21 We measure the monetary policy stance using the shadow federal funds rate developed by Wu and Xia (2016), as it reects that balance sheet policy is now part of the conventional tool kit of modern central banking. If we only used the federal funds rate, we would omit relevant information. This is especially the case given our recent sample period, with the primary response to the COVID shock occurring through balance sheet policies. In our specications, beyond the variables related to equity and crypto prices, we account for a set of variables that proxy for global economic activity. Specically, we include: (i) the spread between tenand two-year yields on US government bonds, reecting investors expectations of future economic growth; (ii) the dollar index, to proxy for the status of international trade and credit owswhich the literature has shown to be cyclical (e.g.,

21In addition, in Table A.5 in Appendix A, we estimate a simple monthly regression of the crypto factor and bitcoin on the Wu and Xia (2016) shadow rate and on monetary policy shocks from Bu et al. (2021). For the latter, we use latest updated series, which includes data up to end-2021. Although the sample size is very small, we still nd a signicant negative eect of US monetary policy on the crypto cycle. 24 Bruno and Shin, 2022); (iii) oil and gold prices, as they are usually associated with the economic cycle; and (iv) the VIX to capture anticipated future uncertainty and eective risk-aversion. Table 6: VAR specications Variable Ordering (1) (2) (3) (4) (5) Interest Rates Wu-Xia Shadow FFR Average S-FFR (BOE,ECB,Fed) X X X X X Conjuncture 10-2 Y Treasury Yield Spread Dollar Index VIX Oil Gold X X X X X X X X X X X X X X X X X X X X X X X X Equity Variables Aggregate Equity Risk Aversion S&P500 Global Equity Factor X X X X X X

Crypto Factors Aggregate Crypto Risk Aversion Bitcoin Crypto Factor First Generation Factor Smart Contracts Factor DeFi Factor Metaverse Factor IoT Factor X X X X X X X X X

X Notes: This table shows the selection and ordering of variables in each of our VAR specications. Column (1) is our baseline specication. Column (2) tests whether the baseline results are determined by the construction of the crypto and equity factors. Column (3) explores if the crypto factor is aected by other (major) monetary policies. Column (4) investigates the heterogeneous eects of the responses by crypto sub-classes. Finally, column (5) tests whether US monetary policy aects the crypto factor via the risktaking channel (as in Miranda-Agrippino and Rey, 2020). Data is from January 2018 to March 202, with the exception of column (4) which is from 2021 due to data availability. 25 Figure 8 reports the most relevant cumulative impulse response functions for the rst specication in Table 6. Overall, the signs of the responses are consistent with the literature. A Fed monetary contraction leads to an increase in the VIX and to a decline in the