Connectedness between DeFi, cryptocurrency, stock, and safe-haven assets

Finance Resea ch Le e s 53 (2023) 103692 A ailable online 10 Feb ua y 2023 1544-6123/© 2023 The Au ho (s). Published by Else ie Inc. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). Connec edness be ween DeFi, c yp ocu ency, s ock, and sa e-ha en asse s And ea Ugolini a , Juan C. Rebo edo b , c , * , Walid Mensi d , e a Depa men o Economics, Managemen and S a is ics. Uni e si y o Milan-Bicocca, I aly b Depa men o Economics, Uni e sidade de San iago de Compos ela, Spain c ECOBAS Resea ch Cen e, Spain d Depa men o Finance and Accoun ing, Uni e si y o Tunis El Mana and IFGT, Tunisia e Depa men o Economics and Finance, College o Economics and Poli ical Science, Sul an Qaboos Uni e si y, Musca , Oman ARTICLE INFO JEL classi ica ion: C32 G15 R30 Keywo ds: DeFi C yp ocu ency Sa e-ha en asse Financial condi ions Spillo e s ABSTRACT This pape examines e u n spillo e s wi hin and be ween di e en DeFi, c yp ocu ency, s ock, and sa e-ha en asse s. Fo he pe iod Janua y 2019 o Ma ch 2022, we ind ha DeFi and c yp ocu ency asse ma ke s exhibi s ong wi hin-ma ke and be ween-ma ke e u n spillo e s, ha s ock and sa e-ha en ma ke s show weak connec edness, and ha sa e-ha en asse s a e mino ecei e s and ansmi e s o be ween-ma ke spillo e e ec s. The connec edness be ween ma ke s is ime- a ying and e eals s uc u al changes in ea ly 2020. Fu he mo e, we documen ha inancial condi ions shape he dynamics o e u n spillo e e ec s be ween ma ke s. 1. In oduc ion In ecen yea s, digi al asse ma ke s ha e gained popula i y among in es o s; howe e , hose ma ke s a e ex emely ola ile, wi h unexpec ed up u ns o down u ns ha ha e impo an size e ec s on hei ma ke capi aliza ion ha make hem pa icula ly agile (Taleb, 2021). Along wi h c yp ocu encies, di e en kinds o in es able c yp o asse s ha e been in oduced in o he ma ke , ep e- sen ing new oppo uni ies o in es men in as -g owing echnology-backed asse classes. One such asse class gaining p ominence – as hey a e suppo ed by inancial se ices, including lending, bo owing, spo ading, online walle s, and de i a i es – is decen alized inance (DeFi) asse s, which a e aded pee - o-pee on he basis o blockchain echnology wi h no cen al au ho i y (Guba e a, 2021; Sch¨ a , 2021; Yousa e al., 2022). Such new asse s o e new oppo uni ies o in es men in e ms o bo h pe o mance and hedging abili ies. In his pape , we assess how DeFi asse s a e ela ed o o he asse classes, such as c yp ocu encies and he usual s ocks and sa e- ha en asse s. This in o ma ion is c ucial o in es o po olio and isk managemen decisions, mainly unde ex eme ma ke ci - cums ances when hedging and sa e-ha en ea u es o DeFi asse s could be pa icula ly use ul o p o ec ion agains downside isk. Recen e en s such as he global pandemic (COVID-19) and geopoli ical con lic ( he Russian-Uk aine mili a y con lic ), as well as soa ing ene gy cos s, deglobaliza ion, and s onge egula ion o c yp ocu encies ha e inc eased spillo e s among in e na ional ma ke s, making hedging di icul and cos ly (Ha and Nham, 2022; Mai a e al., 2022; Bossman e al., 2023). The e o e, s udying * Co esponding au ho . E-mail add ess: [email p o ec ed] (J.C. Rebo edo). Con en s lis s a ailable a ScienceDi ec Finance Resea ch Le e s jou nal homepage: www.else ie .com/loca e/ l h ps://doi.o g/10.1016/j. l.2023.103692 Recei ed 22 No embe 2022; Recei ed in e ised o m 3 Feb ua y 2023; Accep ed 8 Feb ua y 2023 Finance Resea ch Le e s 53 (2023) 103692 2 whe he and how DeFi asse s a e connec ed wi h o he inancial asse s in he ligh o ecen e en s p o ides aluable in o ma ion on he po en ial a ac i eness o DeFi asse s o hedge o sa e ha en in es men po olios. P e ious esea ch shows ha digi al asse s a e weakly co ela ed wi h inancial and commodi y ma ke s, and he ac ha hey can be used o hedge agains downside s ock ma ke mo emen s (see, e.g., Cao and Xie, 2022; Guesmi e al., 2019). Rega ding DeFi asse s, Yousa e al. (2022) poin s o s ong connec edness be ween DeFi and con en ional cu ency ma ke s, mainly in ea ly 2020. Yousa and Ya o aya (2022a) ind no e idence o he ding, excep o ime- a ying he ding o c yp ocu ency and DeFi asse s o e sho in- es men ho izons and o low- ola ili y days. In con as , Co be e al. (2021) show ha DeFi asse alues luc ua e independen ly o con en ional c yp ocu encies, p o iding hedging abili y o digi al in es o s. Simila ly, Yousa and Ya o aya (2022b) ind ha DeFi asse s a e decoupled om adi ional asse classes. Pi˜ nei o-Chousa e al. (2022) show ha DeFi okens se e as sa e-ha en asse agains s ock ma ke ola ili y. Ce ik e al. (2022) ind ha DeFis ha e he p ope y o sa e-ha en asse s o s a egic commodi y (c ude oil and gold) ma ke s. Uma e al. (2022) show an inc easing in e dependence be ween DeFi, NFT and inancial ma ke s mainly du ing he pandemic c isis. Finally, in analyzing p ice explosi eness in DeFi alues, Wang e al. (2022) epo ha DeFi p ices a e highly co ela ed wi h c yp ocu ency ma ke unce ain y. This pape con ibu es o exis ing esea ch by s udying spillo e s wi hin and be ween DeFí asse s (including BAT, Make , LINK, and SNX), c yp ocu encies (including Bi coin, E he eum, Te he , and BNB), con en ional s ocks (including ma ke s in Japan, US, UK, and Eu ope), and adi ional sa e-ha en asse s (including gold, he USD, and US T easu y bills). We use he block agg ega ion p ocedu e o G eenwood-Nimo e al. (2015, 2016), based, in u n, on he connec edness app oach o Diebold and Yilmaz (2014). Fo he u bulen 2018–2022 pe iod, ma ked by he COVID-19 pandemic, he c yp ocu ency p ice c ash, and he mili a y con lic in Uk aine, we ind ha all ma ke s a e mainly a ec ed by hei own shocks, and ha s ong spillo e s occu wi hin each asse class. We also ind s ong e idence o spillo e s be ween DeFi and c yp ocu ency ma ke s, bu weak e idence o spillo e s be ween sa e-ha en ma ke s and he emaining asse ma ke s. Finally, we documen ha inancial condi ions, as gi en by gold and s ock ma ke ola ili y, illiquidi y, c yp ocu ency ma ke ola ili y, e m sp ead, and economic policy unce ain ies all play a ele an ole in shaping he dynamics o ne spillo e s o he DeFi, c yp ocu ency, s ock, and sa e-ha en asse ma ke s. The pape is o ganized as ollows: Sec ion 2 discusses he me hodology and da a, Sec ion 3 discusses he esul s, and Sec ion 4 concludes he pape . 2. Me hods and da a 2.1. Modelling connec edness We s udy e u n spillo e connec edness be ween ou ma ke blocks: DeFi (d), c yp ocu ency (c), s ock (s), and sa e-ha en ( ) asse s, assuming ha asse e u ns in hose ma ke s a e endogenously de e mined by a ec o au o eg essi e (VAR) model wi h p lags: y = μ +∑ p k=1 Aky −k+ ε ,(1) whe e y =(y d ,y c ,y s ,y )′is a column ec o con aining j- ou column ec o s o e u ns in he ma ke j =d, c, s, ; μ is a column ec o o cons an s; and A is a (4 ×4) coe icien block ma ix, whe e each block A jj accoun s o eedback e ec s be ween asse e u ns in ma ke j, and A ji accoun s o eedback e ec s be ween asse e u ns in ma ke s j and i, wi h j, i =d, c, s, . ε is a s ochas ic column ec o wi h ze o mean and a (4 ×4) block a iance-co a iance ma ix Σ, wi h blocks Σ jj accoun ing o a iance-co a iance be ween j-ma ke e u ns, and Σ ji accoun ing o co a iances be ween e u ns in ma ke s j and i. F om he mo ing a e age (MA) ep esen a ion o p ice dynamics as pe Eq. (1) we ha e: y =∑ ∞ w=1 Bw μ +∑ ∞ w=1 Bw ε −w,(2) wi h he MA coe icien s B w =A 1 B w −1 +⋅ ⋅ ⋅ +A p B w −p , B w =0 o w <0, and B 0 equal o he iden i y ma ix. Re u n spillo e s be ween asse e u ns a e ob ained on he basis o he h-s ep ahead o ecas e o a iance decomposi ion (Pesa an and Shin, 1998) as: θh i←j= σ −1 jj ∑H−1 h=0(e′ iBwΣej)2 ∑H−1 h=0(e′ iBwΣB′ wei),(3) whe e σ jj is he j-diagonal elemen o Σ, and e i is a column ec o o ze os wi h 1 o i s i h componen . Hence, he alue o θh i←j depends on bo h he eedback e ec s be ween asse e u ns and he a iance-co a iance ma ix. Diebold and Yilmaz (2014) use he in o ma ion in Eq. (3) o build a connec edness ma ix, wi h θh i←j componen s a anged in ows epo ing in o ma ion on he impac o a shock om ma ke j o ma ke i (in ows), o om ma ke i o ma ke j wi h θh i←j componen s a anged in columns. By no malizing θh i←jin ows o sum 1 ( θh i←j), o al spillo e s o ma ke i a e compu ed as  θh i←⋅ =∑d j=1,j∕=i θh i←j, and o al spillo e s om ma ke i o di e en ma ke s a e compu ed as  θh ⋅←i=∑d j=1,j∕=i θh j←i. Hence,  θh i←i+ θh i←⋅ =100%. By ex ending he Diebold-Yilmaz app oach as in G eenwood-Nimmo e al. (2015, 2016), we can assess connec edness wi hin and A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 3 be ween ma ke blocks, as gi en by he block o m o he connec edness ma ix: ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ Θh d←dΘh d←c Θh c←dΘh c←c Θh d←sΘh d← Θh c←sΘh c← Θh s←dΘh s←c Θh ←dΘh ←c Θh s←sΘh s← Θh h←sΘh ← ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ,(4) whe e Θh j←j and Θh j←l include wi hin-block ma ke j spillo e s and spillo e s om ma ke block l o ma ke block j, espec i ely. Hence, we can easily ob ain o al spillo e s om di e en ma ke blocks o ma ke block j and also he e e se spillo e e ec s. F om es ima es o pa ame e ma ices o he VAR model, we can ob ain he bi a ia e and block ela ionship me ics as in Eq. (4), and e alua e s a is ical signi icance by Mon e Ca lo simula ion o he VAR model so as o build 95% con idence in e als. 2.2. Da a The da ase consis s o p ices o ou asse classes: (a) DeFi asse s (including BAT, Make , LINK, and SNX); (b) c yp ocu ency asse s (including Bi coin, E he eum, Te he , and BNB); (c) s ocks ( ep esen ed by he Nikkei 225 o Japan, Eu o S oxx 50 o Eu ope, FTSE 100 o he UK, and S&P 500 o he USA); and (d) sa e-ha en asse s (including gold, a ade-weigh ed index o he USD, and 3-mon h US T easu y bills). Da a we e sou ced om Bloombe g o daily pe iods om 1 Janua y 2018 (wi h he s a ing da e delimi ed by da a a ailabili y) o 18 Ma ch 2022. The sample pe iod includes se e al key economic and poli ical e en s such as he Bi coin p ice c ash (wi h a p ice all o 65% in Feb ua y 2018), he COVID-19 pandemic wi h i s di e en a ian s and wa es, he global all in oil demand, and he mili a y con lic in Uk aine. Table 1 summa izes he main s a is ical ea u es o he da a. A e age daily p ice e u ns a e nea ze o; DeFi and sa e-ha en asse s exhibi he highes and lowes ola ili ies, espec i ely; asse e u ns a e asymme ic and lep oku ic; no mali y is ejec ed; and all e u n se ies a e s a iona y a he 1% le el. 3. Empi ical e idence 3.1. Connec edness es ima es We compu ed e u n spillo e s as pe Eq. (3) om he es ima ed VAR model in Eq. (1) wi h 1 lag, selec ed acco ding o he Bayesian in o ma ion c i e ion (BIC). The e idence in Table 2 shows ha all ma ke s a e mainly in luenced by hei own shocks. Fo example, 31% o o al shocks o BAT come om hei own shocks, while BAT con ibu es 9.3%, 10.3%, and 8% o he o ecas ing a iance o Make , LINK, and SNX, espec i ely. I also con ibu es 33.5% o he o ecas ing a iance o c yp ocu ency, 9.1% o s ocks, and unde Table 1 Desc ip i e s a is ics. Mean S d. de . Max Min Skewness Ku osis JB ADF PP KPSS DeFi asse s BAT 0.001 0.073 0.303 -0.595 -0.486 9.472 1868.408* -9.557*** -35.237*** 0.064 Make 0.001 0.073 0.442 -0.881 -1.338 27.767 27,071.808* -9.840*** -36.994*** 0.094 LINK 0.003 0.078 0.479 -0.662 -0.520 11.129 2929.634* -9.877*** -33.327*** 0.105 SNX 0.002 0.096 0.543 -0.515 0.265 6.778 634.815* -8.058*** -35.623*** 0.251 C yp o asse s Bi coin 0.002 0.045 0.203 -0.465 -1.045 15.473 6977.370* -9.128*** -33.431*** 0.175 E he eum 0.001 0.060 0.354 -0.551 -0.829 12.761 4276.465* -9.238*** -34.273*** 0.450* Te he 0.000 0.004 0.053 -0.053 0.311 49.598 94,743.029* -13.026*** -61.450*** 0.007 BNB 0.004 0.064 0.529 -0.543 -0.193 14.959 6245.657* -8.700*** -32.404*** 0.165 S ocks Nikkei 225 0.000 0.012 0.077 -0.063 -0.089 7.534 898.057* -9.348*** -32.446*** 0.055 S&P 500 0.000 0.013 0.090 -0.128 -0.984 20.862 14,087.699* -8.935*** -39.974*** 0.071 FTSE 100 0.000 0.012 0.087 -0.115 -1.181 18.651 10,930.191* -9.188*** -33.628*** 0.071 Eu o S oxx 50 0.000 0.013 0.088 -0.132 -1.155 18.867 11,215.886* -9.273*** -33.715*** 0.045 Sa e-ha en asse s Gold 0.000 0.010 0.058 -0.051 -0.305 8.705 1436.229* -10.453*** -32.848*** 0.086 USDX 0.000 0.004 0.020 -0.015 0.265 4.973 182.079* -10.958*** -30.563*** 0.137** T-Bill 0.000 0.003 0.012 -0.015 -0.086 6.364 495.038* -9.286*** -33.449*** 0.628 No es: This able p esen s summa y s a is ics o he analyzed DeFi, c yp ocu ency, s ock, and sa e-ha en asse s. JB deno es he Ja que-Be a s a is ic, wi h he as e isk deno ing ejec ion o he null o no mali y a he 5% le el. ADF, PP, and KPSS deno e he augmen ed Dickey-Fulle es , Phillips- Pe on uni oo es , and he one-sided Kwia kowski-Phillips-Schmid -Shin s a iona i y es , espec i ely, wi h an as e isk indica ing ejec ion o he null hypo hesis. A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 4 Table 2 Connec edness ma ix be ween a iables. BAT Make LINK SNX Bi coin E he eum Te he BNB S&P 500 FTSE 100 Eu o S oxx 50 Nikkei 225 Gold USDX T-Bill BAT 31.06 9.36 9.91 5.75 11.82 14.03 0.54 11.07 1.95 1.48 2.10 0.50 0.10 0.24 0.10 [29.64, 31.50] [8.79, 9.78] [9.38, 10.40] [5.40, 6.28] [11.25, 12.15] [13.25, 14.26] [0.18, 1.72] [10.40, 11.46] [1.88, 2.84] [1.38, 2.90] [1.94, 3.48] [0.47, 1.56] [0.09, 0.70] [0.21, 1.17] [0.07, 1.10] Make 9.32 30.06 9.07 5.80 11.51 15.84 1.15 10.05 1.90 1.57 2.35 0.24 0.35 0.70 0.10 [8.68, 10.72] [27.16, 30.55] [8.26, 9.72] [5.34, 6.64] [10.40, 11.96] [14.57, 16.31] [0.51, 3.33] [9.14, 10.63] [1.65, 2.86] [1.24, 3.88] [1.86, 4.66] [0.12, 1.38] [0.17, 1.11] [0.32, 1.58] [0.02, 0.97] Link 10.30 9.66 32.00 7.05 10.91 14.33 0.53 9.71 1.42 1.44 2.06 0.11 0.11 0.36 0.02 [9.70, 10.99] [8.89, 10.17] [29.60, 32.43] [6.39, 7.60] [10.19, 11.56] [13.27, 14.75] [0.08, 1.75] [9.01, 10.06] [1.30, 2.81] [1.20, 3.25] [1.69, 4.16] [0.10, 0.83] [0.04, 0.80] [0.19, 1.72] [0.01, 0.90] SNX 8.05 8.50 9.33 42.66 8.69 11.81 0.59 6.92 0.65 0.61 1.18 0.05 0.39 0.55 0.04 [7.61, 9.04] [8.01, 9.16] [8.89, 10.47] [39.25, 43.14] [8.16, 9.15] [11.22, 12.37] [0.04, 2.23] [6.41, 7.60] [0.57, 1.55] [0.47, 1.58] [0.95, 2.08] [0.03, 0.84] [0.33, 1.56] [0.52, 2.03] [0.02, 1.10] Bi coin 10.55 10.54 9.52 5.66 27.74 18.43 0.34 12.09 1.21 1.19 1.67 0.12 0.36 0.53 0.05 [9.94, 10.83] [9.87, 10.65] [9.11, 10.36] [5.32, 6.09] [26.35, 28.03] [17.50, 18.61] [0.05, 1.56] [11.44, 12.55] [1.16, 2.02] [1.04, 2.86] [1.45, 3.44] [0.10, 0.63] [0.32, 0.78] [0.41, 1.27] [0.03, 0.79] E he eum 11.05 12.85 10.97 6.81 16.30 24.79 0.50 11.71 1.29 1.15 1.64 0.21 0.26 0.43 0.03 [10.64, 11.89] [12.17, 13.01] [10.51, 11.52] [6.45, 7.07] [15.46, 16.67] [23.61, 25.00] [0.13, 1.46] [11.12, 11.97] [1.20, 2.14] [1.03, 2.61] [1.41, 2.87] [0.17, 0.96] [0.23, 0.77] [0.31, 1.35] [0.03, 0.97] Te he 1.01 2.94 0.70 0.77 0.30 0.66 81.55 0.74 3.44 2.30 1.88 1.33 0.07 0.44 1.86 [0.38, 2.00] [0.56, 4.10] [0.13, 2.09] [0.03, 3.34] [0.06, 2.21] [0.30, 2.77] [72.22, 87.76] [0.47, 2.12] [1.68, 6.21] [1.48, 4.76] [1.29, 4.25] [0.33, 3.03] [0.02, 2.03] [0.05, 3.30] [1.22, 6.43] BNB 10.87 10.04 9.26 4.89 13.42 14.53 0.41 30.62 1.48 1.38 1.78 0.29 0.41 0.54 0.09 [9.98, 11.29] [9.10, 10.34] [8.55, 9.83] [4.41, 5.50] [12.41, 14.59] [13.29, 14.80] [0.24, 1.99] [27.31, 30.54] [1.33, 2.83] [1.27, 3.61] [1.60, 3.77] [0.17, 1.62] [0.31, 1.07] [0.25, 1.70] [0.02, 1.54] S&P 500 2.83 3.10 2.22 0.83 2.42 2.68 3.29 2.17 36.63 15.27 17.43 5.69 0.52 1.08 3.83 [2.36, 5.01] [2.14, 4.85] [1.63, 4.64] [0.51, 3.15] [1.64, 4.04] [2.14, 4.97] [1.28, 5.89] [1.63, 4.02] [31.46, 38.99] [12.87, 16.51] [14.68, 18.52] [4.75, 6.81] [0.03, 1.67] [0.64, 2.60] [3.04, 4.51] FTSE 100 1.76 1.25 1.62 0.68 1.55 1.85 1.40 1.37 13.46 35.98 27.61 6.41 0.24 0.53 4.29 [1.58, 2.75] [1.07, 2.09] [1.35, 3.25] [0.41, 1.80] [1.29, 1.97] [1.45, 2.81] [0.69, 2.35] [1.27, 2.06] [12.75, 14.35] [33.98, 36.78] [26.13, 28.32] [6.02, 7.28] [0.04, 0.98] [0.23, 2.17] [4.00, 4.98] Eu o S oxx 50 2.40 1.90 2.09 1.10 2.12 2.42 1.20 1.86 14.40 25.75 34.07 5.56 0.25 0.86 4.03 [2.14, 3.87] [1.51, 2.51] [1.73, 3.55] [0.77, 2.49] [1.66, 2.88] [1.92, 3.48] [0.57, 2.42] [1.63, 2.79] [13.62, 15.55] [23.98, 26.43] [31.79, 34.96] [5.19, 6.17] [0.01, 1.05] [0.55, 1.85] [3.73, 4.91] Nikkei 225 2.11 1.81 1.47 0.90 1.98 1.72 0.49 1.15 16.60 11.67 15.09 40.90 0.24 0.71 3.15 [1.25, 5.00] [1.16, 5.51] [0.72, 4.86] [0.10, 3.09] [0.56, 4.44] [0.82, 4.70] [0.14, 1.45] [0.49, 3.55] [11.52, 17.79] [8.90, 13.18] [12.49, 17.83] [31.33, 47.82] [0.05, 1.24] [0.18, 3.17] [1.92, 4.41] Gold 0.37 0.89 0.50 0.86 1.41 1.07 0.51 1.35 0.49 0.38 0.76 0.22 74.22 8.93 8.04 [0.22, 1.91] [0.45, 3.08] [0.11, 2.13] [0.57, 3.19] [0.98, 5.09] [0.71, 2.75] [0.04, 1.78] [0.76, 3.67] [0.10, 4.20] [0.07, 1.71] [0.02, 2.60] [0.10, 3.61] [62.74, 76.23] [7.64, 10.68] [6.53, 11.77] USDX 1.31 2.47 0.93 1.39 2.29 2.15 0.47 2.04 4.68 2.44 3.86 0.93 8.66 64.92 1.48 [0.67, 3.75] [1.22, 6.14] [0.55, 3.42] [0.90, 3.36] [1.43, 5.06] [1.37, 5.02] [0.07, 2.77] [1.28, 6.30] [1.93, 6.58] [0.76, 3.47] [1.61, 5.39] [0.22, 2.85] [6.56, 10.34] [50.41, 71.66] [1.22, 5.24] T-Bill 0.24 0.25 0.05 0.07 0.12 0.15 2.10 0.07 6.99 7.85 7.69 3.11 5.78 1.25 64.27 [0.17, 2.11] [0.12, 3.43] [0.03, 1.39] [0.03, 1.56] [0.10, 2.44] [0.08, 2.61] [0.91, 6.67] [0.05, 1.40] [6.11, 8.89] [6.85, 8.54] [6.76, 8.66] [2.74, 4.26] [4.97, 7.60] [0.94, 3.25] [54.82, 63.81] No es: The able epo s esul s on spillo e s among he DeFi, c yp ocu ency, s ock, and sa e-ha en asse s indica ed in he i s ow and i s column using a o ecas ho izon o h =10 ading days o he (no malized) spillo e me ic in Eq. (3). Repo ed o he ma ke in each ow is he o ecas e o a iance ha is explained by he ma ke s indica ed in he columns, wi h alues adding 100%. Repo ed o he ma ke in each column is he con ibu ion o he a iance o ha ma ke o he ma ke s indica ed in he ows. 95% con idence in e als epo ed in squa ed b acke s a e compu ed using 10,000 Mon e Ca lo simula ions o he educed- o m VAR model. A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 5 2% o sa e-ha en asse s. Two-way spillo e s o he o he asse classes also e eal ha own shocks a e mo e ele an han shocks om o he asse s. Table 3 con i ms ha ma ke s a e la gely a ec ed by hei own shocks, e.g., s ock and sa e-ha en ma ke s. In con as , DeFi ma ke s show a lowe own in luence, ecei ing a g ea e impac om o he ma ke s (mainly om c yp ocu ency) and con ibu ing mos shocks (29.6%) o he c yp ocu ency ma ke . O e all, DeFi and he c yp ocu ency ma ke s a e closely in e - connec ed, and also ela i ely disconnec ed om bo h s ock ma ke s, as epo ed by Yousa and Ya o aya (2022b), and sa e-ha en ma ke s; he la e , in u n, show weak connec edness, as would be expec ed om hei sa e-ha en na u e, a esul ha is consis- en wi h Ce ik e al. (2022). Fig. 1 plo s connec edness and he size and di ec ion o spillo e s wi hin and be ween DeFi, c yp ocu ency, s ock and sa e-ha en asse ma ke s. DeFi, c yp ocu ency, and s ock ma ke s a e ne con ibu o s o spillo e s, whe eas sa e-ha en asse s a e a ne ecei e o spillo e s. A s onge bidi ec ional spillo e is obse ed be ween DeFi and c yp ocu ency ma ke s, e lec ing g ea e in eg a ion be ween hose ma ke s – consis en wi h he e idence epo ed by Yousa e al. (2022) bu a odds wi h ha epo ed by Co be e al. (2021). Sa e-ha en asse s ecei e spillo e s om all ma ke s, ansmi ing some isk o s ock ma ke s and negligible isk o he DeFi and c yp ocu ency ma ke s. S ock ma ke s show bidi ec ional connec edness wi h all ma ke s. Wi hin DeFi asse s, BAT, Make , and LINK a e ne con ibu o o spillo e s, whe eas SNX is a ne ecei e . In he c yp ocu ency ma ke , he connec edness ne wo k shows signi ican bidi ec ional linkages be ween Bi coin and E he eum, consis en wi h he indings o Beneki e al. (2019). BNB is connec ed wi h bo h Bi coin and E he eum, while Te he , in e es ingly, is disconnec ed om he h ee emaining c yp ocu encies, poin ing o po en ial implica ions o hedging agains downwa d c yp ocu ency p ice mo emen s. Fo s ock ma ke s, he e is high connec edness, wi h he Japanese s ock ma ke a ne ecei e o spillo e s om o he s ock ma ke s. In con as , sa e-ha en asse s a e weakly con- nec ed, sugges ing possible di e si ica ion e ec s be ween gold, T easu y bills, and he USD index. O e all, equi y in es o s may conside DeFi asse s, c yp ocu encies, gold, T easu y bills, and he USD index o hedge hei posi ions agains s ock p ice downwa d mo emen s. 3.2. Spillo e dynamics and inancial condi ions We explo e whe he spillo e s change o e he sample pe iod by es ima ing hose spillo e s o a daily olling window o 220 ading days, which allows eedback e ec s and a iance-co a iance ma ix o swing o e he sample pe iod, and hus connec edness alues as pe Eq (4). The g aphical e idence in Fig. 2 shows ha spillo e s ose du ing he i s COVID-19 wa e (Ma ch-Ap il 2020) and om ea ly 2021: (a) spillo e s om DeFi asse s o c yp ocu encies and ice e sa exhibi ed he same pa e ns, anging om 22% in Decembe 2020 o abo e 35% in Ma ch 2022, he la e e lec ing he mili a y con lic in Uk aine; (b) spillo e s be ween sa e-ha en asse s and DeFi and c yp ocu ency asse s we e smoo he han hose be ween DeFi asse s and s ock ma ke s; (c) be ween Janua y 2020 and Janua y 2021, spillo e s be ween s ocks and c yp ocu ency and DeFi asse s inc eased, and dec eased o sa e-ha en asse s, unde lying he impo ance o adding sa e-ha en asse s o equi y-DeFi o equi y-c yp ocu ency po olios; and inally, (d) spillo e s om sa e-ha en asse s o he o he ma ke s we e lowe han hose om DeFi asse s, c yp ocu encies, and s ocks o sa e-ha en asse s. We examine whe he ime- a ying ne spillo e s a e shaped by inancial ma ke condi ions. Pa icula ly, we conside : (a) un- ce ain y as gi en by ola ili y in he gold and s ock ma ke s (CBOE gold and VIX indices); (b) illiquidi y in he in e bank ma ke as gi en by he TED sp ead (3-mon h LIBOR based on he USD minus he 3-mon h T easu y yield); (c) T easu y sp ead (US go e nmen 10-yea yield minus US go e nmen 3-mon h yield); (d) c yp ocu ency ma ke ola ili y (see Wang e al., 2022) as gi en by he VCRIX index (Royal on VCRIX C yp o Index; see Kim e al., 2021); and (e) he Economic Policy Unce ain y (EPU) index. Table 4 epo s es ima ed impac s o six con ol a iables on ne spillo e s o each o he ou asse classes. Gold ola ili y has a nega i e and signi ican impac on ne spillo e s o DeFi, c yp ocu ency, and s ock ma ke s, implying ha an inc ease in gold un- ce ain y educes ne spillo e s in hose ma ke s. In con as , e ec s a e posi i e o sa e-ha en asse s. VIX has no impac on ne spillo e s o c yp ocu ency and sa e-ha en asse s, bu does posi i ely in luence ne spillo e s in DeFi ma ke s. Likewise, he impac o VIX on ne spillo e s in he s ock ma ke s is nega i e, implying ha a ise in VIX educes ne spillo e s in s ock ma ke s. As o he illiquidi y impac , TED posi i ely a ec s ne spillo e s in all ma ke s, wi h he excep ion o sa e-ha en asse s whe e he sign is Table 3 Connec edness ma ix o DeFi, c yp ocu ency, s ock, and sa e-ha en asse ma ke s. DeFi C yp o S ocks Sa e-ha en DeFi 59.47 34.88 4.90 0.77 (55.25, 62.14) (32.02, 37.82) (4.21, 10.16) (0.50, 3.68) C yp o 29.61 63.53 5.59 1.27 (26.78, 32.48) (57.99, 68.16) (4.17, 11.90) (0.79, 5.50) S ocks 7.02 7.42 80.63 4.93 (5.11, 14.60) (4.79, 13.46) (70.36, 86.82) (3.60, 8.39) Sa e-ha en 3.11 4.58 13.13 79.18 (1.67, 11.82) (2.59, 15.18) (9.08, 20.26) (65.27, 86.86) No es. This able p esen s e idence o ma ke connec edness among DeFi, c yp ocu ency, s ock, and sa e-ha en ma ke s using a o ecas ho izon o h =10 ading days and he block agg ega ion p ocedu e o G eenwood-Nimmo e al. (2015). Repo ed o he ma ke in each ow is he ac ion o he o ecas e o a iance ha is explained by he ma ke s indica ed in he columns, wi h alues adding 100%. Repo ed o he ma ke in each column is he con ibu ion o each ma ke indica ed in he ows. 95% con idence in e als epo ed in ound b acke s a e compu ed using 10,000 Mon e Ca lo simula ions o he educed- o m VAR model. A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 6 nega i e. The T easu y sp ead impac s nega i ely on spillo e s o bo h he DeFi and sa e-ha en asse ma ke s, and posi i ely o c yp ocu ency and s ock ma ke s. The VCRIX con ibu es posi i ely o ne spillo e s o DeFi, s ock, and c yp ocu ency ma ke s, and nega i ely o ne spillo e s o sa e-ha en asse ma ke s. Finally, a ise in EPU educes spillo e s in all ma ke s, excep o sa e-ha en asse s. O e all, hose esul s highligh he ele ance o inancial ma ke condi ions in shaping spillo e s ac oss ma ke s, a inding ha is consis en wi h Rebo edo e al. (2021) o commodi y ma ke s. 4. Conclusions This s udy examines spillo e s be ween ou ma ke blocks, namely DeFi (BAT, Make , LINK, and SNX), c yp ocu encies (Bi coin, E he eum, Te he , and BNB), s ock ma ke s (Japan, US, UK, and Eu ope), and sa e-ha en asse s (gold, USD index, and US T easu y bills) using he Diebold and Yilmaz (2014) spillo e index and he G eenwood-Nimmo e al. (2015, 2016) me hodology. Addi ionally in es iga ed is he impac o ce ain inancial condi ions on he size o spillo e s, including implied gold ola ili y, equi y ma ke unce ain y (VIX), TED sp ead, T easu y sp ead, he Royal on VCRIX C yp o Index, and he EPU index. Ou esul s e eal ha all ma ke s a e mainly a ec ed by hei own shocks. O all he asse classes, DeFi asse s and c yp ocu encies exhibi he highes spillo e s, while he sa e-ha en asse s a e hose leas connec ed wi h o he asse s. Spillo e s among ma ke s a e dynamic and each hei highes le el be ween ea ly 2020 and ea ly 2021. Finally, he gold ola ili y index, VIX, TED, T easu y sp ead, Royal on VCRIX C yp o Index, and EPU index all impac on ne spillo e size wi hin each asse class. These indings a e help ul o in es o s and po olio manage s and ha e implica ions o he design o policies. CRediT au ho ship con ibu ion s a emen And ea Ugolini: Me hodology, So wa e, Da a cu a ion, W i ing – o iginal d a , W i ing – e iew & edi ing. Juan C. Rebo edo: Concep ualiza ion, Me hodology, W i ing – o iginal d a , Fo mal analysis, W i ing – e iew & edi ing, Funding acquisi ion. Walid Mensi: Concep ualiza ion, W i ing – o iginal d a , W i ing – e iew & edi ing. Fig. 1. Connec edness wi hin and be ween DeFi, c yp ocu ency, s ock, and sa e-ha en asse ma ke s. A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 7 Decla a ion o Compe ing In e es The e is no con lic o in e es among au ho s. Da a a ailabili y Da a will be made a ailable on eques . Fig. 2. Connec edness dynamics o DeFi, c yp ocu ency, s ock, and sa e-ha en asse ma ke s. Table 4 Financial ma ke condi ions and ne spillo e s in he DeFi, c yp ocu ency, s ock, and sa e-ha en asse ma ke s. DeFi C yp o S ocks Sa e-ha en β 0 4.553*** -8.059*** 4.245*** -0.739 (0.615) (0.691) (0.502) (1.353) Gold Vol -0.495*** -0.628*** -0.415*** 1.538*** (0.054) (0.06) (0.044) (0.118) VIX 0.114*** 0.051 -0.138*** -0.027 (0.921) (1.035) (0.752) (2.025) TED 14.101*** 10.084*** 1.294* -25.479*** (0.921) (1.035) (0.752) (2.025) TS -4.248*** 3.809*** 2.840*** -2.400*** (0.244) (0.275) (0.199) (0.538) VCRIX 0.003*** 0.008*** 0.003*** -0.013*** (0.001) (0.001) (0.001) (0.001) EPU -0.027*** -0.031*** -0.012*** 0.069*** (0.002) (0.002) (0.001) (0.004) Adj. R 2 0.647 0.622 0.656 0.694 No es. This able p esen s e idence on he e ec s o inancial ma ke condi ions as gi en by he ola ili y o gold, s ocks, c yp ocu encies, illiquidi y, T easu y sp ead, and economic policy unce ain y on ne spillo e s in he DeFi, c yp ocu ency, s ocks, and sa e-ha en asse ma ke s. T-s a is ics a e epo ed in ound b acke s. *, ** and *** indica e signi icance a he 10%, 5%, and 1% le els, espec i ely. A. Ugolini e al. Finance Resea ch Le e s 53 (2023) 103692 8 Acknowledgemen s We would like o hank wo anonymous e iewe s o cons uc i e and insigh ul commen s. Juan C. Rebo edo acknowledges inancial suppo p o ided by he Agencia Es a al de In es igaci´ on (Minis e io de Ciencia, Inno aci´ on y Uni e sidades) unde esea ch p ojec wi h e e ence PID2021–124336OB-I00 co- unded by he Eu opean Regional De elopmen Fund (ERDF/FEDER). Re e ences Beneki, C., Koulis, A., Ky iazis, N., Papadamou, S., 2019. In es iga ing ola ili y ansmission and hedging p ope ies be ween Bi coin and E he eum. Res. In . Bus. Financ. 48, 219–227. Bossman, A., Guba e a, M., Teplo a, T., 2023. Asymme ic e ec s o geopoli ical isk on majo cu encies: ussia-Uk aine ensions. Financ. Res. Le . 51, 103440. Cao, G., Xie, W., 2022. Asymme ic dynamic spillo e e ec be ween c yp ocu ency and China’s inancial ma ke : e idence om TVP-VAR based connec edness app oach. Financ. Res. Le . 49, 103070. Ce ik, E., Gunay, S., Za a , M., Des ek, M., Bugan, M., Tuna, F., 2022. The impac o digi al inance on he na u al esou ce ma ke : e idence om DeFi, oil, and gold. Resou . Policy 79, 103081. Co be , S., Goodell, J.W., Gunay, S., Kaskaloglu, K., 2021. DeFi okens a sepa a e asse class om con en ional c yp ocu encies? A ailable a SSRN: h ps://ss n.com/ abs ac =3810599. Diebold, F.X., Yilmaz, K., 2014. On he ne wo k opology o a iance decomposi ions: measu ing he connec edness o inancial i ms. J. Econom. 182, 119–134. G eenwood-Nimmo, M., Nguyen, V.H., Ra e y, B., 2016. Risk and e u n spillo e s among he G10 cu encies. J. Financ. Ma k. 31, 43–62. G eenwood-Nimmo, M.J., Nguyen, V.H., Shin, Y., 2015. Measu ing he Connec edness o he Global Economy. Uni e si y o Melbou ne, Mimeo. Wo king Pape . Guba e a, M., 2021. Lowe e e sal limi o he Eu opean Cen al Bank deposi a e and sus ainabili y o adi ional banking business model. J. Financ. Econ. Policy (2021). h ps://doi.o g/10.1108/JFEP-07-2020-0151. Guesmi, K., Saadi, S., Abid, I., 2019. Po olio di e si ica ion wi h i ual cu ency: e idence om bi coin. In . Re . Financ. Anal. 63, 431–437. Ha, L., Nham, N., 2022. An applica ion o a TVP-VAR ex ended join connec ed app oach o explo e connec edness be ween WTI c ude oil, gold, s ock and c yp ocu encies du ing he COVID-19 heal h c isis. Technol. Fo ecas . Soc. Change 183, 121909. Kim, A., T imbo n, S., H¨ a dle, W.K., 2021. VCRIX - a ola ili y index o c yp o-cu encies. A ailable a SSRN: h ps://ss n.com/abs ac =3480348. Mai a, D., Rehman, M., Dash, S., Kang, S.H., 2022. Do c yp ocu encies p o ide be e hedging? E idence om majo equi y ma ke s du ing COVID-19 pandemic. The No h Am. J. Econ. Financ. 62, 101776. Pesa an, H.H., Shin, Y., 1998. Gene alized impulse esponse analysis in linea mul i a ia e models. Econ. Le . 58, 17–29. Pi˜ nei o-Chousa, J., L´ opez-Caba cos, M., Se ic, A., Gonz´ alez-L´ opez, I., 2022. A p elimina y assessmen o he pe o mance o DeFi c yp ocu encies in ela ion o o he inancial asse s, ola ili y, and use -gene a ed con en . Technol. Fo ecas . Soc. Change 181, 121740. Rebo edo, J.C., Ugolini, A., He nandez, J.A., 2021. Dynamic spillo e s and ne wo k s uc u e among commodi y, cu ency, and s ock ma ke s. Resou . Policy 74, 102266. Sch¨ a , F., 2021. Decen alized inance: on Blockchain- and sma con ac -based inancial ma ke s. Fede al Rese e Bank o S . Louis Re . 103 (2), 153–174. Taleb, N.N., 2021. Bi coin, cu encies, and agili y. Quan i a i e Financ. 21 (8), 1249–1255. Uma , Z., Pola , O., Choi, S., Teplo a, T., 2022. Dynamic connec edness be ween non- ungible okens, decen alized inance, and con en ional inancial asse s in a ime- equency amewo k. Paci ic-Basin Financ. J., 101876 Wang, Y., Ho ky, F., Baals, L.J., Lucey, B.M., Vigne, S.A., 2022. Bubbles all he way down? De ec ing and da e-s amping bubble beha iou s in NFT and DeFi ma ke s. J. Chinese Econ. Bus. S ud. 20 (4), 415–436. Yousa , I., Nekhili, R., Guba e a, M., 2022. Linkages be ween DeFi asse s and con en ional cu encies: e idence om he COVID-19 pandemic. In . Re . Financ. Anal. 81, 102082. Yousa , I., Ya o aya, L., 2022a. He ding beha io in con en ional c yp ocu ency ma ke , non- ungible okens, and DeFi asse s. Financ. Res. Le . 50, 103299. Yousa , I., Ya o aya, L., 2022b. S a ic and dynamic connec edness be ween NFTs, De i and o he asse s: po olio implica ion. Glob. Financ. J. 53, 100719. A. Ugolini e al.