Contributed by Devashish Sharma, email@example.com
Regulations against cryptocurrencies have globally increased since 2018. Existing evidence suggests that geopolitical events, the Covid-19 pandemic and changes in monetary policy are amongst the most significant factors affecting overall price movements for cryptocurrencies. Econometric tests conducted in the blog find a relation between cryptocurrency prices and prices of other equity assets but not between crypto prices and crypto specific regulation. The tests also show that cryptocurrencies behave like equity. There are some considerations related to this analysis which are not covered in this blog and will be followed up later with a detailed analysis. They are all addressed in Appendix 4.
Survey of existing literature
Governments across the world have viewed cryptocurrencies and their impact on traditional financial systems with suspicion. Since 2018, many governments have announced policies to check the misuse of cryptocurrencies. The crash in cryptocurrency prices and collapse of exchanges have reinforced regulatory concerns. Research, however, finds little evidence of policies and regulatory events affecting the price of cryptocurrencies. Feinstein and Werbach (2021) have surveyed a number of policies during the period 2013 to 2019 and concluded that policy regulations do not have lasting effects. Auer and Claessens (2018) also found evidence of a reactionary and unsustained change in returns (1.52 per cent increase in prices and 3.13 per cent decline in volume in a period of 24 hours after a regulatory change). Shanaev et al (2019) examine movement of prices for 120 different policy events. The results again are mixed with only anti-money laundering policies seen to have an effect.
Five episode breaks across crypto
Regulatory action on cryptocurrencies has significantly increased since the time covered in existing literature. This blog analyses the closing prices of the three most popular cryptocurrencies namely Bitcoin, Ethereum and XRP from May 2019 to September 20221. The average price of these cryptocurrencies vary significantly. The average price since May 2022 until February 2023 hovered around USD 24675.33 for Bitcoins, USD 1655 for Ethereum and USD 0.39 for XRP.
The time series data are presented in Figure 1. A structural break-test analysis2, shows partitions of five segments across four break dates, represented by yellow vertical lines in each of the graphs. The break dates and segmentation are fairly uniform across the three cryptocurrencies. While segments 1 and 2 show a flat trend, 3 and 4 are periods of high growth and 5 is the period of decline. For XRP, most of the growth is in a narrower period over segment 4. Average price levels increased by 2.8, 7.9 and 1.5 times for Bitcoin, Ethereum and XRP, respectively between period 1 and period 5. Appendix 1 provides details of the break dates and average price.
The time series for total volume traded follows a slightly different pattern. Growth is steepest in the second segment and volumes are much more volatile with several peaks, instead of the clear humps observed in the case of prices.
The major policy regulation3 taken into account for econometric tests do not coincide with the break dates in price changes in countries that see active crypto trading. Fifteen such regulations have been taken into account. All changes considered are in the form of acts, regulations and bans. Appendix 2 provides a more detailed explanation of the types of policies taken into account (summary of policies). This is not an exhaustive list of policy changes, but it is indicative of strict regulations that can potentially affect crypto trading. In line with findings from the literature, we observe no significant impact of crypto specific regulatory/policy events on crypto prices (refer to methodology section in Appendix 3 for elaboration).
Figure 1: Price Movements for Bitcoin, Ethereum and XRP
Figure 2: Total Volume of Cryptocurrency Traded (USD Billion)
Traditional policy regulations work better than crypto specific regulations
The price breaks registered from segments 4 to 5 can be explained by China’s outright ban on crypto, global inflation, increase in interest rates, the effects of the Russo-Ukraine war, etc.4 There is evidence to show that geopolitical fluctuations and monetary policies determine the concurrent movement of cryptocurrency prices. In a study on prices of Bitcoins, Aysan et al(2018) conclude that geopolitical and economic policy uncertainty indices have a significant impact on the same. Kyriazis (2020) also concludes that geopolitical fluctuations cause major changes in returns of cryptocurrencies. Corbet et al (2017) conclude that cryptocurrencies are susceptible to changes in interest rates. They add that economic factors affect cryptocurrencies as much as they affect fiat currency.
It is noticeable that cryptocurrency price movements are not affected much by crypto specific regulations. Literature suggests that the traditional tools of monetary policy work as effectively for cryptocurrencies as well. More importantly, the econometric tests in this blog show a significant positive impact of equity on crypto prices. Cryptocurrencies exhibit the properties of equity which make them susceptible to similar fluctuations. The similarity between these two assets explains why traditional policy tools affect cryptocurrencies more than crypto specific policies.
Auer, R., & Claessens, S. (2020). Cryptocurrency Market Reactions to Regulatory News. Federal Reserve Bank of Dallas, Globalization Institute Working Paper 381, 2-16.
Aysan, A. F., Demir, E., Gozgor, G., & Lau, C. K. (2018). Effects of the Geopolitical Risks on Bitcoin Returns and Volatility. Research in International Business and Finance, Volume 47(C), 511-518.
Bai, J., & Perron, P. (2003). Computation and Analysis of Multiple Structural Breaks. Journal of Applied Econometrics,18(1), 1-22.
Corbet, S., McHugh, G., & Meegan, A. (2017). The Influence of Central Bank Monetary Policy announcements on Cryptocurrency return volatility. Investment Management and Financial Innovations, Volume 14 (4), 60-72 .
Feinstein, B. D., & Werbach, K. (2021). The Impact of Cryptocurrency Regulation on trading Markets. Journal of Financial Regulation, 7(1), 48–99.
Kyriazis, Ν. (2021). The efects of geopolitical uncertainty on cryptocurrencies and other fnancial assets. SN Business and Economics, Volume 1 (1), 1-14.
Shanaev, S., Sharma, S., Ghimire, B., & Shuraeva, A. (2019). Taming the Blockchain Beast? Regulatory Implications for the Cryptocurrency Market. Research in International Business and Finance, Volume 51 (C), 1-23.
Summary of structural break tests
Summary of policies
We use a generalised least squares MLE approach to model three equations of cryptocurrency prices. The primary variable of concern is the dummy variable. We assign dummy values for each month that policy (given in the table in Appendix 2) was implemented. The model estimated is of a monthly frequency for the period of March 2017 to November 2022. This was the period in which there were large variations in cryptocurrency prices. Given the fairly recent rise in the popularity of cryptocurrencies, most of the policies implemented globally also lie within this time frame. The data specifics are as follows.
1. msci respresents the performance of selected global stock markets. An increase in the index value implies better stock market performance in selected countries
2. gpr refers to the geopolitical risk index. An increase in the value of the index indicates an increase in global geopolitical risks.
3. epu refers to economic policy uncertainty. An increase in the value of the index indicates an increase in global economic uncertainties.
4. APSP crude oil index refers to the benchmark prices of oil globally. An increase in the value of the index would indicates an increase in global prices of crude oil
Our objective is to analyse if policy regulations have an impact on the considered cryptocurrency prices along with a set of control variables. Various aspects such as geopolitical risks, policy uncertainties, global stock market performance and crude oil prices are considered in the literature for modelling. Such variables can be expected to affect the prices of cryptocurrencies. We estimate the following equations separately.
We include two autoregressive components till order 2 to avoid endogeneity/ spurious results. These components are denoted by ∅. For the stability and stationarity of the model, the condition is that │∅│< 1. All the three models show stable results for autoregressive components ∅ which implies that the gls estimation is also stable. The β’s denote the coefficients of each variable while D denotes the coefficient of the policy implementation dummy variable. are the error components for the corresponding equations. The results show that none of the policies resulted in a change in prices. While the impact of implementing crypto regulatory policies is insignificant, there were some interesting results.
The above table gives the coefficients of the GLS estimation. It can be seen that the policy dummy is statistically insignificant, indicating that policy interventions had no effects globally. The growth rates of the prices of Bitcoins and Ethereum show a positive significant relation with the MSCI. This indicates that if global stock markets perform better, then it effectively increases the growth rates of changes in the prices of the two dominant cryptocurrencies. This is not true for XRP. This can ascribed to the stable nature of the currency because it is pegged to the dollar.
Some methodological issues need to be addressed. There is abundant data available for cryptocurrencies on a daily basis. This is not true for other indicators. To build a comparable and consistent econometric model across the three cryptocurrencies, we use monthly data with 70 observations. A GLS estimation using MLE has been employed. GLS controls for heteroscedasticity, resulting in coefficients with less variance. The estimation is as follows.
where corresponds to the coefficient estimates. x is an NxK dimensional matrix (for N observations and K parameters) of the explanatory variables and y is the dependent variable of dimensions Nx1. Ω here is the variance covariance matrix of errors (Ω=ee’) accounting for shifts in heteroscedasticity.
There are some methodological issues that need to be addressed. Figure 3 shows the plots of the fitted lines from the model and actual lines for each equations (Bitcoin, Ethereum and XRP). Of the panels shown below, the lower panel shows that linear regression fitted values while the upper panels show a polynomial local regression (loess). It can be seen that while there is some degree of correlation between the actual and fitted lines, it is not very strong for any of the equations. In fact, a more non linear fit on the top panel is a more accurate fit. This implies that there is some degree of non linearity that needs to be accounted for in the modelling approach. Such issues could be tackled by including more data and variables by utilising other methods and models. The specifics of a more detailed modelling can be explored in a more extensive study.
The caveats listed below lie beyond the scope of this blog. While these problems need to be addressed, it would require a more rigorous methodological/econometric approach.
⦾ An important point of discussion is the nature of causation of policy. The direction of effects between crypto price fluctuations and policy – whether there is a unidirectional relationship, a simultaneity or reverse causality must be determined. This would reveal more information that policy makers can work upon.
⦾ It is suspected that crypto prices share a long term relation with USD. Changes in the strength of the dollar causes changes in crypto-currency prices. Inferences from a study on cointegration would provide policy implications. Furthermore, one can also analyse different kinds of volatility. This again has not been explored in the blog.
⦾ There was no global crackdown on crypto-currencies before 2017. It was only after 2017 that policy interventions started globally. This makes it a little harder to model the effects of policy on crypto-currencies.
⦾ While data on cryptocurrency prices are available on a daily basis, the data for covariates are not available with as much frequency. This makes it harder to fit models for more robust inferences. More importantly, Covid seems to have largely inflated the value of crypto-currencies. This has caused larger deviations from expected trends for the considered cryptocurrencies. The inherent trends for each currency have not been captured. It is only the average prices that have shifted pre and post Covid. Hence, it is thus equally important to look at more natural trends of such currencies as opposed to the ones fuelled by Covid.
1. Daily data for structural break test taken from coinmarketcap.com. https://coinmarketcap.com/.
2. Bai Perron (2003) break test.
3. Refer to table in appendix 2
4. Rise and fall of cryptocurrencies; looking at the cryptocurrency crash in 2022. https://www.financialexpress.com/blockchain/rise-and-fall-of-cryptocurrencies-looking-at-the-cryptocurrency-crash-in-2022/2628151/.
5. Average price before Covid (May 2019- July 2020)
6. Average price after Covid (March 2022-February 2023)
Please provide your contact details to hear from us on IPCIDE's research.