝜂𝑖 = 2.0266). While beef and goat meat were gross substitutes, beef and

chicken were gross complements.

Table 4 presents the compensated own and cross-price elasticities. Again, with the

exception of mutton, own-price elasticities for beef, goat and chicken were negative, consistent

with the inverse relationship between price and demand. All pairs of goods were substitutes

except beef and chicken, beef and mutton and chicken and mutton.

Summary and Conclusions

In this study, the linear approximate almost ideal demand system (LA-AIDS) has been used to

estimate aggregate demand for meat products in Nigeria using household survey data from the

World Bank’s Living Standards Measurement Study. Three of the four own-price elasticities

were negative and consistent with demand theory. Mutton had a positive own-price effect which

is consistent with similar studies. The results suggest that all the meat products considered were

normal goods as revealed in their positive expenditure elasticities. Our results indicate that beef

is a necessity and also suggest that goat and mutton are price sensitive. Hence, policies that cause

variations in the prices of goat and mutton would potentially cause welfare distortions.

1
Blanciforti and Green (1983) explain that negative 𝛽𝑖′ 𝑠 indicate necessities while positive 𝛽𝑖′ 𝑠 indicate luxuries. A
negative 𝛽𝑖 implies that 𝜂𝑖 < 1 while a positive 𝛽𝑖 implies that 𝜂𝑖 > 1.

13
 Some of the limitations of this research were due to the proportion of missing

observations in the data. Future research should consider incorporating socio-demographic

variables as well as model expenditures on meat products at a disaggregated level. This will

enable all stakeholders in the meat industry capture the various factors affecting household

expenditures.

14
References

Adetunji, M. O and M. O. Rauf. (2012). "Analysis of Household Demand for Meat, in

Southwest, Nigeria." Global Journal of Science Frontier Research Agriculture and
Biology, 15-21.
Alston, J. M., K. A. Foster, and R. D. Green. (1994). "Estimating Elasticities with the Linear
Approximate Almost Ideal Demand System: Some Monte Carlo Results." The Review of
Economics and Statistics, 76(2), 351-356.
Bett, H. K., M. P. Musyoka, K. J. Peters, and W. Bokelmann. (2012). "Demand for Meat in the
Rural and Urban Areas of Kenya: A Focus on the Indigenous Chicken." Economics
Research International, 1-11.
Biesalski, H.-K. (2005). "Meat as a Component of a Healthy Diet - Are there any Risks or
Benefits if Meat is Avoided in the Diet?" Meat Science, 70, 509-524.
Blanciforti, L. and R. Green. (1983). "An Almost Ideal Demand System Incorporating Habits:
An Analysis of Expenditures on Food and Aggregate Commodity Groups." The Review
of Economics and Statistics, 65(3), 511-515.
Cashin, P. (1991). "A Model of the Disaggregated Demand for Meat in Australia." Australian
Journal of Agricultural Economics, 35(3), 263-283.
Castellón, C.E., T. Boonsaeng and C. E. Carpio. (2015). "Demand System Estimation in the
Absence of Price Data: An Application of Stone-Lewbel Price Indices." Applied
Economics, 47(6), 553-568.
Deaton, A. (1997). The Analysis of Household Surveys: A Microeconometric Approach to
Development Policy. Baltimore, Maryland: The Johns Hopkins University Press.
Deaton, A. and Muellbauer, J. (1980). "An Almost Ideal DemandSystem." The American
Economic Review, 312-326.
Fashogbon, A. E. and O. A. Oni. (2013). "Heterogeneity in Rural Household Food Demand and
its Determinants in Ondo State, Nigeria: An Application of Quadratic Almost Ideal
Demand System." Journal of Agricultural Science, 5(2), 169 - 177.
Green, R. and J. M. Alston. (1990). "Elaticities in AIDS Models." American Journal of
Agricultural Economics, 72(2), 442-445.
Igwe, K. C. and O. N. Onyekwere. (2007). "Meat Demand Analysis in Umuahia Metropolis Abia
State, Nigeria." Agricultural Journal, 2(5), 550-554.
Jumah, A., A. O. Dipeolu, I. A. Ayinde and K. Adebayo. (2008). "An LA-AIDS Analysis of
Cassava Food Products Demand in Lagos." The Journal of Developing Areas, 41(2), 110-
117.

15
National Bureau of Statistics. (2016). LSMS - Integrated Surveys on Agriculture General
Household Survey Panel 2015/2016. Nigerian National Bureau of Statistics.
Nayga, Jr. R. M. (1995). "Microdata Expenditure Analysis of Disaggregate Meat Products."
Review of Agricultural Economics, 17(3), 275-285.
Obayelu, A. E., V. O. Okoruwa and O. I. Y. Ajani. (2009). "Cross-sectional Analysis of Food
Demand in the North Central Nigeria." China Agricultural Economic Review, 173 -193.
Pereira, P. M. C. C and A. F. R. B Vicente. (2013). "Meat Nutritional Composition and Nutritive
Role in the Human Diet." Meat Science, 93, 586-592.

16
Table 1: Descriptive Statistics of Variables used in the LA-AIDS Model

Variable Unit N Mean Std Dev Minimum Maximum

Beef Quantity kg 1935 1.16 1.07 0.00 15.30

Goat Quantity kg 554 1.16 0.95 0.02 15.00

Chicken Quantity kg 404 1.55 0.93 0.20 10.00

Mutton Quantity kg 361 1.02 0.71 0.20 6.00

Beef Price (N/kg) 2412 1253.89 586.63 464.14 2530.00

Goat Price (N/kg) 2412 860.08 93.31 758.44 1000.00

Chicken Price (N/kg) 2412 980.81 107.41 834.00 1133.14

Mutton Price) (N/kg) 2412 712.02 116.69 525.20 896.09

Expenditure (N/kg) 2412 1730.00 1826.54 2.53 21945.12

1$USD = N 312.81
The regions are comprised of the following states: North Central – Plateau, Kogi, Benue, Nasarawa,
Niger, Kwara and FCT Abuja; North East – Adamawa, Taraba, Bauchi, Borno, Gombe and Yobe; North
West – Kano, Jigawa, Kaduna, Zamfara, Katsina, Kebbi and Sokoto; South East – Abia, Anambra,
Ebonyi, Enugu and Imo; South South – Akwa Ibom, Cross River, Bayelsa, Delta, Edo and Rivers; South
West – Lagos, Ekiti, Osun, Ogun, Ondo and Oyo.

17
Table 2: Parameter Estimates of the LA-AIDS Model
Meat
𝛼𝑖 𝛾𝑖1 𝛾𝑖2 𝛾𝑖3 𝛾𝑖4 𝛽𝑖
Products
Beef 0.6776* 0.0867* 0.0281 -0.0155 -0.0993* -0.1272*
(w1) (0.0102) (0.0184) (0.0148) (0.0104) (0.0101 (0.0104)

Goat 0.1243* -0.2560 0.1491* 0.0788* 0.0174**

(w2) (0.0096) (0.0297) (0.0235) (0.0189) (0.0079)

Chicken 0.0570* -0.01129 -0.1223* 0.0889*

(w3) (0.0106) (0.0342) (0.0232) (0.006)

Mutton 0.1411* 0.1427* 0.0209*

(w4) (0.0086) (0.0219) (0.0058)
Note: One asterisk (*) indicates significance at the 1% level and two asterisks (**) indicates
significance at the 5% level. Values in parentheses are the standard errors.

18
Table 3: Uncompensated Expenditure and Price Elasticities in the LA-AIDS Model
Meat 𝜀𝑖𝐵 𝜀𝑖𝐺 𝜀𝑖𝐶 𝜀𝑖𝑀 𝜂𝑖
Products
-0.5864* 0.0427 -0.1205* -0.1550* 0.8191*
Beef
(0.0442) (0.0230) (0.0194) (0.0161) (0.0147)

0.0012 -2.8499* 1.1445* 0.5786* 1.1256*

Goat
(0.1418) (0.2144) (0.1721) (0.1369) (0.0568)

-1.8272* 1.7055* -0.5717 -1.3333* 2.0266*

Chicken
(0.1928) (0.2764) (0.3997) (0.2708) (0.0653)

-1.8547* 1.0967* -1.5497* 1.0161* 1.2916*

Mutton
(0.1944) (0.2649) (0.3270) (0.3058) (0.0804)
Note: One asterisk (*) indicates significance at the 1% level. Values in parentheses are the
standard errors. Elasticities are calculated from equations (9), and (10) for expenditure (𝜂𝑖 ) and
uncompensated (Marshallian) own - (𝜀𝑖𝑖 ) and cross-price (𝜀𝑖𝑗 ) elasticities respectively.

19
Table 4: Compensated Expenditure and Price Elasticities in the LA-AIDS Model
Meat 𝜀𝑖𝐵 𝜺𝒊𝑮 𝜺𝒊𝑪 𝜺𝒊𝑴 𝜼𝒊
Products
Beef -0.0103 0.1562* -0.0495** -0.0963* 0.8191*
(0.0363) (0.0236) (0.0201) (0.0164) (0.0147)

Goat 0.7928* -2.6940* 1.2420* 0.6592* 1.1256*

(0.1196) (0.2147) (0.1729) (0.1370) (0.0568)

Chicken -0.4020** 1.9863* -0.3961 -1.1882* 2.0266*

(0.1635) (0.2765) (0.4007) (0.2709) (0.0653)

Mutton -0.9464* 1.2756* -1.4378* 1.1085* 1.2916*

(0.1612) (0.2651) (0.3278) (0.3061) (0.0804)

Note: One asterisk (*) indicates significance at the 1% level and two asterisks (**) indicates
significance at the 5% level. Values in parentheses are the standard errors. Elasticities are
calculated from equations (9), and (11) for expenditure (𝜂𝑖 ) and compensated own - (𝜀𝑖𝑖 ) and
cross-price (𝜀𝑖𝑗 ) elasticities respectively.

20