DOES INFLATION UNCERTAINTY DECREASE WITH INFLATION? A ...

DOES INFLATION UNCERTAINTY DECREASE WITH INFLATION? A GARCH MODEL OF INFLATION AND INFLATION UNCERTAINTY Alfred Barimah1 and Franklin Amuakwa-Mensah

Abstract In the present study, we examine the relationship between inflation and inflation uncertainty in Ghana from 1964:04 to 2012:12 using two-step procedure. At the first step, GARCH(1,2)-M model of monthly inflation data is estimated and the conditional variance from these estimates is used to as proxy for inflation uncertainty. Then, the Granger causality tests between actual inflation and our generated inflation uncertainty series are performed. Two main results follow from this paper. First, we find strong statistical support for the Friedman-Ball hypothesis: inflation significantly raises inflation uncertainty in Ghana over the full sample period and two subsamples at different lag lengths. Second, we also find evidence of inflation uncertainty affecting inflation in the long run as suggested by the Cukierman and Meltzer Hypothesis. Thus, results of this study have significant implications for Ghanaian Inflation Targeting (IT) efforts as well as the literature focusing on the relationship between inflation and inflation uncertainly in developing countries that are contemplating adopting inflation targeting. The policy implication is to aim at low average inflation rates in order to reduce the negative consequences of inflation uncertainty. Keywords: GARCH Models; Inflation uncertainty, Friedman-Ball Hypothesis; Cukierman-Meltzer Hypothesis; VAR model JEL Classification: E310; E300; C220; O550

1

Dr, Alfred Barima is a Lecturer at the Department of Economics, University of Ghana, Legon while Franklin Amuakwa-Mensah is with the Institute of Statistical, Social and Economic Research, University of Ghana, Legon

Vol. 12, No.1

Alfred Barimah and Franklin Amuakwa-Mensah

INTRODUCTION One of the most remarkable macroeconomic achievements in Ghana has been the maintenance of single digit inflation continuously for 30 months since May 2010. Prior to this single digit inflation record, the inflation rate had accelerated steadily from an annual average of 9.79% during the 1960s to 38.6 during the 1970s and further to a record high of 49.5% in the 1980s before easing to 28.2% in the 1990s. The declining trend in inflation started from the period 2000-2009 with an average rate of 19.6% and during the entire period of inflation targeting (May 2010-to date) average inflation was 9.6%. Regarding volatility clustering, our data reveals that those periods with the high average inflation were in general the ones with the largest standard deviation. Ghana is a good candidate for volatility analysis given her chucked history of high and volatile inflation, especially for much of the 1977-78 and 1983. Even in most of the much talked-about stable years of 1993 –2012, there were large swings in the inflation rate (see figure 1). It is thus empirically important to ascertain whether the recent low inflation figures in Ghana have any implications on inflation uncertainty. Figure 1: Inflation Trend for 1993-2012

Source: Authors’ computation based on data from GSS

33

Vol. 12, No.2

Journal of Monetary and Economic Integration

The main purpose of this study is to examine the relationship between inflation and its uncertainty in Ghana over the period 1964:04-2012:122. This paper estimates an AR(3)-GARCH-in-Mean model that incorporates simultaneous feedback effects of inflation and inflation uncertainty. We generate a measure of monthly inflation uncertainty and investigate the link between inflation and inflation uncertainty using Granger tests. Following the adoption of inflation targeting (IT) as the main monetary policy thrust in May in 2007 we see from figure 3 (in appendix) that average inflation has been low and falling particularly since May 2010. The question we ask is: how does inflation affect inflation uncertainty in Ghana and does inflation uncertainty fall with the rate of inflation? To the extent that uncertainty reduces with lower inflation, can it be concluded that the lower inflation these days have lowered the real cost of inflation via the inflation uncertainty channel? The objective of this paper is therefore to investigate the relationship between inflation and inflation uncertainty in Ghana over the full sample period and three other sub-periods that have important implications for inflation uncertainty3. Following recent empirical studies, we first derive a measure of inflation uncertainty from a generalized autoregressive conditional heteroscedasticity (GARCH) model of inflation and study the nexus between inflation and inflation uncertainty in a bivariate VAR context. We then identify the direction of causality between inflation and inflation uncertainty using Granger causality. To our knowledge, there is no similar empirical study in the literature concerning the inflation and inflation-uncertainty relationship in Ghana that uses the inflation figures for the 1964:04-2012:12 period. This study differs from prior works on Ghana in three main respects. First, our GARCH-in-Mean model that controls for the simultaneous feedback effects between inflation and inflation uncertainty provides a theoretical basis for testing whether inflation granger causes inflation uncertainty. Second, the issue of whether inflation uncertainty varies with the rate of inflation can be formally investigated for the sub-period 2007-2012. Finally, the results of this study will have significant implications for Ghanaian Inflation targeting regime as well as the literature focusing on the relationship between inflation and inflation uncertainly in developing. The rest of the paper is structured as follows. Section 2 contains a brief overview of the history of inflationary process in Ghana and a review of theoretical and empirical literature connecting between inflation and inflation uncertainty is undertaken in 2

The annualized year-on year inflation rates as reported by the Ghana Statistical Service are available from March 1964 and as would be explained below the estimations in this paper cover the period 1964:04 - 2012:12. 3

These periods are the period of sustained democracy (1993:01 to 2012:12), inflation targeting period (2007:5 to 2012:12) and the continuous single digit inflation period (2010:6 to 2012:12).

34

Vol. 12, No.1


section three. Section four introduces the data and methodology while, the model and the estimation results are contained in section five. Finally, section six concludes. INFLATIONARY DEVELOPMENTS IN GHANA This section briefly describes the theories of inflation and inflationary trends in Ghana during the last six decades. Whereas the demand pull inflation results from excess demand of goods and services over supply, cost push inflation results from increases in cost of production which is independent of demand. Demand pull inflation is linked to the monetarist and Keynesian theories since increases in money supply and expansionary fiscal policies lead to excess demand and thus inflation. Supply side pressures emanating from excessive wage demands and unfavourable supply conditions have all been found to be significant determinants of inflation in Ghana. Specifically, higher than average growth in money supply that are often accompanied by huge fiscal slippages and production bottlenecks have been identified as important explanatory factors in Ghana’s inflation (Kwakye, 2010). According to Table 1, average inflation for the last five decades are 9.8%, 38.6%, 49.5%, 28.2% and 19.6% whilst the average inflation for the entire period is 29.6%. Figure 2 (see appendix) plots the 12-months inflation rate in Ghana between April 1964 and December 2012. The series display that high volatility overtime and historical inflationary peaks were usually followed by periods of diminishing inflation rate. The only period when low and single digit levels of inflation persisted for an extended period was the post-inflation targeting years. Following the adoption of inflation targeting (IT) framework in May 2007, Ghana’s inflation performance has been considered satisfactory; the country has since experienced the lowest average inflation that culminated into thirty continuous months of single digit inflation. The average inflation since the adoption of IT is 12.7% and that for the single digit inflation period is 9% (see panel B of Table 1). Before then inflation had largely been in the double digit range. In general, compared to the situation in the 1970s and the early 1980s when average inflation peaked at 50%, Ghana’s current inflation performance can be considered as satisfactory. From a lower average rate in the 1960s, there has been steady increase in the rate of inflation from 38.6% in the 1970s to 49.5% in the 1980s.

35

Vol. 12, No.2


Table 1: Inflation for various decades and Sub-periods Mean Standard Inflation Dev.

N

Panel A Period (In Decades) 1964 – 1969

9.78

12.15

69

1970 – 1979

38.64

37.67

120

1980 -1989

49.5

42.83

120

1990 – 1999

28.24

16.28

120

2000 – 2009

19.57

8.49

120

2010 – 2012

9.56

1.54

36

1964 – 2012

29.62

30.65

585

Sustained Democracy Period

24.97

13.86

240

Inflation Targeting Period

12.70

4.36

66

Single Digit Inflation Period

9.01

0.39

31

Panel B

The inflationary experience in most of the period 1972–1982 was largely due to expansionary fiscal and loose monetary policies. The futile attempt at using controls (i.e. fixed exchange rate, interest rates, import licensing, and administered prices for goods and services) to hold down inflation also accounted for the sharp decline in the size of the economy. A series of changes in government dominated mainly by the military was witnessed within the latter part of the period (1978-81). These governments pursued expansionary fiscal policy programmes supported by monetary expansion as huge budget deficits became widespread and were financed by bank loans to the government. It is therefore not surprising that the inflationary effects of the higher fiscal deficits manifested in the highest ever inflation rate of 174% in June 1983.Among major exogenous factors that contributed to the poor inflation performance over this period were the various negative shocks the economy experienced over of this period. These included the oil price shock of 1979-80 and the mounting debt problem. Ghana was also hit hard by the decline in commodity prices in the early 1980s, with the terms of trade in 1982 being less than half the 1970 level. The severe drought in 1983 accounted for the highest ever increase in food prices.

36

Vol. 12, No.1


The implementation of the Economic Recovery Programme (ERP) in 1983 has been largely credited with the slow-down in inflation since 1984. The set of policy measures undertaken under the reforms was grouped under three main themes namely, the Economic Recovery Programme I (1983–1986), the Economic Recovery Programme II (July 1987–July 1990) and the accelerated growth phase (1993–2000). Among the reasons that necessitated the reforms include the massive expenditure witnessed in 1992 as a result of the first democratic election of the fourth republic which contributed significantly towards inflation rates reaching 25 per cent in 1993 which further increased to 59.4 per cent in1995. Further, as much as persistent shortage of food results in increase food prices have significantly contributed to the rise in consumer price index during the 1990s, excessive money supply as a result of extensive financing of government deficits, have through the transactions demand for real balances, influenced the rate of change in the general price level. Among the most successful and significant measures taken during the stabilization phase of the ERP were realistic exchange rate policy and fiscal discipline. In the preceding decade, inflation had averaged 66% annually with wide fluctuations from year to year. During the early reform years, inflation came down from 50% per annum in the early reform years to 27% for 1987–1993. Even though there were occasions when inflation was brought down to appreciable levels, the actual inflation outturn during the reform years came nowhere near the targets that were set for each fiscal year: over the entire period of the reforms, 1983–2000, inflation averaged 34% per annum. As a means of price stabilization, monetary policy of the Bank of Ghana (BOG) was changed from directly targeting monetary aggregates such as reserve money and the money supply to inflation targeting (IT). In line with the broad objective of Ghana’s medium term economic programme, the government set an inflation target of 13 percent by end of 2002. To meet the set target, the government together with the monetary authorities put in place prudent fiscal management and tight monetary policy stance. These measures coupled with the slow pace of depreciation of the cedi led to a deceleration on the year-on-year inflation rate from 21.3 percent in 2001 to 15.2 percent as at the end of December 2002. Thereafter, the inflation rate has been oscillating at an average rate of 19.57 between the period of 2000 to 2009 as a result of high crude oil price and increased fiscal indiscipline in election years. However, there has been a recent decline in the inflation rate due to effective implementation of inflation targeting measures which was fully legislated in 2007. To be effective, IT requires certain policy and structural conditions which include autonomy of the central bank to conduct monetary policy freely, fiscal discipline and exchange rate flexibility, both of which facilitate monetary policy conducted mainly via interest rates (Kwakye, 2010). The presence of these conditions was mainly responsible for the lowest average inflation rate of 9.56 per cent between the periods 2010 to 2012.

37

Vol. 12, No.2


In sum, we see from Table 1 that inflation and its standard deviation follow an invented U-shaped path as those periods with lowest average inflation were in general the ones with the smallest standard deviation. The question we seek answers in the next section concerns the impact the current progress in taming inflation has on inflation uncertainty. LITERATURE REVIEW Inflation-inflation uncertainty nexus: The theoretical background Inflation and its effects on the real sector has been the subject of enormous research efforts in the inflation-inflation uncertainty literature. By making the value of future nominal payments to be unknown, uncertainty about future levels of inflation distorts the saving and investment decisions of agents and thus the often observed reduction in consumption, investment and economy growth in countries suffering from high inflation volatility can be attributed to inflation uncertainty. On how inflation uncertainty affects the real sector, three distinct channels have been identified. First, by increasing long-term interest rates, inflation uncertainty makes investment in long-term debt riskier and given that investment is a major component of aggregate demand, the efficient allocation of investible funds by financial sector participants is impeded. Second, inflation uncertainty results in partial indexation of contractual payments and with the resultant increases in uncertainty about wages, rent, taxes, depreciation, and profits, firms are thus forced to reduce their desired capital stocks. Finally, inflation uncertainty forces firms to shift their allocation of resources from more productive to protective activities such as improved forecasting skills about inflation and hedging activities via derivatives which are more costly for small enterprises and households (Golob, 1994). On the origins and sources of inflation uncertainty, Ball (1990) believes that inflation uncertainty emanates from the uncertainty of the monetary policy regime, known as “regime uncertainty”. When there is high inflation, policymakers face a dilemma: on one hand, the monetary authority may fear that the high inflation will trigger a recession in the economy and on the other hand, they would like to reduce inflation. Because the public is unaware of the inclination of policymakers, it will remain highly uncertain of the future course of inflation (Ball, 1992; Okun, 1971; Friedman, 1977). Therefore, economic agents will be highly uncertain about the future taste of policymakers and consequentially about future inflation. There is also the belief that that inflation

38

Vol. 12, No.1


uncertainty arises because of the unknown magnitude of a change in price level due to a given change in money supply (Holland, 1993a). There are four well-developed theories (hypotheses) that imply a relationship between inflation and inflation uncertainty; each of these makes a distinct prediction regarding the interrelationship between inflation and uncertainty. These hypotheses are (i) the Friedman-Ball hypothesis of Friedman (1977) and Ball (1992), (ii) the Cukierman and Meltzer hypothesis of Cukierman and Meltzer (1986) and Cukierman (1992), (iii) the Pourgerami and Maskus hypothesis of Pourgerami and Maskus (1987) and (iv) the Holland hypothesis of Holland (1995). The Friedman-Ball hypothesis predicts that causality is found to run from inflation to inflation uncertainty. Friedman (1977) postulates that by creating political pressure to reduce inflation, high inflation results in future inflation uncertainty because policy makers may fear the resultant recessionary effects and therefore be reluctant to lower inflation. Ball (1992) formalizes the view of Friedman in a game-theoretic setting among the monetary authority and the public to demonstrate that inflation uncertainty tends to be higher during periods of high inflation. Ball’s model assumes two policy makers, who alternate power stochastically, of which only one is willing to disinflate the economy through a recession. For the low levels of inflation observed in the economy, both policy makers aim to keep inflation at these levels that give rise to low inflation uncertainty in the eyes of economic agents. However, for the high levels of inflation, the public is uncertain for how long it will take the policy makers to disinflate the economy. In this case, uncertainty regarding future monetary policy would be greater and as a result inflation would be able to cause inflation uncertainty. Considering the reverse linkage that inflation variability leads to higher inflation, Cukierman and Meltzer (1986) employ the Barro-Gordon set-up to show how opportunistic central bank behaviour (the creation of inflation surprise in order to stimulate output growth) raises the optimal average rate of inflation as a result of an increase in the public’s uncertainty about money growth and inflation. Cukierman and Meltzer postulate that central banks prefer to either expanding output by making monetary surprises or keeping inflation at low levels. During periods of increased uncertainty, monetary policy tends to be discretionary due to the lack of commitment mechanisms so a central bank central has increased incentive for acting opportunistically in order to stimulate output growth by making monetary surprises. Thus, inflation uncertainty leads to higher money growth rates and inflation than what is expected by economic agents due to opportunistic central bank behaviour. There are also the “Pourgerami and Maskus’’ and “Holland’’ Hypotheses that reject the harmful effect high inflation has on predictability of prices so a negative relation between inflation and inflation uncertainty is thus postulated by these hypotheses. In

39

Vol. 12, No.2


contrast to Friedman-Ball Hypothesis, Pourgerami and Maskus (1987) argue that higher inflation leads economic agents to invest more in generating accurate predictions, which reduces their prediction error. Therefore, with rising inflation, agents may forecast inflation better due to the improved prediction capacity. Similarly, Holland (1995) asserts that higher inflation variability lowers inflation due to stabilization motives of policymakers. In the so-called “stabilizing Fed hypothesis”, Holland assumes that stabilization tendencies of central bank increase in high inflation periods in order to reduce the welfare costs of disinflationary policies when inflation uncertainty is high. In sum, we see from the conclusions from these theories that the causality between inflation and inflation uncertainty is uncertain which implies that ultimately, the effects can only be assessed empirically. Thus, the applicability and relevance of any of these theories to the Ghanaian case would be determined jointly by the sign and significance of the inflation and uncertainty variables in the causality analysis. Empirical Literature review In testing the different hypotheses indicated above, generalized autoregressive conditional heteroscedasticity (GARCH) models and its extensions are common methods used in the inflation-inflation uncertainty literature since the estimated conditional volatilities can serve as better proxies for inflation uncertainty. Studies examining the link between inflation and inflation uncertainty fall into two groups. The first group tests the Friedman-Ball hypothesis by including the level or the one-periodlag of the inflation rate into the conditional-variance equation of a GARCH process. A positive and significant coefficient for the inflation variable in the variance equation provides support for the Friedman-Ball hypothesis. The second group makes use of a two-step procedure by estimating the conditional variance of inflation, as a measure of inflation uncertainty and then perform the Granger causality tests between these generated conditional variance measures and inflation series. Granger causality tests involving the rate of inflation and the conditional variance estimates and the sign and the significance of the causal effect in the causality analysis are used to support one of the four hypotheses. Using various methodologies and datasets, the studies that have examined the linkages between inflation and inflation uncertainty can be grouped into those that support or reject the aforementioned hypotheses. In general, while majority of the studies support the Friedman-Ball Hypothesis much more than Cukierman–Meltzer Hypothesis, there is also very little support for the Pourgerami and Maskus Hypothesis and Holland Hypothesis. The conclusions of these studies are summarized under Panels A, B, C and D of Table A1 in the appendix.

40

Vol. 12, No.1


For the studies that have direct bearing on the data and methodology employed in our study, Fountas (2010) uses a GARCH-in-Mean (GARCH-M) model that augmented the conditional variance equation with lagged inflation using annual inflation data spanning over one century for 22 industrial countries. He finds significant evidence for a positive effect of inflation uncertainty on inflation supporting the Cukierman–Meltzer hypothesis. By employing exponential GARCH models and impulse-response functions, Hegerty (2012) uses monthly data from 1976 to 2011 to examine inflation and inflation uncertainty links for nine Sub-Sahara African countries (that is, Burkina Faso, Botswana, Cote d’Ivoire, Ethiopia, Gambia, Kenya, Nigeria, Niger and South Africa).It was found that inflation increases fuel uncertainty in all countries, while the reverse relationship holds for only half of the countries (that is, Burkina Faso, Gambia, Kenya and Nigeria). Also, multi-country Granger causality tests suggest that regional spillovers are weak except for South Africa and country pairs such as Burkina Faso and Cote d’Ivoire. In using data which is based on periods of implementation of IMF/World Bank economic adjustment and recovery programs for Senegal, Ghana and Uganda and GARCH framework, Sintim-Aboagye and Byekwaso (2005) find unidirectional links between inflation and uncertainty for Ghana and Uganda supporting Friedman-Ball hypothesis. However, Senegal demonstrated a bidirectional causal relationship between inflation and uncertainty which affirms both Friedman-Ball and CukiermanMeltzer Hypotheses. Sintim-Aboagye et al. (2012) utilized a GARCH model and found that countries with high central bank independence in the short run experience unidirectional links between inflation and uncertainty, affirming Friedman-Ball hypothesis. On the other hand, the low central bank independent countries display a bidirectional causal links in the short run, thus providing support for both the FriedmanBall and Cukierman-Meltzer hypothesis. In the long run however, both low and high central bank independent countries either separately or jointly provided significant evidence supporting the Friedman-Ball hypothesis. Thus, they concluded that their empirical investigation suggests that the Friedman-Ball hypothesis is a long run phenomenon, irrespective of the differences in degrees of central bank independence. Our brief review of the large body of empirical literature on inflation-inflation uncertainty shows the bulk of these studies make use of data from industrialized and emerging market economies. Specifically, while Fountas (2010) applied a GARCH-inMean to inflation data for 22 industrialized countries, Neanidis and Sava (2011) examined the cases of new EU member states. For the other cited works, their GARCH models are applied to data from Asian and Latin American countries (see table A1 in the appendix). As far as the empirical literature on Ghana is concerned, there is as at now no study that uses Ghanaian data to investigate the real effects of inflation by

41

Vol. 12, No.2


simultaneously controlling for inflation and inflation uncertainty in a GARCH framework. We therefore propose GARCH-M model that is able to control for simultaneous feedback effects between inflation and inflation uncertainty in Ghana.

METHODOLOGY AND MODEL SPECIFICATION AND DATA Methodology Following Fountas (2010), we propose the following AR(p)-GARCH-Mean Model: p

 t  0  i t i   D t21   t

(2)

i 1 k

q

i 1

i 1

t2  0   i t2i   i t2i   t 1

(3)

Where π denotes inflation rate, D4 is a dummy variable with one for extreme high inflation rates and zero for otherwise, δ2 is a measure of inflation uncertainty conditional on past information and εt is the disturbance term at time t. Equation (2) is a mean inflation equation which follows an autoregressive, AR(p) process. The optimal lag length (p) in equation (2) is selected based on the Box-Jenkins procedures and information criterion (eg. AIC and SC). Following the discussion on the year-on-year definition of inflation in the preceding section, an AR specification for the inflation equation should be adequate to control for seasonal effects. The inclusion of lagged volatility and inflation in two equations is a direct way of assessing the applicability of either Friedman-Ball or Cukierman hypothesis for Ghana. If  is non-zero, the impact of inflation uncertainty has effects on inflation; when  is positive an increase in inflation causes more inflation uncertainty.

Using standard Box-Jenkins techniques, we find an AR (3) to be the best fitting time series model for Ghanaian inflation over the full sample period5. Our choice of AR for the mean equation is supported by the work of Grier and Perry (1998) who used AR 4

Fountas (2001) uses dummy variables to capture structural breaks during high inflation times using annual data for the U.K, Hwang (2001) and Levent (2010) incorporate dummy variables for the U.S. and Turkey respectively. 5

We performed ARCH-LM test using inflation series where inflation is modelled as AR(3) series which is suggested by Final Prediction Error (FPE) criteria. The FPE allows us to select the optimum lag order such that the errors of the regression are no longer auto correlated.

42

Vol. 12, No.1


(12) process in their study of the G-7 countries. For the conditional variance equation, GARCH (1, 2) model augmented with lagged inflation provides the best fit for inflation uncertainty6:

 t  0  1 t 1  2 t 2  3 t 3   D t21   t t2  0  1 t21  1 t21  2t22   t 1

(4) (5)

There are at least two advantages for using GARCH-Mean model over the standard GARCH models. First, we are able to account for the simultaneous dependence of the two variables on each other. Second, causality analysis is justified only if the relation between inflation and inflation uncertainty is determined from the mean and variance equations.

Causality Relationships between Inflation and Inflation Uncertainty To complement the results from the mean and volatility equations, we use Granger noncausality analysis to investigate the causality relationship between the rate of inflation and inflation-uncertainty. The following kth order vector autoregressive [VAR(k)] framework is the basis of our causality analysis: k

k

i 1

i 1

UNCt  c0  iUNCt i   i INFt i   t ..............................(6) k

k

i 1

i 1

(6)

INFt  b0   i INFt i  iUNCt i  ut ...............................(7) (7)

Where co and b0 denote the constant terms in the Granger regressions and (k) represents the lag length chosen for the causality analysis. In the above single equations VAR(k) model, Granger non-causality hypotheses are tested with Wald statistics at lag interval to be specified by appropriate lag length selection criterion. In equation (6) the null hypothesis is that inflation does not Granger cause inflation uncertainty and likewise the null hypothesis in equation (7) is that inflation uncertainty does not granger cause

6

Other representations of the GARCH process are possible for the conditional inflation variance. We considered other estimations, but found that the GARCH(1,2) model is the best for the 1964:04-2012:12 period.

43

Vol. 12, No.2


inflation. The rejection of the latter (former) hypothesis implies an acceptance of the Friedman-Ball (Cukierman-Meltzer) hypothesis. The sign-consistency of the findings is checked with respect to the sign corresponding to the sum of the estimated β i or αi coefficients Data We use monthly CPI data obtained from Ghana Statistical Service and calculate inflation rate on a year-on-year (y-o-y) basis by taking the percentage change in the CPI with its value from the corresponding month in the previous year:

 CPIt  CPIt 12   *100 CPIt 12  

t  

(1)

The sample data set covers the period from April 1964 to December 2012 and includes 585 monthly observations. There are several advantages to using equation (1) for calculating inflation rates as compared to the month-on-month formula. Most importantly, inflation rates calculated on a y-o-y basis are implicitly seasonally adjusted as each month is compared with the corresponding month in the previous year. Thus, these inflation rates not only reflect the underlying trend in the data, they are also capable of capturing deviations from expected seasonal behaviour (Ryan and Milne, 1994; Cheem, 2003). With the data implicitly seasonally adjusted, an Autoregressive (AR) process will be sufficient to model the inflation process. Table 2 below presents the descriptive statistics for Ghana’s inflation series. The inflation rate has a high standard deviation that exceeds the mean value by nearly one percentage point. The Kurtosis statistic (7.59) shows that the distribution of the inflation rate is non-normal and skewed to the right and as indicated by the large value of both the Shapiro-Wilk W and Q statistics the data deviates from normality. As indicated by the large value of the LM(12) statistic which turns out to be significant at the 1% level, the inflation series in Ghana exhibit ARCH effects.

44

Vol. 12, No.1


Table 2: Summary Statistics for Inflation

Descriptive Statistics Mean Median Minimum Maximum Standard Deviation Skewness Kurtosis Shapiro-Wilk W

29.62 19.91 -12.08 174.14 30.64 2.15 7.59 98.57 (0.000) 1075.57 (0.00) 105.18 (0.00)

Portmanteau (Q) Statistics LM (12)

In order to test the stationary properties of the inflation series, we employ six unit root procedures - the Augmented Dickey-Fuller (ADF), Elliot et al. (1996) DF-GLS, Phillips and Perron (1998) P-P, Kwiatkowski-Phillips-Schmidt-Shin (KPSS), ZivotAndrews (ZA) and Clemente-Montanes-Reyes tests. Panel A in Table 3 shows the results for the ADF, DF-GLS, P-P, and KPSS tests. Irrespective of the lag lengths used, all four tests show that the inflation series in Ghana is stationary. In Panels B and C, the results from the ZA and Clemente-Montanes-Reyes tests show that the null hypothesis of a unit root with no break against the alternative of a stationary process with break(s) is rejected. The break date turns out to be July of 1977 for ZA test. For the Clemente-Montanes-Reyes test, the two break dates are July 1975 and July 1983. All these dates correspond to periods where the highest ever rates of inflation were recorded in Ghana. The results from these tests indicate that the monthly inflation data in our empirical analysis follows a stationary process over the 1964-2012 periods and so we can estimate inflation uncertainty using GARCH modelling.

45

Vol. 12, No.2


Table 3: Unit Root Test Results Panel A: Unit Root with no Structural Break Lags ADF DF-GLSS P-P 2 -5.247 -4.9 -3.484 4 -4.943 4.602 -3.945 6 -4.98 -4.63 -4.148

KPSS 1.32 0.821 0.614

Panel B: Zivot-Andrews test with one Structural Break Test Statistics Break Date ZA test -6.591 1977:07 Panel C: Clemente-Montanes-Reyes unit-root with two Structural Breaks Test Stat. Break Dates du 1 4.298 (0.00) 1975:07 du 2 -4.412 (0.00) 1983:07

EMPIRICAL RESULTS Estimates of Inflation Uncertainty The results in Table 4 show the maximum likelihood estimates of AR(3)-GARCH(1,2)M model for inflation rate that has three lags of inflation and a dummy variable that captures the dampening effect of the ERP/SAP in 1983. Diagnostic tests on the residuals of both the mean and conditional variance equations are shown in the lower part of panel A and B. The mean and variance models exhibit productive power with the statistical significance of the overall F-statistics at the 1 percent level. The Q and LM Tests of the residuals in the mean and conditional variance equations reveal no serial correlation and autoregressive conditional heteroskedasticity. In Figure 2 below, mean inflation and uncertainty derived from equations (4) and (5) are depicted. Mean inflation is approximated well by this model and our conditional variance measure exhibits instability over time, and its surge varies directly with level of inflation rate. Thus, AR(3)-GARCH-M model seems to fit both the mean and variance of Ghanaian inflation well and as such the conditional variance from this model can be used to proxy inflation uncertainty.

46

Vol. 12, No.1


Figure 2: Inflation and Inflation Uncertainty

Source: Authors calculation

In Panel A of table 5, the results for the mean equation indicate that the coefficient for the dummy variable for the ERP/SAP period is insignificant at conventional levels. In addition, the coefficient of the conditional variance in the mean equation is positive but insignificant. There is thus no contemporaneous relationship between inflation uncertainty and inflation. For the conditional variance equation, our estimation results indicate that the coefficients for the ARCH term is significant at the 10 percent level while that of the two GARCH terms are all significant at the 5 percent level (Panel B of table 4). The results strongly support the presence of a positive relationship between the level of inflation and its uncertainty. It can thus be concluded that the high inflation in Ghana for more than past two decades has been costly not only via the commonly known channels of distorted prices and income; but also via the inflation uncertainty channel. The reported coefficient of lagged inflation in the variance equation is positive and highly significant. This supports the Freidman-Ball hypothesis that inflation uncertainty increases with the level of inflation. The numerical estimate shows that if inflation increases by one unit, its conditional variance rises by 0.07. We turn to Granger causality tests to determine the direction of causality between inflation and its volatility.

47

Vol. 12, No.2


Table 4: AR (3) –GARCH-M Model for inflation and inflation Uncertainty in Ghana Panel A: Mean Equation (AR 3)

Constant

 t 1  t 2  t 3

InfDum

2

Model Diagnostics Q Test

LM Test

Coefficient 0.671 1.451 -0.316 -0.159 -0.470 0.001

Std. Error 0.369 0.061 0.094 0.053 0.371 0.004

Prob. 0.069 0.000 0.001 0.003 0.205 0.791

R-Squared 0.982 Q (1) 1.725 [0.725]

F-Statistics 278.324 Q (4) 3.281 [0.527]

Prob. 0.000 Q (8) 4.315 [0.381]

ARCH (1) 1.65 [0.328]

ARCH (4) 3.78 [0.256]

ARCH (8) 4.97 [0.245]

Coefficient -6.152 0.374 0.444 0.284 0.066

Std. Error 1.81 0.144 0.16 0.126 0.014

Prob. 0.001 0.01 0.006 0.024 0

Q (1) 1.725 [0.624]

Q (4) 3.285 [0.576]

Q (8) 4.315 [0.357]

ARCH (1) 1.65 [0.328]

ARCH (4) 3.78 [0.251]

ARCH (8) 4.97 [0.215]

Panel B: Variance Equation: GARCH (1,2) – M

Intercept ARCH (1) GARCH (1) GARCH (2)

 t 1

Model Diagnostics Q Test

LM Test

48

Vol. 12, No.1


Granger Causality Test The results of Granger-causality tests for Ghanaian inflation and inflation uncertainty for the full sample period are reported in Table 5. Panel A shows that over the full sample period, the null hypothesis that inflation does not Granger-cause inflation uncertainty is rejected at the 0.01 across lags 4 to 12. Furthermore, since the sum of the coefficients is positive in all cases, these results indicate that an increase in the Ghanaian inflation rate "Granger-causes" greater inflation uncertainty. These results support the Friedman-Ball Hypothesis that an increase in inflation causes an increase in inflation uncertainty. Table 5: Granger Causality Test between Inflation and Inflation Uncertainty over the Period 1964:04 – 2012:12 Panel A: H0 : Inflation uncertainty does not Granger-cause Inflation (1964 - 2012) Lags F-Statistics 4 20.23***(+) 8 36.66***(+) 12 68.06***(+)

Panel B:

H0 : Inflation does not Granger-cause Inflation Uncertainty (1964 - 2012)

Lags 4 8 12

F-Statistics 84.19***(+) 103.50***(+) 123.46***(+)

Note : *** indicates significance at the 0.01 level.

In Panel B, the null hypothesis that uncertainty does not Granger-cause inflation is also rejected at the 0.01 level for lags 4, 8, and 12. The sum of the coefficients on lagged uncertainty in the inflation equation is also positive, indicating that increased inflation uncertainty leads to higher future inflation over the full sample period. Thus, we interpret the finding to mean that increased inflation first raises inflation uncertainty, which in turn raises actual inflation. On the other hand, lower inflation outturns lead to lower future uncertainty which also directly reduce the rate at which prices go up. So, there is evidence of bi-directional causality between inflation and inflation uncertainty which supports the Friedman-Ball as well as Cukierman and Meltzer hypotheses. Are these results altered when our analysis is restricted to sub-periods that relate to the various political, social and economic policy regimes in Ghana? These issues are discussed for the three sub-periods we identified in section 2 above.

49

Vol. 12, No.2


We next investigate the relationship between inflation and inflation uncertainty in three sub-periods and report these results in panels A, B and C in Table 6. These periods have been selected to reflect the major changes in the political, economic and policy environment in Ghana. The selected sub-periods are 1993-2012, 2007- 2012, and 2010–2012. The first sub-period, which corresponds to the sustained democracy period, captures the role democracy has on inflation uncertainty. Davis and Kanago (1998) find that political instability may generate both high and uncertain inflation uncertainty. For the second period, we look at the role of inflation targeting in reducing inflation uncertainty. Onur Tas (2009) demonstrates that inflation uncertainty is significantly lower after the adoption of inflation targeting policies for both emerging and developed countries. Finally, in the third sub-period which pertains to the period Ghana has experienced the longest and sustained period of single digit inflation, we are particularly interested in assessing whether low and single digit inflation have any effect on inflation uncertainty. In each sample period, the procedure outlined in section 4 above is used to re-estimate equations (6) and (7). Table 6: Granger Causality Tests for Inflation and Inflation Uncertainty for Sub-Periods Panel A: Sample Period: 1993 to 2012 December (Sustained Democracy Period)

Lags 4 8 12

H0: Inflation does not Granger-cause Inflation Uncertainty F-Statistics 12.92**(+) 18.94**(+) 38.54***(+)

H0: Inflation Uncertainty does not Granger-cause Inflation F-Statistics 1.643 7.92 54.58***(+)

Panel B:Sample Period: 2007 May: 2012 December (Inflation Targeting Period) Lags F-Statistics F-Statistics 4 43.04 ***(+) 10.79**(+) 8 64.46***(+) 25.33**(+) 12 58.37 ***(+) 61.37***(+) Panel C:Sample Period: 2010 June-2012 December (Continuous Single Digit Inflation Period) Lags F-Statistics F-Statistics 4 39.51***(+) 6.62 8 140.57***(+) 25.42**(+) 12 58.37 ***(+) 97..43***(+) Note : *** and ** indicate significance at the 0.01 and 0.05 levels respectively

50

Vol. 12, No.1


For all three sub-sample periods, as was the case for the full sample period, the effect of inflation on inflation uncertainty is consistently positive and significant (see panels A-C, Table 6). At all lag lengths and in all sample periods, we find that higher inflation is unambiguously associated with higher average inflation uncertainty at the 0.01 level of significance. Therefore, we find strong statistical support for the Friedman-Ball hypothesis. Given that the cost of inflation is positively related to the level of inflation, the higher average inflation (24.97%) during the first sub-period suggests that the real cost of inflation is higher in the sustained democracy period. The low inflation for the inflation targeting and the single digit inflation periods means that low inflation is good but its impact on inflation uncertainty is maximized in the long run. Test results for whether inflation uncertainty lowers or raises subsequent inflation are mixed for the sub-periods. For the sustained democracy period (1993-todate), panel (A) in Table 6 shows that the Cukierman and Meltzer hypothesis holds only at lag 12 and above. This finding implies that at 4 and 8 lags, there is no statistically significant relationship between inflation uncertainty and inflation. The interpretation that can be put forward for this finding is that the euphoria that greeted the move to the fourth republican constitutional rule did not translate into lowering inflation uncertainty. The effect inflation targeting has on inflation uncertainty is provided by the bidirectional causality results for the 2007-2012 period (Table 6, Panel B). At the 1 percent significant level and at all lag lengths, both the Friedman-Ball and Cukierman and Meltzer hypothesis hold. The reverse causality that runs from inflation uncertainty to inflation for this period implies that the low and declining average inflation have not lowered inflation uncertainty. The results for the single digit inflation also confirm the findings for the inflation targeting period. For the single digit inflation period, inflation uncertainty lowers inflation at 8 and 12 lags at the 5% and 1% level of significance respectively. At lag 4 however, we find no statistically significant relationship between inflation uncertainty and inflation. Thus, inflation uncertainty does not translate into higher inflation in the short run. Thus, in the short run, economic agents could be unsure of the sustainability of inflation rates at the current low level.

51

Vol. 12, No.2


CONCLUSION This study explored the linkage between inflation and inflation uncertainty in Ghana over the period 1964:04–2012:12 by using AR(3)-GARCH(1,2)-M model. We find overwhelming evidence that increased inflation significantly raises inflation uncertainty in Ghana between 1964 and 2012 and in various sub-samples. The evidence on the effect of inflation uncertainty on average inflation is mixed and depends on the time period examined. The results from the inflation equation show that rising inflation increases inflation uncertainty, providing empirical support for Friedman's hypothesis. The significant inflation coefficient could be interpreted to mean that lower inflation reduces inflation uncertainty in Ghana. For the Granger causality tests, the overwhelming evidence that causality runs from inflation to its uncertainty for both long and short horizons confirms the results from the inflation equation that lower inflation causes lower inflation uncertainty. Given the results that inflation uncertainty affects future inflation only in the long run; our analysis suggests that a tough anti-inflation stance will be successfully in reducing long-run uncertainty in Ghana. Since the benefits of keeping inflation low and predictable are enormous, the major contributory factors and drivers of current inflation must at all times be explained to the general public in order to help rationalize their inflation expectations. Our results also have implications for the formulation of policies targeted at lowering inflation uncertainty. Moreover, based on findings of our work, and in agreement with Friedman-Ball hypothesis, we do conclude that a stable inflation will result in reducing inflation uncertainty which in turn can improve economic performance in Ghana. Thus, the results of our study justify low and declining inflation as the main target of monetary policy of the Bank of Ghana.

52

Vol. 12, No.1


REFERENCES Ball, L. P., (1992), “Why Does High Inflation Raise Inflation Uncertainty?” Journal of Monetary Economics”, 29, p. 371-388. Bhar R., Mallik, G., (2010), “Inflation, Inflation Uncertainty and Output Growth in the USA”, Physica A, 389, p. 5503-5510. Caporale, B. and Caporale, T. (2002), Asymmetric effects of inflation shocks on inflation uncertainty, Atlantic Economic Journal, Vol. 30, pp. 385-8. Caporale, G.M., L. Onorante, and P. Paesani, (2009), ―Inflation and Inflation Uncertainty in the Euro Area,‖CESifo Working Paper No. 2720. Caiman, T. and D. Jansen (1988), “Estimates of the Variance of U.S. Inflation Based Upon the ARCH Model: A Comment”, Journal of Money, Credit, and Banking, Volume 20, Issue 3, pp. 409-21. Cukierman A., Meltzer A., (1986), “A theory of Ambiguity, Credibility and Inflation under Discretion and Asymmetric Information”, Econometrical, 54, p. 10991128. Daal, E., A. Naka, B. Sanchez (2005), ‘Re-examining inflation and inflation uncertainty in developed and emerging countries’, Economics Letters, 89, 180186. Doe, S. K. (2012), “The relationship between Inflation, Inflation Uncertainty and Interest Rate in Ghana”. A dissertation submitted to the Department of Economics, KNUST Engle, R. (1982), “Autoregressive Conditional Heteroskedasticity with Estimates of the Variance of United Kingdom Inflation”, Econometrical, Volume 50, Issue 4, pp. 987-1008. Engle, R. (1983), “Estimates of the Variance of U.S. Inflation Based Upon the ARCH Model”, Journal of Money, Credit, and Banking, Volume 15, Issue 3, pp. 286301. Erkam, S. and T. Çavusoglu. (2008). “Modelling Inflation Uncertainty in Transition Economies: The Case of Russia and the Former Soviet Republics.” Economic Annals 53, 178-179: 44-71.

53

Vol. 12, No.2


Foster, E. (1978), “The Variability of Inflation”, Review of Economics and Statistics, Volume 60, Issue 3, pp. 346-350. Fountas S., 2001, “The Relationship between Inflation and Inflation Uncertainty in the UK; 1885-1998”, Economics Letters, 74, p. 77-83. Fountas S., Ioannidis, A., Karanasos, M., (2004), “Inflation, Inflation Uncertainty and a Common European Monetary Policy” Manchester School, 72, p. 221-242. Fountas, S., Karanasos. M., (2007), “Inflation, Output Growth, and Nominal and Real Uncertainty: Empirical Evidence for the G7”, Journal of International Money and Finance, 26 (2007), p. 229-250. Fountas S., (2010), “Inflation, Inflation Uncertainty and Growth: Are They Related”, Economic Modeling, 27, p. 869-899. Friedman, M., (1977), “Nobel Lecture: Inflation and Unemployment”, Journal of Political Economy, 85, p. 451-472. Girijasankar Mallik, Anis Chowdhury, (2011) "Effect of inflation uncertainty, output uncertainty and oil price on inflation and growth in Australia", Journal of Economic Studies, Vol. 38 Iss: 4, pp.414 - 429 Grier K. B., Grier, R., (2006), “On the Real Effects of Inflation on Inflation Uncertainty in Mexico“, Journal of Development Economics, 80, p. 478-500. Grier K. B., Perry M. J., (1998), “On the Inflation and Inflation Uncertainty in the G-7 Countries”, Journal of International Money and Finance, 17, p. 671-689. Hegerty, S. W. (2012), “Does High Inflation Lead to Increased Inflation Uncertainty? Evidence from Nine African Countries”. African Economic and Business Review Vol. 10, No. 2. Holland, A. S., (1995), “Inflation and Uncertainty: Test for Temporal Ordering”, Journal of Money, Credit and Banking, 27, p.827-837. Hutchful, Eboe. 2002. Ghana’s Adjustment Experience, The Paradox of Reform. United Nations Research Institute for Social Development (UNRISD). London: James Currey; Oxford: Woeli. Hwang, Y., (2001), “Relationship between Inflation and Rate and Inflation Uncertainty”, Economic Letters, 173, p.179-186.

54

Vol. 12, No.1


Jha, R. and Dang. T. (2011), “Inflation variability and the relationship between inflation and growth” ASARC Working Paper Jiranyakul, K., Opiela P. T., 2010,”Inflation and Inflation Uncertainty in the ASEAN5 Economies”, Journal of Asian Economics, 21, p. 105-112. Kapur, I., M.T. Hadjimichael, P. Hilbert, J. Schift and P. Szymczak. 1991. Ghana: Adjustment and Growth, 1983–91. Washington, D. C.: IMF. Karahan, O., (2012), “The Relationship between Inflation and Inflation Uncertainty: Evidence from the Turkish Economy”, Elsevier, International Conference of Applied Economics, Procedia Economics and Finance 1 ( 2012 ) 219 – 228 Karanasos, M., and J. Kim, 2005, •On the Existence or Absence of a Variance Relationship: A Study of Macroeconomic Uncertainty,. WSEAS Transactions on Computers, Vol. 4, pp. 1475.82. Keksek, S., Orhan, M., (2008), “Inflation and Inflation Uncertainty in Turkey”, Applied Economics, 42: 10, p. 1281-1291. Kontonikas, A. (2004), “Inflation and Inflation Uncertainty in the UK.Evidence from GARCH Modelling”, Economic Modelling, Volume 21, Issue 3, pp. 525-543. Korap, L. (2010). "Threshold GARCH modelling of the inflation & inflation uncertainty relationship: historical evidence from the Turkish economy," MPRA Paper 31765, University Library of Munich, Germany. Kwakye, J. K. (2010), “Assessment of Inflation Trends, Management and Macroeconomic Effects in Ghana”. IEA MONOGRAPH, NO 28 Logue, D. and T. Willet (1976), “A Note on the Relation between the Rate and Variability of Inflation”, Economica, Volume 43, Issue 17, pp. 151-58. Mladenoviæ, Z. (2009), “Relationship between Inflation and Inflation Uncertainty: The Case of Serbia”, Yugoslav Journal of Operations Research, Vol. 19 (2009), Number 1, 171-183. Mohammad A. M. (2009). “A GARCH Model of Inflation and Inflation Uncertainty in Iran. Nas, T. F., J. Perry., 2000, “Inflation, Inflation Uncertainty and Monetary Policy in Turkey: 1960-1998”, Contemporary Economic Policy, 18, 2, p. 170-180.

55

Vol. 12, No.2


Neanidis, C. K., Savva, S. C., 2011, “Nominal Uncertainty and Inflation: The role of European Union Membership”, Economic Letters, 112, p. 26-30. Okun, A., 1971, “The Mirage of Steady Inflation” Brooking Papers on Economic Activity, 2, p. 485-498. Özdemir, A. Z., M. Fisunoğlu, (2008) “On the Inflation-Uncertainty Hypothesis in Jordan, Philippines and Turkey: A Long Memory Approach” International Review of Economics & Finance, 17, 1, p.1-12. Payne, E. J., (2008), “Inflation and Inflation Uncertainty: Evidence from the Caribbean Region” Journal of Economic Studies, 35, 6, p.501- 511. Pourgerami, A., Maskus, K., (1987), “The Effects of Inflation on the Predictability of Price Changes in Latin America: Some Estimates and Policy Implications”, World Development, 15, 1, p. 287-290. Tas K. O. (2009), “Inflation targeting and inflation. TOBB University of Economics and Technology, Department of Economics. Work Pap. No.:09-07. Ankara. Tevfik F. Nas and Mark J. Perry (2001). “Inflation and Output Growth in Turkey, 19631999” University of Michigan-Flint. Thorton J., 2007, “The Relationship between Inflation and Inflation Uncertainty in Emerging Market Economies”, Southern Economic Journal, 73, 4, 858-870. Thorton J., 2008, “Inflation and Inflation Uncertainty in Argentina, 1810-2005”, Economic Letters, 98, p.247-252. Salmanpour, A. and Bahloli, P. (2011), “Inflation, Inflation Uncertainty and Factors Affecting Inflation in Iran”. World Applied Sciences Journal 14 (8): 12251239.IDOSI Publications. Samimi, A. J., Mohamadreza, A. and Ghader (2012), “Inflation and Inflation Uncertainty: Evidence from MENA”, Universal Journal of Management and Social Sciences. Sintim-Aboagye, H. and Byekwaso, S. (2005) “Inflation Uncertainty and Inflation: Implications of Adjustment and Economic Recovery Programs in Sub-Saharan Africa”. CERAF.

56

Vol. 12, No.1


Ungar, M., Zilberfarb, B., 199) “Inflation and Its Unpredictability- Theory and Empirical Evidence”, Journal of Money, Credit and Banking, 25, p. 709-720. Wilson, B. K., (2006), “The Links between Inflation, Inflation Uncertainty and Output Growth New Time Series Evidence from Japan”, Journal of Macroeconomics, 28, p. 206-220.

APPENDIX Figure 3: Trend of Inflation, 1964-2012

Figure 3 Inflation and Predicted Inflation from Model

57

Vol. 12, No.2


Table A1: Summary of Empirical Literature Author(s) Panel A: Fountas and Karanasos (2007)

Data

Methodology Friedman-Ball Hypothesis monthly data for Univariate GARCH the G7 covering the 1957-2000

Findings

Hwang (2001)

Monthly data from US covering 1926 to 1992

ARFIMA-GARCH models

Inflation affects its uncertainty weakly negatively whereas uncertainty affects inflation insignificantly

Wilson (2006)

Japanese inflation data spanning from 1957 to 2002

Bivariate EGARCHM model

Increased uncertainty is associated with higher average inflation and lower average growth

Fountas (2001)

UK inflation data for 1885-1998

GARCH

Supports Hypothesis

Kontonikas (2004)

UK data over the period 1972-2002

Symmetric, asymmetric and component GARCH-M models

Positive relationship between past inflation and uncertainty

Thornton (2007)

Monthly inflation uncertainty data from 12 emerging market economies which spans for 48 years. Caribbean Countries

GARCH model

Higher inflation rates increased inflation uncertainty in all the economies

ARMA-GARCH models

Results for Bahamas and Barbados support the Friedman-Ball Hypothesis whereas that of Jamaica advocate Holland Hypothesis

Argentina’s annual data between 1810 and 2005

GARCH model

Positive short run relation between the mean and variance of inflation

Payne (2008)

Thornton (2008)

58

Findings support FriedmanBall hypothesis.

Friedman-Ball

Vol. 12, No.1


Panel B: Neanidis and Sava (2011)

Cukierman-Meltzer Hypothesis New EU member states GARCH-M Model and candidate countries

Bhar and Mallik (2010)

United States from 1957 to 2007

Panel C: Jiranyakul and Opiela (2010)

Fountas et al. (2004)

Nas and Perry (2000)

Erkam and Cavusoghlu*

Inflation uncertainty increases inflation in the majority of the countries in the pre-EU accession period Inflation uncertainty increases inflation significantly

EGARCH-M model and Bivariate Granger causality test Friedman-Ball Hypothesis and Cukierman-Meltzer Hypothesis Data from Asian EGARCH model and Inflation can lead to Countries (Indonesia, Granger causality test inflation uncertainty Malaysia, Philippines, and uncertainty can Singapore and Thailand) lead to inflation over the period 19702007 Data from five European EGARCH model Inflation causes countries inflation uncertainty for France and Italy, but not Germany. Also, they found out that uncertainty causes inflation for France and Germany with a negative sign. Monthly inflation GARCH models Increased inflation uncertainty data from significantly raises Turkey covering 1960inflation uncertainty. 1998 However, the effect of inflation uncertainty on average inflation is mixed. Data from seven ARCH modelling Friedman-Ball transitional framework hypothesis was economies (Armenia, supported in Azerbaijan, Georgia, Azerbaijan, Russian Kazakhstan, Kyrgyz Federation and Republic, Russian Ukraine. However, Federation and Ukraine) Cukierman-Meltzer hypothesis is favoured in Kyrgyz Republic and in Russian Federation using a different model.

59

Vol. 12, No.2


Salmanpour and Bahloli (2011)

Inflation and inflation uncertainty data for Iran covering 1989-2003

GARCH model

Mixed results based on the periods under consideration.

Mladenović (2009)

Data from Serbia covering 2001 – 2007

GARCH

Korap (2010)

Monthly data from Turkish covering 19602008.

Threshold Modeling

High inflation leads to high uncertainty, while high uncertainty negatively affects the level of inflation at long horizon Inflation leads to inflation uncertainty. However, higher inflation uncertainty lowers inflation.

Mohammad (2009)

Monthly data from Iran spanning from 1959:03 –2008:02

TGARCH and CGARCH models

Ozdemir and Fisunoğlu (2008)

Jordanian, Philippine and Turkish CPI-based inflation series from 1987 to 2003

ARFIMA-GARCH

Chowdhury (2011)

Monthly data from India spanning 1954‐2010

GARCH model

Panel D: Caporale Caporale (2002)

GARCH

Bi-directional causality, and increased inflation raises inflation uncertainty Results support Friedman and Ball hypothesis. Also, they found a weak to support the Cukierman and Meltzer hypothesis. Both Friedman-Ball and CukiermanMeltzer hypotheses hold simultaneously in India

Other Macroeconomic Variables and Inflation Uncertainty and

Monthly U.S. data for the period 1961-2000

TARCH

60

Negative inflationary shocks result in greater inflation uncertainty than positive shocks.

Vol. 12, No.1


Tas (2009)

IFS inflation data for 19 inflation targeting countries as at October 2008.

PARCH approach

Inflation targeting countries have lower inflation variances after inflation targeting. Also, the decrease in inflation uncertainty is higher in emerging economies compared to developed countries.

Jha and Dang (2011)

WDI data covers 182 developing countries and 31 developed countries for the period 1961-2009.

GLS and 2SLS

For the developing countries, when the rate of inflation exceeds 10 per cent inflation variability has a negative effect on economic growth. For the developed countries, there is no significant evidence that inflation variability is detrimental to growth.

61