How does the self-sufficiency rate affect international price volatility transmissions in the wheat sector? Evidence from wheat-exporting countries

The paper addresses how international wheat price shocks transmit to domestic markets in wheat-exporting countries and whether higher self-sufficiency rates reduce these volatility transmissions. Although much prior research focuses on importing countries, food insecurity persists in exporters as well. The authors highlight substantial co-movement between global wheat prices and retail flour prices in exporters (Canada, Kazakhstan, UK, USA), suggesting external shocks can affect vulnerable households even in exporting regions. The study aims to quantify causal links between world and local prices, measure time-varying volatility connections, and identify determinants—especially the role of wheat self-sufficiency and substitution with maize or rice—that modulate international-to-local volatility pass-throughs. The work is policy-relevant given recent crises (e.g., 2008 food price spike, COVID-19 disruptions) that destabilize prices and supply chains.

A large literature examines price transmission, often within developing countries and from global to local markets (e.g., Abdulai 2000; Baulch 1997; Conforti 2004; Minot 2011; Mundlak and Larson 1992; Robles et al. 2010). However, most studies focus on importing countries and use error-correction frameworks for price levels, with limited attention to volatility transmission. Ceballos et al. (2017) is one exception, using a T-BEKK-GARCH model in developing countries. There is an ongoing policy debate about the effectiveness of self-sufficiency/autarky (Hamilton 1918; Keynes 1933; Kako 2009; Clapp 2017). Quantitative analyses of self-sufficiency’s stabilizing effects exist (Tanaka and Hosoe 2011; Tanaka 2018; Guo and Tanaka 2019) but often rely on CGE models subject to parameter uncertainty. This study fills gaps by focusing on exporting countries, modelling volatility transmissions with DCC-GARCH methods, and empirically identifying determinants—especially self-sufficiency and substitution with maize/rice—of time-varying volatility pass-throughs.

Data and variables: Monthly international wheat price (IMF, U.S. No. 2 Hard Red Winter) and monthly domestic retail flour prices for four exporters—Canada (Statistics Canada, flour-based mixes), Kazakhstan (FAO GIEWS, wheat milling soft), UK (ONS, self-raising flour), USA (BLS, white flour)—from Jan 2006 to Dec 2017. Local prices converted to USD using FRED exchange rates. All series seasonally adjusted with X-13-ARIMA and transformed to continuously compounded monthly returns ln(X_t/X_{t-1})×100. Stationarity checked via ADF, PP, and KPSS; all series stationary in first differences. Structural breaks in mean and variance tested using Bai-Perron; no breaks detected. Step 1: Univariate volatility modelling. For each series r_it (i=1 global, 2 local), estimate AR(1)-GJR-GARCH(1,1): r_it = ω + φ r_{it-1} + ε_it, ε_it|I_{t-1} ~ N(0, h_it); h_it = π + (λ + κ I_{t-1}) ε_{it-1}^2 + γ h_{it-1}, where I_{t-1}=1 if ε_{it-1}<0 else 0. Obtain standardized residuals ξ_it. Step 2: Causality tests. Using standardized residuals, implement Cheung-Ng CCF-based tests for causality-in-mean and causality-in-variance between international and each local market over lags 1–12 months to detect lead-lag relationships. Step 3: Time-varying correlations. Estimate bivariate conditional correlations between global and local return volatilities using Engle’s DCC, Hafner-Franses G-DCC, and asymmetric variants (A-DCC, AG-DCC). Conditional variance-covariance H_t = Γ_t Θ_t Γ_t, with Γ_t = diag(√h_it). Correlation dynamics governed by P_t evolution (standard DCC: P_t = (1−α−β) P + α ξ_{t-1} ξ_{t-1}′ + β P_{t-1}; G-DCC and asymmetric extensions allow matrix parameters and sign asymmetries). Parameters estimated via Gaussian QMLE with BFGS. Model selection via BIC; compute time-varying DCC ρ_{12,t}. Step 4: Determinants of volatility transmission (panel). Construct yearly DCC per country by averaging monthly DCCs. Regress yearly DCC_it on: (i) SSR_it (self-sufficiency rate of wheat = Production/(Production + Import − Export)); (ii) SSR_it and MAIZE_it (log maize consumption); (iii) SSR_it and RICE_it (log rice consumption). Annual SSR and consumption from FAOSTAT (2006–2017). Specification tests: heteroskedasticity (modified Wald), autocorrelation (Wooldridge), cross-sectional dependence (Pesaran). Estimation uses FGLS allowing heteroskedasticity/autocorrelation as baseline; robustness via Prais-Winsten with panel-corrected standard errors (PCSEs).

- Causality: CCF tests show bi-directional causality-in-mean and in-variance between international and local wheat markets. International price returns lead local returns at a one-month lag in all four exporters (e.g., lag 1 IWP→local significant at 10–5%: Canada 1.671*, Kazakhstan 2.390**, UK 1.723*, USA 2.551**). Feedback from local to international prices is also observed at various lags, confirming two-way linkages. - DCC model selection and parameters: Standard DCC best by BIC for Canada, Kazakhstan, UK; G-DCC selected for USA. Mean-reversion (α+β<1) holds. In GJR-GARCH, ARCH and GARCH terms are significant in most cases; asymmetry (leverage) significant for Canada and UK, indicating negative shocks increase volatility more than positive shocks. - Time-varying correlations: Dynamic conditional correlations (DCCs) vary over time and across countries. Descriptive statistics of DCCs (mean, min, max, sd): • USA: mean 0.274; min −0.046; max 0.996; sd 0.154 (highest average and variability). • Kazakhstan: mean 0.181; min −0.052; max 0.345; sd 0.047 (most stable). • UK: mean 0.155; min −0.283; max 0.488; sd 0.100. • Canada: mean 0.148; min −0.140; max 0.638; sd 0.097. Patterns include a spike post-2008 in Canada and a downward trend during the 2007–2008 crisis in Kazakhstan. - Panel determinants of DCC (yearly): Higher wheat self-sufficiency reduces volatility transmission. Estimated coefficients (DCC on SSR): Model 1 (SSR only) FGLS β≈−0.011*** (SE 0.003); PCSE β≈−0.013*** (SE 0.004). Including substitutes: • Model 2 (SSR, MAIZE): SSR β≈−0.016** (FGLS), −0.031*** (PCSE); MAIZE β≈−0.003** (FGLS), −0.006*** (PCSE). • Model 3 (SSR, RICE): SSR β≈−0.021*** (FGLS), −0.038*** (PCSE); RICE β≈−0.004** (FGLS), −0.007*** (PCSE). All significant with expected negative signs, indicating that higher SSR and greater consumption of maize or rice attenuate international-to-local volatility pass-throughs. - Overall: International-local volatility linkages are positive on average and time-varying; self-sufficiency and substitution with maize/rice act as buffers against external volatility.

The findings directly address the core question: higher self-sufficiency rates mitigate the transmission of international wheat price volatility to domestic retail flour markets in exporting countries. The one-month lead of global prices over local prices suggests information and shocks propagate quickly, but elevated self-sufficiency weakens this linkage, likely by reducing reliance on trade-exposed channels. Bi-directional causality indicates exporters’ domestic dynamics can also influence world prices, so stabilizing local markets can feed back positively to global stability. The significant, negative effects of maize and rice consumption on DCCs support a substitution mechanism: when wheat prices become volatile or high, consumers shift toward alternative cereals, dampening the pass-through to local wheat-related retail prices. Time-variation across countries and episodes (e.g., 2007–2008 crisis) underscores that exposure and resilience differ by structural characteristics of markets, supply chains, and policies. Together, these results imply that bolstering self-sufficiency and encouraging diversified cereal consumption can reduce vulnerability of domestic markets to external shocks, benefiting food-insecure populations even within exporting nations.

This study contributes by: (1) documenting bi-directional and time-varying volatility linkages between international and domestic wheat markets in exporting countries; (2) showing that international prices lead domestic prices by about one month; and (3) identifying that higher wheat self-sufficiency and greater maize/rice consumption significantly reduce international-to-local volatility transmission. Policy implications include raising self-sufficiency (potentially via revenue-neutral measures), promoting dietary diversification across cereals, and investing in production stability and forecasting to shield domestic consumers from global turbulence (as highlighted during COVID-19 and export restrictions). The balance between exporting and importing regions must be considered to avoid global destabilization, as increasing self-sufficiency in exporters may affect importers. Future research should compare results using mixed-frequency correlation models such as DCC-MIDAS for aggregating monthly to yearly DCCs, and assess the cost-effectiveness and welfare impacts of self-sufficiency policies and consumption diversification strategies.

- Data scope: Only four exporting countries are analyzed due to monthly retail flour price availability, limiting generalizability. - Price measure: Domestic series use retail flour prices rather than farm-gate or wholesale prices; results may differ across market levels. - Frequency mismatch: Self-sufficiency and consumption variables are annual; monthly DCCs are averaged to annual values, potentially smoothing dynamics. Alternative aggregation (e.g., DCC-MIDAS) is not implemented. - Model assumptions: GJR-GARCH-DCC relies on QMLE and specific correlation dynamics; although multiple DCC variants and BIC selection are used, model misspecification risk remains. - Policy evaluation: The study identifies correlations and determinants but does not quantify the fiscal costs, trade-offs, or welfare impacts of raising self-sufficiency; external validity across different institutional contexts is uncertain.