II. The Exchange Rate Pass-Through in Tunisia1
1. Given the high degree of openness of the Tunisian economy, the exchange rate channel of monetary policy plays a central role in the transmission mechanism of monetary policy, especially as Tunisia moves closer to a free floating exchange rate system and adopts an inflation targeting framework. Understanding the transmission channels of monetary policy is key for the conduct of an efficient monetary policy. This is particularly true within an inflation targeting framework where policy relies heavily on model-based forecasting of inflation for monetary policy formulation.
2. Therefore, it is crucial to assess the exchange rate pass-through given that the degree of consumer price responsiveness to exchange rate changes has important implications for monetary policy. Exchange rate pass-through to domestic prices measures the extent to which fluctuations in the nominal exchange rate affect consumer prices through the changes in the prices of imported goods.2 Consumer prices are affected directly by the change in the prices of imported finished and intermediate goods, but also indirectly through the effects of exchange rate movements on aggregate demand. For instance, an exchange rate depreciation affects net exports, which in turn influences domestic prices through aggregate demand, putting upward pressure on domestic prices. However, consumer prices may not be very responsive to the exchange rate when variations in imports costs are absorbed by intermediaries in the distribution channel. Therefore, the degree of exchange rate pass-through will depend, among other things, on the competitive nature of the domestic market for importable goods and pricing-to-market behavior.
3. The aim of this paper is to econometrically estimate the degree of exchange rate pass-through in Tunisia. The methodology relies on time series and panel estimation methods, using both monthly and quarterly data for a basket of 43 consumption products during 1995-2006.
4. The paper is organized as follows. Section B discusses the determinants of exchange rate pass-through and summarizes the results of some empirical studies on developing and emerging economies. Section C describes the fluctuations in the nominal effective exchange rate in Tunisia, the underlying factors, and the resulting effects on inflation. Section D presents the econometric methodology and the estimation results. Section E concludes and offers some policy recommendations.
B. Theory and Empirical Studies on Exchange Rate Pass-Through
5. The law of one price posits that the exchange rate pass-through to import prices (in domestic currency) is always immediate and complete. However, absolute purchasing power parity (PPP) theory has been generally rejected statistically. Several theoretical arguments have been put forward to explain partial or incomplete exchange rate pass-through.
6. The “pricing-to-market” theory argues that pass-through could be incomplete because foreign producers may accept lower profit margins to preserve their market share. Although an exchange rate depreciation increases the cost of imported intermediate goods, imperfectly competitive firms may choose to totally or partially absorb the increase in production cost, reducing the pass-through to consumer prices. Moreover, distribution services provide some insulation for consumer prices of traded goods, as they dilute the import content of final consumption goods and as distributors may actively adjust profit margins to absorb currency fluctuations (Campas and Goldberg, 2006).
7. Taylor (2000) argues that the pass-through rises with the level of inflation. In a model of firm behavior based on staggered price setting and monopolistic competition, the author shows that a credible low inflation regime leads to a lower pass-through, or conversely the persistence of high inflation is positively correlated with the level of pass-through. The underlining assumption is that improved credibility and effectiveness of monetary policy in maintaining a low inflation rate will lower the pass-through as firms would expect a change in costs or prices to be less persistent. Consequently, they are less prone to change prices in response to a given exchange rate shock because they expect the monetary authority to act strongly to stabilize domestic inflation.
8. Other factors such as nominal rigidities and slow adjustment of good prices could make domestic prices less responsive to exchange rate movements. In the later case, the pass-through is delayed but might not be incomplete.
9. Whether partial or complete, exchange rate pass-through is an important factor that determines the effectiveness of exchange rate adjustments in achieving or maintaining a sustainable external balance. Expenditure switching policies aimed at addressing balance of payments difficulties rely on the role of the exchange rate in shifting consumer purchase to domestically produced goods from imported goods. If the degree of pass-through is high, the required exchange rate adjustement needed to correct an unsustainable current account position will be relatively small. The converse holds if the pass-through is low.
10. Under a flexible exchange rate regime, a low exchange rate pass-through could help stabilize both output and inflation. Devereux (2001) shows that in a small open economy with high exchange rate pass-through, there is a significant trade-off between output volatility and inflation volatility. But with limited or delayed pass-through, this tradeoff is much less pronounced. A flexible exchange rate can deliver both lower output variance and lower inflation. Therefore, a low exchange rate pass-through could provide greater flexibility in the pursuit of an independent monetary policy and make it easier to implement inflation targeting (Choudhri and Hakura, 2001).
A brief empirical literature review
11. The relationship between exchange rates and good prices has been extensively analyzed. But only a few studies pertain to developing and emerging countries. These studies can be split into two broad categories: country-specific and cross-country studies.
12. Most country-specific studies find low exchange rate pass-through. Among these, Mwase (2006) uses structural vector autoregression models to provide evidence of an incomplete pass-through to inflation in Tanzania. Furthermore, the degree of pass-through has decreased over time, possibly because of declining inflation since the mid-1990s, higher productivity, and increased competition. However, Khundrakpam (2007) does not find evidence of a decline in exchange rate pass-trough to domestic prices in India, despite a lower inflationary environment. Bhundia (2002) also finds a low pass-through in South Africa as exchange rate fluctuations are absorbed at the intermediate stage of production.
13. Cross-country studies provide a strong support for Taylor (2000)’s theory. Choudhri and Hakura (2001) find that the pass-through is positively correlated with the inflation rate across a large sample of countries.4 The authors use quarterly data over the period 1979-2000 and provide evidence for incomplete pass-through in most countries, including Tunisia. In line with this study, Devereux and Yetman (2003) show that estimated pass-through are positively associated with average inflation rates, but the relationship is nonlinear. Pass-through rises with inflation but at a declining rate. While Choudhri and Hakura’s (2001) results suggest that inflation dominates other macroeconomic variables in explaining cross-country differences in pass-through, Goldfajn and Werlang (2000) find, in a sample of 71 countries including Tunisia, that the real exchange rate misalignment is the most important variable explaining exchange rate pass-through for emerging countries, whereas it is the initial inflation for developed countries. Aside from inflation and exchange rate misalignment, factors such as per capita income, tariffs, wages, long-term exchange rate variability, and trade openness have been also evidenced as determinants of exchange rate pass-through (see Goldfajn and Werlang, 2000; Frankel, Parsley, and Wei, 2005).
C. Exchange Rate and Inflation in Tunisia
14. During the 1990s, Tunisia adopted a real effective exchange rate (REER) targeting policy aimed at preserving the competitiveness of the country. This policy, consisting in adjusting periodically the nominal exchange rate so as to maintain the REER constant, proved to be fairly successful as the country avoided the pitfalls of its REER targeting, which generally are a persistently high inflation and an exchange rate misalignment.5 The absence of significant shocks, combined with a prudent macroeconomic policy mix and price rigidities, were the main factors underlying the relatively low inflation (5.1 percent on average over 1990-99) and the absence of significant exchange rate misalignment the country had experienced while pursuing the REER targeting policy.
15. Since 2000, the central bank has implemented a more flexible exchange rate policy and adopted broad money as nominal anchor. The nominal effective exchange rate has depreciated by about 20 percent since 2000 (Figure II.1). Two factors help explain the gradual depreciation of the dinar. First, the depreciation reflects a series of negative shocks (a series of severe droughts and the events of September 2001 in the U.S. and 2002 in Djerba). Second, the euro has strengthen rapidly against the dollar, which contributed to a depreciation of the Tunisian currency, since the euro is estimated to account for at least two-thirds of the dinar’s currency basket. Although the dinar has depreciated significantly, the latest estimates suggest that the REER appears in line with economics fundamentals. At the same time, inflation has been subdued, averaging only 2.7 percent over 2000-06 compared with 5.1 percent during the 1990s.
Figure II.1.Trends in Nominal Effective Exchange Rate (NEER), 1990–2006
Source: IMF (2006).
D. The Degree of Exchange Rate Pass-Through to Consumer Prices
16. The following log-linear regression specification captures the dynamic relationship between the consumer price inflation (henceforth, referenced simply as inflation) and year-on-year changes in the nominal effective exchange rate while controlling for other factors affecting inflation:
Where the index t refers to time, cpit is the log of the aggregate consumer price index, neert is the log of the nominal effective exchange rate defined as foreign currencies per unit of domestic currency (therefore an increase in the index implies a nominal appreciation), Xt is a vector of control variables—including the log of the monetary aggregate M4 and a set of dummy variables capturing the changes in administered food and fuel prices—and εt is the error term.
17. An important variable that is missing in the set of control variables Xt is the import price index, which is not available on a monthly basis but is available on a quarterly basis. Given the relatively small number of quarterly observations in the sample, the model with import prices as explanatory variable is estimated using panel data for the 43 groups of goods and services included in the Tunisian CPI. The corresponding model for panel data can be written as:
Where indexes i and t refer to the cross-sectional and time dimensions, cpiit is the log of the consumer price index for group of goods and services i at time t, neert has the same definition as in equation (1). Note however that this variable has no cross-sectional index i because it only varies across time given that the effective exchange rate is the same for all groups of goods and services. Yit is a vector of control variables—including the log of the monetary aggregate M4 and import price PMt—, ui is a good-specific fixed effect, and εit is the error term. The dummy variables for administered prices are not included given that the model includes fixed effects.
18. Equations (1) and (2) were estimated by Ordinary Least Square (OLS) using respectively monthly and quarterly data from 1995 to 2006.6 To avoid the problem of spurious regression, all the variables entering equations (1) and (2) have been tested for unit roots. Augmented Dickey-Fuller and Phillips-Perron unit-root tests, performed on all variables entering the two equations, suggest that all variables are I(1) in levels, and therefore I(0) in first difference (Appendix II.1) In other words, all variables entering equations (1) and (2) are stationary in first differences. To eliminate seasonality the Δ operator refers to year-on-year percentage changes.
19. The appropriate lag structure (i.e., the parameters Mand N) is determined by using the Bayesian Information Criteria (BIC). The lags structures of inflation and nominal effective exchange rate year-on-year changes capture inflation persistence and allow for a parsimonious but still flexible parameterization of the dynamics of the pass-through of exchange rate fluctuations to consumer prices. The main variable of interest is the long-run exchange rate pass-through given by the formula below:
In addition to the long-run exchange rate pass-through, impulse response functions will also be used to trace out the detailed dynamics of the exchange rate pass-through.
Results from times-series data
20. Overall, the model has a good fit (Table II.1 and Figure II.2). All variables have the expected sign and are generally statistically significant at least at the 5 percent significant level. As shown by the adjusted R2, the model has a good fit, explaining more than 90 percent of the variability in the aggregate inflation rate. The Durbin-Watson test suggests that the absence of serially correlated errors could not be rejected, as a result, standard errors are corrected for autocorrelation as well as heteroscedasticity.
|Food price dummy||0.002||0.002|
|Fuel price dummy||0.001||0.001|
|Total effect of NEER change||-0.008||-0.008||-0.010||-0.010|
|Long run pass-through||0.065||0.075||0.089||0.093|
|Long run impact of M4||0.165||0.170||0.177||0.179|
|Root Mean Square Error||0.003||0.003||0.003||0.003|
|Durbin-Watson test (prob)||0.70||0.53||0.84||0.69|Figure II.2.Quality of the Prediction: Actual Versus Fitted Inflation Rate
Source: Authors’ calculations.
21. Estimates of the long-run pass-through imply a 10 percent nominal depreciation of the dinar that translates into an increase in inflation in the 0.7-0.9 percentage point range, depending on model specification. Equation (1) in Table II.1 shows that a 10 percent depreciation in the NEER will increase inflation by 0.65 percentage points. Although the estimated pass-through is significant, it could be even higher if the price of some imported goods was liberalized as administrated prices account for almost a third of the CPI basket. Equations (2)-(4) control for the administered nature of the price of food and fuel by including two dummy variables. Food and fuel are two important goods with heavy import content that are included in the CPI basket. The food price (respectively fuel prices) dummy variable takes 1 when food prices (respectively fuel prices) increase and 0 otherwise. When controlling for administrated food prices (equation (2), the degree of pass-through increases significantly (by about 15 percent). The pass-through strengthens even more when the dummy variable for fuel prices is included (equation (3)), jumping from 0.65 to 0.89 (a 37 percent increase). Including both dummy variables raises the degree of pass-through to 0.93, which implies a 43 percent increase. As a result, a 10 percent depreciation in NEER is associated with a 0.93 percentage point increase in inflation (equation (4)). Interestingly, this long-run pass-through estimate for Tunisia is very close to the one found in Choudhri and Hakura (2001) using quarterly data over 1979-2000. According to the authors’ findings, the long run pass-through was 0.09 after 4 quarters and 0.10 after 20 quarters.
22. Monetary aggregate M4 is significantly correlated with inflation. The coefficient associated with M4 is positive and significant at 5 percent in all regressions, suggesting that a monetary contraction reduces inflation. Section D of the previous chapter explains why in Tunisia M4 captures the stance of monetary policy better than M3.
23. The impulse response function shows that the pass-through process takes about 18 months to complete (Figure II.3), reflecting the persistence of the inflation process. This is consistent with the results from other studies.
Figure II.3.An Impulse Response Function of a 1 percent Depreciation in the NEER
Source: Authors’ calculations.
Results from panel data
24. This sub-section compares the results from the time-series model with that of the panel data model. As discussed earlier, the time-series model does not include import prices as they are not available on a monthly basis. Import prices are an important explanatory variable in the sense that it controls for the dollar price of imports (i.e., imports prices in foreign currency) when included with the nominal effective exchange rate in the estimated equation. Therefore, equation (2) includes quarterly import prices and is estimated using panel data to increase the sample size. To facilitate the comparison between the time-series model (which excludes import prices) and panel data model, the latter will be estimated with and without import prices. In equation (2) the dependent variable is the price index of each of the 43 CPI components at the two-digit level of desegregation. All variables are measured at the quarterly frequency.
25. Once import prices are controlled for, the exchange rate pass-through strengthens to about 1.2 percentage point increase in inflation for a 10 percent increase in the NEER. The estimation results of the panel model are given in Table II.2. The first column provides the estimation results of equation (2) without import prices and without fixed effects. The estimate of long-run pass-though is 0.072, very close to the benchmark estimate of 0.065 from the time-series model. The estimate of pass-through remains almost unchanged when using fixed effects (0.068). The long-run exchange rate pass-through strengthens when the import index is included—0.118 when equation (2) is estimated with fixed effects and 0.11 when estimated without fixed effects. After controlling for changes in import prices, a 10 percent depreciation in the NEER will translate into an increase in inflation of 1.2 percentage points according to the model without fixed effects and 1.1 percentage points according to the model with fixed effects. The fact that the panel model yields similar results with and without fixed effects implies that it captures relatively well the heterogeneity in inflation dynamic across the 43 categories of products.
A Panel Approach, 1995–2006
|Without fixed effects||With fixed effects|
|Total effect of NEER change||-0.027||-0.048||-0.030||-0.053|
|Long run pass-through||0.072||0.118||0.068||0.110|
|Long run impact of M4||0.129||0.109||0.131||0.107|
26. The estimates from panel data are relatively close to those of the time-series model. It is worth noting that the time-series model without the dummy variables for the administered prices of food and fuel products is closely related to the panel data model without the import price index while the time-series model with these dummy variables corresponds to the panel data model with the import price index among its explanatory variables. The reason is that the dummy variables for food and fuel products capture indirectly the increase in the import prices of these two categories of goods.7 Similarly, controlling for the dollar price of imports indirectly controls for administered prices of imported goods. The estimation results tend to confirm this interpretation. The time-series model without the administered price dummy variables yields an estimate of the exchange rate pass-through of 0.065, which is very close to the estimate of 0.068 from the panel data model without the import price index. Similarly, the time series-model with the dummy variables for the administered price of food and fuel products yields an estimate of the exchange rate pass-though of 0.093 relatively close to the estimate of 0.11 from the panel data model with the import price index. It is important to stress, however, that both the use of dummy variables for the administered price of food and fuel products in the time-series model and the inclusion of the import price index in the panel model do not control fully for the effect of a large share of administered prices in the CPI, especially for the administered price of goods and services domestically produced.
27. The effect of the monetary aggregate M4 on inflation also strengthened and is even more powerful than that of exchange rate. A 10 percent increase in M4 leads to an increase in inflation in the 1.1-1.8 percentage point range, which is higher than that of the effect of the nominal effective exchange rate.
28. Given the large size and the parsimony of the panel data model, it fits the data quite well. The model explains about 50 percent of inflation variations across the 43 groups of goods and services during 1995 to 2006.
29. However, the results presented so far may be subject to endogeneity bias as the causality may also run from inflation to the exchange rate. To deal with endogeneity issues, a Generalized-Method of-Moments (GMM) dynamic panel estimator is used. It also addresses potential biases associated with measurement errors and omitted variables.8
30. The results obtained using the System-GMM estimator are consistent with the previous findings. A 10 percent depreciation of the dinar would increase inflation by about 0.9 percentage points (Table 3, column 1). As expected, the pass-through strengthens to 1.2 percentage points after controlling for changes in import prices. The results are statistically meaningful as the Hansen over-identification test and the second-order autocorrelation test validate the use of the system GMM estimator and the suitability of lagged variables as instruments.
An Instrumental Variable Approach, 1995–2006
|Total effect of NEER change||-0.036||-0.051|
|Long run pass-through||0.087||0.12|
|Long run impact of M4||-0.131||-0.118|
E. Conclusions and Policy Recommendations
31. This paper analyzes the exchange rate pass-though, an important channel of transmission of monetary policy. Time series and panel data models provide close estimates of exchange rate pass-through:
- Combining the results from the time-series and panel models imply that the estimate of exchange rate pass-through is in the 0.09-0.12 range, depending on the estimation method and model specification. This implies that a 10 percent nominal depreciation of the dinar would induce an increase in inflation in the 0.9-1.2 percentage points.
- The estimation results also show that the liberalization of administered prices would increase the degree of pass-through, although it is difficult to precisely quantify by how much.
- The relatively low degree of pass-through in Tunisia is consistent with the finding in previous studies showing that the degree of pass-through in low inflation countries tend to be lower than that in high inflation countries.9
- The estimation results also point to the fact that the actual monetary policy framework, based on broad money targeting, combined with a supporting exchange rate policy can be used effectively to control inflation, including imported inflation.
- While parsimonious, the time-series and panel data models capture fairly well the inflation dynamics in Tunisia, and provide estimates of pass-though that are consistent with the estimates found in the literature. Furthermore, the estimation method controls for potential endogeneity of the explanatory variables, the problem of omitted variables, and measurement errors.
The main policy recommendations are the following:
- It is worth stressing that these econometric estimates of the degree of pass-through as well as the results of the previous chapter are conditional on the current monetary policy framework. It is quite likely that the transmission mechanism will evolve as the monetary framework moves gradually toward inflation targeting. Therefore, it is critical to build the research capacity of the BCT to continue to develop and update its analytical toolkit for the transmission mechanism of monetary policy and for inflation forecasting.
- For monetary policy to effectively control inflation, the share of administered prices, which stands currently at 32 percent, should be reduced significantly through price liberalization to avoid inconsistencies between monetary and fiscal policies. Price liberalization is also consistent with the absence of fiscal dominance required under the inflation targeting framework. While price liberalization may induce a short-term increase in inflation, which would require a monetary policy response, monetary policy will become more flexible and more efficient. Furthermore, price deregulation would dampen fiscal pressures, which should translate into a less inflationary fiscal position.
- While the pass-through is relatively low, it remains significant. Thus, the econometric study highlights the importance of closely coordinating monetary and exchange rate policies. This is particularly important in view of Tunisia’s transition toward inflation targeting.
- As the Tunisian economy continues its integration to the world economy through the liberalization of the capital account, the relatively low pass-through implies that an orderly move to a free-floating exchange rate should improve Tunisia’s inflation-output variability trade-off as research shows.10 In other words, as the capital account opens up the adoption of a more flexible exchange rate could help Tunisia stabilize both output and inflation.
ArellanoM. and S.Bond1991 “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations” Review of Economic Studies Vol. 58 pp. 277–97.
ArellanoM. and O.Bover1995 “Another Look at the Instrumental Variable Estimation of Error-Components Models” Journal of Econometrics Vol. 68 pp. 29–52.
BhundiaA.2002 “An Empirical Investigation of Exchange Rate Pass-Through in South Africa” IMF Working Paper No. 02/165.
BlundellR. and S.Bond1998 “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models” Journal of Econometrics Vol. 87 No. 1 pp. 115–43.
CampaJ. M. and L.Goldberg2006 “Pass Through of Exchange Rates to Consumption Prices: What Has Changed and Why?” Federal Reserve Bank of New York Staff Report No. 261.
ChoudhriE. U. and D.Hakura2001 “Exchange Rate Pass-Through to Domestic Prices: Does the Inflationary Environment Matter?” IMF Working Paper No. 01/194.
DevereuxM.2001 “Monetary Policy, Exchange Rate Flexibility, and Exchange Rate Pass-through” In Revisiting the Case for Flexible Exchange RatesBank of Canada, pp. 47–82.
DevereuxM. and J.Yetman2003 “Price-Setting and Exchange Rate Pass-Through: Theory and Evidence” In Price adjustment and monetary policyBank of Canada, pp. 347–71.
FanizzaD.N.LaframboiseE.MartinR.Sab and I.Karpowicz2002 “Tunisia’s Experience with Real Exchange Rate Targeting and the Transition to a Flexible Exchange Rate Regime” IMF Working Paper No. 01/194.
FrankelJ. A.D.Parsley and S.Wei2005 “Slow Pass-through Around the World: A New Import for Developing Countries?” NBER Working Papers 11199.
GoldfajnI. and S.Werlang2000 “The Pass-through from Depreciation to Inflation: A Panel Study” Department of Economics PUC-RioBrazil.
KhundrakpamJ. K.2007 “Economic Reforms and Exchange Rate Pass-Through to Domestic Prices in India” BIS Working Papers No. 225.
MumtazH.O.Oomen and J.Wang2006 “Exchange Rate Pass-Through into UK Import Prices” Bank of England Working Paper No. 312.
MwaseN.2006 “An Empirical Investigation of the Exchange Rate Pass-Through to Inflation in Tanzania” IMF Working Papers 06/150.
Appendix II.1. Stationarity Tests with Monthly Data
|Level||First difference||1 percent critical value||Number of lags*||Result|
Based on the BIC criterion.
Based on the BIC criterion.
|Level||First difference||1 percent critical value||Number of lags*||Result|
|Without trend||-3.55||-9.51||-3.48||2||I(0) without trend,|
|With trend||-2.57||-9.95||-4.01||and I(1) with trend|
Based on the BIC criterion.
Based on the BIC criterion.
|(year to year change)||Obs||Mean||Std. Dev.||Min||Max|
Monthly data unless indicated.
Monthly data unless indicated.
|CPI||Consumer price index|
|M4||Monetary aggregate M4||Tunisian Authorities|
|PM||Industrial production index|
|NEER||Nominal effective exchange rate expressed as the weighted average of the dinar exchange rate vis-à-vis the currencies of the main partners. An increase in the nominal effective exchange rate implies an appreciation of the dinar.||International Monetary Fund|
Prepared by Abdelhak Senhadji and Kangni Kpodar.
Exchange rate pass-through has been used interchangeably to mean exchange rate pass-through to import prices and to consumer prices. In this paper, exchange rate pass-through will refer to the later unless otherwise indicated.
In this section, exchange rate pass-through refers to exchange rate pass-through to import prices, which in turn impact consumer prices.
Fanizza D, N. Laframboise, E. Martin, R. Sab, and I. Karpowicz, 2002, “Tunisia’s Experience with Real Exchange Rate Targeting and the Transition to a Flexible Exchange Rate Regime,” IMF Working Paper No. WP/02/190.
The sample period was determined by data availability. It also corresponds to the post-stabilization period during which inflation averaged about 3.8 percent.
By definition, the dummy variables for food and fuel products take one when their prices increase and zero otherwise. Given that these increases happen when the import price of these goods increases, these dummy variables indirectly capture increases in their import prices.
The GMM dynamic panel estimator offers two different approaches. The first-differenced GMM, developed by Arellano and Bond (1991), takes the first difference of the equation to remove time invariant specific effects and instruments the right-hand side variables with appropriate lags of the specified variables in levels. However, the efficiency of this approach is weakened when lagged values in levels are weak instruments for the first differenced variables. Arellano and Bover (1995) and subsequently Blundell and Bond (1998) suggested the System GMM estimator which combines the first-differenced GMM with an additional set of equations in levels where the right-hand side variables are instrumented with suitable lagged first differences. Evidence from Monte Carlo simulations showed that the System GMM estimator performs better than the first-differenced GMM estimator.
Choudhri and Hakura (2001) find that the average degree of pass-through for 37 countries with low inflation is 0.16. In that study, Tunisia’s degree of pass-through for the period 1979-2000 was 0.10 compared to 0.29 for Morocco.