Spatial price transmission in the onion markets of Bangladesh: An application of NARDL approach

**Background:** Onion is a critical crop in Bangladesh, with annual demand of 2.3–2.5 million metric tons and imports of 0.9–1.1 million tonnes (approximately 40% of demand) from India and neighboring countries. Domestic onion production growth has remained at only 1.85% annually from 2010–11 to 2019–20. Price fluctuations are frequent and sudden, creating misconceptions and undesired perceptions among the population. Despite the importance of understanding price transmission between imported and local onions, no prior study had applied the NARDL approach to analyze asymmetric price transmission in the Bangladesh onion market.

**Methods:** The study employed monthly time series data from January 2006 to December 2020 (180 observations) collected from the Department of Agricultural Marketing, Bangladesh. Variables included local wholesale price (LWP), local retail price (LRP), imported wholesale price (IWP), and imported retail price (IRP), all measured in Bangladeshi Taka (BDT) per kg. Unit root tests (Augmented Dickey–Fuller and Phillips–Perron) confirmed all variables were stationary at first difference. The NARDL model was used to capture positive and negative shocks in both the short and long run. Two models were estimated: Model 1 examined transmission from imported wholesale to local wholesale prices; Model 2 examined transmission from imported retail to local retail prices. The Bounds cointegration test, Wald test for asymmetry, and Pairwise Granger causality tests were also applied. Diagnostic tests included the Ramsey RESET test, Breusch–Godfrey LM test for serial correlation, Breusch–Pagan–Godfrey test for heteroscedasticity, and CUSUM/CUSUMSQ stability tests.

**Key Results:** The Bounds cointegration test confirmed long-run relationships in both models (Model 1 F-statistic = 15.39; Model 2 F-statistic = 13.44, both exceeding upper bounds at 1% significance). The Wald test showed that for Model 1, long-run asymmetry was not significant (F = 1.710582, p = 0.1928) but short-run asymmetry was significant (F = 2.827599, p = 0.0546). For Model 2, long-run asymmetry was significant (F = 9.276738, p = 0.0027) while short-run asymmetry was not (F = 0.357377, p = 0.5508). In Model 1, the long-run coefficients showed that a 1% increase in imported wholesale price leads to a 1.090% increase in local wholesale price, while a 1% decrease leads to a 1.394% decrease, indicating negative shocks have a larger effect. In Model 2, the long-run coefficient for positive shocks (lnIRP⁺ = 0.8606) was greater than for negative shocks (lnIRP⁻ = 0.8399), indicating positive shocks transmit more intensely at retail level. The Granger causality test revealed unidirectional causality from imported wholesale to local wholesale prices (F = 3.26908, p = 0.0130) and from imported retail to local retail prices (F = 5.20475, p = 0.0006). Diagnostic tests confirmed model stability: Ramsey RESET p-values were 0.2430 (Model 1) and 0.2531 (Model 2); LM serial correlation p-values were 0.1707 and 0.2636; Breusch–Pagan–Godfrey p-values were 0.1227 and 0.5996. CUSUM and CUSUMSQ tests confirmed parameter stability for both models.

**Clinical Implications:** This is not a clinical study. The findings have policy implications for agricultural market regulation in Bangladesh. The asymmetric price transmission suggests that onion traders may exercise market power, particularly at the retail level where positive price shocks transmit more strongly than negative ones. Policymakers should consider developing policies to promote competition in the onion supply chain, support storage and distribution infrastructure to prevent short-run asymmetries, and explore public-private partnerships for price management. The study also highlights the need to boost domestic onion yield to reduce dependence on imports and stabilize domestic market prices.