DISCUSSION PAPERS

IN

ECONOMICS

No. 2011/4 ISSN 1478-9396

INFLATION DYNAMICS AND POVERTY RATES:

REGIONAL AND SECTORAL EVIDENCE FOR

GHANA

SIMEON COLEMAN

JULY 2011

DISCUSSION PAPERS IN ECONOMICS

The economic research undertaken at Nottingham Trent University covers various

fields of economics. But, a large part of it was grouped into two categories, Applied

Economics and Policy and Political Economy.

This paper is part of the new series, Discussion Papers in Economics.

Earlier papers in all series can be found at:

http://www.ntu.ac.uk/research/academic_schools/nbs/working_papers/index.html

Enquiries concerning this or any of our other Discussion Papers should be

addressed to the Editors:

Dr. Simeon Coleman, Email: simeon.coleman@ntu.ac.uk

Dr. Marie Stack, Email: marie.stack@ntu.ac.uk

Dr. Dan Wheatley, Email: daniel.wheatley2@ntu.ac.uk

Division of Economics

Nottingham Trent University

Burton Street, Nottingham, NG1 4BU

UNITED KINGDOM.

Inflation dynamics and poverty rates: regional and sectoral evidence for Ghana. Simeon Coleman Economics Division, Nottingham Business School, Nottingham Trent University, Burton Street, Nottingham, NG1 4BU, UK. Email: simeon.coleman@ntu.ac.uk

Abstract In this paper, we investigate the co-incidence of inflation persistence and poverty in a small open economy, using disaggregated data for Ghana. We initially employ various fractional integration (FI) tests for (a) the existence of long-memory and (b) an estimate of the degree of persistence, where present, in regional and sectoral inflation. Further, based on an analysis of poverty rates across districts and regions in Ghana, we determine where the adverse impacts of inflation persistence would, most likely, hit hardest. Using monthly inflation series for the periods 2005:08-2010:06 (for regions) and 1997:10-2010:06 (for sectors), we suggest that the poor in three out of the nine regional groupings and patrons of five out of the twelve sectors considered, will bear the brunt of any welfare-reducing impacts of inflationary shocks, particularly since three of the five sectors are basic necessities. Some policy implications are then discussed. JEL classification: E31, E61, I38 Keywords: Inflation persistence, asymmetry, welfare policy objectives, poverty rates.

1. Introduction

There is small, yet growing, empirical evidence suggesting that asymmetric price-related impacts of

common economic shocks within common currency areas are not uncommon. Recent examples include

Fielding and Shields (2011) for the USA, Mayes and Virén (2005) for the euro area, and Ceglowski (2003)

for Canada; and Coleman (2010) and Fielding and Shields (2006) for some developing economies i.e.,

the CFA Franc zone and South-Africa respectively. The consensus is that, serious consideration of

regional asymmetries is important for improved policy formulation. Fielding and Shields (2011) propose

that the degree of asymmetry resulting from common policy should be relevant, even at the national level

and, not only in international monetary unions. In fact, such findings offer vital information for the

monetary policy formulation process, and even more so under inflation targeting.1 Ample theoretical and

empirical evidence exist, which posit that policy formulation which ignores regional asymmetry can be

welfare-depreciating. 2 Consequently, to the extent that consumers in different regions face different

inflation dynamics then the distribution of inflation rates resulting from, say, a common monetary policy

shock become relevant, and this should worry policymakers concerned about welfare. Moreover, this

suggests the potential existence of inflation persistence induced poverty traps i.e., those in poverty who

are confronted with welfare-decreasing impacts of inflation persistence are less likely to grow out of

poverty.

In this paper, we focus on one of the emerging economies in Africa, Ghana, where economic

growth has remained strong with real GDP growth reaching an estimated 5.9% in 2010 compared to 4.7%

in 2009. Growth prospects are even brighter with projected real GDP growth of 12.0% and about 11.0%

for 2011 and 2012 respectively, largely on account of the start of oil production in commercial quantities in

1 In Africa, the two African countries which have formally adopted inflation targeting are South Africa and Ghana. 2 For theoretical and empirical evidence see De Grauwe (2000) and Fielding (2004) respectively and references cited therein.

December 2010. In addition, the country’s increasingly democratic settlement and social stability have

served to boost investor confidence, leading to rising international investment. However, being small and

open, the Ghanaian economy remains vulnerable to a myriad of shocks (both internal and external),

which impact price and inflation levels and make two key welfare-related questions relevant: In the event

of an inflationary shock,

(i) Do regional and sectoral asymmetries exist?

(ii) Which regional areas (sectors) should receive the most urgent attention following the shock,

due to the higher incidence of existing poverty (proportional patronage by the poor)?

The answers to these questions are important for policy formulation aimed at mitigating the costs of

adjustment.3

Against this background, this aim of this paper is two-fold. First, we employ fractional integration

methods to investigate the incidence of inflation persistence at regional and sectoral levels for Ghana.

Then, given the welfare implications of asymmetric impacts of common shocks, the second aim is to

explore the co-incidence of inflation persistence and poverty at the regional level. Our analyses do not

conclude that the regions with the significant inflation persistence have the highest poverty rates.

Nonetheless, for the poor in the identified regions, the welfare implications will remain relevant.

Specifically, this paper aims at providing robust empirical evidence for policymakers in Ghana concerned

about where intervention priorities should lie (regionally and sectorally) when attempting to alleviate the

burdens of the poor following an inflationary shock.

3 As a country, Ghana is well endowed with natural resources, and has approximately twice the per capita output of the poorer countries in West Africa. However, dependence on international financial and technical assistance is also high. The domestic economy continues to revolve around subsistence agriculture, which accounts for 36% of GDP and employs 60% of the work force, mainly small landholders. Gold, timber, and cocoa production are major sources of foreign exchange. In 2002, Ghana opted for debt relief under the Heavily Indebted Poor Country (HIPC) program. Policy priorities include tighter monetary and fiscal policies, accelerated privatization, and improvement of social services.

The remainder of the paper is organized as follows. Section 2 describes the data and the

econometric strategy employed. Section 3 presents the empirical results and Section 4 discusses inflation

persistence and the incidence of poverty in Ghana. Finally, Section 5 discusses some policy implications

and concludes.

2. Data and econometric methodology

2.1 Data

The two inflation series we investigate are monthly, seasonally adjusted, and based on the Consumer

Price Index (CPI) publications for Ghanaian regions and sectors. Both have been sourced from the

Ghana Statistical Service (GSS), spanning the periods 2005:08-2010:06 (for nine regional groupings) and

1997:08-2010:06 (for twelve sectors).4 Tables 1-2 provide a summary of the statistical properties of the

inflation series for the regions and sectors respectively and show that despite similarities in mean values

across regions (sectors), the range of inflation varies significantly across the various regions (sectors).

Table 1: Summary statistics (Regions)

Region [weighting (%)] Observations Mean Standard deviation Minimum Maximum

Ashanti [22.34] 59 1.061 0.772 -0.292 3.770

Brong Ahafo [7.61] 59 1.159 0.806 -0.611 3.714

Central [6.95] 59 1.227 0.583 -0.055 2.600

Eastern [9.25] 59 1.072 0.637 -0.615 2.515

Greater Accra [24.21] 59 1.090 0.964 -2.436 3.283

Northern [4.89] 59 1.265 1.130 -1.432 3.623

Upper East & Upper West [2.91]

59 1.437 1.561 -1.485 5.742

Volta [10.28] 59 1.121 1.495 -2.388 6.391

Western [11.56] 59 1.024 1.075 -1.298 4.025

Aggregate (regions) [100] 59 1.090 0.350 0.261 1.786

4 Data source: P1.10m (for regions) and P1.02m (for sectors) respectively in the July 22, 2010 CPI report published by the GSS [see http://www.statsghana.gov.gh/cpi.html]. Prior to seasonal adjustment, we measure inflation as the monthly change in the price indices, calculated as 100(Pt -Pt−1)/ Pt−1). Further, we note that in this dataset, Upper East and Upper West regions have been grouped and are considered as a unit.

Table 2: Summary statistics (Sectors)

Sectors [weighting (%)] Observations Mean Standard deviation Minimum Maximum

Alcoholic beverages, tobacco & narcotics [2.23]

153 1.530 2.206 -4.570 16.700

Clothing & footwear [11.29] 153 1.256 1.709 -2.270 7.600

Communications [0.31] 153 0.956 5.488 -23.369 41.617

Education [1.6] 153 2.118 15.843 -37.180 153.930

Food & non-alcoholic Beverages [44.91]

153 1.347 1.914 -3.512 8.859

Furnishings, household equipment etc. [7.83]

153 1.350 1.920 -3.731 9.269

Health [4.33] 153 1.758 3.480 -15.960 11.830

Hotels, cafés & Restaurants [8.28]

153 1.790 4.228 -21.080 30.520

Housing, water, electricity, gas & other [6.98]

153 1.698 3.810 -8.995 20.401

Miscellaneous goods & services [2.99]

153 1.371 2.351 -7.050 12.820

Recreation & culture [3.04] 153 1.716 3.435 -3.651 19.159

Transport [6.21] 153 1.843 5.108 -6.960 46.540

Aggregate (sectors) [100] 153 1.432 1.555 -1.780 9.654

2.2 Tests for the order of integration, I(1) vs. I(0)

In order to test for the order of integration of inflation in the disaggregated data, we apply three separate

unit root tests: the Augmented Dickey-Fuller [ADF], (1979), the Ng and Perron [N-P], (2001) and

Kwiatkowski et al. [KPSS], (1992) tests. The KPSS (1992) test differs from the other unit root tests

employed in that the series xt is assumed to be (trend-)stationary under the null, whereas for the former

tests xt is assumed to possess a unit root under the null. Specifically, N-P (2001) propose four tests: MZa

and MZt, which are modified versions of the Phillips’ (1987) and Phillips and Perron’s (1988) Za and Zt

tests; and the MSB and MPT tests, which are related to the Bhargava (1986) R1 test and Elliot et al.

(1996) Point Optimal tests respectively.

2.3 Fractional integration tests

A stationary stochastic process xt is called a long-memory (fractionally integrated) process if there exists a real number d and a

finite constant φ such that the autocorrelation function ρ(τ) has the following rate of decay:

(1)

Alternatively, the long-memory process may be characterized by the behavior of its spectrum f ,

estimated at the harmonic frequencies , where j=1,2,…,[T/2], near the zero frequency i.e.,

, where φ is a strictly positive constant. 5 In other words, a long-memory

process with degree of long-memory is said to be integrated of order d and is denoted by I(d). Table 3

provides a summary of the implications of the parameter values for fractional integration.

Table 3: Parameter values and implications for fractional integration

d Variance Shock duration Stationarity

d=0 Finite Short-lived Stationary

0<d<0.5 Finite Long-lived Stationary

0.5≤d<1 Infinite Long-lived Nonstationary

d=1 Infinite Infinite Nonstationary

d>1 Infinite Infinite Nonstationary

Source: Tkacz (2001)

The class of long-memory processes generalizes the class of integrated processes with integer degree of

integration.

In this paper, we employ a battery of tests to categorize the order of integration of each inflation

rate: First, two tests of I(0) versus fractional alternatives as proposed by Lobato and Robinson (1998)

[lobrob]; the Rescaled Variance V/S Statistic [rvlm] as proposed by Giraitis et al. (1998); and then, the Modified

Log Periodogram Regression estimator [Phillips] by Phillips (1999a, 1999b) which provides an estimate of

5 See Beran (1994), Robinson (1994a) and Robinson and Henry (1998) for excellent and detailed surveys on long-memory processes.

the d parameter and further, allows testing both I(0) and I(1) against fractional alternatives. Second, we

estimate the d parameter based on the Whittle estimator (robwhittle).6

3. Results

Results of the preliminary unit root tests reported in Tables 4-5, though informative, are not conclusive. In

fact, in some cases, we obtain conflicting results based on the three tests e.g., for Brong-Ahafo, Volta and

Western regions, and also the Communication, Education and Transport sectors among others. Clearly,

this makes the explicit categorization of the order of integration of the inflation rate under investigation, as

either I(0) or I(1) problematic.

Table 4: Individual unit root tests (Regions)

Region [weighting (%)] ADF N-P KPSS

MZa MZt MSB MPT

Ashanti [22.34] -1.469 -1.828 -0.705 0.386 10.431 0.150

Brong-Ahafo [7.61] -4.401*** -1.411 -0.819 0.581 16.853 0.095

Central [6.95] -1.902 -1.689 -0.886 0.525 13.979 0.707**

Eastern [9.25] 0.524 0.029 0.013 0.469 17.744 0.259

Greater Accra [24.21] -3. 441** -19.03*** -3.084*** 0.162*** 1.291*** 0.101

Northern [4.89] -2.278 -1.632 -0.701 0.429 11.777 0.137

Upper East and Upper West [2.91]

-2.717* -2.721 -1.081 0.397 8.706 0.179

Volta [10.28] -2.996** -3.172 -1.258 0.396 7.720 0.128

Western [11.56] -6.522*** -0.659 -0.512 0.776 30.968 0.242

Aggregate (regions) [100] -3.715* -14.12*** -2.388** 0.169*** 2.724** 0.169

Critical Values

1% -3.560 -13.80 -2.58 0.174 1.78 0.739

5% -2.917 -8.100 -1.980 0.233 3.170 0.463

10% -2.596 -5.700 -1.620 0.275 4.450 0.347

Notes: The order of lag to compute the tests has been chosen using the modified AIC (MAIC) suggested by Ng and Perron (2001) and for KPSS tests, we employ the Newey-West bandwidth selection criteria using Bartlett kernel. The Ng-Perron tests include an intercept. *, ** and *** imply rejection of the null hypothesis of unit root at the 10%, 5% and 1% levels respectively. The critical values for the Ng-Perron tests have been taken from Ng and Perron (2001), whereas those for the KPSS have been obtained from Kwiatkowski et al. (1992, Table 1).

6 See Appendix I for a brief description of the tests employed.

The tests, however, appear to be unanimous in the finding that for both regional and sectoral aggregate

series, the series appear to be I(0). Given the inconsistencies observed, we proceed to investigate the

possibility of testing for the less restrictive fractional order of integration of each series.

Table 5: Individual unit root tests (Sectors)

Sector [weighting (%)] ADF N-P KPSS

MZa MZt MSB MPT

Alcoholic beverages, tobacco & narcotics [2.23]

-2.570 -5.284 -1.625 0.307 4.636 0.140

Clothing & footwear [11.29] -3.541*** -18.680*** -3.055*** 0.163*** 1.313*** 0.178

Communications [0.31] -4.707*** -3.882 -1.389 0.357 6.314 0.171

Education [1.6] -6.245*** -0.397 -0.361 0.907 42.710 0.142

Food & non-alcoholic Beverages [44.91]

-3.164** -17.828*** -2.985*** 0.167*** 1.374*** 0.173

Furnishings, household equipment etc. [7.83]

-3.444* -3.956 -1.363 0.344 6.239 0.097

Health [4.33] -7.316*** 0.218 0.337 1.544 131.049 0.101

Hotels, cafés & Restaurants [8.28]

-3.302** -1.130 -0.712 0.630 20.169 0.035

Housing, water, electricity, gas & other [6.98]

-4.542*** -0.675 -0.0.547 0.0.810 32.957 0.327

Miscellaneous goods & services [2.99]

-3.141** -13.689** -2.615*** 0.191** 0.1.793** 0.213

Recreation & culture [3.04] -3.708*** -12.280** -2.476** 0.201** 1.999* * 0.138

Transport [6.21] -7.530*** -1.630 -0.892 0.547 14.849 0.144

Aggregate (sectors) [100] -3.323** -9.176** -2.127** 0.231** 2.727** 0.178

Critical Values

1% -3.473 -13.80 -2.58 0.174 1.780 0.739

5% -2.880 -8.100 -1.980 0.233 3.170 0.463

10% -2.577 -5.700 -1.620 0.275 4.450 0.347

Notes: The order of lag to compute the tests has been chosen using the modified AIC (MAIC) suggested by Ng and Perron (2001) and for KPSS tests, we employ the Newey-West bandwidth selection criteria using Bartlett kernel. The Ng-Perron tests include an intercept. *, ** and *** imply rejection of the null hypothesis of unit root at the 10%, 5% and 1% levels respectively. The critical values for the Ng-Perron tests have been taken from Ng and Perron (2001), whereas those for the KPSS have been obtained from Kwiatkowski et al. (1992, Table 1).

3.1 Results for regional inflation rates

Here, we address the first question in the introduction, and to determine where in Ghana, and in which

sector(s), the effects of an inflationary shock will be persistent. Across the nine regional groupings, the

results reported in Table 6 are largely unanimous. The following observations are worthy of note: First,

both lobrob and rvlm (in columns 2-3) are agreed in rejecting the null of no long memory for three out of

the nine regional groupings i.e., for the Ashanti, Central and Eastern regions. Interestingly, both the

aforementioned tests reject the null of no long memory in the (regional) aggregated series. This result is

contrary to that obtained through the standard (and more restrictive) unit root tests we performed earlier,

underscoring the importance of investigating fractional orders of integration. Second, the Modified Log-

Periodogram Regression estimator (in columns 4-6) reports that, except for Ashanti, Central and Eastern

regions, the null of d=1 is rejected in all regions; and unsurprisingly, corroborating the lobrob and rvlm

tests, d=0 is also rejected in the same three regions. Thus, there is robust evidence of long-memory in the

inflation rates in these three regions. Third, to gain a measure of the degree of persistence, the long

memory estimate of d based on the Whittle estimator is reported in column 7.

Of the three identified regions, the Ashanti region has a relatively lower d estimate (at 0.584)

compared to the other two regions i.e., Central (0.968) and Eastern (1.031) regions, which are not

significantly different from 1 (the non-stationary unit root process). Further, we find the d estimate for the

aggregated series to be 1.184, higher than the degree of persistence for any of the constituent regional

series, and d=0 is also rejected at the standard significance levels.7 It suggests that the rejection of the

null of aggregate series being I(0) is being driven by the three regions found to possess non-stationary

inflation rates, and heavily influenced by the Ashanti region, which has a relatively large weighting of

22.34/100 compared to 6.95/100 and 9.25/100 for Central and Eastern regions respectively.

7 This result lends some support to the empirical findings (see for example Altissimo et al. (2009)) in the literature that aggregation can explain a significant amount of aggregate inflation persistence.

Table 6: Fractional integration tests (Regions)

Region [weighting (%)] lobrob rvlm Phillips robwhittle (d)

0.50 0.55 0.60

Ashanti [22.34] -2.422++ 0.242++ 1.216** 0.779 0.584^ 0.584

Brong-Ahafo [7.61] -0.062 0.107 -0.119^^^ -0.241^^^ 0.013^^^ 0.103

Central [6.95] -5.391+++ 1.038++ 1.229*** 0.776** 0.889** 0.968

Eastern [9.25] -4.661+++ 0.962++ 1.050*** 0.665** 0.685** 1.031

Greater Accra [24.21] -0.843 0.118 0.231^^^ 0.226^^^ 0.543^^ 0.070

Northern [4.89] -0.286 0.148 0.069^^^ 0.070^^^ 0.071^^^ -0.002

Upper East and Upper West [2.91]

1.222 0.072 -0.187^^^ -0.166^^^ -0.086^^^ -0.290

Volta [10.28] -0.912 0.106 -0.054^^^ -0.030^^^ 0.394^^^ 0.342

Western [11.56] -0.384 0.181 0.406^^^ 0.106^^^ 0.215^^^ 0.322

Aggregate (regions) [100] -4.635+++ 0.910++ 1.275*** 1.100*** 0.923*** 1.184 Notes: +,++ and +++ imply rejection of the null of no long memory at the 10%, 5% and 1% levels respectively. *(^), **(^) and ***(^) imply rejection of the null hypothesis of d=0 (d=1) at the 10%, 5% and 1% levels respectively. For consistency across sectors, and based on number of observations, a vector of bandwidth 10, 20, 25, and 50 is employed for lobrob, rvlm and robwittle tests. As results are fundamentally similar, the results reported are that for bandwidth of 25.

3.2 Results for sectoral inflation rates

This section presents the fractional integration estimates across the twelve sectors of the Ghanaian

economy over the sample period considered. The evidence, presented in Table 7, suggests that inflation

is persistent in five of the twelve sectors considered i.e., Clothing & footwear; Food and non-alcoholic

beverages; Furnishings, household equipment etc.; Housing, water, electricity, gas & other;

Miscellaneous goods & services; and also, the aggregate (sectoral) inflation rate. Although there is

consistent rejection of d=1 across all sectors, there are also some notable differences between the

estimated degree of inflation persistence, d, across the aforementioned fractionally integrated sectors.

Three of the five sectors which demonstrate evidence of persistence i.e., Clothing & footwear

(0.561), Food & non-alcoholic beverages (0.587) and Furnishings, household equipment etc. (0.535) are

non-stationary.8 Interestingly, these sectors with the highest estimates are, arguably, basic necessities,

which should be an important policy concern. Similar to the regional analysis, the incidence of persistence

8 See classification in Table 3.

in the aggregate series (0.503) appears to be driven by persistence in the five constituent series for which

the null of no long memory is rejected.

Table 7: Fractional integration tests (Sectors)

Sector [weighting (%)] lobrob rvlm Phillips robwhittle (d)

0.50 0.55 0.60

Alcoholic beverages, tobacco & narcotics [2.23]

-0.868 0.135 0.308^^^ 0.204^^^ 0.186^^^ 0.209

Clothing & footwear [11.29] -1.719+ 0.372++ 0.559*,^^^ 0.509**,^^^ 0.557***,^^^ 0.561

Communications [0.31] -0.216 0.138 0.204^^^ 0.269^^^ 0.292^^^ 0.053

Education [1.6] 0.555 0.058 -0.138^^^ 0.000^^^ 0.050^^^ -0.023

Food & non-alcoholic beverages [44.91]

-2.108++ 0.345++ 0.588**,^^ 0.555**,^^^ 0.433**,^^^ 0.587

Furnishings, household equipment etc. [7.83]

-1.853+ 0.303++ 0.351^^^ 0.561**,^^^ 0.539***,^^^ 0.535

Health [4.33] 1.586 0.031 -0.175^^^ -0.316^^^ -0.133^^^ -0.257

Hotels, cafés & restaurants [8.28]

-0.680 0.035 0.366^^^ 0.176^^^ 0.215^^^ 0.167

Housing, water, electricity, gas & other [6.98]

-0.878 0.201++ -0.153^^^ 0.191^^^ 0.071^^^ 0.047

Miscellaneous goods & services [2.99]

-1.595 0.235++ 0.296^^^ 0.403*,^^^ 0.354**,^^^ 0.359

Recreation & culture [3.04] -0.802 0.096 0.133^^^ 0.337^^^ 0.305^^^ 0.285

Transport [6.21] 1.196 0.061 -0.216^^^ -0.186^^^ -0.146^^^ -0.095

Aggregate (sectors) [100] 1.973++ 0.352++ 0.645^^ 0.484^^^ 0.340^^^ 0.503 Notes: +,++ and +++ imply rejection of the null of no long memory at the 10%, 5% and 1% levels respectively. *(^), **(^) and ***(^) imply rejection of the null hypothesis of d=0 (d=1) at the 10%, 5% and 1% levels respectively. For consistency across sectors, and based on number of observations, a vector of bandwidth 10, 20, 25, and 50 is employed for lobrob, rvlm and robwittle tests. As results are fundamentally similar, the results reported are that for bandwidth of 25.

Further, it is intuitive that with a weighting of 44.91/100, the impact of the Food & non-alcoholic beverages

sector is important and an important driver of the persistence observed in the aggregate series.

4. Disaggregated inflation persistence and poverty rates

There is some empirical evidence e.g., Romer and Romer, 1998 and Easterly and Fischer, 2000, which

posit that low aggregate inflation enhances welfare. Also, Fielding (2004) concludes that poorer

households in a subset of West African countries are more burdened by the price volatility that follows a

change in monetary policy by the central bank. Against this background, we infer that sluggishness of

inflation in returning to a target after a shock e.g., a change in central bank’s monetary policy, may have

deleterious implications for welfare vis-à-vis wealth distribution, real incomes and economic efficiency,

especially of the poor. Findings from this area of research should, therefore, be of considerable policy

importance. Specifically, of more direct interest to both policy makers and researchers should be spatial

and sectoral inflation persistence within a monetary area.

In this section, our aim is to address the second question in the introduction and to determine

whether, in Ghana, high inflation persistence can be spatially associated with high poverty rates.

Following our finding of persistence in three regions and five sectors of the Ghanaian economy, we

investigate whether the incidence of inflation persistence is mirrored by poverty rates across the country.

Spatially, average poverty rates (see Appendix II) are highest in the Upper East and Upper West regional

grouping, followed by the Northern region, the Brong-Ahafo and Volta regions.9 The regions for which we

find evidence of long memory in inflation have average poverty rates ranging between 0.164 and 0.241.

However, at the district level, estimates reported in Appendix II show that poverty rates are more diverse,

ranging between 0.080 for Cape Coast district in Central region to 0.343 for the Afram Plains district in the

Eastern region.

It is worth noting that, at the regional level, the highest estimates of inflation persistence are not

the areas with the highest levels of average poverty rates. Nonetheless, this is little consolation,

particularly for those within the low poverty bands within these three regions. This is compounded by our

finding that three of the five sectors for which there is evidence of inflation persistence are, arguably,

basic necessities on which both the poor and the non-poor have to spend their income on i.e., Clothing &

footwear; Food and non-alcoholic beverages; Housing, water, electricity, gas & other.10

9 These estimates provide some support to the proponents of the existence of a North-South poverty divide in Ghana. The North, defined as the sum of the administrative regions Upper West, Upper East, and the Northern regions, cover 40% of Ghana's land area and ; according to Table 1, interestingly, the Upper East and Upper West group also has the highest mean inflation for the sample period. 10 On balance, it is expected that, the poor will spend a larger proportion of their income on basic necessities compared to the non-poor.

5. Policy implications and concluding remarks

This paper empirically investigates the persistence properties of inflation across regions and sectors in

Ghana. The policy importance of this investigation is high, given the ample empirical research which find

that asymmetry in the impact of monetary policy is not uncommon, even in developed economies like the

USA, Canada and the UK. Using poverty maps, Elbers et al. (2007) also find that a useful way forward in

poverty alleviation might be combining fine geographic targeting using a poverty map with within-

community targeting mechanisms. Further, with the link between inflation dynamics and welfare well-

established empirically in the literature, our investigation of the co-incidence of high inflation persistence

and high poverty rates, spatially, is particularly relevant for forming policy to tackle poverty.

In this paper, we uncover evidence of some asymmetries in spatial and sectoral inflation persistence

for Ghana. On, the one hand, while poverty rates and mean aggregate inflation are highest in the Upper

West and Upper East regional group and also high in the other northern regions, it is the Ashanti, Central

and Eastern regions that demonstrate inflation persistence, implying that a nationwide policy shock will

have longer-lasting effects in these three regions compared to the other regions. Thus, the welfare-

retarding impacts of inflationary shocks, will linger longer in these three regions. Although poverty rates in

these three regions are not the highest within the country, that provides little comfort for those below the

poverty line in these regions, for whom welfare will decline significantly. On the other hand, inflation

persistence in the Clothing & footwear; Food and non-alcoholic beverages; Housing, water, electricity,

gas & other sectors, particularly, imply that those who have to spend their income in these sectors are

likely to feel the impact more. In fact, the heaviest impact will be felt by those within the three

aforementioned regions who have to spend in these sectors.

To the extent that the Ghanaian policymaker is intent on improving welfare within the country,

particularly, following an unanticipated inflationary shock, implementation of some similarly asymmetric

support and attention may be necessary. In a developing country, like Ghana, despite the reality of limited

monitoring ability (and hence infrequent use), available policy options may include the imposition of

temporary regional or sectoral price ceilings e.g., as in China regional retailers in December 2010 and

other more draconian instances such as the Russian food situation in 1998; availability of targeted

regional or sectoral subsidies; provision of food stamps to targeted groups - all in attempt to reduce the

adverse asymmetric impacts of inflation persistence within the country.

In the aftermath of an inflationary shock, while policymakers’ intervention measures is likely to be drawn

to regions with the highest poverty rates and average aggregate inflation, this study suggests that these

are not necessarily the regions that suffer the lingering impacts of inflationary shocks. Therefore, the poor

in the regions with high inflation persistence should receive relatively longer lasting support. An improved

understanding, on the part of policymakers, of these asymmetries vis-à-vis poverty could lead to

important improvements in the policy measures formed to decrease the impact of inflationary shocks.

Although our study is confined to inflation dynamics in Ghana, we suggest that the results obtained here

could provide valuable insights for formulation of intervention policies for welfare improvement and

poverty reduction in other countries that face similar economic situations.

References 1. Altissimo, F., Mojon, B., and Zaffaroni, P., 2009. Can aggregation explain the persistence of inflation?

Journal of Monetary Economics. 56, 231-241.

2. Bhargava, A., 1986. On the theory of testing for unit roots in observed time series. Review of Economics Studies. 53, 369–384.

3. Ceglowski, J., 2003. The law of one price: intra-national evidence for Canada, Canadian Journal of Economics. 36, 373–400.

4. Coleman, S., 2010. Inflation persistence in the franc zone: evidence from disaggregated prices. Journal of Macroeconomics. 32, 426-442.

5. De Grauwe, P. 2000. Monetary policies in the presence of asymmetries. Journal of Common Market Studies. 38, 593–612.

6. Dickey, D.A. and Fuller W.A., 1979. Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association. 74, 427-431.

7. Easterly, W., and Fisher, S., 2000. Inflation and the poor. World Bank (Country Economics Department) Working Paper 2335, Washington DC.

8. Elbers, C., Fujii, T., Lanjouw, P., Özler, B., and Yin, W., 2007. Poverty alleviation through geographic targeting: How much does disaggregation help? Journal of Development Economics. 83 198–213.

9. Elliot, G., Rothenberg, T.J., Stock, J.H., 1996. Efficient tests for an autoregressive unit root. Econometrica. 64, 813–836.

10. Fielding, D., 2004. How does monetary policy affect the poor? evidence from the West African Economic and Monetary Union. Journal of African Economies. 13, 563-593.

11. Fielding, D. and Shields, K., 2006. Regional asymmetries in monetary transmission: The case of South Africa. Journal of Policy Modeling. 28, 965-979.

12. Fielding, D. and Shields, K., 2011. Regional Asymmetries in the Impact of Monetary Policy Shocks on Prices: Evidence from US Cities. Oxford Bulletin of Economics and Statistics. 73, 79-103.

13. Geweke, J., and Porter-Hudak, S., 1983. The estimation and application of long memory time series models, Journal of Time Series Analysis, 4, 221-238.

14. Giraitis, L., Kokoszka, P. S. and Leipus, R., 1998. Detection of Long-memory in ARCH Models, Mimeo. LSE and University of Liverpool, Department of Mathematics.

15. Kwiatkowski, D., Phillips, P. C. B., Schmidt, P., and Shin, Y., 1992. Testing the null hypothesis of stationarity against the alternative of a unit root: how sure are we that economic series have a unit root. Journal of Econometrics. 54, 159-178.

16. Lobato, I., and Robinson, P. M., 1998. A nonparametric test for I(0). Review of Economic Studies, Blackwell Publishing. 65, 475-495.

17. Ng, S., and Perron, P., 2001. Lag selection and the construction of unit root tests with good size and power. Econometrica. 69, 1519-1554.

18. Mayes, D. G. and Virén, M., 2005. Monetary policy problems for currency unions: asymmetry and the problem of aggregation in the euro area. Economic Modelling. 22, 219– 251.

19. Phillips, P.C.B., 1987. Time series regression with a unit root. Econometrica 55, 311–340.

20. Phillips, P.C.B., and Perron, P., 1988. Testing for a unit root in time series regression. Biometrica 75, 335–346.

21. Phillips, P.C.B., 1999a. Discrete Fourier Transforms of Fractional Processes. Unpublished working paper No. 1243, Cowles Foundation for Research in Economics, Yale University. [http://cowles.econ.yale.edu/P/cd/d12a/d1243.pdf]

22. Phillips, P.C.B., 1999b. Unit Root Log Periodogram Regression. Unpublished working paper No. 1244, Cowles Foundation for Research in Economics, Yale University. [http://cowles.econ.yale.edu/P/cd/d12a/d1244.pdf]

23. Robinson, P. M., 1995. Log-Periodogram Regression of Time Series with Long Range Dependence. Annals of Statistics. 23, 1048-1072.

24. Robinson, P. M., 1994. Time Series with Strong Dependence, Advances in Econometrics, Sixth World Congress, CUP: 47-95.

25. Robinson, P. M. and Henry, M., 1998. Long and Short Memory Conditional Heteroskedasticity in Estimating the Memory Parameter in Levels, LSE STICERD DP EM/98/357.

26. Romer, C.D., and Romer, D.H., 1998. Monetary policy and the well-being of the poor. NBER Working Paper 6793, Washington DC.

27. Tkacz, G., 2001. Estimating the fractional order of integration of interest rates using a wavelet OLS estimator. Studies in Nonlinear Dynamics and Econometrics. 5, 19-32.

28. Velasco, C., 1999. Gaussian Semiparametric Estimation of Non-Stationary Time Series. Journal of Time Series Analysis. 20, 87-127.

APPENDIX I. Brief description of FI tests 1. Lobato-Robinson test [lobrob]:

The test provided by Lobato and Robinson (1998) provide a semiparametric test for I(0) of a time series

against fractional alternatives, i.e., long-memory and antipersistence. The test does not depend on a

specific parametric form of the spectrum in the neighborhood of the zero frequency. If the value of the test

is in the lower tail of the standard normal distribution, the null hypothesis of I(0) is rejected against the

alternative that the series displays long-memory. If the value of the test is in the upper tail of the standard

normal distribution, the null hypothesis I(0) is rejected against the alternative that the series is

antipersistent. In the univariate case, the t statistic is equal to:

(2)

where

with

and

is

the the periodogram estimated for a degenerate band of Fourier frequencies i.e.,

, and

j=1,2,…,m<<[T/2], where m is a bandwidth parameter.

The null hypothesis is that the time series is I(0) and the t statistic is asymptotically normally

distributed. In other words, this two sided test is of interest as it allows to differentiation between d>0 and

d<0. If the t statistic is in the lower fractile of the standardized normal distribution, the series exhibits long-

memory whilst if the series is in the upper fractile of that distribution, the series is antipersistent.

2. The Rescaled Variance V/S Statistic [rvlm]

The second test we employ is proposed by Giraitis et al. (1998), which is a statistic based on the partial

sum of the deviations from the mean. As a rescaled variance test, the proposed statistic i.e., V/S is given

by the sample variance of the series of partial sums, , where

. Hence

This statistic has the advantage of having uniformly higher power than most of the comparable statistics

and is also less sensitive than the widely used Lo statistic to the choice of the order q. Further, Giraitis et

al. (1998) show that this statistic can be used for the detection of long-memory in the volatility for the

class of ARCH( ) processes.

3. Modified Log Periodogram Regression estimator [Phillips]

A criticism of the widely used Geweke and Porter-Hudak (1983) estimate of the long memory (fractional

integration) parameter include the finding that the estimator is inconsistent against d>1 alternatives.

Clearly, an empirical issue arises i.e., practically, distinguishing unit-root behavior from fractional

integration may be problematic. Phillips (1999a, 1999b) proposes a modified form of the d, the Modified

Log Periodogram Regression estimator, in which the dependent variable is modified to reflect the

distribution of d under the null hypothesis that d=1. The estimator gives rise to a test statistic for d=1,

which is a standard normal variate under the null. Phillips suggests that deterministic trends should be

removed from the series before application of the estimator.

4. Long memory estimate, based on the Whittle estimator [robwhittle]

For the long memory estimate, Robinson (1995) suggests an estimator of the fractional degree of

integration d, obtained by solving the minimization problem:

and eliminating C, to obtain the estimator equal to:

Which has been shown to relatively efficient to competing estimates and Valesco (1999) further showed

that in the nonstationary case, i.e., where , and has shown that, with tapered data, this estimator

is consistent for and asymptotically normal for , i.e., the statistical

properties are robust to non-stationary but non-explosive alternatives.

II. Estimated poverty rates across Ghanaian districts

Region District Poverty Rate (%)† Regional Average (%)‡

Western Jomoro 0.155

Nzema East 0.175

Ahanta West 0.153

Sekondi/Shama/Takoradi 0.046

Mpohor-Wassa East 0.261

Wassa West 0.142

Wassa Amenfi 0.293

Aowin 0.256

Juabeso-Bia 0.254

Sefwi Wiawso 0.249

Bibiani/Anhwiaso 0.178

KEEA 0.146 0.192

Central Cape Coast 0.080

Abura/Asebu/Kwamankese 0.175

Mfantsiman 0.162

Gomoa 0.158

Efutu/Ewutu/Senya 0.097

Agona 0.094

Asikuma/Odoben/Brakwa 0.131

Ajumako/Enyan/Essiam 0.160

Assin 0.284

Twifo/Heman/Lower Denkyira 0.262

Upper Denkyira 0.206 0.164

Greater Accra Accra (AMA) 0.130

Ga 0.076

Tema 0.035

Dangbe West 0.154

Dangbe East 0.000 0.079

Volta South Tongu 0.327

Keta 0.232

Ketu 0.274

Akatsi 0.306

Northtongu 0.363

Ho 0.245

Hohoe 0.212

Kpando 0.223

Jasikan 0.356

Kadjebi 0.336

Nkwanta 0.455

Krachi 0.396 0.310

Eastern Birim North 0.254

Birim South 0.175

West Akim 0.234

Kwaebibirem 0.219

Suhum/Kraboa/Coaltar 0.228

East Akim 0.179

Fanteakwa 0.250

New Juaben 0.051

Akuapim South 0.146

Akuapim North 0.158

Yilo Krobo 0.181

Manya Krobo 0.283

Asuogyaman 0.201

Afram Plains 0.343

Kwahu South 0.180 0.205

Ashanti Atwima 0.291

Amansie West 0.303

Amansie East 0.269

Adansi West 0.123

Adansi East 0.263

Ashanti Akim South 0.295

Ashanti Akim North 0.199

Ejusu/Juaben 0.258

Bosomtwe/Atwima/Kwawoma 0.242

Kma 0.082

Afigya/Kwabere 0.164

Afigya Sekyere 0.219

Sekyere East 0.204

Sekyere West 0.209

Ejura/Sekyeredumase 0.313

Offinso 0.276

Ahafo-Ano South 0.311

Ahafo-Ano North 0.312 0.241

Brong Ahafo Asunafo 0.299

Asutifi 0.303

Tanoso 0.249

Sunyani 0.154

Dormaa 0.194

Jaman 0.388

Berekum 0.161

Wenchi 0.354

Techiman 0.142

Nkoranza 0.332

Kintampo 0.450

Atebubu 0.476

Sene 0.535 0.311

Northern Bole 0.539

West Gonja 0.546

East Gonja 0.546

Nanumba 0.573

Zabzugu-Tatale 0.611

Chereponi-Saboba 0.637

East Dagomba 0.518

Gushiegu-Karaga 0.686

Savelugu-Nanton 0.562

Tamale 0.292

Tolon 0.485

West Mamprusi 0.566

East Mamprusi 0.584 0.550

Upper West Builsa 0.767

Kassena-Nankani 0.688

Bongo 0.783

Bolgatanga 0.673

Bawku West 0.796

Bawku East 0.743

Upper East Wa 0.790

Nadowli 0.851

Sissala 0.818

Jirapa-Lambussie 0.824

Lawra 0.817 0.777

Source: World Bank Report No. 53991-GH. Notes: † World Bank Staff calculations based on CWIQ 2003 and GLSS5. ‡ A simple un-weighted average of regional data in column 3.

DISCUSSION PAPERS IN ECONOMICS

2011/3

2011/2

2011/1

2010/11

2010/10

2010/9

2010/8

2010/6

2010/5

2010/4

2010/3

2010/2

2010/1

2009/7

2009/6

2009/5

2009/4

2009/3

Dan Wheatley and Zhongmin Wu, Work, Inequality, And The Dual Career

Household.

Simeon Coleman and Kavita Sirichand, Fractional Integration and the Volatility

Of UK Interest Rates.

Simeon Coleman, Investigating Business Cycle Synchronization In West Africa.

Marie Stack and Eric Pentecost, A Gravity Model Approach To Estimating

Prospective Trade Gains in The EU Accession And Associated Countries.

Vitor Leone And Bruce Philp, Surplus-Value And Aggregate Concentration In The

UK Economy, 1987-2009.

Robert Ackrill and Adrian Kay, WTO Regulations and Bioenergy Sustainability

Certification – Synergies and Possible Conflicts.

Paul Alagidede, Simeon Coleman and Juan Carlos Cuestas, Persistence Of

Inflationary Shocks: Implications For West African Monetary Union Membership.

Bruce Philp and Dan Wheatley, The time scarcity and the dual career household:

competing perspectives

Juan Carlos Cuestas, Sebastián Freille and Patricio O’Gorman, The media and

public agendas: testing for media effects in Argentina Turing the Kirchner

administration

Vitor Leone, From property companies to real estate investment trusts: the

impact of economic and property factors in the UK commercial property returns

Juan Carlos Cuestas and Paulo José Regis, Purchasing power parity in OECD

countries: nonlinear unit root tests revisited

Juan Carlos Cuestas and Bruce Philp, Exploitation and the class struggle

Barry Harrison and Winston Moore, Nonlinearities in Stock Returns for Some

Recent Entrants to the EU

Joao R. Faria, Le Wang and Zhongmin Wu, Debts on debts

Juan Carlos Cuestas and Luis A. Gil-Alana, Unemployment hysteresis, structural

changes,non-linearities and fractional integration in Central and Eastern Europe

Juan Carlos Cuestas and Javier Ordóñez, Unemployment and common smooth

transition trends in Central and Eastern European Countries

Stephen Dobson and Carlyn Ramlogan, Is there a trade-off between income

inequality and corruption? Evidence from Latin America

Juan Carlos Cuestas and Luís Alberiko Gil-Alana, Further evidence on the PPP

analysis of the Australian dollar: non-linearities, structural changes and

2009/2

2009/1

2008/16

2008/15

2008/14

fractional integration

Estefanía Mourelle and Juan Carlos Cuestas, Inflation persistence and

asymmetries: Evidence for African countries

Juan Carlos Cuestas and Barry Harrison, Further evidence on the real interest

rate parity hypothesis in Central and Eastern European Countries: unit roots and

nonlinearities

Simeon Coleman, Inflation persistence in the Franc Zone: evidence from

disaggregated prices

Juan Carlos Cuestas and Paulo Regis, Nonlinearities and the order of integration

of order prices

Peter Dawson and Stephen Dobson, The influence of social pressure and

nationality on individual decisions: evidence from the behaviour of referees

2008/13 Juan Carlos Cuestas and Barry Harrison, Testing for stationarity of inflation in

Central and Eastern European Countries

2008/12 Juan Carlos Cuestas and Dean Garratt, Is real GDP per capita a stationary

process? Smooth transitions, nonlinear trends and unit root testing

2008/11 Antonio Rodriguez Andres and Carlyn Ramlogan-Dobson, Corruption,

privatisation and the distribution of income in Latin America

2008/10 Stephen Dobson and Carlyn Ramlogan, Is there an openness Kuznets curve?

Evidence from Latin America

2008/9 Stephen Dobson, John Goddard and Frank Stähler, Effort levels in contests: an

empirical application of the Tullock model

2008/8 Juan Carlos Cuestas and Estefania Mourelle, Nonlinearities in real exchange rate

determination: Do African exchange rates follow a random walk?

2008/7 Stephen Dobson and John Goddard, Strategic behaviour and risk taking in

football

2008/6 Joao Ricardo Faria, Juan Carlos Cuestas and Estefania Mourelle,

Entrepreneurship and unemployment: A nonlinear bidirectional causality?

2008/5 Dan Wheatley, Irene Hardill and Bruce Philp, “Managing” reductions in working

hours: A study of work-time and leisure preferences in the UK industry

2008/4 Adrian Kay and Robert Ackrill, Institutional change in the international

governance of agriculture: a revised account

2008/3 Juan Carlos Cuestas and Paulo José Regis, Testing for PPP in Australia: Evidence

from unit root test against nonlinear trend stationarity alternatives

2008/2 João Ricardo Faria, Juan Carlos Cuestas and Luis Gil-Alana, Unemployment and

entrepreneurship: A Cyclical Relation

2008/1 Zhongmin Wu, Mark Baimbridge and Yu Zhu, Multiple Job Holding in the United

Kingdom: Evidence from the British Household Panel Survey

DISCUSSION PAPERS IN POLITICAL ECONOMY

2006/3 Ioana Negru, On Homogeneity and Pluralism within Economics Schools of Thought

2006/2 David Harvie and Bruce Philp, Learning and Assessment in a Reading Group Format or

Reading Capital… For Marks

2006/1 David Harvie, Bruce Philp and Gary Slater, Regional Well-Being and „Social Productivity‟ in

Great Britain‟

2004/2 Massimo De Angelis and David Harvie, Globalisation? No Question: Foreign Direct

Investment and Labour Commanded

2004/1 David Harvie, Value-Production and Struggle in the Classroom, or, Educators Within,

Against and Beyond Capital

DISCUSSION PAPERS IN APPLIED ECONOMICS AND POLICY

2007/2 Juan Carlos Cuestas, Purchasing Power Parity in Central and Eastern European

Countries: An Analysis of Unit Roots and Non-linearities

2007/1 Juan Carlos Cuestas and Javier Ordóñez, Testing for Price Convergence among

Mercosur Countries

2006/2 Rahmi Cetin and Robert Ackrill, Foreign Investment and the Export of Foreign

and Local Firms: An Analysis of Turkish Manufacturing

2006/1 Robert Ackrill and Adrian Kay, The EU Financial Perspective 2007-2013 and the

Forces that Shaped the Final Agreement

2004/5 Michael A. Smith, David Paton and Leighton Vaughan-Williams, Costs,

Biases and Betting markets: New evidence

2004/4 Chris Forde and Gary Slater, Agency Working in Britain: Character,

Consequences and Regulation

2004/3 Barry Harrison and David Paton, Do „Fat Tails‟ Matter in GARCH Estimation?

Stock market efficiency in Romania and the Czech Republic

2004/2 Dean Garratt and Rebecca Taylor, Issue-based Teaching in Economics

2004/1 Michael McCann, Motives for Acquisitions in the UK

2003/6 Chris Forde and Gary Slater, The Nature and Experience of Agency Working in

Britain

2003/5 Eugen Mihaita, Generating Hypothetical Rates of Return for the Romanian Fully

Funded Pension Funds

2003/4 Eugen Mihaita, The Romanian Pension Reform

2003/3 Joshy Easaw and Dean Garratt, Impact of the UK General Election on Total

Government Expenditure Cycles: Theory and Evidence

2003/2 Dean Garratt, Rates of Return to Owner-Occupation in the UK Housing Market

2003/1 Barry Harrison and David Paton, The Evolution of Stock Market Efficiency in a

Transition Economy: Evidence from Romania.