Dynamics and volatility transmission of tea prices:implication for small tea growers in Kerala

Abstract

Tea (Camellia sinensis, f: Theaceae) is an important plantation crop that originated in the borderlands of China, India and Burma. It is a perennial tropical crop that requires high elevations and a humid climate. Globally, major producers of tea are China, India, Sri Lanka and Kenya, where China is the leading producer followed by India. In India, tea plantations were established during the colonial period and are cultivated mainly in Assam and West Bengal in the North and Kerala, Tamil Nadu and Karnataka in the South. This study mainly focuses on Kerala, which is the second largest tea producer in South India. In India, tea is cultivated by Small Tea Growers (STGs) who cultivate on a land area of less than 25 acres and big tea growers who cultivate on much larger area. Usually, the leaves harvested by STGs are sold either through intermediaries or directly to the auction centres. However, the tea industry faces persistent challenges, among which price volatility is the most critical one faced by growers, especially STGs. Price instability has profound implications for STGs, who are highly vulnerable to income fluctuations due to limited capital and market access. Volatile prices affect their economic welfare and contribute to long-term uncertainty in the sector. Accordingly, this study was done with the following objectives: (a) to analyse the volatility and transmission of tea prices, (b) to assess the impact of price fluctuations on the welfare of STGs, and (c) to identify suitable forecasting models for predicting tea prices. Data on area, production, productivity and annual auction prices were collected from the Tea Board of India, Indian Tea Association and international sources, and primary data were collected through field survey from 200 STGs across Idukki and Wayanad districts of Kerala. In order to analyse the effect of trade liberalisation on the area, production, productivity and auction prices, the data were segmented into two groups, i.e., pre-liberalisation period (1980-2000), post-liberalisation period (2001- 2024) and the overall period (1980-2024). Analytical tools included growth and instability models like Compound annual growth rate (CAGR) per cent per annum, Cuddy-Della Valle Index (CDVI) and Coefficient of Variation (CV). Decomposition analysis was done using Hazell’s decomposition model. The ARMA-APARCH model was employed for the estimation of volatility persistence, and Johansen’s cointegration (ii) and Granger causality tests were used for price transmission analysis. In order to study the influence of price volatility on the welfare of STGs, econometric models such as Propensity Score Matching (PSM), Augmented Inverse Probability Weighting (AIPW), and Inverse Probability Weighted Regression Adjustment (IPWRA) were used. To identify the best forecasting model, traditional time-series models (ARIMA, SARIMA), regression models (Linear, SVR, GBR and RFR), deep learning techniques (RNN and LSTM), and hybrid combinations (ARIMA-LSTM and ARIMA-GARCH) were evaluated based on statistical performance metrics including RMSE, MSE, MAE, MAPE and R2 and residual diagnostic tests. In North India, the CAGR of area, production, and productivity showed a steady increase in both pre- and post-liberalisation periods. In South India, however, the growth rates declined after liberalisation, with slower expansion in area and productivity. In both regions, growth in annual auction prices decreased after trade reforms. In North India, instability in area, production, and productivity increased after liberalisation. In contrast, South India experienced a decrease in instability during the post-liberalisation period. In both regions, instability in annual auction prices reduced after liberalisation. In Kerala, the growth rate in area was very low, such that it could not be offset by a marginal increase in productivity. This resulted in much lower growth in production. The decomposition analysis revealed that the price effect contributed the most to the variation in tea income across all regions. The trend analysis revealed a significant positive and consistent upward movement in monthly auction prices for both North and South India. The seasonal analysis revealed distinct intra-year price movements in both regions, with North Indian prices showing stronger seasonality. Cyclicality was more evident in North Indian tea prices, showing longer and sharper price cycles driven by market demand and production adjustments. In contrast, South Indian prices showed shorter, milder cycles. In North India, volatility persistence (β) was present in the pre- liberalisation period, indicating strong clustering of shocks, whereas it increased after liberalisation, showing the influence of increased trade activity in North India after implementing the trade reforms. The overall period also reflected a considerable persistence level, confirming that liberalisation increased long-term volatility persistence in North Indian tea prices. Volatility persistence in South India declined (iii) notably from the pre-liberalisation to the post-liberalisation period, indicating quicker dissipation of price shocks after trade reforms. Compared with North India, South Indian prices showed lower overall persistence and weaker volatility clustering, suggesting a more stable and less reactive market environment. Price transmission between markets was stronger in North India, showing greater integration with both domestic and international markets, while South Indian markets exhibited weaker linkages. In the post-liberalisation period, transmission improved significantly in both regions, reflecting enhanced market connectivity and faster price adjustments. The Granger causality test revealed bidirectional causality among the domestic and international markets before and after liberalisation. During the post-liberalisation period, causal linkages strengthened such that overall, this confirmed improved price transmission and greater interdependence between domestic and global tea markets. In both Idukki and Wayanad, labour costs constituted the major share of the total cost. The cost of cultivation was higher in Idukki (₹2,44,551 ha⁻¹) than in Wayanad (₹2,15,567 ha⁻¹), mainly due to greater labour use. However, Wayanad achieved higher gross returns and a higher benefit-cost ratio compared to Idukki. This was because in Wayanad, the sample STGs followed organic cultivation method and sold the tea leaves at a higher fixed price. In Idukki, the tea leaves were sold through intermediaries, while in Wayanad, it was sold directly to the Bought Leaf Factory (BLF). Marketing efficiency was notably higher in Wayanad (5.94%) than in Idukki (1.82%) due to the absence of intermediaries in Wayanad. The wider price spread and higher marketing margin in Idukki indicate reduced producer share and lower overall efficiency compared to Wayanad. Constraints in tea cultivation were analysed using the Garrett ranking technique. It was found that the most critical constraint identified in Idukki was price volatility, followed by issues like high wage rates and climate change. The most severe issue in Wayanad was labour shortage, followed by disease and pest incidence. Welfare of the STGs was analysed using PSM, AIPW and IPWRA. It was found that on average, the annual household consumption expenditure reduces by Rs. 22,916 to Rs. 34,661. Income inequality was analysed using the Gini coefficient and the Lorenz curve. It was found that there were significant inequalities between the STGs in Idukki and Wayanad districts in terms of their gross annual income. (iv) The forecasting modelling demonstrated that while classical models like ARIMA and SARIMA captured linear trends effectively, they failed to model complex nonlinear dependencies inherent in tea price data. Deep learning models, particularly LSTM, outperformed traditional methods, and the ARIMA–LSTM hybrid model achieved the highest accuracy. For both North Indian and South Indian price series, the ARIMA-LSTM hybrid achieved the lowest RMSE and best overall predictive accuracy, outperforming both traditional and machine-learning approaches. This finding underscores the potential of combining statistical and machine learning approaches for improved predictive performance in agricultural commodity markets. From the study, it was revealed that STGs in Idukki who faced price volatility had a welfare reduction in consumption expenditure ranging from Rs. 22,916 to Rs. 34,661 per year. This indicates the need for intervention in price stabilisation. It was also found that the cost of production per kg was Rs. 22 in Idukki. So, whenever the price falls below the minimum cost of production, especially during peak production months, the government may provide a Price Deficiency Payment (PDP), which could also be observed in the case of other plantation crops like rubber. Based on the empirical evidence, the study recommends providing financial aid for the establishment of cooperative and SHG-based bought-leaf factories (BLFs) to ensure fair pricing and reduce dependence on private intermediaries. The development of a digital market intelligence platform integrating real-time price data, predictive analytics, and advisory services would empower growers to make better marketing decisions. Additionally, the study advocates promoting crop diversification, which helps to mitigate the risk of the STGs. The results emphasise that trade liberalisation and technological integration have enhanced market connectivity but simultaneously exposed domestic producers to external price shocks. Strengthening institutional mechanisms, ensuring access to reliable price information, and encouraging cooperative market structures can mitigate risks and enhance sustainability. The research contributes to the academic understanding of price dynamics in perennial crops and provides actionable insights for policymakers, emphasising that managing volatility through predictive modelling and institutional innovation is key to securing the livelihood of smallholders. In conclusion, the study affirms that effective price stabilisation policies, digital forecasting tools, and (v) cooperative empowerment are vital for achieving inclusive growth in Kerala’s tea sector. By integrating advanced econometric techniques with machine learning models, this research not only enriches the literature on agricultural price volatility but also provides a policy-relevant framework for ensuring the long-term viability and economic security of STGs in volatile global commodity markets.