Howoptimal is Ghana’s single-digit inflation targeting? An assessment of monetary policy effectiveness in Ghana Richard Amoatey, Richard K. Ayisi and Eric Osei-Assibey Department of Economics, University of Ghana, Accra, Ghana Abstract Purpose – The purpose of this study is twofold. First, to estimate an optimal inflation rate for Ghana and second, to investigate factors that account for the differences between observed and target inflation. Design/methodology/approach – The paper explored the questions within two econometric frameworks, the Autoregressive Distributed Lag (ARDL) and Threshold RegressionModels using data spanning the period 1965–2019. Findings – The study estimated a range of 5–7% optimal inflation for Ghana. While this confirms the single-digit inflation targeting by the Bank of Ghana, the range is lower than the central bank’s band of 6–10%. The combined behaviours of the central bank, banks and external outlook influence inflation target misses. Practical implications – The study urges the central bank to continue pursuing its single-digit inflation targeting. However, it implies that there is still room for the Bank to further lower the current inflation band to achieve an optimal outcome on growth and welfare. Again, the Bank should commit to increased transparency and accountability to enhance its credibility in attaining the targeted inflation. Originality/value – The study is one of the first attempts in Africa in Ghana to estimate an optimal inflation target and investigate the underlying factors for deviation from the targets. Keywords Optimal inflation rate, Monetary policy, Ghana Paper type Research paper 1. Background The optimal level of inflation to target for an economy has engaged the attention of many academics and policymakers (Bhattarai, 2014), particularly because inflation brings forth both positive and negative economic outcomes that require a balance to maximize economic gains. Theoretically, an effective monetary policy aimed at controlling inflation requires in principle that, the gap between actual inflation and target rates should be close (possibly zero) and consistent with growth (Svensson, 2010). On that account, the key question for a developing country like Ghana is the level of inflation that is ideal and can promote economic welfare and stimulate growth. Since adopting the Inflation Targeting (IT) framework in 2002, Ghana’s central bank has failed to achieve its target rate of 8% [1], though it has succeeded in stabilizing prices after years of high inflation in the 1980 and 1990s. Available data reveal that apart from the period December 2010 through to 2011 and 2019 when the country attained single-digit inflation, the central bank has not been able to attain its inflation target. The central bank’s inability to consistently achieve its inflation target is supported by available data. Figure 1 presents the inflation gap for Ghana under the IT regime [2]. It is evident from Figure 1 that the inflation gap has remained largely positive within the entire IT period. This evidence suggests that monetary policy has been ineffective in achieving the desired level of inflation. Why monetary policy is unable to achieve its desired outcome requires investigating. Two important questions that the paper investigated are: Optimal inflation rate for Ghana The current issue and full text archive of this journal is available on Emerald Insight at: https://www.emerald.com/insight/2040-0705.htm Received 29 March 2023 Revised 13 September 2023 Accepted 4 October 2023 African Journal of Economic and Management Studies © Emerald Publishing Limited 2040-0705 DOI 10.1108/AJEMS-03-2023-0119 https://doi.org/10.1108/AJEMS-03-2023-0119 (1) Is the deviation resulting from a sub-optimal inflation target rate? Or (2) The deviation results from ineffective monetary policy. The debate on optimal inflation is a very important discussion in the conduct of monetary policy. This is against the backdrop that inflation is a necessary evil, which has the parallel effect of serving as an incentive to production and negatively impacting general welfare. In effect, central banks would target a non-zero inflation rate by forming a balance between the two opposing effects (Adam et al., 2022). Ghana practising inflation targeting for about two decades has constantly missed inflation targets [3]. Are inflation target misses due to sub-optimal inflation targets or the ineffectiveness of monetary policy? The study explored this objective by estimating an optimal inflation rate for Ghana using two different econometric methods – the threshold and Autoregressive Distributed Lag (ARDL) modelling approaches. This study estimated an optimal inflation rate of 5–7% which agrees with the inflation target rate of the Bank of Ghana (BOG). The rest of the paper is organized as follows: Section 2 provides a brief review of the literature, Section 3 espouses the methodology adopted, Section 4 presents the results and discussions and Section 5 concludes the study with policy recommendations. 2. Literature review In the literature, many theoretical explanations have been proposed for inflation (Totonchi, 2011). Notwithstanding, there is a consensus that two foundational theories of inflation are the demand-pull and cost-push theories (Dastgerdi, 2020; Nigusse et al., 2019; Shaikh et al., 2022). However, the detailed factors that drive inflation are enshrined in theories such as the Neoclassical Monetarism theory, New Keynesian theory, Post-Keynesian theory, Modern Monetary Theory, Political Economy Theory, Sociological theory and Structuralist Theory (Barta et al., 2023). According to Abuselidze (2019), inflation in developing countries should be analysed from perspectives different from that of a developed country. As Dastgerdi (2020) points out, inflation in industrial countries is largely supported by demand-pull and cost-push theories but these theories do not completely cover the reasons for inflation in developing countries. Lapavitsas (2022) has emphasized that structural inflation theory espouses the causes of 0 5 10 15 20 25 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Ghana Infla on Rate Infla on Gap Source(s): Authors’ compilation with data from the Bank of Ghana Database Figure 1. Graph of inflation and inflation gap AJEMS inflation in developing countries. The structuralist theory considers the essential characteristics of inflation by identifying its chronic and recurrent nature. That notwithstanding, the appropriate theory that underlines inflation dynamics in any country is largely an empirical issue. Irrespective of the source of inflation in any economy, the most optimal rate requires to be determined. To that extent, countries have adopted Inflation targeting, either explicitly or implicitly, as a policy framework to target inflation and anchor inflation expectations. In inflation targeting, the policymaker commits to achieving an inflation target. According to Bernanke et al. (1999), “inflation targeting is a framework for monetary policy characterised by the public announcement of official quantitative targets (or target ranges) for the inflation rate over one or more time horizons, and by the explicit acknowledgement that low stable inflation is monetary policy’s primary long-run goal. Among other important features of inflation targeting are vigorous efforts to communicate with the public about the plans and objectives of the monetary authorities, and, in many cases, mechanisms that strengthen the central bank’s accountability for attaining those objectives” (p. 4). It is generally agreed inflation targeting requires a central bank can conduct monetary policy with some degree of independence and a sole focus on inflation, a key requirement for achieving its policy goal. Nevertheless, inflation-targeting monetary authorities frequently fail to achieve their inflation target. Many explanations are provided. Ehrmann (2021) claims, for example, that point objectives are more likely to be missed than range targets, although Kahn and Parrish (2020) note exceptional conditions such as shocks in the price of oil. According to Albagli and Schmidt-Hebbel (2003), inflation targets would be frequently missed if institutions were weak, the central bank was not independent and risk premiums were large. The necessity of an empirical effort to determine the ideal inflation cannot be overstated if target misses could be attributed to non-optimal targets. For emergingmarkets and developing countries, studies on optimal inflation largely focus on the associations between inflation and economic growth, per capita GDP or some macroeconomic indicator exploring the nonlinearities to determine an ideal inflation rate. Various studies in developing and emerging market economies have followed such an approach. Dammak and Helali (2017) studied the inflation–economic growth nexus in the case of Tunisia for the 1993–2012 period employing a threshold regression model and found that an inflation rate lower than 3.48% fosters economic growth. Khan and Hanif (2020) consider the issue of optimal inflation from a different perspective by looking at the relationship between institutional quality on the one hand, and inflation and economic growth on the other hand for some countries. They observe the existence of a minimum level of institutional quality which has a significant effect on inflation and economic growth. Adaramola and Dada (2020) also sought to find the level of inflation that was not detrimental to economic growth by considering the effect of inflation on economic growth in Nigeria. Abdulqadir et al. (2020) investigated the optimal inflation targets in an appropriate exchange rate policy framework in 15 major oil exporting countries in Sub-Saharan Africa. They found an optimal inflation rate of 14.47%. Within the Ghanaian context, there are some studies worth mentioning. A seminal work was done by Frimpong andOteng-Abayie (2010) analysing the threshold effect of inflation on economic growth in Ghana for the period 1960–2008 using a threshold regression model. The data set employed includes the growth rate of GDP, the growth rate of the aggregate labour force, the growth rate of money supply and the growth rate of terms of trade (TOT). Their Optimal inflation rate for Ghana result was finding an inflation threshold level of 11% above which inflation starts to significantly hurt economic growth. In another study, Musah et al. (2019) examined the threshold effects of inflation on economic growth in Ghana and also observed that inflation negatively affects overall growth but estimated a threshold inflation rate of 21.57%. For the Ghanaian setting, two salient issues observed are the use of a single methodology in estimation and varying estimates and non-consistency in results. The contribution of this present paper is by employing two different methodologies with a more recent dataset. 3. Methodology 3.1 Variable selection and data source The variables were selected considering the inflation-growth nexus with a special focus on Per Capita GDP growth. Neoclassical growth theory considers the capital stock, labour force and technological progress as the main determinants of growth whereas endogenous growth theory considers human capital, knowledge and new technologies (Piętak, 2014). Following Fischer (1993), Barro (1995), Tien (2021) and Siddik (2023) and principally Mohsin Khan and Senhadji (2001) we select variables utilizing the neoclassical growth theory underscored by the growth characteristics of Ghana and existing studies. Again, we consider data constraints and adjust accordingly. The variables were selected from the World Bank’s Development indicators (WDI) TheWorld Bank (2019) and the descriptions are discussed below: (1) Growth rate of GDP Per Capita (PCI): dependent variable employed in the estimation. It is computed as the percentage change in real GDP divided by population. (2) Inflation Rate (INFRATE): computed as the annual percentage change in consumer price index (CPI) with 2010 as the base year as the main explanatory and threshold indicator. Growth is expected to be lower at high rates of inflation over long-term economic growth but unaffected or even positive when inflation is lower (Mandeya and Ho, 2021). (3) Broad Money as a Percentage of GDP (M2) or Money Supply: used as one of the explanatory variables to represent an index of financial depth in a country at a particular point in time, as indicated by King and Levine (1993). The variable M2 as a proportion of GDP has been used as a proxy for financial development. Its computation is based on the annual percentage change in money and quasi-money as a percentage of real GDP. The expected sign of the coefficient for the broad money supply is positive. (4) Gross Capital Formation (GCF). Gross Capital formation was formerly called gross domestic investment. It is computed as the annual percentage change in gross capital formation as a percentage of GDP. Given the relevance of capital formation in growth modelling, this variable is included as a proxy. Gross capital formation is expected to be positive in this model (5) General government final consumption expenditure (GOVT_EXP): General government final consumption expenditure was previously known as general government consumption. It includes all government current expenditures for purchases of goods and services (including compensation of employees). It also includes most expenditures on national defence and security but excludes government military expenditures that are part of government capital formation. This variable is expected to be positive because of the enormous role of the government in the economy of Ghana. AJEMS 3.2 Econometric approach The main goal of the study is to estimate the optimal inflation rate for Ghana. To this end, we employed the well-known Autoregressive Distributed Lag (ARDL) model as proposed by Pesaran et al. (2001) and the Threshold Regression Model as proposed by Hansen (2000). The generalized form of the study model is represented as follows: PCI z}|{ GrowthrateofPerCapitaGDP ¼ f INFRATE zfflfflfflfflfflffl}|fflfflfflfflfflffl{ ;Inflation GCF ;GOVTexp;M2 zfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{ ControlMacroeconomicVariables ! (1) The econometric equation is specified as a growth model augmented with inflation as PCIt ¼ β0 þ β1INFRATEt þ β2GCFt þ β3GOVTexpt þ β4M2þ μt (2) Where variable source and expected signs are as previously defined in section 3.1 and βs represent the marginal effects and μt is the white noise error term. 3.2.1 Autoregressive distributed lag model.ARDL modelling technique, an Ordinary Least Square (OLS) based model applies to stationary and non-stationary time series. Being a least squares-based model implies it can be estimated as least squares regressions using lags of both dependent and independent variables. Equation (1) is specified in an ARDL framework as follows: ΔPCIGRt ¼ α0 þ Xn i¼1 βiΔPCIGRt−1 þ Xn i¼1 ∙γiΔINFRATEt−1þ þ Xn i¼1 ∙γiΔM2t−1 þ Xn i¼1 ∙γiΔGCFt−1 þ DINFRATEþ w1PCIt−1 þ w2INFRATEt−1 þ w3GCFt−1 þ w4GOVTexpt−1 þ w5M2t−1 þ μt: (3) ΔPCIGRt ¼ α0 þ Xn i¼1 βiΔPCIGRt−1 þ Xn i¼1 ∙γiΔINFRATEt−1þ þ Xn i¼1 ∙γiΔM2t−1 þ Xn i¼1 ∙γiΔGCFt−1 þ DINFRATE þ∅ECMt−1 þ μt (4) where equation (2) is the ARDL specification and equation (4) is the Error Correction Model (ECM) specification. D_INFRATE is a dummy variable generated for selected levels of inflation from 5 to 30%. Δ represents the first difference, and ∅ coefficients of parameters. Optimal lag orders which are possibly different across variables were obtained by minimizing the Akaike information criterion (AIC). The bounds procedure for testing the existence of a long-run relationship is based on the Error Correction (EC) representation of the ARDLmodel. Indeed, the dynamic error correction model (ECM) as shown in equation (4) can be derived fromARDL through a simple linear transformation. The ECM term is typically the cointegrating relationship. Furthermore, an ARDL model is diagnosed by testing for serial correlation in the residuals, checking the stability as well and performing a bounds test. 3.2.2 Estimation procedure of the ARDLmodel. Examining the data from equation (3), the ARDL bound test for cointegration is applied, which is also a Wald test. The null hypothesis of theWald test suggests cointegration exists whereas the alternative implies the existence of Optimal inflation rate for Ghana cointegration. If the computed F-statistic falls below the lower bound we would conclude that the variables, are I (0), so no cointegration is possible, by definition. If the F-statistic exceeds the upper bound, we conclude that we have cointegration. Finally, if the F-statistic falls between the bounds, the test is inconclusive Pesaran et al. (2001). 3.2.3 Threshold regression. This article follows the work of Khan and Senhadji (2001) and adopts the threshold regression of Hansen (1999, 2000) in addressing the non-linearity perceived in the growth and inflation relationship. This serves as a robustness check to the estimate from the ARDL. The threshold model takes the form PCIt ¼ ( β10 þ β11INFRATE þ β12GCF þ β13GOVTexp þ β14M2þ εt β20 þ β21INFRATE þ β22GCF þ β23GOVTexp þ β24M2þ εt if INFRATE ≤ γ INFRATE > γ (5) with PCI denoting per capita GDP growth rate as a regime-dependent variable, INFRATE representing inflation and the threshold variable Inflation (INFRATE), the threshold variable is used to split the sample into the regimes and the control variables as defined previously. Where I (•) is the indicator function, and the regimes are separated into the lower and beyond the optimal threshold parameter γwhile β1 and β2 are the regression slope under each regime and the error term u iidð0; σ2Þ. 3.2.3.1 Test for threshold effect. The test for the existence of threshold effect involves a hypothesis test that the equation can be divided into regimes depending on the value of the threshold variable. Upon the detection of at least one threshold value, the relationship between the dependent variable and threshold can be deemed as nonlinear. In the testing [procedure, the null hypothesis of linearity Ho: β1 ¼ squaredis tested against the alternative that the coefficient varies across the regimes. This test employs a likelihood ratio test where; F1 ¼ � S0 � S1bγ�bσ2 (6) where S0∧S1bγ are sums of squared errors for both the null and alternative hypotheses respectively. Since the F1 has a non-standard distribution Hansen (1996) shows that theWald LM static for each possible value of λ ith inference can be computed based on the bootstrap calculation of a Wald or LM statistic for each possible value of λ and subsequently basing inferences on the supremum of the Wald or Lm statistic across all possible λs: Furthermore, the limiting distribution of their supremum statistic is non-standard and depends on numerous model-specific nuisance parameters. Obtaining an estimate of λwhich is based on the minimisation of the residual sum of squares computed across all possible values of λ, as estimates of the B1(λ and B2(λ yt ¼ xtβ þ ztδ1Ið−∞ < wt ≤ γÞ þ ztδ2Iðγ < wt < ∞Þ þ et (7) For a threshold regressionwith two regions identified by a threshold value ‘λ’, the threshold of the estimator is given as�bλ� ¼ argminγeΓST1 ðγÞWherereΓ ¼ ð−∞;∞Þ (8) the λ is obtained based on theminimized residual sum of squares computed across all possible values of λ; estimates of the slope parameters followed trivially as bβ1ðbλÞ and bβ2ðbλÞ. AJEMS The impact of inflation considering the other control variables on per capita GDP growth will bebβ1 when the inflation rate is less than or equal to v andbβ2 when inflation is greater than the threshold value. If the null hypothesis is not rejected then the threshold model would transform into a linear model implying that a threshold effect does not exist. 3.2.3.2 Asymptotic distribution of threshold estimate. Understanding the characteristics of parameter estimates andhypothesis testing in thismodelling framework depends heavily on the asymptotic distribution of threshold regression. The asymptotic distribution of the threshold estimate is testedwith the null hypothesis;H0 ¼ γ ¼ γ0 using the likelihood ratio statistic test of LR1ðγÞ ¼ 24 � S1 � S1bγ�bσ2 35 (9) The asymptotic confidence interval is known as cðβÞ ¼ −2log � 1� ffiffiffiffiffiffiffiffiffiffiffi 1� β p � (10) Such that for a given of β, the null hypothesis of γ ¼ γ0 is rejected if LR1ðγÞ exceeds cðβÞ. The model can be modified if a double threshold exists. In the estimation, we compute the critical values at the asymptotic distribution of the LR, at the 1%, 5 and 10% significance levels and find. 4. Results and discussions We began by presenting the characteristics of the variables in Table 1 showing the summary statistics. The statistics reveal that between 1965 and 2019 the average inflation rate (INFRATE) was approximately 27.16%, with the highest being 122.87% and the lowest – at 8.42%. Within the same sample period, the average annual per-capita GDP growth (PCI) is about 1.02%, with a maximum of 11.28% and a minimum of �14.45%. Average annual Per capita GDP growth (PCI) is far lower than the average inflation rate. Further, a correlation analysis suggests that the correlation between the variables is low. Table 2 presents the correlation results, and it shows that the lowest correlation among the variables is 0.01 which is between per capita GDP growth (PCI) and final government Variable Mean SD Min Max PCI 1.130535 4.40386 �14.50853 11.31545 INFRATE 27.16475 27.10922 �8.422486 122.8745 GCF 16.28002 7.555804 3.37764 30.04927 GOVT_EXP 10.98494 2.313663 5.86129 16.76471 M2 22.78945 5.689418 11.30499 34.10823 Source(s): Authors’ estimation with data from WDI, 2019 PCI INFRATE GCF M2 GOVT_EXP PCI 1.0000 INFRATE �0.3643 1.0000 GCF 0.3483 �0.4000 1.0000 M2 0.2229 �0.2522 0.5424 1.0000 GOVT_EXP 0.0119 �0.2173 0.0991 0.1551 1.0000 Source(s): Authors’ estimation with data from WDI, 2019 Table 1. Summary statistics Table 2. Correlation matrix Optimal inflation rate for Ghana expenditure (GOVT_EXP). The highest correlation among any of the variables is 0.54 between gross capital formation (GCF) and broad money (M2). To ascertain the applicability of the ARDL model and to prevent any spurious analysis, we subjected the variables to a unit-root test using the Augmented Dickey-Fuller (ADF) test. The ADF test results as shown in Table 3 indicates a mixture of I (0) and I (1) variables, appropriate for ARDL modelling. 4.1 Results from the auto-regressive distributed lag (ARDL) model The study fits the ARDL model as specified in equation (3) using the Akaike Information Criteria to select the optimal lag length and incorporate a dummy variable INFRATE � k. INFRATE is the observed inflation rate while k is an arbitrarily selected inflation rate from 5% to 30%. The dummy variable D_t is generated as INFRATE– k. When the dummy variable D t is 1 then inflation is greater than k% and D t is 0 when the inflation rate is less than or equal to k%. The estimation involved running equation (3) twenty-six (26) times with the dummy inserted consecutively. The dummy variables will be used separately in the equation such that it will be the only item that would change in the estimation and the optimal value of k(observed/actual inflation) is obtained from the equation with the lowest residual sum of squares or maximum R2. The significance of the dummy will determine the overall optimality of those inflation rates. If D t is not significant then itmeans inflation rates lower than kare not optimal, while a range of significant inflation rates may also be determined. The result of the estimation of the ARDL model is presented in Table 4. Accordingly, k is the observed inflation rate put into equation (2). k values 5 to 9 declared single-digit inflation rate and 10 to 30% as double-digit respectively, then the significance of the coefficient of D t ðINFRATE − kÞ where k represents inflation from 5–30% would establish the optimality or otherwise of the single-digit or double-digit inflation rate respectively. From the result in Table 4, it is seen that inflation rates(k) of 19%, 20%, 21% and 22% have the highest R2 and are also significant at the 5% significance level. Furthermore, the first ten values of the selected inflation rates have four of them with double digits (i.e. 19%, Variable Model Level t- statistic 1st Difference t-statistic Conclusion PCI None �4.5023*** �5.730549*** I (0) Intercept �5.399*** �12.62013*** Intercept and trend �5.5483*** �5.728079*** INFRATE None �1.593721 �12.74817*** I (1) Intercept �2.536288 �12.74817*** Intercept and trend �4.516948 �12.51140*** GCF None �0.570123 �6.378851*** I (1) Intercept �1.703443 �6.328836*** Intercept and trend �2.958621 �6.224476*** M2 None �0.153069 �7.064946*** I (1) Intercept �1.741525 �7.003379*** Intercept and trend �2.071525 �6.933596*** GOVT_EXP None �1.025046 �7.783881*** I (1) Intercept �3.210004** �7.722896*** Intercept and trend �3.410158 �7.642048 Source(s): Authors’ estimation with data from WDI, 2019 Table 3. Stationarity test results AJEMS 20%, 21% and 22%) while three of them are single digits (i.e. 5%, 6% and 7%). This implies that there are two sets of significant inflation rates; on one hand, single-digit rates 5, 6 and 7% respectively and on the other hand are double-digit rates, that is 19%, 20%, 21% and 22% also being significant at the 5% level. Considering the coefficient of the single-digits that is 5, 6 and 7% being larger and thus more negative than the double-digit inflation rates, it presupposes that the single-digit rates have a larger inverse impact on the growth of Per Capita Income (PCI). What this means is that if inflation is larger than these single-digit rates the impact on Per Capita Income is larger. The conclusion, therefore, is that single-digit inflation rates have a much larger effect on welfare as measured by the growth of per capita income (PCI). Hence, inflation rates between 5–7% are optimal per the data employed for this study. Further, we ascertain the validity of the above results using the coefficient diagnosis (bound) test, which result is presented in Table 5. We show in Table 5 that the computed F-statistic of 5.136 is above the lower and upper bound levels at the 10 and 5% significant levels, respectively. Hence, the null hypothesis of no cointegration cannot be rejected, k Coefficient Std. Error t-statistics Prob R2 Adj R2 F-value 5 �8.260088 3.793424 �2.18 0.037 0.6328*** 0.4437 0.0014 6 �8.260088 3.793424 �2.18 0.037 0.6328*** 0.4437 0.0014 7 �8.260088 3.793424 �2.18 0.037 0.6328*** 0.4437 0.0014 8 �2.321623 1.865426 �1.24 0.221 0.5097 0.3714 0.0008 9 �2.771939 1.710435 �1.62 0.113 0.5224 0.3877 0.0008 10 �2.400193 1.688225 �1.42 0.163 0.5154 0.3787 0.0010 11 �2.780088 1.493367 �1.86 0.071 0.5724 0.4061 0.0014 12 �1.912189 1.499464 �1.28 0.211 0.6244 0.4131 0.0036 13 �2.091718 1.466253 �1.43 0.162 0.5236 0.3732 0.0016 14 �2.091718 1.466253 �1.43 0.0016 0.5236 0.3732 0.0016 15 �1.739223 1.375886 �1.26 0.214 0.5103 0.3722 0.0012 16 �2.755907 1.477814 �1.86 0.070 0.5725 0.4063 0.0014 17 �2.938242 1.432738 �2.05 0.048 0.5803 0.4170 0.0011 18 �2.728998 1.466194 �1.86 0.072 0.6200 0.4242 0.0022 19 3.284562 1.473254 �2.23 0.033 0.6350*** 0.4470 0.0013 20 3.835134 1.565703 �2.45 0.020 0.6447*** 0.4616 0.0010 21 �3.835134 1.565703 �2.45 0.020 0.6447*** 0.4616 0.0010 22 �3.835134 1.565703 �2.45 0.020 0.6447*** 0.4616 0.0010 23 0.3084246 1.629002 0.19 0.851 0.4907 0.3471 0.0022 24 0.3084246 1.629002 0.19 0.851 0.4907 0.3471 0.0022 25 0.252078 1.592926 0.16 0.875 0.4906 0.3469 0.0022 26 �0.6388085 1.696462 �0.38 0.709 0.4921 0.3488 0.0021 27 �0.0162326 1.854859 �0.01 0.993 0.4902 0.3465 0.0022 28 �0.9832936 1.881384 �0.52 0.604 0.4938 0.3510 0.0020 29 �0.9832936 1.881384 �0.52 0.604 0.4938 0.3510 0.0020 30 3.447786 1.934309 1.78 0.082 0.5286 0.3957 0.0007 Source(s): Authors’ estimation with data from WDI, 2019 10% 5% 1% p-value I (0) I (1) I (0) I (1) I (0) I (1) I (0) I (1) F 2.564 3.786 3.080 4.445 4.276 5.954 0.003 0.024 Note(s): HE: no level relationship F 5 5.136 Source(s): Authors’ estimation with data from WDI, 2019 Table 4. Auto Regressive distributed lag model estimation results Table 5. Result of coefficient diagnosis (bounds test) Optimal inflation rate for Ghana suggesting a long-run relationship among the variables. The confirmation of a long-run relationship among the variable reinforces the optimality of the threshold band from 5–7%. 4.2 Results from the threshold regression model To circumvent the issues of non-stationarity in the data, we differentiate the data to its level of stationarity for the estimation of the threshold model, the results of which are presented in Table 6. In the estimation of the threshold model, we trim the data by 20% and estimated a threshold of 7.71% that splits the sample into two regimes; regime 1, corresponds to the sample for which the inflation rate is less than or equal to 7.71% signifying a relatively low inflation regime, and regime 2, which corresponds to the portion of the sample for which inflation is greater than 7.71% signifying a relatively high inflation regime. The results established that whereas an inflation rate below 7.71% generates a positive impact on GDP per capita, with a coefficient of 0.0453, the rate above 7.71% generates a negative impact on per capita GDP growth such that when inflation increases by about 1% it leads to decrease in the per growth rate of GDP per capita by 0.15%. This finding supports a low inflation target, suggesting a single-digit inflation rate as optimal. In summary, the study estimated the optimal inflation rate to be between 5–7% using the ARDL model. This optimal rate is reinforced by the threshold method which concluded that there existed a threshold of 7.71% inflation above which inflation was not conducive to the growth of per capita GDP Therefore, regarding the first objective, the study concludes that the optimal inflation for Ghana is estimated to be within the single-digit band. 4.3 Unmasking the drivers of the inflation gap Establishing that the estimated optimal inflation rate corroborates the target inflation rate of the central bank, the question we interrogate is why the central bank has constantly missed the target. We can evaluate this situation under two scenarios: 1) the gap arising from how fast monetary policy affects its goals, that is smoothness in monetary transmission and/or; 2) deficiency inmonetary policy credibility that adversely affectsmonetary policy effectiveness. Ghana as a small open economy whose financial system is highly integrated with the global market may not enjoy monetary independence and credibility. Since monetary independence and credibility is a cornerstone in the inflation targeting framework to shape agents’ response to policy, lack of credibility may cause policy target to deviate from the actual. As a result, the study explored the drivers of the inflation gap by identifying the effect of monetary independence, exchange rate stability and financial sector conditions on the gap. The study adopted the Aizenman et al. (2008) approach to construct the monetary policy independence (MPI) and exchange rate stability (ERS) indices. The MPI index is constructed using the inverse of the annual correlation between the monthly interest rates of Ghana and that of the United States. The United States was chosen as the benchmark country because, Variables Regime 1 INF≤ 7.71% Regime 2 INF>7.71% Dependent variable: PCI INFRATE 0.4532 �0.1493 GCF 0.1144 0.47056 M2 �0.4356 �1.1682 GOVT_EXP 0.72412 �1.1721 CONSTANT 2.8495 1.9744 Source(s): Authors’ estimation with data from WDI, 2019 Table 6. Result of threshold regression result AJEMS according to Shambaugh (2004), Ghana’smonetary policy ismost closely related to that of the United States. The MPI index is mathematically expressed as: MPI ¼ 1� corrðii:ijÞ � ð−1Þ 1� ð−1Þ (11) Where 0≤MPI ≤ 1. A value of zero implies a complete absence of monetary policy independence or an exact correlation between the interest rate of Ghana and the United States, whereas a value equal to one assures complete monetary policy independence of the Ghanaian economy. The ERS index is constructed using the annual standard deviation of the monthly exchange rate between Ghana and the United States as ERS ¼ 0:01 0:01þ stdevðΔlogðexchangerateÞÞ (12) The study, therefore, presents the effect of the financial structure, the behaviour of the policy maker captured by monetary independence, and the external effect captured by exchange rate stability on the inflation gap.Within the study period, we estimate the average monetary policy independence and exchange rate stability as 0.65 and 0.185 respectively. This suggests that while Ghana has experienced independent monetary policy, its exchange rate has been very volatile. The study also estimated a very weak negative correlation between monetary independence and exchange rate stability of about�0.07. This reveals that the MPC sets the directions of monetary policy with little or no cognizance of the stability of the Ghana cedis. The variation in the inflation gap is estimated to be influenced by variations in monetary policy independence, the structure of financial structure (approximated by private sector credit) and exchange rate stability. At a 1% significant level, the combined effect of monetary independence, financial structure and exchange rate stability explains about 48% of variations in the inflation gap. However, independently these variables do not explain the variations in the inflation gap statistically. A unit change in exchange rate stability, monetary independence and structure of the financial sector resulted in 0.03-, 0.88- and 12-unit changes in the inflation gap respectively, though not statistically significant. 5. Concluding remarks The paper has sought to investigate the optimal inflation rate target of the Central Bank of Ghana and to determine whether the deviation from the target rate is due to a sub-optimal target or otherwise. We established through both the ARDL and threshold regression approaches that the target rate of 8%± 2 target pursued is ideal in terms of economic growth. Therefore, the inflation deviation from the target rates could not be due to a sub-optimal target. Implying that although the central bank has consistently missed its inflation target, the single-digit target is however ideal and worthy of pursuit. Unmasking the drivers of the gap, the study established that the combined effect of own policy behaviour (monetary independence), external outlook (exchange rate stability) and the financial structure affect actual inflation deviation from the target rate. We, therefore, recommend that inflation has to be brought under control within growth- enhancing ranges. This stems from the regular occurrences of target misses resulting from overshooting the target evidenced by the high number of positive inflation gaps identified in this study, and the optimal inflation rate estimated. Again, the Central Bank is urged to maintain its single-digit inflation target as it would aid in achieving robust economic growth and consequently improve the growth of per capita GDP. Optimal inflation rate for Ghana Finally, the central bank of Ghana as a policymaker must commit to policy rules as a key requirement of successful policy outcomes would hinge the credibility of the policy-making body. All policymakers must strive to commit themselves to achieving targets and stimulating credibility as the study established that their policy behaviour is related to the conduct of monetary independence and has large effects on achieving targets. Notes 1. The Bank of Ghana’s inflation target is within a band of 8% ± 2 2. The gap is the difference between the target and actual inflation rate. A zero gap indicates the desired rate was achieved. A positive gap indicates that the target was not achieve while a negative gap indicates that the actual rate was below the target rate. 3. A forward-looking rule based monetary policy framework unofficially adopted by the Bank of Ghana in year 2002 and officially in the year 2006. References Abdulqadir, I.A., Chua, S.Y. and Mohd, S. (2020), “Investigation of optimal inflation targets for 15 major oil exporting Sub-Saharan African countries. A panel threshold estimation”, Journal of Economics, Finance and Administrative Science, Vol. 25 No. 49, pp. 45-60. Abuselidze, G. (2019), “Modern challenges of monetary policy strategies: inflation and devaluation influence on economic development of the country”, Academy of Strategic Management Journal, Vol. 18 No. 4, pp. 1-10. Adam, K., Gautier, E. and Weber, H. (2022), Why Central Banks Should Aim for a Positive Inflation Target, Deutsche Bundesbank. https://cepr.org/voxeu/columns/why-central-banks-should-aim- positive-inflation-target-0. Adaramola, A.O. and Dada, O. (2020), “Impact of inflation on economic growth: evidence from Nigeria”, Investment Management and Financial Innovations, Vol. 17 No. 2, pp. 1-13. Aizenman, J., Chinn, M.D. and Ito, H. (2008), “Assessing the emerging global financial architecture: measuring the trilemma’s configurations over time”, National Bureau of Economic Research Working papers, doi: 10.3386/w14533. http://www.nber.org/papers/w14533. Albagli, E. and Schmidt-Hebbel, K. (2003), “By how much and why do inflation targeters miss their targets?”, In Conference on Monetary Policy and Learning. Barro, R., J. (1995), “Inflation and Economic Growth”, National Bureau of Economic Research Working Paper Series, Vol. 5326, doi: 10.3386/w5326. Barta, A., Gerrard, J., Steffen, J. and Zaspel, F. (2023), “Inflation in economic theoryj exploring economics”. Bernanke, B.S., Laubach, T., Mishkin, F. and Posen, A. (1999), Inflation Targeting: Lessons from the International Experience, Princeton University Press, Princeton New Jersey. Bhattarai, B. (2014), Inflation and Growth: Estimation of Threshold Level of Inflation in Nepal, Central Department of Economics Ii Tribhuvan University, Kirtipur Kathmandu. Dammak, T.B. and Helali, K. (2017), “Threshold effects on the relationship between inflation rate and economic growth in Tunisia”, International Economic Journal, Vol. 31 No. 2, pp. 310-325, doi: 10. 1080/10168737.2017.1289546. Dastgerdi, H.G. (2020), “Inflation theories and inflation persistence in Iran”, Zagreb International Review of Economics and Business, Vol. 23 No. 2, pp. 1-20. Ehrmann, Michael (2021), “Point targets, tolerance bands or target ranges? Inflation target types and the anchoring of inflation expectations”, Journal of International Economics, Vol. 132, doi: 10. 1016/j.jinteco.2021.103514. AJEMS https://cepr.org/voxeu/columns/why-central-banks-should-aim-positive-inflation-target-0 https://cepr.org/voxeu/columns/why-central-banks-should-aim-positive-inflation-target-0 https://doi.org/10.3386/w14533 http://www.nber.org/papers/w14533 https://doi.org/10.3386/w5326 https://doi.org/10.1080/10168737.2017.1289546 https://doi.org/10.1080/10168737.2017.1289546 https://doi.org/10.1016/j.jinteco.2021.103514 https://doi.org/10.1016/j.jinteco.2021.103514 Fischer, S. (1993), “The role of macroeconomic factors in growth”, Journal of Monetary Economics, Vol. 32 No. 3, pp. 485-512, doi: 10.1016/0304-3932(93)90027-D. Frimpong, J.M. and Oteng-Abayie, E.F. (2010), “When is inflation harmful? Estimating the threshold effect for Ghana”, American Journal of Economics and Business Administration, Vol. 2 No. 3, pp. 232-239. Hansen, B.,E. (1996), “Inference when a nuisance parameter is not identified under the null hypothesis”, Econometrica: Journal of the econometric society, Vol. 64 No. 2, pp. 413-430. Hansen, B.E. (1999), “Threshold effects in non-dynamic panels: estimation, testing, and inference”, Journal of Econometrics, Vol. 93 No. 2, pp. 345-368. Hansen, B.E. (2000), “Sample splitting and threshold estimation”, Econometrica, Vol. 68 No. 3, pp. 575-603. Kahn, G.A. and Parrish, K. (2020), “Conducting monetary policy with inflation targets”, in Handbook of Monetary Policy, Routledge, pp. 485-512. Khan, M. and Hanif, W. (2020), “Institutional quality and the relationship between inflation and economic growth”, Empirical Economics, Vol. 58, pp. 627-649. Khan, M.S. and Senhadji, A.S. (2001), “Threshold effects in the relationship between inflation and growth”, IMF Staff Papers, Vol. 48 No. 1, pp. 1-21, doi: 10.2307/4621658. King, R.G. and Levine, R. (1993), “Finance and growth: schumpeter might be right”, The Quarterly Journal of Economics, Vol. 108 No. 3, pp. 717-737. Lapavitsas, C. (2022), “The return of inflation and the weakness of the side of production”, The Japanese Political Economy, Vol. 48 Nos 2-4, pp. 149-169, doi: 10.1080/2329194X.2022.2142613. Mandeya, S.M.T. and Ho, S.-Y. (2021), “Inflation, inflation uncertainty and the economic growth nexus: an impact study of South Africa”, MethodsX, Vol. 8, 101501. Musah, A., Sare, Y.A. and Ibrahim, M. (2019), “Examining the threshold effects of inflation on economic growth in Ghana”, Ghanaian Journal of Economics, Vol. 7 No. 1, pp. 5-23. Nigusse, T., Tadesse, T. and Melaku, T. (2019), “Supply and demand side determinants of inflation in Ethiopia: auto-regressive distributed lag model (ARDL)”, International Journal of Commerce and Finance, Vol. 5 No. 2, pp. 8-21. Pesaran, M.H., Shin, Y. and Smith, R.J. (2001), “Bounds testing approaches to the analysis of level relationships”, Journal of Applied Econometrics, Vol. 16 No. 3, pp. 289-326. Piętak, Ł. (2014), “Review of theories and models of economic growth”, Comparative Economic Research. Central and Eastern Europe, Vol. 17 No. 1, pp. 45-60. Shaikh, P.A., Muhammad, F. and Khan, S.K. (2022), “The dynamic theories of inflation”, Pakistan Journal of International Affairs, Vol. 5 No. 2, pp. 956-971. Shambaugh, J.C. (2004), “The effect of fixed exchange rates on monetary policy”, The Quarterly Journal of Economics, Vol. 119 No. 1, pp. 301-352. Siddik, M.N.A. (2023), “Does macroeconomic stability promote economic growth? Some econometric evidence from SAARC countries”, Asian Journal of Economics and Banking, Vol. ahead-of-print No. ahead-of-print, doi: 10.1108/AJEB-05-2022-0052. Svensson, L.E. (2010), “Inflation targeting”, Handbook of Monetary Economics, Vol. 3, pp. 1237-1302. The World Bank, World Development Indicators (2019). [Data file]. available at: http://data.worldbank. org/indicator Tien, N.H. (2021), “Relationship between inflation and economic growth in Vietnam”, Turkish Journal of Computer and Mathematics Education (TURCOMAT), Vol. 12 No. 14, pp. 5134-5139. Totonchi, J. (2011), “Macroeconomic theories of inflation”, Paper presented at the International Conference on Economics and Finance Research. Optimal inflation rate for Ghana https://doi.org/10.1016/0304-3932(93)90027-D https://doi.org/10.2307/4621658 https://doi.org/10.1080/2329194X.2022.2142613 https://doi.org/10.1108/AJEB-05-2022-0052 http://data.worldbank.org/indicator http://data.worldbank.org/indicator Further reading Ambrocio, G., Ferrero, A., Jokivuolle, E. and Ristolainen, K. (2022), “What should the inflation target be? Views from 600 Economists’” CEPR Discussion Paper No. DP17289, available at SSRN: https://ssrn.com/abstract54121474 Arsi�c, M., Mladenovi�c, Z. and Nojkovi�c, A. (2022), “Macroeconomic performance of inflation targeting in European and Asian emerging economies”, Journal of Policy Modeling, Vol. 44 No. 3, pp. 675-700, doi: 10.1016/j.jpolmod.2022.06.002. Banerjee, R.N., Contreras, J., Mehrotra, A. and Zampolli, F. (2020), “Inflation at risk in advanced and emerging market economies”, Bank for International Settlements Working Paper 883. Blanco, A. (2021), “Optimal inflation target in an economy with menu costs and a zero lower bound”, American Economic Journal: Macroeconomics, Vol. 13 No. 3, pp. 108-141. Brown, R.L., Durbin, J. and Evans, J.M. (1975), “Techniques for testing the constancy of regression relationships over time”, Journal of the Royal Statistical Society. Series B (Methodological), Vol. 37 No. 2, pp. 149-192, available at: http://www.jstor.org/stable/2984889 De Grauwe, P. and Ji, Y. (2019), “Inflation targets and the zero lower bound in a behavioural macroeconomic model”, Economica, Vol. 86 No. 342, pp. 262-299. Duong, T.H. (2022), “Inflation targeting and economic performance over the crisis: evidence from emerging market economies”, Asian Journal of Economics and Banking, Vol. 6 No. 3, pp. 337-352. Gong, D. and Qian, Z. (2022), “Inflation targeting and financial crisis”, Applied Economics, Vol. 54 No. 41, pp. 4782-4795. Khan, M.S. and Ssnhadji, A.S. (2001), “Threshold effects in the relationship between inflation and growth”, IMF Staff Papers, Vol. 48 No. 1, pp. 1-21. Olamide, E., Ogujiuba, K. and Maredza, A. (2022), “Exchange rate volatility, inflation and economic growth in developing countries: panel data approach for SADC”, Economies, Vol. 10 No. 3, p. 67. Poku, K., Opoku, E. and Agyeiwaa Ennin, P. (2022), “The influence of government expenditure on economic growth in Ghana: an Ardl approach”, Cogent Economics and Finance, Vol. 10 No. 1, 2160036. Quartey, P. (2010), “Price stability and the growth maximizing rate of inflation for Ghana”, Modern Economy, Vol. 1 No. 03, p. 180. Tamborini, R. (2022), “Beware of pitfalls in the European Central Bank’s review of monetary policy strategy”, The Economists’ Voice, Vol. 19 No. 1, pp. 15-23. Yilmazkuday, H. (2022), “COVID-19 and Monetary policy with zero bounds: a cross-country investigation”, Finance Research Letters, Vol. 44, 102103. Corresponding author Richard Amoatey can be contacted at: ramoatino@gmail.com For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: permissions@emeraldinsight.com AJEMS https://ssrn.com/abstract=4121474 https://ssrn.com/abstract=4121474 https://doi.org/10.1016/j.jpolmod.2022.06.002 http://www.jstor.org/stable/2984889 mailto:ramoatino@gmail.com How optimal is Ghana's single-digit inflation targeting? An assessment of monetary policy effectiveness in Ghana Background Literature review Methodology Variable selection and data source Econometric approach Estimation procedure of the ARDL model Threshold regression Test for threshold effect Asymptotic distribution of threshold estimate Results and discussions Results from the auto-regressive distributed lag (ARDL) model Results from the threshold regression model Unmasking the drivers of the inflation gap Concluding remarks Notes References Further reading