Modeling price behavior and convenience yield in Indian commodity futures markets

Abstract

Organized commodity derivatives markets in India are relatively new. After the commencement of trading of futures on the national commodity exchanges (NCDEX, MCX, and NMCE) in 2003-2004, the trading volume and value have increased mani fold. Despite this fact, the futures markets have been often widely criticized for speculative trading. Understanding the importance of commodity futures market in a country like India, and the debate on speculative activity requires a systematic enquiry and examination of the behaviour of these futures markets. The emerging futures markets are usually characterized by low liquidity, thin trading, and consequentially, returns exhibit higher sample average, low correlations with developed market returns, non-normality, better predictability, higher volatility, and offer small sample size for research (Bakaert & Harvey, 1997 & Antoniou and Ergul, 1997). Unfortunately, relatively less empirical work has been done in emerging markets contexts. In the Indian context, most of these studies either have serious methodological limitations or are too narrow in terms of commodity/sample size coverage. It is in the context that we undertake the investigation of Indian commodity futures markets under three main objectives (1) price dynamics of futures markets (2) dynamic interlink ages of spot and futures markets and (3) price dynamics of spot and futures markets within the framework of the theory of storage. In this study, futures contracts on four agricultural commodities: Soybean, Maize, Castor seed, and Guar seed, three industrial metals: Aluminum, Copper and Zinc, two precious metals: Gold and Silver, and two energy com modifies: Crude oil and Natural gas, are empirically investigated. Other than Gold and Silver, all other non-agricultural commodity futures in India use spot prices of international markets (LME / NYMEX) for settlement. To investigate the price dynamics of futures markets, we examine (a) the return and volatility characteristics (b) the relationship between futures trading activity and futures return volatility and (c) the international linkages of Indian futures markets. Traditional unit root tests {ADF, PP, and KPSS) and variance ratio test indicate that futures prices are gradually becoming more efficient in terms of randomness. The non-linearity in futures returns commonly known as volatility clustering (ARCH test) is found. Both unconditional and conditional volatility of futures re-turns are estimated and it is found that the volatility of agricultural commodities is lower than non- agricultural commodities. The ratio of overnight volatility and trading volatility is very high {30-50 times) for non-agricultural commodities as compared to the agricultural commodities (2-3 times}. In Indian commodity futures markets, and more specifically, for nonagricultural commodities, we do not find support of Samuelson hypothesis which states that the futures volatility is inversely related to time to maturity. Illiquidity and infrequent trading in the next to near month futures of most of the non-agricultural commodities is probably the cause of higher volatility. The results of augmented GARCH model to test the contemporaneous relationship between futures trading activity and futures volatility indicate a positive and significant correlation between volatility and trading volume for all the commodities. Although volume parameters are significant, volatility is mainly explained by its own lagged values. For most of the commodities, we find insignificant relationship between volatility and open interest. Results on the dynamic relationship between futures trading activity and conditional futures volatility indicate that the overnight volatility drives the futures trading volume but not the open interest. This result is more prominent in case of non - agricultural commodities, where speculation ratio is higher as compared to the agricultural commodities. We investigate the international linkages of Indian commodity futures markets with other developed markets using return and volatility spillovers. We find that futures prices of agricultural commodities traded at NCD- EX and CBOT, prices of precious metals traded at MCX and NYMEX, prices of industrial metals traded at MCX and LME and prices of energy commodities traded at MCX and NYMEX are integrated. In bivariate model, we found bidirectional return and volatility spillovers between MCX and LME markets. However, the effect of LME on MCX is stronger than the effect of MCX on LME. After understanding the price dynamics in the futures markets, we investigate the spot-futures dynamics together and examine some of the important roles of futures markets such as (a) effect of futures trading activity on spot price volatility (b) price discovery in Indian commodity futures markets (c) the efficiency of Indian commodity futures markets and (d) the hedging effectiveness of futures. The unexpected futures trading volume, which is considered as a proxy of information, affects the spot price volatility contemporanously for all commodities. Results on the dynamic relationship between spot price volatility and futures trading activity (unexpected) suggest that in most of the commodities, the unexpected futures trading volume causes spot price volatility whereas for the agricultural and industrial metals, spot price volatility affects the futures trading activity. It is interesting to note that the effect of open interest on spot volatility is not significant. Price discovery role of Indian commodity futures market is investigated through return and volatility spillovers between spot and futures prices. It is found that Indian commodity futures markets do not dominate the price discovery process as they do in other developed markets. In case of agricultural commodities and industrial metal- s, the price discovery takes place in both spot and futures markets. For the precious metals and energy commodities, which are more tradable in nature, futures markets are not affected by spot markets. On further analysis of price discovery process in agriculture commodities during the harvest and the lean period, we find that in the harvest period, when the futures volume is high, futures markets lead the spot market whereas in lean period both markets jointly perform the function of price discovery. Nevertheless, in the recent two years, the price discovery role of futures markets has strengthened compared to the initial two years. As far as the long-run efficiency is concerned, we find that the near month futures prices of most of the commodities are integrated with the spot prices. The integration relationship is not found for the next to near months futures contracts, where futures trading volume is low. We find support for the hypothesis that thinly traded contracts fail to forecast future spot prices and are inefficient. The un- biasedness hypothesis is rejected for most of the commodities. The short-run efficiency is tested with time varying risk premium and it is found that for all commodities, some inefficiency exists in the short-run. We do not find support of time varying risk premium in the Indian commodity futures markets. We investigate the risk management role of Indian commodity futures markets by estimating the constant and dynamic hedge ratios using VECM / VAR and VECM-MGARCH respectively. In case of agricultural commodities, the futures markets in India provide higher hedging effectiveness (30-70%) as compared to industrial metals and energy commodities (less than 20%) when LME / NYMEX cash prices are used . The results are similar whether hedging is done using constant hedge ratio or dynamic hedge ratio. It is found that the near month futures provide higher hedging effectiveness than the next to near month futures. We revisit the price dynamics of spot- futures markets in the framework of the theory of storage and investigate the following issues: (a) the characteristics of convenience yield, (b) the asymmetric volatility of convenience yield, and (c) the non-linear and asymmetric error correction mechanism between spot and futures prices with convenience yield. The net convenience yield of most of the commodities exhibits high autocorrelation. However, the speed of mean reversion is highest in case of Gold & lowest for agricultural commodities. The GARCH parameters estimated from mean reverting equation reveal that the lagged net convenience yield has higher effect on the current net conveniences yield than any demand and supply shock. Seasonality is the NCY is observed for the agricultural commodities but is not statistically significant. The correlation between the net convenience yield and the futures returns is higher than the net convenience yield and the spot returns. This result is opposite to the implications of the theory of storage. The asymmetry in volatility of the net convenience yield is modeled as an EGARCH process and it is found that net convenience yield of most of the agricultural commodities do not exhibit asymmetric volatility except for Crude oil. In case of agricultural commodities, when the spot prices are higher than the futures prices ( backwardation ), the volatility of basis is higher than the volatility of basis when market is in contango. We perform an extensive investigation of the price dynamics of spot and futures markets in Indian commodity markets within the framework of the theory of storage. The asymmetry and non- linearity in error correction mechanism are found in the mean spot returns. In case of most of the non-agricultural commodities, w- hen spot prices are higher than the futures prices (backwardation) , spot prices converge rapidly to equilibrium. We also find the evidence of non-linearity in the error correction mechanism, i e, the prices converge more rapidly when the basis or net convenience yield is high in absolute value. We find that the basis affects the spot returns volatility and that this effect is asymmetric. Spot prices are more volatile when the spot prices are higher than the futures prices than the reverse is true. The findings and implications of the study are potentially useful for commodity exchanges, policy makers and participants besides throwing light on the functioning of commodity futures markets in India for the researches.