Demand for money in Lao PDR and policy implications

LIST OF VARIABLES M Total amount money in circulation T Level of transactions Y Output V Transaction velocity of money P Prevailing price level M d Money demand MD Total dem

MINISTRY OF EDUCATION AND TRAINING VIET NAM

NATIONAL ECONOMICS UNIVERSITY

DECLARATION

I hereby declare that this doctoral dissertation “Demand for money in Lao PDR and policy implications” is written by

me and has not been published yet

I am responsible for my declaration

Somphao Phaysith

ACKNOWLEDGEMENT

First of all, as the Governor of the Bank of the Lao PDR, I am very pleased and greatly honored for taking the opportunity to take part in the doctoral program organized jointly between the National University of Laos and National Economics University of Vietnam After a number of years following this program, my study is devoted to the title: “Demand for Money

in Lao PDR and Policy Implications”

I would like to express my heartfelt appreciation to the Lao People’s Revolutionary Party and Ministry of Education and Sport of Lao PDR for granting me this opportunity to undertake this rewarding doctoral program and achieve the result successfully as my wish comes true

As one of the candidates for this program, I would like to extend my gratitude and appreciation to the leaders and management staff of these two universities for making this program possible to enhance capacity building for Lao leaders, who are well qualified from the program During the five-year program, I have got the close guidance and knowledge from the lecturers of the two universities as well as relevant authorities for their support that enable

me to ease difficulties and hurdles and to complete the program successfully

My achivement is also due to valuable academic support from Prof Dr Tran Tho Dat, my academic advisor and close supervisor for my research, who enable me to fulfill all requirements of the doctoral program on a timely manner My special thank also goes to Assoc Prof Lai Phi Hung, the program coordinator I would also like to extend my profound appreciation of wholehearted assistance and cooperation between our two nations, Lao PDR and Vietnam

My research is carried out with the staff development grant from the Bank of the Lao PDR Furthermore, I shall not forget and pay tribute to kind support and sincere assistance rendered by the BOL staff, namely Ms Fongchinda Sengsourivong, Mr Soulysak Thamnuvong, and Mr Khamkeo Visisombath

Lastly, my profound thanks go to my beloved family, including my wife and my two daugthers for their encouragements, care and motivation during

my research They always provide me very important supports, and love during the period of hardship All of their supports provided are the power injecting to me to strive the various difficulties and to complete the program successfully as the task assigned by the Party and Lao people

TABLE OF CONTENTS

DECLARATION i

ACKNOWLEDGEMENT ii

TABLE OF CONTENTS iv

ABBREVIATION vi

LIST OF TABLES vii

LIST OF FIGURES viii

LIST OF VARIABLES ix

INTRODUCTION 1

CHAPTER I: OVERVIEW OF THEORETICAL AND EMPIRICAL STUDIES ON MONEY DEMAND 7

1.1 A brief theoretical overview 7

1.1.1 Quantity theory of demand for money 7

1.1.2 Keynesian approach (John Maynard Keynes) 10

1.1.3 Friedman’s model of the demand for money 14

1.2 Some empirical problems in estimating money demand functions 15

1.2.1 Functional forms 16

1.2.2 Choice of variables 17

1.2.3 Empirical estimation problems 20

1.3 Some Asia-specific studies on the money demand function 23

Conclusion of chapter 1 28

CHAPTER II: LAO FINANCIAL SYSTEM AND MONETARY POLICY 31

2.1 Economic development 31

2.2 Overview of monetary developments in Lao PDR 34

2.3 Banking system 35

2.3.1 Mono-bank system 37

2.3.2 Banking system reform 40

2.4 Monetary Policy 47

2.4.1 Monetary instruments 47

2.4.2 Target of monetary policy 51

2.5 Dollarization 54

2.6 Bond market 56

2.7 Stock market 58

2.8 Interbank market 59

2.9 Exchange rate policy 60

Conclusion of chapter 2 66

CHAPTER III: DEMAND FOR MONEY IN LAO PDR 67

3.1 The theory-based money demand function for Lao PDR 67

3.2 Data description and issues 69

3.2.1 Definition of money 69

3.2.2 Scale variable 71

3.2.3 Opportunity costs 71

3.3 Unit root and co-integrated test 73

3.3.1 Unit root test 73

3.3.2 Johansen co-integration test 74

3.4 Estimating money demand function for Lao PDR by using ECM 76

3.4.1 Engle and Granger: Error Correction Models 76

3.4.2 Estimated results and hypothesis testing 77

Conclusion of chapter 3 92

CHAPTER IV: POLICY IMPLICATIONS 93

4.1 Lao economic development strategy 93

4.1.1 Macro-economic targets 93

4.1.2 Targets of Economic Sectors 94

4.2 Monetary policy recommendations 98

4.2.1 Monetary instruments 99

4.2.2 Choosing intermediate target 99

4.2.3 Increasing banking supervisor 101

4.3 Some recommendations to Lao government 102

4.3.1 Dedollarization 102

4.3.2 Stimulate financial market development 104

CONCLUSION 106

REFERENCE 108

APPENDICES 114

ABBREVIATION

Lao PDR Lao People’s Democratic Republic

Laos Used under monarchy regime

BOL Bank of the Lao PDR

MOF Ministry of Finance

BCEL Banque Pour le Commerce Exterieur or Bank for Foreign

Trade Lao LDB Lao Development Bank

APB Agricultural Promotion Bank

SOCBs State-Owned Commercial Banks

GDP Gross Domestic Product

Kip Khangkhay Banknotes issuing by tri-parties of the coalition

government in 1962 Kip Potpoi Liberated currency under Lao Patriotic Front

NPL Non-Performing Loans

RR Reserve Requirement

CIB Credit Information Bureau

LIST OF TABLES

Table 1.1: Estimated income elasticity and interest rate semi-elasticity for

selected Asian countries 26

Table 3.1: Unit Root Test Results 78

Table 3.2: Likelihood ratio test: lag lengths test 79

Table 3.3: Johansen test: the λ max and λtrace tests results at 5% 80

Table 3.4: Diagnostic tests for the short-run dynamic money demand model, equation (3.9) 82

Table 3.5: Short-run dynamic estimation, dependent variable is ∆lnrm t 90

Table 3.6: Long-run relation, dependent variable is lnrm t 91

Table 4.1: Nonperforming loans ratio and number of banks in Lao PDR 102

LIST OF FIGURES

Figure 1.1: Determinants of money demand function 30

Figure 2.1: GDP growth (1991-2011) 33

Figure 2.2: GDP by Components (billion Kip) 33

Figure 2.3: GDP growth, Inflation rate, M2 growth, M2 to GDP and FX to total deposit 35

Figure 2.4: Interest rates of deposits and loans in Kip 50

Figure 2.5: Monthly loans interest rates by currency (%) 50

Figure 2.6: Money supply- M2 (billion Kip) 52

Figure 2.7: Deposits by currency (%) 55

Figure 2.8: Relationship between M2, inflation and exchange rate (%) 63

Figure 2.9: Exchange rate (Kip/USD) movement (%) 65

Figure 2.10: Average exchange rate (Kip/USD) 65

Figure 3.1: Cumulative Sum of Squares of Recursive Residuals 82

Figure 4.1: Money and interest rate targets 100

Figure 4.2: Lao money multiplier and velocity, January, 1999 – September, 2012 101

LIST OF VARIABLES

M Total amount money in circulation

T Level of transactions

Y Output

V Transaction velocity of money

P Prevailing price level

M d Money demand

MD Total demand for money

M1 Sum of transaction and precautionary demands (narrow money)

M2 Speculative demand (broad money)

t c Transaction costs

r Interest rate

M/P = rm Demand for real balances

πe Expected inflation rates

i Nominal interest rate

GNP Gross National Product

NNP Net National Product

GDP Gross Domestic Product

ln Natural logarithm

∆ First differenced term

cpi Consumer price index

er Exchange rate

t p x

to assess credit risks and fund allocation according to the government plans Furthermore, governments have strong controls over the financial systems, such

as setting interest rate ceilings, entry barriers, and interference in credit allocation According to McKinnon (1973) and Shaw (1973), such financial repression leads to reduction in private savings, thereby decreasing the resources available to finance capital accumulation, with a negative impact on growth The Lao PDR is in the process of transition toward a market economy, the requirement of government’s operations to change its intervention methods and scope is a fundamental process One of these fundamental changes is the way the government conducts its monetary policy as well as the development of the financial system, especially the banking system to achieve desirable economic growth The relationship between monetary growth and economic growth is still

in debate Hossain and Chowdhury (1996, p.126) argued that there is a clear correlation between money supply growth and economic growth in the long-run Other studies in the field also point out the importance of monetary policy, especially, in managing the demand side of the economy

In order to make the monetary policy more efficient and effective, the prerequisite for the monetary authority must be able to predict the demand for money with acceptable accuracy Theoretical views of the demand for money have usually been based on the consideration of money as a medium of

exchange The Keynesian models represent money with the role of transactions and the role of the store of value Analytically, demand for money is the sum of three components: transactions, precautionary, and speculative demand Milton Friedman argued that physical goods should be regarded as a substitute for money and that higher expected rates of inflation should induce a portfolio shift from money to physical assets as well as financial assets (hard currencies) in the context of transition and underdeveloped countries The fact that money does not bring interests like other alternative assets means that money holders should receive compensation in some other forms Following the idea of Keynes, many economists attempted to model the demand for money based on the above consideration Although they are different technically, the main idea is still based on the so-called ‘liquidity services’, which money gives in compensation for the interest earning foregone

The relationship between the demand for money and its key determinants is an important building block in macroeconomic theories and is

a crucial component in the conduct of monetary policy (Goldfeld, 1974) Even in the era of inflation targeting, a well-specified money demand function

is of utmost importance for the effective implementation of monetary policy – especially to track both, the interest rates and the stock of money – in order to access the impact of monetary policy upon the economy As a result, the issue

of long-run relationship between broad money, its determinants and also the stability of the demand for money has always been in the center of research Stable money demand is particularly important for policy makers to choose a credible monetary policy instrument For instance, unstable money demand caused by the financial reforms of the late 1970s, the financial innovations and the financial integration induced many central banks in

developed countries to switch from monetary targeting to the interest rate targeting as a monetary policy instrument The same view was proposed by Poole (1970) who showed that the interest rate should be targeted if the money demand function is unstable However, monetary targeting can play an important role in the formulation of an efficient monetary policy strategy, even though the monetary policy of developed countries typically uses an interest rate target as a policy instrument Monetary aggregates can be appropriate indicators for future inflation in the medium term and long-term

As mentioned by Valadkhani (2006) an emerging consensus among economists is that it is not advisable to concentrate exclusively on a single policy instrument while neglecting another important information variable Both interest rates and monetary aggregates are important in selecting appropriate monetary policy actions Monetary aggregates, however, will only

be related to the real economy if the money demand function is stable Thus, the stability of money demand entails whether monetary targeting is an appropriate guide to policy

Due to its particular importance, the money demand function was studied extensively in many countries as can be seen from a large body of literature

on theoretical as well as empirical studies of the demand for money, discussed

in Chapter I However, these studies had been largely carried out in developed countries One explanation for that is the lack of good quality data in developing countries Lao PDR is not an exception in that respect Indeed, the poor quality and short time horizon of Lao economic data is well known among researchers One main reason is that Lao PDR just adopted the system

of national accounts (SNA) in 1993 and the statistical system is still underdeveloped

The dissertation titled “Demand for money in Lao PDR and policy implications” can be viewed as the earliest research in the case of Lao PDR

This fact, while telling us about many difficulties, also can be viewed as an encouraging attempt for us since ‘learning by doing’ in many cases may be the best strategy

Objectives of research

The main purpose of this dissertation is to estimate the money demand function for Lao PDR during the period of the first quarter of 1993 to the second quarter of 2010 (1993Q1-2010Q2), using available data Taking into account the limitations mentioned above, the conclusion of the dissertation should be viewed as a suggestion However, by doing so it is hopefully to establish an appropriate framework for future studies in this field once the comprehensive data are available In addition, the results of this study can hopefully reveal some important issues and areas required to improve, especially data problems which will raise awareness among policy makers to improve the quality of the economic database of the country

Research Questions

The main research question: What is the money for demand in the Lao PDR?

The specific research questions:

1 What is the behavior of money demand patterns?

2 How has the money demand contributed to the monetary stability in monetary policy?

3 What are direct and indirect factors influencing the demand for money

uses a basic and popular theoretical framework surrounding the money demand analyzed from various empirical works After formulating the theoretical model, this dissertation adopts a framework of the Error Correction Model, which is widely used, to analyse the money demand in both developed and developing economies, to examine factors driving money demand balances in Lao PDR over a period of 1993-2011 This econometric model, an Error Correction Model is believed to be well-suited for the empirical investigation through rigorous empirical testing

Contribution of the study

To the best knowledge of the author, the current study represents the first attempt to examine the factors influencing money demand for Lao PDR This study will provide a quantifiable estimation of the money demand in Lao PDR for the first time by using quarterly data The outcome of the study can be useful for the purpose of conducting monetary policy for policy makers of the country’s central bank The economic variables which are used to conduct monetary policy identified in this study will be helpful to systematically consider monetary policy and facilitate policy discussions in the country The outcome will provide information needed for key decisions to formulate the future design of monetary and exchange rate policies, which will significantly impact overall macroeconomic stability The findings will be useful for central bankers to understand the factors influencing money demand in the Lao PDR, especially taking into account of dollarization problems prevailing in the economy The study also updates the database of the Lao PDR financial statistics One important contribution of this study is in constructing economic variables especially annual GDP data to quarterly GDP data This provides a good starting point to study the relationships between the money balance and other economic variables in the economic framework Further, this study can be adopted to estimate money demand in other similar developing countries as the Lao PDR

Database

The data used in this analysis is taken from the Bank of the Lao PDR The

estimated sample uses quarterly data in the period from Q1/1993 to Q2/2010 Structure of dissertation

Besides the introduction, conclusion, appendices, references, this dissertation includes 4 chapters as follows:

Chapter I: Overview of theoretical and empirical studies on money demand

This chapter reviews the main theories of money demand in order to explore what factors can affect the demand for money It also presents some empirical studies on money demand The lessons learned from the literature survey will help select the appropriate modeling framework and choose suitable variables in the following chapters

Chapter II: Lao financial system and monetary policy

The overview of financial system development of Lao PDR will be presented in this chapter In particular, major developments of the banking sector in the last 20 years will be reviewed The monetary policy will be sketched with emphasis on the new role of the Bank of the Lao PDR The current pros and cons of BOL monetary policy raises the need to estimate the money demand function

Chapter III: Demand for money in Lao PDR

In this chapter, the empirical framework of money demand function for Lao PDR will be formulated and estimated, the specific problems in the case

of Lao PDR will also be discussed

Chapter IV: Policy implications

Based on empirical estimation of chapter III, some policy implications for not only the Lao government but also for the BOL will be suggested in this chapter

CHAPTER I OVERVIEW OF THEORETICAL AND EMPIRICAL STUDIES ON MONEY DEMAND 1.1 A brief theoretical overview

1.1.1 Quantity theory of demand for money

The quantity theory of demand for money proposes a direct and proportional relationship between the quantity of money and the prevailing price level This relationship emerges within the classical equilibrium framework using two separate, but equivalent expressions The first expression is associated with American economist Irving Fisher and is called the “equation of exchange” The second expression is associated with Cambridge University’s Arthur C Pigou is called the “Cambridge approach”

or the “cash balance approach”

• Fisher’s “equation of exchange”

The Fisher’s equation of exchange provides an important relation between four macroeconomic variables to determine the nominal value of aggregate income.1 The four variables in the equation of exchange are the

total amount money in circulation, M, an index of the total value of aggregate transactions, T, the price level of articles traded, P, and a proportionality factor V denoting the “transaction velocity of money” The equation is given

One should note that while Fisher developed the algebraic formulation of the equation of exchange in 1911,

it was John Stuart Mill who originally stated the relation, expanding on the ideas of David Hume

economists believed that wages and prices were completely flexible, they posited that the level of aggregate output produced in normal economic period,

Y, would remain at the full employment level, so Y is by definition is nation’s

total potential level of output Fisher assumed that the ratio between the level

of transactions, T, and output, Y, is reasonably stable ( Y = ×t T ) and hence T

can be treated as constant in the short-run

Fisher believed that the velocity of money, V, is determined by the

institutions in an economy, because these directly affect the way in which individuals conduct transactions For example, if consumers use charge accounts and credit cards to conduct their transactions, and consequently use money less often when making purchases, less money is required to conduct

the transactions generated by nominal income (M decreases relative to PT)

Hence, velocity, defined as ( )PT /M , will increase On the other hand, if consumers find it more convenient to purchase items with cash or checks (both of them are counted as money), more money is used to conduct the transactions generated by the same levels of nominal income, hence velocity will fall Fisher theorized that institutional and technological features of the economy that affect velocity change only slowly over time, so velocity can safely be considered constant in the short-run By dividing both sides of the

equation of exchange by V, we obtain the money demand function:

to conduct transactions and have no freedom of action in terms of the amount they want to hold The demand for money is determined by the level of

transactions generated by the level of nominal income, PY, and by the

institutions in the economy that affect the way people conduct transactions

that determine velocity, V, and hence k Therefore, Fisher’s quantity theory of

money suggests that the demand for money is purely a function of income Interest rates have no effect on the demand for money

• Cambridge approach to money demand

A group of classical economists, including Alfred Marshall and Arthur C Pigou in Cambridge, England, studied the demand for money by considering how much individuals want to hold, given a set of circumstances This approach is different from Fisher’s approach to money demand Fisher held the central assumption that individual demand for money is driven by the institutional environment, as this is the main factor that affects whether individuals use money (i.e., cash and check) to conduct transactions In the Cambridge model, the individual demand for money is not completely bound

by institutional constraints such as whether one can use credit cards to make purchases Instead, individuals desire money because money is a medium of exchange and a store of wealth Cambridge economists concluded that money demand would be proportional to nominal income and expressed the demand for money function as:

In the short–run, k is the constant of proportionality and money demand

does not depend on the interest rate However, money demand can depend on the interest rate when velocity is not constant over time

From our discussion so far, the quantity theory of money emerges as the theory with a simpler approach to estimating money demand The estimating equation is

MV =PY (1.4)

Where M denoted nominal money stock, V denotes the income velocity

of circulation, P denoted the prevailing price level and Y denoted the real

income

Note that the elegant expression for money demand given by the quantity theory of money relies on the assumption of constant velocity In reality, however, the velocity isdetermined by technological and/or institutional factors, which experience large changes, especially during periods of financial liberalization In these cases, equation (1.4) cannot capture the complex relationship between the money demand and other macroeconomic variables Hence, we will turn to two other approaches to the theory of money demand: the Keynesian approach and Friedman’s modern quantity theory approach Both approaches consider the demand for money as part of the general problem

of wealth allocation, but place emphasis on different aspects of the problems

1.1.2 Keynesian approach (John Maynard Keynes)

In 1936, Keynes famous book, “The General Theory of Employment, Interest and Money”, abandoned the classical view of economics In its place,

he offered a theory of demand for money that emphasized the importance of interest rates Keynes’ theory of money demand (referred to as liquidity preference theory), focuses on the factors that influence individual decision-making He postulated that there are three motives driving the demand for money: a transaction motive, a precautionary motive, and a speculative

motive With this view, money demand is a function of real income (Y) and the interest rate (r)

Equation (1.5) has the key implication that velocity is not constant and is positively correlated with interest rates, which fluctuates substantially Initially, Keynes suggested a liquidity-preference schedule like the following equation,

where dM1 /dY >0 The speculative demand depends only on the level of

interest rate, r, where dM2 /dr <0

The following sections review the work of other economists who have further developed the Keynesian theory of money demand

1.1.2.1 Transaction demand- Baumol-Tobin model

The transaction motive is the demand for money rising from the use of money as regular payment for goods and services The trade-off is between the amount of interest an individual forgoes by holding money, the costs and conveniences of holding money Baumol and Tobin set up their model by analyzing a per-period model of an individual managing his wage in every time

period, assuming that the person is paid in bonds an amount Y at the beginning

of a period An additional assumption is that the amount Y is spent uniformly

over the time period Each time the individual requires money for consumption,

he has to convert bonds to money and incur two costs in the process:

• Transaction costs, (t c), which represent the direct cost of converting bonds into money

• Interest opportunity costs, which are forgone interest earned by converting bonds into money

Solving to find the optimal number of transactions yields the minimum sum of transaction costs and interest opportunity costs This solution can be used to derive a function of individual-level money demand that can be aggregated across individuals The aggregate money demand function can then be expressed as:

within a given time interval (e.g., 1/t), where t is the number of transactions

within a period of time (e.g., one week) Assuming that money balances are

bounded below by zero and bounded above by h, Miller and Orr solve for the optimal return level z Note that the bounds imply that if the balances are below zero, then z dollars of bonds are converted to money On the other hand,

if the balances are above h, then z dollars of bonds are converted to a bond

2

Note that there are the same determinants of money demand as in the transaction model

Miller and Orr formulate the problem such that the optimal level of z

minimizes the expected sum of transaction costs and opportunity costs Using the same framework, they also solve for the optimal size of average cash

has a negative relationship with the interest rate, r The level of balances has a

positive relationship with income, where income is expressed as the variances

of changes in cash balances, m2 t

1.1.2.3 Speculative demand

Speculative demand for money arises from uncertainties about the money value of alternative assets that an individual can hold The portfolio balance model, developed by Tobin, has provided the most convincing theory

of speculative demand for money Based on an assumption that the return on bonds is risky while the return on money is certain, Tobin’s model analyzes

an individual’s optimal portfolio allocation between bonds and money The three most important components of this model are:

− The probability distribution of capital gain from bonds, where the risk

is measured by the standard deviations from the mean,

− The budget constraint, which represents the individual’s feasible portfolio allocation, and

− The utility map of the individual; (represented by indifference curves), which show the individual’s preferences over different portfolio allocations Tobin’s model implies that for a given level of wealth W, the amount of wealth to be help in the form of money is inversely related to the prevailing

interest rate Note that the distinguishing feature between different portfolios allocated is the balance between profitability and risk

Although the Keynesian approach to analyzing the demand for money focuses on the three motives for holding money, the models do not allow us to uniquely identify an individual’s particular motive for holding money However, this is not an important weakness of these models because all three motives together influence an individual’s optimal level of money holding Next we will review Friedman’s work (1956) as he developed a theory on the demand for money based on another approach: producer and consumer behavior

1.1.3 Friedman’s model of the demand for money

In 1956, Milton Friedman developed the modern quantity theory of demand in a famous article, “The quantity theory of money: A restatement”

He simply stated that the demand for money must be influenced by the same factors that influence the demand for any other asset An individual’s demand for money should be a function of his wealth and his expected relative (to money) return on alternative investments

Friedman developed his theory on the demand for money within the context of the traditional microeconomics theories of consumer behavior and

of the producer demand for input Consumers hold money because it yields a direct utility stemming from the convenience of holding an immediate form of payment Producers hold money because it is a productive asset which smooths the payment and expenditure streams over time Therefore, the sum

of demand for money by both consumers and producers is the demand for real balances Intuitively, this demand should depend on the level of real income (or real output) as well as on the returns of alternative assets such as bonds or durable goods (for consumers) Therefore, the equation below gives us the demand function for real balances:

( 1 2 )

/ , , , , n

Where rm is the demand for real balances and sequence r1, r2,…, rn

represent the real rates of return on alternative (i.e., non-money) assets

In particular, Friedman considers durable goods as an important category of alternative assets to money for consumers With this view, the demand for

consumers’ durable goods depends on expected inflation rates, π e, then, the demand function for real balances also depends on the expected rate of inflation

1.2 Some empirical problems in estimating money demand functions

Money demand functions were first conducted in developed countries where financial systems developed and central banks realized the role of money demand in conducting monetary policy However, lately there has been considerable interest among several other industrial and developing countries All empirical studies base on a conventional textbook formulation of a simple theoretical demand for money function, rm= f r Y( , ), relating demand

for real money balances (rm) to a measure of transactions or scale variables (Y) and the opportunity costs of holding money (r) However, the demand for

money functions estimated for difference countries are not the same Because

of the differences in definitions of dependent variables, available of scale variables, financial development,…

This section will discuss some issues related to estimating money demand function

lnrm t =b lnY t +b r t (1.12)

If the coefficients of a regression equation such as (1.12) were estimated using time series data, there would be no dynamic dependence in the model due to the lack of lagged explanatory variables However, time series data are typically dependent on their previous values, i.e , the data observed during the period t −1 and earlier Therefore, lagged explanatory variables are necessary to obtain the correct model specification using time series data Generally, errors in model specification are often reflected in specific tests such as those for residual autocorrelation and heteroscedasticity For example, a naive application of a static regression such as (1.12) to time series data will likely exhibit a strong autocorrelation in the residuals This autocorrelation occurs because information concerning dynamic relations through the data is not reflected in any way by the static regression, so all this information is captured in the “unmodeled” part of the model, i.e., the residual From a purely statistical perspective, it is hence appropriate for models with time series data to include relevant lagged variables

To sum up, the standard demand for money function should include lagged dependent variables as one of the explanatory variables The lagged variable lnmt-1 has also been justified on the basis of dynamic models that incorporate partial adjustment or adaptive expectations

In studies of Asian countries (reviewed in the next section), the definition of

narrow money, M1, included only cash and demand deposits while in the American studies, M1 includes cash, travelers’ checks, demand deposits and

other checkable deposits (Goldfeld and Sichel, 1990)

Economists have shown that studies that interchange the use of M1, M2,

or M3 to estimate the demand for money face the problem of estimating

heterogeneous assets For example, cash and demand deposits may differ significantly in terms of transaction costs, financial risks, and the ease of concealing illegal or tax-evading activities One solution is to separately estimate the demand functions for cash and demand deposits This approach has yielded more robust empirical results, but it does not resolve the underlying empirical difficulties Any analysis on Lao PDR will face similar issues regarding the definition of money and should leverage the advances made by economists to deal with these empirical difficulties

Scale variables

Recently, scale variables were typically created by using data on a

country’s GNP, permanent income or wealth, measured in real terms A number of other related variables that move together with Gross National Product (GNP) such as net national product (NNP) and GDP have also been heavily utilized in creating scale variables without any significant differences induced by their substitution 3 Traditionally, GNP has been used for transaction-oriented models, while the modern-quantity theories relied on permanent income Permanent income is most frequently measured as an exponentially weighted average of current and past values of GNP Since permanent income is often viewed as a proxy for wealth, the direct measures

of wealth have also been used Given that financial transactions can generate

a demand for money, the use of wealth is also consistent with a transaction view of the demand for money

Empirical studies show that the choice of the scale variable must be tied

to the sample period, the definition of money, and the specific country context When transactions-oriented models began to “misbehave”, it is natural to examine whether permanent income or wealth might improve the model’s results In more recent years, research on how to obtain scale variables has focused on two measures of transactions: the need to construct more comprehensive transaction measures, and the disaggregation of transactions into its various components, reflecting the observation that not all transactions are equally “money intensive”

The construction of more comprehensive measures is motivated by the fact that even GNP is less inclusive than a more general measure of transactions In particular, the GNP excludes all scales of intermediate goods transfer, purchases of existing goods, and financial transactions, all of which may contribute to the demand for money The GNP may also overstate

3

IMF Working paper, “Survey of literature on demand for money”, 1999, P.21

transactions because it includes impute items While these shortcomings of using GNP have been recognized for a long time, data limitations have prevented the construction of a more general measure of transactions Some researchers such as Judd and Scadding (1982a) substitute debits to GNP, but the effect of this substitution is small

Whichever measure of transactions is ultimately chosen, the question of whether it can be disaggregated into several scale variables remains an open question For example, if real GNP is the basic scale variable, one might separately enter some different components of GNP as each category is likely

to generate a different payment need For example, one might posit that consumption is the most money-intensive component of GNP However, while there are some promising indications, there is no clear evidence that disaggregated measures of GNP yield worthwhile improvements to understand the aggregate demand for money Economic aggregate proxies for scale variables in estimating demand for money function depend much on the development of statistical systems and available data

Opportunity costs of holding money

Interest rates in money demand includes two groups: the own-rate of money and the rate of return on alternative assets Tobin (1958) and Klein (1974) argue that both of these rates are important and should be included in any model for money demand However, perhaps for ease of empirical analysis, several economists in the U.S treat the explicit own-rate as zero in the period before 1980

Regarding the return on alternative assets, the researcher has several choices Those with a transactions view of demand for money use more short-term rates like the yields on government securities, commercial paper, or saving deposits The argument for using these alternatives is that they are

closer substitutes for money and hence their yields are especially relevant when deciding how much cash to hold Those considering a less narrow view

of the demand for money can use a boarder set of alternatives that includes the return on equities, yields on long-term government or returns on corporate bonds Previous work has also incorporated foreign interest rates and the expected rate of depreciation for domestic currency

In countries where the financial sector is not highly developed and also suffers from hyperinflation, the expected rate of inflation is also a useful variable to calculate the opportunity cost of holding money The reasons for using this variable are four-fold4: first, there is a limited substitution between

money and alternative assets due to the underdeveloped financial markets

outside the banking system; second, interest rates may show insufficient

variations for a long period of time because they may be regulated by the

government; third, payment of interest is legally prohibited in some countries; and finally, time series data on the interest rates may simply not be available

1.2.3 Empirical estimation problems

The ‘missing money”

The number of empirical studies on the demand for money function exploded between the 1960s and the 1970s with American economists leading the movement, according to Tomoo Yoshida (1990), these studies stayed with the frame of a particular adjustment mechanism to estimate the “standard” money demand function according to traditional macroeconomic theory It is interesting to note that in the U.S., the interest rate generated a significant effect on the amount of money demanded Equation (1.13) is a simplified specification of a model from this time period:

where rm (equal to M/P) is the real demand money, Y is real income and

r is the interest rate Note that rm and Y are transformed to the natural

logarithmic scale in equation (1.13)

Goldfel (1973) is particularly representative of the studies during this period In fact, Goldfel’s work is often regarded as the culmination of the far-ranging research that took place during this period Equation (1.13) includes a one-period lag in the real money balance as an explanatory variable in

addition to income (Y) and interest rate (r) The basic form of this model has

survived time and many researchers still use this specification

In the mid-1970s, the consensus on the stability of the money demand function began to break down In 1976, Enzler, Johnson and Paulus pointed out that the poor performance of the 1973 models on the out-of-sample, post-1974 period The problem was that the 1973 models consistently overestimated money demand in the post-1974 period Goldfel (1976) coined the term “missing money” to explain this phenomenon He argued that the overestimates were at least partly attributable to a shift in the money demand function resulting from financial innovation such as the introduction of NOW accounts and MMMFs

In the U.S., tracing the “missing money” renewed interest on studying the stability of the money demand function These studies can be divided into three broad categories:

• Attempts to improve the explanatory power of the money demand function by including explanatory variables such as wealth and bank debits,

• Attempts to incorporate the explicit effect of financial deregulation into the money demand function by introducing the racket effect of high interest rate, and

• Attempts to better incorporate effects of financial deregulation on money supply by adopting artificial money supply aggregates as additional model components, i.e., as additional dependent variables

Spurious regression

The problem of spurious regression arises when a regression equation contains independent variables that are random walk or non-stationary variables Nelson and Plosser (1982) conducted ADF tests using annual data on key macroeconomic variables in the U.S., such as nominal and real GNP, industrial production index The number of people employed, the unemployment rate, the GNP deflator, the consumer price index, the nominal and real wage indices, the money supply, the interest rate and the price of publicly traded stocks To their own surprise, Nelson and Plosser could not reject the null hypothesis that each of these variables followed a random walk pattern, with only one exception: the unemployment rate

In time series analysis, the properties of each series of data must be clearly known since estimation methods such as ordinary least squares only work when data series is stationary If the regression equation includes one or more non-stationary variables, familiar statistics such as the coefficient of determination and t-values no longer have conventional distributions Consequently, these types of regression equation are likely to lead us into incorrect inferences

To address this issue, Phillips (1986) investigated the case of regression between two mutually independent random walk variables The highlights of his findings are summarized below:

• Since the conventional t-values do not have a limiting distribution, it does not have an asymptotically correct set of critical values If the set of critical values for a normally distribution variables is used, the probability of finding a significant relationship between the two variables will increase with the sample size

• The coefficient of determination (R2

) has a non-degenerate limiting distribution, hence we expect moderate values of R2

• As the sample size approaches infinity, the Durbin-Watson (DW) ratio approaches zero

To avoid the problems with spurious regression as demonstrated by Granger and Newbold (1974) and Plosser and Schwert (1978), Hafer and Hein (1980) and Gordon (1984b) showed that it is preferable to estimate a money demand function such as (1.14) that contains a first differenced term

In this case, the specification becomes:

1.3 Some Asia-specific studies on the money demand function

A large body of literature is available to estimate money demand functions Initial work in this area was confined primarily to industrial countries, especially the U.S and the U.K However there has also been considerable attention paid to studying the money demand function in developing countries in Asia and South East Asia Various central bank officials realized that understanding the money demand function is a cornerstone of monetary policy In this section, the set studies are carefully chosen on the basis of potential relevance to the Lao PDR context

• Fan and Liu (1970) 5

demand for real money balances The authors use actual income (not permanent income) in their analyses and justify this by arguing that the concept of permanent income is meaningless in developing country economies that are not stable

Fan and Liu find that the opportunity cost of holding money is given by the call money interest rate in Japan, Korea, Philippines, Thailand, and Myanmar For countries like Taiwan, India, Pakistan and Sri Lanka, the opportunity cost of holding money is given by the government bond yield Fan and Liu found that the effects of interest rate on the demand for money are significant and have mixed signs In some countries, it has positive effects, but in the others, negative effects

In the case of Lao PDR, the interest rate has an effect on the behavior of the money market For context, the government has liberalized the economy, including money market If the difference between the interest rate and the return one can obtain via the black market is very high, then many commercial banks face difficulties in collecting money via savings from consumers However, the number of private commercial banks is increasing

in Lao PDR This market entry partly reflects the observation that the country’s interest rates are responsive to the market’s signal Thus, the higher the interest rate, the lower the demand for money In the Lao PDR, individuals face the negative opportunity costs of holding money

• Aghevli et al (1979)

Hossian and Chowdhury (1996, p 165) also highlight the study of

Aghevli et al (1979), which covers six Asian countries: Indonesia, Malaysia,

Philippines, Singapore, Sri Lanka and Thailand The study treats the demand for money as a function of actual income, expected rate of inflation and lagged real money balances In their model, the opportunity cost of holding

money is given by the expected inflation rate and not the interest rate The period of data analyzed is 1975 to 1978, as these are three years during which all of the countries in the study had been following a policy of financial repression by controlling interest rates In these countries, the monetary authorities typically kept the interest rate below the market-clearing rate The estimated result from using a partial adjustment mechanism framework shows that the demand for real money balance has a significant relationship with real income, the rate of inflation and the lagged real money balance The only exception to this is Singapore, where the coefficients are not significant at the 95% confidence level

Estimating the money as demand function by including the expected inflation rate as an explanatory variable has been used to study hyperinflation

as well For example, Kate Phylaktic and Mark P Taylor use this type of model to study the monetary policies in Argentina, Bolivia, Brazil, Chile and Peru during the 1970s and 1980s These studies find that inflation is a statistically significant and useful predictor of the demand for money Thus, studies that are undertaken to understand hyperinflation during the 1980s and 1990s in Lao PDR would likely benefit by incorporating inflation as an explanatory variable in the model

• Khan (1980)

Khan studied the demand for money in 11 countries: seven Latin American countries and India, Malaysia, Philippines and Thailand The study

is worth noting because of the amount of data used in the empirical analyses,

as Khan studied these countries from 1962 to 1976 (Hossian and Chowdhury

1996, p 165) The study confirms the results documented by Aghevli et al

(1970): there are significant relationships between real money balances, actual income, and expected rate of inflation

• Tseng and Corker (1991)

Tseng and Corker specified an error-correlation model for their study Their study uses the data from the period of financial liberalization (1970-1989) and shows that interest rates had moved in line with the actual economic situation The implication is that the opportunity cost of holding money is given by the interest rates and not the expected rate of inflation Table 2.1 shows that income and interest rates have a significant relationship with the demand for money The sign of income is positive and interest rate is negative Therefore, in order to preclude spurious results, we must test for unit roots before choosing an estimating equation Another observation for table 1.1 is that in the hyperinflation countries, inflation is a better predictor

of money demand than the interest rate

It is useful to note that in financially liberalized countries, income and interest rates are fitted explanatory variables

Table 1.1: Estimated income elasticity and interest rate semi-elasticity

for selected Asian countries

Period of study (year and quarter) Country Income elasticity

Interest rate Semi-elasticity 1974II – 89IV Indonesia 1.16(28.1) -0.66(3.0) 1970I – 89IV Korea 0.79 (35.6) -0.84 (3.6) 1970I – 89IV Malaysia 1.11 (74.0) (No data) 1970I – 89IV Myanmar 1.27 (37.9) (No data) 19970I – 89IV Nepal 1.75 (48.2) (No data) 1973I – 89IV Philippines 0.67 (9.0) -1.16 (4.4) 1975I – 89IV Singapore 0.86 (55.4) -1.17 (5.65) 1978I – 89IV Sri Lanka 0.92 (12.8) -1.60 (5.6) 1977I – 89IV Thailand 0.85 (40.8) -1.53 (6.1)

Note: Parentheses indicate t-statistics Source: Tseng and Corker (1991)

• Watanabe S and Pham T B (2005) 6

Watanabe and Pham tried to find the relationship between broad money and inflation by decomposing the demand for money into the demand for domestic currency and the demand for foreign currency in view of the fact that Vietnam is highly dollarized The empirical analysis employs co-integration, error correction model, impulse response and variance decomposition and uses quarterly data over the period 1993-2004

This paper found that the long run demand for real broad money of domestic currency is determined by real income, domestic interest rate, inflation rate and rate of return of USD deposits, which satisfies the standard properties of the demand for money The long run demand for real foreign currency deposit is determined by real income and the difference between rates of returns of foreign and domestic currency deposits, that represents asset substitution Demand for real foreign currency deposits is found to be very sensitive to the difference between rates of returns of foreign and domestic currency deposits, especially exchange rate depreciation A positive difference between rates of returns of foreign and domestic currency deposits triggers a shift from domestic financial assets and real assets to foreign financial assets, resulting in a lower inflation rate and a strong increase in demand for foreign money When this effect is very strong, then even an increase in broad money, including foreign currency deposits, may negatively correlate with inflation

• Nguyen, D H., and W D Pfau, (2010) 7

This paper investigated the money demand function in Vietnam by using

6

Watanabe, S and Pham, T.B (2005) “Demand for Money in Dollarized, Transitional Economy: The Case

of Vietnam” Paper presented at the 1st VDF-Tokyo Conference on the Development of Vietnam

7

Nguyen, D H., and W D Pfau, (2010) "The Determinants and Stability of Real Money Demand in Vietnam, 1999-2009." GRIPS Discussion Paper 10-14 Tokyo: GRIPS

co-integration analysis and a reduced-form short-run error correction model over the period of 1999-2009 It found evidence for a co-integrating relationship between the real money demand, income, the foreign interest rate, and the real stock price More importantly, statistical tests show that real money demand in Vietnam is stable in this period

Although Watanabe and Pham found a statistical significance for domestic interest rates in 1993-2004 for the long-run analysis of real money demand, the estimation in this study provides no evidence that domestic interest rates influenced long-run real money demand in the period from 1999-2009 This implies that a rigid interest rate policy might become ineffective when the economy is opening, following milestones such as the establishment of the Vietnam-US Trade Agreement in December 2001 and accession to the WTO in November 2006 In the context of Vietnam's increasing integration into the world economy, interest rate policies need to be reformed towards market principles in order to improve the effectiveness of monetary policy

Conclusion of chapter 1

The demand for money is a proportion of income level In addition, most

of the models assume a relationship between money demand and transactions The specific relationship is context-specific with regard to the industrial structure, financial development, and institutional arrangements of a particular country in a given time period For example, two countries with the same income level may have very different aggregate demands for money if one country has an agro-based economy while the other has a service-based economy We might expect the country with an agro-based economy to have

a higher aggregate demand for money, perhaps because the transaction velocity in agriculture is lower than that in services Another example of how two countries with the same income level can have a developed credit system,

as this can reduce the demand for narrow money, M1

A demand for money is constrained by a measure of wealth that can be proxied by either income or permanent income

The demand for money fluctuates with changes in the opportunity costs

of holding money This opportunity cost depends on the relative return on non-money assets such as other financial investments and real goods

Expectations are important The demand for money depends not only on the prevailing level of factors such as the interest rate and inflation, but also

on the future expected values of each of these factors

In the case of dollarization, interest rates of dollar and exchange rates are also an interesting explanation for demand for money balances

In the developed countries, the nominal interest rate considers an appropriate proxy for the opportunity cost of holding money, whereas the weak financial markets and administrative interest rates are the overriding feature in most of the developing countries In most developing countries the nominal interest rate is institutionally determined; it does not fully capture the opportunity cost of holding money Furthermore, administrative nominal interest rates are not often adjusted for changes in inflation and consequently the real interest rate becomes negative To overcome this problem, , researchers often use the consumer price index as the proxy for the interest rate variable In fact, the asset substitution in developing countries usually occurs between money and real assets as inflation hedges and not between money and other financial assets Thus the expected rate of inflation rather

than the nominal interest rate can be regarded as a better proxy for the opportunity cost of holding money in developing countries

In the next chapter, an overview of the financial system and current monetary policy will be presented With this, the researcher can choose suitable variables for estimating demand for money function for Lao PDR

Figure 1.1: Determinants of money demand function

Vector of Opportunity Cost Variable

Nominal Interest rate

Inflation rate Exchange rate

Foreign Interest rate

Assets Substitution Currency Substitution

Foreign Factors