Appendix I. Recent MCM Technical Assistance to the RMA on Central Banking Issues
Appendix II. Key Features of a Liquidity Forecasting Framework
Appendix III. Modeling Currency in Circulation – Structural Time Series Approach
1. The augmented ARIMA model for the purpose of banknote modeling has the following structure:
Dayi is the dummy variable that takes the value of 1 on day i of the week and 0 otherwise.
Weeki is the dummy variable that takes the value of 1 during week i of the year and 0 otherwise.
Week Position is the dummy variable that takes the value of 1 during the first week of a given month and 0 otherwise.
Monthi is the dummy variable that takes the value of 1 during month i of the year and 0 otherwise.
Public Holiday is the dummy variable that takes the value of 1 when a day is a public holiday and 0 otherwise.
Public Holiday-i is the dummy variable that takes the value of 1 i day(s) before the public holiday and 0 otherwise.
Public Holiday+i is the dummy variable that takes the value of 1 i day(s) after the public holiday and 0 otherwise.
One Off Factor is the dummy variable that takes the value of 1 when the one-off factor occurred and 0 otherwise.
ARi are autoregressive terms.
MAi are moving average terms.
εi are residuals.
2. To be noted, the position of a public holiday within a week is not neutral and should also be tested. For instance, if a public holiday occurs just before or after a weekend, it should have a stronger effect on CIC than a public holiday occurring in the middle of a week. Before starting the estimation process, the usual stationary test (such as the Durbin-Watson test) should be implemented regarding the explained variable (∆CICt). As for explanatory dummy variables, it is advisable to test a very large scope and to progressively discard non-significant variables. Should residual autocorrelation issues occur, AR and MA should be tested and introduced.
3. Once the specification is established, the forecasting power of the model should be tested using an “out-of-the-sample” testing procedure (straightforward to compute in the available Eviews Software). For a given set of historical data (for example, from January 1, 2006, to August 31, 2015), the idea is to run, in a first step, the regression on a subsample (for example, from January 1, 2006, to July 30, 2015) and to implement the forecast on the other part of the sample (from August 1, 2015, to August 31, 2015). Forecasts are then compared to historical data. Eventually, the model specification and forecasting performances should be reassessed regularly (such as every three months).
Appendix IV. Simulated effects of proposed Option 2 for hydro project flows
1. This Appendix presents an estimate of the effects of Option 2 for hydro project flows (sell INR to the RMA only when there is a need for payments in local currency) on liquidity conditions. The implementation of Option 2 would: (i) considerably reduce the amount of sweeping accounts of banks to the RMA, and (ii) reduce by the same amount the RMA’s foreign reserves (in gross and net terms). Volatility in liquidity conditions would decrease, as well as the optical volatility in the RMA’s foreign reserves.
Although the legal tender is the ngultrum, Indian rupee banknotes circulate freely in Bhutan.
The authorities noted that the successful introduction of new monetary instruments requires the development of the interbank market, including short-term instruments to be used as collateral.
Appendix II describes key features of a liquidity forecasting framework.
Currently, only the balances of the Government subsidiary accounts are included in the sweeping arrangement. The Letter of Credit and Project Letter of Credit accounts are not swept but they do not hold any cash; they only record the limit for withdrawal authorized by the DPA. The balances of the Government Current Deposits (CD) accounts, Trust Funds and Revolving Funds are not included in the sweeping arrangement. Information on these account is available in the Annual Financial Statements of the Government. As of June 2016, balances in those accounts amounted to about 2.7 billion ngultrums, of which 1.5 billion for the Bhutan Health Trust Fund.
The mission was made aware of a project being developed by the government to the effect of setting up a Sovereign Wealth Fund that could play in role in the recording and channeling of the loan and grant components of the hydro flows (in addition to the flows related to the sale of electricity to Indian, and possibly a fraction of the RMA profit remitted to the government). Under such a scenario, the loan and grant component of the hydro projects would first be channeled to an account managed by the RMA (logged in the Sovereign Wealth Fund, and not in the balance sheet of the RMA). Disbursements would occur as and when the Hydro Projects need liquidity. As regards domestic liquidity conditions, such a framework would achieve a similar result to what is propose by the mission. The RMA indicated that SARTTAC TA support may be requested once a draft of the Royal Charter related to the Sovereign Wealth Fund is available.
In this section, all mentioned static levels from the RMA’s balance sheet are as of end April 2017.
In Bhutan, while Government accounts are recorded in the balance sheet of the RMA, at the beginning of the day their balance is credited to the current account of the Bank of Bhutan (BOB). During the day, the BOB handles government transactions through its branches, as instructed by the DPA. At the end of the day, the balance is swept back to the RMA. This mechanism does not change the approach regarding the calculation of this autonomous factor. Because of the sweeping, the liquidity impact of those Government accounts is the same as if they were kept permanently with the RMA.
By convention, the RMA’s foreign currency liabilities to the Government are not included in the calculation of the Government accounts autonomous factor, because they are already included in the NFA autonomous factor.
Sometimes the data download can fail for technical issues, or be very time-consuming.
Because of this difficulty, the Excel file designed by the mission, based on monthly data for the period December 2014 – April 2017, had to be prepared largely on a manual basis. For the necessary move to daily data analysis, an automatic process would be much preferred.
This model prevails in a number of leading central banks. For instance, each division involved in the implementation of monetary policy at the European Central Bank (ECB) typically comprises about 15 experts (with profiles focused on finance and/or economics), and 5 research analysts and trainees (with profiles focused on IT). On top of the daily IT support provided from these internal resources within each division, the ECB’s IT department provides support e.g. for project management and the development of new applications.
Except for the overdraft facility, which can be activated for only three days, and at very penal conditions (16 percent), for a bank in violation of the Cash Reserve Ratio (CRR).
20 percent for banks, 10 percent for non-bank financial institutions.
T-bonds and T-bills, due to their liquidity and low credit risk, are normally the first candidate vehicle for a repo market. There is currently no T-bond in Bhutan (even if the Treasury discussed vague plans for a T-bond issuance with the mission), and T-bills are not dematerialized (paper transactions), making the development of a secondary market difficult.
NFA are categorized as an autonomous item regardless of the exchange rate regime. This assumes that under flexible exchange rates the central bank does not intervene in the very short run and NFA are, therefore, constant. On the other hand, under fixed exchange rates NFA changes are outside the control of the central bank as it is committed to intervene to hold the exchange rate stable.