Analysing the impact of the financial sector, open economy, and climatic factors on agriculture value added in India—an ARDL approach

1 Introduction

Agricultural productivity, a key component in achieving the Sustainable Development Goals set out in the United Nations' 2030 Agenda, is essential for maintaining food security (Fusco et al., 2023). Food security is desirable regardless of political systems and socioeconomic conditions (Jerzak and Smiglak-Krajewska, 2020). It is a primary concern in developing countries due to population growth and intensifying natural disasters, such as floods, droughts, extreme temperature fluctuations, and rainfall, which can imperil food security (Ahmed et al., 2017). Increasing agricultural production and diversifying agricultural land use are strategies to ensure a better food supply (Stocking, 2003; Smyth et al., 2015; FAO, 2024a; World Bank Group, 2024). Thus, the agricultural and related sectors are vital to the global food system and promote livelihood, wellbeing, environmental responsibility, and economic growth (Liu, 2014; Dinar et al., 2012; Djokoto et al., 2022).

According to a Food and Agricultural Organisation (FAO, 2024b) report, 60 per cent of working-class individuals earn their income from agriculture. Although agriculture's contribution to global GDP is only estimated at 4 per cent, growth in this sector increases the income of the poor two to four times more effectively than in other sectors (Townsend, 2015). Thus, contributing to employment opportunities also, thereby making it an instrument for reducing poverty in developing or underdeveloped countries two to three times more effectively (UNCTAD, 2018).

Considering India, it is one of the world's largest countries in terms of agricultural production, with more than half of its land being cropland (Teluguntla et al., 2015). It accounts for 16 per cent of the GDP (Ministry of Finance, 2024). Approximately 46 per cent of the nation's labor force was employed in the agricultural sector in 2016, whereas 64.6 per cent of the population depends on agriculture either directly or indirectly, which plays a crucial role in rural development (Sharmiladevi, 2023). India is the second-largest producer of rice, cotton, tea, groundnuts, wheat, sugarcane, fruits, vegetables, and goat meat (Vidya and Kadam, 2023). Importantly, Indian agriculture is dominated by small and marginal landholdings, which account for more than 80 per cent of total operational holdings. These structural characteristics shape the finance–productivity relationship, as smallholders are more vulnerable to climatic shocks, credit risk, and price instability. Consequently, macroeconomic variables may transmit asymmetrically across farm sizes, making institutional credit access and market integration critical determinants of agricultural value added. Due to the anticipated population growth of over 1.6 billion people by 2050 (United Nations, 2017), India's food supply must be significantly boosted in the coming decades. Studies (Mauser et al., 2015) point out that India has significant potential to boost agricultural output by adopting novel crop varieties and improved management practices.

Over the past few decades, India's Liberalisation, Privatisation, and Globalisation (LPG) policy was instrumental in increasing agricultural output (Pingali, 2012). Since the introduction of LPG in the 1990s, India's agriculture has undergone numerous changes, impacting its growth for various reasons, including agroclimatic and ecological conditions, cropping patterns, investment, fertilizer application, infrastructure support, and yield levels, and agricultural exports, demonstrating their positive effects on Indian agriculture (Chatterjee, 2016). Particularly in states like Maharashtra, the boom in agricultural crop exports has expanded cropping areas and led to a notable transformation in the value of horticulture harvests, including fruits and vegetables. India has been positioned as the largest exporter of agricultural products, with $53.15 billion in exports in 2022–2023, which includes rice, sugar, spices, and marine products (Kumar, 2024; Rajput and Hasan, 2019).

It is also important to note that, in India, agricultural value added, or agrarian output, has become more closely connected to the outside world since the introduction of LPG, thereby affecting economic growth. For example, India's food security has been affected by the Russia–Ukraine war, especially in rural regions, where inflation disproportionately harms food and nutritional security. Though India's share of global agricultural imports increased from 0.56% in 2003 to 1.37% in 2023, indicating India's resilience to imports, such conflicts have led to rising fuel and fertilizer costs, which have negatively impacted the agricultural sector as a whole (Ritter et al., 2023), challenging output and low income (Searchinger et al., 2019; Ben Slimane et al., 2016). The interconnectedness of these geopolitical events significantly affects international trade, foreign direct investment (FDI), food production, agricultural supply chains, and economic growth, puzzling agricultural value added (Jiang et al., 2018). Recalling the research of Professors Deepak Nayyar and Abhijit Sen, who state that separating the effects of trade policies on domestic agriculture is difficult because agrarian development and international trade are intertwined. Thus, the interactions between changes in national and international trade policies determine the long-term effects of trade policies (Nayyar and Sen, 1994), which is still relevant in the present era. From this discussion, it can be inferred that trade, inflation, FDI, and economic growth often significantly affect agricultural output.

FD is projected to influence agricultural growth, as it forms savings, collects cash, and enables simple access to credit and loans for impoverished farmers and other agricultural organizations, increasing agrarian investment (McKinnon, 1974). Different studies have used various proxies for financial development when examining its relationship with agricultural growth. For instance, (Nakazi and Sunday 2020) and Peng et al. (2020) used credit to the agriculture sector as a proxy for FD, whereas Oteng-Abayie et al. (2019) used it as a proxy for credit extended to the private sector. Furthermore, Agbenyo et al. (2019) used domestic credit to the private sector and money supply as indicators of FD. Despite these efforts, existing research still lacks a more accurate and consistent set of variables for measuring FD, particularly in the agricultural context. In the Indian context, agricultural finance functions through Regional Rural Banks (RRBs), cooperative credit societies, NABARD refinancing mechanisms, and commercial bank penetration in rural areas. Despite policy mandates requiring banks to allocate a fixed share of credit to agriculture, this remains largely uneven across regions and farm sizes, with small and marginal farmers often facing liquidity constraints. Thus, investigating macro-level financial indicators without acknowledging these institutional transmission channels risks overlooking structural frictions in credit allocation. Thus, there is a need to further explore more suitable financial indicators to better understand the association between FD and agricultural growth in India. For instance, King and Levine (1993) argue that GCF reflects long-term investment in productive assets and infrastructure, hence enhancing the agricultural productivity and sectoral growth. This was supported by Benhabib and Spiegel (2000) and Timmer (2002). MS reflects the macroeconomic liquidity conditions that lead to transaction, input purchases and market engagement in agriculture (Gurley and Shaw, 1967; Odhiambo, 2009), whereas DC represents the accessibility of formal financial resources necessary for funding farming inputs and facilitating technological adoption, thus enhancing AVA (Beck et al., 2000; De Gregorio and Guidotti, 1995; Rioja and Valev, 2004; Levine, 2005).

It is also essential to understand the environmental impacts on agriculture. Factors such as biodiversity, water usage, biomass appropriation, soil degradation, and land conversion significantly affect agricultural productivity (Ramirez-Contreras et al., 2022). In India, these challenges are more pronounced due to the extensive use of chemical fertilizers, irrigation, and land conversion for urbanization, resulting in biodiversity loss across many regions. These environmental pressures not only impact short-term crop yields but also undermine the long-term sustainability of the agricultural sector (Tesfahunegn and Gebru, 2019; Ahmed et al., 2022). India's agricultural activities are increasingly susceptible to the impact of climatic variation. For instance, Barrios et al. (2010) stress that climate change increases the likelihood of altered rainfall patterns and higher temperatures, as well as the frequency of extreme weather events such as droughts and floods. These shifts threaten the reliability of India's agricultural output, as Indian agriculture is highly dependent on seasonal monsoons. In regions like Maharashtra and Tamil Nadu, droughts have led to severe crop failures, while floods in states like Bihar have disrupted planting seasons. As climate change intensifies, developing nations like India face greater challenges in maintaining food security and economic stability. Therefore, it is crucial to integrate environmental sustainability into agricultural policies to ensure that agricultural growth in India remains resilient, productive, and ecologically viable amid climate variability. Despite the relevance of FDI and financial development to agricultural development, studies on FDI and how credit accessibility in the primary sector influences agriculture are scarce, with few exceptions such as Sharmiladevi (2023) and Chaung (2019).

To conceptualize the linkages between trade, inflation, FDI, and economic growth, this study employs a structural transmission framework in which financial development, openness, and climatic factors influence agricultural value added (AVA) through distinct but interrelated channels. Financial development improves the AVA by easing credit constraints, facilitating capital deepening, enabling adoption of yield-improving technologies, and improving input intensity. Open economy elements impact AVA through export-demand expansion, price transmission mechanisms, imported intermediate inputs (e.g., fertilizers, machinery), and foreign capital inflows. Climatic variables affect AVA through yield variability, production risk, and resource-use efficiency, particularly in nations such as India having a monsoon-dependent agrarian system. These channels jointly determine both short-run output adjustments and long-run equilibrium agricultural growth.

Grounded in the endogenous growth framework (Rebelo, 1991), the model assumes that sustained agricultural growth arises from continuous capital accumulation, technological diffusion, and productivity-enhancing investment. Financial development proxies—gross capital formation (GCF), domestic credit (DC), and money supply (MS)—enter the empirical specification as determinants of capital deepening and liquidity expansion, thereby affecting long-run agricultural value added. These are the primary channels through which financial growth affects agricultural value added—specifically investment, liquidity, and access to finance. Within this framework, openness variables capture external knowledge spillovers and market expansion effects, while climatic variables represent productivity shocks that may alter steady-state growth.

The present study has three main contributions to economic literature: (1) introduction of open economy variables [such as FDI, trade (Tr), and exchange rate (ER)]; (2) application of gross capital formation (GCF), domestic credit to the private sector (DC), and money supply (MS) as proxies of financial development; and (3) climatic variables including rainfall (RN), temperature (TEM), and carbon dioxide emission (CA), thereby investigating the effect of the above-mentioned factors on agriculture value added of India for the period from 1991 to 2023.

Research focusing on these open-economy factors, financial development, and climatic factors addresses endogeneity issues and ensures stability by employing econometric models. In this paper, the researchers have used the ARDL (auto-regressive distributed lag) model with cointegration, bounds test and error correction. As the presence of major structural events- such as the 1991 economic liberalization, the global financial crisis, the COVID-19 pandemic, and the Russia–Ukraine conflict- raises the possibility of regime shifts in the finance–agriculture nexus. Such events may induce structural breaks in the long-run relationship among the variables. Accordingly, this study formally tests for structural stability within the ARDL framework to ensure that long-run coefficients are not biased by potential regime changes.

Further, Granger causality is applied to establish the direction of causality relations. The introductory phase of the study is followed by a literature review in Section 2. Methodology, model specification and estimation techniques were presented in Section 3. In Section 4, the results are analyzed. The discussion and conclusions are given in Section 5.

2 Review of literature

2.1 Agriculture and open economy factors

The initial strand of the literature review primarily focuses on open-economy factors that impact agriculture, such as FDI, trade, and exchange rates. Numerous studies have investigated the relationships among agriculture, FDI, exchange rates, and trade. For instance, in a survey conducted by Edeh et al. (2020), who employed quarterly time-series data from 1981 to 2017 in Nigeria and used the ARDL model to investigate relationships between FDI and the agricultural industry, reported that FDI has a more significant short-term impact on agriculture than a long-term impact. Further, they added that economic deregulation negatively affected agricultural output. Nyiwul and Koirala (2022) researched the contribution of foreign capital inflows to the growth of the farming sector in 16 developing countries. They established a bidirectional causality between agricultural value added and FDI. They suggest that FDI positively affects agricultural value added over the medium to long term. The result implies that countries with high FDI transaction costs can improve agriculture by removing these barriers. Norton et al. (1992) investigated the effect of foreign investment on agricultural production from 1970 to 1985 and reported that foreign aid positively affects agricultural production in Asia and Sub-Saharan Africa, whereas in Latin America and the Middle East, it has an adverse effect. Santangelo (2018) reported a positive but insignificant association between FDI and agricultural growth, suggesting that FDI neither directly influences agricultural growth nor mitigates its effect through other macroeconomic factors. This outcome was in line with Epaphra and Mwakalasya (2017), who used data from 1990 to 2015 to examine the relationship between agricultural value added and FDI in Tanzania. They reveal that FDI inflows have not affected Tanzania's agriculture value-added-to-GDP ratio. These findings contradicted those of Dhungana and Ghimire (2013), who reported a significant and inverse relation.

Studies concerning FDI with agriculture and food security show mixed results. Wimberley and Bello (1992) state that there is an inverse relationship between food security and FDI, whereas Jenkins and Scanlan (2001) and Ben Slimane et al. (2016) report that FDI positively affects food security. In a panel of 55 host developing nations between 1995 and 2009, Ben Slimane et al. (2016) identified that FDI in the agricultural sector increases food security. Dogan's (2022) empirical investigation across 56 developing nations found that FDI in agriculture negatively impacts food security in host nations. FDI is more beneficial in areas with stronger land governance (Iddrisu et al., 2015). Mihalache-O'keef and Li (2011) report that FDI in the primary sector negatively impacts food security, but in the secondary sector, it increases food security. Further, a crucial distinction in this literature concerns aggregate FDI inflows vs. agriculture-specific FDI. While macro-level FDI may stimulate growth through capital accumulation and spillover effects, agriculture-targeted FDI directly impacts farm-level technology adoption, agro-processing infrastructure, and export competitiveness. In India, the share of FDI directed explicitly toward agriculture and allied activities remains modest relative to manufacturing and services, potentially explaining the heterogeneous empirical findings reported across studies.

With regard to the nexus between agricultural growth and trade openness, results are mixed—positive results (Potelwa et al., 2016), negative results (Djokoto, 2013), and mixed results (Sultana and Sadekin, 2023). Using data from 1995 to 2015 in Sotamenou and Nehgwelah (2018) found that free trade policies facilitated the smooth flow of agricultural produce. Similarly, factors such as FDI, the interest rate, and the real effective exchange rate positively and significantly affect agricultural value added. This was in line with the findings of Potelwa et al. (2016). Regarding the relationship between trade openness and agriculture, two studies conducted in Ghana over distinct periods−1995–2009 (Djokoto, 2013) and 1980–2013 (Iddrisu et al., 2015)—report different findings. According to Djokoto (2013), trade negatively influences agricultural performance in the short run, but the relationship between FDI and trade with agriculture is insignificant in the long run. However, Iddrisu et al. (2015) found that trade openness had a positive and significant effect on agriculture in the long term. The divergent findings of these two investigations suggest that the time period and statistical method employed significantly influenced the study outcome. Notably, most empirical studies operationalise trade openness using aggregate trade-to-GDP ratios rather than sector-specific agricultural trade intensity. Such macro measures may dilute interpretability, particularly in agrarian economies where agricultural exports and agri-import dependence follow distinct dynamics from total trade flows. For India, agricultural trade composition—comprising cereals, horticulture, marine products, and processed food—may transmit differently to agricultural value added compared to overall trade openness.

An investigation by Sharmiladevi (2023) using data from 2000–2022 for India revealed that trade openness and agricultural value added are statistically significant and correlated in the short and long run. The stock of FDI and inward FDI do not impact agricultural value added. According to a study by Sultana and Sadekin (2023) examining the influence of foreign investment in the agricultural sector of Bangladesh, FDI has a long-term and significant adverse effect on agriculture, although FDI does not significantly impact the nation's agricultural sector; however, in the case of trade, FDI has a positive but negligible impact on agricultural output in the long run and a negative and insignificant effect in the short run.

In the case of exchange rates, the existing literature offers mixed results. A pioneering study by Johnson et al. (1977) examined the effects of foreign commercial policy and exchange rates on the price of wheat grown in the United States of America and reported that foreign commercial policy had a more significant impact on their domestic wheat price than U.S. policy after using a deterministic short-run forecasting model to analyse wheat prices internationally. They found that a dollar depreciation raised demand for wheat exports, reducing domestic supply and wheat prices. Through the monetary policy channel, Orji et al. (2022) investigate how exchange rates affect agricultural value added in the Economic Community of West African States (ECOWAS) between 2013 and 2019. The findings of using the fixed-effect OLS with the Driscoll-Kraay model show that the exchange rate's influence on the agriculture sector's value added is heterogeneous and significant. Employing data from 1980 to 2016, Nwalem et al. (2023) examine how investment and exchange rates affect Nigerian agricultural output, specifically groundnut production. Using the vector error correction model, Nwalem et al. (2023) demonstrated that public agricultural expenditure and exchange rates positively contribute 0.3% to Nigeria's groundnut production.

Similarly, in their investigation of the effects of certain macroeconomic factors on Nigeria, Asaleye et al. (2023) reported a positive, statistically significant correlation between the exchange rate and agricultural sector production. Recent research by Iorember et al. (2024) examined how Nigerian agricultural value addition was affected by exchange rates, energy costs, and sectoral spending between 1981 and 2020. According to Kernel Regularised Least Squares (KRLS) results, the exchange rate has a positive and substantial effect on agricultural value-added. In the Indian context, exchange rate movements exert dual effects: currency depreciation may enhance export competitiveness for rice, spices, and marine products, while simultaneously increasing the cost of imported fertilizers, machinery, and energy inputs. Empirical evidence from India suggests that the net effect depends on the degree of export orientation and input import intensity within the agricultural sector (Yadav and Chattopadhyay, 2024; Sinha and Sengupta, 2017).

Beyond traditional trade measures, participation in agricultural global value chains (AVA) has emerged as a critical determinant of value addition. Integration into higher segments of AVA- such as processed food exports and branded agricultural commodities- facilitates technology transfer, quality upgrading, and market access. Further, from a structural transformation perspective, trade and capital inflows can alter the agricultural sector's productivity by shifting resources from subsistence farming toward more commercialized and export-orientated production. However, premature exposure to global competition may also heighten vulnerability to price volatility and external shocks. Thus, the agriculture–openness nexus must be interpreted within a broader structural transformation framework.

2.2 Agriculture and climatic factors

The second segment of the literature examines how agriculture is affected by climatic factors. Climate variables impact the agriculture operations through multiple channels: productivity (temperature and rainfall), fertilization effects (CO₂), and environmental degradation. The literature, however, provides inconclusive evidence regarding both the direction and magnitude of these effects. For instance, in the case of the causal influence of CO₂ emissions on agriculture, empirical research has produced mixed results. Appiah et al. (2018) reported a direct causal relationship between CO₂ emissions and agricultural production indexes in both the short- and long-term, using panel data from Brazil, India, China, and South Africa. Edoja et al. (2016) found a one-way causative relationship between CO₂ emissions and agricultural production in Nigeria, employing autoregressive distributed lag, vector autoregressive, and Granger causality techniques. By analysis of 184 nations from 1984 to 2014, Ayyildiz and Erdal (2020) identified a bidirectional correlation between CO₂ emissions and the Crop Production Index (CPI) in the short term for lower-middle, higher-middle, and high-income countries. Another study by Edoja et al. (2016) examined the dynamic relationship among CO₂, agricultural production, and food security in Somalia from 1961 to 2010 using time-series data. The findings indicate a one-way causal relationship and a long-term correlation between CO₂ and food security. Another empirical work by Sarkodie and Owusu (2017) finds bidirectional causation between the crop production index and CO₂ emissions and unidirectional causality from the Livestock Production Index (LPI) to CO₂ emissions in Ghana, based on time-series data from 1960 to 2013. Pata's (2021) investigation of the BRICS nations from 1971 to 2016 indicates a two-way causal relationship between agribusiness and environmental degradation. According to the report, agriculture both contributes to and suffers from environmental contamination, as burning farmland, draining wetlands, agrochemicals, breeding ruminant animals that emit methane, and using fossil fuels in production (Flachowsky, 2015). Druckman and Jackson (2009) and Hillier et al. (2009) did empirical research to determine the carbon footprint and impact of agronomic output using different econometric methodologies. The concept of maintaining a minimal global carbon footprint will go a long way toward improving agricultural productivity, thereby contributing to environmental sustainability. These findings suggest a bidirectional relationship between agriculture and environmental degradation. While CO₂ may enhance productivity through photosynthesis in the short run, long-run environmental stress, and extreme climatic variability tend to suppress agricultural value added.

Beyond carbon emissions, temperature and precipitation represent direct climatic production constraints. Certain studies have examined how precipitation and temperature affect agriculture (Choudhary and Gupta, 2024; Chandio et al., 2020a; Pickson et al., 2023; Ghosh et al., 2023). For instance, Choudhary and Gupta (2024) investigated how climate variability affected important crops farmed in India using the panel data from 1970 to 2020, employing the ARDL method. The results reveal that, in the long term, maximum temperature has a detrimental influence on crop output, whereas carbon dioxide emissions have a favorable impact. An increase in crop productivity was reported in the short run with higher maximum temperatures and rainfall, whereas a decrease in minimum temperature decreased crop production. Given India's monsoon-dependent agriculture and dominance of smallholders, climatic shocks may generate disproportionate output volatility compared to other emerging economies. In a more thorough investigation, Chandio et al. (2020a) used the autoregressive distributive lag (ARDL) technique to analyse the association between climatic variance and agricultural production. Findings reveal that CO₂ significantly increases agricultural productivity in the short and long term, and environmental factors such as temperature and rainfall negatively affect agricultural output. More recently, Pickson et al. (2023) employed the ADRL approach to evaluate the effects of climatic and non-climatic factors on cereal production in China. It was reported that while precipitation improves cereal production in both the short and long term, mean temperature, temperature variability, and CO₂ negatively affect crop production. Higher photosynthetic rates and water loss are associated with CO₂ emissions, which ultimately increase yield (Chandio et al., 2020b). However, higher temperatures would lower rice yields by damaging pollen viability and spikelet sterility (Tesfahunegn and Gebru, 2019). Most research shows that crops grown at higher temperatures and with less precipitation have lower quality and yield (Babel et al., 2011; Shrestha et al., 2017; Asumadu-Sarkodie and Owusu, 2016). To conclude, the evidence indicates asymmetric climatic effects. Moderate increases in rainfall may raise the output, whereas persistent temperature increases and climate variability lead to a decline in agricultural productivity, particularly in developing economies.

In another study using the ARDL model, Ghosh et al. (2023) investigated how climate change indicators have affected Bangladeshi agriculture from 1980 to 2014. In the short term and long term, carbon emissions have a negative and significant effect on agricultural value added, whereas, in the short term, average rainfall positively impacts agriculture. The findings also demonstrate a unidirectional causal association among agricultural production, average rainfall, and carbon emissions. Climate variables negatively impact grain productivity and agricultural value-added, including average temperatures and carbon dioxide (CO₂) concentrations (Pickson et al., 2023). Moreover, an inverted U-shaped relationship exists between financial development and agricultural value-added and cereal production (Chandio et al., 2022).

Despite extensive global evidence, limited studies integrate climatic factors with open-economy variables within a unified empirical framework, particularly for India. Given India's monsoon-dependent agriculture and dominance of smallholder farming, analysing the joint macroclimate transmission mechanism is essential. Moreover, recent global disruptions and climate-related extreme events may generate structural changes in agricultural output dynamics, underscoring the requirement for stability diagnostics in empirical modeling.

2.3 Agriculture and financial development factors

The third strand of literature mainly focuses on the nexus between agriculture and financial development. Numerous studies have investigated the relationships between agriculture and factors such as GCF, GDP, DC, and MS. Financial development affects agriculture through capital accumulation, liquidity provision, risk diversification, and intermediation efficiency. By relaxing credit constraints and improving access to formal finance, financial deepening can enhance input adoption, mechanization, and ultimately agricultural value added (AVA). From a theoretical perspective, financial repression- characterized by interest rate controls, credit ceilings, and directed lending- may distort capital allocation and suppress productivity. In agricultural economies, however, directed credit policies may either alleviate structural financing gaps or generate inefficiencies depending on institutional quality. Awokuse (2009) examined the dynamic relationship between agricultural productivity and economic growth in 15 Latin American, Asian, and African countries from 1971 to 2006, employing the ARDL model. The results show that agriculture is a crucial component of economic expansion. The distribution of private and public resources for infrastructure development and agriculture is vital. Another study by Awunyo-Vitor and Sackey (2018) examined the relationship between foreign investment in the agricultural sector and economic growth using the Granger causality test. They found a positive and significant relationship between economic growth and the agricultural industry. However, the study results demonstrate that neither economic growth nor FDI in agriculture is the cause of the other.

A study by Ogundari (2022) examined the association between Sub-Saharan Africa's economic growth and agricultural production. Using panel data covering 35 countries from 1981 to 2010, he found that economic growth does not affect agricultural production; agricultural growth, however, positively affects economic growth in both the short and long run. Certain studies clearly state the relationship between FDI in agriculture and economic growth. Chandio et al. (2019) examined the relationship between FDI in the agricultural sector and economic growth in Pakistan between 1991 and 2013. Long-term economic growth is positively and significantly affected by FDI in the agricultural industry, according to the ARDL, DOLS, FMOLS, and CCR models. An investigation by Iddrisu et al. (2015) revealed no significant relationship between GCF and agriculture in the short run. However, the study by Sultana and Sadekin (2023) showed that GCF has a positive relationship with AVA in the long run, although it has a negative relationship with GCF in the short run. These mixed findings suggest that macroeconomic stability conditions the finance–agriculture relationship.

It should be noted that persistent inflation may erode real credit availability, distort input prices, and weaken agricultural productivity. Studies on inflation and agricultural sector growth have also produced mixed results. Mekonen (2020) conducted a study to examine the impact of inflation on Ethiopia's agricultural sector between 1980 and 2018. The long-term relationship between inflation and agricultural growth is negative and significant, indicating that inflation hinders rather than promotes agricultural expansion. Kaur (2013) reported that agricultural growth negatively influenced inflation in India. Furthermore, according to research by Mkhatshwa et al. (2015), there is an inverse relationship between inflation and the agricultural sector in Swaziland. Similarly, Janet (2024) reported an inverse association between agriculture and inflation in Nigeria via the ARDL model. Research by Sultana and Sadekin (2023) revealed that factors such as inflation, GCF, and GDP positively impact agriculture in the long run. In contrast, employment in agriculture has a negative long-term impact. In the short run, inflation, GCF, and GDP negatively influence agriculture.

Using a variety of proxies for financial development, multiple studies have found a substantial mixed relationship between financial development and agricultural value added. For instance, Oteng-Abayie et al. (2019) investigated the impact of financial development on agricultural output in Ghana using the ARDL technique. Financial development was measured using the Cobb-Douglas production function, which included the money supply and private-sector credit. The study concluded that financial development, including the money supply and domestic credit, had a negligible detrimental impact on the expansion of agricultural production. The study found that agricultural capital negatively affected agricultural value added. This outcome may reflect structural inefficiencies or weak credit allocation mechanisms. Another investigation by Oluwatoyese et al. (2016) showed that while interest rates, commercial bank loans to agriculture, and food imports significantly affect agricultural production, exchange rates, inflation rates, and unemployment rates have negligible effects on Nigeria's agricultural output. Agbenyo et al. (2019) studied the cointegration analysis of agricultural growth and financial inclusion in Ghana. The Johansen Cointegration methodology and the Fully Modified Ordinary Least Square method (FMOLS) were used to estimate the long-term link between Ghana's agricultural growth and financial inclusion using time series data from 1980 to 2014. Domestic credit to the private sector and agricultural credit were used as proxies for financial development. The study found that domestic credit has a strong but negative impact on the private sector on agricultural value added. The study also showed that loan interest rates, a stand-in for farmers' access to financial services, notably favored agricultural growth. Additionally, using the autoregressive distributed lag (ARDL) model for estimation, Adabor and Essah (2024) investigate the relationship between financial development and agricultural production. According to the ARDL data, financial development has a beneficial and considerable long-term and short-term impact on increased agricultural production. In contrast, net domestic credit and the ratio of liquid liabilities to GDP are adverse to agricultural development, and domestic credit to the private sector is beneficial. Nonetheless, three measures of financial development have a favorable effect on agricultural production, suggesting that a healthy financial system fosters agricultural expansion. Similarly, Iwegbu and De Mattos (2022) investigate how globalization and FD affect the agricultural production performance of BRICS and WAMZ member nations—using money supply and domestic credit to the private sector as proxies for financial development (FD). The study demonstrates that both WAMZ and BRICS member nations' agricultural production growth rates are not accelerated by broad money to GDP. For member nations with poor agricultural production growth trajectories, the negative impact of vast money expansion in WAMZ is much more pronounced. Additionally, improving financial development by expanding private-sector credit boosts agricultural output growth in BRICS but not in WAMZ; the effect is more pronounced in BRICS members with high agricultural output growth rates. Recent finance–growth literature proposes a non-linear (inverted-U) relationship, whereby financial deepening enhances productivity up to a threshold, beyond which excessive credit expansion may generate inefficiencies and resource misallocation. Whether such non-linearity characterizes the finance–agriculture nexus remains an empirical question.

The possibility of reverse causality between agricultural value added and financial deepening cannot be ignored, as rising agricultural income may stimulate savings mobilization and credit demand. To address potential feedback effects, causality analysis within a dynamic ARDL framework should be employed to distinguish short-run and long-run directional relationships. To conclude, the evidence suggests that the finance–agriculture nexus is context-specific, institution-dependent, and potentially non-linear, warranting country-specific empirical investigation.

2.4 Research gap

While a wide array of studies has examined the relationship between agriculture and various macroeconomic, climatic, and financial factors globally and in developing countries, significant gaps remain in India. First, much of the existing literature on agriculture and open-economy dynamics, such as FDI, trade, and exchange rates, has been conducted in cross-country or regional settings, often yielding mixed results. However, studies focusing specifically on India rarely integrate all these open-economy variables into a single empirical framework. Agriculture operates within an interconnected macro-financial-climate system, where openness, credit conditions, and environmental shocks jointly shape production outcomes rather than functioning as isolated determinants. Second, although the importance of financial development for agriculture is acknowledged, few studies have jointly considered gross capital formation (GCF), domestic credit to the private sector (DC), and money supply (MS) as simultaneous determinants of agricultural performance. Most existing analyses employ fragmented financial proxies, failing to capture the broader picture of financial sector development in India's agricultural context. Even where ARDL techniques are applied in Indian studies, the focus remains limited to selected variables, without adopting a fully integrated macro-financial-climate specification capable of capturing joint transmission mechanisms. It is also important to distinguish between domestic credit to the private sector (DC) and agricultural credit. DC captures overall financial deepening and economy-wide intermediation efficiency, whereas agricultural credit reflects sector-specific resource allocation. This distinction allows identification of whether broad financial development or targeted credit policies drive agricultural performance. Third, while the impact of climate change on agriculture has received growing attention, studies in the Indian context often isolate climatic factors such as rainfall, temperature, and CO₂ emissions rather than examining their combined and dynamic effects over time using advanced econometric techniques. Furthermore, the interactions between these climatic variables and economic and financial indicators remain largely unexplored. Moreover, major structural reform episodes in India—including the 1991 liberalization reforms, the 2008 global financial crisis, and the 2016 monetary and institutional adjustments—may have altered the transmission dynamics between finance, openness, and agriculture, yet these regime shifts are rarely incorporated into integrated empirical analyses. Further, potential bidirectional causality between agricultural value added and financial deepening raises endogeneity concerns, necessitating a dynamic modeling strategy with causality diagnostics to properly identify directional effects. Therefore, the current study aims to fill this gap by investigating the influence of open economy factors (FDI, trade, and exchange rate), financial development indicators (GCF, DC, and MS), and climatic factors (rainfall, temperature, and CO₂ emissions) on agricultural value added in India from 1991 to 2023, using a comprehensive and integrated analytical approach. The ARDL framework is particularly appropriate in this context due to its suitability for small samples, mixed orders of integration, and its ability to estimate both short-run dynamics and long-run equilibrium relationships, offering advantages over traditional VECM approaches that require uniform integration order and larger sample stability. This multifactorial perspective is crucial to understanding the broader determinants of agricultural performance and informing holistic policy interventions.

3 Methodology

3.1 Data

The present research examined the dynamic short- and long-run impact of open economy factors such as foreign direct investment (FDI), trade (Tr), and exchange rate (ER); financial development factors such as gross factor production (GCF), money supply (MS), and domestic credit to the private sector (DC); and climatic factors such as rainfall (RN), temperature (TEM), and carbon dioxide emission (CA) along with GDP on agriculture value added (AVA) in India by employing the ARDL method. A time series dataset spanning over 33 years, from 1991 to 2023, was used in the investigation. Data on agricultural value added (AVA), foreign direct investment (FDI), trade (Tr), exchange rate (ER), gross capital formation (GCF), money supply (MS), domestic credit to the private sector (DC), and GDP were sourced from World Development Indicators (WDI) (World Bank Group, 2024). Additionally, the Climate Change Knowledge Portal (2024) provides data on average temperature (TEM) and rainfall (RN). Agricultural value added is the dependent variable in the current study, and all other variables included in the model are treated as independent.

We used EViews 12, an exhaustive econometric software, for all our data analysis and econometric modeling. This software was selected to ensure the validity and dependability of the findings because it has a strong capacity to manage sophisticated econometric methods and diagnostic tests. The factors were converted into logarithms to ensure the data was normally distributed. Though the time-series sample consists of 33 annual observations (1991–2023), the ARDL framework is specifically employed for small-sample applications and provides unbiased long-run estimates even in limited time dimensions. Further, lag length was selected using the Akaike Information Criterion (AIC) to avoid over-parameterization and preserve degrees of freedom. Table 1 provides a detailed overview of factors used in the study and information on sources, rationale, unit of measurement, and data description.

Table 1. Description of variables.

^*represent World Development Indicators.

3.2 Model specification

The ARDL model was employed to accomplish the objectives. Similar models were used by studies such as Iddrisu et al. (2015), Edeh et al. (2020), Iorember et al. (2024), and Sultana and Sadekin (2023) while analysing the influence of FDI on the agricultural sector along with other factors such as exchange rate, trade, and economic growth, and studies of Anh et al. (2023) and Hassan and Mohamed (2024), focusing on how climatic factors such as rainfall and precipitation impact agricultural productivity. Domestic credit to the private sector (DC) is used as a proxy for aggregate financial deepening rather than agriculture-specific lending. The intention is to capture economy-wide liquidity conditions and intermediation efficiency that indirectly affect agricultural investment and capital formation. Whereas, CO₂ emissions are employed as a macro-level proxy for environmental pressure and climate stress. Although not agriculture-specific, total emissions reflect broader climatic dynamics effecting temperature variability, rainfall patterns, and ecological sustainability, which in turn affect agricultural productivity. Although potential interaction effects between financial development and climatic variables may exist, incorporating multiple interaction terms within a small-sample time-series framework would substantially reduce degrees of freedom and may introduce multicollinearity among regressors. Accordingly, the baseline specification focuses on direct effects to ensure parsimony and parameter stability, while leaving interaction dynamics for future research with larger datasets or panel structures. Further, model stability is examined using CUSUM and CUSUMSQ tests to verify the absence of structural instability over the sample period, particularly in light of major macroeconomic and geopolitical events. And given the inclusion of multiple macro-financial indicators within a relatively small-sample framework, multicollinearity was assessed using Variance Inflation Factor (VIF) diagnostics. The results indicate no evidence of severe multicollinearity. Detailed statistics are reported in Appendix Table A1.

3.3 Theoretical model

The present investigation employs the following model, an alternative and extended version of the model used by Sharmiladevi (2023), Hassan and Mohamed (2024), and Iorember et al. (2024), which serves as the theoretical foundation. In this regard, we use the following equation.

AVAit=f(FDIit,ERit,Trit,GCFit,DCit,MSit,RNit,

TEMit,CAit.GDPit)………(1)

Where,

AVA_it = Agriculture value added

FDI_it = Foreign direct investment

ER_it = Exchange rate

Tr_it = Net trade in goods and services

GCF_it = Gross Capital Formation

DC_it = Domestic credit

MS_it = Money supply

RN_it = Rainfall

TEM_it = Temperature

CA_it = Carbon dioxide emission

GDP_it = Economic growth

t represents “time,” i represents “observation,” and f represents “function.”

From Equation 1, we derive Equation 2.

AVAit=FDItβ1.Trtβ2.ERtβ3.GCFtβ4.DCtβ5.MStβ6.RNtβ7.

TEMtβ8.CAtβ9.GDPtβ10.vt… … …(2)

The multiplicative log-linear specification implies constant elasticities, which is consistent with production-theoretic foundations and allows direct interpretation of coefficients as long-run elasticities. This functional form is widely adopted in macro-agricultural growth models and facilitates comparability with prior empirical studies. At the outset, this study's primary objective is to answer the following questions. Did India's AVA get influenced by the abovementioned factors? If yes, what type of impact does it have on AVA? However, earlier studies have shown mixed results. Few studies have reported a positive relationship (Epaphra and Mwakalasya, 2017), a negative relationship (Sultana and Sadekin, 2023), or no significant relationship between FDI and agricultural value added (Sharmiladevi, 2023). In our current study, we assume that FDI, ER, Tr, GCF, RN, and GDP may have a positive effect on AVA in the short and long runs, whereas CA, DC, and MS may have an adverse effect on AVA in the short run, although they may have a positive influence in the long run. We also assume that TEM may have a positive impact in the short run; however, it has a negative impact in the long run.

3.4 Unit root test

The integration level of the series should be determined before determining if there is a cointegration relationship between the factors. Stationarity tests are widely used to prevent spurious regression, especially when analysing time series data (Engle and Granger, 1987). The Augmented Dickey-Fuller test (Dickey and Fuller, 1979) was employed to test this hypothesis. In this test, the alternative hypothesis is that the series is stationary, meaning there is no unit root. According to the null hypothesis, the series has a unit root, indicating that it is not stationary. The variables are said to be cointegrated if a linear combination of the factors is stationary. The cointegrating relationship—the tendency for the factors to change together—is studied via the Engle and Granger (1987) approach, the Johansen and Juselius (1992) method, or the ARDL approach. However, the Engle and Granger (1987) approach or the Johansen and Juselius (1992) method can be employed only when the variables are integrated in the same order. If the factors are integrated in unequal order, the ARDL approach may be utilized.

3.5 ARDL model

Pesaran et al. (2001) introduced the autoregressive distributed lag model. The ARDL techniques are suitable for generating both short- and long-run components between the factors for a small sample size where the sequence in which the variables are integrated is in a different order. The ARDL approach works well when the autonomous factor in the model is I(0) or I(1) or is jointly cointegrated (Fosu and Magnus, 2006; Zhang et al., 2022). However, it is not effective when there is an I(2) in any variable. The dependent-independent relationship was determined via the following model.

AVAt=α0+α1FDIt+α2Trt+α3ERt+α4GCFt+α5DCt+α6MSt+α7RNt+α8TEMt+α9CAt

+α10GDPt+vt………(3)

α₀ is the constant; α₁, α₂, α₃, α₄, α₅, α₆, α₇, α₈, α₉, α₁₀ are the coefficients and v_t is the error term. The following model is developed by transforming all the factors into the natural log.

LnAVAt=α0+LnFDIt+LnTrt+LnERt+LnGCFt+LnDCt+LnMSt+LnRNt+LnTEMt+LnCAt

+LnGDPt+vt………(4)

Equation 4 may be expressed as follows in ARDL form:

ΔInAVAt=α0+∑i=1n1α1iLnFDIt−i+∑i=1n2α2iΔLnTrt−i+∑i=1n3α3iΔLnERt−i+∑i=1n4α4iΔLnGCFt−i+∑i=1n5α5iLnDCrt−i+∑i=1n6α6MSit−i+∑i=1n7α7iLnRNt−i+∑i=1n8α8iLnTEMt−i​+∑i=1n9α9iΔLnCAt−i+∑i=1n10α10iΔLnGDPt−i​+β1LnFDIt−1+β2LnTrt−1+β3LnERt−1​+β4LnGCFt−1+β5LnDCt−1+β6LnMSt−1​+β7LnRNt−1+β8LnTEMt−1+β9LnCAt−1+β10LnGDPt−1+et

………(5)

Where α₀ indicates a drift element, Δ represents the first difference among the variables, and e_t is the white noise error term. According to Gujarati (2009), the term “white noise error term” refers to an uncorrelated random error term with zero mean and constant variance σ². In Equation 3, the coefficients from 2nd to 10th (α_1i to α_10i) indicates the relationship in short and long- term, similar to coefficients from 11th to 20th (β₁ to β₁₀) in Equation 5. Applying the ARDL bound test method, whether the long term relationship exists is checked and further, f-statistic is employed to evaluate the hypothesis.

H0: There is no cointegration. (β1= β2= β3= β4 = β5 = β6 = β7 = β8 = β9 = β10)

H1: Cointegration exists. (β1≠ β2≠ β3≠ β4≠ β5 ≠ β6≠ β7≠ β8 ≠ β9≠ β10)

If the computed F-statistic exceeds the upper bound critical value, the null hypothesis of no cointegration is rejected. If it falls below the lower bound, the null cannot be rejected. Values between the bounds yield inconclusive results. On the other hand, the assumption made by Pesaran et al. (2001) allows for introducing an unrestricted error-correction model, in which the long-run elasticities are the negative coefficients of a one-lag dependent variable. The ECM version of the ARDL model is shown in Equation 6 as follows.

ΔInAVAt=α0+∑i=1n1α1iLnFDIt−i+∑i=1n2α2iΔLnTrt−i​+∑i=1n3α3iΔLnERt−i+∑i=1n4α4iΔLnGCFt−i+∑i=1n5α5iLnDCrt−i+∑i=1n6α6MSit−i​+∑i=1n7α7iLnRNt−i+∑i=1n8α8iLnTEMt−i+∑i=1n9α9iΔLnCAt−i+∑i=1n10α10iΔLnGDPt−I

+γECt-1+et ……………(6)

Where γ represents the speed of adjustment and EC are the residuals obtained from Equation 5.

3.6 Granger causality test

The present investigation also aims to assess the causal impact of the variables involved in the study. Therefore, the Granger causality test (Granger, 1969) was employed to determine if there was any causal relationship among the factors. Granger causality can be stated as follows:

“If the prediction of one time series is improved by incorporating the knowledge of second time series, then the latter is said to have a causal influence on the first.” (Granger, 1969)

In other words, it is mainly used to test if one time series can accurately predict another. Time series x is referred to Granger causal y, often through a sequence of t-tests and F-tests on lagged values of x (and with lagged values of y also given), where x provides statistically significant information about future values of y (Wei, 2013). The time series of two factors results in data length t, including x_t and y_t (t = 1, 2, 3… t), indicating their results at time t. However, X_t and Y_t can be modeled via a bivariate autoregressive model (Equations 7, 8).

xt=α1+∑i=1nβiYt-1+∑i=1nμiXt-1+ut ………(7)

yt=α2+∑i=1nδiYt-1+∑i=1nϑiXt-1+et ………(8)

where,

n = Total number of lags.

α₁, α₂, βi, δi, ϑi, μi are parameters for assessment. The F test may be employed to determine the Granger causality between x and y, and ordinary least squares can be used to evaluate the coefficients.

There are four main basic outcomes for the Granger causality test regarding the association between x and y, such as unidirectional causality between x and y either from left to right ( → ) or from right to left (←); bidirectional causality between x and y indicated by (←), where both parameters cause each other; and, finally, no presence of causality (≠) among x and y (Wei, 2013).

4 Results

4.1 Descriptive data

According to the descriptive analysis shown in Table 2, the mean value addition in agriculture (AVA) during the study period was 2.46E+11. The lowest agricultural value added was in 1991, and the highest was in 2023. Similarly, the average FDI during the period was 2.21E+10, ranging from 73,537,638 in 1991 to a maximum of 6.44E+10 in 2020. The mean exchange rate (ER) was 50.4328. With a range of 16.9877 to 50.4328, the mean trade (Tr) in the study region was 37.0731. The mean gross capital formation (GCF) during the period was 4.64E+11. The average domestic credit (DC) over the period is 39.903, ranging from 22.5108 in 1995 to 54.5053 in 2022. The mean money supply (MS) was 66.2467, ranging from 43.4179 to 87.6767. The average rainfall (RN) was 1,193.43, ranging from 1,002.06 in 1992 to 1,373.32 in 2019. Similarly, the average temperature (TEM) during the period was 24.2306, with a range of 23.55°C in 1997 and 24.85°C in 2016. The average carbon dioxide emission (CA) over the period is 1.2807, ranging from 0.7312 in 1991 to 2.0549 in 2023. The average GDP for the study period was 1.40E+12.

Table 2. Descriptive data.

Source: Author's own evaluation.

4.2 Unit root test

The initial process in performing a cointegration analysis is determining if the variables under investigation have unit roots. Consequently, we first employ ADF test statistics to examine the unit root characteristics of the employed data. Before the unit root test, the variables were transformed to natural logs to address heteroscedasticity. Although alternative unit root tests such as Phillips–Perron could provide additional robustness, the ADF test remains appropriate in small samples when optimal lag length is carefully selected. Moreover, the ARDL methodology is robust to mixed orders of integration, provided no variable is integrated of order two, a condition satisfied in the present study. Both intercept-only and intercept-with-trend specifications were examined depending on the graphical characteristics of each series to ensure consistent deterministic components.

Table 3 displays the ADF outcomes for the level and first difference of the relevant variables. It is concluded that all factors except RN and TEM are stationary at the first difference. Therefore, factors such as AVA, FDI, ER, Tr, GCF, DC, MS, CA, and GDP are integrated into order 1, I(1). That is, the null hypothesis of a unit root cannot be rejected at level but is rejected at first difference for I(1) variables. Since the variables are stationary at different levels, the application of the ARDL method can be justified. Hence, the factors AVA, FDI, ER, Tr, GCF, DC, MS, CA, and GDP are first-differenced for further analysis.

Table 3. Results of Augmented Dickey-Fuller unit root test.

Source: Author's own evaluation.

4.3 Lag selection criteria

It is important to choose a suitable lag order for the variables before applying the ARDL bound test to check if integration exists between the dependent variable LnAVA and explanatory variables LnFDI, LnER, LnTr, LnGCF, LnMS, LnDC, LnRN, LnTEM, LnCA, and LnGDP. The study employed the vector autoregressive (VAR) model's optimal lag-order criteria to determine the optimal lag. The findings in Table 4 present the full set of lag selection criteria for applying the ARDL bound test. Based on Akaike Information Criteria (AIC), the model performs better at lag 2 than at lag 1.

Table 4. Optimal lag order selection based on VAR information criteria.

Source: Author's own evaluation.

^*denotes the lag order selection criteria.

4.4 Testing the cointegrating relationship

Having confirmed that the factors chosen for the current study are not integrated in the same order, the next step is to test for cointegration among the model's variables (Epaphra and Mwakalasya, 2017; Pesaran et al., 2001). The primary goal of the cointegration test is to analyse the equilibrium of long-run relationships among the selected variables. For the cointegration examination, the ARDL bound test was applied. The lag length, chosen based on the lowest Akaike Information Criterion (AIC) value, was used to estimate the f-statistic. The bounds test was conducted under Case III specification (unrestricted intercept and no deterministic trend), consistent with the underlying data structure. The number of regressors included in the unrestricted error correction model was k = 10.

The results of the ARDL bound test, used to examine cointegration among the factors, are displayed in Table 5. According to the bounds testing procedure, if the computed F-statistic exceeds the upper bound critical value, the null hypothesis of no cointegration is rejected; if it falls below the lower bound, the null cannot be rejected. The estimated f-statistic value (23.1235) is greater than 10 per cent, 5 per cent, 2.5 per cent, and 1 per cent of the critical upper limit in order zero and one, respectively, indicating that there is a long-term association among the variables and accepting the alternative hypothesis. The relatively high F-statistic reflects strong joint significance among regressors rather than model misspecification, particularly given subsequent diagnostic tests confirming parameter stability and absence of serial correlation.

Table 5. ARDL bound test for cointegration between the agricultural values added and explanatory variables.

Source: Author's own evaluation.

4.5 Long- and short-run estimations of factors via the ARDL method

In this study, we find long-term cointegration between the studied variables. Moreover, ARDL techniques were used for producing long- and short-term estimations between the dependent variable LnAVA and explanatory variables LnFDI, LnER, LnTr, LnGCF, LnMS, LnDC, LnRN, LnTEM, LnCA, and LnGDP. Table 6 presents the long-run results of the ARDL approach. In the long run, LnFDI, LnER, LnGCF, LnRN, LnCA, and LnGDP positively and significantly impact the agricultural value added. Whereas, LnTr and LnTEM have a negative and significant impact on LnAVA. Though LnMS has a negative coefficient, it is insignificant.

Table 6. Long term elasticity estimates from the ARDL method.

Source: Author's own evaluation.

The ARDL findings for the short run are presented in Table 7. LnFDI, LnGCF, LnTEM, and LnMS have a positive and significant impact. Factors such as LnER, LnTr, and LnCA negatively affect LnAVA. However, LnDC has a negative but insignificant impact on LnAVA. The error correction term [ECM (−1)] is negative and statistically significant, confirming long-run convergence. The magnitude (−1.219) suggests an adjustment speed of approximately 121.9%, indicating an overshooting correction mechanism whereby deviations from equilibrium are corrected within one period with mild oscillatory convergence dynamics.

Table 7. Short term dynamic estimates and error correction results.

Source: Author's own evaluation.

4.6 Granger causality

Based on the value of f-statistics, the association among the parameters indicates the incidence of more significant causality. Lag selection for Granger causality followed information criteria to minimize residual autocorrelation; results remained stable across alternative lag structures. Table 8 displays the pairwise Granger causality and the direction of causality. The findings indicate a unidirectional causality between LnAVA and LnRn, LnAVA and LnTEM, and LnAVA and LnCA because of statistically significant results. For instance, LnAVA → LnCA suggests that agricultural value-added causes CA. Furthermore, there is bidirectional causation between LnAVA and LnFDI, LnAVA and LnTr, and LnAVA and LnGDP. This indicates that agricultural value-added influences FDI and vice versa. But the present research is unable to find a causal association among LnAVA and LnER, LnAVA and LnDC, LnAVA and LnMS, and LnAVA and LnGCF.

Table 8. Paired Granger causality test between agricultural value added and selected variables.

“ → ” represent unidirectional causality, “↔” represent bidirectional causality, and “≠” represent no causality; ^***, ^**, and ^* denote significance at the 1%, 5%, and 10% levels, respectively.

4.7 Diagnostic test

The diagnostic test findings are displayed in Table 9. Based on Jarque-Bera's normality test, the data distribution is almost normal, with a value of 0.214 and a probability of 0.8983. There may not be serial correlation in the model, according to the Breusch-Godfrey L-M test of correlation, which yields a value of 2.3938 and a corresponding probability of 0.9564. The results of the Breusch-Pagan-Godfrey and ARCH heteroskedasticity tests are 1.043 and 1.7479, respectively, with a corresponding probability that they do not exhibit any observable heteroskedasticity. The high p-value in the Breusch–Godfrey test indicates the absence of serial correlation rather than overfitting. Combined with heteroskedasticity and stability diagnostics, this supports model adequacy. Figures 1, 2 display the CUSUM and CUSUMQ results. CUSUM and CUSUMQ plots confirm parameter stability without structural breaks. No sub-sample instability was detected, suggesting consistent model behavior across the study period.

Table 9. Results of model diagnostic and stability test.

Source: Author's own evaluation.

5 Discussion

Examining the impact of an open economy and financial and climatic factors with value added in agriculture (AVA) in India, which spans 33 years, from 1990 to 2022, reveals noteworthy correlations. The study employed the ADF unit root test to test stationarity. The findings show that the variables such as AVA, FDI, ER, Tr, GCF, DC, MS, CA, and GDP are integrated only at the first difference. On the other hand, variables such as RN and TEM are becoming stationary at their own levels. Hence, the factors AVA, FDI, ER, Tr, GCF, DC, MS, CA, and GDP are first-differenced for further analysis. Following the stationarity test, an appropriate lag has been selected. The research applied the lag-order criterion based on the VAR. Based on AIC, lag 2 has been selected, as the model performs better on lag 2 than on lag 1. Given the relatively small sample (33 observations), lag length was restricted to a maximum of 2 to mitigate over-parameterisation risks, and model adequacy was validated through stability and diagnostic tests.

The long-term coefficient of FDI (LnFDI) is positive and significant, suggesting a direct impact of FDI on value addition in agriculture. A 1 per cent rise in FDI increases AVA by 0.77 per cent. More developed infrastructure, efficient farming methods, and easier access to new technology all lead to FDI's having an advantageous impact on agricultural value addition. Moreover, FDI strengthens supply chains, fosters agribusiness growth, and increases capital availability, thereby enhancing agricultural value addition by boosting productivity, reducing post-harvest losses, and improving market accessibility. These findings were consistent with Nyiwul and Koirala (2022) and Awunyo-Vitor and Sackey (2018), who reported a positive relationship between FDI and AVA. And it contradicts the findings of Edeh et al. (2020), Epaphra and Mwakalasya (2017), Sharmiladevi (2023), and Ogbanje and Salami (2022), who reported mixed results of negative impacts on agricultural value addition in the long run (Edeh et al., 2020; Epaphra and Mwakalasya, 2017) and no significant impact on FDI in the long run (Sharmiladevi, 2023; Ogbanje and Salami, 2022). In the case of trade (LnTr), it significantly impacts value addition in the long term, with a negative coefficient of −0.29. This negative impact of trade is more closely explained by structural and policy-related variables. As classical trade theory predicts, gains from trade are based on comparative advantage; however, these gains are conditional on productivity, competitiveness, and institutional readiness (Ricardo, 1817; Krueger, 1978). Following trade liberalization in 1991, India's agriculture was exposed to more import competition with a large-increased farm productivity and infrastructure, leading to displacement effects and reduced domestic value conditions (Krueger, 1990; Timmer, 2002). Furthermore, trade openness has increased price instability, which disproportionately impacts smallholders and can suppress agricultural value addition in the absence of sufficient support systems (Anderson and Martin, 2006). Increased competition from imported agricultural products may have a detrimental effect on value addition by lowering domestic pricing and decreasing profitability. This finding was contrary to the study by Potelwa et al. (2016), which reported a positive relationship between trade and AVA, and the Verter (2016) finding of a negative effect in the short run.

For the exchange rate (LnER), a positive influence on agricultural value added (LnAVA) is observed, with a coefficient of 0.18. This was in line with Ogbanje and Salami (2022). Currency depreciation increases agricultural exports' competitiveness abroad, increasing demand and income for home producers, according to the positive effect of exchange rates on agricultural value added in the long run. Increased investment in agriculture, the adoption of new technologies, and improved infrastructure can result from this increase in export revenue, thereby boosting the value addition and productivity of the agricultural sector. The findings further reveal that the impact of financial factors such as GCF, MS, and DC on AVA differs. GCF (LnGCF) has a positive and significant effect, with a long-run coefficient of 0.31. Sultana and Sadekin (2023) also had similar findings. Increased investment in physical assets, such as infrastructure, storage, irrigation, and machinery, boosts agricultural production, according to the positive and significant long-term impact of Gross Capital Formation (LnGCF) on agricultural value added. Further capital formation promotes modernization and sustained growth in the agriculture sector.

In the case of domestic credit (LnDC), the agricultural value added is negatively affected. A one per percent rise in the LnDC decreases the AVA by 0.13 per cent. This is similar to the result of Oteng-Abayie et al. (2019). Increased credit may not be used efficiently for productive agricultural investments. Further, the negative coefficient of domestic credit indicates that credit expansion alone does not guarantee productive investment. Inefficient allocation, high repayment burdens, and limited financial literacy may divert credit to non-productive uses. Credit targeting and financial inclusion policies are therefore critical. Although the impact of MS (LnMS) is negative, it is insignificant. The insignificance of money supply indicates that liquidity may not transmit effectively to agriculture, possibly due to financial intermediation constraints or preference for non-agricultural investment. Monetary expansion alone is therefore insufficient to stimulate agricultural value addition without complementary institutional mechanisms.

Regarding the climatic factors, rainfall (LnRN) has a positive and significant impact with a value of 0.39. A 1% increase in LnRN increases AVA by 0.39%. Ghosh et al. (2023) also reported similar findings. Accordingly, timely and sufficient rainfall improves crop yields, facilitates irrigation in areas that receive rainfall, and reduces reliance on artificial water sources, thereby directly increasing agricultural output and value addition. Furthermore, this demonstrates that rainfall significantly influences agricultural production, emphasizing the need for water management and conservation measures to improve agricultural productivity, particularly in rain-fed agrarian systems. In the case of temperature (LnTEM), it has a negative and significant impact on the AVA with a value of −0.58. An increase in LnTEM by 1 per cent; AVA will decrease by −0.58. This was in line with the findings of Choudhary and Gupta (2024) and Pickson et al. (2023). Particularly in climate-sensitive areas, higher temperatures can negatively impact agricultural production and output by increasing insect infestations, accelerating evapotranspiration, reducing crop yields, and causing heat stress.

However, carbon dioxide emission (LnCA) has a positive impact on AVA, with a value of 0.47, which is in line with Pata (2021), Chandio et al. (2020b), and Ghosh et al. (2023). This outcome aligns with earlier research by Ghosh et al. (2023), Pata (2021), and Chandio et al. (2020b). The CO₂ fertilization effect, in which higher carbon dioxide levels improve crop photosynthesis and water-use efficiency, may account for the positive connection. This impact might lead to higher yields, especially when other growth conditions, such as water and nutrients, are adequate. However, such a result was contradicted by (Pickson et al. 2023).

For the economic growth variable, as measured by GDP, the effect on AVA is significantly positive, with a coefficient of 1.32. Studies of Awunyo-Vitor and Sackey (2018) and Sultana and Sadekin (2023) also reported a positive association between GDP and AVA in the long run. However, Ogundari (2022) reported no significant relationship between GDP and AVA. The elasticity exceeding unity (1.32) does not imply structural expansion of agriculture relative to other sectors; rather, it reflects the responsiveness of agricultural value addition to overall economic expansion through demand linkages, infrastructure spillovers, and capital deepening effects. This result should also be interpreted in the context of agriculture's declining share in GDP over time, where value responsiveness may increase even as structural share diminishes. However, the positive correlation also suggests that higher public and private investment, better infrastructure, easier access to technology, and strengthened institutional support all contribute to agricultural development during general economic expansion. The demand for agricultural goods rises as an outcome of economic growth, encouraging increased output and value addition. This outcome emphasizes the close relationship between the expansion of the agricultural sector and overall economic performance.

The results of the short-term impact show that LnFDI, LnGCF, LnMS, LnRN, LnTEM, and LnGDP have positive impacts, with values of 0.51, 0.29, 0.24, 0.86, 0.27, and 0.2, respectively. The positive impact of LnFDI is in line with Norton et al. (1992). However, this was contradicted by the findings of Dhungana and Ghimire (2013) and Iddrisu et al. (2015), who reported a negative relationship, as initially most FDI may be in the industrial or service sectors to increase and accelerate profits. The positive effect of rainfall (LnRN) and temperature (LnTEM) is supported by the earlier study by Choudhary and Gupta (2024). Similarly, Ghosh et al. (2023) also support the favorable impact of rainfall in the short run. The relatively high short-run elasticity of rainfall (0.86) reflects India's strong dependence on monsoon-driven agriculture, where rainfall shocks immediately translate into output fluctuations, particularly in rain-fed regions. At the aggregate level, such responsiveness may capture broad climatic sensitivity rather than micro-level agronomic response coefficients.

However, LnTr negatively affects LnAVA, with a value of −0.22, implying that a 1% increase in trade decreases agricultural value added by 0.22. In the study by Aweryo-Vitor and Sackey (2023), trade was positively related. Similarly, carbon dioxide emissions (CA) have a negative impact, with a value of −0.13. Temperature variations can impact agricultural yields, growing seasons, and the kinds of crops that can be grown in a particular area. Increased CO₂ levels can affect agricultural output, photosynthesis, and plant development. Crop yields and farm incomes in tropical countries like India are negatively affected by rising CO₂ emissions, which exacerbate climatic variability through higher temperatures, unpredictable rainfall, and a rise in the frequency of catastrophic events (Schlenker and Roberts, 2009; Lobell and Gourdji, 2012). This is supported by India's policy experience, as despite climate-resilient agriculture programmes, the environmental stress increases, limiting the agricultural productivity and value addition (IPCC, 2014; Dell et al., 2012). It should also be taken into account that the contrasting short-run negative and long-run positive effects of CO₂ emissions may indicate non-linear or threshold dynamics, potentially reflecting short-term climatic stress alongside longer-term CO₂ fertilization effects. However, non-linear specifications were beyond the scope of the present linear ARDL framework and remain an avenue for future research. However, LnER and LnDC negatively affect LnAVA: a 1 per cent increase in LnER and LnDC decreases AVA by −0.04 and −0.11, respectively; the effect is insignificant.

Finally, the study applied the Granger causality test to analyse the causality relationship between the variables. It was found that there is a unidirectional relationship between LnAVA and LnRn, in which rainfall increases agricultural value. Rainfall directly influences crop health, irrigation needs, and soil moisture, thereby boosting agricultural productivity. Similarly, LnAVA, LnTEM, and LnCA have a unidirectional relationship. These findings aligned with the results of Appiah et al. (2018). Because agricultural production is sensitive to specific climatic and atmospheric variables, the unidirectional links imply that changes in these environmental elements (rainfall, temperature, and CO₂) drive changes in agricultural value added. Specific causal effects are, therefore, likely to occur (Appiah et al., 2018).

The introduction of cash, technology and knowledge, foreign investment, or LnFDI (foreign direct investment) and LnAVA (agricultural value-added) can propel expansion in the agricultural sector. Meanwhile, as investors may look for profitable opportunities in agriculture, a strong agricultural industry might draw in additional foreign direct investment. According to Nyiwul and Koirala (2022), these results were consistent. As with LnAVA and LnTr (trade), a higher agricultural value can improve exports and trade opportunities, and the agricultural sector can develop due to increased trade openness and demand for agricultural products. AVA and GDP (Gross Domestic Product) are two examples of how agricultural production may have a substantial impact on the economy as a whole. On the other hand, economic expansion might raise incomes, thereby boosting demand for agricultural goods and starting a vicious cycle. The interdependence of various industries is shown in these reciprocal linkages.

6 Conclusion

In conclusion, the examination of value added in agriculture (AVA) in India, spanning 33 years from 1990–2022, reveals noteworthy correlations with various open-economy, financial, and climatic factors. Using advanced econometric methods and the autoregressive distributed lag (ARDL) model, FDI, GCF, and RN show a positive association with AVA in the short and long terms. In contrast, Tr and CA show negative, statistically significant short-term associations with AVA but a positive association in the long term. DC has detrimental short- and long-term effects on VAA. FMOLS, DOLS, and CCR further validated these results. According to the study, there is a proven bidirectional causal relationship among GDP, Tr, FDI, and agricultural value added. Similarly, a bidirectional causality exists between the area under cultivation and direct credit. With an ECM coefficient of −13.87, it may be inferred that the rate at which errors and shocks from the previous year will be adjusted in the current year is 100.8%. The R-squared value of 91.8 per cent indicates the model's fitness.

Since FDI has a positive impact on agriculture in the long run, though negative in the short run, and given the established bidirectional causality between agriculture and FDI, stakeholders in government and the farming sector should move quickly to increase FDI to boost output. Enacting new laws to provide concessionary and additional incentives, such as extra tax holidays, to attract foreign businesses interested in the agriculture industry should be part of these efforts. Notably, numerous developed and developing countries, including Canada, Bangladesh, Latin America, and India, are placing undue pressure on the environment, resulting in the depletion of natural resources, environmental degradation, and temperature increases with the aim of increasing agricultural. Similarly, financial factors such as GCF and MS have a positive impact in the short run; though DC had a negative value, it is insignificant. First, it means that well-functioning financial institutions and markets stimulate agricultural growth, as all new financial development indices have a favorable impact on agricultural production growth. Second, the findings suggest that equal attention, effort, and resources should be allocated to the development of the various subsectors or components of the financial sector when developing policies to boost financial development and stimulate the growth of the agriculture sector. This is because FD indexes include data from a variety of financial industry proxies, capturing various aspects of FD.

Similarly, RN has a positive impact on both the short and long run. CA has a positive association in the long run, though it has a negative impact in the short run. Subsequent research may prioritize increasing agricultural output to ensure food security while maintaining environmental stability. To enhance agricultural production over time, policymakers should prioritize investments in agricultural infrastructure, technology, and human capital development. To enhance long-term agricultural development and productivity, initiatives to improve access to markets, knowledge, and inputs should be prioritized. The government should establish policies and measures to reduce carbon dioxide emissions and address the effects of climate change on agriculture, such as supporting renewable energy sources and sustainable land management practices. Rural development initiatives must be supported to harness the rural population's potential to increase agricultural production, which can be achieved through investments in education, healthcare, and agricultural extension services. Policymakers should improve water management and conservation initiatives to secure the long-term use of water resources for agriculture, such as irrigation projects and rainwater gathering techniques.

However, there are negative aspects of increased agricultural production: pollution, harm to human health, afforestation, weather changes, and CO2 emissions. Given the variety of agroclimatic zones in India, efforts should be made to improve farmers' knowledge of which alternative crops are best suited to each zone, so that their planting patterns can be planned. While multinational and global entities must ensure that they are meeting the specific development requirements in the investing countries by adhering to fair and ethical business practices when making investments abroad, governments must start focussing on the present and future domestic developmental prospects when receiving foreign capital.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frevc.2026.1648191/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Martinson Ankrah Twumasi, Chengdu University of Technology, China

Reviewed by: Souryabrata Mohapatra, Indian Institute of Technology Jodhpur, India

Amber Gul, Sichuan Agricultural University, China