JOURNAL REPRINT SERIES
Institute of Food and Agricultural Sciences
THE DERIVED DEMAND FOR IMPORTED
CHEESE IN HONG KONG
Andrew A. Washington and Richard L. Kilmer
JRTC 02-2 December 2002
INTERNATIONAL AGRICULTURAL TRADE
AND POLICY CENTER
MISSION AND SCOPE: The International Agricultural Trade and Policy Center (IATPC) was
established in 1990 in the Food and Resource Economics Department (FRED) of the Institute of
Food and Agricultural Sciences (IFAS) at the University of Florida. Its mission is to provide
information, education, and research directed to immediate and long-term enhancement and
sustainability of international trade and natural resource use. Its scope includes not only trade
and related policy issues, but also agricultural, rural, resource, environmental, food, state,
national and international policies, regulations, and issues that influence trade and development.
The Center's objectives are to:
Serve as a university-wide focal point and resource base for research on international
agricultural trade and trade policy issues
Facilitate dissemination of agricultural trade related research results and publications
Encourage interaction between researchers, business and industry groups, state and
federal agencies, and policymakers in the examination and discussion of agricultural
trade policy questions
Provide support to initiatives that enable a better understanding of trade and policy
issues that impact the competitiveness of Florida and southeastern agriculture
specialty crops and livestock in the U.S. and international markets
THE DERIVED DEMAND FOR IMPORTED CHEESE IN HONG KONG
Andrew A. Washington
(Assistant Professor, Department of Economics, Southern University)
Richard L. Kilmer
(Professor, Food and Resource Economics Department, University of Florida)
This article was published in the International Food and Agribusiness Management Review,
Volume 5(1), Washington, Andrew A. and Richard L. Kilmer, "The derived demand for
imported cheese in Hong Kong," pages 75-86, Copyright (2002), and is posted with permission
from Elsevier Science http://www.elsevier.com/locate/intagr. Copies of the article can be
downloaded and printed only for the reader's personal research and study.
Abstract: The objective of this paper was to provide the U.S. dairy industry with empirical
estimates of Hong Kong's derived demand for imported cheese differentiated by source country of
production. These estimates were used to simulate the effects of EU subsidy reductions on the U.S.
share of Hong Kong cheese imports. Simulation results suggested that Oceania was the primary
beneficiary from EU subsidy reductions. Hong Kong cheese imports from the U.S. were
expected to increase by 12 percent if subsidy reductions continue at the same pace as the 1994
GATT agreement and 21 percent if reductions were twice the pace.
Key Words: Dairy, Cheese, Imports, Demand, GATT, Subsidy.
The Derived Demand for Imported Cheese in Hong Kong'
The Federal Agriculture Improvement and Reform Act of 1996 (FAIR Act) legislated a
phase out of dairy price supports by January 1, 2000. This phase-out was recently extended through
2001. As a result of the General Agreements on Tariffs and Trade (GATT), subsidized U.S. dairy
exports, as well as exports from other countries, were reduced 21 percent by the year 2000 with
more reductions expected as trade negotiations continue. These changes in policy and the reduction
of the number of dairy farms have made gaining a larger share of world dairy exports more
important to U.S. producers/processors. Evidence of this concern was the willingness of the industry
to create and fund the U.S. Dairy Export Council (USDEC) in 1995. USDEC has nine offices
located in nine countries around the world.
Given the increased emphasis on the export market, the dairy industry was in need of
demand information by country and by individual dairy product. Import demand work for
various countries (Dobson, 1992; Dobson, 1995; Zhou and Navakovic) has been minimal and
qualitative in nature. However, the dairy industry was in need of a quantitative analysis of import
demand for various countries in order to provide the U.S. dairy industry with the demand
characteristics of their chosen markets. The U.S. dairy industry needed information upon which
to build export markets and they needed help now in lieu of assistance in the future when the
export markets were developed.
Of all international markets investigated by USDEC, Hong Kong had been identified as
having significant growth potential in demand for U.S. dairy products. This identification was based
1 This article was published in the International Food andAgribusiness Management Review, Volume 5(1),
Washington, Andrew A. and Richard L. Kilmer, "The derived demand for imported cheese in Hong Kong," pages
on (1) milk and dairy product consumption per-capita in Hong Kong tended to be higher than in
other Asian countries (USDEC, 1996), (2) per-capita milk consumption had grown by nearly 10
percent annually since 1990 (FAO Statistics), (3) Hong Kong had experienced vast economic
growth the last two decades (Central Intelligence Agency, 1999), (4) increased trade with Hong
Kong could further expose U.S. dairy products to the emerging dairy markets of China since Hong
Kong was a major re-exporter to China, and (5) the European Union (EU) had supplied Hong Kong
with over 70 percent of its dairy imports which were expected to decline as the removal of export
subsidies progressed. Additionally, due to limited resources, Hong Kong was nearly 100 percent
dependent on dairy imports to satisfy demand. Regarding trade restrictions, Hong Kong pursued a
policy of non-interference in custom practices where dairy imports were subject to import licensing
for sanitary conditions and record keeping purposes only (USDEC, 1996; U.S. State Department,
1997; WTO, 1998). USDEC had an office in Hong Kong, which demonstrated their belief in Hong
Kong's demand potential.
Since 1991, growth in U.S. cheese exports had been strong. For the period 1991 to 1997,
the quantity of cheese exported from the U.S. grew by 20 percent per year. Dobson (1995) notes
that there is great potential in U.S. cheese exports due to the fact that it is highly differentiable
and may command a premium price in international markets. Zhou and Novakovic also note that
in Asian markets, U.S. cheese is considered high quality when compared to cheese from
Australia and New Zealand.
For the last 40 years, Hong Kong cheese imports grew by over 9 percent annually.
However, during this time period, less than 10 percent was imported from the US. The two
regions that dominated this market during this period were Oceania (48 percent) and the EU (30
75-86, Copyright (2002), and is posted with permission from Elsevier Science
htI \\ \\ \ .elsevier.com/locate/intagr.
percent) (FAO Statistics). Given that EU cheese exports to Hong Kong were subsidized and the
U.S. subsidized no dairy exports to Hong Kong, could the U.S. obtain the EU lost sales when
dairy export subsidies were removed by EU, particularly in the cheese market?
The objective of this paper was to provide the U.S. dairy industry with empirically estimated
elasticities of Hong Kong's derived demand for imported cheese differentiated by country of
production. These estimates were used to assess the relative competitiveness of cheese imported
from the U.S. with cheese imported from other countries. The elasticities also were used to simulate
the effects of EU subsidy reductions on the U.S. share of Hong Kong cheese imports. Past studies
that assessed the demand for imports differentiated by source country of production used a utility or
consumer approach to obtain import demand equations. However, given that imported dairy
products were purchased by firms, and that a significant amount of transformation and/or value
added took place after goods reached the importing country, this article estimated demand from a
production approach where imports were inputs into production processes.
Specific objectives were (1) to econometrically estimate the derived demand for imported
cheese in Hong Kong, (2) to utilize the empirically estimated import demand parameters to provide
empirical measures of the sensitivity of demand to changes in total imports, own price, and the
prices of cross country substitutes, and (3) to estimate the effects of export subsidy reductions on the
derived demand for imported cheese in Hong Kong.
The econometric model used to estimate the derived demand for imported cheese into
Hong Kong was the differential factor allocation model (DFAM) derived from the differential
approach to the theory of the firm (Laitinen). This model is given by
fitD xit = OiD Xt + ij D w, + it (1)
where Dx = log(x,)-log(x,_l) and Dw, = log(w,)-log(wt_) were the log change in quantity and price
respectively from period t-1 to t; x, and w, were respectively the quantity and price of Hong
Kong's imported cheese from source country I; f = (f + ft-1)/2, where f was the ith factor
share of total cost; DX was a version of the Divisa input index, where DXt = =1 fitD xit
zi 's were the price coefficients and 0i was the marginal share of the ith input in marginal cost.
Both were parameters to be estimated.1
A key feature of the DFAM was that production theory was imposed on the system to
determine if the data was consistent with theory. Homogeneity and symmetry were imposed and
tested, and negative semi-definiteness was checked by inspection of the eigenvalues of the price
coefficient matrix. The homogeneity property in the DFAM model was satisfied when
- = 0. Symmetry was satisfied when nsz = 7ji (Washington).
When applied to the estimation of the derived demand for cheese imports into Hong
Kong, equation (1) was the ith derived demand equation for imported cheese into Hong Kong
from exporting country i, where i e (US, Oceania, EU, ROW). ROW was the rest of the world,
which in this instance was an aggregation of all imports of cheese into Hong Kong not imported
from the US, Oceania, or the EU. Oceania was an aggregation of Australia and New Zealand.
The Divisa input index was an index of total cheese imports into Hong Kong.f was the total cost
of cheese from source country i divided by the total cost of all cheese imported into Hong Kong.
The w,'s were the prices for imported cheese charged by the exporting countries. x, was the
quantity of cheese imported into Hong Kong from the ith exporting source.
Estimation of the system of equations represented by equation (1) were accomplished by
using the LSQ procedure in the econometric program package, Time Series Processor (TSP),
version 4.4. This procedure used the multivariate Gauss-Newton method to estimate the
parameters in the system. The output from LSQ included parameter estimates, standard errors,
probability values, a goodness of fit measure for each equation (R2), the Durbin Watson statistic
for each equation, and the log likelihood function value for the system (Hall and Cummins,
The system goodness of fit measure used was (Bewley (1986))
R =1 1 (2)
1 + W l(T k)(n 1)
where W* was the Wald statistic that forced all the coefficients in the system to zero; Twas the
number of observations; n was the number of equations in the full system; and k was the number
of regressors in each equation.
The test for AR(1) in the DFAM model was accomplished using the likelihood ratio (LR)
test where the DFAM with AR(1) imposed was the unrestricted model and the DFAM without
AR(1) was the restricted model. In this study, the estimate of the autocorrelation parameter p
was obtained using full maximum likelihood estimation where p was common across equations.
This procedure was found in Berndt and Savin (1975), Green et al. (1978) and Beach and
The DFAM allows for homogeneity, symmetry, and negative semi-definiteness to be
tested, imposed, or checked. The homogeneity property was satisfied when Y cj =0, which
implied that arn = --,1 -'2 i-... n. Imposing this restriction on equation (1) yielded
,D,= ODX,7++ z (D w, Dwnt) + ,. (3)
Equation (3) was estimated using the LSQ procedure in TSP. The resulting log likelihood value
was obtained from the estimation procedure and used in a LR test to determine if the
homogeneity constraint was valid. For this test, the homogeneity-constrained model was the
restricted model that was compared to the unconstrained system. The symmetry constrained ML
estimator was obtained using the LSQ procedure in TSP as well. The property of negative semi-
definiteness was verified by inspection of the eigenvalues of the price coefficient matrix.
Mean-based elasticities were calculated using the constrained parameters resulting from
the estimation procedure. These elasticities were
Equation (4) is the conditional own and cross price elasticity. This is evaluated at the mean factor
share. Equation (4) is the percentage change in the quantity demanded of an imported dairy
product from the ith source country resulting from a 1 percent change in the price of that same
product from source country. Equation (5) is the Divisa index elasticity, which reflected the
effects of a change in the Divisa index on imports from the various source countries. Given that
this index was proportional to total imports, this elasticity reflected the effects of total import
changes on source-specific imports.
Future imported quantities of imported dairy products resulting from reductions in EU
export subsidies were simulated until year 2003, which was the first half of the new World Trade
Organization's (WTO) implementation period. In order to assess the effects of subsidy reductions
on the quantity of imported cheese demanded by Hong Kong, a person must know how subsidy
reductions affected the price that an individual exporting country charged. Since export subsidies
were a policy exclusive to the exporting country, the importing country only realized a lower
price for the products exported under subsidy. Given that imported products were differentiated
by country of origin, the EU-cheese market was viewed as a separate market when analyzing the
effects of export subsidy changes. When subsidies were reduced, this resulted in a fall in the total
exported, thereby increasing the import price of EU-cheese. The increase in the import price was
the only change realized in the Japanese market for EU-cheese. This indicated that a reduction in
export subsidies could be simulated in the DFAM by increasing the price of the subsidized
commodity. However, the effect of a subsidy reduction on prices was still needed. Gardner
(1987) shows that the elasticity of demand price with respect to a one percent change in a
producer subsidy payment is
O%AV 1- /8
where P is the demand price; Vis the subsidy payment; 7 and E were the own price demand and
supply elasticities respectively. Applying equation (6) to export subsidies, it became the
percentage change in the import price of the subsidized product resulting from a 1 percent
change in export subsidy payments. The change in price was used in either of the two forecasting
procedures to assess the changes in import demand.
The Commodity Trade Statistics section of the United Nations provided the data used in
this article. Imported quantities were in metric tons and values were in $1000US. Source
countries were the U.S., Oceania, and the EU. The time period for the data set was from 1962 to
1998. The value of imports was on a cost, insurance, and freight (CIF) basis, which included the
cost of the product, the insurance paid, and the transportation cost. Commodity prices were
calculated by dividing the value of the commodity imported by the quantity, which resulted in a
per-unit cost per kilogram measure. The rest of the world quantities and values were calculated
by subtracting the total quantity and value imported from the U.S., Oceania, and the EU.
3. Empirical Results
Results indicated that the hypothesis of no autocorrelation was rejected at the .05
significance level (Table 1). Therefore, autocorrelation must be accounted for and the order of
this process is AR(1). In addition to autocorrelation, LR tests also were used to test if the data
satisfied the economic properties, homogeneity and symmetry. The results of these tests are in
Table 1. LR tests indicate that the property of homogeneity and symmetry could not be rejected.
The property of negative semi-definiteness was verified by inspection of the eigenvalues of the
price coefficient matrix. This property is validated when all of the eigenvalues are less than or
equal to zero. All eigenvalues were non-positive. Eigenvalues that had zeros up to the fifth
decimal place were considered to be zero.
Since homogeneity and symmetry could not be rejected, results have homogeneity and
symmetry imposed. Results also have AR (1) imposed as well. Since symmetry was imposed, the
lower triangular portion of the price coefficient matrix was exactly equal to upper triangular
portion, and was left blank. The marginal share (0,) coefficients for each equation in the system
were all significant at any reasonable significance level. The marginal shares also were all
positive, indicating that as total imports grew, imports from the individual exporting countries
grew as well. All own-price coefficients were negative, which was to be expected, and with the
exception of the ROW, the own-price coefficients were all significant by at least the .10
significance level. With the exception of the Oceania/ROW cross-price coefficient, all cross-
Likelihood ratio test results for AR(1) and economic constraints
Model Log-likelihood LR 2
AR(1) 213.946 4.008 3.84(1)a
Unrestricted b 213.946
Homogeneity 212.172 3.306 7.81(3)a
Symmetry 210.814 2.715 7.81(3)
a The number of restrictions are in parenthesis.
b The Unrestricted model and the AR(1) model are the same model since No-AR(1) was
price coefficients were positive, indicating that these goods were substitutes. The negative cross-
price coefficient for Oceania and the ROW indicated a possible complementary relationship
between the imports from these two sources; however, these goods were independent given that
the parameter estimate was not statistically different from zero. Of the cross-price coefficients,
three were significant. These were the U.S./EU, U.S./ROW and the Oceania/EU cross-price
coefficients. All indicated that goods from these sources are substitutes (Table 2).
Table 3 presents the Divisia import and conditional own and cross-price elasticities for
Hong Kong's derived demand for imported cheese. The Divisia import elasticities for the U.S.,
Oceania, EU, and the ROW were 1.284, 1.013, 1.196, and .459 respectively. Of these elasticities,
all were significant by at least the .10 significance level. The Divisia import elasticity was
similar to a total import elasticity, which indicated the percentage change in imports from the
exporting countries given a percentage change in total imports. Given that the Divisia index was
proportional to a percentage change in the total quantity imported, the Divisia elasticity indicated
DFAM parameter estimates for Hong Kong imports of cheese
Price Coefficients, Factor
Exporting U.S. Oceaniaa EU ROWb Shares, 0,
U.S. -.1040 .0217 .0512 .0311 .0864
(.0131)c*** (.0265) (.0263)* (.0185)* (.0251)***
Oceania -.1446 .1844 .-0615 .4823
(.0874)* (.0774)** (.0485) (.0645)***
EU -.3192 .0836 .3597
(.1026)** (.0579) (.0647)***
ROW -.0533 .0716
Equation R2 .63 .47 .43 .20
System R =.91
a Australia and New Zealand aggregation.
b ROW= rest of the world.
Asymptotic standard errors are in parentheses.
*** Significant level= .01
** Significant level= .05
* Significant level =. 10
a similar relationship as the total import elasticity. Of all the Divisia elasticities, the U.S.
elasticity was the largest. This indicated that as total imports of cheese into Hong Kong
increased such that the Divisia index increases, U.S. cheese imports into Hong Kong increased
by a larger percent when compared to the increase in imports from other exporting sources.
The own-price elasticities of Hong Kong's derived demand for imported cheese from the
U.S., Oceania, EU and the ROW were -1.546, -.304, -1.061 and -.342 respectively. Except for
Hong Kong Divisia and price elasticities of the derived demand for imported cheese
Exporting Divisia Conditional
Country Import Own-Price Conditional Cross-Price
U.S. Oceaniaa EU ROWb
U.S. 1.284c -1.546 .323 .761 .462
(.373)d (.195) (.393) (.391) (.274)
Oceania 1.013 -.304 .046 .387 -.129
(.136) (.183) (.056) (.163) (.102)
EU 1.196 -1.061 .170 .613 .278
(.215) (.339) (.087) (.257) (.192)
ROW .459 -.342 .200 -.394 .537
(.250) (.344) (.118) (.311) (.372)
a Australia and New Zealand aggregation.
b ROW = rest of the world.
' Italics indicate that the elasticity was significant by at least .10.
d Asymptotic standard errors are in parentheses which were obtained using the Delta
Method in TSP.
the ROW, all own-price elasticities were significant at the .10 significance level. Elasticities
indicated that the derived demand for U.S. and EU cheese in Hong Kong was highly elastic, with
the demand for U.S. cheese being the most elastic of all the exporting sources. The derived
demand for Oceania cheese was inelastic, indicating that cheese from Oceania imported into
Hong Kong was less responsive to price changes than imports from other sources.
Cross-price elasticities indicated a substitutional relationship between exporting sources
for cheese imports into Hong Kong. The cross-price elasticity that stands out was the U.S.-EU
elasticity, which says that if the price of EU cheese increased by 1 percent, the quantity
demanded for U.S. cheese increased by .761 percent. The Oceania-EU cross-price elasticity
indicated that imports from Oceania increased by .387 percent. This suggested that the U.S.
stands to benefit more than Oceania (percentage-wise) when EU dairy export subsidy reductions
lead to increases in the EU cheese price (Table 3).2
Out of commitment to the Uruguay Round (UR) General Agreement on Tariffs and Trade
(GATT), the EU has agreed to reduce export subsidy expenditures by 36 percent during the
period 1995 to 2000. The question thus arises, how would import quantities change given the
continuation of this policy or that new trade policy was more aggressive?
Equation (6) was used to assess the percentage change in demand price resulting from a
percentage change in a producer subsidy payment. Zhou et al. (1998) indicates that the supply
elasticity for the EU is .65 for all milk produced and the own-price demand elasticity for cheese -
0.40. Using these elasticities in equation (6), the elasticity of the cheese demand price with
respect to a subsidy payment was -0.619. A 36 percent reduction over a six-year period was a 6
percent per year reduction on average. Using -0.619, a 6 percent subsidy reduction resulted in a
3.7 percent increase in the demand price per year. A 72 percent subsidy reduction over a six-year
period resulted in a 7.43 percent per year increase in the demand price. These percentages were
use to simulate the effects of EU subsidy reductions at the current rate and twice the current rate.
Since the UR GATT implementation period ended in 2000, the 72 percent reduction was applied
to the period 2001 to 2003.3
Table 4 presents the expected quantities of cheese imported into Hong Kong if the
upcoming World Trade Organization (WTO) agreement continued subsidy reduction at the
current rate or twice the rate of the UR GATT agreement. If reductions continued at the same
pace, imports of U.S. cheese into Hong Kong for the period 1999 to 2003 were expected to
increase from 761 to 851 metric tons. Imports from Oceania were expected to increase from
Hong Kong cheese imports given a 36 and 72 percent EU export subsidy reduction:
Year U.S. Oceaniaa EU ROWb
36% Subsidy Reduction: 1999-03
1999 760.91 5,293.04 1,356.34 130.55
2000 782.41 5,369.17 1,302.87 133.16
2001 804.52 5,446.39 1,251.51 135.81
2002 827.26 5,524.72 1,202.18 138.52
2003 850.64 5,604.17 1,154.79 141.28
36% Subsidy Reduction: 1999-00
72% Subsidy Reduction:2001-03
1999 760.91 5,293.04 1,356.34 130.55
2000 782.41 5,369.17 1,302.87 133.16
2001 826.63 5,523.61 1,200.15 138.47
2002 87335 568248 1 10553 14399
5,293 to 5,604 metric tons for an overall increase of 311 metric tons. Imports from the EU were
expected to decrease by 202 metric tons. Imports from all other sources were expected to
increase by 11 metric tons. If subsidy reductions for the WTO agreement were twice the UR
GATT rate, imports from the U.S. for the same period was expected to increase from 761 to 923
metric tons, an increase of 162 metric tons. Imports from Oceania were expected to increase to
5,846 metric tons, an increase of 553 metric tons. Imports from the EU were down by over 338
metric tons. Imports from all other sources were expected to increase by 19 metric tons.
Overall, simulation results suggested that Oceania was the primary beneficiary from EU
subsidy reductions for both rates of reduction. This was in terms of quantity. In terms of
percentages, Hong Kong imports of cheese from the U.S. were expected to increase by
approximately 12 percent if policy continued and approximately 21 percent if reductions in
subsidies were twice the current rate. Oceania imports increased approximately 6 and 10 percent,
4. Summary, Conclusions, and Implications
This study assessed the competitiveness of U.S. cheese imported into Hong Kong. Given
the possible elimination of U.S. dairy price supports and the 21 percent quantity and 36 percent
expenditure reduction in subsidized dairy exports, U.S. producers of dairy products have gained
interest in obtaining a greater market share of international markets. Given that the U.S. has had a
relatively small market share of world dairy trade, the degree to which U.S. products compete in
international markets was unknown. This article gives the U.S. dairy industry a snapshot of how
cheese products have been competing in Hong Kong from past to present.
When total cheese imports into Hong Kong change, U.S. cheese imports will change by a
larger percent than that of other exporting sources. This means that as Hong Kong's per-capita
income changes and/or population increases, the percentage change in the United States' exports
into Hong Kong will change by more than any other country and the U.S. market share will
The derived demand for U.S. and EU cheese in Hong Kong is highly elastic, with the
demand for U.S. cheese being the most elastic of all the exporting sources. This means that a
change in the U.S. price will cause a larger change in the quantity consumed of U.S. cheese than
that of any other country. If the U.S. reduces the price of its cheese, the total revenue for the U.S.
will increase because the quantity sold will increase by a larger percentage than the decrease in
price. Furthermore, when the U.S. lowers price, this reduces the quantity sold by the EU and the
ROW. This will increase the market share of the US. Oceania is not likely to lower price because
this will decrease its total revenue.
A competitive relationship between exporting sources for cheese imports into Hong Kong
exists between the U.S. and the EU and the ROW. Thus, if the price of EU or ROW cheese were
to change, the quantity of U.S. cheese demanded will also change in the same direction as the EU
and ROW price. Therefore, as EU cheese price increases as the EU reduces its export subsidies,
the U.S. will increase imports into Hong Kong. A competitive relationship does not exist
between the U.S. and Oceania. This indicates that the U.S. is not competitive with Oceania, the
largest importer into Hong Kong.
Finally, the U.S. will gain a larger market share in Hong Kong when the EU decreases its
export subsidies which will increase the price of EU cheese imported into Hong Kong. In terms
of percentages, Hong Kong imports of cheese from the U.S. will increase by approximately 12
percent if policy continued and approximately 21 percent if reductions in subsidies were twice
the current rate. Oceania imports will increase approximately 6 and 10 percent, respectively.
Thus, the U.S. will increase its market share.
In conclusion, the U.S. does not compete with Oceania, the largest cheese importer in
Hong Kong. Oceania is not likely to reduce price because it will reduce total revenue and the
quantity increase will be marginal. In fact, a price change by Oceania does not change the
quantity of U.S. cheese demanded. The demand for U.S. cheese and the demand for Oceania
cheese are independent. The U.S. competes with the EU and the ROW in the Hong Kong cheese
market. The U.S. has incentive to decrease price in order to increase total revenue and increase
market share. The implications are that the U.S. must compete on the basis of price. Second, the
U.S. must determine the characteristics of the cheese that is sold by Oceania and compare them
to the U.S. cheese. This will help the U.S. to become competitive with Oceania through product
An objective of this study was to simulate the effects of the removal of export subsidies
on Hong Kong cheese imports. Given the left-hand side of equation (1), quantity forecasts were
not easily obtained. There were two methods for obtaining quantity forecasts with the DFAM.
The first method was a model-based approach, which used the estimated model as a means of
forecasting future quantities. The model-based forecasting equation for the DFAM was
0, DXt + Ty (log w,t log w,t_-)
x1t = exp + log xt-1 (Al)
It lt wIt-lXlt-l
i=1 t i=1 _i_
Use of equation (Al) required that the Divisa index and prices were exogenous, where
the only unknowns were the individual quantities. Given prices, the Divisa index, and all lag
values, equation (Al) results in a system ofi equations with i unknowns which can be solved for
the x's using the SOLVE procedure in TSP. This procedure uses a Gauss-Seidel algorithm.
The second method was the elasticity based approached, similar to the approach used by
Kastens and Brester (1996). The elasticity-based forecasting equation for the DFAM was
,t = -J + E, [DX, ] x, + x,1 (A2)
where Ej and E, were the price and Divisa elasticities evaluated at the mean. Both procedures
used results from the estimation procedure where the economic properties of homogeneity and
symmetry were imposed.
The first step in the forecasting and simulation procedure was to determine which of the
two approaches was most accurate in terms of forecasting. To determine which method was best,
each of the DFAM systems were estimated using all except the last 5 years of the data sets.
Using both model-based and elasticity-based forecasting methods to forecast the remaining
years, the precision of each of these methods determined which of the two procedures to use in
forecasting and simulating future periods.
Hong Kong's derived demand system for imported cheese was re-estimated using all of
the available years except the last five (1994-1998). Once new estimates were obtained, model-
based and elasticity based forecasting equations were used to forecast imported quantities for the
remaining years. This was done to determine which of the two equations forecasted with the
most precision. Precision of forecasts was determined by the absolute percentage difference in
the forecasted and actual quantities. Table Al presented the absolute percentage difference in the
actual and forecasted values. Results showed that the forecasting precision of the elasticity-based
approach was a 14 percent improvement over model-based forecasts on average. For each
country, forecasts improved by as much as 39 percent when using elasticities instead of the
model to forecasts.
Percentage differences in the actual quantities and forecasts for Hong Kong cheese imports:
Year U.S. Oceaniaa EU ROWb Overall
1994 45.8 6.3 4.5 12.0
1995 58.1 11.0 1.2 25.6
1996 29.1 16.0 16.4 25.7
1997 23.9 16.8 7.4 115.5
1998 53.1 16.2 12.0 124.3
Average 42.0 13.2 8.3 60.6 31.0
1994 23.7 1.4 5.8 21.2
1995 11.6 1.6 2.1 10.0
1996 18.8 6.4 18.9 16.9
1997 40.7 3.1 9.4 27.8
1998 62.7 0.6 18.7 31.9
Average 31.5 2.6 11.0 21.5 16.7
a Australia and New Zealand aggregation.
b ROW = rest of the world.
1. Davis used the differential production approach to estimate the import demand for broad
aggregates. Washington used the differential production approach to estimate source
specific import demand.
2. Elasticity comparisons were not based on statistical tests, but on the relative magnitude of
3. Of the two forecasting methods considered, the elasticity-based method was selected
because it had more precise "in sample" forecasts. Forecast selection results were
summarized in the appendix.
The authors wish to express their appreciation to two anonymous journal reviewers for their
helpful comments. This research was supported by the Florida Agricultural Experiment Station,
and approved for publication as Journal Series No. R-07930.
Beach, C.M., & J.G. MacKinnon. 1979. "Maximum Likelihood Estimation of Singular Equation
Systems with Autoregressive Disturbances." International Economic Review, 20: 459-
Berndt, E.R. & N.E. Savin. 1975. "Estimation and Hypothesis Testing in Singular Equation
Systems with Autoregressive Disturbances." Econometrica, 43: 937-958.
Bewley, R. 1986. Allocation Models Specification, Estimation, and Applications. Cambridge,
Central Intelligence Agency (CIA). 1999. WorldFactbook 1999. Washington D.C.,
Davis, G.C. 1997. "Production Aggregation Bias as a Specification Error in Demand Systems."
American Journal ofAgricultural Economics, 79: 100-109.
Dobson, W.D. 1992. "The Deregulation of the U.S. Dairy Industry: Implications for World Dairy
Markets." Agribusiness, 8: 457-471.
Dobson, W.D. 1995. "Prospects for U.S. Dairy Exports." University of Wisconsin-Madison,
Department of Agricultural Economics Staff Paper No. 382, 20 pages.
Food and Agricultural Organization of the United Nations Statistics (FAO Statistics). 1999.
Rome, Italy, www.fao.org.
Gardner, B.L. 1987. The Economics ofAgricultural Policies. New York, NY: Macmillan.
General Agreements on Tariffs and Trade. 1992. Trade Policy Review Japan Volume 1 Geneva,
General Agreements on Tariffs and Trade. 1995. Trade Policy Review Japan Volume 1 Geneva,
Green, R.D., Z.A. Hassan, & S.R. Johnson. 1978. "Maximum Likelihood Estimation of Linear
Expenditures Systems with Serially Correlated Errors." European Economic Review, 11:
Hall, B.H. & C. Cummins. 1998. Reference Manual Version 4.4. Palo Alto, CA: TSP
Kastens, T.L. & G.W. Brester. 1996. "Model Selection and Forecasting Ability of Theory-
Constrained Food Demand Systems." American Journal ofAgricultural Economics, 2:
Laitinen, K. 1980. The Theory of the Multiproduct Firm. New York, NY: North Holland
Theil, H. 1971. Principles ofEconometrics. New York, NY: John Wiley and Sons.
U.S. Dairy Export Council (USDEC) 1996. Comprehensive Reference Document on World
Dairy Trade Arlington, VA: USDEC.
U.S. Department of State. 1997. "Hong Kong." Country Reports On Economic Policy and Trade
Practices. Washington, D.C.
Washington, A. 2000. The Derived Demandfor Imported Dairy Products in Selected
International Markets. Ph.D dissertation University of Florida, Gainesville.
World Trade Organization. 1998. Trade Policy Review Hong Kong, China 1998. Geneva,
Zhou, M. and A.M. Novakovic. 1996. "Exporting to China: Possibilities and Challenges for U.S.
Dairy Industry." Agribusiness, 12: 1-13.