demand for fish and its substitutes malaysia: evidence of habit formation and...

Pertanika J. Soc. Sci. & Hum. 7(2): ] 11 - 119 (1999) ISSN: 0128-7702© Universiti Putra Malaysia Press

Demand for Fish and its Substitutes in Malaysia: Evidence of HabitFormation and Structural Change

NIK MUSTAPHA R. ABDULlAH and AHMAD ZUBAIDI BAHAROMSHAHFaculty ofEconomics and Management

Universiti Putra Malaysia43400 UPM Serdang, Selangor, Malaysia

Keywords: demand, Iish, habit fonnation, structural change

ABSTRAK

Kajian ini membincangkan isu perubahan citarasa pengguna terhadap permintaan ikan dandaging di Malaysia. Kedua-dua kesan tabiat berterusan dan perubahan struktur mengenaikegemaran pengguna terhadap ikan dan penggantinya dikaji secara empirik dengan menggunakandata dari 1960 hingga 1990 dan pendekatan Sistem Permintaan Hampir Unggul (AIDS). Denganmenggunakan model dinamik AIDS, terdapat pembentukan tabiat yang berterusan di dalampermintaan terhadap ikan, ayam dan khinzir tetapi kuantiti daging lembu dan daging kambingyang dibeli dimasa lepas akan mengurangkan agihan belanjawan semasa ke atas daging-dagingtersebut. Perubahan struktur mengenai kegemaran pengguna diuji dengan menggunakan prosedurCUSUMSQ dan didapati tiada terdapat selangan struktur bagi persamaan ikan. Ini menunjukkanbahawa citarasa pengguna tidak beruba.h terhadap ikan, konsisten dengan anggapan a priori,tetapi penduduk yang lebih peka kepada cara hidup yang lebih sihat, mengubah kegemaranmereka daripada daging merah dan sekarang meminta lebih banyak daging putih, termasuk ikan.

ABSTRACf

This paper addresses the issue of changes in consumers' tastes in the demand for fish and meatproducts in Malaysia. Both, habit persistent effect and structural change in consumers' preferencestowards fish and its substitutes are empirically examined using data from 1960 to 1990 with anAlmost Ideal Demand System (AIDS) approach. In the dynamic AIDS model, it was found thatthere was a pervasiveness of habit formation in the demand for fish, chicken and pork but theamount of beef and mutton purchased during the last period tends to lower current budgetallocation on these meat products. The structural change in consumers' preferences were testedusing CUSUMSQ procedure and it was found that there was no structural break in the fishequation. This indicates that consumers' tastes do not change with respect to fish, consistent witha priori expectation, but the more health-conscious population are changing their preferencesaway from the red meats and currently demanding a bigger amount of white meat, which includesfish.

INTRODUCTION

Malaysia is a comparatively small, multi-racialand multi-religious country in Southeast Asia.For the past several years, the Malaysian economyhas been growing at the rate of eight per centper annum. Slightly less than half of itspopulation are Malays, followed by Chinese,Indians and other minority races. The Malaysbeing Moslems cannot eat pork while the Indiansbeing predominantly Hindus are prohibited fromconsuming beef. Whilst meat-based protein

consumption is dependent on the racial andreligious fabrics of the population fish is basicallyacceptable to all irrespective of income levels.Thus, it is only natural to expect that fish tendsto dominate over other meat consumption inthis country.

Around mid-seventies and early eightiesthere was a sudden increase in the demand forseafood worldwide. This shift in the consumptionpattern was partly attributed to a very importantdiscovery in the medical field on the relationship

Nik Mustapha R. Abdullah and Ahmad Zubaidi Baharomshah

TABLE 1Average per capita consumption and budget

shares for fish and meat products in Malaysia,1960-1990

by Blanciforti and Green (1983), Eales andUnnevehr (1988), Fulponi (1989), Moschini andMeilke (1989), and Chalfant et al. (1989). Severalstudies have also been carried out in this regionusing the same model. Examples are studies byAhmad Zubaidi and Zainal Abidin Mohamed(1993) and Nik Mustapha et al. (1994). Exceptfor Blanciforti and Green (1983), all these studieshave applied the linear approximation of AIDSmodel (LA/AIDS) by using the Stone's index toobtain price and income elasticity estimates.

Most of these applications however utilizedstatic demand model (static AIDS) in whichconsumers are assumed to fully andinstantaneously adjust their optimal purchase ofcommodities to current changes in prices andincome. One problem of the static AIDS modelis that it ignores the features of persistence inhabits and the possibility of dynamic behavior inconsumer demand. It has been suggested thatinappropriate specification of the dynamicbehavior may account for the rejection oftheoretically based demand conditions (Deatonand Muellbauer, 1980). Several studies that havesuccessfully incorporated this habit formationproblem were Blanciforti and Green (1983),Burton and Young (1992) and Molina (1994),among others.

Another important aspect of demand studiesis the issue of structural change in demand forfish and meat products. Statistical instability isoften interpreted to mean that some underlyingparameters in the utility function have changed.For example, changes in own-price elasticity(taste) or income elasticity affect the stability ofthe demand equations. Structural instabilitymay result in model that may be correct in onesample period but not in another period.Structural change in Malaysian fish and meat

between saturated fats and nutrition and health.Seafood was found to be particularly healthybecause of its Omega-3 fatty acids which wasoften associated with reduced heart disease andneurological disorders (Lees, 1988). Thediscovery has affected the pattern of meatconsumption worldwide including many of thedeveloping economies.

The importance of fish products (whitemeat) in the consumer's diet and budget shareis not entirely new in Malaysia. Being a relativelypoor country then and maritime in nature, fishhas naturally been a life line for the majority ofits population. In 1990 per capita fishconsumption was estimated at 37.5 kg per annumwhile per capita consumption for beef, poultry,mutton and pork were relatively lower at 3.49kg,20.40kg, 0.50kg and 10.29kg, respectively (DVS,1992). The high per capita consumption of fishrelative to other meat products is not only amanifestation of the health concerns but mostimportantly fish is the cheapest form of proteinmeat available and acceptable to all races andreligions in Malaysia.

Figs. 1 and 2 illustrate the per capitaconsumption trends for [lsh and other meats inMalaysia from 1960 to 1990. The summarystatistics of these trends are given in Table 1.The most striking feature in these diagrams isthe steady increase in the per capita consumptionof fish and chicken, both ofwhich are categorizedas white meat, while the consumption of otherred meats remains steady or even declining. Theper capita fish consumption in particular hasincreased almost threefold since 1960 whilechicken consumption has doubled during thesame period. One possible explanation for thisphenomena is that the relative prices of chickento beef, and fish to beef have declined over thesampled period. In terms of average budgetshare, fish captures 57.09% of the consumers'total meat expenditures, pork 32.9%, chicken13.45% while beef and mutton account for 4.36%and 2.09%, respectively. Coincidentally thisconsumption pattern is similar compared to thosein other countries of the Pacific rim regionwhere the principal product is unequivocallyfishery products (Capps et aI., 1994).

Since the introduction of the almost idealdemand system (AIDS) by Deaton andMuellbauer (1980), many applications of thismodel have been made to analyze consumerdemand for food. These have included studies

Commodity

FishChickenPorkBeefMutton

Average percapita

consumption (kg)

37.359.84

15.881.670.65

Average budgetshare of meatproducts (%)

57.0913.4532.9

4.362.09

112 PertanikaJ. Soc. Sci. & Hum. Vol. 7 No.2 1999

Demand for Fish and its Substitutes in Malaysia: Evidence of Habit Formation and Structural Change

YEAR

Fig. 1. Per capita consumption for fish and meat, 1960-1990

~ ..,. _----~~ .._-----

Fig. 2. Budget share for fish and meat, 1960-1990

demand is an important concern because suchchanges may necessitate corresponding changesin the fish and meat industry. The industry'sresponses however should be based on soundeconomic modeling and economic forecast.Models that do not explicitly account for thechanging nature of fish and meat demand maybe inappropriate, and thus may suggest industrychanges that are inconsistent with economicreality. Previous studies that dealt with this issueusing different statistical approaches includeMartin and Porter (1985), Chalfant and Alston(1988), Moschini and Meilke (1989), Chen andVeeman (1991), Burton and Young (1992) andEdwards (1992). However, the results reportedin these earlier studies were mixed.

No previous studies have so far been carriedout to investigate these issues in this part of theworld. This study is believed to be the firstattempt in this country to address such research

which undoubtedly can be very useful towardsunderstanding consumers' behavior in manydeveloping countries concerning demand forprotein-based meats. The objective of this paperis therefore two folds: The dynamic form ofAIDS that incorporates a habit effect in theconsumer expenditure function will first beestimated empirically. The purpose is to examinewhether this specification yields empirical resultsthat are consistent with the economic theory ofconsumer behavior. The other objective is todemonstrate evidence of structural change inthe demand for fish and meat products, to assesswhether or not consumers' preferences (tastes)towards fish and other meat products havechanged over the past three decades.

This paper is organized as follows. Sectiontwo discusses the specification of dynamic AIDSmodel. This is followed by the description of thedata used in the study and the estimation

PertanikaJ. Soc. Sci. & Hum. Vol. 7 No.2 1999 113


procedure. A brief note on the testing forstructural change is given in section four. Sectionfive presents the empirical results and discussionwhile the last section provides the concludingcomments.

MODEL SPECIFICATION

The Almost Ideal Demand System developed byDeaton and Muellbauer (1980) as expressed inexpenditure share form is:

S; =a i +2:y ij lnPj +~;(lnE-lnP) (1)J

wherei, j = 1, ..... , n refers to meat groupsSi = the expenditure share of the ith

meat groupPj = pricesE = total expenditure on all

commodities in the system, anda p ~ij' ~; = demand parameters to be

estimated.In equation (1), P is a price index defined by

In P = a 0 + 2: a kIn Pk+k j

S; = a i * +diqit-l + 4Yij In Pj +J

f3i(lnE-lnP), i,j=1...,5 (3)

where

In P = a o + 2:(a i *+diqit_l)ln Pj +J

(4)

Equation (3) is popularly known as the dynamicAIDS.

The adding-up condition in the modifiedsystem applies if:

La. * = 1 L Y = L A. = L dq. = 0 (5)I '1J PI 1 n-l

As in the case of the original AIDS model,the adding up restrictions hold only locally. Therestriction L diqit_l = 0 requires that at least oneof the cij is negative. While a positive signindicates habit persistence, a negative sign impliesinventory depletion effects. The condition ofhomogeneity and symmetry remain as L y. = 0and y.. = y.. , respectively. The habit persist~nce

lJ JIextension adds n parameters to the static AIDSmodel.

n

InP*= 6Wk 10gPk (6)

where Wk is the mean of the budget share of

the kd1 commodity.

THE DATA AND ESTIMATIONPROCEDURES

The dynamic AIDS model will be estimated inthis paper. The annual time series data from1960-1990 on prices for fish and four other meatgroups namely chicken, pork, beef and mutton,and income were used in this study. Consumptionfigures were obtained from the Department ofthe Veterinary Services (DVS), while data onprices were obtained from various FederalAgricultural Marketing Authority (FAMA)bulletins. Population, income and consumerprice index (CPI) were collected from variousMalaysia Plans and Economic Reports. All retailprices and income data were deflated by CPI(1980=100). The per capita consumption figureswere derived by dividing the total consumptionfor fish and each of the other meat groups withtotal consuming population.

The AIDS system summarized in equation(3) is non-linear in parameters. In order topermit this equation to be expressed in linearform equation (4) is often replaced by an indexdeveloped by Stone (l953). The index is:

(2)

(2b)

(2c)

2: Y;j = 0, (homogeneity)J

Y;j = Yji (symmetry)

X2: 2: Yk In Pk In Pk J J

To be consistent with the fundamental postulatesof demand theory, the following conditions musthold in terms of parameter restrictions:

Lai = 1, f Yij = 0, L~i = 0 (adding-up) (2a)

The standard AIDS specification in (1) is oftendescribed as static AIDS.

To incorporate consumption habit variablesinto the AIDS model, the "dynamic translating"procedure proposed by Pollak (1970) and Pollakand Wales (1981) is adopted here. Followingthis procedure, the original demand function isreplaced by a new system that contains translatingparameters, and it is assumed that only the (iparameters depend on the habit persistencevariables. Applying this procedure to the AIDSmodel and specifying the linear dynamictranslating parameter as a i = a i* + diqit_J' wheredi is the coefficient that measures the impact ofprevious consumption on the currentexpenditure share of meat type i, the habitpersistence version of the AIDS model becomes:

114 PertanikaJ. Soc. Sci. & Hum. Vol. 7 No_ 2 1999


In this paper, the dynamic AIDS model withhomogeneity and symmetry restriction imposedwas estimated using the Zellner's SeeminglyUnrelated Regression (SUR) method. l Thispermits cross equation restrictions to be imposedand with iterative solutions, estimates areMaximum Likelihood. The mutton equation wasdeleted due to the adding-up restrictions.

TEST OF STRUCTURAL CHANGE

There have been many developments whichcould have resulted in a systematic change in-the demand for fish and meat. For example, theattitude of some consumers towards red meatappears to have changed following the discoveryof the relationship between saturated fats andhealth. The health concerns may have produceda shift in fish and meat preferences. In addition,recent developments in the fast-food industry,for example, have introduced new forms ofproducts available to consumers and this mayhave directly affected the demand for fish andmeat.

In this study we examined the stability (orconstancy of parameters) of the meat demandequations using tests based on recursive residualsintroduced by Brown, Durbin and Evan (1975).The testing was undertaken using single equationmethod to avoid the spillover of anymisspecification from one equation to another.The approach has the advantage over the others(for example, Chow tests) since it does notrequire prior knowledge of the shifts but testsfor the presence of such occurrence over thesample period.2

Briefly, the cumulative sum (CUSUM) andcumulative sum of squares (CUSUMSQ) arebased on the one-step-ahead forecast errorsderived using recursive updated parameterestimates. Accordingly, a change in the structureovertime will result in the recursive residuals tohave non-zero mean. The CUSUM andCUSUMSQ of these residuals are used to test for

structural change. If the plot of the CUSUM orCUSUMSQ sample path moves outside criticalregion, and in this case 5% significant level, thenull hypothesis of stability overtime of theintercept and slope parameters is rejected(assuming the model is correctly specified).3 Inthis study we only report the CUSUMSQ since itis known to be more powerful than the CUSUMtest.

RESULTS AND DISCUSSION

Estimates of the structural parameters for thedynamic AIDS model are given in Table 2. Somevery interesting results can be summarized fromthis table. Nineteen of the thirty coefficientsestimated are significantly different from zeroand the minimum budget shares, (X.;, are allbetween zero and one for each meat type,indicating satisfactory fit. The coefficient 1\ areall negative and highly significant for each meattype except mutton. This implies that with theexception of mutton, fish and all other meatproducts are necessities.

The Durbin-Watson (D.W.) statistics in allexcept the pork equation are quite high,suggesting that serial correlation is not a problem.The low D.W statistic in the pork equation maysuggest some sort of misspecification problem.We also compare the results of dynamic AIDSmodel with the static AIDS model. Based uponlikelihood ratio test, it was found that thedynamic specification is preferred over the staticmodel. The R2 values for the fish, chicken, porkand beef equations are 0.87, 0.93, 0.86 and 0.86,respectively, indicating that the model performsreasonably well in terms of explanatory power.This suggests that the AIDS model with habitformation variable (i.e. the Dynamic AIDS) is anacceptable specification of the demand modelfor meats.

It is also interesting to note that the habitpersistence variable, d;, are positive andstatistically significant for fish, chicken and pork

The demand restrictions were tested using the log-likelihood ratio tests. In general the results obtained reject theserestrictions. At conventional significant levels, these restrictions were imposed in the final estimations of the dynamicAIDS model.

2 A major weakness of this approach is that it is difficult to distinguish between the effects of true structural changeand model misspecification. To overcome this problem we examined alternative specifications such as static, dynamicand dynamic with diagonal autoregressive AIDS models. The results of our preliminary investigations suggest thatthe dynamic AIDS specification is the preferred model based on conventional standards.The CUSUMSQ has a beta distribution with mean of (t-k)/(T-K). For more detailed discussion on the test, seeHarvey (1990).

PertanikaJ. Soc. Sci. & Hum. Vol. 7 No.2 1999 115


but negative and insignificant for beef andmutton. This indicates the pervasiveness of habitformation in fish, chicken and pork but theamount of beef and mutton purchased duringthe last period tends to lower current budgetallocation on these meat products. Thesignificant coefficients on habit persistencevariables suggest that this feature has someinfluence on consumer's budget share allocationfor fish, chicken and pork.

The above results concur with some of theprevious studies in that the inclusion ofconsumption habit effect improves theconsistency between theory and data. Whilechanges in relative prices, total expenditure andconsumers' habit explain some of the variationin fish and meat consumption, a considerableportion of the observed changes in fish andmeat expenditure patterns over the past thirtyyears is also consistent with a structural change

in consumer preferences.The results of the test for structural change

in fish and meat demand in Malaysia using theCUSUMSQ procedure are presented in Fig. 3.Using a 5 percent significant level, it is interestingto note that there is no structural breaks in thefish equation. This indicates that there is nostructural change in consumer preferences (ie.no change in taste) as far as fish is concerned.Being the cheapest source of protein, fish usedto be and continues to be consumed by the vastmajority of the population, either in fresh orprocessed forms. The favorable medical discoveryregarding seafood products has not resulted inthe change in tastes or preferences as reportedin many other studies.

However, structural breaks are detected, asexpected, in chicken, beef and porkconsumption. For beef in particular, structuralbreaks occur in the mid-seventies and the later

TABLE 2Maximum likelihood estimates of dynamic AIDS with for

homogeneity and symmetry restrictions

-0.032(-0.69)

-0.0158 0.0096(-3.29) * (2.57) *

-0.0060 -0.0100 0.0096(-2.37) * (-3.22) * (-2.57) *

0.0254 0.0034 0.0126 -0.0605(3.89) * (0.22) (2.31) *

-0.0165 -0.0323 -0.0062 0.0724(-18.79)* (-12.03) * (-9.90)*

0.0009 0.0011 -0.0001 -0.0035(-4.13)* (3.34)* (-0.14)

0.1234 0.2456 0.0408 0.0199(-21.32) * (-14.80)* (8.96) *

1.71 1.07 1.45

0.93 0.86 0.86

Fish

Yll 0.0047(0.25)

Y21 -0.0004(-0.09)

Y31 -0.0173(-1.71)

Y41 -0.0061(-2.75)*

Y51 0.0191(0.93)

13, -0.0174(-13.95)*

d] 0.0016(7.16)*

Constant 0.5703(18.09) *

D.W. 1.47

R2 0.87

Chicken Pork Beef Mutton

116

Log-likelihood value = 510.308

Note: t - values in parenthesis* - significant at 5%

1, 2, ....... , 5, where 1 = fish, 2 = chicken, 3 = pork, 4 = beef and5 = mutton.

PertanikaJ. Soc. Sci. & Hum. Vol. 7 No.2 1999


,", {;";~

J;. );.1" .,".ll.';x..tr ~.....

.'.No. ':1:':(:;r;;()". .~~r-.~~ .t....>::Q'

~~~ :;...·~':r

.~'.

~ff' .~.{·f...,

PertanikaJ. Soc. Sci. & Hum. Vo!. 7 No.2 1999 117


part of the eighties. For pork, it was evident inthe late seventies while the same phenomenonoccurred in the early eighties for chicken. Itappears that structural changes may have causedthe demand for some red meats to decline.Interestingly enough, these trends are consistentwith those findings in Australia (Martin andPorter, 1985), the USA (Edwards, 1992) and theGreat Britain (Burton and Young, 1992). Forexample, Burton and Young (1992) reportedthat there was a reduction in the budget sharesfor beef, pork and mutton during the lateseventies and eighties, but changes in tastes inrecent years tended to be in favor of fish andchicken. These changes in fish and meatexpenditures and consumption pattern areusually alleged to be associated with increasinghealth concerns regarding diets. Other possiblecauses of these changes include the changingnature of the poultry, beef and pork productsand the growth of the fast food outlets.

CONCLUSION

This study has addressed at least two importantissues in the demand for fish and meat productsin Malaysia. Aspects of consumers' tastes wereincorporated in the analysis of the Almost IdealDemand System (AIDS) for fish and other meatproducts through the inclusion of habitformation variable. The structural changes inthe consumer preferences were also tested.

The results of the Almost Ideal DemandSystem with habit formation (dynamic AIDS)performed creditably from the statistical pointof view. It was found that there was strongpersistence in the consumption for fish, chickenand pork but past purchases tend to lower budgetshare allocations for beef and mutton duringthe period from 1960 to 1990. These findingsare consistent with the trends observed in theper capita consumption and budget share wherefish, chicken and pork tend to dominate overbeef and mutton.

In the structural change tests, it was foundthat there was no structural change in consumerpreferences for fish but changes in taste wereobserved in chicken, beef and pork consumption.This occurrence is consistent with studies carriedout elsewhere, where the health consciouspopulations are now consuming lesser amountof red meat products but an increasing quantityof white meat such as fish.

Both results suggest several point of interestto Malaysian policy makers, planners and fishand meat traders. The consumers are persistentlydemanding more fish and their tastes andpreferences towards fish have not changed overthe past three decades. Fish remains as the mostpopular meat item and has acquired a veryimportant position in the diet. It is thereforeimperative that continuous and stable supply offish at reasonable price be made available to thepopulation. Any short fall in the supply of fishwill bound to have a negative impact on theconsumer's diet, nutrition and the health ofMalaysians at large.

Several strategies could be adopted by policymakers to ensure a sustainable supply of fish inthis country. Increase importation of fish fromneighboring countries would probably be theeasiest option. However this option willunnecessarily put a pressure on the country'salready strained balance of payment, and willnot be politically and economically viable. Theother alternative is to increase fish productionlocally. This proposition however will furthercreate problems on our already overexploitedfishery resources. Increasing fish price and theuse of more efficient fishing technology will notredress the problem at hand but unnecessarilyputting severe pressure on fish resources. Inorder to ensure sustainable fish supply, thisrenewable resources must be managed efficiently.Various management regimes are discussed inmany fishery economics literatures and some ofthem especially the fishing effort reductionmeasures are currently being instituted by fisheryadministrators. In some areas the results are veryencouraging where the fish production improvesand in the long run, these fishery resourcesconservation measures will help mitigate thegrowing fish demand in this country.

The movement towards an increasedimportance of white meats has some importantimplications on the beef and other red meatindustries as well. Our results suggest that theseindustries need to make quality adjustment inproduction since consumers are concerned withthe potential danger oflarge intake of cholesteroland other saturated fats.

Further research should be carried out inthis aspect using other approaches. Consumers'responses to price and income changes beforeand after the structural change has occurredshould be studied before any effective policies

118 PertanikaJ. Soc. Sci. & Hum. Vol. 7 0.21999


regarding fish and meat industries be drawn upand implemented to satisfy the presently moreeffluent and health-conscious consumers in this

country.

REFERENCES

AHMAD ZUBAIDI B. and ZAINAL ABIDIN MOHAMED. 1993.Demand for meat in Malaysia: an applicationof the almost demand system analysis. PertanikaJ. Soc. Sci & Humanities 1: 91-95.

BIANCIFORTI, L. and R. GREEN. 1983. The almostideal demand system incorporating habits. Ananalysis of expenditure on food and aggregatecommodity groups. Rev. Econ. and Statistics 5:511-515.

BROWN, RL., J. DURBIN, and J.M. EVANS. 1975.Technique for testing the constancy ofregression relationships over time. RoyalStatistical Soc. J. Series B 37: 149-163.

BURTON, M. and TREVOR YOUNG. 1992. The structureof changing tastes for meat and fish in GreatBritain. Euro. Rev. of Agric. Econ. 19: 165-180.

CAPPS, 0., R TSAJ, R KJRByand G.W. WILLiAM. 1994.A comparison of demands for meat productsin the Pacific rim region. J. of Agric. ResourceEcon. 19: 120-224.

CHALFANT, JAMES A. and J.M. ALSTON. 1988.Accounting for changes in tastes. J. of PolioEcon. 96: 391-410.

CHALFANT,J., R GRAY and K WHITF.. 1989. EvaluatingPrior Beliefs in a Demand System: The Case ofMeats Demand in Canada. Working Paper 483.Berkeley: Universiti of California, Departmentof Agriculture Economics and NaturalResources.

CHE ,P.Y. and M.M. VEEMAN. 1991. An almost idealdemand system analysis for meats with habitformation and structural change. Can. J. Agric.Econ. 39: 223-235.

DEATON, A. and J. MUELLBAUER. 1980. An almostideal demand system. Amer. Econ. Rev. 70: 312-326.

DEPARTMENT OF VETERINARY SERVICES (DVS). 1992.Livestock Statistics. Ministry of Agriculture,Malaysia.

EALES, J.S. and LJ. UNNEVEHR. 1988. Demand forbeef and chicken: separability and structuralchange. Amer. J. Agric. Econ. 70: 521-532.

EDWARDS, STEVEN F. 1992. Evidence of structuralchange in preference for seafood. MarineResource Econ. 7: 141-151.

FULPONI, L. 1989. The almost ideal demand system:an application to food and meat groups forFrance. J. Agric. Econ. 40: 82-92.

HARVEY, A. 1990. The Econometric Analysis of TimeSeries. 2 edn. Cambridge: MIT Press.

LEES, RS. 1988. The Impact of Dietary Fat on HumanHealth. MITSG 88-3, Massachusetts Institute ofTechnology Sea Grant, Massachusetts Instituteof Technology, Cambridge, MA.

MARTIN, WILL and D. PORTER. 1985. Testing forchanges in the structure of the demand formeat in australia. Austr. J. Agric. Econ. 29: 16-31.

MOLINA, J.A 1994. Food Demand and Spain. Anapplication of the almost ideal system. J. Agric.Econ. 45: 252-258.

MOSCHlNI, G. and KD. MEILKE. 1989. Modelling thepattern of structural change in the U.S. meatdemand. Amer. J. Agric. Econ. 71: 253-261.

NIK MUSTAPHA R A, AG. ROSLAN and D. POERWONO.1994. An almost ideal demand system analysisof fresh fish in Semarang, Indonesia. J. Intn.Food and Agribusiness Marketing 6: 19-28.

POLLAK, RA 1970. Habit formation and the dynamicdemand functions. J. Polito Econ. 78: 745-763.

POLLAK, RA and T. WALES. 1981. Demographicvariables in demand analysis. Econometrica 49:1533-1551.

STONE, J.RN. 1953. The Measurement of Consumers'Expenditure and Behaviour in the United Kingdom,1920-1938. Vol. 1, Cambridge: CambridgeUniversity Press.

(Received: 30 October 1996)

Pertanika.J. Soc. Sci. & Hum. Vol. 7 No. 21999 119

demand for fish and its substitutes malaysia: evidence of habit formation and...

Documents