FORECASTING INSTABILITY IN PRICES AND ARRIVALS OF SELECTED VEGETABLES IN NORTHERN INDIAN MARKETS
Loading...
Date
2023-11-02
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
UHF,NAUNI
Abstract
The present investigation titled "Forecasting instability in prices and arrivals of selected
vegetables in Northern Indian markets" explores the arrival and price dynamics of daily used vegetables of
Indian household such as chilly, garlic, ginger, onion, and potato using different statistical analyses. Data
on the monthly arrivals and prices of these selected vegetables were considered for the markets like
Chandigarh, Delhi, Dehradun, and Shimla from the period 2008 to 2009. Descriptive statistics along with
CDVI were calculated for the study data which confimed the presence of instability in the arrivals and
prices ranging from medium to high. The CGR computed for the data showed that the arrivals growth rate
was high in the Dehradun market whereas the price growth rate was high in Dehradun and Chandigarh
markets. Different trend equations were fitted for the data which showed that the cubic model fits well for
arrivals data whereas the cubic and power models fit well for price series. Seasonality indices computed
for the data using Ratio-to-Moving averages confirmed that the arrivals and prices are in perfect relation.
The structural break points computed for the arrivals and prices confirmed the presence of more drifts in
the price series rather than the arrivals series of vegetables across the years. Different methods were
employed for examining the presence of spatial market cointegrations among the markets. The Granger
causality test unveiled the presence of unidirectional and bidirectional relationships among the markets
across different markets. ARDL test was employed for chilly and ginger whereas the Johansen
cointegration test was employed for garlic in which both the test confirms the presence of long-run and
short-run relationships. The presence of short-run relationships in onion and potato was confirmed using
the VAR model. Different forecasting techniques such as ARIMA, ANN, ARIMA-ANN, ARMA
GARCH, EEMD-ARIMA, and EEMD-ANN were employed for the data. The best-performing model was
selected using the different error measure criteria and the selected model was used for future forecasting.
Among all the models, EEMD-ANN, followed by EEMD-ARIMA, emerged as the best-fitting model for
the arrivals and prices series, while ANN and ARIMA-ANN also demonstrated good fits for certain data sets.