Professional Documents
Culture Documents
Objective
20.1 Introduction
20.2 Problems in the Estimation
20.3 Indian Data- Demand function
20.4 Summary
20.5 Glossary
20.6 Model Questions
20.7 Further Readings
20.1: Introduction
Already studies the different theories of demand for money propounded by different
schools of economic thought. And also identified important determinants of demand for
money. In this lesson, made an attempt to explain how an empirical Demand Function
for Money in India would be estimated. The theoretical discussions outcomes are the
demand for money of a country is an increasing function of its wealth or its income, and
decreasing function of rates of returns from various non–monetary assets. Many
economists made attempts to estimate empirical demand functions for India. Among
them, Dew Laidler’s work of 1960 is an important contribution in the field of demand
function for money. In India also many economists like, A.M. Khusro (1976), A.K.
Lahiri (1977), A. Vasudevan (1977), S.B. Gupta (1977), Kaur (1998), SB Gupta, (1977)
done estimations on demand function of money in India.
Before we present their findings, it will be useful to enlist, the important theoretical and
statistical problems that arise in estimating the demand functions. They are as follows:
What will be the measure of money supply that will be relevant currently, M 1, M2,
M3, or M4?
Which rate or rates of interest and returns should be taken into account?
Depending on the theory, one may specify the function in terms of real or nominal
balances. However, from a rational point of view, real balances concept is
appropriate because what people want to hold is certain purchasing power and not
a quantity of money.
In the asset portfolio of the public, money and a variety of assets compete with
each other. They yield different rates of return with different degrees of risk and
convenience. In view of this, one cannot rely on a single rate of return, but at the
same time, for empiricists, incorporation of all these returns in the function is
neither practicable nor meaningful (due to problems of multi-collinearity and loss
of degrees of freedom). Hence, each researcher has been found to be selecting a
few important representative rates of returns arbitrarily.
Basutkar, Tirupati, the main objective of his paper on Money Demand in India is to
derive money demand function for the Indian economy over the period 2004-2014.
Secondly, it underscores the effectiveness of Broad Money (M3) as a significant
contributory factor towards the derivation of money demand function in India. Right
from the classical version of quantity theory of money to more modern Friedman’s
version, there exist plenty of approaches, both at theoretical and empirical levels, to
compute money demand function. Baumol (1952) and Tobin (1956) independently
developed a model based on the tradeoff between the liquidity and its cost.
Empirically, the real demand for money function can be computed of as follows:
Md (= M) = β1 + β2(Y) – β3(i) + u (II)
P
Following the Quantity Theory of Money approach, money supply should be equal to
money demand. Any excess in money supply will be reflected in the rise of inflation
rate.
Empirical Estimation of Money Demand function: Following the methodology
developed in section two, both narrow and broad money variables for 2004-2014 were
divided by WPI for the same period to arrive at the real money demand function for the
Indian economy. Nominal GDP were weighted with respect to WPI to arrive at real
GDP values for the said period. As for the opportunity cost of holding money, inter-
bank call rate for the stated period, were made use of. Derivation of regression equation
using the processed data revealed the following estimates:
Regression:
M1/WPI v/s Real_GDP, Call rate
Model Summary
Derivation of demand function using narrow money (M1) data for 2004-2014 does not
yield a convincing equation especially with low Durbin–Watson statistic and a greater
than 5% p-value for call rate coefficient. The situation changes entirely when we use
broad money (M3) data.
Regression:
M3/WPI v/s Real_GDP, Call rate
Model Summary
Regression equation:
Both Durbin-Watson statistic and p-value for all the coefficients are within acceptable
range thereby making the money demand function using M3 more representative.
At the outset, Equation III confirms to the theoretical underpinnings of money demand
function, i. e. demand for money is positively correlated to GDP and negatively
correlated to interest rate. That with regards to Indian economy, real demand for money
increases by 0.011 points for every one point increase in real GDP and decreases by
6.950 for every one point increase in the call rate, the proxy for opportunity cost of
holding money.
In keeping with the objective, it can be stated with a fair degree of confidence that broad
money (M3) is far more significant in deriving money demand function as compared to
narrow money (M1) in India, over the period 2004-2014. From an empirical point of
view, if the policy objective is to influence money demand function in India, targeting
M3 may yield better results.
(B) Amrita Sarkhel etal: Estimation of the Money Demand function for India and
China (1995-2015)
Estimation of the Money Demand functions for India and China Amrita Sarkhel etal. the
main objective is to study the variation in money demand function of two developing
countries, India and China. To study the practical implementation of the money demand
function and how strongly demand for money is correlated to changes in interest rates
and national income.
We have extracted data from different sources for GDP(USD Billions), M1(USD
Billions), Interest rate( Call money rate%) and CPI and then we have done a regression
analysis with dependent variables being Real money demand and independent variable
being GDP and interest rate.
Comparing the behavior of money demand function with reference to GDP and Interest
Rate. We are using the below function for our calculation:
L(Y, r) = 𝑀 𝑃 𝑑 = L0 + L1 Y − L2 𝑟
M= M1= Money Supply P= Price level Y= Real GDP at market price r= Call Money
Rate L0=Autonomous demand L1= Income sensitivity of money demand L2= Interest
sensitivity of money demand.
We are estimating the money demand of India and China for the last 21 years (1995-
2015)
Sample Data for India: The money demand is estimated linearly using regression and
checked at 5% level of significance. Real GDP is taken at the end of the year with base
year as 2005. Money supply is taken on monthly average. Money supply is divided by
CPI to get real money balance. Interest rate is money call rate and it is average of
months. CPI is taken as monthly average SOURCES: Real GDP- International Monetary
Fund, USA CPI & M1- Trading Economics Interest Rate – OECD Database
From the above statistical analysis, we can derive the money demand function of India as
above:
Sample Data for China: The money demand is estimated linearly using regression and
checked at 5% level of significance. Real GDP is taken at the end of the year with base
year as 2010. Money supply is taken on monthly average. Money supply is divided by
CPI to get real money balance. Interest rate is money call rate and it is average of
months. CPI is taken as monthly average
From the above statistical analysis, we can derive the money demand function of China
as above:
Interest sensitivity of demand for real money (L2) is lower for India (0.12) than china
(0.55) The income sensitivity of demand for real money ( L1) is lower for India (0.001)
than China (0.004) .The Autonomous Demand (L0) is lower for India (2.04) than
China(5.24)
At the outset, both the money demand functions of China and India are conforming with
the theoretical demand function i.e. Money demands is directly proportional to real GDP.
Money demand is inversely proportional to interest rate.
For both the countries, real money demand rises less than proportionately than the rise in
real income
Since the slope of the LM curve for India is steeper any changes in fiscal policy will be
less effective as compared to China.
Demand function of money in India estimated by SB Gupta for the period 1950-76 is
presented in the fallowing. His regression equation is:
The above function is linear one as it is in the log form. Gupta holds that the demand
function in logarithmic form gives a better result compared to the simple linear form
(except for interest rate variable). In the equation, (M/P) d denotes demand for real
balances, Y the real national income, P the market price level, r the twelve month time–
deposit rate of commercial banks, R2 the adjusted coefficient of determination, D the
Durbin-Watson statistic (as a measure of the first order serial correlation among the
residuals) and the figures in the brackets are t values of the coefficients.
The equation indicates a very good fit as the independent variables are able to explain
94% of the variations in the log (M/P); further, there is no serial correlation among the
residuals at 1% level of significance.
Main Findings:
A rise in the price level lowers the demand for money. Many countries, including
India which experience rapid inflation, are characterised by negative relation
between demand for money and expected rate of inflation, and therefore, It means
that higher the expected inflation rate, the lower will be the demand for money.
Another demand functions for money estimated by Gian Kaur on his PhD work, for
India, covering the period 1950-85. However, as the results were not found to be
statistically significant and consistent, he restricted the study period to 1950-80. In order
to test whether the demand for money remained stable, he divided the time span into to
sub-periods (1950-51 to 1962-63 and 1963-64 to 1979-80). The model used by him to
estimate the demand for money is also in log form and is presented below:
Where ‘M’ is the desired amount of money holding. ‘Y’ is the income, ri, is the rates of
interest, ‘P’ is the price level, Ut is the disturbance term. B1, B2, B3 are the elasticities of
demand for money with respect to income, rate of interest and price level in that order.
Kaur estimated a number of demand functions both for M1 and M3 in real as well as in
nominal terms. He estimated an aggregate (i.e., money components taken as a whole)
function as also component–wise (currency, Demand Deposits & Time Deposits)
functions. The Reason for estimating component–wise functions are that over time
structural changes in the components of money supply are likely to take place. For e.g. in
India, the share of currency in the total money supply decreased from about 70% in the
1950s to 60% in the 1970, and further to 50% in the 1990s to 57% (in M 1). There
occurred significant changes in the relative shares of demand and time deposits in M 3 to
capture the impact of these changes, Kaur estimated component-wise functions.
Main Findings:
Whether money is defined inclusive or exclusive of time deposits, is another question that
one should dwell on. More than 90% of the changes in money demanded have been
explained by three factors, interest rate, and income and price level. However, the role of
interest rate and income elasticities differ substantially depending on the definition of
money balances used. Demand for money (both M1 and M2 i.e. narrow and broad
definitions respectively) was found to be highly income elastic. As noted in the findings
of Gupta, income elasticity was greater than unity, indicating that cash balances were
treated as a luxury good in India. The estimates of income elasticities are 1.048 and
1.583 for M1 and M3 respectively. The implication of higher income elasticity with
respect M3 is that as real incomes rise, the public prefer to hold their money more in the
shape of time deposits, rather than demand deposits and currency. With regard to the
impact of interest rate on the demand for money, a negative relation was noticed. The
negative relation holds well whether money is defined inclusive or exclusive of time
deposits (i.e. M1 or M2) and whether the interest rate pertains to the short-term or long-
term. However, for the 1 year commercial banks deposit rate and for the government
securities, the demand for real M1 and M3 balances was interest elastic.
With regard to the relation between the demand for real and nominal balance and price
level negative relation was found. It means that when price level rises, demand for real
balances decreases.
Kaur also tested whether the demand function for money in India was stable or unstable.
To test it, the method of dummy variables and BDE (Brown, Durbin and Evans) tests
were used. They revealed that more inclusive is the definition of money (i.e. M 3 rather
than M1), greater was the instability. Time deposits demand was found to be more
unstable than demand deposits demand. It means that the demand for real M 3 is more
unstable than the same for real M1. The only exception he found was with regard to the
demand for real currency balance (Note: the difference between nominal and real
balances is that nominal balances adjusted for prices will give real balances (M/P = Real
Balances). In other words, in Kaur’s period of study, the demand for real balance in
terms of M1 and M2 (i.e. M1 / P and M2 /P) was unstable and for real currency demand, it
was stable.
Based on the empirical studies, we may conclude that demand for money in India
depends, among other factors, mainly on three factors viz. Income, interest rate and price
levels. But we must also keep in mind all the statistical and conceptual limitation of such
empirical studies. Further, the study also brings out the fact that expected change in
prices and changes in the composition of rural urban income shares effect the demand for
money in India. It appears that the demand for money function in India is somewhat
unstable. However, the issue of stability or instability remains unsettled. For policy
purpose, there is need to study disaggregate broad sector-wise demand functions such as
the demand functions for the business, rural & urban households, government, etc.
20.4 Summary
In this lesson, an attempt was made to explain aspects regarding the Empirical Demand
Function for Money in India. If any economy wants to ensure efficiency of its monetary
policy, a prerequisite is the proposition of a relevant demand function for money.
There are many theoretical and statistical problems that arise in estimating the demand
functions and in the case of underdeveloped countries like India the task becomes even
more complex.
Based on the results of the empirical studies, we may conclude that demand for money in
India depends, among other factors, mainly on three factors viz. Income, interest rate and
price levels. But we must also keep in mind all the statistical and conceptual limitation of
such empirical studies. Further, the studies also brings out the fact that expected change
in prices and changes in the composition of rural urban income shares effect the demand
for money in India. It appears that the demand for money function in India is somewhat
unstable. However, the issue of stability or instability remains unsettled. For policy
purpose, there is need to study disaggregate broad sector-wise demand functions such as
the demand functions for the business, rural & urban households, government, etc.
20.5 Glossary
Elasticity of Money
Demand Deposits
Time Deposits
Narrow Money
Broad Money
2. Amrita Sarkhel etal. “Estimation of Demand for money in India and China”Goa
Institute of Management. Goa
5. Indian Economic Journal Special Numbers on Monetary economics’ for various years.