Use-case 2 evaluates the effects of different regulations on the policy goals of regulators relating to currency in its use as a means of day-to-day exchange. Similar to use-case 1, this was achieved through two separate regressions, wherein each of the two policy goals related
to the use-case were regressed against the regulations being evaluated. These policy goals are
‘Inflation’ and ‘Volatility of exchange rate’.
The approach for use-case 2 was slightly different to that of use-case 1. This is due to the fact that, whilst regression 3 takes the same functional approach to regressing the policy goal on the regulation variables, regression 4, evaluating the second policy goal of use-case 2,
‘Volatility of exchange rate’ uses cross-sectional data. Consequently, the independent variables, those being the regulations and control variables, needed to be adapted to suit a cross-sectional approach, the approach to this explained in the methodology section.
The Regulations under consideration were the same as those used in the regressions of use-case 1, being primarily sourced from the Basel III banking regulation framework, with an additional regulation variable ‘Reserve requirements’. Regression 3 also included the dummy control variables ‘dummy_EURO’ and ‘dummy_Peg’.
As with use-case 1, regulations that were evaluated to have significant, and desirable effects, on the policy goals were added to the ‘sub-basket’ of optimal regulations for use-case 2. For the sub-basket of use-case 2, regulations from regressions 3 and 4 were considered.
4.2.1 Regression 3: Inflation
Similar to the first 2 regressions, a Hausman test was performed for regression 3. Once again, the random effects model provided inconsistent estimators, therefore a fixed effects approach was employed.
Regression 3 also evaluated the effect of Basel III regulations, on the policy goal of low inflation, under the policy label: ‘Inflation’. In addition to the Basel III regulations, the model for regression 2 also included the variable ‘Reserve requirements’ and control variables
‘dummy_EURO’ and ‘dummy_Peg’.
The policy goal ‘Inflation’ is measured through the annual inflation rate per country, using GDP deflator data from the World Bank (World Bank, 2020d). “The GDP implicit deflator is the ratio of GDP in current local currency to GDP in constant local currency” (Kok & Ersoy, 2009). It is generally agreed that a low and stable inflation rate is optimal. Whilst regression 4 will evaluate the stability of the inflation rate, regression 3 focuses on the absolute value of
inflation. Consequently, a desirable effect on inflation is considered to be a decrease in inflation rate. This would be represented by a negative coefficient of the regressor.
The initial regression provided the following output:
Regression 3: Inflation rate
(Regression 3: output 1)
Of the variables included in the model for regression 3, two regressors were found to be significant. These are observed in the above output as ‘Regulation1Reserverequiremen’ and
‘dummy_Reg2’. These regressors represent the regulations ‘Reserve requirements’ and the dummy for ‘Countercyclical capital buffer’, henceforth referred to as such for clarity.
Similar to regressions 1 and 2 of use-case 1, the variable ‘Reserve requirements’ is measured as the “ratio of bank liquid reserves to bank assets, given as a percentage” (World Bank, 2020b). Whilst the variable ‘Countercyclical capital buffer’, was a dummy variable
measuring whether a country had a countercyclical capital buffer regulatory requirement for banking in a given year.
The coefficient of the variable ‘Reserve requirements’ was observed to be 0.1263952, significant at less than 0.01% level. The coefficient value of 0.1263952 can be interpreted as meaning that for each unit increase in a bank’s ratio of liquid reserves to bank assets, the country’s Inflation rate is expected to increase by 0.126%, ceteris paribus. A desirable effect on the policy goal ‘Inflation’ occurs when a regulation has a negative coefficient, as this would lead to a decrease in the inflation rate. Under this criterion, the regulation ‘Reserve requirements’ should not be included in the sub-basket of regulation for use-case 2.
The second significant regulation, ‘Countercyclical capital buffer’ had a coefficient of -2.781626, significant at the 0.9% level. This can be interpreted as meaning that countries that had countercyclical capital buffers implemented saw on average Inflation rates that were 2.782% lower than those that did not have a countercyclical capital buffer requirement, ceteris paribus. As mentioned previously, for the policy goal ‘Inflation’, a desirable
regulatory effect would see a decrease in the Inflation rate when regulation is implemented.
As this is the case for the regulation ‘Countercyclical capital buffer’, there is a strong case for its inclusion in the sub-basket for use-case 2.
Whilst the ‘Reserve requirements’ regulation was not found to independently have a
desirable impact on the policy goal ‘Inflation’, it is necessary to also evaluate whether there may be a desirable interaction effect between a ‘Reserve requirements’ regulation and the
‘Countercyclical capital buffer’ regulation. This is achieved through the inclusion of an interaction variable between the two regressors.
The output of this new regression is observed below:
Regression 3: Inflation rate including interaction effect variable
(Regression 3: output 2)
From the above output, including the interaction effect, it is clear that there is no significant interaction effect between the variables ‘Reserve requirements’ and ‘Countercyclical capital buffer’, with the interaction variable having a p-value of 0.056. Furthermore, when the interaction effect is implemented, the variable ‘Countercyclical capital buffer’ becomes insignificant. The purpose of evaluating the interaction effect was to determine whether the joint implementation of regulations for both ‘Reserve requirements’ and a ‘Countercyclical capital buffer’ could lead to a desirable effect on the policy goal of ‘Inflation’, despite only
‘Countercyclical capital buffer’ having a desirable effect when considering the regulations independently. As there is no significant interaction effect between the variables, with only
‘Countercyclical capital buffer’ having an independently desirable effect on the policy goal,
only ‘Countercyclical capital buffer’ can be recommended for inclusion in the sub-basket for use-case 2.
4.2.2 Regression 4: Volatility of exchange rate
Unlike regressions 1-3, regression 4 uses cross-sectional data instead of panel data.
Consequently, it is not necessary to perform a Hausman test for fixed or random effects.
Initially a multiple linear regression was performed, regressing the ‘Volatility of exchange rate’ variable on the modified independent variables from regressions 1-3.
Following on from regression 3, the policy goal ‘Volatility of exchange rate’ is
calculated using the Standard deviation of moving average of exchange rate. As mentioned in the regression 3 section, the goal for inflation is to maintain a low and stable rate. Regression 4 captures the stable aspect of this goal. Consequently, regulations that decrease the volatility of the exchange rate are considered to be desirable, which would be observed as a negative coefficient for the regressor in the regression output.
The output of this regression is observed below:
Regression 4: Exchange rate volatility
(Regression 4: output 1)
The above regression output indicates that none of the included regulations were found to have a significant effect on the policy goal ‘Volatility of exchange rate’. Furthermore, many of the regulations were omitted due to collinearity. This is likely due to the fact that many regulations were implemented at similar times, resulting in identical values when converting the variables from panel to cross-sectional format.
What can be concluded from the insignificant results is that none of the Basel III banking regulations had a significant impact on the stability of the exchange rate of the countries included in the model. This does not discount the use of ‘Volatility of exchange rate’ as a policy goal; as the literature indicated that it is a key policy goal for the central bank when considering currencies. On the contrary, it indicates that regulations are likely not the primary drivers of exchange rate volatility. This is backed up by the relatively high adjusted
R-squared of the regression, 0.4937. This value indicates that the regressors included in the model provide a moderate level of explanatory power. From this it can be inferred that the regulations included are appropriate predictors of the policy goal ‘Exchange rate volatility’, however, they simply do not have significant effects.
To conclude the analysis of regression 4, evaluating the impact of the regulations on the policy goal of ‘Exchange rate volatility’, it is not possible to infer any significant regulations from regression 4. Therefore use-case 2 must rely solely on the regulations indicated as desirable by regression 3.