An Enquiry into Efficiency of Futures Trading in Agricultural Commodities in India Ashwini Kumar, IES Ministry of Agriculture
Economics of Futures Trading Objectives Price discovery Hedge against risk Trade facilitation Heterogeneity of firms’ behaviour Zero-sum nature Representative individual?
Perspectives Risk Management Perspective Interaction between hedgers ( risk avert) and risk premium earners Transaction cost / Arbitrage Perspective Firms benefit from arbitrage because of their better position in terms of transaction cost. Speculators? Contribute to liquidity and forecasting ability.
Commodity Futures Markets in India Indian Agriculture Prominent sector Source of livelihood for majority Susceptible to weather fluctuations Fragmented Agricultural Markets Inequality in distributional benefits
Commodity Futures Markets in India Long History Reference in Kautilya’s Arthashastra Several forward markets/ Satta in late 19 th -early 20 th century Cotton Trade Association, Bombay, 1875 Specialised in trading of a particular commodity/ group of commodities
Commodity Futures Markets in India Independent India Forward Contracts (Regulation) Act, 1952 Forward Makets Commission in futures trade banned in all major agricultural commodities Khusro Committee Kabra Committee Recommended forward trading in 17 commodity groups.
Commodity Futures Markets in India National Agricultural Policy, Envisaged use of futures contracts. Watershed year Ban on futures trading of all commodities lifted. 3 new Nation-wide multi-commodity exchanges, MCX, NCDEX & NMCE. Electronic trading. Phenomenal growth in turnover since
Efficiency of Futures Markets Efficient market => Market prices reflect all informations Nobody can earn excess profits in a systematic manner. Random walk.
Data and Methodology Two indices of NCDEX NCDEXAGRI- index of spot prices FUTEXAGRI- index of futures prices Identical basket of commodities and same base. FUTEXAGRI constructed on prices of the nearest month expiry contract. Data from 01/Jan/2007 to 03/Oct/ days Opening values of every day.
Descriptive Statistics FUTEXAGRINCDEXAGRI MEAN MEDIAN MAXIMUM MINIMUM STD DEV SKEWNESS KURTOSIS
Econometric tests Tests for stationarity Augmented Dickey Fuller (ADF) Test Philips-Peron (PP) Test Johansen’s Cointegration Test Granger Causality Test Impulse Response Function
Findings Unit Root tests Both indices are not stationary in level form. First Difference of log form, i.e., rates of growth series of these indices are stationary. It implies that while it may not be possible to predict future values, the rate of growth of either of the two series is predictable.
Findings Contd.. Johansen Cointegration test Assuming Linear deterministic trend, and Assuming no deterministic trend. There are two cointegrating equations implying that rates of growth of the two indices have long-term relationship.
Causality Findings Granger Causality Test results imply No causality in any direction Rate of growth in futures prices do not depend on rate of growth in spot prices and similarly the other way round.
Impulse Response Function Results imply that Rate of growth of futures prices get affected by any exogenous shock in rate of growth in spot prices but not vice versa. In case of an exogenous shock to rate of growth in spot prices futures prices take longer to stabilize than the spot prices themselves.
Conclusions Futures market not efficient in short run. Change in spot prices are found to affect futures prices. Effect of change in futures prices on spot prices is found to be minimal.
Thank You
Unit Root Tests Results VariablesADF StatisticPP Statistic FUTEXAGRI Level Form Level form of Log First Difference form of Log NCDEXAGRI Level Form Level form of Log First Difference form of Log Critical Values ADFPP 1% % %-3.21
Johansen cointegration test result Hypothesized No. of CE(s) Eigen valueLikelihood Ratio5% Critical Value1% Critical Value None** At most 1** Hypothesized No. of CE(s) Eigen valueLikelihood Ratio5% Critical Value1% Critical Value None** At most 1**
Granger Causality Test result Null Hypothesis:ObsF-StatisticProbability LN_NCDEX_101 does not Granger Cause LN_FUTEX_ LN_FUTEX_101 does not Granger Cause LN_NCDEX_