The African Financial Review July-August 2014 | Page 14

capture expectations on inflation, we specify a dynamic panel
model for prices in WAMZ. We employed a panel data analysis
because it considers various cross sectional units (countries in
WAMZ) over time. This study covers four countries in WAMZ
(The Gambia, GHANA, Nigeria and Sierra Leone) from 1980
to 2011. Based on the direct view of banking crises, we employed
the banking crises dates as presented by Demirgüç-Kunt and
Detragiache (2005), Jácome (2008), Laeven and Valencia (2008)
and Reinhart and Rogoff (2008). These crises periods are based
on expert judgments and they are used as control variables
to determine its effect on inflation. Following Martinz Periz
(2000 and 2002) and Kaehler (2010) we adopted the modified
monetarist inflation theory that is augmented with crises dummy.
Inflation in this case will depend on a vector of explanatory
variables and a control variable. The dummy control variable
employed in this study is the banking crises which takes the
value of 1 during crises and 0 otherwise. The empirical model
is specified as:
CPI_(i,t)=
θ_i+θ_1 CPI_(i,t-1)+ θ_2 (M2) _(i,t)+ θ_3
INTR_(i,t)+ θ_4 AINT_(i,t)+ θ_5 EXC_(i,t)+ θ_6 Y +θ_(7 )
BC_(i,t)+ε_(i,t) (1)
Where the index “i” represents panel members or number of cross
sectional units and t = 1, ..., T denotes the time dimension. In this
study, we considered four cross sectional units and all variables are
in log form except those rates. “M2” is the nominal broad money
stock, “Y” represents a measure of real income as a scale variable
(Gross Domestic Product), “INTR” is the own interest rate on
money (demand deposit), “AINT” is the alternative interest rate
(treasury bill or savings rate) , “EXC” is the exchange rate, and
“CPI” captures the inflation rate (consumer price index). BC
is a dummy variable that reflects the years of banking crises in
WAMZ countries. The disturbance term “ε_(i,t)” is assumed to
be a white noise error process. The a priori expectations for the
parameters are:
θ_1>0,θ_2>0,θ_3<0,θ_4<0,θ_5>0 ,θ_6>0 and θ_7>0

Table 2. A panel regression result on inflation in
WAMZ.
Variable

Coefficient

Constant
INFL(-1)
M2
Y
INTR
AINT
EXC
BC

-4.278*** (3.190)
0.3387*** (6.206)
-0.0127 (-0.271)
0.6677*** (2.745)
-0.5322*** (-3.606)
0.9824*** (5.804)
-0.0580*** (-4.007)
0.2160*** (2.851)

Number of oberservation: 115
Sargan over-identification test: 253.683***
Wald (joint) test: 258.799 ***
Note: *,** and *** signifies 10, 5 and 1% respectively. t-statistic in parenthesis.

14 | The African Financial Review

Estimation procedure

In estimating Equation 1, using fixed or random effect model,
several econometrics problems may arise such as; first, timeinvariant country characteristics (fixed effects), such as geography
and demographics, may be correlated with the explanatory
variables. Secondly, the presence of the lagged dependent variable
gives rise to autocorrelation. Thirdly, the lagged dependent variable
may correlate with the error term. Due to the above problems, a
Generalized Method of Moments “GMM” estimator is proposed.
In addition, Kaehler, (2010) asserted that using dynamic panel
model estimated with the general method of moments (GMM)
all estimated coefficients are highly significant in explaining
variations in inflation. This study therefore employs dynamic panel
model to ascertain the claims of Kaehler, (2010) in WAMZ. To
determine the relationship in the above equation, we will employ
a one- step system GMM estimator. The GMM estimator implies
that the regression is time-differenced in order to remove crosssection specific effects. It estimates in a system the regression
equations in differences and levels, each with its specific set
of instruments. Relative to conventional instrumental variable
methods, it improves substantially on the weak instruments
problem through more formal checks of the validity of the
instruments and provides for potentially improved efficiency.
Data are sourced from WDI-Online (2010), IFS CD-ROM
(2010), index mundi (online) and Central banks statistical bulletin
for various countries.

Result and discussion

This section of the paper empirically analyse the objectives of
the paper, by employing various statistical and econometrics
techniques. Table 1 presents a vivid description of the variables
used in the study.

The lowest rate of inflation is 0.8% while the maximum is
178.7%. On the average, inflation stood at 24.2% with a standard
deviation of 27.7. The average money growth and exchange rate
in WAMZ is 1.3 and 248.9 with a standard deviation of 1.2
and 730.16 respectively. In addition, log of output growth rate
on the average remained at 1.6 with a standard deviation of 1.3.
The banking crisis variable has the mean of 0.17 with a standard
deviation of 0.38.

Table 2 presents the result of the relationship between
banking crises and inflation in WAMZ. The result shows the
efficacy of the explanatory variables in the model and the
instruments employed are valid. In addition, the immediate past
value of inflation, interest rate, exchange rate, output and banking
crises are the major determinants of current rate of inflation
in WAMZ. Immediate past value of inflation exerts a positive
and significant impact on current inflation rate in WAMZ. The
positive impact is statistically significant at 1% level of significance.
This corroborates with empirical studies on inflation (Kaehler,
2010; Egwaikhide et al., 1994). A 100% increase (decrease) in
expected inflation will increase (decrease) current inflation rate
by 33.8%. This implies that the public expectations on inflation
increase current inflation in WAMZ. There is a negative
relationship between broad money and inflation in WAMZ and
this is not statistically significant. The negative relationship is not
also consistent with a-priori expectation and findings from some
studies (Ocran, 2007).

Income increment tends to increase inflation in WAMZ.
There is a positive and significant relationship between income