Price Transmission in Thai Aquaculture Product Markets: An Analysis along Value Chain and across Species

We did not find any evidence of asymmetric price transmission in walking catfish except in long-run,vannamei shrimp and tilapia; however, it is evident in Thai seabass market; wholesaler

Price Transmission in Thai Aquaculture Product Markets: An Analysis along Value

Chain and across Species KEHAR SINGH

Former Research Associate, Aquaculture/Fisheries Centre,

University of Arkansas at Pine Bluff, AR 71601, USA

Presently Research Scientist (Agricultural/Resource Economics),

Canada Excellence Research Chair - Aquatic Epidemiology (CERC),

Atlantic Veterinary College, Charlottetown, PE C1A 4P3, Canada

Email: kesingh@upei.ca

MADAN M DEY *

Professor

Aquaculture/Fisheries Centre,

University of Arkansas at Pine Bluff,

1200 North University Dr., Mail Slot 4912, Pine Bluff, AR-71601, USA

Former Graduate Assistant, Aquaculture/Fisheries Centre,

University of Arkansas at Pine Bluff, AR 71601, USA

Presently Ph D Student, School of Economic Sciences,

Washington State University, Pullman WA 99164, USA

Email: umesh.bastola@wsu.edu

* Corresponding Author

Price Transmission in Thai Aquaculture Product Markets: An Analysis along Value

Chain and across Species Abstract We have examined the presence of price transmission asymmetry along the value chain, and the price transmission across four main aquaculture species in Thai fish market This is an attempt to contribute to the horizontal and vertical price transmission in the seafood markets literature including the price transmission asymmetry in the developing countries We did not find any

evidence of asymmetric price transmission in walking catfish (except in long-run),vannamei shrimp and tilapia; however, it is evident in Thai seabass market; wholesalers exercising some market power In most of the cases, none of the species considered affect significantly prices of other species at the same level ofvalue chain

Key words Vertical price transmission, price transmission asymmetry, price

transmission across species, price transmission models, Thai fish market

Running Title Price Transmission in Thai Fish Market

JEL Classification C22, D4, Q13

Horizontal and vertical prices linkages are important areas of research in the food markets The extent to which a price shock at one market/level of value chain affects a price in other market/value chain level provides an assessment of the functioning of markets The number of studies on horizontal price linkages in the seafood markets in the developed world has increased recently; however, it is hard to find studies in the developing countries There are limited studies on vertical price transmission including the asymmetric price transmission in seafood markets in the world Lack of the price transmission studies in seafood producing developing countries is primarily due to unavailability of the time series price data across species, markets and along the value chain

The present study is an attempt to contribute to the horizontal and vertical price transmission in the seafood markets literature including the price transmission

asymmetry in the developing countries We have examined the presence of price transmission asymmetry along the value chain, and the price transmission across four main aquaculture species in Thai fish market The fish species considered in the

analysis are vannamei shrimp (Penaeus vanamei), tilapia (Oreochromis niloticus),

walking catfish (Clarius sp.) and seabass (Lates calcarifer).

The fisheries sector including aquaculture plays a vital role in the food security and economy of Thailand In 2009, total fisheries production in the country was 3.78 million tons equivalent to 140,000 million baht (4,700 million US$) in value The

contribution of individual management sub-sectors to the total production included: marine capture (58%), inland capture (6%), coastal aquaculture (22%), and fresh water culture (14%) Marine capture fishery is mainly for exports while the coastal and fresh

water aquaculture is for domestic consumption Vannamei shrimp (Penaeus vanamei)

constitutes 60% of total coastal aquaculture culture production Seabass (Lates

calcarifer) is the main marine finfish cultured in Thailand; about 63% of the total of

marine fin fish farms cultured seabass during 2007 (Department of Fisheries, 2007)

Tilapia (Oreochromis niloticus) and walking catfish (Clarius sp.) account for 32% and

19% of total fresh water production, respectively

Recent studies on the spatial price linkages in seafood markets in the developed world include Nielsen (2004); Asche et al (2005); Nielsen (2005); Nielsen et al (2007);

(2009); Jimenez-Toribio, Guillotreau, and Mongruel (2010); Asche et al (2012) Nielsen (2004) found that the ‘Law of One Price’ is in force between the Norwegian and

Danish herring markets Asche et al (2005) examined market integration between wild

and farmed salmon on the Japanese market and found that the species were close substitutes on the market, and that the expansion of farmed salmon had resulted in price decreases for all salmon species Nielsen (2005) identified strong integration of

European cod markets and partially integrated saithe markets Nielsen et al (2007)

found that markets for farmed trout are related toothed fish markets in Germany, and that markets for these trout are more closely linked to markets for captured fish than to farmed salmon Using import price data from Japan, United States, and European

fresh and frozen tilapia fillets lie in different market segments, while fresh and frozen catfish fillets compete in the same market Norman-Lopez (2009) showed that fresh farmed tilapia fillets compete with wild whole red snapper, wild fresh fillets of seabass, and back flounder in the U.S market.Nielsen, Smit, and Guillen (2009) identified a loose

form of market integration between 13 fresh and seven frozen fish species in Europe

They found that the Law of One Price is in force on the fresh market within the segments

of flatfish and pelagic fish in Europe Jimenez-Toribio, Guillotreau, and Mongruel (2010) examined the degree of integration between the world market and the major European marketplaces of frozen and canned tuna through both vertical and spatial price

relationships They found that the European market for final goods segmented between the Northern countries consuming low-priced canned skipjack tuna imported from Asia (mainly Thailand) and the Southern countries (Italy, Spain) processing and importing yellowfin-based products sold at higher prices Asche et al (2012) used detailed data on shrimp prices by size class and import prices to conduct a co-

integration analysis of market integration in the U.S shrimp market They found a significant evidence of market integration, suggesting that the ‘Law

of One Price’ holds for this industry

The literature analyzing vertical price linkages has concentrated on evaluations

of the links between farm, wholesale and retail prices (Vavra and Goodwin 2005) The price relationships along the value chain provide insights into marketing efficiency, and consumer and farmer welfare (Aguiar and Santana 2002) It is to mention here that the relationships between two stages in the value chain are well developed by the theory of derived demand; however, the high data requirements to estimate such relationships often make it impossible to estimate Therefore, analysis of just prices at different levels

of the market chain is more commonly employed

Vertical price linkages in seafood markets are not studied much A few recentstudies to site are: Jimenez-Toribio, Garcia-del-Hoyo, and Garcia-Ordaz (2003); Guillen and Franquesa (2008); Jimenez-Toribio, Guillotreau, and Mongruel (2010) Jimenez-Toribio, Garcia-del-Hoyo, and Garcia-Ordaz (2003) used prices concerning ex-vesselmarkets, wholesale markets and foreign trade to study the impact of vertical integration

on price transmission in the fishing distribution channel of the Striped Venus (Chamelleagallina) Using weekly data, Guillen and Franquesa (2008) analyzed the pricetransmission elasticity of the main twelve seafood products in the Spanish market chain(Ex-vessel, Wholesale and Retail stages) Jimenez-Toribio, Guillotreau, and Mongruel(2010) tested vertical price relationships between the price of frozen tuna paid by thecanneries and the price of canned fish in both Italy and France The two species show

an opposite pattern in prices transmission along the value chain: price changes alongthe chain are far better transmitted for the “global” skipjack tuna than for the more

“European” yellowfin tuna

The asymmetric price transmission, i.e., increasing and decreasing prices at one

level of value chain transmit at different rates to another level, has received

considerable attention in agricultural economics Meyer and von Cramon-Taubadel (2004); Frey and Manera (2005) provide reviews of the literature on asymmetry price transmission However, the issue of asymmetric price transmission has been

overlooked in fish and fish product market studies (Jaffry 2005) A few studies to

mention are Jaffry (2005); Garcia (2006); Guillen and Franquesa (2008), Matsui et al (2011); and Nakajima et al (2011) Gonzales et al (2003) detected the asymmetric price

transmission in the distribution of wild cod and farmed salmon Jaffry (2005) found asymmetry in price transmission in the whole hake value chain in France Garcia (2006)studied the hake prices transmission along the Spanish market chain Guillen and Franquesa (2008) investigated the price transmission asymmetry in the main twelve seafood products in the Spanish market chain (ex-vessel, wholesale and retail levels)

Matsui et al (2011) analyzed Japanese blue fin tuna market and discussed that entities

having the market power shifted from upstream to downstream by tuna market structure

change Using a threshold autoregressive rolling window regression model, Nakajima et

al (2011) studied blue fin tuna market in Japan The findings of this study supported

those of Matsui et al (2011).

Common explanations of the existence of asymmetric farm-retail price

transmission in the food sector include: market power, search costs, consumer

response to changing prices, producer adjustment cost, and the behavior of markups over the business cycle (Jaffry 2005) The presence of asymmetric price transmission isoften considered as an evidence of market failure (Meyer and Cramon-Taubadel 2004) Peltzman (2000) found that asymmetric pricing is not just anecdotal, it’s closer to

universal, and asymmetric pricing to be as common in unconcentrated industries as it was in concentrated industries

Methodology

We have used following procedure to fulfill the objectives of the study:

i) Testing for a presence of the unit-root, Granger causality, and cointegration;

ii) Testing for the price transmission asymmetry along the value chain; and

iii) Specifying and estimating the price transmission models

Unit Root, Granger Causality and Cointegration Tests

Important issues in the price transmission analysis are: a) stationarity/non-stationarity ofthe time series, b) the Granger causation, and c) co-integration of non-stationary time series having same order of integration Addressing these issues is important to decide

on the regression model to adopt for the price transmission analysis stationarity and cointegration) and the R.H.S variables in the model (the Granger causation) If the series under study are stationary at levels, one can use traditional econometric tools like ‘ordinary least square’ estimation procedure to determine

(stationarity/non-relationships between those series The non-stationary series having unit root may be co-integrated if their order of integration is same; one can use the ‘error correction models’ to determine the relationships The ‘models in difference’ can be used for non-cointegrated series having unit root

There are two types of tests used to test whether a time series is stationary or not: the unit root tests and the stationarity tests The unit root tests test the null of a unit root against an alternative of stationarity, or mean reversion If the unit root null

hypothesis is rejected, then the series is said to be stationary The presence of a unit root in the time series representation of a variable has important implications for both the econometric method used and the economic interpretation of the model in which that variable appears The Augmented Dickey Fuller (ADF) test of Dickey and Fuller (1979), the generalized least squares ADF (DF-GLS), the Point Optimal tests (PT) of Elliott, Rothenburg, and Stock (ERS) (1996), and the Phillips-Perron test (Phillips and Perron 1988) are commonly used univariate unit root tests The stationarity tests test the null hypothesis of stationarity against a unit root alternative If the test fails to reject the null, the time series is said to be stationary The tests most widely used are those of

Kwiatkowski, Phillips, Schmidt, and Shin (KPSS) (1992); Saikkonen and Luukkonen (1993); Leybourne and McCabe (1994).

As is well known in the applied economics literature, even a test with DF-GLS’s favorable characteristics may still lack power to distinguish between the null hypothesis

of nonstationary behavior (I(1)) and the stationary alternative (I(0)) The Ng-Perron test (Ng and Perron 2001) modifies the Phillips and Perron (1988) test in a number of ways

in order to increase the test’s size and power This testing procedure ensures that rejections of the null hypothesis of the unit root are not due to a low probability of

non-rejecting a false null hypothesis, while rejections are not related to size distortions The Ng-Perron test constructs four test statistics that are based upon the GLS de-trended data These test statistics are modified forms of Phillips (1987) Zα statistics and Phillips and Perron (1988) Zt statistics, the Bhargava (1986) R1 statistic which is built on the work of Sargan and Bhargava (1983), and the ERS (1996) Point Optimal statistic Keeping in view the improved size and power of Ng-Perroni (2001) test over other univariate unit root tests, we have used the same to test the null hypothesis of presence

of unit root in the series

The next step is to determine whether the series having unit root are cointegrated

or not Cointegration between two time series integrated of same order can be tested with either by the Engle and Granger (1987) test or by the Johansen (1988) test; we have used the latter one The Johansen (1988) cointegration test is an unrestricted cointegration test; Gonzalo (1994) discussed advantages/disadvantages of this test

The issue of testing whether or not a variable precedes another variable, i.e., the

Granger causality (Granger 1969), is increasingly gaining attention in empirical research

(Hatemi-J 2012) We followed the Toda and Yamamoto (1995) procedure to test for the

Granger causality: i) determining maximum order of integration of two series, ii) setting

up a VAR model in levels, iii) selecting appropriate maximum lag length for variables in the VAR model, iv) testing for serial autocorrelation in the model, v) re-estimating the VAR model with appropriate lag length, and vi) testing the null hypothesis As discussed

earlier seabass farm, wholesale and retail price series, and tilapia retail price series price series are I(1), and all other series are (I(0) We have estimated appropriate

maximum lag order using: i) FPE (Final prediction error), ii) AIC (Akaike information criterion), iii) SIC (Schwarz information criterion), and iv) HQIC (Hannan-Quinn

information criterion) Then we have estimated the VAR model with lag order equal to maximum lag length selected using different information criteria plus maximum order of integration of two series Then we conducted (post-estimation test) to check for

autocorrelation in the model using the Lagrange-multiplier test (H0: no autocorrelation atlag order) If autocorrelation is found in the selected lag length, we increased the lag length until autocorrelation issue resolved and re-estimated the model In the end we, tested the null hypothesis using the Wald test, which has asymptotically chi-square

distributed with p degree of freedom under the null hypothesis For this test, we included

only lag length selected on the basis of different information criteria; extra lags

(maximum order of integration and increased lags to resolve autocorrelation) used are just to fix up the asymptotics

Testing for the price transmission asymmetry along the value chain

Meyer and von Cramon-Taubadel (2004) provide a survey of the asymmetric price transmission methods The results of the Johansen (1988) cointegration test, which will

be discussed in the succeeding section, shows that none of the series having unit-root are cointegrated Therefore, we followed the Houck (1977) and Ward (1982) approach This approach basically splits the change in explanatory variable into positive and negative changes

We have considered three levels along the value chain: farm, wholesale and retail Based on the pair-wise Granger causality test, we determined the direction of causation The Granger causality test, which will be discussed in the results and

discussion section, shows unidirectional in some cases and bidirectional causation in other cases; however, in some of the cases the price at one level of value chain (e.g wholesale) is caused by the prices at other levels of value chain (farm and wholesale) Depending on these results, we have extended the Houck (1977) and Ward (1982) model to consider two regressors The empirical model used in this paper for testing its asymmetry can be expressed as:

[ (ln )] [ (ln )] [ (ln )] [ (ln )] , ln

0 0

0 0

m m l j p

l l l j p

l l

P

where, cum and ln stand for cumulative and natural logarithmic value,

respectively Subscripts ‘i’, ‘j’ and ‘k’ stands for value chain level; ‘l’ and ‘m’ denote lag

number; t is the time; 0

*

lnln

lnP i = P t − P t= ; ∆(lnP t+)=lnP t −lnP t−1, if lnP t >lnP t−1 and 0

otherwise; and ∆(lnP t−)=lnP t −lnP t−1, if lnP t <lnP t−1 and 0 otherwise εt is the error

(1)

component If the price series on the LHS of the equation are stationary at levels withouttrend, we did not use the time as a variable on the RHS of the equation.

The null hypotheses of no difference tested against the alternate hypotheses of inequality are as follows:

m l

S :α+ =β+ 3 :α+ ≠β+ =

1

3 0

m l

S :α− =β− 4 :α− ≠β− =

1

4 0

l l L

o l l o

l l

H Null

1 1

1 1 1

1 1

2 1 1

l l L

q m m o

l l

H Null

1 1

3 1 1

l l L

q m m o

l l

H Null

1 1

3 1 1

(2.7)(2.8)

The equality of the coefficients of the positive change and negative change

H provides the evidence whether degree of positive changes in two

regressors on the changes in the dependent variable are significantly different from each other or not in short run and long run, respectively Similarly rejection of null

0

4 0

L

S and H

H provides evidence of significant difference in the influence of negative changes in two independent variables on the dependent variable

Specifying and Estimating the Price Transmission Models

We have identified the regressors based on the Granger causality test results If a price series on the left-hand side (LHS) and the right-hand side (RHS) price series (price series which are the Granger cause of the series on the RHS) do not have unit root, we have used a price transmission in levels (eq 3.1) However, if any of the price series on the LHS and RHS have a unit root and two or more price series are not cointegrated,

we have used a model in difference (eq 3.2)

, ) (ln )

(ln )

(ln ln

0

j

l jk l

jlk k

v

l iv l

ilv l

ik l

jlk k

iv l

ilv l

ik l

where subscript ‘i' and ‘j’ denote the species, ‘v’ and ‘k’ denote value chain level,

‘l’ denote lag order and ‘t’ denote time ‘P’ stands for price series, ‘ln’ is the natural

logarithmic value, α denote parameter and εt is the error component If the price series

on the LHS of the equation are stationary at levels without trend, we did not use the time as a variable on the RHS of the equation Since we have used logarithmic form,

(3.1)

(3.2)

therefore, the estimated parameters (α) are price short run price transmission

elasticities The long run elasticities along the value chain ( LR

l

il l

jl

LR S l

il l

STATA12, we tested the null hypotheses given in equations 2.1 to 2.8

We have used monthly price data on different fish species at different levels of supply chain, collected by different agencies The time period of data used ranges from January 2001 to October 2010 (Appendix 1) Data on farm-gate price and wholesale level price of seabass, catfish, and tilapia were obtained respectively from Office of Agriculture and Cooperative and Fish Market Organization under Ministry of Agriculture and Cooperatives, Thailand, while the retail prices were obtained from Ministry of

Commerce Prices on black tiger shrimp were obtained from central shrimp wholesale

market, Sakot Sarom, Thailand

Results and Discussion

Unit Root, Granger Causality and Cointegration

Table 1 presents the unit root test results for different time series under study The price series namely, shrimp farm, shrimp wholesale, shrimp retail, walking catfish wholesale,

(4)

tilapia farm and tilapia wholesale are stationary at levels, whereas the price series namely walking catfish farm and walking catfish retail are trend stationary (table 1) Seabass farm, wholesale and retail price series, and tilapia retail price series price series have unit root These series are stationary in first difference without a linear trend; we have taken the liberty not to present these results in table 1.

For the pair wise Granger causality test, tables 2A and 2B present the selected number of lags Tables 3A (along the value chain) and 3B (across the species) show results of the pairwise Granger causality test The test rejected following null

hypotheses (table 3A):

i). Shrimp wholesale price does not Granger cause shrimp farm price,

ii). shrimp retail price does not Granger cause shrimp wholesale price,

iii). shrimp wholesale price does not Granger cause shrimp retail price,

iv). shrimp retail price does not Granger Cause shrimp farm price,

v). walking catfish farm price does not Granger cause walking catfish

wholesale price,

vi). walking catfish retail price does not Granger cause walking catfish

wholesale price,

vii). seabass farm price does not Granger cause seabass wholesale price,

viii). tilapia wholesale price does not Granger cause tilapia farm price,

ix). tilapia retail price does not Granger cause tilapia farm price, and

x). tilapia farm price does not Granger cause tilapia retail price

Across the value chain, the Granger causality tests rejected following null

hypotheses up to 0.10 levels of significance (tables 3B):

i). Walking catfish retail price does not Granger cause seabass retail price,

ii). walking catfish retail price does not Granger cause tilapia retail price,

iii). walking catfish retail price does not Granger cause shrimp retail price,

iv). walking catfish farm price does not Granger cause seabass farm price,

v). walking catfish farm price does not Granger cause tilapia farm price,

vi). shrimp wholesale price does not Granger cause walking catfish wholesale

price,

vii). seabass wholesale price does not Granger Cause walking catfish

wholesale price, and

viii). tilapia wholesale price does not Granger cause walking catfish wholesale

a Granger cause of walking catfish farm and retail prices, and tilapia wholesale price

Our results suggest that prices in the Thai fish sector are not determined at one end and then passed down or up along the supply channel That is, pricing patterns in the Thai fish sector are not just cost or demand driven We found the direction of

causality from retail to farm prices in vannamei shrimp; however, the direction of

causality also found from wholesale to retail prices In case of walking catfish, the pricing patterns are both supply and demand driven The retail market shocks in case of

tilapia are directly transmitted to farmers, and vice-versa The wholesale prices of

seabass adjust to shocks in farm prices; however, shocks in retail market remains confined to retail market Tiffin and Dawson (2000) while studying the United Kingdom lamb market found that lamb prices were determined in the retail market, and then passed upward along the supply chain Goodwin and Holt (1999) and Goodwin and Harper (2000) found that retail market shocks were confined in retail markets for the most part, but farm markets adjusted to shocks in wholesale markets However, Ben-

Kaabia, et al (2002) found both supply and demand shocks were fully passed through the marketing channel; i.e., they found complete price transmission Saghaian (2007)

found that beef price causality in the U.S markets at different levels of the supply

channel are bi-directional, influencing and being influenced by each other at each stage

We have tested the cointegration along value chain for seabass; and at the retaillevel of value chain among seabass and tilapia Other price series are either stationary

at levels or trend stationary or there is only one price series having unit root atfarm/wholesale level of value chain Table 4 presents the results of the JohansenCointegration test The Trace and Eigen value statistics failed to reject the nullhypothesis of maximum rank equal to ‘0’ in all other cases, which shows absence ofcointegration between those price series

Price Transmission Analysis

Equation 3.1 and 3.2 are a general model used to study the price transmission relations

in Thai fish market These models have AR-terms; therefore, it is necessary to decide the number of lags of AR terms We have selected the lags using FPE, AIC, HQIC and SBIC criteria (table 5)

elasticity (sum of coefficients) of wholesale price with increasing retail prices (0.62) is significantly lower than decreasing retail prices (1.14) This means positive demand shocks in the walking catfish retail market are transmitted at a lower rate than negative shocks to the walking catfish wholesale market in the long run

Table 6B provides the estimates of the price transmission models for the walking catfish farm, wholesale and retail prices As stated earlier, we did not find any of the price series along the value chain and across the species at the same level of value chain as a Granger cause for farm and retail prices (tables 3A and 3B) Also these price series are trend stationary (table 1), and lag length selection criteria showed optimum lag length three for farm prices and lag length two for retail price (table 5) The

estimated models show very low but positive trends in walking catfish farm and retail prices (table 6B) Both farm and retail current prices of walking catfish are positively influenced by its previous month prices and negatively with two month lagged price (table 6B)

Walking catfish wholesale price series is influenced by its farm and retail prices, and also vannamei shrimp, seabass and tilapia wholesale prices Seabass wholesale price has unit root, and walking catfish wholesale price is stationary at levels without

trend (table 1) Therefore, we have used model in difference without trend The

estimates of the model (table 6B) show that walking catfish farm price do not have any significant influence on its wholesale price, whereas its retail price affected its wholesaleprice significantly Walking catfish current month retail price does not affect its current wholesale price, whereas one and two month lagged retail price has positive (short run elasticity = 1.25) and negative (short run elasticity = -0.97), respectively, on walking catfish wholesale prices Two month lagged vannamei shrimp wholesale price affects walking catfish current month wholesale price significantly (short run elasticity = 0.40) Current seabass wholesale price has negative and previous month has positive

influence on walking catfish wholesale price Only current month tilapia wholesale pricesinfluence walking catfish wholesale prices significantly In nutshell a positive and a negative changes in current month tilapia and seabass whole prices, respectively, lead

to a positive change in current month walking catfish wholesale price The reverse is true for effects of previous month wholesale prices of tilapia and seabass on current month wholesale price of walking catfish

Vannamei Shrimp

We have presented the estimated price transmission asymmetry models (eq 1) for vannamei shrimp farm, wholesale and retail prices in table 7A, and the asymmetric pricetransmission hypotheses tests results in table 7B.hypothesis None of the estimated coefficients in vannamei shrimp farm price model are statistically significant up to 0.10 levels of significance However, in case of wholesale/retail price models, current price coefficients of positive as well as negative cumulative changes in retail/wholesale pricesare significant, and magnitudes of coefficients are almost equal This means absence of

asymmetric price transmission in Thai vannamei shrimp markets at farm, wholesale andretail levels of value chain This is confirmed by the hypotheses test results given in table 7B.

Vannamei shrimp farm, wholesale and retail prices are stationary in levels

without trend Walking catfish retail price, which is trend stationary at levels, is the Granger cause of vannamei shrimp retail price At the same level of value chain, price ofnone of the species understudy is the Granger cause of vannamei shrimp farm and wholesale prices The test results showed the absence of asymmetric price

transmission in Thai vannamei shrimp market along the value chain Therefore, we haveused model given in equation 3.1 (table 7C) to work out price transmission

wholesale price is very low (0.30) Current and one month lagged vannamei shrimp wholesale/retail prices affect current vannamei shrimp retail/wholesale prices

significantly The log run price transmission elasticity of vannamei shrimp

wholesale/retail price with respect to its retail/wholesale price is 0.84/0.77

Seabass

All seabass price series have the unit roots (table 1); however, they are not cointegrated(table 4) Seabass farm price is the Granger cause of its wholesale price; the hypothesis

of the Granger causality is rejected in other price pairs of seabass along value chain Keeping in view these results, we have estimated equation 1 for seabass wholesale price (table 8A) The coefficient of current cumulative positive change in seabass retail price is significant; however, the coefficient of current cumulative negative change in retail seabass price is non-significant up to 0.10 levels of significance The coefficients

of lagged (one and three month lags) cumulative negative change in retail seabass price are significant too This means that if the seabass wholesalers pay higher prices (say 1%) to the farmers, they immediately receive higher prices (0.58%) from the

seabass retailers However, if the wholesalers pay lower prices to the farmers, they do not pass the decrease to the retailers immediately They pass around 20% of decreasedprice to the retailers in next month and about 26% in third month Less than 50% of decrease and 70% of increase in wholesalers’ purchase price is passed to the retailers

in the long run This indicates, and is confirmed by asymmetry hypotheses tests results (table 8A), presence of short run as well as long run asymmetric price transmission between seabass price in Thailand It is to mention here that seabass production is mainly based on cage culture, which requires very high investments Seabass farmers are well organized too Retailers have very low, if any, control over prices

We have estimated models in difference given in equation 3.2 for seabass farm, wholesale and retail prices (table 8B) One month lagged seabass retail and farm pricesinfluence respective prices One month lagged farm price of walking catfish affects seabass farm price Seabass current farm price is only factor which affects seabass wholesale price significantly Three month lagged walking catfish retail price has

significant influence on seabass retail price

Tilapia wholesale and retail prices are the Granger cause of tilapia farm price, and wholesale and farm prices of its retail price Tilapia retail price have unit root, whereas wholesale and farm price series are stationary The results of the asymmetric price transmission model shows that six month lagged cumulative positive change in

wholesale price and five month lagged cumulative change in retail price affect tilapia farm price significantly (table 9A); however, there is no evidence of asymmetric price transmission in tilapia markets along the value chain in Thailand (table 9B)

The estimates of the price transmission model (table 9C) for tilapia retail price show that walking catfish retail price influence tilapia retail price significantly (price transmission elasticity in current month = 0.32, and long run price transmission elasticity

= -0.06) Recent historical prices affect tilapia prices at all levels of value chain

Conclusions and Policy Implications

We have examined the presence of price transmission asymmetry along the value chain, and the price transmission across species in Thai fish market This is an attempt

to contribute to the horizontal and vertical price transmission in the seafood markets literature including the price transmission asymmetry in the developing countries

We found unidirectional Granger causation in some cases and bidirectional Granger causation in other cases; however, in some of the cases the price at one level

of value chain is Granger caused by the prices at other levels of value chain Therefore,

we have extended the Houck (1977) and Ward (1982) asymmetric price transmission model to consider two regressors, which allow the researchers to test the hypotheses

“whether degree of positive/negative changes in two regressors on the changes in the

dependent variable are significantly different from each other or not in short run and long run” We estimated the price transmission relationships using regressors along the value chain and across the species at the same level of value chain.

There is no evidence of short run asymmetric price transmission from either retail

or farm level to wholesale level; however, there is weak evidence of long run

asymmetric price transmission from retail to wholesale price We did not find any

evidence of asymmetric price transmission in Thai fish market for vannamei shrimp and tilapia in short- and long run Short run and long run price transmission asymmetry is evident in Thai seabass market; wholesalers exercising some market power

In most of the cases, none of the species considered affect significantly prices of

other species at the same level of value chain The exceptions to this are: i) walking catfish price affects tilapia price at retail level in short as well as long run, ii) three month lagged walking catfish retail price affects seabass current retail price, iii) one month lagged walking catfish farm price influences seabass farm price, and iv) vannamei

shrimp two month lagged price, current tilapia price and current and one month lagged seabass price affect significantly walking catfish prices at wholesale level In all these cases, the price transmission elasticities are positive except for long run elasticity in case i (where it is negative but close to zero) and current month seabass wholesale price in case iv where it is -1.11 These results indicate lack of competition among different species in Thai seafood market However, walking catfish faces some

competition from tilapia in short run at wholesale level

Price transmission relationships along the value chain shows that walking catfish retail prices (one month and two month lagged) influence significantly its wholesale

price in short run Vannamei shrimp retail and wholesale prices affects each other in short run as well as long run Vannamei shrimp’s current wholesale price also influencesits current farm price Seabass current farm price affects its wholesale price None of the prices along value chain in tilapia affect each other significantly.

The results of the study have important policy implications Various studies (Dey

et al 2008a; Dey et al 2008b) indicate that, given elastic income elasticity of demand

for fish, there will be tremendous increase in demand for various types of fish in

Thailand over time due to population growth and increases in per capita income Dey

et al (2008a) also indicates that fish exports from Thailand are expected to rise

particularly of tilapia, cultured shrimp and high-value marine fish like seabass It is projected that consumer prices of the various species studied are expected to rise fasterthan the posited inflation rate of 3.5%during 2005-2020, except for tilapia (with a yearly

rise of 2.6%) (Dey et al 2008a) The findings of no asymmetric price transmission of

retail prices of aquaculture products , indicating that increases in the retail price of the aquaculture products are likely to pass fully to the primary markets , are beneficial to aquaculture farmers in the country In recent years, almost all increases in fish

production have come from aquaculture sector However, increasing fish supply from aquaculture will exert a downward pressure on prices of aquaculture products But if market prices fall due to the expansion of products, retailers might also be able to easilypass through falling prices to farmers, and thereby farmers’ revenue might fall Thus, there is a need to monitor the likely effect of aquaculture expansion on farm prices Theaquaculture products should have a favorable market outlook to ensure economic viability of the concerned farm enterprises

Aquaculture harvests are seasonal in nature Like in other developing countries, many fish farmers in Thailand are often forced to sell their produces during the

harvesting season If retail and/or wholesale prices drop due to some market

phenomenon, farmers will have to sell their produces at that low price This signifies the importance of better storage facilities and transport infrastructure in rural markets Policies that encourage small-scale farmers to form collective arrangement for

marketing will be helpful