Spillovers between positively and negatively affected service sectors from the COVID-19 health crisis: Implications for portfolio management

The study examines how the COVID-19 pandemic, an exogenous global health shock, altered return and volatility transmission mechanisms among service sectors that were differentially affected: travel and leisure (T&L) was adversely impacted, while healthcare, technology, and telecommunications generally benefited. Prior crises such as the GFC were endogenous to financial systems; COVID-19 uniquely disrupted real and financial sectors via lockdowns and mobility restrictions. The paper seeks to quantify dynamic return and volatility spillovers between T&L and the three positively affected sectors within four key tourism regions (Europe, Eastern Europe, Asia-Pacific, North America) and to draw implications for portfolio allocation and hedging during crisis versus non-crisis periods. Motivated by sectoral heterogeneity documented during COVID-19 and the rising importance of industry effects for portfolio design, the authors ask which sectors transmit or receive shocks, how connectedness evolves over time (especially during COVID-19), and how these dynamics inform optimal portfolio weights and hedge ratios for investors exposed to T&L risk.

The literature documents strong spillovers and contagion during COVID-19 across markets and sectors. Studies focusing on T&L show spillovers from news and media-based indices (panic, media hype, contagion) to tourism equities (Wang et al., 2021; Zargar and Kumar, 2021), heightened connectedness within global T&L stocks with peaks around March and November 2020 (Hadi et al., 2022), and increased systemic importance of smaller tourism firms and downside risk contagion (Shahzad et al., 2022). Sectoral heterogeneity during COVID-19 is widely reported (Goodell and Huynh, 2020; Izzeldin et al., 2021; Liu et al., 2022; Ramelli and Wagner, 2020). Inter-sector work specific to tourism and healthcare in the US shows negative bidirectional return spillovers and effective hedging by healthcare against tourism risk (Salisu et al., 2021). Broader connectedness studies highlight time-varying and crisis-sensitive linkages across regions and sectors (Rehman et al., 2022; Cheng et al., 2022; Mensi et al., 2021a,b). This paper extends prior work by encompassing healthcare, technology, and telecommunications alongside T&L, covering multiple regions central to global tourism flows, and linking connectedness to actionable portfolio decisions using DCC-GARCH-based allocation and hedging metrics.

Data: Daily USD-denominated STOXX sector indices for healthcare (HEALTH), technology (TECH), telecommunications (TELEC), and travel and leisure (TRAVEL) across Europe, Eastern Europe, Asia-Pacific, and North America (total 16 indices). Sample: 01/01/2013–11/12/2021, split into pre-COVID (01/01/2013–12/30/2019) and COVID period (12/31/2019–11/12/2021). Returns are computed as 100*ln(P_t/P_{t-1}). Preliminary tests show non-normality, serial correlation, volatility clustering, and stationarity of returns. Spillover analysis: The Diebold and Yilmaz (2012, 2014) generalized VAR-based forecast error variance decomposition (FEVD) framework is used to compute total spillover index (TSI), directional spillovers (to/from), net directional spillovers, and net pairwise spillovers. A VAR(1) is selected via AIC; 10-step-ahead FEVD; rolling 200-day windows for time-varying system-wide connectedness. Dynamic correlations: Engle’s DCC-GARCH model is estimated in two steps: univariate GARCH(1,1) for each return series followed by the DCC process R_t from standardized residuals. A multivariate t-distribution is employed for quasi-maximum likelihood estimation to accommodate fat tails. The DCC captures time-varying conditional correlations between T&L and each of the other sectors per region. Portfolio design and hedging: Using DCC outputs, the study computes optimal portfolio weights (OPW) for two-asset portfolios consisting of T&L and one other sector j∈{HEALTH, TECH, TELEC} via Kroner and Ng (1998), imposing bounds [0,1]. Risk-minimizing hedge ratios (HR) for a $1 long in T&L hedged with sector j are computed as β_t = h_{TL,j}/h_j (Kroner and Sultan, 1993). Hedging effectiveness index (HEI) (Ederington, 1979) assesses risk reduction relative to unhedged T&L; ΔHEI compares COVID-19 vs pre-COVID performance. Figures illustrate time variation in connectedness, correlations, hedge ratios, and portfolio weights.

- Market behavior and summary stats: During COVID-19, all sector volatilities rose; T&L was the most volatile across regions. T&L returns underperformed and turned negative in three of four regions, whereas technology delivered the highest mean returns (e.g., North America TECH mean 0.141%). - System-wide return connectedness (TSI): North America exhibits the highest return TSI at 61.71%; Asia-Pacific 56.37%; Europe 53.47%; Eastern Europe lowest at 42.08%. Thus, in most regions, over half of a sector’s return forecast error variance is due to non-own shocks. - System-wide volatility connectedness (TSI): Highest in North America (62.87%), followed by Asia-Pacific (54.04%) and Europe (48.97%); Eastern Europe is lower (35.85%). Rolling analysis shows sharp peaks during COVID-19, exceeding 70% in all regions. - Directional spillovers and roles by sector: Healthcare and technology are consistent net transmitters of both return and volatility shocks in Europe and Asia-Pacific; telecommunications is generally a receiver there. In North America, healthcare and telecommunications strongly transmit returns shocks (>70% contribution to others on average) while T&L is a net receiver. In Eastern Europe, telecommunications and T&L are net transmitters of returns, yet for volatility, healthcare and technology remain net transmitters while T&L and telecommunications are net receivers. - T&L’s position: T&L is predominantly a net receiver of return and volatility spillovers during the COVID-19 period across regions, with brief exceptions (e.g., around the 2015 Chinese stock market episode). Pairwise net spillovers confirm that T&L received more shocks than it transmitted during the pandemic. - Dynamic correlations: DCC correlations between T&L and other sectors spike at COVID onset, then decline to levels often below pre-pandemic norms. North America shows relatively strong co-movement overall, but COVID triggers a noticeable decrease in correlations across pairs. - Portfolio and hedging implications: Average hedge ratios pre-COVID often exceed 0.5; hedging costs rose at the pandemic onset (HR spikes). During COVID-19, hedging effectiveness improves in Europe and Eastern Europe across sectors (positive ΔHEI), deteriorates in Asia-Pacific and for two sectors in North America, with telecommunications showing notable improvement in North America. Optimal portfolio weights for the non-T&L sector rise during COVID-19 across regions. Illustratively, during COVID-19 the average non-T&L weight ranges from 49% (Asia-Pacific technology) to 97% (Eastern Europe telecommunications), implying substantial down-weighting of T&L to minimize risk. In North America pre-COVID, the minimum-variance portfolio occasionally allocated zero to T&L when paired with telecommunications.

The analysis demonstrates that COVID-19 substantially intensified both return and volatility spillovers among key service sectors, with healthcare and technology frequently acting as shock transmitters and T&L as a shock recipient. This pattern is consistent with sector-specific pandemic effects: mobility restrictions and demand shocks impaired T&L, while healthcare, technology, and telecom benefited from heightened health demand and digitalization. Regional differences in connectedness (e.g., higher in North America) reflect varying sectoral compositions and market structures. Elevated and time-varying correlations and spillovers during COVID-19 diminish diversification benefits, necessitating adaptive portfolio strategies. The portfolio results indicate that investors can reduce risk by overweighting healthcare, technology, and telecommunications relative to T&L, especially during crisis periods, and by dynamically adjusting hedge ratios as conditions evolve. Improved hedging effectiveness in Europe and Eastern Europe during COVID-19 suggests that these markets offered more efficient risk transfer vis-à-vis T&L exposure, while Asia-Pacific and parts of North America required more cautious hedging due to reduced effectiveness.

The paper documents substantial and time-varying return and volatility spillovers between the adversely affected travel and leisure sector and the positively affected healthcare, technology, and telecommunications sectors across Europe, Eastern Europe, Asia-Pacific, and North America. Spillovers intensify markedly during COVID-19, with T&L becoming a net receiver of shocks. For portfolio design, investors should increase allocations to healthcare, technology, and telecommunications to reduce risk during crises, and employ dynamic hedging as hedge ratios vary across regimes. The study shows that optimal diversification during COVID-19 can improve risk reduction by roughly 17%–42% relative to pre-pandemic. Future research could revisit these relationships post-crisis, consider other systemic events such as the Russia-Ukraine war, and apply alternative connectedness frameworks (e.g., frequency- and quantile-based approaches).

The analysis focuses on four service sectors and four regions with consistent sectoral indices, excluding regions like the Middle East and Africa due to data limitations. The event focus is the COVID-19 pandemic; results may differ under other shocks. Findings are based on STOXX sector indices in USD and a sample ending 11/12/2021; results may change with extended data. While DCC-GARCH with t-innovations addresses fat tails, model choice and parameterization may affect estimates, and full model parameter outputs are not reported.