The objective of the study is to examine the impact of external debt on inflation by highlighting the moderating role of imports in the external debt–inflation nexus.
The autoregressive distributed lag model is employed to show short-run versus long-run dynamics using data from Guinea for the period of 1986–2021.
The findings reveal that imports negatively affect inflation in the short run, while (public) external debt and imports drive long-run inflation. Furthermore, imports positively moderate the external debt–inflation nexus in the long run. The threshold of imports above which external debt has a positive effect on inflation is roughly 34 (% of GDP).
Although this research reports interesting findings, the small sample size represents a structural limitation of our findings. Another limitation is data quality, a common problem in research.
The contribution is threefold. First, Guinea is an African country where inflation is relatively higher. In general, there is a lack of studies analyzing simultaneously the effect of external debt and trade on inflation, and statements concerning the role of imports on inflation remain conjectural. To fill this gap, this research simultaneously highlights their impacts on inflation. Second, for the sensitivity analysis, this research analyzes the effects of overall trade and public external debt on inflation. Third, to our knowledge, it is the first time a study explores the moderating role of imports in the external debt–inflation nexus.
1. Introduction
A fundamental challenge in open countries is controlling or keeping inflation relatively low. Recent studies are motivated by the increased share of debt, especially public debt observed in the world due to the COVID-19 pandemic (Ando et al., 2025; Diamond et al., 2025). Traditional views emphasize the primary role of monetary policies in determining the price level while usually marginalizing the role of fiscal policies. This implies a certain independence of the central bank in managing the inflation rate. Friedman (1968) affirms that the monetary authority can manage the inflation rate, especially in the long run. However, if the monetary policy instrument is open market operations, Sargent and Wallace (1981) argue that the monetary authority’s ability to manage inflation in a monetary economy is very limited, even though there is a close connection between the monetary base and price level. They explain the link between government debt and inflation by exploring the interactions between monetary and fiscal policy. These interactions imply two polar forms of coordination: a coordination scenario where monetary policy dominates fiscal policy and a coordination scenario where fiscal policy dominates monetary policy. In both the “monetary dominance” regime and the “fiscal dominance” regime, deficits force the central bank to monetize revenue, leading to a rise in the inflation rate.
The fiscal theory of the price level (FTPL) is a complement to Sargent and Wallace's “unpleasant monetarist arithmetic”. It refers to the non-Ricardian or “fiscal dominance” regime. According to this theory, the price level is determined to equate the real value of nominal government debt with the present value of primary government budget surpluses. The fiscal theory implies that the government can directly target the price level using fiscal variables alone (Daniel, 2001). Indeed, an increase in government spending leads to an increase in aggregate demand and hence inflation pressures. External debt is more compatible with the FTPL, which is extended to the open economy FTPL (Daniel, 2001; Bajo-Rubio et al., 2009). Daniel (2001) argues that when all governments follow a “no-surplus” policy, the open-economy FTPL is assimilated to the closed-economy fiscal theory. Indeed, in an open economy, when monetary policy rules are carefully and independently adapted, governments can determine price and exchange rates. Specifically, price and exchange rate are determined when each government sets a policy aimed at avoiding an intertemporal budget surplus. Furthermore, any government that aims to improve the welfare of its citizens would adopt this type of “no-surplus” policy. The “fiscal view” of inflation has been prominent in the developing country literature, which has long recognized that less efficient tax collection, political instability and more limited access to external borrowing tend to lower the relative cost of seigniorage and increase dependence on the inflationary tax (Catão and Terrones, 2005). The empirical effect of (external) debt on inflation is unclear (Essien et al., 2016; Aimola and Odhiambo, 2021a,b; Saani et al., 2024; Sharaf et al., 2025).
Furthermore, openness, through reduced tariffs, increases imports and reduces inflation (Romer, 1993). Romer (1993) argues that there is a negative correlation between openness and inflation due to the dynamic inconsistency of discretionary monetary policy. The theoretical literature identifies some mechanisms through which trade affects inflation. These channels include the real exchange rate depreciation, terms of trade, welfare, competition and firms’ mark-ups (Romer, 1993; Lane, 1997; Alfaro, 2005; Corsetti and Pesenti, 2005; Watson, 2016). According to this literature, these channels are seen as instrumental to the adjustment of the relative prices and hence inflation. The recent literature development is based on the general equilibrium model of optimal monetary policy among interdependent economies with nominal rigidities, imperfect competition in production and forward-looking price setting (Alfaro, 2005; Corsetti and Pesenti, 2005; Watson, 2016).
From an empirical point of view, Carluccio et al. (2023) argue that statements concerning the role of imports on inflation remain rather conjectural. The few empirical studies include (1) Lin et al. (2017) and Herawati and Sidik (2022), who find a negative effect of imports on inflation in African and ASEAN countries, respectively, (2) while Carluccio et al. (2023) and Jingjing et al. (2025) report a positive effect of imports on inflation in France (micro analysis) and Afghanistan, respectively. In parallel, several studies explore the effect of trade (including imports) on inflation. Yiheyis (2013) finds a positive effect of trade on inflation at the African level, while the abundant literature on the effect of trade on inflation in other regions is inconclusive (Romer, 1993; Lane, 1997; Ghosh, 2014; Munir et al., 2015; Sepehrivand and Azizi, 2016; Mukhtar et al., 2019; Luangaram and Wongpunya, 2024). In particular, some studies investigate the effect of external debt on inflation in Africa, especially in Nigeria (Essien et al., 2016; Aimola and Odhiambo, 2021b), while there are almost no studies for the case of Guinea. Recent data show that between 2010 and 2022, the minimum value of the inflation rate in Nigeria is about 8.04% in 2014, while the highest value is about 18.85% in 2022. Similarly, Guinea registers roughly the same inflation rates between 2010 and 2022 (minimum value of 8.2% in 2015 and maximum value of 21.4% in 2011). Based on this literature and descriptive evidence, the questions are as follows: Does external debt affect inflation in Guinea? Do imports play a moderating role in the debt–inflation link?
The contribution of this study to the literature on the implications of external debt and imports for inflation in Guinea is threefold. First, the aforementioned descriptive evidence shows that Guinea is an African country where inflation is relatively higher. Furthermore, in general, there is a lack of a simultaneous analysis of the effect of external debt and trade on inflation. In particular, recently, Carluccio et al. (2023) argue that empirical literature is scarce and statements concerning the role of imports on inflation remain rather conjectural. To fill this gap, this research simultaneously highlights the effects of both external debt and imports on inflation in Guinea. Second, according to Lane (1997), imports represent a good measure of the share of tradables in consumption if, for example, consumers have a love of variety in tradables. However, for the sensitivity analysis, this research uses an alternative proxy of trade (overall trade) and debt (public external debt). Third, to our knowledge, it is the first time a study explores the moderating role of imports in the external debt–inflation nexus. Computed threshold is important for practitioners in economic development.
The objective is to analyze the effects of external debt on inflation by highlighting the moderating role of imports in the external debt–inflation nexus: specifically, to (1) study the effects of external debt and imports on inflation and (2) test whether imports play a moderating role in the debt–inflation nexus. The main findings show that imports negatively affect inflation in the short run, while (public)external debt and imports drive long-run inflation. Furthermore, the threshold of imports above which external debt has a positive effect on inflation is roughly 34 (% of GDP).
2. Literature review
2.1 Theoretical literature
Openness, in terms of trade and external debt, affects inflation. According to Friedman (1968), monetary policy cannot permanently influence the levels of real output, unemployment or real rates of return on securities, while the inflation rate can be managed by the monetary authority, especially in the long run. However, if monetary policy is seen as open-market operations, Sargent and Wallace’s “unpleasant monetarist arithmetic” shows that the ability of the monetary authority to manage inflation in a monetary economy is very limited, even though the monetary base and price level remain closely connected. To this end, Sargent and Wallace (1981) consider the particular case where monetary and fiscal policies are coordinated in a certain way and the public’s demand for interest-bearing government debt has a certain form.
The FTPL is a complement to Sargent and Wallace’s “unpleasant monetarist arithmetic”. According to this theory, the price level is determined to equate the real value of nominal government debt with the present value of primary government budget surpluses. Furthermore, external debt is more compatible with the FTPL, which is extended to the open-economy FTPL. The literature discusses the implications of the FTPL on inflation targeting in open economies (Daniel, 2001; Bajo-Rubio et al., 2009). Indeed, Daniel (2001) argues that when all governments follow a “no-surplus” policy, the open-economy FTPL is assimilated into the closed-economy fiscal theory. Indeed, in an open economy, when monetary policy rules are carefully and independently adapted, governments can determine prices and exchange rates. Specifically, price and exchange rate are determined when each government sets a policy aimed at avoiding an intertemporal budget surplus. Furthermore, any government that aims to improve the welfare of its citizens would adopt this type of “no-surplus” policy.
Finally, the conventional view of trade liberalization suggests that trade (increased imports through reduced tariffs or increased exports through subsides) leads to a reduction in inflation (Romer, 1993). The literature highlights some mechanisms that link trade and inflation. Romer (1993) argues that there is a negative correlation between openness and inflation. The theoretical literature argues that trade openness leads to lower average inflation because (1) surprise monetary expansion results in the real exchange rate depreciation and (2) the harms or real depreciation is greater in more open countries, and the gains of surprise expansion are decreasing in the degree of openness (Romer, 1993). The theoretical explanation provided by Rogoff (1985) is that more open countries gain less from unanticipated inflation. Unanticipated monetary expansion leads to the real exchange rate depreciation, which in turn leads to a negative terms-of-trade effect. The more open the country, the more the real exchange depreciates, thus reducing incentives to undertake monetary expansion (Alfaro, 2005). However, this explanation concerns only large economies.
Lane (1997) proposes alternative mechanisms by providing an equilibrium relationship (from welfare micro-foundations) between inflation and trade that holds even for a country too small to affect international relative prices. Lane (1997) develops a model where traded sector output is a constant endowment of a homogeneous good, which is a perfect substitute for foreign output. The mechanism linking the welfare effects of monetary surprises to trade does not rely on a large-country terms of trade effect but rather is due to imperfect competition and nominal price rigidity in the non-traded sector. Assuming that the government chooses a monetary policy based on welfare, this produces a negative correlation between trade and the incentive to unleash a surprise monetary expansion. Corsetti and Pesenti (2005) conclude that in a global economy framework, policies exclusively attempting to stabilize domestic prices and output gap may actually result, on average, in inefficiently high consumer prices of imports and, therefore, suboptimal welfare levels for domestic consumers. According to Watson (2016), trade affects inflation through the competition channel, which in turn influences prices. Also, trade negatively affects firms' mark-ups, explaining the negative link between trade and inflation.
2.2 Empirical studies
Focusing on SSA countries, we identify three groups of studies: (1) the first group explores the effect of external public debt on inflation (Essien et al., 2016; Sumba et al., 2024), (2) the second group explores the effect of total external debt on inflation (Helmy, 2022) and (3) the last group investigates the interaction between total public debt and inflation (Aimola and Odhiambo, 2021a,b; Saani et al., 2024). Essien et al. (2016) use the vector autoregression (VAR) model to conclude that external public debt has no effect on inflation in Nigeria over the period of 1970–2014. However, applying the generalized method of moments (GMM) in 45 SSA countries from 2005 to 2022, Sumba et al. (2024) reveal a positive effect of external public debt on inflation. Helmy (2022) uses autoregressive distributed lag (ARDL) cointegration analyses from 2000 M1 to 2020 M1 to investigate the relationship between external debt and inflation in Egypt. The findings reveal that external debt raises inflation in both the short and long run in Egypt. Aimola and Odhiambo (2021a) explore the link between total public debt and inflation in Ghana over the period of 1983–2018. Applying the ARDL model, the results indicate that public debt exerts inflationary pressures. However, using the same model, Saani et al. (2024) report a negative relationship between total public debt and inflation for the period of 1990–2022 in Ghana. In Nigeria, based on the ARDL model over the period of 1983–2018, Aimola and Odhiambo (2021b) find an insignificant effect of total public debt on inflation in both the short and long run.
Concerning other countries or regions, Ekinci (2016) studies the link between external debt and inflation in Turkey between 2003 and 2015 using a simple linear regression analysis. The results show that external debt exerts inflationary pressures in Turkey. Recently, based on the ARDL approach from 1970 to 2020, Sharaf et al. (2025) also report a positive effect of external debt on inflation in the long run in Jordan. Other studies explore the total public debt–inflation nexus. Prabheesh et al. (2024) investigate the effect of public debt on inflation in 10 emerging market economies during the COVID-19. Using the panel vector autoregression (PVAR) regressions on monthly data, they find a positive effect of public debt on inflation. Also, applying the panel Granger causality test for a sample of 39 developing countries from 1990 to 2020, Shah et al. (2024) show the existence of a positive bidirectional relationship between public debt and inflation.
Regarding the specific effect of imports on inflation, the empirical literature remains insufficient (Carluccio et al., 2023). Lin et al. (2017) investigate the effect of imports on inflation in SSA countries from 1985 to 2012. Applying the 2SLS and GMM technique, they find a significant negative (insignificant positive, respectively) effect of imports on inflation. Also, using a panel least squares, Herawati and Sidik (2022) find a negative impact of imports on inflation in ASEAN countries from 2006 to 2019. However, at the micro level, Carluccio et al. (2023) report a positive effect of imports on inflation in France over the period of 1994–2014. Also, Jingjing et al. (2025) explore the impacts of imports on inflation in Afghanistan. Using the ARDL model from 1990 to 2023, they document a positive effect of imports on inflation. Beyond the specific effect of imports on inflation, other studies focus on the effect of trade on inflation using various techniques, such as ARDL, VAR and GMM. The first empirical literature reports a negative effect of trade on inflation (Romer, 1993; Lane, 1997; Terra, 1998), while others report positive or ambiguous effects of trade on inflation (e.g. Alfaro, 2005; Yiheyis, 2013; Ghosh, 2014; Munir et al., 2015; Sepehrivand and Azizi, 2016; Mukhtar et al., 2019; Luangaram and Wongpunya, 2024).
The above literature reveals that there is a lack of studies exploring the main determinants of inflation in Guinea. Furthermore, our study differs from the existing literature by simultaneously analyzing the effects of debt and imports on inflation. Finally, our study investigates the moderating role of imports in the external debt–inflation nexus.
3. Methodology
3.1 Model
The first objective of this work is to analyze the effects of external debt (and imports) on inflation in Guinea. Based on the literature (Ekinci, 2016; Essien et al., 2016), the baseline model is as follows:
π is the dependent variable, inflation. REDS, the ratio of external debt stocks, is the explanatory variable of interest. For robustness checks, we use beyond the ratio of total external debt stocks (RTEDS, % of GDP) and the ratio of public external debt stocks (RPUEDS, % of GDP). The assumption is that external debt increases inflation. X is the vector of control variables selected from the existing literature. These are trade openness captured by imports (M), money stock (MON), real GDP per capita (GDPpc) and exchange rate (EXR) (Romer, 1993; Lin et al., 2017).
3.2 Estimation techniques and diagnostic tests
Generally, the ARDL approach developed by Pesaran et al. (2001) is preferred as the estimation of this model provides common tests. This approach is based on the error correction model (ECM). Equation (1) can take the following form in the context of the ARDL specification:
θ1, θ2, θ3, θ4, θ5 and θ6 represent the coefficients of the long-run effects, while λ1, λ2, λ3, λ4, λ5 and λ6, capture the short-run effects. Some reasons justify this choice. Firstly, unlike the Johansen technique, which requires that all variables are integrated of the same order, I(1), ARDL is valid even if all explanatory variables are of the same degree of integration, I(0) or I(1), or not. However, we must ensure that the degree of integration of the dependent variable is 1, I(1) (Jordan and Philips, 2018). This indicates that the unit root test is not necessary; it is conducted only to ensure that the variables are not I(2). Secondly, this approach is robust for small and finite sample sizes. Finally, the ARDL approach provides unbiased results as it controls for the possible existence of endogenous variables in the model. For robustness checks, the fully modified OLS (FMOLS) technique will be employed if the regressors are I(1). Indeed, Pesaran et al. (2001) argue that if the regressors are I(1), the ARDL approach is directly comparable to the semi-parametric, FMOLS method for estimation of cointegrating relations. The FMOLS technique is free from endogeneity issues, small sample size bias and serial correlation.
Furthermore, the ARDL bounds testing method introduced by Pesaran et al. (2001) is implemented to examine cointegration among variables. This approach offers the possibility to test for the existence of a long-run relationship among the regressors (cointegration) by implementing an F-test. The null hypothesis implies that the t-type and F-type tests of no cointegration are confirmed. Thus, the procedure consists of the comparison of the computed values of F-statistics (or t-statistics) to critical values. If the calculated F statistic is greater than the upper bound critical value, it means that there is a long-run relationship among regressors. However, if it is below the lower bound, the null hypothesis of no cointegration cannot be rejected. Nevertheless, if the calculated value of the F-statistic lies between the bounds, the test is inconclusive.
Also, the ARDL bounds testing procedure is sensitive to the selection of the lag structure. The inappropriate selection of lag length may lead to biased findings. In this vein, it is imperative to have exact information about the lag order of the series to avoid the problem of bias of the ARDL F-statistics (Kripfganz and Schneider, 2023). The literature often explores the Akaike information criterion (AIC) for the selection of lag length. The hypotheses of the bounds test are fulfilled when diagnostic tests, such as tests for autocorrelation, heteroscedasticity and normality of the errors, are satisfied (Kripfganz and Schneider, 2023). Other diagnostic tests include the Ramsey RESET test and the cumulative sum (CUSUM) test for stability. Besides, for the unit root test for time series, the literature often applies both the Dickey–Fuller generalized least-squares (DF-GLS) and Phillips–Perron (PP) unit-root tests (Jordan and Philips, 2018).
Finally, the research analyzes the moderating role of imports in the external debt–inflation nexus. To this end, we introduce the interaction term, captured by the product between external debt and imports, in Eq. (2):
Importantly, the interpretation of the coefficients associated with external debt and imports in Eq.(2) differs from the interpretation of these coefficients in Eq.(3). In this type of literature, following Brambor et al. (2006) and Mamba (2021, 2025), the components of the interaction term in Eq.(3) do not capture unconditional effects but conditional effects. In the long run, for example, the estimated value of Ɵ2 captures the marginal effect (ME) of external debt on inflation when imports take the value zero. A positive (negative) sign on the significant coefficient associated with the interaction term (Ɵ7) means that imports positively (negatively) moderate the external debt–inflation nexus, implying that imports play an amplifying (a reducing) role in the relationship between external debt and inflation. In the long run, we can compute the total ME of external debt on inflation as follows:
3.3 Data
Data are collected from the World Bank’s World Development Indicators (WDI) database. Due to data availability (some data are missing in the more recent WDI database), the sample covers the period of 1986–2021. Table A1 in Appendices defines the variables, while Table A2 in Appendices presents descriptive statistics. In line with Romer (1993) and Lane (1997), the research uses the natural logarithm of the dependent variable to mitigate the heteroscedasticity issue.
4. Findings and discussion
4.1 Correlation matrix, multicollinearity and unit-root tests
Table 1 reports the correlation matrix and the multicollinearity. Table 1 reveals that there is a strong correlation among explanatory variables. The minimum correlation is larger than 0.5, revealing a problem of multicollinearity. Overall, the correlation appears to be stronger between the exchange rate and each of the variables in the model. The multicollinearity test through the variance inflation factor (VIF) shows that the value of the mean VIF is roughly 9, which is greater than 5. In this case, the literature argues that the specification suffers from the multicollinearity issue (Mamba and Balaki, 2022; Ali and Mamba, 2025; Mamba, 2025; Mamba et al., 2025). As shown in Table 1, this multicollinearity problem is attributed to the exchange rate variable, which has the highest value of the individual VIFs. Excluding exchange rate from the specification, the value of the mean VIF tends roughly towards 5.
Correlation matrix and multicollinearity test
| LnRTEDS | LnM | LnMON | LnGDPpc | LnEXR | |
|---|---|---|---|---|---|
| LnRTEDS | 1.000 | ||||
| LnM | −0.806 | 1.000 | |||
| LnMON | −0.789 | 0.754 | 1.000 | ||
| LnGDPpc | −0.836 | 0.651 | 0.873 | 1.000 | |
| LnEXR | −0.825 | 0.781 | 0.939 | 0.918 | 1.000 |
| VIF | 5.680 | 4.460 | 8.62 | 9.970 | 15.850 |
| Mean VIF = 8.920 | |||||
| Mean VIF = 5.360 (without exchange rate) | |||||
Table 2 reports unit-root tests, including the DF-GLS and PP tests. Since the dependent variable (inflation) is stationary at the first difference, this implies that the ARDL technique may be suitable (Jordan and Philips, 2018). At the 5% threshold, all variables are stationary in the first difference. These findings imply that there may be a cointegrating relationship between inflation and regressors.
Phillips–Perron and DF-GLS tests for a unit root in time series
| Variable | Dickey–Fuller generalized least-squares (DF-GLS) unit root test | Phillips–Perron test statistic | ||||
|---|---|---|---|---|---|---|
| Level | First difference | Decision | Level | First difference | Decision | |
| Ln(π) | −1.83(1) | −3.68(1)** | I(1) | −2.79*(3) | −6.03(3)*** | I(1) |
| LnRTEDS | −1.49(1) | −3.86(1)*** | I(1) | −0.58(3) | −6.18(3)*** | I(1) |
| LnRPUEDS | −1.49(1) | −3.79(1)*** | I(1) | −0.60(3) | −6.45(3)*** | I(1) |
| LnM | −1.09(3) | −5.91(2)*** | I(1) | −1.11(3) | 7.67(3)*** | I(1) |
| LnMON | −3.27(1)* | −4.88(2)*** | I(1) | −0.52(3) | −4.96(3)*** | I(1) |
| LnGDPpc | −1.25(1) | −3.87(1)*** | I(1) | 1.60(3) | −4.33(3)*** | I(1) |
| LnEXR | 1.86(1) | 5.19(1)*** | I(1) | −2.02(3) | −4.62(3)*** | I(1) |
Note(s): The choice of the optimal lag structure (reported in parentheses) of the DF–GLS test is based on the Schwarz information criterion (SIC). For the Phillips–Perron (PP) test, Newey–West lags are in parentheses and represent the optimal lag structure. *p < 0.1; **p < 0.05; ***p < 0.01
4.2 ARDL co-integration tests
This section explores the co-integration test by employing the bounds testing method proposed by Pesaran et al. (2001). We recall that the correct ARDL F-statistics are chosen depending on the AIC for the selection of lag length (Table 3). In all ARDL models, the findings confirm the existence of a long-run relationship between inflation and other variables. Indeed, the calculated F statistics are greater than the upper bound critical value at least at the 5% threshold, implying the rejection of the null hypothesis of no level relationship.
Co-integration test through the ARDL bounds testing method
| Nature of ARDL | Critical values (lower bounds): I(0) | Critical values (upper bounds): I(1) | Computed F-statistic | Proxy of debt | ||||
|---|---|---|---|---|---|---|---|---|
| 10% | 5% | 1% | 10% | 5% | 1% | |||
| ARDL(1,0,2,1,2) | 3.03 | 3.47 | 4.40 | 4.06 | 4.57 | 5.72 | 6.46 | RTEDS |
| ARDL(2,0,2,1,2,2) | 2.75 | 3.12 | 3.93 | 3.79 | 4.25 | 5.23 | 5.71 | RTEDS |
| ARDL(2,0,2,1,2,2) | 2.51 | 3.09 | 4.55 | 3.97 | 4.78 | 6.81 | 5.15 | |
| ARDL(1,0,2,1,2) | 3.03 | 3.47 | 4.40 | 4.06 | 4.57 | 5.72 | 7.28 | RPUEDS |
| ARDL(1,0,2,0,0,0) | 2.75 | 3.12 | 3.93 | 3.79 | 4.25 | 5.23 | 6.72 | RTEDS_M |
4.3 ARDL specification: direct and combined effects of external debt and imports on inflation
Table 4 displays the normality test, Ramsey RESET test for a good model specification, White's test for heteroscedasticity, Breusch–Godfrey LM test for autocorrelation and the CUSUM test for stability. Except for the findings in column (4) (valid at the 5%, but not at the 10%, threshold for normality and autocorrelation tests), which also suffer from the multicollinearity issue, these tests validate the reported findings and suggest that the assumptions underlying the bounds test are met. Table 4 shows the statistical significance of the negative speed-of-adjustment coefficient at the 1% threshold in linear and non-linear regressions, suggesting that there will be a long-run equilibrium convergence when there is a shock to any of the regressors. Before applying the extended ARDL models, columns (1–2) display the effects of external debt and imports on inflation, respectively, by starting the regressions with the ARDL(1,0) models. These columns show the significant and positive effects of external debt and imports on inflation in the long run. Similarly, the findings in column (3) reveal a significant and positive effect of external debt on inflation in the long run at the 1% threshold. Ceteris paribus, a 1% increase in total external debt leads to an increase of 1.24% in inflation. Helmy (2022) and Sharaf et al. (2025) also report a significant positive effect of external debt on inflation. Table A3 in Appendices indicates that the findings based on the use of the FMOLS technique are consistent with the ARDL findings. Furthermore, the significant and positive effect of external debt on inflation does not change with the inclusion of the exchange rate variable. Finally, using an alternative proxy of debt (public external debt) in column (5), we also find that debt significantly spurs inflation. Besides, we realize that there is a small difference, in terms of the magnitude, between the coefficient of total external debt (1.243) and that of public external debt (1.151). This confirms the idea that public external debt is more inflationary.
Long- and short-run effects of external debt and imports on inflation
| ARDL | ARDL | ARDL | ARDL | ARDL | ARDL | |
|---|---|---|---|---|---|---|
| Variables | (1,0) | (1,0) | (1,0,2,1,2) | (2,0,2,1,2,2) | (1,0,2,1,2) | (1,0,2,0,0,0) |
| Total external debt (1) | Total external debt (2) | Total external debt (3) | Total external debt (4) | Public external debt (5) | Total external debt (6) | |
| Long run | ||||||
| L1.LnInflation | −0.238*** | −0.306** | −0.820*** | −0.792*** | −0.876*** | −0.949*** |
| (0.093) | (0.119) | (0.188) | (0.173) | (0.186) | (0.195) | |
| L1.LnRatio of debt | 0.539*** | 1.243*** | 1.872*** | 1.151*** | −9.622*** | |
| (0.089) | (0.422) | (0.507) | (0.336) | (2.156) | ||
| L1.LnImports | 0.629*** | 2.062** | 3.692*** | 2.107*** | −10.192*** | |
| (0.078) | (0.774) | (0.916) | (0.683) | (2.198) | ||
| L1.LnBroad money | 5.653*** | 4.595*** | 5.485*** | 4.945*** | ||
| (1.097) | (1.125) | (0.976) | (0.881) | |||
| L1.LnGDP per capita | 7.845*** | 2.265 | 8.243*** | 2.903* | ||
| (2.488) | (1.589) | (2.271) | (1.690) | |||
| L1.LnExchange rate | −1.799*** | |||||
| (0.477) | ||||||
| L1.Ratio of debt*LnImport | 2.727*** | |||||
| (0.557) | ||||||
| Short run | ||||||
| L1.ΔLnRatio of debt | −0.579 | |||||
| (0.481) | ||||||
| ΔLnImports | −1.280* | −3.023*** | −1.337* | |||
| (0.714) | (1.022) | (0.680) | ||||
| L1.ΔLnImports | −1.461** | |||||
| (0.715) | ||||||
| ΔLnBroad money | −3.549*** | −4.005*** | −3.752*** | −3.234*** | ||
| (1.027) | (1.272) | (1.003) | (0.856) | |||
| L1.ΔLnBroad money | −2.770*** | −2.766*** | −2.944*** | −2.768*** | ||
| (0.907) | (0.929) | (0.885) | (0.810) | |||
| ΔLnGDP per capita | −13.853** | −8.795 | −14.687** | |||
| (6.128) | (5.558) | (5.942) | ||||
| L1.ΔLnGDP per capita | −9.296* | −12.478** | −9.717* | |||
| (5.139) | (5.332) | (4.966) | ||||
| ΔLnExchange rate | 2.483*** | |||||
| (0.834) | ||||||
| Constant | −56.003*** | −24.532** | −61.331*** | 11.577 | ||
| (17.882) | (11.56) | (17.739) | (15.595) | |||
| R-squared | 0.122 | 0.122 | 0.656 | 0.511 | 0.680 | 0.679 |
| Normality test: χ2 | 1.270 | 0.760 | 1.670 | 5.810* | 0.510 | 1.090 |
| Ramsey RESET test: F | 0.883 | 0.883 | 0.450 | 1.140 | 0.440 | 0.380 |
| Durbin–Watson test | 1.928 | 1.911 | 2.375 | 2.438 | 2.348 | 2.213 |
| Breusch–Godfrey LM test | 0.041 | 0.091 | 2.545 | 3.114* | 1.855 | 0.800 |
| White’s test: χ2 statistic | 10.35* | 3.530 | 34.000 | 34.000 | 24.980 | 34.000 |
| CUSUM test statistic | 0.521 | 0.521 | 0.366 | 0.391 | 0.227 | 0.735 |
Note(s): *p < 0.1; **p < 0.05; ***p < 0.01. Standard errors in parentheses
Considering the trade liberalization variable, the findings in column (3) highlight a significant and positive correlation between imports and inflation in the long run at the 5% threshold, while imports have a significant and negative effect on inflation in the short run at the 10% threshold. (The findings in column 4 are marginalized because they suffer from the multicollinearity issue.) Ceteris paribus, a 1% increase in imports results in an increase of 2.06% in inflation in the long run. Carluccio et al. (2023) and Jingjing et al. (2025) report a significant and positive effect of imports on inflation. However, Lin et al. (2017) document a significant and negative effect of imports on inflation in SSA countries, and this difference can be explained by the heterogeneous characteristics of countries. The positive effect of imports on inflation is robust, in terms of statistical significance, when we use the FMOLS technique. Performing sensitivity analysis, the findings, not reported here, show that overall trade is positively but insignificantly correlated with inflation. Focusing on other control variables, Table 4 shows a positive effect of broad money and real GDP per capita on inflation at the 1% threshold in linear regressions in the long run. The short-run analysis reveals a significant and negative effect of the first differences of broad money and real GDP per capita and the first-lags of these differences on inflation. Alfaro (2005) reports mixed effects of GDP per capita on inflation in the long run. However, in column (4), exchange rate has a significant and negative effect on inflation in the long run while exerting inflation pressure in the short run at the 1% threshold. We keep in mind that although diagnostic tests validate the findings in column (4) at the 5% threshold, they are marginalized because they suffer from multicollinearity.
For the non-linear regression, we find a significant and negative correlation between external debt and inflation in the long run. Following the interpretation provided by Brambor et al. (2006) and Mamba (2021), this implies that the ME of external debt on inflation is roughly −9.622 when imports take the value zero. Alternatively, the ME of imports on inflation is roughly −10.192 when external debt is zero. The interaction coefficient is significant and positive (2.727) in the long run, suggesting that the ME of external debt on inflation depends on the level of imports. In other words, imports positively moderate the external debt–inflation nexus in the long run, implying that imports play an amplifying role in the relationship between external debt and inflation.
In the long run, we can compute the total ME of external debt on inflation as follows:
Allowing Eq. (5) to be equal to zero, we compute the threshold of the imports above which the ME of external debt on inflation is positive. The threshold value is given by the fraction 9.622/2.727 = 3.528. This means that the threshold of imports above which external debt has a positive effect on inflation is roughly 3.528. Without ln, through exponential transformation (Mamba, 2025), this threshold is 34.056 (% of GDP). Computed threshold is important for practitioners in economic development.
5. Conclusion and policy implications
The research investigates the effects of external debt on inflation and the moderating role of imports in the external debt–inflation nexus in Guinea over the period of 1986–2021. Computed threshold is important for practitioners in economic development. For robustness checks, beyond the ARDL estimator, the research applies the FMOLS technique, which is free from endogeneity issues, small sample size bias and serial correlation. We obtain a statistical significance of the negative speed-of-adjustment coefficient at the 1% threshold, suggesting that there is a long-run equilibrium convergence when there is a shock to any of the regressors. The findings reveal that imports negatively affect inflation in the short run, while external public debt and imports drive long-run inflation. A similar finding is obtained when public external debt is used. Also, the positive effect of external debt and imports on inflation is confirmed when we use the FMLOS technique. Furthermore, imports positively moderate the external debt–inflation nexus in the long run, implying that imports play an amplifying role in the relationship between external debt and inflation.
Relying on the above findings, we can formulate the following policy implications. The Guinean government should strengthen the management of external debt issuance to stabilize inflation and achieve a suitable economic performance. A well-managed external debt contributes to stabilizing the economy. Also, directing external debt toward financing infrastructure, education or healthcare projects can increase the production capacity of the economy, which can stimulate the supply of goods and services and reduce inflationary pressures. Pursuing a more liberal trade policy, especially with tariffs cut, would not be beneficial in reducing inflation rates. Therefore, the Guinean government needs to reduce its reliance on imports by encouraging the consumption of domestic manufacturing goods rather than imported manufacturing goods. Furthermore, the Guinean government needs to keep the degree of its openness under the computed threshold of imports above which external debt has a positive effect on inflation.
Research limitations: Although this research reports interesting findings, the small sample size represents a structural limitation of our findings. Another limitation is data quality, a common problem in research.
Future research: Future research can analyze the direct and combined effects of external debt and trade on economic growth or sectoral growth.
The supplementary material for this article can be found online.
