3. As an investment
3.2.3 Determining the regulations that are typically employed by regulators, in pursuit of the policy goals
Following the determination of the policy goals for each use-case it was necessary to evaluate what regulation is currently employed for assets in each of the three use-cases. As this paper seeks to obtain estimates on the average effect that different regulations have at achieving policy goals, it was important to maximise the breadth of data included in the regressions. In pursuit of this, the 50 largest economies as of 2018 were chosen for evaluation in this
research.
The initial plan for curating a list of potential regulations was to individually select
regulations based on existing literature. During the data gathering stage however, it became apparent that the data reporting standards were not consistent between countries when trying to cultivate this list of regulations, even within the 50 largest countries by GDP. Furthermore, it was impossible to determine whether independently enforced regulations were proportional across countries.
A solution to this for use-cases 1 and 2 was found in the Basel III banking regulation framework. The Basel committee, through the Basel III monitoring report, provided data on 27 regulations related to banking and currency, sorted by country. In implementing the Basel III banking regulatory guidelines, countries aim to execute consistent regulations such as capital requirements for equity investments in funds, monitoring tools for intraday liquidity management, and net stable funding ratios. The monitoring report provided data on the year that countries had issued rules, or whether a country was yet to make progress towards a regulation’s implementation. Furthermore, where countries already had similar regulations in place, the Basel committee had already assessed the existing regulations to determine
whether they were similar enough to the Basel regulations to be considered sufficient. This negated the issue of gauging similarity between countries’ regulatory approaches that was initially faced when cultivating a list of appropriate regulations.
Nonetheless, the downside to taking this approach is that the list of regulations included in the models is still not exhaustive, relying solely on the regulations included in the pre-set Basel III framework. This poses a threat of omitted variable bias in the models; it is difficult to account for all potential regulations and factors that could influence the policy goals,
especially when it is known that there are additional regulations that could not be feasibly included in the model but can be expected to influence the policy goals. The challenge in finding consistent regulation across countries was further compounded by the fact that the method of regression required a complete dataset for all variables due to using panel data. A possible argument to dispute the above is that the Basel III framework is designed to be as exhaustive as possible. By considering all of the regulations included in the framework in the model, it was possible to provide a best estimate on the effects of the different, relevant, regulations on policy goals. Whilst some regulations may not be accounted for, those included in the model should provide an almost exhaustive list.
These limitations in relying solely on pre-set frameworks such as Basel III will be evaluated further in the results and discussion sections.
Full regulations included in each regression:
It was mentioned above that the central bank is responsible for the regulation of currency and that the implication of this for this paper is that the central bank is the regulatory body in question for both use-cases 1 and 2. Consequently, the same regulations are applicable to both use-cases 1 and 2, with one exception. The policy goals that were evaluated in use-case 1; ‘Liquidity of reserves’ and ‘Return on reserves’, as well as the first policy goal in use-case 2; ‘Inflation’, had panel data dependent variables. However, the second policy goal of use-case 2; ‘Volatility of exchange rate’ did not have panel data. Instead there was only 1 data point per country. Consequently, the way the regulations were coded had to be adapted for that policy goal (Volatility of exchange rate).
Regulations being considered in regressions 1-3:
The regulations under consideration are the full list of regulations as prescribed by the Basel III international regulatory accord, alongside interaction variables between significant variables where more than one regulation was found to be significant in a regression. The following is a complete list of regulations that are considered for regressions 1, 2, and 3.
Regulations not sourced from the Basel III accord are marked with a *.
Furthermore, also included in the below table are groupings for the regulations; this describes the type of regulation. The Basel III regulation variables were coded in a specific way. They were appointed the value ‘No’ in all years where the regulation was not implemented. ‘Yes’
for all years following the publishing of a draft or final rules of the regulation, whichever
comes first. Consultative papers, whilst indicated in the Basel III implementation progress report, were not considered as sufficient implementation of regulation due to the fact that a consultative paper does not necessarily lead to implementation. The justification behind taking the draft publish year as the regulation implementation year is that from the point where the draft is confirmed the regulation is guaranteed to be implemented in the future. By considering this as the point where the regulation begins, it is possible to capture the full effect of the regulation over the transitory period of implementation. This is important because changes may begin from the moment that regulatory implementation is anticipated.
By including this in the variable it is possible to also account for the omitted variables that are correlated with a regulation’s anticipated implementation but not included in the regression. As these omitted variables change as a result of the regulation’s anticipated implementation, they can be considered as an extension of the effect of a regulation’s implementation without having to include them independently in the model.
If a regulation was not applicable to a country, for example, because no banks in the country are on the list of G or B-SIBs, it was coded as ‘No’ for all years for that variable.
For variables relating to information disclosure, where multiple disclosures are covered by a single variable, the implementation year is taken as the year in which a majority of the information is disclosed.
When only limited information was available, such as when it was only stated that a draft rule had been published, with no date of publishing given, the targeted implementation date was used as the regulation implementation date.
If it was noted that a country already had their own regulations in place that were sufficient as to not require implementation of a Basel regulation, then the date of this regulatory
implementation was externally sourced and used as the date of implementation. Whilst the data were coded as strings, upon implementation into the analysis program, Stata, they were re-coded into float format, with ‘No’ replaced by 0 and ‘Yes’ replaced by 1. Dummy
variables were then created based on these numerical values.
Regulations included in the Basel III standards were grouped by type, for example: ‘capital’,
‘leverage’, or ‘SIB’. When reviewing the data, it became clear that regulations within these groups frequently had similar or identical implementation dates. This is likely due to the process of rolling out the Basel regulations, with a focus on one section at a time. This is also
reflected by the targeted implementation dates, which are similar within the regulation-groups mentioned above. This did not however present as an issue in the results.
Regulation: Grouping according to Basel III:
1. Reserve requirements * N/A, not Basel III 2. Countercyclical capital buffer Capital
3. Margin requirements for non-centrally cleared derivatives
Capital
4. Capital requirements for CCPs Capital 5. Capital requirements for equity investments in
funds
Capital
6. SA-CCR Capital
7. Securitisation framework Capital
8. TLAC holdings Capital
9. Revised standardised approach for credit risk Capital 10. Revised IRB approach for credit risk Capital
11. Revised CVA framework Capital
12. Revised minimum requirements for market risk
Capital
13. Revised operational risk framework Capital
14. Output floor Capital
15. Exposure definition (Revised version) Leverage
16. G-SIB requirements SIB
17. D-SIB requirements SIB
Regulation: Grouping according to Basel III:
18. Leverage ratio buffer SIB
19. Interest rate risk in the banking book IRRBB 20. Monitoring tools for intraday liquidity
management
Liquidity
21. Net stable funding ratio (NSFR) Liquidity 22. Supervisory framework for measuring and
controlling large exposures
Large exposures
23. Revised Pillar 3 requirements Disclosure 24. CCyB, Liquidity, Remuneration, Leverage
ratio (Revised)
Disclosure
25. Key metrics, IRRBB, NSFR Disclosure 26. Composition of capital RWA overview,
Pruential value adjustments, G-SIB indicators
Disclosure
27. TLAC Disclosure
28. Market risk Disclosure
29. Member of EU * N/A, not Basel III
30. Pegged to dollar * N/A, not Basel III
Table 2 (Regressions included in Regulations 1-3, with groupings)
Regulations being considered in Regression 4:
Regression 4, relating to the policy goal ‘Volatility of exchange rate’, took a different format to the previous 3 regressions. Whilst, as with regression 3, it related to Bitcoin’s use as a day-to-day means of exchange, regression 4 relied on a different data format. Specifically, unlike regressions 1-3, which are panel data sorted by country, regression 4 used cross-sectional data. This is because the policy goal, volatility of exchange rate, was measured through the
standard deviation of the moving average of the exchange rates over the time period in question (1989-2018).
To obtain figures for this variable, the daily exchange rates of different currencies to the US dollar were obtained from the IMF for the time period of January 3rd 1994 to 31st December 2018. From the raw data, the 1 year moving average of the exchange rate was calculated.
Following this, the standard deviation of this moving average was calculated per country to provide a value of exchange rate volatility.
For European countries the standard deviation of the yearly moving average exchange rate was taken up until they joined the Euro.
Due to the fact that the regression relies on cross-sectional data, the regulation variables used in regressions 1-3 had to be adapted. As the policy goal measures the average volatility in the exchange rate, it is logical that the regulation variables could be converted from dummies into an average of the dummy values. To achieve this, the average value of the dummy variables for each country were counted. This provided a value between 0 and 1, representing the portion of the time period for which a certain regulation was implemented. As regression 4 relies on the same data set for the regulation implementation data, the same core
characteristic of the regulations’ implementation still holds, namely, once a regulation was implemented, it remained implemented for all successive years. Therefore, whilst the value for the regulation variables represents the percentage of the time-period for which a given regulation was implemented, it also can be used to infer when a regulation was implemented (for example, a value of 0.5 would mean that a regulation was implemented halfway through the time period; in 2004).
Control variables included in the regressions for use-case 1&2:
The variable EU: (Use-cases 1&2 only)
The variable “EU” is a dummy variable for whether a country is a member of the EU. For the purposes of this research only countries that are fully integrated members of the common currency area are considered to be members of the EU. If a country was a member of the EU at time of research, then it is marked as ‘Yes’ for all years in the study, not just the years when it was a full member. This is due to the fact that the process of joining the EU is not instantaneous, with a long build-up period where policy goals and regulatory approaches are aligned. Due to the infeasibility of accounting for variances in the timing of these alignments it was determined that a simpler dummy approach to indicate whether the country is a
member of the EU is sufficient. This is further justifiable by the fact that the majority of regulations in question were only implemented after the EU countries were already members, thus the EU dummy serves more as an indicator that those countries have the same regulation implementation schemes.
Pegged to the US Dollar: (Use-cases 1&2 only)
This is a dummy variable for whether a country had its currency pegged to the US dollar in a given year. For the purposes of this research the US dollar is considered to be pegged to itself, to ensure that the stability of the US dollar relative to the US dollar is not falsely attributed to the unpegged currencies.
Regulations being considered in Regressions 5 & 6:
Use-case 3 treats Bitcoin as a security; it is necessary to understand how regulations that affect the exchange of securities could affect the policy goals of regulators.
There are two policy goals that were considered for use-case 3: ‘Economic growth’, and
‘Sustainable development’. Unlike use-cases 1 and 2, there was no clear global framework for regulating securities. This posed challenges when seeking to consistently evaluate the impact of regulations across countries. A satisfactory solution was found by individually curating a smaller list of regulations based on the policy goals of use-case 3. Namely,
‘Carbon tax’, a dummy that describes whether a country has a carbon tax in a given year,
‘MAD’, a dummy that represents whether a country has agreed to sign the Market Abuse Directive, and ‘TPD’, a dummy to describe whether a country has agreed to sign the transparency directive. Whilst there were significantly less regulations included in the regressions for use-case 3, the included regulation dummies cover the weaknesses of Bitcoin for the use-case and are expected to have a significant impact on the policy goals.
The dummy variable ‘Carbon tax’, describing whether a country has a carbon tax initiative in place for a given year is included as it puts a price on pollution. Bitcoin, specifically Bitcoin mining, the process through which new Bitcoins are created and transactions are confirmed, uses significant quantities of energy, estimated to peak at over 100 terawatt-hours per year (CBECI, n.d.). Of this, much of the energy consumption occurs in China, a country well known for its high reliance on non-renewable, highly polluting energy sources
(Digiconomist, 2021). A carbon tax, therefore, if implemented on Bitcoin activities, could, if
found to be a significant driver of sustainable growth, lead to a shift in Bitcoin energy sources towards renewable, green alternatives.
The MAD (Market Abuse Directive), according to Directive 2014/57/EU: “establishes minimum rules for criminal sanctions for insider dealing, for unlawful disclosure of inside information and for market manipulation to ensure the integrity of financial markets in the Union and to enhance investor protection and confidence in those markets” (Directive 2014/57/EU, 2014). At present the Bitcoin market has no legal rules against manipulation of the market. This is one of the key weaknesses of Bitcoin, as large holders of the
cryptocurrency have the power to significantly impact the price without punishment. This variable is used to measure whether imposing punishments for market manipulation could be used for the Bitcoin market to prevent manipulation, without leading to economic decline.
Similarly, the TPD (transparency directive) “addresses corporate reporting and disclosure”
(Christensen et al., 2010). According to the European Securities and Markets Authority (ESMA), the TPD “aims to ensure transparency of information for investors through a regular flow of disclosure of periodic and on-going regulated information and the
dissemination of such information to the public. Regulated information consists of financial reports, information on major holdings of voting rights and information disclosed pursuant to the Market Abuse Directive (Directive 2004/109/EC, 2004).” It has been noted previously that Bitcoin’s anonymity can cause issues with regard to criminal activity. It can be inferred that regulation akin to the TPD, specifically applied to Bitcoin, requiring specific reporting of transactions, holdings and general Bitcoin activity by market participants could reduce the frequency and severity of criminal activities with Bitcoin. Furthermore, increased
transparency could lead to wider adoption of Bitcoin as an investment as its perceived safety increases, leading to growth of Bitcoin as an investment medium for economic growth.
As the MAD and TPD variables only apply to the EU, with the regressions using panel data and within-group fixed effect models meaning that data must be available for all variables for a country to be included in the regression, it is only possible to evaluate EU countries for use-case 3. Whilst this does not fully account for the full global variance in economic and
environmental situations, it should provide enough data to extrapolate approximate effects for the remaining countries.