China's renminbi (RMB) is gaining international prominence, with its inclusion in the Special Drawing Rights (SDR) currency basket in 2016. The onshore and offshore RMB markets are interconnected, influencing each other through various channels like stock and bond connections. The offshore market, with diverse participants and fewer restrictions, impacts the onshore market. Conversely, the larger onshore market, reflecting fundamental RMB information, influences the offshore market. The COVID-19 pandemic, declared a global pandemic in March 2020, significantly impacted the RMB exchange rate, an emerging market currency with inherent risk. The RMB's central parity rate against the US dollar showed initial depreciation followed by rapid appreciation in 2020. This study examines the volatility spillovers between onshore and offshore RMB markets during this period, focusing on the impact of the COVID-19 crisis and the dominant market in price discovery.

The literature review covers five areas: (i) COVID-19's impact on international financial markets (stock, oil, bond, cryptocurrency markets), showing significant risk spillovers and negative impacts; (ii) the lead-lag relationship between onshore and offshore RMB markets, with studies showing varied results depending on the methodology and time period; (iii) attribution of RMB exchange rate pricing power, with previous studies suggesting changes over time and varied dominance between onshore and offshore markets; (iv) impact of exchange rate reforms (particularly the "8.11" reform) on the relationship between the two markets, generally showing increased inter-market correlations and spillovers; and (v) the application of wavelet transforms to volatility spillover analyses in financial markets, highlighting its ability to capture time-frequency characteristics and non-linear behavior. Existing research often focuses on time-domain analysis, neglecting the time-frequency characteristics and heterogeneity of investor behavior.

The study employs both discrete wavelet transform (DWT) and continuous wavelet transform (CWT) analyses. DWT, using the maximal overlap discrete wavelet transform (MODWT) with Daubechies filter of length 8, decomposes the RMB onshore (CNY) and offshore (NDF) logarithmic return series into smooth and detailed components across different timescales. CWT is used to analyze the power spectrum and cross-wavelet power (coherence) to reveal co-movements, and phase difference to determine lead-lag relationships between CNY and NDF. The data consists of daily CNY (RMB against the US dollar spot exchange rate) and NDF (one-year offshore RMB non-deliverable forward) data from May 1, 2012, to April 30, 2021, obtained from the Wind database. The sample period is divided into pre-COVID-19 (May 1, 2012 – January 22, 2020) and post-COVID-19 (January 23, 2020 – April 30, 2021) periods for comparative analysis.

Using DWT, the study reveals that the correlation between CNY and NDF increases with timescale (investment horizon) both before and after the COVID-19 outbreak, indicating stronger linkages at longer horizons. Post-COVID-19, the correlation coefficients are significantly higher across all timescales, suggesting heightened volatility spillovers. Wavelet cross-correlation analysis shows bidirectional spillovers at longer timescales, with the onshore market (CNY) exhibiting a stronger lead over the offshore market (NDF) at the longest timescale (64-128 days) during the post-COVID-19 period. CWT analysis shows that the onshore market (CNY) displays higher power at shorter timescales (2-48 days), particularly after the "8.11" reform and during the COVID-19 outbreak, suggesting dominance of short-term investors. The offshore market (NDF) exhibits higher power across various timescales, indicating a broader mix of short, medium, and long-term investors. The cross-wavelet power spectrum reveals significant co-movements between CNY and NDF across various time periods and frequencies, particularly after the "8.11" reform and during the COVID-19 crisis, reflecting heterogeneous investor behaviors across investment horizons. The wavelet coherence spectrum illustrates a shift in lead-lag relationships after the "8.11" reform, with the onshore market leading at shorter timescales and the offshore market at longer timescales. During the COVID-19 crisis, the onshore market consistently leads the offshore market, especially at shorter to medium timescales.

The findings demonstrate that the interconnectedness of onshore and offshore RMB markets intensifies at longer investment horizons, with the COVID-19 pandemic exacerbating this linkage. The increased volatility spillovers post-COVID-19 are attributed to factors such as investor panic, risk hedging behavior, and the pandemic's impact on the real economy. The onshore market's lead in price discovery during the crisis highlights its informational advantage and the role of central bank interventions. The "8.11" reform improved the short and medium-term price discovery capabilities of the onshore market but further reforms are needed to enhance its long-term abilities. The results support the idea of heterogeneous investor behavior driving the time-frequency variations in volatility spillovers.

This study provides novel insights into the time-frequency dynamics of RMB volatility spillovers, particularly during the COVID-19 crisis. The findings highlight the importance of considering both timescale and the heterogeneous nature of investor behavior in understanding these markets. The onshore market's leading role during the crisis underscores its importance for investors and policymakers. Future research could incorporate other offshore RMB products and analyze linkages with other international financial markets to provide a more comprehensive understanding of RMB internationalization.

The study uses only CNY and NDF to represent onshore and offshore markets, neglecting other offshore products and their potential interactions. Future research could expand to include these factors. Additionally, while the study analyzes the impact of the COVID-19 crisis, other macroeconomic factors and policy interventions could be explored for a more comprehensive understanding of the observed volatility spillovers.