Evaluating the Trade-growth Nexus in India Using Hybrid Econometrics and Explainable Machine Learning

Suneel Kumar Duvvuri, Balayya Rajana, Sanjeev Kumar Chejarla, Prasad Teja Dakey, K. Ramachandra Rao and Thilothu Rao Gandam
Additional contact information
Suneel Kumar Duvvuri: Department of Computer Science, Government College (Autonomous), Rajahmundry, Andhra Pradesh, India.
Balayya Rajana: Department of Economics, Government College (Autonomous), Rajahmundry, Andhra Pradesh, India.
Sanjeev Kumar Chejarla: Department of Economics, Government College (Autonomous), Rajahmundry, Andhra Pradesh, India.
Prasad Teja Dakey: Department of Economics, ESLA, SRM-AP University, Amaravati, Andhra Pradesh, India.
K. Ramachandra Rao: School of Economics, University of Hyderabad, Hyderabad, Telangana, India.
Thilothu Rao Gandam: Centre for Economics and Social Studies, Hyderabad, Telangana, India.

Post-Print from HAL

Abstract: The trade–growth relationship in India has been widely studied, with evidence suggesting a long-run association but considerable variation driven by structural shifts, policy changes, and global economic conditions. However, existing literature highlights potential asymmetries and nonlinearities in this nexus, motivating the use of hybrid econometric and machine learning approaches for deeper analysis. This paper examines the relationship between trade openness and economic growth in India using annual data from 1975 to 2024. ARDL and NARDL models estimate long-run equilibrium and test for asymmetric adjustment; Random Forest and XGBoost models provide predictive benchmarks, with SHAP values used to interpret feature contributions. The ARDL results confirm a stable cointegrating relationship. World GDP growth has a positive and significant effect on domestic output (β = 0.657, p < 0.05), and gross capital formation contributes positively at a weaker level (β = 0.130, p < 0.10). The error correction term (ECT = −1.197, p < 0.01) indicates fairly rapid reversion toward equilibrium following a shock. NARDL estimates show sign differences between the effects of positive and negative trade shocks, but a Wald test fails to reject symmetry (p = 0.522). The asymmetry, in other words, is apparent rather than statistically meaningful. Trade openness also does not register as a significant short-run growth driver. Among the models tested, NARDL edges out ARDL on prediction error (RMSE = 2.59 vs. 2.74), while both machine learning models perform worse (XGBoost: 3.77; Random Forest: 3.89). SHAP analysis identifies lagged trade openness, global growth, and capital formation as the dominant contributors. Trade openness is associated with growth in India, but the relationship is conditional. Global economic conditions and domestic investment drive outcomes more consistently - a finding that argues for stable trade policy and attention to underlying macroeconomic fundamentals.

Date: 2026-05-11
References: Add references at CitEc
Citations:

Published in Journal of Economics and Trade, 2026, 11 (1), pp.454-476

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:hal:journl:hal-05619581

Access Statistics for this paper

More papers in Post-Print from HAL
Bibliographic data for series maintained by CCSD ().