SPATIAL INTEGRATION AND PRICE FORECASTING ANALYSIS OF MAJOR TOMATO WHOLESALE MARKETS IN INDIA: AN APPLICATION OF SARIMA MODEL 3541
Loading...
Files
Date
2022-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
JAU JUNAGADH
Abstract
Tomato, Growth rate, instability, Trends, Seasonality, Market
integration, Granger causality, Price shocks, forecasting
Tomato is the second most important vegetable crop under cultivation next
only to potatoes in the world. The present study has been undertaken to analyze the
growth rate, instability, the behaviour of arrivals and prices, market integration,
market behaviour to price shocks among the tomato markets. Major eight tomato
markets of India were selected based on the largest average volume of arrivals during
the study period. Time series data regarding monthly tomato arrivals and prices in
selected markets were collected for the period 2010 to 2021 from secondary sources.
Appropriate econometric tools like CGR, CDVI, trend analysis, seasonal indices,
johansen cointegration, granger causality test, VECM, impulse response function and
variance decomposition besides, SARIMA model were employed to meet the
objectives of the study.
In overall, growth rate of arrivals showed that Kolar market recorded a highest
positive significant value of 0.11 per cent and in case of prices Coochbehar registered
non-significant highest growth rate of 0.21 per cent. In month wise also Kolar market
showed higher growth rate of arrivals compared to other markets and in case of prices
Coochbehar showed higher growth rate with more than 10 per cent in all the months.
Instability analysis of arrivals and prices showed that the variability of arrivals is
higher than the prices.
The trend component of arrivals showed negative trend except in Azadpur and
Kolar market while in case of prices all the study markets showed highly significant
positive values over the study period the with the highest annual increase in the prices
was observed in the Coochbehar market (Rs.305.04/q). The seasonal indices of
tomato arrivals had not shown a unique seasonal pattern while prices of study markets
showed a uniform pattern with highest during July. Correlation coefficient of market
arrivals and prices over the months and years showed the inverse relationship between
them with some exceptions. The stationarity tests showed that the all the markets were
stationary at level itself. Engle-granger and Johansen cointegration test confirmed the
presence of long run relationship with each other and thus are integrated so that there
is a smooth transmission of prices across the spatially separated markets.
To ascertain the speed of adjustment among the markets for long run
equilibrium vector error correction model was employed the coefficient of error
correction term of Azadpur market revealed that about 56 per cent adjustment towards
the long run equilibrium occurs in the period of one month. Granger causality test
revealed the presence of bidirectional relationship between the pair’s viz., Azadpur–
Ahmedabad, Kanpur-Azadpur, Coochbehar-Bargarh, Kanpur-Bargarh and Kanpur Coochbehar. Kolar market was the lead tomato market as it was influencing prices of
other selected tomato markets of India except Coochbehar market.
Impulse response function showed that the one-unit standard deviation shock
to any of the market result in the immediate decline in all the other selected markets
during the past 4-5 months thereafter the prices getting stabilized with exception in
Coochbehar market the shock originating from Coochbehar is less transmitted to the
other selected tomato markets. Variance decomposition analysis results showed that
impact of variable on self is greater than the impacted caused by other variables in all
the selected tomato markets of India.
In terms of price forecasting, SARIMA models were found to be the best fit
among different models to forecast tomato prices in all the markets. The SARIMA
model (1,0,3) (1,0,2)12 was selected for predicting onward tomato prices in
Ahmedabad (Gujarat) market while (2,0,1) (1,0,1)12, (3,0,2) (2,1,2)12, (2,0,3) (1,0,1)12,
(3,0,3) (2,0,1)12, (2,0,6) (1,1,3)12, (2,0,4) (1,0,1)12, and (2,0,2) (1,1,0)12 were selected
for Azadpur (New Delhi), Bargarh (Odisha) and Coochbehar (West Bengal) markets,
Kanpur (Uttar Pradesh), Kolar (Karnataka), Mulakalacheruvu (Andhra Pradesh) and
Nashik (Maharashtra). Similarly, SARIMA models were found useful for forecasting
of tomato prices in all the markets with MAPE values ranging from 17.7% in Azadpur
(New Delhi) market to 36.3% in Mulakalacheruvu (Andhra Pradesh) market