The fisheries is one of the key sec

The fisheries is one of the key sectors in India with around 6.7 million people dependent on it for livelihood (GoI, 2001). The sector is undergoing fast transformation, from subsistence level to a multimillion industry. The fish production (both marine and inland) of the country has increased from 0.75 million tonnes in 1950-51 to 6.4 million tonnes in 2003-04 (Narayankumar and Sathiadas, 2006). The marine landings alone were valued at Rs 13019 crores at the landing centre level in 2004, while the value at the final consumer point was estimated at Rs 22,653 crores (Sathiadas, 2005). For the past many years, marine exports have been a substantial
source of foreign-exchange earning to India’s exchequer and hence are accorded utmost priority.
However, the domestic fisheries marketing system in the country has long been neglected due to various reasons. It deserves its due share primarily because around 85 per cent of the total fish production is consumed within the country. It is, therefore, important to develop a strong network of efficient marketing system within the country so that a substantial chunk of country’s fish production is efficiently managed and delivered to the consuming
masses, while not negating the due share of the fishermen. Essentially, an efficient marketing system is one where there is a perfect market integration and full price transmission, with instantaneous price adjusment to changes from within or outside the
system. Such a system would enable the producers, middlemen and consumers in the marketing chain
to derive maximum gains. It would also help in elimination of unprofitable arbitrage and isolation of spatially differentiated markets and would ensure that efficient allocation of resources across space and time is achieved (Nkang et al., 2007). In the fish marketing system, price movements in different markets depend to a large extent on the cross market movement of available catch, which in turn, is governed by the demand and supply factors. The extent of price transmission from one market to the other and its direction are the important aspects to
be looked into, as these would provide valuable information on the degree of integration, and in turn, the efficiency of these markets. In the present paper, the degree of spatial market integration between the major coastal markets in India has been studied using monthly retail price data on important fish species. The study has highlighted the supply side constraints,
which are essentially the major factors responsible for poor integration between the markets.

Data and Methodology
For the study, monthly price data for a ten-year period from January 1998 to December 2007 were
collected on important marine fish species, viz.mackerel, sardine, pomfret and shrimp from the major coastal states of India. The states covered were Andhra Pradesh, Gujarat, Karnataka, Kerala, Maharashtra, Orissa, Tamil Nadu and West Bengal.
The retail markets around major landing centres in each of these states were selected for this purpose. The data were collected through regular and systematic primary surveys conducted by the Central Marine Fisheries Research Institute (CMFRI),
Cochin.
Analytical Framework
Two price series belonging to spatially separated markets are said to be integrated if there exists a long-term equilibrium relationship between them.
The degree of transmission of price signals between these two markets can be obtained by fitting a classical regression model given by Equation (1):
Yt = β0 + β1 Xt + et …(1)
where,
Yt = Price at the dependent market,
Xt = Price at the independent market,
β0 = Constant,
β1 = Long-run elasticity of price transmission, and
et = Error-term.
However, assumptions of the classical regression model necessitate that both Yt and Xt variables should be stationary and the errors should have a zero mean and finite variance. A stationary series is one whose parameters (mean, variance and autocorrelations) are independent of time. Regression between two nonstationary variables may result in spurious relationship with high R2 and t-statistics that appear to be significant, but with the results of having no
economic meaning. Under such circumstances, the series have to be first checked for stationarity. If a time series requires first order differencing to be stationary, then it is said to be I (1), which means integrated of the order one. I (2) series requires
differencing twice to become stationary and so on. If it is verified that both the series are stationary, then the classical regression model would hold good and the β coefficient would
represent the coefficient of price transmission. However, if the two series prove to be non-stationary but integrated of the same order, the validity of regression can be checked by testing the residuals of the regression for stationarity. Engle and Granger
(1987) had demonstrated that, if the residuals from such a regression turn out to be stationary, then the series are co-integrated and there existed a long-run relationship between the two series. Engle-Granger theorem states that if a set of variables are cointegrated
of order (1, 1), then there exits a valid error-correction representation of the data. Converse
of this theorem also holds good, that is, if an errorcorrection model (ECM) provides an adequate
representation of the variables, then they must be co-integrated. However, if the series are integrated of different orders, the regression equations using such variables would be meaningless and it can be concluded that there cannot exist any long-term relationship between the two.