ASSESING SEASONAL PRICE BEHAVIOUR OF SELECTED DRIED FISH VARIETIES IN SRI LANKA

Dried fish plays an important role in the national economy in multiple ways including minimizing post-harvest losses of fish providing a source of animal protein, enlarging livelihoods, utilizing idling labor into the production process and as a dish, especially for rural poor. Dried fish prices have increased drastically from 2012 to 2019 compared to fish and chicken. Being the key determinant of demand, price is crucial on consumption, dried fish demand and industry performance: increasing prices, decrease the demand, which adversely effects on dried fish industry from 2012 pushing dried fish consumption to third place. Therefore, this study aims to assess the price behavior of selected dried fish varieties and appropriate price forecasting models that could be feed into policy formulation for reasonable prices and price variation during the year. A quantitative approach was adopted gathering national average price data from secondary sources such as Hector Kobbekaduwa Agrarian and Research Institute, Department of Census and Statistics and other relevant institute. Three dried fish varieties which are highly consumed by the populace were selected representing large pelagic-skipjack tuna (Katsuwonus pelamis) and double-spotted queenfish (Scomberoides lysan) and small pelagicgold-striped sardinella (Sardinella gibbosa). Average monthly retail price data from January 2007 to December 2019 were considered to analysis. Seasonal price variation and price indices were calculated for the three varieties in terms of real market prices. Analysis revealed highest dried fish prices in June and July months, corresponding with south-west monsoon. In contrast, lower prices have been reported during the north-east monsoon for large pelagic, months of January and February. Gold-striped sardinella indicated the highest price index in middle month of the both monsoon seasons. Highest seasonal price indexes for large pelagic varieties are indicated during May-September following the south -west monsoon. The Real Market Price (RMP) of all three dried fish varieties demonstrated a fluctuating pattern with a slight increment throughout the year. The same fluctuating patterns are observed in relation to moving average price, seasonality impact removed price and seasonality around moving average price of each three varieties separately. However, prices of three varieties are strongly correlated (r=0.941<, p=0.00) with each other. The cubic price forecasting model is the best fit model of price forecasting for all the analyzed varieties. Besides, the Quadratic model can be used to predict the price of analyzed large pelagic varieties. The ARIMA analysis revealed that ARIMA (2,1,12) is the best fit model for price forecasting of skipjack tuna while ARIMA (0,1,0) for double spotted queen and gold stripped sardinella forecasting. Having a distinct seasonality impact on dried fish prices, the policies need to be focused on price stabilizing mechanisms to assure a certain and stable performances in the dried fish industry in Sri Lanka ensuring affordability to the majority in Sri Lanka.


INTRODUCTION
The dried fish industry in Sri Lanka operates as a cottage level industry (Koralagama and Bandara 2018;Piyasiri et al. 2018) yet plays an important role in the economy with multiple contributions including, a source of animal protein intake, enlarge employment opportunities, utilize idling labors-female in particular and reduce post-harvest losses 22 sensitivity (Koralagama et al. 2021). Thus, low income people tend to buy low price dried fish products from village fair and other rural dried fish markets whereas high income earners consumes expensive dried fish varieties with innovative and attractive packages under a specific brand name from hypermarkets (Faruque et al. 2012).

Production and consumption
Double-spotted queenfish (Scomberoides lysan), skipjack tuna (Katsuwonus pelamis), smooth-belly sardinella (Amblygaster clupeoides), seer fish (Scomberomorus commersoni), and giant catfish (Arlus thalassinus) are the main marine dried fish varieties produced in Sri Lanka (MFARD 2019;DCS 2016). Local production satisfies only 65% of the local demand thus, dried fish varieties such as sprats, skipjack tuna, smoothbelly sardinella and shark are being imported, which accounts for about 35% of the local demand (MFARD 2019). Consumption pattern of several dried fish varieties are different based on the type-small and large pelagic varieties (DCS 2016).
Accordingly, Skipjack tuna (113.15g), double -spotted queenfish (74.82g) and shark (84.19g) are the highest consuming largepelagic varieties whereas sprats (487g), smooth-belly sardinella (76.41g), gold-striped sardinella (43.69g) and trenched sardinella (22.85g) are the highly consuming small pelagic varieties (DCS 2016). Despite the lower production, sprat is highly consumed by Sri Lankan hence the local demand is fulfilled by the imported sprats that accounts for 72. 2%;approximately 23,232 Mt in 2006-2018(MFARD 2019. The monthly expenditure on the highly consumed dried fish varieties are given in Table 1. Sprat importation has increased by 36% in 2018 compared to 1995.Other dried fish varieties have decreased by 70.3% (MFARD 2019). Skipjack tuna and double spotted queen fish are the highest and second highest expenditure dried fish varieties representing large pelagic group while smooth-belly sardinella is the third highest expenditure and consumed variety representing small pelagic group (DCS 2016  Prices of dried fish products are mainly depending on the availability of the products in the market places (Carlucci et al. 2015). Besides, dried fish prices are important to monitor industry development and market functions (Adenegan and Bolarinwa 2010). Prices have been identified as the sign of the relative scarcity or abundance of a given product; prices also serve as incentive to direct the allocation of economic resources and to a large extent they determine the structure and rate of economic growth (Aswathy and Abdussamad 2013). Information on prices is important to producers, traders and consumers to understand the market trend and better planning for policy decisions (Murray and Little 2000;Ravallion 1986). Despite the market prices follow the demand and supply theories, many scholars have researched on few other variables that influence on the pricing mechanism including the variety, market type, importation, seasonal variation of production, transport mode and quality are another paramount (Hirimutugoda et al. 2014;Hosseini et al. 2012 ;Shamsuddoha 2007).
Further, dried fish prices also vary based on the geographical area, marketing systems and the number of intermediaries and product availability (Hasan et al. 2016;Weerahewa and Kodithuwakku 2013). However, due to one or multiple factors, the dried fish prices have being increasing from 2012 pushing down the consumption to the third place from the animal protein sources. Increasing prices and declining demand affects the industry performance (Koralagama et al. 2021), affecting both processors (Koralagama and Bandara 2018) and rural consumers (DCS 2016), where both ends are marginalized. In this regard, price behavior analysis and forecasting is important on dried fish. Although the fresh fish market is fully backed by the local production, the dried fish demand is compensated by the imported dried fish varieties. However, dried fish prices are increasing (HARTI 2019; MFARD 2019). Thus, this study explores the price behavior of the mostly consumed dried fish varieties by analyzing the impact of seasonality on the prices. Further, the study attempts to calculate a seasonality index to detect the percentage share of variation/fluctuation for better forecasting and prediction that would enable for price controlling mechanism in the future for efficient policy decisions. The next section of this paper briefs the methodology carried out in data collection and analysis, which is followed by the results and discussion zooming into seasonality variations and price indices. Finally the paper concludes by highlighting the price behavior of dried fish and pragmatic solutions for the variations. tuna, double-spotted queenfish and goldstriped sardinella were considered as the highly consumed dried fish varieties in Sri Lanka (DCS 2016) representing locally produce large pelagic and small pelagic varieties (see table 01). Weekly retail price data, particularly market price data for the period of 2007-2019 available at the Hector Kobbekaduwa Agrarian Research and Training Institute of Sri Lanka, which is the main research institute with a large data bank, were used as the main data source for the analysis.

Analytical procedure
Real Market Prices (RMP) were calculated for the generalization and to extract the absolute prices avoiding inflation by using Nominal Market Price (NMP). Colombo consumer price index, which uses 2006/2007 years as the base year (CCPI) (2006/2007 = 100) was considered. The CCPI gives the local government, businesses, citizens and idea about price changes in the Sri Lankan economy. Besides, the Colombo consumer price index is the widely used tool in adjusting for inflation and, by proxy, for the effectiveness of the Sri Lankan government's economic policy (Alibuhtto and Peiris 2012; Sandika 2011). The calculation procedure of RMP is given in the following equation (i).
Price is a value generated through the effect of four components namely trend, cyclical, seasonal, and random/irregular parts which are according to classical price multiplicative model as (Price = Trend x Cyclical x Seasonal x Random/Irregular components). In dried fish price variation and production, seasonality is one of the vital characteristics (Koralagama and Bandara 2018; Hirimutugoda et al. 2014). Therefore, the analysis focused on the seasonal components thereby removing other components (viz; trend, cyclical and random) from dried fish price series. This method is important as it enables to predict the future stock price movements based on recently available data. Trend regression was calculated as per the equation given below (iii). There, (a) constant and trend coefficient (b) were estimated to determine each trend Where; Ti = trend value during the period I; a = the constant-coefficient as estimated by the regression analysis; b = the trend coefficient estimated by the regression; ti = the value of the variable (dried fish price) during the period i.
Seasonal Index (SI) is the standard indicator of price comparison, which can be used to compare the movement of prices. It shows the average price variation of products SI states on percentage price increment or decrement of the commodity within the observed time period. It helps to forecast percentage changes of profit for traders during each month of the year. See equation (iv).

Where: Pi= Price component; CMA= central moving average
Calculation of central moving average (CMA) is important to develop the seasonal price index. CMA is a simple, technical analytical tool. It is usually calculated to determine the trend direction of a product or to identify its support and resistance levels. CMA represents the trend and cyclical components of the original series, and it removes seasonality and the random factor. The formula is given below (v).
Thus, CMA will now be, Where: CMA = Central Moving Average; P = Nominal price; n = Number of periods.
The SI is already deflated as it is calculated by dividing the nominal price series (the original price) by another price series (the SI). Seasonality Removed Price (SRP) from another price series can be calculated as in the equation (vi), it is used to remove the seasonal impact of price out of four main components in price series.
Where, SRP=Seasonality removed price; Pi = Real price; SI= Seasonal index Seasonality around moving average price (SAMAP) from another price series was calculated as follow (see equation vii), Where, SAMAP= Seasonality around moving average price, Pi = Nominal/Real price, SI= Seasonal index. It is important to remove the effect of irregular movement from the SI values. It calculates the average of SI for each month over the different years, then adjusting SI figure series by the adjusting factors.
Karl Pearson's product movement correlation coefficient has been used to assess the correlation between prices of different dried fish varieties and the same was applied to calculate the correlation by using SPSS version 22.
Different time series models such as linear (Y = a + bX), Quadratic (Y = a + bx+ cx2) Cubic (Y = a + bx+ bx2 + cx3) and power were applied to the determination of the best fit model. R2was used to easy and accurate determination of suitable model.

RESULTS AND DISCUSSION Behavior of average RMP and seasonal price index
This section elaborates the results of the price analysis in assessing monthly real market price variation of skipjack tuna (balaya), double spotted queenfish (katta) and goldstriped sardinella (salaya). The study uses RMP; hence, NMP was converted to RMP (see equation i) (annex i for NMP). Prices of skipjack tuna and gold-striped sardinella reveal clear price increment in the month of July however the prices are gradually increasing from March to July and then decrease thereafter. In contrast, prices of double spotted queenfish show a clear increase of price in May where the prices are gradually increasing from January to May then slightly decrease till December. Peak prices are reported during south-west monsoon season from May to September period while the lower prices in north-east monsoon. Furthermore, gold stripped sardinella price indicate comparatively highest prices in January and February following the north-east monsoon. Price increment of gold-stripped sardinella initiate with the beginning of the north-east monsoon. Large pelagic fish varieties that used for dried    It has recorded that the fish production during the south-west monsoon (May-September) and north-east monsoon (December-February) comparatively lesser than other months of the year due to adverse weather condition and low community participation for fishing (Dayalatha 2020;Murray and Little 2000: 30;FAO 1984: 40). The small-pelagic fish that used to dry are mainly coming from west and north coast and large-pelagic from south and west coast in Sri Lanka (Koralagama et al. 2021;Amarasinghe 2020;MFARD 2019). Due to the lesser fish production during south -west and north-east monsoon in northern, eastern, western and southern coast (Murray and Little 2000: 30;FAO 1984: 40), it is perceived that the large pelagic raw fish production comparatively low during southwest monsoon while small pelagic comparatively low during the both monsoon season in Sri Lanka (Koralagama 2009) so as the dried fish. Therefore, production quantity of small pelagic dried fish is comparatively decreased in both monsoon periods and large pelagic in south-west monsoon (Murray and Little 2000). Low supply of dried fish varieties such during two monsoon periods causes to highest percentage of seasonal price increment for large pelagic in south-west and small pelagic in both monsoon periods (Murray and Little 2000;FAO 1984). Similar results have been obtained by a few researches conducted on Inidan and Bangladesh fisheries (Vishwanatha et al. 2018;Omar et al. 2018), where has the same monsoon pattern as in Sri Lanka.

Behavior of monthly RMP and different price components
The behavior of monthly RMP variation of skipjack tuna, gold striped sardinella and double spotted queen fish are given in figure  RMP variation is not well maintained during the observed period of different three varieties comparatively nominal market price variation (see annex ii). It can be identified; inflation is the main factor that affects the price increment of observed dried fish varieties. Therefore, processor income levels were not increased by a considerable amount comparatively market price increment of the three dried fish varieties.
Furthermore, the nominal fish prices have expanded over time, real price is consistent and occasional in Sri Lanka market (Perera et al. 2016). There is the fluctuating pattern in nominal fish price over the years but not in real price. Thusly, the evident expanding The comparison of real market price variation, moving average price variation, seasonality impact removed price variation and seasonality around moving average price variation of different three dried fish varieties are shown in following figure 4.
Various fluctuating pattern of different price components of each analyzed three dried fish varieties are comparatively illustrated in above figure. Besides, each price components show the same fluctuating pattern within the same dried fish variety at most of the month. Inflation is the rate at which the estimation of a price is falling and consequently, the general level of price for goods and services is rising. 29 As inflation increments to the highest levels, price dissemination isn't even symmetric. At the point when the yearly inflation rate arrives at 130 percent, there is an equivalent chance of discovering real prices above or below the market average. Accordingly, inflation is the one of critical factor which affected to variation of each price components of analyzed varieties.

Relationships and forecasting models
In statistical analysis, a correlation matrix is used to determine the degree to which a pair of dependent and independent variables are linearly related each other. A strong positive correlation is noted (table 2) among all three varieties on the highest r-value (r~1) of each variety. Price of one variety has a direct impact on the prices of other varieties as well. Price increment of one variety would increase the other varieties and vice versa. Therefore, it demonstrates that most of the factors are the same which affect to increase the price of all three varieties.
The results of Perera et al. (2016) reveals that the wholesale price and retail price of fish in the Sri Lanka market moderately correlated. The highest correlation can be seen between Colombo and Kalutara retail market fish prices while the lowest correlation between Colombo and Kandy wholesale market fish prices.  spotted queenfish are also being imported to Sri Lanka (MFARD 2019; HARTI 2014). The price of imported dried fish mainly affect to price variation of local produce varieties (lbid). Accordingly, timely importation quantity changes are not considerably affected to price variation of gold stripped sardinella in local market. Therefore, timely price increment of gold stripped sardinella is higher than other observed varieties due to the lower competition from imported varieties. The resulted ARIMA model in this analysis with (0,1,0) is identical with the findings of Mustapa and collegues (2019), which was conducted for fresh fish and vegetables in Malaysia. Further, ARIMA model with (2,1,1) is identical to fish wholesale price analysis of India. Accordingly, ARIMA

CONCLUSION
This paper attempted to analyze the price behavior of three dried fish varieties, namely Katsuwonus pelamis (skipjack tuna), Scomberoides lysan (double-spotted queen fish) -large pelagic and Sardinella gibbosa (gold-stripped sardinella) -small-pelagic, which are highly consumed by Sri Lankans. The price behavior was analyzed by developing the price indices. Large pelagic dried fish varieties showed relatively higher prices in the months of June and July months and lowest in November -February, which follows the fresh fish abundance during northeast monsoon season, hence lower fresh fish prices and vice versa. Small pelagic varietiesgold-striped sardinella reported highest price in May/June while lowest in November/ December still following the monsoon impact on fresh fish availability and prices.
Seasonal price index (SI) of skipjack tuna and double-spotted queenfish showed a high SI during the months of May-September whereas the SI is decreased from November to February. In contrast, gold stripped sardinella has higher SI in months of January and July while lowest in December. The indices Southwest monsoon (May-September) was identified as the highest percentage of price increase period for large pelagic group. Both north-east monsoon (December-February) and south-west monsoon season were identified as the highest percentage of price increase season for small pelagic varieties. In general, all three varieties showed higher prices during May-September (south-west monsoon). Impacts of imports of dried fish products to local price determination may not affect to considerably for gold stripped sardinella. Because of that reason price of gold stripped sardinella is highly increased with the time than other analyzed varieties. Coefficient of determination revealed that cubic price forecasting model was the best fit model of price forecasting for all the analyzed varieties. Besides, the Quadratic model can be used to predict the price of analyzed large pelagic varieties. The ARIMA analysis revealed that ARIMA (2,1,12) was the best fit model for price forecasting of skipjack tuna while ARIMA (0,1,0) for double spotted queen and gold stripped sardinella forecasting.
The study concludes that the seasonality of fresh fish is one of the important factors of variation of dried fish prices in Sri Lanka. Therefore, price stabilization policies need to be implemented during the south-west and north-east monsoon season particularly to expand storage facilities during the peak season and to ensure ceiling and floor prices, which would help to limit the speculative practices in the markets. Expansion and further development of price forecasting models can recommend for decision-making on future prices.