Demand Elasticities: A Comperative Study

Zehra KASNAKOGLU Erkan ERDIL

Haluk KASNAKOGLU

Middle East Technical University

Department of Economics

06531 Ankara, Turkey

E.mail: zkasnak@metu.edu.tr

1. INTRODUCTION

The subject of this study is to provide a consistent set of econometrically estimated demand parameters (own price, cross-price and income elasticities) for 20 food/feed commodities and 42 countries/regions. The estimated parameters are to be used in the development of a global modelling framework that will form the basis for linking consistently the markets of basic food commodities in order to facilitate the situation and short-term outlook for basic foodstuffs, as well as to update the demand parameters of the World Food Model (WFM).

The nature of the planned utilization of the estimated parameters, imposes certain restrictions on the econometric estimation of the demand functions:

1. Functional Form Selection and Specification: As the estimated parameters will be used in large scale models with many countries and commodities in the form of elasticities rather than demand functions, it is necessary that the demand equations to be estimated have a constant elasticity form.

1.

2. Variable Selection and Specification: As the estimated elasticities will be used in large scale models with many countries and commodities the set of variables used, and specification of the used variables should be the same for different countries. That is use of variables with different lag structures for different countries and/or same variables deflated and/or scaled differently for different equations and countries, which can improve the statistical significance of the estimates, are not desirable in this study.

2.

3. Data Selections and Specification: Since the domestic prices for all commodities and/or all countries and regions are not readily available, the international prices are used for all commodity groups and countries, as an approximation to domestic prices. This probably is the most important restriction imposed on the validity of the estimated elasticities for two main reasons: First of all changes in international prices may not reflect changes in domestic markets. This is less of a problem to the extent the countries and commodities in question are freely traded in international markets and hence, the transmission elasticities are close to 1 or, at least the price trends in domestic markets are the same as those in international markets. Secondly, the definition of international prices and specification of international prices for different countries is not an easy task. The data availability restricts the international prices used to average annual export or import prices of a country. The methodology used is to use export prices if the country is in general an exporter and the import prices if the country in general is a net importer of the commodity in question. This approach is more of a problem, when the country in question does not trade the concerned commodity(international prices are not available) or trades in small amounts(international prices are nor representative) and/or switches from net exporter to net importer position and v.s.

2. THE MODEL

The formal models used in elasticity estimations are given below:

ln Q_ij = a_0ij + a_1ij lnP_ij + a_2ijlnY_i + S b_kij lnP_ki + a_3ijT + u_ij

where:

Q_ij = Domestic consumption of j^th commodity in i^th country/region (mt)

P_ij = Price of j^th commodity in i^th country/region(US$/mt)

Y_ij = Real GDP per capita of i^th country/region in 1987 US$ prices

P_ki = Price of k^th commodity which is either a substitute or a complement

for j^th commodity in i^th country/region (US$/mt)

T = Time trend

Since loglinear demand functions are employed in the study, the estimated slope coefficients will give price and income elasticities. Thus:

a_1ij = Own price elasticity of commodity j in country i

a_2ij = Income elasticity of commodity j in country i

b_kij = Cross price elasticity of k^th and j^th commodity in country i.

The commodities are divided into four main groups:

i) Meat and meat products (ovine, bovine, poultry, pork)

ii) Milk and milk products (milk, cheese, butter, dry milk)

iii) Grains (wheat,maize,rice)

iv) Oil and oil products (soybean, perennial, other)

v) Feeds (wheat, maize, rice, coarse grains feeds; soybean, perennial, other oil cakes)

In the demand function for a given product, price of that product and the prices of all other products in the same group are included as explanatory variables. From the other groups main or representative products are chosen and the prices of those products are included in the models.

Since the time series data on the prices of products in most of the countries are not available, instead of the domestic prices of these products, international prices are used. If the country in question is mainly an exporter of the commodity, then the export prices are used, if the country is an importer of the commodity in the period studied, then the import prices are adopted. The real per capita GNP of the country, measured in constant 1987 US$ is also included as an explanatory variable.

The models are first estimated by using Ordinary Least Squares(OLS) estimation technique. Since most of the models estimated by OLS method are autocorrelated, AR(1) estimation process is used next in order to eliminate the autocorrelation problem. In these estimations an iterative approach is used. Some of the estimated slope coefficients (estimated elasticites) which are found to be statistically insignificant are deleted from the regression models after economic considerations and data analysis and the models are reestimated. Own price variable is never deleted from the models, even if insignificant since it is considered as a theoretically must variable.

If the above classical estimation techniques failed for a commodity, then the Bayesian Estimation procedure is applied under proper inequality constrains such as a_1ij < 0 and a_2ij >0.

3. SUMMARY OF FINDINGS

Own price, cross price and income elasticities are estimated and analysed for each of the 20 products for each of the 40 country/regions. In this paper we only present a summary of aggregated results for the world, 10 sub-regions and Turkey¹. Table 1 presents the average elasticities estimated for the 40 country/regions and 20 products. The averages in Table 1 are simple arithmetic averages of the estimated non-zero elasticities. In Table A1, average elasticities are also presented for each of the 10 sub-regions. These aggregate tables are produced a) to give the reader a general idea about the magnitudes of the elasticities estimated, b)to provide a bench mark for elasticities estimated for individual countries, c) to provide approximations for elasticities not being able to estimate.

_________________________________________

¹Results for individual products and countries can be found in Kasnakoğlu, Z., E. Erdil and H. Kasnakoğlu (2000). “Estimation of Demand Elasticities” FAO Working Paper, Rome, April.

Table 1: Aggregate Demand Elasticities

WORLD	Bovine	Ovine	Poultry	Pork	Wheat	Maize	Rice	Milk	Butter	Dry milk	Cheese	Soya Oil	Perr. Oil	Other Oil	Wheat Feed	Maize Feed	Rice Feed	Coarse Grains	Soya Feed	Other Feed	Income
Bovine	-0.19	0.11	0.16	0.16	0.09	-	-	-	-	-	0.11	-	-	-	-	-	-	-	-	-	0.63
Ovine	0.10	-0.16	0.11	0.10	0.16	-	-	-	-	-	0.13	-	-	-	-	-	-	-	-	-	0.61
Poultry	0.12	0.14	-0.13	0.10	0.09	-	-	-	-	-	0.17	-	-	-	-	-	-	-	-	-	0.77
Pork	0.12	0.09	0.09	-0.09	0.07	-	-	-	-	-	0.11	-	-	-	-	-	-	-	-	-	0.64
Wheat	0.11	-	-	-	-0.14	0.11	0.09	-	-	-	0.22	-	-	-	-	-	-	-	-	-	0.60
Maize	0.21	-	-	-	0.16	-0.16	0.11	-	-	-	0.31	-	-	-	-	-	-	-	-	-	0.67
Rice	0.11	-	-	-	0.13	0.11	-0.12	-	-	-	0.16	-	-	-	-	-	-	-	-	-	0.53
Milk	0.11	-	-	-	0.10	-	-	-0.06	0.05	0.07	0.14	-	-	-	-	-	-	-	-	-	0.80
Butter	0.22	-	-	-	0.25	-	-	0.17	-0.47	0.32	0.26	-	-	-	-	-	-	-