Price hikes, economic reforms and causality in money, income and prices: evidence from Pakistan.
Husain, Fazal ; Rashid, Abdul
This study extends the analysis of causality by Husain and Rashid
(2008) by examining the shift in the variables due to the price hikes in
Pakistan in the early 1970s. We investigate the causal relations between
real money and real income, between nominal money and nominal income,
and between nominal money and prices using the annual data set from
1959-60 to 2003-04. Moreover, we examine the stochastic properties of
the variables used in the analysis, and take care of the shifts in the
series due to price hikes and liberalisation measures through dummy
variables. The results indicate significant shifts in the variables
during the sample period. In this context, the shift that occurred due
to price hikes in the early 1970s seems to be more important to be
incorporated in the analysis. The study finds the active role of money
as the leading variable in changing prices without any feedback. In the
earlier studies on income the feedback mechanism of money is found
missing perhaps because of overlooking the shift in the macro economic
variables in the early 1970s.
JEL classification: E3, E4, N3
Keywords: Money, Income, Prices, Price Hikes, Causal Relations,
Pakistan
I. INTRODUCTION
Money, Income, and Prices are important macroeconomic variables
which play a crucial role in an economy. In this context, the role of
money in the determination of income and prices has long been debated
particularly between the Keynesians and the Monetarists who hold
opposite views in this regard. The Monetarists claim that money plays an
active role and leads to changes in income and prices. In other words,
changes in income and prices in an economy are mainly caused by the
changes in money stocks. That is, the direction of causation runs from
money to income and prices without any feedback. The Keynesians, on the
contrary, argue that money does not play an active role in changing
income and prices. In fact income plays the leading role in changing
money stocks via demand for money implying that the direction of
causation runs from income to money without any feedback. Similarly,
changes in prices are mainly caused by structural factors.
The empirical evidence in this regard also remains inconclusive.
For example, Sims (1972) examining the causal relationship between money
and income in the US economy found the evidence of a uni-directional
causality from money to income. Similarly, Brillembourg and Khan (1979)
using a longer data set found a unidirectional causality from money to
income and prices in the U.S. as claimed by the Monetarists. However,
the other studies on the issue reported opposite or different results.
For example, Williams, Goodhart, and Gowland (1976) found unidirectional
causality from income to money in the UK economy as suggested by the
Keynesians. On the other hand, Barth and Bennett (1974), and Lee and Li
(1983), found the evidence of a bi-directional causality between income
and money in the economies of Canada and Singapore. However, regarding
money-prices causality, the evidence seems to be consistent as the
results of Williams, Goodhart, and Gowland (1976) and Lee and Li (1983)
are similar to that of Brillembourg and Khan (1979); that is, a
unidirectional causality from money to prices.
In Pakistan too the issue has long been investigated but with
different results. For example, Khan and Siddiqui (1990) found
uni-directional causality from income to money and bi-directional
between money and prices. On the other hand, Bengali, Khan, and Sadaqat
(1999) found a bi-directional causality between money and income and
uni-directional from money to prices. Abbas (1991) also found
bidirectional causality between money and income in Pakistan while
performing the causality test in Asian countries. Jones and Khilji
(1988) while analysing causal relationship between money and prices in
Pakistan found the evidence of a bidirectional causality with money
supply leading. But Siddiqui (1989) found bidirectional causality
between the two with prices leading. Finally, Husain and Rashid (2008)
in a comprehensive investigation on the issue which covers a longer data
set, uses both the real and nominal terms of money and income, and
mindful of the shifts in the series due to the economic liberalisation
programme, found the evidence of a unidirectional causality from income
to money and from money to prices. The study does not find the role of
money in increasing national income even after taking care of
liberalisation measures.
This study, extending Husain and Rashid (2008), attempts an
investigation of the causal relationship between money and income and
between money and prices while being cognisant of another important
shift in Pakistan's economic data. The price hikes in the early
1970s, generally termed as Oil Price Shocks, had significant impact not
only on the economy of Pakistan but also on the economies the world
over. We investigate the causal relations between real money and real
income, between nominal money and nominal income, and between nominal
money and prices using the data set from 1959-60 to 2003-04 with due
regard to the stochastic properties of the variables used in the
analysis. In addition, we take note of the two shifts, that is, the
shifts due to price hikes as well as due to economic reforms together.
The rest of the paper is organised as follows. The next section
discusses the data and outlines the methodology to test the stochastic
properties of the variables and their interrelationship. Section III
presents the descriptive statistics regarding money, income, and prices
as well as the stochastic properties of these variables. Sections IV, V,
and VI examine causal relations between real money and real income,
nominal money and nominal income, and nominal money and prices
respectively. The final section contains the summary and conclusions.
II. DATA AND METHODOLOGY
We use annual data from 1959-60 to 2003-04 to investigate the
causal relations of money with income and prices in Pakistan. The Gross
National Product (GNP) at current prices and constant prices of 1980-81
are used as nominal and real income, broad measure of money (M2) and GDP
Deflator with base 1980-81 are used as Money and Prices, respectively.
Finally, real money is obtained by deflating M2. The principal data
source is the National Accounts of Pakistan, prepared by the Federal
Bureau of Statistics. The other data sources include Economic Surveys by
the Finance Division and Annual Reports by the State Bank of Pakistan.
We start by presenting the descriptive statistics that show the
basic characteristics of the variables used in the analysis. The formal
investigation, however, starts with examining the stochastic properties
of the variables used in the analysis. Hence, the Unit Root Test is
performed on the variables to test the stationarity of the variables. In
this context, the widely used Augmented Dickey Fuller (ADF) test is
used. We also use Phillips-Perron (PP) tests which is robust to a wide
variety of serial correlation and heteroskedasticity, where the
truncation lag parameters are determined following Schwert's
(1987). Next, we apply the Engle-Granger Co-integration test to explore
the long run relations among the variables. Finally, the causal
relationships between these variables are examined through Granger
causality and/or Error Correction Models (ECM). In all cases lag lengths
are decided on the basis of minimum Final Prediction Error (FPE) and
Akaike Information Criteria (AIC).
The sample period, 1959-60 to 2003-04, has been subjected to
various changes due to economic and political events. In this context an
important event that is likely to significantly affect the variables
used in the analysis is the economic liberalisation programme started in
the early 1990s. Husain and Rashid (2008) taking note of the event did
not find any significant change in the role of money in the causality
analysis. We extend their analysis by referring to another event that
significantly affected the macro variables in Pakistan in the early
1970s i.e., the price hikes that in fact affected the economy
significantly around the world. Moreover, we take note of the two events
together. Hence we include a dummy from 1972-73 onwards to reflect the
effects of price hikes and a dummy from 1991-92 onwards to refer to the
economic reforms.
III. MONEY, INCOME, AND PRICES IN PAKISTAN
We start, following Husain and Rashid (2008), by presenting the
descriptive statistics of the variables used in the analysis for both
the full sample and the two subsamples. Though the number of
observations in sub-sample I is very low relative to the observations in
sub-sample II, it can provide some insights regarding the percentage
changes in the variables. The results are shown in Table 1.
The table shows an average annual expansion of around 13 percent in
nominal money. With an expansion of around 7 percent in prices, the real
money has expanded by 6 percent. Similarly, nominal and real incomes
have increased over time with an expansion of 12.6 percent and 5.4
percent respectively. The table further shows the descriptive statistics
for the two sub-samples. Sub-sample I covers the period before the price
hikes, whereas sub-sample II represents the periods after the price
hikes. Moreover, we also conduct the tests for equality of means and
variances between the two sub-samples, The results indicate significant
increase in the means of the two nominal variables along with prices. In
fact, the average expansion in prices has increased by three times in
the second sample. The table also indicates significant increase in
variations in real money as well as in the two nominal variables.
The formal investigation is done through Co-integration and Error
Correction Model framework. At the first step, the variables are tested
for the unit roots by applying both the ADF and PP tests. The results
are reported in Table 2.
The table indicates that the variables are, in general, first
differenced stationary, i.e., I(1). We now proceed to investigate the
causal relation between the two variables by estimating the
co-integrating regression suggested by Engle-Granger. If co-integration
is found, the Error Correction Models are estimated. Otherwise, the
Granger causality equations are estimated. In all cases the lag lengths
are decided on the basis of Log Likelihood, Akaike and Bayesian
information criteria. The next three sections investigate the causal
relations between real money and real income, nominal money and nominal
income, and nominal money and prices.
IV. CAUSALITY BETWEEN REAL MONEY AND REAL INCOME
We start by looking at the causal relation between the two real
variables, real money and real income. In this context, we reproduced
the results reported by Husain and Rashid (2008) indicating no short run
and long run causal relations between the two variables. Table 3(a)
shows the results.
They, however, found significant impact on the relations between
the two variables following the shift due to the start of the economic
reforms in the 1990s. In this study we consider another important shift
during the sample period, the shifts due to price hikes in the early
1970s.
Shifts in Real Money and Real Income Due to Price Hikes
To reflect the shifts in real variables due to the price hikes we
introduce a dummy variable in the analysis that takes the value of one
from 1972-73 onwards. The results are reported in Table 3(b).
The dummy variable in the co-integrating regression is highly
significant indicating significant shift in the relation between real
variables. The ADF and PP tests are also highly significant indicating
the existence of a strong relation between real money and real income in
the long run. The error term in money equation is also highly
significant and verifies the strong long run relation between real
variables. Finally, the analysis indicates a uni-directional causality
from real income to real money in the long run. In the short run the
real variables do not seem to affect each other.
Shifts in Real Money and Real Income Due to Both Prices and Reforms
To incorporate both the shifts, price hikes and economic reforms,
we include another dummy, D2, which takes the value of one from 1991-92
onwards in addition to the dummy for prices, D1. The results are shown
in Table 3(c).
Both the dummies are significant where the dummy for prices has a
greater magnitude. The results remain the same as in the case where only
the dummy for prices is included, that is, a uni-directional causality
from real income to real money in the long run with no short run
causality.
V. CAUSALITY BETWEEN NOMINAL MONEY AND NOMINAL INCOME
We now turn to examine the causal relation between the two nominal
variables using the same procedure adopted earlier. The first set of
results is shown in Table 4(a).
The PP tests in Co-integrating regression are insignificant
rejecting any long run relation between the two nominal variables.
However, the ADF test is significant at 10 percent level of
significance. Hence, we can say that there is a weak evidence of any
long run relation between the two nominal variables. The Error
Correction equations verify the weak long run relation where the error
term is significant at 10 percent in the money equation. The equations
indicate a weak evidence of unidirectional causality from nominal income
to nominal money in the long run with no short run causal effects. If we
assume no Co-integration, the Granger equations show the evidence of
income affecting money at 2nd lag, although the F-test is not
statistically significant.
Following Husain and Rashid (2008) we report the analysis for the
3rd lag too. The results show that the error term in error correction
equations has become insignificant implying no long run relation between
money and income. The equations further show the significant effects of
income on money at 3rd lag verified by F-value. The same result is shown
by Granger equations if we ignore the error term.
Hence, there is evidence of a one-way causation from nominal income
to nominal money although the existence of a long run relation between
the two nominal variables is not clear. There is also persistent
evidence of nominal income affected by its own first lag and affecting
money at three years' lag. We now proceed to take note of the
shifts in nominal variables during the sample period.
Shifts in Nominal Money and Nominal Income Due to Price Hikes
The results reflecting the shifts in nominal variables due to the
price hikes in the early 1970s are reported in Table 4(b) which shows
that as in the case of real variables the shift in the relation of
nominal variables due to the price hikes is very significant. Once
again, the ADF and PP tests have become highly significant indicating
strong evidence of a long run relation between the two nominal
variables. However, the most significant change occurs in the direction
of causality. Now the results show the bidirectional causality between
nominal money and nominal income in the long run. In the short run,
however, no causal relation between the two still prevails. Following
the procedure adopted previously, we do the analysis for the third lag
that also indicates significant change. The persistent three years lag
effect of income on money now disappears.
Hence by incorporating the shifts in the nominal variables due to
the price hikes we have found the feedback mechanism of money in
changing income in Pakistan. It may be mentioned here that Husain and
Rashid (2008) in a similar kind of analysis that took note of the shift
due to the economic reforms did not find such mechanism of money. We now
consider both the shifts, the shifts due to the price hikes and the
economic reforms, together to further explore the issue.
Shifts in Nominal Money and Nominal Income Due to Both Reforms and
Prices
The results reflecting both the shifts are reported in Table 4(c).
The table shows that, as in the case of real variables, the two
dummies are significant and the dummy for prices has greater magnitude.
It can be seen that the result regarding causality is similar to the one
when only one dummy, the dummy for prices, is included.
Hence, the analysis indicates, as in the case of real variables,
significant shifts in the nominal variables during the sample period.
Similarly, the shift that occurred in the early 1970s due to price hikes
seems to be very crucial to be incorporated in the analysis as it
significantly changes the results. The results indicate the existence of
a long run relation between nominal money and nominal income where the
two variables seem to affect each other in the long run. In the short
run, however, the two nominal variables, as the real variables, appear
to be independent of each other.
VI. CAUSALITY BETWEEN NOMINAL MONEY AND PRICES
Finally, we investigate the causal relation between nominal money
and prices using the same procedure adopted in the previous sections.
The first set of results is reported in Table 5(a).
Both the ADF and PP tests are highly significant indicating the
existence of a long run relation between money and prices in Pakistan.
The error correction equations suggest a uni-directional causality from
money to prices in the long run. However, in the short run there is
evidence of prices affecting money at 2nd lag although F-value is not
significant. Once again as in the case of nominal income we do the
analysis for the 3rd lag. However, the result (not reported here) shows
no significant lags in either equation indicating no short run causal
effects. However, it verifies the uni-directional causality from money
to prices in the long run.
Shifts in Money and Prices Due to Price Hikes
The results reflecting the shifts in the two variables due to the
price hikes in the early 1970s are reported in Table 5(b).
The dummy variable in the co-integrating regression shows signs of
a significant shift in the relationship of money to prices. However, the
results remain the same, that is, a unidirectional causality from money
to prices in the long run with indication of prices affecting money at
two years' lag.
Shifts in Money and Prices Due to Both Reforms and Prices
Finally, the results reflecting both the shifts are reported in
Table 5(c).
The table shows that the shift in the money-price relationship is
significant in the case of price hikes but not in the case of reforms.
Once again, the results have not changed. Hence, there is persistent
evidence of a uni-directional causality from money to prices in the long
run.
Hence, we can say that the relationship between money and prices in
Pakistan does not seem to be affected by the shifts in the variables
during the sample period. However, the shift that occurred in the early
1970s due to price hikes seems to be greater in this case too. The
results indicate the existence of a long run relation between money and
prices where money seems to lead prices in the long run. In the short
run there is some indication, though not significant, of prices
affecting money with two years' lag. There is also persistent
evidence of prices affected by their own first lag.
VI. SUMMARY AND CONCLUSIONS
The objective of this study is to extend the analysis of causality
by Husain and Rashid (2008) by taking cognizance of the shift in the
variables due to the price hikes in the early 1970s. Following them we
investigate the causal relations between real money and real income,
between nominal money and nominal income, and between nominal money and
prices using the annual data set from 1959-60 to 2003-04, examining the
stochastic properties of the variables used in the analysis, and in
consideration of the expected shifts in the series through dummy/ies.
The formal analysis indicates significant shifts in the variables
during the sample period. These shifts include the price hikes in early
1970s and the start of the economic reforms in early 1990s. In this
context, the shift occurred in the early 1970s seems to be more
important to be incorporated in the analysis. In particular, it seems to
be very crucial in the case of nominal variables as it has significantly
changed the results.
The analysis further indicates the existence of a long run
relationship between real money and real income provided that shifts in
these variables are given consideration. Moreover, real income seems to
be the leading variable that affects real money in the long run. In the
short run, the two real variables appear to be independent of each
other. Similarly, when money and income are expressed in nominal terms,
there is evidence of a one-way causation from income to money although
the existence of a long run relationship between them is not clear.
However, the relationship between the two nominal variables is
significantly affected by the shift due to the price hikes in the early
1970s. Taking note of the shift indicates the existence of a strong long
run relation as well as bi-directional causality between nominal money
and nominal income. The results do not change if we also include the
shift representing reforms. In the short run, however, the two nominal
variables, like real variables, appear to be independent of each other.
As regards the money-price relationship in Pakistan, the analysis
shows a long run relation between the two where money seems to lead
prices in the long run. In the short run there is some indication,
though not significant, of prices affecting money with two years lag.
These findings regarding money-price relationship are not affected by
the shifts during the sample period.
Finally it can be said that the study finds an active role of money
in the Pakistani economy as it is found to be the leading variable in
changing prices without any feedback. In the case of income, the study
finds the feedback mechanism of money generally missing in earlier
studies which may be because of not taking note of the shift in the
macro economic variables in Pakistan in the early 1970s.
Limitations of the Study
At the end we would like to point out the limitations which can be
considered in future research. This study mainly follows Husain and
Rashid (2008) and uses the same sample period, data sources and
methodology, except that it examines the impact of a different shift,
that is, the shift due to price hikes. Both studies confine to the
Bi-variate Causal analysis. However, the extension of the analysis to
Multi-variate causal analysis may provide better insights regarding the
role of these variables. Similarly, the two studies are based on annual
data covering the period from 1959-60 to 2003-4. As mentioned above,
various economic and political events have occurred with a significant
impact on the macro economic variables. One of the significant event was
the separation of the Eastern wing of the country in 1971 that may cause
significant effects on macroeconomic variables. Therefore, the use of
quarterly data covering the last two decades should be a better option.
On the technical side, the use of recent unit root tests taking
care of structural break in the series would be better than the
conventional tests used in the study. Similarly, the studies have used
the Ordinary Least Square (OLS) approach to test for money, output and
prices causality. However, OLS may not be useful in the presence of
conditional heteroskedastic errors. In this context, the use of Maximum
Likelihood (ML) technique may be better because this technique has a
better power to detect causality.
REFERENCES
Abbas, Kalbe (1991) Causality Test between Money and Income: A Case
Study of Selected Developing Asian Countries (1960-988). The Pakistan
Development Review 30:4, 919-929.
Barth, J. and J. Bannett (1974) The Role of Money in the Canadian
Economy: An Empirical Test. Canadian Journal of Economics May, 306-311.
Bengali, K., A. Khan, and M. Saddaqat (1999) Money, Income, Prices
and Causality: The Pakistani Experience. Journal of Developing Areas 33:
Summer, 503-514.
Brillembourg, A. and M. Khan (1979) The Relationship between Money,
Income, and Prices: Has Money Mattered Historically? Journal of Money,
Credit, and Banking August, 358-365.
Dereyes, F., D. Starleaf, and G. Wang (1980) Test of Direction of
Causation between Money and Income in Six Countries. Southern Economic
Journal October, 477-487.
Husain, F. and Abdul Rashid (2002-2008) Economic Liberalisation and
the Causal Relations Among Money, Income, and Prices: The Case of
Pakistan. Pakistan Journal of Applied Economics 18, 103-121.
Jones, J. D. and N. Khilji (1988) Money Growth, Inflation, and
Causality: Empirical Evidence for Pakistan (1973-85). The Pakistan
Development Review 27:1, 45-88.
Khan, A. and A. Siddiqui (1990) Money, Prices and Economic Activity
in Pakistan: A Test of Causal Relation. Pakistan Economic and Social
Review Winter, 121-136.
Lee, S. and W. Li (1983) Money, Income, and Prices and their
Lead-lag Relationship in Singapore. Singapore Economic Review April,
73-87.
Pakistan, Government of (1998) 50 Years of Pakistan in Statistics.
Islamabad: Federal Bureau of Statistics.
Pakistan, Government of (Various Issues) Economic Survey. Ministry
of Finance.
Pakistan, Government of (Various Issues) Monthly Statistical
Bulletin. State Bank of Pakistan.
Phillips, R., and P. Perron (1988) Testing for a Unit Root in Time
Series Regression. Biometrika June, 335-346.
Schwert, W. (1987) Effects of Model Specification on Tests for Unit
Roots in Macroeconomic Data. Journal of Monetary Economics July, 73-103.
Siddiqui, Anjum (1989) The Causal Relation Between Money and
Inflation in a Developing Economy. International Economic Journal
(Summer), 79-95.
Sims, C. (1972) Money, Income, and Causality. American Economic
Review September, 540-552.
Williams, W., C. Goodhart, and D. Gowland (1976) Money, Income, and
Causality: The U.K. Experience. American Economic Review June, 417-423
Authors' Note: The authors are grateful to the referees for
their useful comments.
Fazal Husain <fazal.husain1960@gmail.com> is Chief of
Research at the Pakistan Institute of Development Economics, Islamabad.
Abdul Rashid <ecp09ar@sheffield.ac.uk> is associated with
International Islamic University, Islamabad.
Table 1
Descriptive Statistics for Growth in Money, Income, and Prices
Real Nominal Real Nominal
Money Money Income Income Prices
Full Sample: (1960-61- 2003-04)
Mean 0.0605 0.1325 0.0540 0.1262 0.0720
Std. Dev. 0.0697 0.0541 0.0242 0.0491 0.0499
Observations 44 44 44 44 44
Sample I: (1960-61-1971-72)
Mean 0.0728 0.1010 0.0646 0.0910 0.0282
Std. Dev. 0.0434 0.0385 0.0263 0.0326 0.0332
Observations 12 12 12 12 12
Sample II: (1972-73-2003-04)
Mean 0.0559 0.1444 0.0500 0.1394 0.0885
Std. Dev. 0.0774 0.0548 0.0225 0.0481 0.0451
Observations 32 32 32 32 32
Equality of Means and Variances
Mean (t-value) 0.91 2.94 ** 1.70 3.81 ** 4.83 **
Variance (F-value)3.18 ** 2.02 * 1.36 2.17 ** 1.84
Note: ***, **, and * represent significance at 1 percent, 5 percent
and 10 percent.
Table 2
Unit Root Tests for Money, Income, and Prices
Levels First Difference
W/O Trend W. Trend W/O Trend W. Trend
ADF
Real Money -0.49 -3.303 -4.957 ** -4.365 **
Real Income -2.837 -1.006 -6.119 ** -6.666 **
Nominal Money 0.314 -3.507 -5.012 ** -4.488 **
Nominal Income -0.399 -1.455 -3.661 ** -3.711 **
Prices 0.089 -2.563 -3.548 ** -3.558 **
PP (W/O Trend) (1=3) (1=9) (1=3) (1=9)
Real Money -0.214 -0.103 -4.886 ** -4.763 **
Real Income -3.104 ** -2.930 ** -6.211 ** -6.745 **
Nominal Money 0.844 1.021 -5.014 ** -4.888 **
Nominal Income -0.151 -0.162 -3.612 ** -3.540 **
Prices 0.487 0.469 -3.489 ** -3.309 **
PP (W. Trend) (1=3) (1=9) (l=3) (1=9)
Real Money -2.54 -2.152 -4.823 ** -4.682 **
Real Income -0.457 -0.556 -7.325 ** -7.290 **
Nominal Money -2.6 -2.433 -5.006 ** -4.852 **
Nominal Income -1.788 -1.992 -3.553 * -3.457 *
Prices -2.779 -2.727 -3.488 * -3.295 *
Note: ** and * represent significance at 5 percent and 10 percent.
Table 3(a)
Causality between Real Money and Real Income
Cointegration (Engle-Granger)
Const. Coeff. ADF PP(1=3) PP(1=9)
RM on RY -1.345 *** 1.035 *** -1.092 -1.387 -1.358
Conclusion: No Co-integration
Granger Causality
Lag 1 DRY DRM
DRY(-1) -0.032 -0.115
DRM(-1) 0.059 0.270
F-value 0.917 0.055
Granger Causality
Lag 3 DRY DRM
DRY(-1) -0.132 -0.348
DRY(-2) 0.267 -0.731
DRY(-3) 0.321 0.729
DRM(-1) 0.086 0.394 *
DRM(-2) -0.012 -0.089
DRM(-3) -0.916 -0.117
F-value 1.313 1.328
Conclusion: No Short run Causality upto three lags
Note: ***, **, and * represent significance at t percent, 5 percent
and 10 percent.
Table 3(b)
Causality between Real Money and Real Income (Prices)
Cointegration (Engle-Granger)
Const. D Coeff. ADF
RM on RY -3.863 *** -0.428 *** 1.259 *** -4.943 ***
Cointegration (Engle-Granger)
PP(1=3) PP(1=9)
RM on RY -4.864 *** -4.940 ***
Conclusion:Evidence of Co-integration
Error Correction Causality
Lag 1 DRY DRM
D 0.059 *** 0.017
e(-1) -0.008 -0.728 ***
DRY(-1) -0.081 -0.446
DRM(-1) 0.058 0.369
F-value 0.711 1.474
Error Correction Causality
Lag 3 DRY DRM
D 0.003 0.009
e(-1) 0.062 -0.752 ***
DRY(-]) -0.180 -0.293
DRY(-2) 0.334 -0.206
DRY(-3) 0.299 -0.334
DRM(-1) 0.087 0.295 *
DRM(-2) -0.019 0.058
DRY(-3) -0.093 -0.044
F-value 1.078 0.632
No Short run Causality
Note: and * represent signiticance at 1 percent, 5 percent and
10 percent.
Table 3(c)
Causality between Real Moizey and Real Income (Prices and Reforms)
Cointegration (Engle-Granger)
Const. D1 D2 Coeff. ADF
RM on RY -2.735 *** -0.353 *** 0.124 *** 1.163 *** -5.238 ***
PP(1=3) PP(1=9)
RM on RY -5.093 *** -5.008 ***
Conclusion: Evidence of Strong Co-integration
Error Correction Causality
Lag 2 DRY DRM
D1 0.003 0.004
D2 -0.026 * 0.005
e(-1) 0.051 -0.929 ***
DRY(-]) -0.338 -0.186
DRY(-2) 0.012 0.171
DRM(-1) 0.099 0.381 **
DRM(-2) 0.022 0.082
F-value 1.137 0.211
Conclusion: Unidirectional Causality from Income to Money in the Long curt
No Short run Causality
Note: ***, **, and * represent significance at 1 percent, 5 percent and 10
percent.
Table 4(a)
Causality between Nominal Money and Nominal Income
Cointegration (Engle-Granger)
Const. Coeff. ADF PP(1=3) PP(1=9)
NM on NY -1.100 *** 1.016 *** -1.859 * -1.525 -1.451
Conclusion: Weak Evidence of Co-integration
Error Correction Causality Granger Causality
Lag 2 DNY DNM Lag 2 DNY DNM
e(-1) -0.037 -0.201 *
DNY(-1) 0.520 ** -0.311 DNY(-1) 0.495 *** -0.196
DNY(-2) -0.012 0.125 DNY(-2) -0.06 0.401 **
DNM(-1) 0.085 0.208 DNM(-1) 0.11 0.261
DNM(-2) 0.019 -0.017 DNM(-2) -0.009 -0.052
F-value 0.182 1.061 F-Value 0.371 2.346
Conclusion: Weak Evidence of Unidirectional Causality from Income to
Money
Error Correction Causality Granger Causality
Lag 3 DNY DNM Lag 3 DNY DNM
e(-1) 0.066 -0.075
DNY(-1) 0.569 ** -0.159 DNY(-1) 0.504 *** -0.097
DNY(-2) -0.069 -0.005 DNY(-2) -0.115 0.097
DNY(-3) 0.209 0.559 ** DNY(-3) 0.150 0.520 **
DNM(-1) 0.020 0.034 DNM(-1) 0.061 0.104
DNM(-2) 0.049 0.017 DNM(-2) 0.019 0.022
DNM(-3) -0.095 -0.025 DNM(-3) -0.111 -0.056
F-value 0.148 2.503 * F-Value 0.288 4.034 **
Conclusion: Unidirectional Causality from Income to Money at 3 Years
Lag
Note: ***, **, and * represent significance at 1 percent, 5 percent
and 10 percent.
Table 4(b)
Causality between Nominal Money and Income (Prices)
Cointegration (Engle-Granger)
Const. D Coeff. ADF
NM on NY -1.846 *** -0.393 *** 1.097 *** -4.631 ***
Cointegration (Engle-Granger)
PP(1=3) PP(1=9)
NM on NY
-4.479 *** -4.407 ***
Conclusion: Evidence of Co-integration
Error Correction Causality
Lag 2 DNY DNM
D 0.066 ** 0.060 *
e(-1) -0.369 *** -0.449 ***
DNY(-1) 0.239 0.046
DNY(-2) 0.205 0.018
DNM(-1) -0.072 0.317 **
DNM(-2) -0.041 -0.011
F-value 0.230 0.063
Error Correction Causality
Lag 3 DNY DNM
D 0.056 * 0.050
E(-1) -0.474 *** -0.359 **
DNY(-1) 0.236 0.045
DNY(-2) 0.098 -0.075
DNY(-3) 0.370 * 0.324
DNM(-1) -0.214 0.191
DNM(-2) 0.000 0.018
DNM(-3) -0.118 -0.065
F-value 1.082 0.753
Conclusion: Bidirectional Causality between Income and Money in
the Long run No Short run Causality
Note: ***, **, and * represent significance at 1 percent, 5
percent and 10 percent.
Table 4(c)
Causality between Nominal Money and Income (Prices and Reforms)
Cointegration (Engle-Granger)
Const D1 D2 Coeff. ADF
NM on NY -1.451 *** -0.321 *** 0.117 ** 1.059 *** -4.597 ***
Cointegration (Engle-Granger)
PP(1=3) PP(1=9)
NM on NY -4.411 *** -4.226 ***
Conclusion: Strong Evidence Co-integration
Error Correction Causality Error Correction Causality
Lag 2 DNY DNM Lag 3 DNY DNM
Dl 0.080 *** 0.064 * Dl 0.064 ** 0.054
D2 -0.022 -0.015 D2 -0.012 -0.007
e(-1) -0.442 *** -0.549 *** e(-1) -0.600 *** -0.438 **
DNY(-1) 0.186 0.001 DNY(-1) 0.193 0.010
DNY(-2) 0.252 -0.132 DNY(-2) 0.192 -0.157
DNM(-I) -0.076 0.312 * DNY(-3) 0.407 ** 0.279
DNM(-2) -0.028 -0.002 DNM(-1) -0.239 0.201
F-value 0.25 0.209 DNM(-2) 0.003 0.021
DNM(-3) -0.098 -0.081
F-value 1.352 0.522
Conclusion: Bidirectional Causality between Income and Money in the
Long run No Short run Causality
Note: ***, **, and * represent significance at 1 percent, 5 percent and
10 percent.
Table 5(a)
Causality between Nominal Money and Prices
Cointegration (Engle-Granger)
Const. Coeff. ADF PP(1=3) PP(1=9)
NM on DF 3.850 *** 1.697 *** -3.696 *** -2.687 *** -2.477 **
Conclusion: Strong Evidence of Co-integration
Error Correction Causality
Lag 2 DDF DNM
e(-1) -0.314 *** -0.071
DDF(-1) 0.589 *** -0.349
DDF(-2) 0.216 0.496 *
DNM(-1) 0.163 0.167
DNM(-2) 0.003 0.045
F-value 0.898 2.446
Conclusion: Unidirectional from Money to Prices in the Long run
Note: ***, **, and * represent significance at 1 percent, 5 percent
and 10 percent.
Table 5(b)
Causality between Money and Prices (Prices)
Cointegration (Engle-Granger)
Const. D Coeff. ADF
NM on DF 3.702 *** -0.172 ** 1.755 *** -3.915 ***
Cointegration (Engle-Granger)
NM on DF PP(1=3) PP(1=9)
-2.924 *** -2.734 ***
Conclusion: Evidence of (Co-integration
Error Correction Causality
Lag 2 DDF DNM
D 0.097 *** 0.038
e(-1) -0.462 *** -0.054
DDF(-1) 0.393 *** -0.324
DDF(-2) 0.244 * 0.490 *
DNM(-1) -0.002 0.153
DNM(-2) -0.046 0.032
F-Value 0.130 2.068
Conclusion: Prices in the Long run
Note: ***, 1 percent, 5 percent and 10 percent.
Table 5(c)
Causality between Money and Prices (Prices and Reforms)
Cointegration (Engle-Granger)
Const. D1 D2 Coeff. ADF
NM on DF 3.556 *** -0.220 ** -0.081 1.799 *** -3.953 ***
Cointegration (Engle-Granger)
PP(1=3) PP(1=9)
NM on DF -2.993 *** -2.787 ***
Conclusion: Strong Evidence of Co-integration
Error Correction Causality
Lag 2 DDF DNM
D1 0.102 *** 0.039
D2 -0.010 -0.008
e(-1) -0.446 *** -0.076
DDF(-I) 0.307 ** -0.311
DDF(-2) 0.164 0.452 *
DNM(-1) -0.039 0.162
DNM(-2) -0.054 0.032
F-value 0.350 1.777
Conclusion: Unidirectional from Money to Prices in the Long run
Note: ***, **, and * represent significance at 1 percent, 5
percent and 10 percent.
