Existing literatures forecasted CIC as it is one of the most significant issues influencing the liquidity of the financial market. Given that the Bank of Mauritius (BOM) is the sole distributor of currency it may not forecast the demand for the currency precisely as it is strongly influenced by the non-banking sector. Therefore this dissertation introduces an ARIMA and regression model to forecast CIC in the case of Mauritius. The BOM did not undertake a similar study earlier; hence this paper is the first study that forecasts the CIC in the case of Mauritius and this study is vital chiefly for policymakers.
1.1 Aims & Objectives
The aim of this research is to forecast CIC in the case of Mauritius using ARIMA and regression model on a monthly basis. Moreover, the other objective is estimating accurate currency demand as this would allow the Central Bank to design monetary policy strategies beforehand and to manage liquidity efficiently. This would help the Central Bank to forecast currency demand precisely and thus leading to an effective monetary policy implementation. Lastly, cointegration analysis is used to analyse the relationship between CIC and Gross Domestic Product (GDP).
1.2 Methodology
To model and forecast CIC, the ARIMA and regression models are used. The Box-Jenkins ARMA models used in forecasting is generally viewed to be an effective forecasting technique and is used widely, for univariate time series.
The sample covers the period of January 2000 to December 2011. The actual data for the period of January 2012 to November 2012 have been used to test the validity of the forecast. In Mauritius, the data for CIC are published by the BOM at monthly intervals and the data is available in the monthly statistical bulletin of the Bank of Mauritius.
The cointegration is performed using the Johansen cointegration test and quarterly time series data about GDP for the period 2000-2011 is gathered from the Central Statistics Office (CSO).
1.3 Organization of study
This piece of work is organised as follows:
Chapter 2: Review of theoretical and empirical literature
The section explores the different views brought forward by various researchers on CIC. This chapter is very important in the sense that the empirical literature provides guidance and relevance to the study.
Chapter 3: Overview of the Mauritian economy
The chapter provides an overview of Mauritius.
Chapter 4: Research Methodology
In this section, we introduce the methodology that we will use for our study. The data obtained came from reliable sources. A time series analysis is carried out using Mauritian monthly data over 2000-2012 to forecast CIC and quarterly time series data about GDP and CIC for the period 2000-2011 are used for the cointegration analysis between CIC and GDP.
Chapter 5: Empirical results
This chapter interprets the results obtained. The desired output is obtained which shows that we are able to reproduce the empirical literature with our methodology.
Chapter 6: Summary
This chapter provides the conclusion of the dissertation. It will sum up with briefings of the previous chapters and main findings in this study. Furthermore, it concludes whether or not the objective of the dissertation is attained and provides some policy recommendations.
2.0 Introduction
The main purpose of this chapter is to study existing literatures on modelling and forecasting Currency in Circulation (CIC). An important function of the Central Bank is to ensure a smooth and efficient supply of banknotes and coins to meet public demand, which is dependent on the level of economic activity, calendar anomalies, and the improvement in the alternative means of payment. Moreover, a cointegration analysis is used to study the relationship between CIC and Gross Domestic Product (GDP).
This chapter is structured as follows: section 2.1 deals with theories based on CIC, monetary base, significance of CIC, importance of forecasting banknotes and changes in the total amount of banknotes in circulation. The next section involves the calendar effect. Section 2.3 explains the findings and models used by other researchers for forecasting CIC and the relationship between CIC and GDP. Then section 2.4 relates the summary of the literature review.
2.1 Theoretical Literature
2.1.1 Currency in Circulation
The Bank of Mauritius (BOM) is the central bank of the country and as the sole currency issuing authority in Mauritius; BOM is responsible for issuing currency to the financial institutions under its supervision when they need it. When financial institutions have more currency than is needed, they send it back to the BOM. The BOM is also responsible to regulate the banking and credit system so as to ensure a proper circulation of credit and a sound financial structure. The BOM Act 2004 specifies that "The primary object of the Bank shall be to maintain price stability and to promote orderly and balanced economic development".
The BOM must devise and implement a number of measures affecting the supply of reserve money, the money supply and the level of interest rates in the economy to accomplish its objective of preserving price stability.
CIC is defined as the total amount of banknotes and coins in circulation outside the central bank. The CIC includes all banknotes in domestic currency that the economic agents that is households, companies and non-residents demand for a particular moment for transaction or as a store of value. The circulation of banknotes and coins to the non-banking sector is mainly carried out by commercial banks.
2.1.2 Monetary base
The monetary base consists of currency outside banks plus reserves balances (deposits held by banks and other depository institutions in their accounts at the Central Bank).
Components of Monetary base
Monetary base = Reserve Money + CIC
Central Bank Balance sheet
Assets
Liabilities
Government securities
CIC
Discount loans
Reserves
The two liabilities on the balance sheet, CIC and reserves, are often referred to as the monetary liabilities. They are an important part of money supply because increases in either of them would lead to an increase in the money supply (everything else being constant). The sum of CIC and reserve is the monetary base.
CIC: It refers to the currency that the Central Bank is issuing and the amount of currency in the hands of the public (outside of banks) and it is an important component of the money supply.
Reserves: All banks have an account with the Central Bank. The reserve consists of deposits at the central bank and currency held by commercial banks. Reserves are assets for the banks but liabilities for the Central Bank, because the banks can demand banknotes at any time and the Central bank is obliged to satisfy its obligation. An increase in reserves leads to an increase in the level of deposits and hence in the money supply.
The volume of monetary base is controlled by the central bank Open Market Operations (OMO). OMO are the principal instrument of monetary policy, consisting of the purchase and sale of Government securities by the central bank. It is the primary determinant of changes in reserves in the banking system. An open market purchase of government security will lead to an expansion of reserves and deposits in the banking system and therefore lead to an expansion of the monetary base and the money supply, while an open market sale of government security will lead to a contraction of reserves and deposits in the banking system and hence to a decline in the monetary base and the money supply.
In Less Developed Countries (LDCs), central bank will adopt an expansionary monetary policy and as a result an expansion of the money supply. This will lead to a decline in real interest rate to encourage borrowing by businesses hence raising investment and boost economic activities in LDCs.
A change in the monetary base is due to the intervention of Central Bank in the foreign exchange market, that is the Central Bank either buys or sells foreign currency called international reserve. If the Central Bank decides to sell its foreign currency in exchange for domestic currency this has two effects:
It reduces the Central Bank holding of international reserves.
Because its purchase of currency removes it from the hands of the public, CIC falls.
Because the monetary base is made up of CIC plus reserves, this decline in currency implies that the monetary base has fallen. Likewise if a Central Bank decides to sell its domestic currency to purchase foreign currency this will result in an equal rise in its international reserves and the monetary base.
2.1.3 Significance of CIC in an economy
Firstly, the most important factor influencing the liquidity of the financial market is the amount of banknotes and coins in circulation. Cabrero et al. (2002) stated that when evaluating the liquidity requirements of the banking system, it is obligatory to take into consideration the expected value of independent liquidity factors that affect the supply of bank reserves. These factors are independent because they are beyond the control of the central bank. For example, CIC is one of the main autonomous factors. It is a liquidity absorbing factor because cash withdrawals from banks will lead to a rise in the level of CIC induce additional refinancing needs of banks which have to meet their reserve requirements.
Furthermore, the share of CIC in money supply that is banknotes in the hands of the public is a key component of the money supply measures. The classical economists explained that changes in the money supply have no influence on output; they argued that the amount of goods and services produced in the economy should not depend on the amount of banknotes in circulation but rather on the economy's productive capacity and that the price level will change according to changes in the money supply. This is what is referred to as money neutrality; money is just a veil that enables transactions to occur. Therefore, if there is a rise in the amount of CIC in the economy, there will be more money chasing more goods and services, hence prices will increase. On the other hand, if money supply declines, there will be less money chasing those goods and services, resulting in a fall in price.
Finally, the ratio of CIC to nominal Gross Domestic Product (GDP), according to the classical view money can be understood in terms of the quantity theory, which links the level of nominal GDP, represented by PT (the price level, P, multiplied by real GDP, T) to the product of the money supply, M, and the velocity of money V.
MV=PT
M is the money supply which includes currency and bank deposits available to the public.
V is the velocity of money. It basically explains how quickly the money supply is turned over.
P is the price level.
T is the total amount of transactions.
So MV=PT means that the total GDP at the current price level is equal to the total money supply multiplied by how often it is turned over.
2.1.4 Importance of forecasting banknotes
Given that the Central bank has the exclusive right to issue currency; it cannot determine exactly the total amount of CIC as that demand is influenced by the non-banking sector. Therefore, Central banks should focus on forecasting banknotes in circulation because providing a precise forecast would enable the central bank to:
Implement effective monetary policy so they can manage liquidity efficiently both in terms of absolute size and in terms of instability
To maintain price stability
Hlavacek et al. (2005) stated that since the objective of the Central Bank is to maintain price stability, it needs an accurate estimate of money market liquidity. However, the liquidity is influenced by independent factors that are not under full control of the Central Bank. The most significant independent factor is CIC, which is difficult to measure as it is strongly influenced by many seasonal factors. To get over the problem Central Banks have been using mathematical models of CIC as supportive tools and most of them have used linear ARMA models.
Balli and Elsamadisy (2010) forecasted CIC because an accurate prediction of CIC would help to stabilize the money market in the short run and it certainly helps to decrease instability in money market rates, thus resulting in higher economic growth.
According to Fischer (2004) the total amount of banknotes and coins is thoroughly controlled by the monetary authorities, as central banks are, in principle, able to decide exactly about the amount of currency they put into circulation and know the exact amount of remaining currency. Moreover, central banks usually accommodate the total demand for banknotes irrespective of its origin. This notwithstanding, central banks are not able to follow the way that currency takes once put into circulation. Therefore, there is little straight statistical information on where the currency circulates, who holds the currency (residents or non-residents) and on the motives for which the currency is held (for transactions, hoarding or illegal purposes). Hence for monetary policy purposes, the amount of currency held for transactions within the domestic currency area is of particular interest and it should be forecasted.
2.1.5 Changes in the total amount of banknotes in circulation
An enormous number of issues and situations cause a change in the total amount of CIC that are clearly uncontrollable. Variations in the total amount of CIC directly have an impact on the liquidity of the banking sector.
When banknotes and coins is returned to the central bank the amount of CIC reduces, hence liquidity of the banking sector increases. Likewise, a cash withdrawal that is, the total amount of CIC rises will lead to a decrease in the liquidity of the system. Thus changes in the overall amount of CIC directly influence the liquidity of the banking sector.
Variations in banknotes in circulation are of interest to policy makers because an increase in CIC entails a decline in deposits and as a result it leads to a decline in the availability of loanable funds for investment, which is essential for economic growth. Moreover, a rise in CIC indicates that there in inflation in the economy.
With financial innovation, that is there is an improvement in the alternative means of payment, the rate of growth in the series of banknotes in circulation has fallen. For example significant changes in the application of electronic technology such as the availability of credit cards which give access to the use of Automatic Teller Machines (ATMs), Electronic Fund Transfer at Point of Sale (EFTPOS) and electronic banking services and overdraft facilities have reduced the need of banknotes.
CIC is a function of interest rate. Interest rate denotes the opportunity cost of holding money, and as a result, the higher the interest rate the lower the level of CIC. While, the rate of interest is lower, the level of CIC rises.
Another aspect that determines CIC is the GDP growth, as people's incomes increase in nominal terms, CIC is likely to rise and as income decreases the amount of banknotes in circulation declines.
Moreover, there are other irregular phenomena that have an impact on the series of CIC. For example, according to Cabrero et al. (2002) the amount of banknotes in circulation rose rapidly in late 1999, because of the Y2K computer bug which led to a strong increase in demand for banknotes.
2.2 The Calendar Effect
Calendar effects are anomalies that relate to the calendar such as the changing month length, the effect of religious and public holidays and the day of the week effect. For example, calendar anomalies have been documented by Balli and Elsamadisy (2010) and they are characterized as international phenomena.
Calendars exert intense effect on the cultural, social and economic behavior of the people. Almost all countries follow the Gregorian calendar to set its working days and holidays according to this calendar. The Gregorian calendar is having 365 or 366 days in every year. However, holidays do not follow the Gregorian calendar, as they are based on the festivals and religious observances of different ethnic groups in the country, whose date vary from year to year. This makes it challenging to model high frequency time series.
The amount of CIC rises just before the weekend when ATM networks are filled for the weekend to withstand all the shopping activity and declines after the weekend that is the trading day effect. It also declines before the middle of the month and increases towards the end of the month as a result of the payment of incomes. The amount of CIC increases during the summer holidays and towards the end of the year, mostly around Christmas. Public holidays also have a strong influence on the amount of banknotes in circulation.
There are 2 different categories of the public holiday effect in Mauritius during the year, one is the fixed public holidays and second one is the religious holidays whose dates vary from year to year.
Fixed Holidays
New Year January 1st and 2nd.
Abolition of Slavery Day February 1st
Independence/Republic Day March 12th.
Labor Day May 1st.
All Saints Day November 1st.
Christmas December 25th.
Moving Holidays
Thaipoosam Cavadee January/February.
Maha shivaratree February.
Chinese Spring Festival January/February.
Ougadi March.
Id-El-Fitr May/June.
Ganesh Chaturthi September.
Divali October/November.
Fixed holidays
Public holidays which are fixed to a specific date and their positions in the month do not change. For example during the Christmas shopping season and the payment of annual bonuses there will be an increase in the series of banknotes in circulation.
Moving holidays
Festivals which do not strictly follow the Gregorian calendar, the dates of religious festivals change over time. That is, though the holiday occurs at almost the same month each year, the exact date changes. There is higher demand for banknotes during festive seasons. Therefore, there is an increase in the amount of banknotes in circulation just before the public holidays and decreases after the holiday.
Intra-monthly effect
Most of the payments (wages, salaries and pensions) to individuals are paid at the end of the month. Consequently banknotes in circulation increases at this time. As households make their regular payments, the money goes to the corporate sector and flows back to the banking system and then the demand for currency decreases until salaries are paid again.
2.3 Empirical Literature
This section provides a brief description of relevant empirical studies of modeling and forecasting CIC. The main findings have been recorded, merged and compared where appropriate to build a model and forecast CIC. Finally, the link between CIC and GDP is examined.
2.3.1 Forecasting CIC
CIC is the most important independent feature in the framework of liquidity, both in terms of size and instability. Therefore, researchers have tried to develop forecasting means that will minimize the forecast errors. By improvement in the forecasting techniques, some researches have obtained precise estimations in recent years.
For example some papers have predicted the CIC based on the theory of transaction, portfolio demand for money and univariate time series models. Bhattacharya and Joshi (2000) have predicted the banknote in circulation for India with weekly data sets from 1992-93 to 1999-00. Another study by Dheerasinghe (2006) have predicted CIC for Sri- Lanka with monthly, weekly and daily data sets for the period 2000-2005 and data for 2006 were used for post validity test. All 3 models used to forecast CIC in Sri-Lanka provided close predictions during the post sample period. These methodologies work well for low frequency data but faces restrictions with high frequency data series. Hence Bhattacharya and Joshi (2000) suggests a substitute way of modeling the weekly growth of CIC by including the 'day of the month' effect whereas Dheerasinghe (2006) proposes another approach in modeling the high frequency data by decomposing the trend, the seasonal and the cyclical components.
Balli and Elsamadisy (2010) forecasted CIC for the State of Qatar using both regression model and Autoregressive Integrated Moving Average (ARIMA) model. To obtain a precise forecast, they forecasted the CIC using daily, weekly and monthly data. They also took into consideration the effects of holidays, weekends and public holidays on currency holding. Both models provided satisfactory results, however, the ARIMA model provides better estimations for short term forecasts compared to the error terms of regression model which is higher. But, for long term forecasts, both models suffer from larger forecast errors.
Lang et al. (2008) forecasted CIC outside banks in Croatia using two econometric models for the short term forecast. In the past forecasting were done by the Croatian National Bank (CNB) staff with expertise in predicting banknotes in circulation outside banks. Hence Lang et al. (2008) forecasted CIC using both Regression model and ARIMA model. In order to obtain precise forecasts it is required to formalize its weekly, monthly and annual patterns. However, they concluded that both models provided satisfactory forecasts. But the regression model is better than the ARIMA model on the intervals up to 5 days ahead, whereas the ARIMA is better for long-term horizons.
In a recent study, Güler and Talasli (2010) model the daily series of banknotes in Turkey for the period of September 2004 to January 2009. Their main motive was to construct an econometric model to predict daily CIC, as the CIC is the most important factors influencing the liquidity of the Turkish banking system. Hence it is fundamental for the Central Bank of the Republic of Turkey (CBRT) to produce accurate forecasts of CIC. Since the total amount of CIC is out of control of the Central Bank. It is required to construct an ARIMA model. They concluded that the forecasting performance of the ARIMA based approach is better than the expert judgments. However, the model has to be constantly checked to improve the quality of the model and make adjustment whenever needed.
Hlavacek et al. (2005) forecasted the banknotes in circulation for the Czech Republic using both feed-forward structured neural network model and ARIMA model for the period July 2003 to June 2004. The purpose of their study was to introduce feed-forward model to forecast banknotes and compare with an ARIMA model. They obtained satisfactory results using the ARIMA approach for long term horizon (forecasts more than 2 weeks ahead) and short term horizon ( for at most 2 weeks ahead), however, they stated that the feed-forward neural network model is a better approach for analysis of time series of CIC compared to the ARIMA approach.
Cabrero et al. (2002) model the daily series of CIC in the framework of managing of the European monetary system using the ARIMA approach and Structural Time Series approach (STS). So far, the forecasting of CIC has been computed by the National Central Banks (NCBs) of the Euro system and the quality of NCBs forecasts has been good up to now. Yet, there are 2 reasons for predicting CIC in the Euro area directly. First, this prediction can be used to supplement and improve the National forecasts and in addition, the introduction of Euro banknotes in 2002 and the free movement of banknotes within the Euro area may make the NCBs prediction less dependable. Results presented in their study shows that these 2 approaches are influential and show a performance which is up to the standards of the current aggregated forecast approach employed by the Euro system. They concluded that the ARIMA provides satisfactory forecasts over 5 days and above, while the STS approach is best over horizons of 1 to 4 days. Therefore the best forecasting model is a combination of the ARIMA and STS models.
2.3.2 CIC and GDP a cointegration analysis
According to the Cambridge approach where MV=PT, any increase in M (Money Supply) will bring about a proportionate rise in P (Price level). As a result MV=PT means that total GDP at current price level is equal to the total money supply multiplied by how often it is turned over. To test whether this statement is valid empirically, Bednarik (2010) examined the Czech economy using graphical analysis and Vector Auto-Regression (VAR) models and found that there is indeed a strong relationship between GDP and money supply. Another study by Zapodeanu and Cociuba (2010) analyzed the data of GDP after removing seasonality and of money supply through the Augmented Dickey-Fuller (ADF) test and found that the two variables are non-stationary. They further applied the Engle-Granger method and show that the two variables have a cointegrating relationship and concluded that the best model explaining the relationship between money supply and GDP is the DVAR autoregressive the difference model. A recent analysis conducted by Mohamadpour et al. (2012) attempted at unveiling the relationships that exists between GDP and monetary policy in Malaysia. Quarterly data from 1991 to 2011 r unveil to be stationary after the first difference. Hence, they applied the cointegration analysis and Vector Error Correction Model (VECM) to examine the link between GDP and money supply. Findings suggested that the VECM analysis specifies that money supply are statistically significant and have a long term influence on GDP. Thus implying that increasing money supply in Malaysia will have a positive impact on GDP.
2.4 Conclusion
This chapter reviews the theoretical and empirical literature of modeling and forecasting CIC and it explores the different views brought forward by various researchers on CIC. Furthermore, the relationship between GDP and CIC are explored. This chapter is very important in the sense that the empirical literature provides guidance and relevance to the study.
