Exchange Rate Pass-Through to Domestic Prices: Evidence Analysis of a Periphery Country

1 Introduction

The exchange rate is considered to be among the main determinants for improving economic activities and lowering prices. In other words, the deterioration of any national currency will cause an appreciation in importation prices leading to their increase domestically. The countries of The Middle East and North Africa regions experienced numerous political and social changes in the aftermath of the 2011-Jasmine Revolution. This is due to the fact that governments should implement policies contributing to achieving economic inclusion and support marginalized social groups (Elnabawy & Abonazel, 2020; Smaili & Ben Aissa, 2018; Turna & Özcan, 2021). Taking into account the implications for economic stability in general as well as the potential effects of the exchange rate on domestic prices (Aguir & Mahaman Boubacar, 2025), it is imperative to quantify the transmission effect of the exchange rate on inflation (exchange rate pass-through [ERPT]) which is, nowadays, a major concern for the Central Bank of Tunisia (BCT) and the governors.

Since the Revolution, Tunisia has experienced deep changes at the economic and political levels. In fact, inflation has become growingly unstable (Trabelsi & Ben Khaled, 2023). This is due to uncertainties arising from the political transition, causing the economic activities to destabilize and costs of goods and services prices to increase (Mimoun et al., 2024). To boost economic growth, frequently funded through debts, the government has expanded public expenditure. Therefore; additional pressure on prices has been exerted causing social pressures and wages to rise in the public sector as a result of increased demand leading to higher prices (Karshenas et al., 2014). Regardless of nominal wage growth, high inflation has resulted in diminishing workers’ purchasing power outcoming from real wages diminishing.

The Tunisian currency has significantly depreciated as an aftermath of declining foreign exchange reserves, contracting exports, and reduced investments. Subsequently, import costs increased leading to further inflation (Souffargi & Boubaker, 2024). High unemployment, notably among young people and graduates, restricts households’ capacity raising their earnings, further worsening their purchasing power decline. The increasing costs of basic goods, including food and energy, push households to reduce their expenditures in other key domains, further worsening their economic hardship (Chiraz & Frioui, 2014).

In order to stabilize the currency, the BCT has raised interest rates as well as intervened in the foreign exchange market although their limited success because of persistent economic burdens (Alsamara et al., 2020; Neaime & Gaysset, 2022). The government has initiated reforms to strengthen economic competitiveness and encourage foreign investment. Nevertheless, the execution of these reforms has been hampered by social unrest and political opposition. (Abid et al., 2014; Nabi, 2021). In fact, Tunisia’ post-revolutionary period has been characterized by major macroeconomic challenges, especially related to inflation, purchasing power, and exchange rate (Abdessalem & Chekki Cherni, 2016; Mbazia, 2017; Nabi, 2021; Ouhibi, 2024).

Tunisia is profoundly influenced by such a dynamic. Therefore, the Tunisian Central Bank’s actions are urged to combat price rise to ensure economic affordability and, thus, set price stability as an explicit objective of the monetary policy, as in an inflation targeting framework (Smaili & Ben Aissa, 2018).

Monetary authorities react to currency fluctuations insofar as they affect consumer prices and then inflation (Ha et al., 2020). This requires studies related to not only the origin of currency movements, but also its economic characteristics in emerging and developing economies (Aguir et al., 2017) where large currency movements are more frequent (Alsamara et al., 2020; Ha et al., 2020; Luong & Xu, 2020).

The ERPT can be reflected in consumer prices through direct and indirect channels (Allegret et al., 2011; Laflèche, 1997; Pain et al., 2008; Villemot et al., 2018). ERPT shapes domestic pricing through variations in imported goods costs (Nagengast et al., 2021). Thus, a decline in import prices is at the origin of a currency appreciation. Likewise, a rise in the prices of imported products following a reduction in currency strength has a significant impact on consumer prices (Aguir, 2018; Ftiti et al., 2017). Yet, this leads to an increase in the marginal cost for producers as well as in the prices of domestic goods (Aguir & Smida, 2015).

Unlike the indirect effect of the exchange rate on domestic prices appearing during currency depreciation, any increase resulting in exports or aggregate demand implies domestic prices increase. This leads, indirectly, to a fall in the unemployment rate because of requiring more labor force (Bouoiyour et al., 2004; Bouveret & Ducoudré, 2008; McCarthy, 2007).

De Bandt et al. (2008) examined the short- and long-term effects of exchange rates on import prices in the Eurozone using cointegration analysis on panel data. Their findings revealed a long-term relationship between exchange rates and import prices during periods of Euro appreciation. Similarly, Bouakez and Rebei (2008) aimed to estimate the degree of ERPT in Canada using a structural dynamic general equilibrium model. They found that whereas the impact of exchange rate fluctuations on import prices has remained steady, its impact on consumer prices has declined over the past few years in relation to a shift in the monetary policy regime.

An and Wang (2012) conducted an analysis to estimate the pass-through of the exchange rate to import prices, production, and consumption, using an structural vector autoregression (SVAR) model for nine OECD countries. They found that the pass-through exchange rate is inferior to the short- and long-term units.

Therefore, for the case of Thailand, using a standard model of Engle and Grager, during the period from 2000 to 2011, Wattanakoon (2011) observed that a change in the exchange rate had an incomplete impact on inflation. He found that the low degree of pass-through varies from 0.02% in the short term to 0.4% in the long term. This can be explained by the non-accommodative government response to inflation aiming to uphold spending power.

Despite the results of these studies, many economists remain cautious about the exchange rate and inflation relationship. Indeed, Lariau et al. (2016) examined the pass-through in Angola and Nigeria. They come across a high degree of pass-through for Angola in the long term. However, for Nigeria, there was no stable relationship between the exchange rate and prices in the long term. Concerning Tunisia, exchange rate changes are of particular importance (Akpan and Akpan, 2012).

In fact, Graph 1 shows stability between the change in the nominal effective exchange rate (NEER), the real effective exchange rate (REER), and the consumer price index (CPI). However, the post-Revolution period seems to be specific given the events that have imposed a sharp rise in the inflation rate and volatility toward lower exchange rates.

Graph 1

NEER, REER, and CPI in Tunisia.

In comparison with a neighboring country: In January 2018, Morocco implemented a more flexible exchange rate system by expanding the fluctuation rate of its currency from ±0.3 to ±2.5% around a central parity (El Aboudi et al., 2023). This initiative sought to reinforce the adaptability of the Moroccan economy to external shocks and boost the competitive advantage of its exports (Pouya et al., 2024). Since the implementation of a more adaptable framework in 2018, the dirham has undergone greater volatility. However, by intervening in the foreign exchange market, the Central Bank has successfully upheld relative stability (Uche et al., 2023). Empirical evidence demonstrates that exchange rate variations only impact domestic prices in Morocco partially. Based on various studies, a 1% depreciation of the dirham leads to a 0.2–0.4% increase in consumer prices (El Hamidi & Karboub, 2023; Lezar, 2023; Oumansour & Azghour, 2024). Still, the degree of pass-through may vary due to different aspects such as the import structure, monetary policies, and international market conditions (Erraitab et al., 2024).

In this study, the empirical estimate of ERPT is presented using monthly data from Tunisia, a country experiencing recurring financial downturns. On the one hand, the country’s sharp fall times frequently coincide with high inflation. This significant connection between currency value and local prices makes it possible to quantify the extent of ERPT.

This article estimates the transmission effect of the exchange rate and import prices on inflation in Tunisia, focusing on the evolution of the latter over time, given the significance of the inflationary trends and monetary governance. We use an autoregressive distributed lag (ARDL) approach that covers the period from m1-2000 to m12-2017. This model has several advantages. First, it deals with the problems of small samples. Then, it does not require the same order of integration.

Although there are researchers estimating the ERPT for Tunisia (Charfi & Kadria, 2016), the key contribution of this article stems from employing the ARDL model and analyzing the impact of the exchange rate as well as import prices on the inflationary environment in Tunisia. The following sections of the article are arranged as follows. Section 2 presents a summary of the pertinent empirical studies. Section 3 outlines the data used and the methodology applied in our empirical analysis. Section 4 details the findings and their interpretations, and the article concludes in Section 5.

2 A Review of the Empirical Literature

Researchers are still interested in analyzing ERPT. In fact, various approaches have been developed for investigating ERPT, at the microeconomic level under different assumptions. Thereby, incentives that aim to protect profits by affecting exchange rates and selling prices are the main causes of ERPT. With this, the pass-through of the exchange rate is based on the structure and state of the national economy such as consumers optimizing their utility by choosing between domestic and imported goods (Krugman, 1987; Obstfeld & Rogoff, 1995).

If we focus on the macroeconomic level, the magnitude of ERPT is directly related to inflation. According to Taylor (2000), any country with minimal inflation tends to find a lower ERPT. Consequently, companies anticipate their future costs in advance and set prices, which ultimately get passed on to expected inflation. If the monetary policy realibility is determined by low expectations of the inflation rate, a stable monetary policy can have a crucial impact on the development of the ERPT (Boubaker & Mouna, 2024; Cooper, 2019; Gagnon & Ihrig, 2004; McCarthy, 2007). A low ERPT can strengthen monetary policy independence and contribute to stabilizing output and inflation (Mishkin, 2009). Nevertheless, weak ERPT could diminish the efficiency of the exchange rate pathway (Jasova et al., 2016). This is often adopted into New Keynesian monetary policy models, where the central bank’s response mechanism considers both the exchange rate gap and the output gap.

The ERPT depends directly or indirectly on the soundness of the Law of One Price, which is influenced by broad structural as well as managerial factors. Frankel et al. (2012) state that any incomplete ERPT theory imperatively begins with the reasons why The Law of One Price fails. In fact, possible obstacles include transportation costs, trade barriers, and distribution as well as retail costs.

Numerous studies analyzed ERPT, focusing on both developed economies (e.g., Gagnon & Ihrig, 2004; McCarthy, 2007; Özyurt, 2016; Taylor, 2000) and developing economies (e.g., Akofio-Sowah, 2009; Choudhri & Hakura, 2006; Lariau et al., 2016; Razafimahefa, 2012; Rodriguez et al., 2024). One of the common findings in these studies is that exchange rate fluctuations limited the impact on inflation via import prices. Additionally, researchers have identified a low ERPT in both developing countries (Lariau et al., 2016; Razafimahefa, 2012) and developed countries (Özyurt, 2016; Taylor, 2000).

The methodologies used in these works generally fall into two categories: single-equation approaches and vector autoregressive (VAR) models. Findings from previous research suggest that a low-inflation environment, which contributes to reduced ERPT in both developed and developing economies, can be explained by the credibility of monetary policies that effectively ensure steady inflation expectations (Gagnon & Ihrig, 2004; McCarthy, 2007; Özyurt, 2016; Taylor, 2000).

By combining two macro and micro economic models, Taylor (2000) studied ERPT in the United States and found that pass-through depreciates when expectations of continued depreciation decline. We also find, according to Taylor, that the prices are fixed by the companies in advance after an anticipation of future costs.

Gagnon and Ihrig (2004) analyzed ERPT across 20 industrialized countries and revealed that low-inflation economies generally show weaker ERPT in comparison with those experiencing higher inflation. Their findings also highlight the significant role of monetary policy in reducing ERPT, attributing this effect to the stabilization of inflation levels. In addition, they noticed that the credibility of the central bank and the fight against inflation lead to finding agents less willing to raising prices due to cost increases (De Mendonça and Tiberto, 2017).

In Tunisia, research on ERPT was developed by Charfi and Kadria (2016) and has mainly focused on two methods which have been carried out: the VAR model and the SVAR model. They showed that the transmission of ERPT is incomplete for the price chain, reaching its lowest level in the CPI.

3 Empirical Methodology

The empirical work is based on monthly data for the period 2000–2020. The NEER is accessible from the database of the BCT. In fact, it specified that an upward variation in the NEER reflects depreciation in the Tunisian dinar against the weighted average exchange rates of its main trading partners. The CPI is sourced from the INS database. In addition to the above variables, we have used import prices as a control variable in line with the majority of empirical studies. The import price index noted (PM) is also extracted from the database of the BCT and measured by fluctuations in the cost of globally sourced goods and services consumed by Tunisians.

Before selecting the appropriate econometric model, we examined the stationarity of the time series for each variable using the augmented Phillips–Perron (PP) and Dickey–Fuller (ADF) tests, with the results presented in Table 1. Since all variables have been stationary with an integration order below 2, the Johansen cointegration method is not suitable. Consequently, we employ the ARDL model introduced by Pesaran et al. (2001) and subsequently utilize the vector error correction model (VECM) to assess the robustness of our findings. The ARDL approach developed by Pesaran et al. along with the VECM technique provides significant benefits. Including conducting causality tests, the study employs both the ARDL approach and the VECM Granger causality technique to carefully study the temporal relationships between variables. This approach reinforces the accuracy and thoroughness of the analysis.

The impact of the exchange rate on the inflation rate is analyzed by applying the cointegration method. Different tests provide a means for analyzing the existence of a cointegration relationship between the variables in an econometric model. To perform this, we have chosen to work with the ARDL model thanks to its reliability and better statistical properties in small samples. Also, according to (Nkoro & Uko, 2016; Pesaran et al., 2001), it takes into account asymmetrical impacts in both the short and long term. In addition, in this model, the lagged features help reduce the risk of endogeneity resulting from reverse causality. However, the cointegration test by ARDL, proposed by Pesaran et al. (1999, 2001), is widely used in research works. We picked this method for its efficiency in studies with limited sample sizes and its applicability to series that are either stationary at level (order 0), integrated at order 1, or a mix of both, contrary to traditional cointegration techniques as those proposed by Engle and Granger (1987), Johansen (1988), or Johansen and Juselius (1990). Yet, this approach becomes unsuitable when the integration order of the series exceeds 1. Another key advantage of this method is its ability to estimate both long-term and short-term dynamics within a single econometric framework, as emphasized by Akpan and Akpan (2012). As informed by the existing literature, our ERPT ARDL model specification lies on the purchasing power parity hypothesis. Therefore, the long-run equation, being expressed in logarithmic form, is presented below:

Here, CPI is the domestic consumer price index of Tunisia, NEER is the nominal effective exchange rate of Tunisia, and PM is the import price index of Tunisia (Dardouri et al., 2024). φ 1 to φ 3 represent the elasticities of the independent variables, φ 0 stands for the constant, ln denotes the natural logarithm, T is the value of the Trend, t is the time, and finally, ε is the term error.

Now, we will present the error correction equation of the ARDL model according to Pesaran et al. (2001); the representation is given as follows:

Here Δ represents the first difference operator, β 0 stands for the constant, β _i , γ _i as well as θ _i are the short-term elasticities. Moreover, α ₁, α ₂, and α ₃ are the model long-term dynamics, ε ∼ iid (0, ϭ) is the error term, and p, q with k representing the lags.

However, following a change in one of its determinants, the dependent variable in equation (1) is unlikely to be immediately reaching its long-term equilibrium level.

Thus, according to Pesaran et al. (2001), the rate of adjustment between the short- and long-term levels can be captured by estimating the error correction version of our ARDL (p, q, r) model shown in the next equation (3) which calculates the long-run elasticites:

Here, ∆ represents the first difference operator, u _{t – 1} represents the error correction term (ECT) of our model, θ is the error correction coefficient related to the speed of the imbalance adjustment between the long and short term of the dependent variable, and p, q, r, s, w, and k are the respective lag lengths. We predict a significant negative sign for the ECT (Gujarati, 2003).

To verify a cointegration relationship, the preliminary step entails identifying the integration order for each variable. This is accomplished through the ADF and PP tests performed under different specifications for assessing whether the series exhibits stationary at the level or after differencing. In these tests, the null hypothesis assumes non-stationarity, whereas the alternative hypothesis suggests stationarity. In the next phase, we employ the Bounds test aimed to investigate the existence of a cointegration relationship. In fact, the approach is mainly based on the Wald F-statistic, where the null hypothesis asserts that there is no cointegrating relationship. The Bounds test undertakes estimating model (1) with reference to the ordinary least squares method. Following this, the F-test is used to evaluate the long-term coefficients joint nullity, which leads to considering the following two hypotheses:

H ₀: α ₁₁ = α ₂₁ = α ₃₁ = 0 (null, as opposed to the alternative hypothesis, there is no evidence of a long-term relationship).

H _A: α ₁₁ ≠ α ₂₁ ≠ α ₃₁ ≠ 0 (alternative with a long-term relationship).

To end with, the third step involves comparing the F statistics calculated with the critical values. Indeed, Pesaran et al. (2001) once estimated the cointegration equations to enable us to derive the long-term elasticities. In the absence of a detected cointegration relationship, the short-term causal relationship is evaluated. Once the model is specified, the general approach includes a series of diagnostic tests, such as to start with (i) normality of residuals (Jarque–Bera normality test); then (ii) serial correlation (Breusch–Godfrey LM test); after that (iii) heteroscedasticity (ARCH test); and finally, (iv) model specification (Ramsey’s regression error specification test [RESET]). These steps are supported by both the cumulative sum (CUSUM) and cumulative sum of squares (CUSUM Square) tests for assessing the model stability. The subsequent section will present and analyze the findings.

To end with, the last procedure incorporates the deployment of a restriction of VAR, the VECM technique for exploring the short-run causality among variables. The ECT acts as the cointegration term, indicating the rate at which endogenous variables adjust toward their long-term equilibrium. The presence of a significant negative ECT coefficient is a key factor that allows short-run dynamic realignments around equilibrium.

Narayan and Smyth introduced the VAR methodology, which is typically as follows:

where β 0 , β i , γ i , θ _i , and φ are factors, and ECT is the error correction term.

4 Empirical Results and Discussions

4.1 Stationary Test and Stability of Variables

In advance of analyzing these variables using Pesaran et al.’s ARDL approach (2001), we noticed that the findings in Table 1 reveal the integration in the first difference I (1) of all the variables, confirming that all of them have an order of integration below 2.

Once establishing the stationarity of the series, it is necessary to assess whether a linear connection between the variables is confirmed. Eviews 10.0 software was used to apply Ramsey’s RESET test and verify the linearity of the series’ structure. Based on the study’s findings, the analyzed series exhibited linearity. The fundamental assumption of Ramsey’s RESET linearity test confirmed the linearity of the series Ramsay and Dalzell (1991); Ramsay and Silverman (1997, 2002, 2005). Hence, the hypothesis of linearity is confirmed as shown in Table 2.

Table 2

Ramsey’s RESET test

	Value	Df	Probability
t-statistic	1.004270	206	0.3164
F-statistic	1.008559	(1, 206)	0.3164

Before proceeding with the ARDL modeling and to emphasize the short- and long-term relations related to Tunisia, we aim to present the time-based development of the variables (Harvey and Leybourne, 2007; Harvey et al., 2008; Hibou, 2011). This will allow more precise identification of simultaneous peaks in the indicators (Graph 2).

Graph 2

The trend of variables.

The analysis of the graphs given indicates long-term stability of the variables. Nonetheless, in the aftermath of the Tunisian Revolution, more specifically in 2012, increased exchange rate volatility marked by peaks and subsequent troughs was observed. This period was notably defined by significant political instability following the Revolution. The graph clearly reveals a substantial impact of ERPT on changes in consumer and import prices. More precisely, a decline in key exchange rates is subsequent to a pronounced increase in both the CPI and import prices.

Stationary test outcomes lead us to examine the relationship between the foreign exchange market and inflation fluctuations through cointegration tests associated with the ARDL approach.

4.5 Robustness Check

Table 7 presents the diagnostic test outcomes in relation to the chosen ARDL model (1, 1, 0, 1). The Jarque–Bera normality test reveals that the residuals follow a normal distribution pattern. The heteroscedasticity test results show that the F-statistic support confirms the null hypothesis of no serial correlation, indicating that there is no evidence related to the heteroscedasticity presence in the residuals.

Table 7

Diagnostic test

χ 2 (serial correlation)1	0.354222 (0.7022)
χ 2 (functional form)2	2.411175 (0.1221)
χ 2 (normality)3	9.220203 (0.009951)
χ 2 (heteroscedasticity)4	1.623925 (0.0758)

¹The Breusch–Godfrey LM test statistic for no serial correlation.

²The Ramsey′s Reset test statistic for regression specification error.

³The Jarque–Bera statistic for normality.

⁴The Breusch–Pagan–Godfrey test statistic for heteroscedasticity.

An essential econometric requirement for an ARDL model is to test parameter consistency, related to the analysis of diagnostic tests. With this aim, CUSUM and CUSUM Square tests are applied to investigate the stability of both short-term and long-term coefficients in the ARDL model. These tests are conducted using recursive residuals obtained from the estimated ARDL model applied in this study (Brown et al., 1975) having their results displayed in Table 7.

It is evident that the CUSUM and CUSUM Square statistics remain largely within the critical threshold at the 5% significance level. Thus, the robustness and stability of the estimated coefficients in the ARDL cointegration model (11, 0, 1) are confirmed.

5 Conclusion

In this study, the role of the ERPT in inflation has been empirically assessed. More specifically, the impact of major exchange rates on price change has been examined. Using a time series of monthly data for the period from January 2000 to December 2020, the ARDL model and VECM are estimated using the terminal test technique for determining the impact of the ERPT on the CPI. This methodology enables us to analyze the short-term dynamics and the long-term effects of the explained variables along with the adjustment speed of the explained variable to the long-term equilibrium trajectory.

To start with, our econometric results indicate that the fluctuation of the NEER impacts the consumer price change in the short and long term.

Second, the findings show that the NEER is crucial for determining long-term inflation, whereas the short-term impact on the nominal exchange rate forecasts is not significant.

These results have several policy implications. One reason is that Tunisia should detain significant foreign exchange reserves to ensure exchange rate stability. Given the opportunity cost of holding a large quantity of foreign reserves, the BCT needs to consider other mechanisms to achieve its primary goal of maintaining price stability.

Also, the results that we found are relevant to Tunisia’s macroeconomic policies. The high overall scale of the ERPT and the speed of adjustment represent a significant and difficult challenge for Tunisian policy. Attempting to control the ERPT to inflation by targeting the change in the exchange rate will be a risky and inevitably unsustainable strategy for a given stock of international reserves. In the case of Tunisia, a floating exchange rate regime with the credibility of monetary policy within the BCT could prove its effectiveness in anchoring inflation expectations.

Tunisian authorities ought to implement measures for mitigating the impact of fluctuations in the NEER on domestic price levels. One idea is to diversify the country’s export base in order to reduce its dependency on a limited number of industries vulnerable to currency fluctuations. This could help stabilize the exchange rate and reduce the impact on domestic prices. In addition, Tunisian authorities could investigate options, such as future contracts as financial instruments, against currency risks. Enhancing both monetary and fiscal policies, coupled with better transparency and accountability in economic decisions, would act as a safeguard against exchange rate fluctuations on domestic prices. Overall, a multi-pronged approach that addresses both macroeconomic factors and specific industry vulnerabilities is crucial to regulate the influence of the NEER on domestic prices in Tunisia.