Price volatility and spot-futures relationship in the Indian castor seed market

Introduction

Castor (Ricinus communis L.) is an important non-edible oilseed crop with significant commercial importance to India. India is a leading producer of castor seed and a major player in the global castor oil trade, which has wide applications in pharmaceuticals, lubricants, paints, coatings and other industries. Gujarat is the major castor-producing state, contributing more than three-fourths of the country's production. In 2025-26, the state accounted for 13.61 lakh tonnes out of India's estimated production of 17.86 lakh tonnes. The export-oriented nature of the castor oil industry and the concentration of cultivation among small and marginal farmers make price fluctuations an important concern for the sector. In India, castor seed futures are traded on the National Commodity and Derivatives Exchange (NCDEX), providing market participants with a means of managing price risk and obtaining forward price signals. During 2025-26, castor futures recorded a traded volume of 19.16 lakh tonnes, equivalent to about 1.07 times the reported annual production (NCDEX, 2026). The close relationship between spot and futures prices is important for understanding price movements in the castor market. Examining the volatility of both markets and their price relationship can help assess how information flows between spot and futures markets and the role of futures prices in the price formation process. Therefore, the study aimed to examine the volatility in spot and futures prices of castor seed in India and to assess the relationship between the two markets.

Data source and research methodology

The study was based on secondary data on monthly spot and futures prices of castor seed for the period July 2004 to December 2024, comprising 246 observations. Futures price data were obtained from the National Commodity and Derivatives Exchange (NCDEX), while spot price data were obtained from the Directorate of Marketing and Inspection (DMI), Ministry of Agriculture and Farmers Welfare, Government of India, for Banaskantha district of Gujarat. Banaskantha was considered for the spot market analysis due to its importance in castor production and trade, with Deesa being the designated basis centre for the NCDEX castor seed futures contract.

Volatility analysis

The volatility of spot and futures prices was examined using ARCH and GARCH models, along with volatility indices to assess the variation in price volatility over the study period.

Arch(Q) model

The ARCH(q) model was used to capture the effect of past squared shocks on the current volatility of castor prices.

Garch(P,Q) model

The GARCH(p,q) model was used to capture both the effect of past shocks and the persistence of conditional volatility in castor prices.

Analysis of spot-futures price relationship

The relationship between spot and futures prices was examined using covariance, correlation, Augmented Dickey-Fuller (ADF) test, Johansen cointegration test, Vector Error Correction Model (VECM), and Granger causality test.

Covariance

The covariance between spot and futures prices was estimated to measure the extent to which the two price series vary together.

Cov =

Where,: Spot price of castor for the observation.

: Futures price of castor for the observation.

: Mean of the castor spot prices.

: Mean of the castor futures prices. n: Total number of paired observations of spot and futures prices.

Index representing the observation.

Correlation coefficient

The correlation coefficient was used to measure the strength and direction of the association between spot and futures prices.

Results and Discussion

Volatility of spot and futures prices

The estimated GARCH (1,1) model was used to examine the conditional volatility of castor spot and futures prices. The results are presented in Table 1. The ARCH coefficients for both spot (1.0095) and futures (0.9837) prices were positive and statistically significant at the 1 per cent level, indicating that recent price shocks had a significant effect on current volatility. However, the GARCH coefficients for spot (-0.0253) and futures (-0.0362) prices were statistically insignificant, indicating limited persistence of past volatility. The results suggest that volatility in both markets was primarily influenced by recent shocks rather than prolonged volatility persistence. The annual volatility indices of castor spot and futures prices were estimated to compare the variation in price volatility acros

The volatility indices showed considerable variation across the study period. Spot price volatility was highest in 2011 (33.09 per cent), while futures price volatility was highest in 2012 (39.74 per cent). In most years, futures prices exhibited higher volatility than spot prices, although the pattern varied across individual years. Towards the end of the study period, volatility declined in both markets, reaching 9.49 per cent for spot prices and 17.52 per cent for futures prices in 2024. Overall, the futures market exhibited relatively higher price variability during most of the study period.

Relationship between spot and futures prices

The descriptive statistics of spot and futures prices were estimated to examine their central tendency and variability during the study period and are presented in Table 3. The mean spot and futures prices were ₹3,848.77 and ₹3,965.04 per quintal, respectively, with futures prices recording a slightly higher average. The coefficient of variation was 39.74 per cent for spot prices and 39.27 per cent for futures prices, indicating broadly similar variability in the two markets. The maximum and minimum prices also indicate significant variation over the study period.

Covariance and correlation analysis was carried out to examine the degree and direction of association between spot and futures prices, with the results presented in Table 4. The spot and futures prices exhibited a very strong positive correlation of 0.9923 in levels, indicating a close association between the two markets. However, the correlation between their first differences was lower at 0.5475, suggesting that shortrun price movements were comparatively less closely related.

The positive association in both cases indicates that the two price series generally moved in the same direction. The stationarity of the spot and futures price series was examined using the Augmented Dickey-Fuller (ADF) test, and the results are presented in Table 5. The ADF test results indicate that both spot and futures price series were non-stationary at levels, as the null hypothesis of a unit root could not be rejected. However, both series became stationary after first differencing, with the test statistics significant at the 1 per cent level under both specifications. Thus, both price series were integrated of order one, I(1), satisfying the requirement for testing cointegration.

Since both spot and futures price series were integrated of order one, Johansen's cointegration test was employed to examine the existence of a long-run equilibrium relationship between the two markets. Both the trace and maximum eigenvalue tests rejected the null hypothesis of no cointegration at the 1 per cent level, while the hypothesis of at most one cointegrating relationship was not rejected. This confirms the existence of one cointegrating relationship between castor spot and futures prices, indicating that the two markets maintained a stable long-run equilibrium relationship during the study period. The Vector Error Correction Model (VECM) was estimated to examine the short-run adjustment of spot and futures prices towards their long-run equilibrium, and the results are presented in Table 7. The estimated cointegrating coefficient of futures prices was -1.0215 and statistically significant at the 1 per cent level, confirming a significant long-run relationship between spot and futures prices. The error correction coefficient for spot prices was -0.1588 and significant at the 5 per cent level, indicating that about 15.88 per cent of the previous period's disequilibrium was corrected through adjustments in spot prices. The adjustment coefficient for futures prices was positive (0.2174) and significant at the 10 per cent level, indicating a comparatively weaker adjustment towards long-run equilibrium.

The results suggest that both markets participate in restoring equilibrium, with a stronger adjustment observed in the spot market.

The Granger causality test was employed to examine the direction of information flow between castor spot and futures prices, and the results are presented in Table 8. The results indicate a unidirectional causal relationship from futures prices to spot prices. The null hypothesis that futures prices do not Granger-cause spot prices was rejected at the 1 per cent level, while the reverse hypothesis could not be rejected. This indicates that past movements in futures prices contain useful information for predicting subsequent movements in spot prices, suggesting that the futures market plays an important role in the price formation process of castor.

Conclusion

The study examined the volatility and relationship between spot and futures prices of castor seed for the period July 2004 to December 2024. The ARCH-GARCH results indicated that recent price shocks significantly influenced volatility in both markets, while volatility persistence was limited. Futures prices exhibited relatively higher volatility in most years. The spot and futures prices showed a strong positive association and shared a stable long-run equilibrium relationship. The VECM results indicated significant adjustment of spot prices towards long-run equilibrium, while Granger causality revealed a unidirectional flow of information from futures to spot prices. Overall, the findings suggest that the castor futures market plays an important role in price formation and provides useful price signals to the spot market.

References

1. Note: *Volatility for 2004 is based on 5 months of data (August-December 2004) as NCDEX castor trading began in July 2004 References 1. Bollerslev T. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics. 1986;31(3):307-327.
2. Dickey DA, Fuller WA. Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association. 1979;74(366A):427-431.