Analysis of main determinants of soya bean price volatility in Malawi
Date
2018-11
Authors
Chimaliro, Aubrey Victor
Journal Title
Journal ISSN
Volume Title
Publisher
African Economic Research Consortium
Abstract
The study primarily focuses on analysing the extent of soybean price volatility in Malawi.
The interest in the study was triggered by the findings from literature noting soybean prices
in Malawi as being particularly volatile. Soybean is one of the most important oilseed crops
in Malawi and has the potential to become a major export crop. It offers good export
prospects to neighbouring countries in Southern Africa, but has also been prioritised for its
potential domestic contribution in Malawi. It is regarded as one of the value chains that
promotes better nutrition in Malawi since most diets are dominated by maize, which
contributes to malnutrition in Malawi.
The study empirically estimates soybean price volatility using a GARCH (1,1) model and
results indicate that both the lagged squared residual and the lagged conditional variance have
an effect on the conditional variance of soybean prices. GARCH terms are significant,
indicating some volatility clustering in monthly returns of soybeans in Malawi, South Africa
and the world. The study confirms that soybean prices in Malawi have been more volatile
relative to South Africa and the USA.
To evaluate the extent to which domestic soybean price volatility can be attributed to regional
and global market volatility, the Engle-Granger procedure was employed to estimate long-run
co-integration between soybean prices in Malawi, South Africa and the world. The prices are
categorised into six pairs and testing the long-run co-integration between these pairs involve
both directions. Five out of the six pairs of prices exhibit long-run co-integrating
relationships.
An error correction model (ECM) is also employed to estimate the speed of adjustment to the
equilibrium for five co-integrated price series. South Africa is the fastest in responding to the
USA price changes, taking two months. Malawi is the second fastest since it takes about four
months for soybean prices to respond to shocks in South African markets. However, it takes
about seven months for the USA soybean prices to respond to price shocks in the South
African markets which is longer than the period that Malawi takes to respond. South Africa
takes nine months to respond to the shocks that occur in the Malawian markets. USA is the
lowest in terms of the speed of adjustment since it takes about sixteen months to respond to
the Malawian market shocks. Therefore, this study agrees with expectation that changes in
the international markets affect the domestic markets. This is so because South Africa is a
small nation in the international soybean market and Malawi is even much smaller – the
volumes traded in these markets are considered too small to have any meaningful impact on
world market prices.
Lastly, to evaluate the influence of shocks on the South African prices, as well as selected
macro-economic variables in Malawi on soybean price volatility in Malawi, the study
employs a vector error correction model (VECM) to evaluate the long- and short-run
relationship between soybean prices and explanatory variables (South Africa soybean prices,
exchange rates and consumer price index). The error correction coefficient of -0.2089 is
negative and highly significant which is in line with expectations. The Johansen test points to
the possibility of 1 or 2 cointegrated relationships between variables. The Wald test results
show that the probability values of the joint F-statistics for South African soybean prices and
Malawi exchange rate are significant at 10% and 5% levels respectively. Based on the
significance of both the error correction term and the joint F-statistics of the two lagged
variables in the Wald test results, the study concludes that South Africa soybean prices and
Malawi exchange rate are significant drivers of volatility in Malawi soybean prices.
Description
Keywords
Price volatility, Price transmission, Generalised Autoregressive Conditional Heteroskedasticity (GARCH) and Vector Correction Model (VECM).