Time-frequency volatility spillovers between Chinese renminbi onshore and offshore markets during the COVID-19 crisis

China is the world's largest emerging market economy and the renminbi (RMB) has become an important international currency (加入 SDR in 2016). Offshore and onshore RMB markets are interconnected via multiple channels (e.g., Stock Connects, Bond Connect). The offshore market is more liberal and sensitive to information, while the onshore market is larger in scale and better reflects fundamentals, implying potential two-way influences. COVID-19, declared a pandemic on March 11, 2020, heightened RMB exchange rate volatility and potentially transmitted risks between onshore and offshore markets. The study focuses on whether the COVID-19 outbreak increased volatility spillovers between CNY (onshore) and offshore (NDF) markets, and which market led price discovery during the crisis. The paper motivates the inquiry via mechanisms of investor sentiment, cross-market contagion, and real-economy feedback and highlights the need for time–frequency methods to capture heterogeneity across investment horizons.

The review spans five areas: (i) COVID-19’s effects on international financial markets, documenting spillovers and higher correlations across stocks, oil, bonds, and cryptocurrencies during the pandemic. (ii) Lead–lag between RMB onshore (CNY) and offshore markets (CNH/NDF): prior evidence finds mutual guidance, changing over reforms and frequencies, with mixed conclusions on which market leads. (iii) Attribution of RMB pricing power: results vary over periods and reforms, with some studies indicating onshore dominance pre-reform and growing offshore influence post-reform, while others note renewed onshore influence after policy adjustments. (iv) Impact of exchange rate reforms (notably the 8/11/2015 reform): generally stronger interdependence and changing spillover directions; reforms raised volatility of pricing differentials and information flows. (v) Wavelet methods in financial spillovers: wavelet correlation, cross-correlation, cross-wavelet power, and coherence capture time–frequency co-movements and lead–lag patterns, handling non-stationarity and heterogeneity across horizons. The review identifies gaps: few studies address FX during COVID-19; prior RMB studies often use time-domain linear models, overlooking time–frequency heterogeneity; wavelet tools can fill this gap.

Data: Daily RMB onshore spot USD/CNY (CNY) and one-year offshore non-deliverable forward (NDF) closing prices from Wind, 05/01/2012–04/30/2021. Returns computed as log differences; after filtering, both series have length 2061. Descriptive stats show skewness, excess kurtosis, significant Jarque–Bera rejecting normality, and stationarity by ADF tests; Pearson correlation is 0.792784 (significant). The sample is split into pre-COVID-19 (05/01/2012–01/22/2020) and post-COVID-19 (01/23/2020–04/30/2021). Methods: (1) Discrete wavelet transform using maximal overlap discrete wavelet transform (MODWT) with Daubechies length-8 filter to decompose each series into detail components D1–D6 (approximate timescales 2–4, 4–8, 8–16, 16–32, 32–64, 64–128 days) and smooth A6. Wavelet correlation coefficients are computed across scales, and wavelet cross-correlation functions assess spillovers and potential lead–lag based on symmetry and lag structure. (2) Continuous wavelet transform (CWT) to compute wavelet power spectra, cross-wavelet power spectra (co-movement significance via Monte Carlo), and wavelet coherence with phase differences to infer directionality (lead–lag) across time–frequency, using the Grinsted-Moore-Jevrejeva MATLAB wavelet package. Visual diagnostics include cones of influence and significance contours. The approach captures multi-resolution dynamics, non-stationarity, and heterogeneous investor horizons.

- Pre-COVID-19 period: Wavelet correlation increases with timescale from 0.68817 at D1 (2–4 days) to 0.93291 at D6 (64–128 days), indicating stronger bidirectional spillovers over longer horizons. Wavelet cross-correlation shows weak spillovers at D1–D3, moderate at D4, and strong at D5–D6 (coefficients >0.5 within ±5 lags). Lead–lag is largely symmetric at D1–D5, but at D6 the curve shifts right, indicating CNY leading NDF in the long run. - Post-COVID-19 period: Correlations are high at all scales and exceed pre-COVID levels, rising from 0.87347 (D1) to 0.94164 (D6), evidencing that COVID-19 increased onshore–offshore volatility spillovers. Cross-correlation within ±5 lags strengthens across scales; notably at D5 (32–64 days) amplitudes are larger on the right, indicating CNY leads NDF (a stronger and earlier leadership than pre-COVID). - Cross-wavelet power: Significant co-movements decompose into multiple sub-spillovers across times and frequencies, becoming more pronounced after the 8/11/2015 reform and after the COVID-19 outbreak. Specific high co-movement clusters post-COVID include: (i) Jan–Mar 2020 at 8–32 days; (ii) Oct–Dec 2020 at 8–16 days; (iii) Dec 2020–Feb 2021 at 0–8 days. - Wavelet coherence and phase: After 8/11/2015, the markets alternately guide each other at high frequencies; onshore tends to lead offshore at medium frequencies (32–128 days), while offshore leads onshore at low frequencies (128–256 days). During COVID-19, arrows predominantly indicate CNY leading NDF on 4–64 day scales, supporting onshore dominance in price discovery. - Power spectra: CNY shows high power at 2–48 and 64–128 days after 8/11/2015 and at 16–48 days post-COVID, implying dominance of short- to medium-term investors. NDF exhibits high power across various scales, with medium/long-term activity (64–128 days) prominent around 2015–2016 but declining thereafter, and increased short-term activity post-COVID. - Overall, spillovers are bidirectional and become stronger with longer horizons; both the 8/11 reform and COVID-19 significantly increased spillovers. Quantitatively, Pearson correlation between return series is 0.792784 (significant); return series are stationary with significant ADF statistics and non-normal by Jarque–Bera.

The study addresses two main questions. First, did COVID-19 increase volatility spillovers between the RMB onshore and offshore markets? Evidence from higher wavelet correlations across all scales and more frequent significant cross-wavelet clusters post-2020 indicates a clear increase. Mechanisms include heightened investor panic and hedging, cross-market arbitrage and position adjustments, and real-economy shocks affecting balance-of-payments expectations. Second, which market dominated price discovery during COVID-19? Wavelet cross-correlation and coherence phase arrows show that the onshore CNY market led the offshore NDF over short-to-medium horizons (4–64 days) during the pandemic, consistent with the onshore market’s larger scale, better access to local information, and central bank interventions. Time–frequency heterogeneity is central: spillovers intensify with longer horizons, and leadership can shift by frequency band and over time. The 8/11/2015 reform enhanced onshore liquidity and transparency, improving onshore leadership at medium horizons, while the relatively liberal offshore market remains more sensitive to long-run global and domestic fundamentals, explaining its leadership at lower frequencies outside the COVID-19 window. These findings imply limited diversification benefits at longer horizons and emphasize horizon-specific strategies for investors and risk managers.

This paper shows that: (i) Volatility spillovers between RMB onshore (CNY) and offshore (NDF) markets are bidirectional and grow stronger with increasing timescales, reducing diversification benefits at longer horizons. (ii) Spillovers decompose into multiple time–frequency sub-spillovers, consistent with heterogeneous investor behaviors. (iii) Both China’s 8/11/2015 exchange rate reform and the COVID-19 outbreak significantly increased cross-market spillovers. (iv) Post-8/11, the onshore market improved its short- and medium-term price discovery, while the offshore market retains relative leadership at long horizons; (v) During COVID-19, onshore CNY dominated price discovery at 4–64 day scales, reflecting local information advantages and policy interventions. Implications include greater short-horizon diversification potential, the need for horizon-specific hedging and monitoring (especially of onshore signals during crises), and continued market-oriented reforms to strengthen long-term price discovery onshore. Future research can incorporate additional offshore instruments (e.g., CNH spot and deliverable forwards) and examine broader linkages between RMB FX and other international financial markets.

The study proxies the offshore market using only the one-year RMB NDF and the onshore market with USD/CNY spot rates, excluding other offshore instruments (e.g., CNH spot, offshore deliverable forwards) that may exhibit substitution and price interactions. Results may thus understate or mischaracterize offshore dynamics. Future work should integrate multiple offshore products and analyze their joint interactions with onshore markets, as well as extend analysis to linkages with other international financial markets to inform RMB internationalization.