Journal Issue
Share
Article

Romania: Selected Issues

Author(s):
International Monetary Fund. European Dept.
Published Date:
March 2015
Share
  • ShareShare
Show Summary Details

Exchange Rate Pass-Through and Inflation Targeting1

Core Questions and Findings

  • How has the degree of exchange rate pass-through to consumer and producer prices in Romania changed since the adoption of inflation targeting? The introduction of inflation targeting has led to a decline in the level and volatility of consumer and producer price inflation, and likely better anchored inflation expectations. The degree of pass-through reduced to below 10 percent for CPI inflation, and to below 25 percent for PPI inflation, compared with a maximum of 45 percent prior to introducing inflation targeting. The reduction in pass-through is typically less prevalent in the lei-euro exchange rate, compared with the lei-dollar exchange rate.

  • What are the main drivers of Romanian inflation dynamics, especially in more recent years with relatively low inflation? The role of the exchange rate in explaining fluctuations in consumer prices has declined. The exchange rate accounts for about 5 percent of fluctuations in CPI inflation since the adoption of inflation targeting, while this number was higher at around 10 to 30 percent prior. For PPI inflation, the share of innovations explained by the dollar-lei exchange rate has declined to about 10 percent, while the importance of the lei-euro exchange rate has been stable. World commodity prices have overtaken the exchange rate as the most important factor in explaining PPI inflation in recent years, after its own innovations, accounting for almost 30 percent of the fluctuations in PPI inflation.

  • What are the potential policy implications from the declined pass-through? The impact of exchange rate pass-through appears to depend on the monetary policy regime and the level of inflation in the economy. The credibility of a low inflation regime, and clear communications from the central bank could help anchor inflation expectations in the economy and reduce in part vulnerabilities of the economy to external shocks. The reduction in the degree of pass-through, combined with a declining share of foreign-currency denominated loans, suggests that Romania could afford to allow more flexibility in its exchange rate, which could in turn lead to a reduction in pass-through. The decline in exchange rate pass-through, if permanent, is a favorable development for sustainable inflation convergence and for meeting the convergence criteria for joining the euro zone in the medium and long run.

A. Introduction and Motivation

1. After long episodes of disinflation, inflation is now close to its target range. Romania experienced elevated levels of inflation in the early 2000s. Consumer and producer price inflation peaked at around 40 percent, and averaged around 25 percent between January 2000 and July 2005 (Figure 1). Since the announcement of moving toward inflation targeting from August 2005, the inflation rate has fallen drastically, to an average of around 5½ percent for both consumer and producer prices (Table 1). In September 2013, CPI entered the National Bank of Romania’s (NBR) target band of 2.5 percent (±1 percent) and fell below the lower bound for most of 2014, partly due to one-off factors, as well as a persistent output gap and falling inflation expectations.

Figure 1.Romania Inflation and Exchange Rates

Source: National Institute of Statistics, Romania; National Bank of Romania; Haver Analytics.

Table 1.The Level and Volatility of Inflation Declined Post Inflation Targeting
PPI inflationCPI inflationLei-dollar ex rateLei-euro ex rate
Full sampleMean1412.633.7
(2000M1–2014M8)Std. Dev.14.612.60.40.7
Pre-ITMean27.924.433.2
(2000M1–2005M7)Std. Dev.15.113.70.40.7
Post ITMean5.45.334
(2005M8–2014M8)Std. Dev.3.72.30.40.4
Source: National Institute of Statistics, Romania; National Bank of Romania; Haver Analytics; IMF staff calculations.

2. The exchange rate has been a key factor for Romania’s monetary policy but the pass-through to inflation has diminished. The exchange rate has been largely flexible with the transition to inflation targeting, and at times, moved significantly contributing to correct external imbalances. At the same time, the exchange rate has continued to play an important role in monetary policy making, including through foreign exchange interventions. During the pre-inflation targeting period, the degree of exchange rate pass-through to inflation was found to be large and relatively fast. It was estimated to have reached at a maximum of 59–72 percent for producer prices and 27–43 percent for consumer prices (see, for example, Gueorguiev, 2003). After nearly a decade of transitioning to inflation targeting, this paper assesses how the pass-through has evolved. The experience of other inflation targeting small open economies suggests that the degree of exchange rate pass-through tend to decline with the introduction of a credible regime of inflation targeting (Floerkemeier, 2013, and Winkelried, 2014). One possible explanation for the declined pass-through is better anchored inflation expectations following the introduction of inflation targeting, with a low and more stable inflation rate leading to reduced pass-through (Choudhri and Hakura, 2006).

Inflation Expectaions Two Years Ahead 1/

(Percent)

1/ Exchange-rate targeting CEE (CEE-ET) includes Bulgaria, Croatia and Lithuania. Inflation targeting CEE (CEE-IT) comprises of Czech Republic, Hungary, Poland and Romania.

Source: Consensus Forecasts.

Box 1.Inflation Targeting Regime in Romania

Inflation targets in Romania are formulated in terms of the annual change in the consumer price index and are set as midpoints within a target band of +/− percentage points. There are two distinct phases in the type and the levels of the inflation targets set by the NBR:

  • 2005–12: the phase of declining inflation targets, set over a 2-year horizon as year-end annual rates, to consolidate the disinflation process and achieve a sustainable annual inflation rate in the medium term;

  • 2013–present: the phase of a flat multi-annual inflation target, an intermediate stage to ensure the transition towards the phase of long-term continuous inflation targeting—in line with the ECB’s quantitative definition of price stability.

Inflation Targeting

(Percent)

Sources: Consensus Forecasts and Haver Analytics.

(Percent)2005200620072008200920102011201220132014
Mid-point target7.55.04.03.83.53.53.03.02.52.5

3. This paper aims to answer two key questions. First, how has the degree of exchange rate pass-through to consumer and producer prices in Romania changed since the adoption of inflation targeting (Section B)? And second, what are the main drivers of Romanian inflation dynamics, especially in more recent years with relatively low inflation (Section C)? To this end, we consider a structural VAR model from January 2000 to August 2014, and investigate the impulse responses and variance decomposition of inflation variables to shocks in the exchange rate. The findings can contribute in shaping the role of the exchange rate in the NBR inflation targeting framework going forward and underpin inflation forecasts.

B. Inflation Targeting: A Regime Shift for the Exchange Rate Pass-Through?

4. To investigate the degree of exchange rate pass-through to inflation, we adopt a structural VAR framework. This is consistent with the existing literature (e.g., McCarthy, 2000, Gueorguiev, 2003, Billmeier and Bonato, 2004, Cozmanca and Manea, 2010). We first conduct the analysis using the full sample, then divide the sample into the pre-inflation-targeting (August 2005) period, and the post-inflation-targeting period, to study possible changes in the pattern of pass-through.2

5. The empirical analysis uses monthly observations of prices, exchange rate, commodity price, output gap, and wages, between January 2000 and August 2014. We consider three exchange rate variables, the lei-dollar exchange rate, the lei-euro exchange rate, and NEER, and two price measures, the CPI and the PPI, capturing inflationary pressures for both consumers and producers. In addition, a commodity price variable is included to capture an exogenous supply shock, while wages are included as a proxy for domestic supply shocks. A demand shock is represented by the output gap, which measures the deviation of industrial production from its trend.

Figure 2.Exchange Rate Pass-Through to CPI and PPI Inflation

Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order). The degree of pass-through is defined as the ratio of the cumulative impulse response in prices to one Cholesky standard deviation shock on the exchange rate.

6. The analysis shows a higher degree of pass-through for PPI than for CPI (Figure 2). In the first twelve months, around 20 to 30 percent of exchange rate is passed to produce prices, with the euro-lei exchange rate having a larger impact than the dollar-lei exchange rate in the full sample (Table 2). For consumer prices, around 15 percent of the impact is felt in the first year, with a slightly higher degree of euro-lei exchange rate transmitted into domestic consumer prices, especially in longer horizon.

Table 2.Estimates of Exchange Rate Pass-Through
US Dollar-Lei Exchange RateEuro-Lei Exchange Rate
PercentFull SamplePre-ITPost ITFull SamplePre-ITPost IT
(2000M1–2014M8)(2000M1–2005M7)(2005M8–2014M8)(2000M1–2014M8)(2000M1–2005M7)(2005M8–2014M8)
MonthsCPIPPICPIPPICPIPPICPIPPICPIPPICPIPPI
10.24.98.79.4-0.14.61.96.65.27.1-1.45.7
35.510.527.733.82.66.19.919.117.722.05.617.0
610.515.034.240.93.96.612.725.419.525.16.718.6
913.918.736.443.44.16.515.228.521.528.86.919.0
1216.121.337.244.44.16.517.431.123.331.66.919.1
1517.723.137.544.84.16.519.233.124.833.76.919.2
1818.824.337.645.04.16.520.734.826.035.26.919.2
2119.625.237.745.04.16.522.036.126.936.56.919.2
2420.125.937.745.14.16.523.037.327.737.56.919.2
Note: Results are based on a structural VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order). The degree of pass-through is defined as the ratio of the cumulative impulse response in prices to one Cholesky standard deviation shock on the exchange rate.

7. The pass-through has reduced sharply since the adoption of inflation targeting, as inflation expectations are better anchored (Figure 3). The reduction is most prevalent in the dollar-lei exchange rate, with the pass-through to consumer prices declining to 4 percent from around 38 percent, and that for producer prices falling to 7 percent from around 45 percent in two year horizon. The reduction can also be seen in euro-lei exchange rate, although to a less extent, with the pass-through to consumer prices declining to 7 percent from 28 percent, while that for producer prices falling to around 19 percent from 38 percent (Table 2).

Figure 3.Exchange Rate Pass-Through pre- and post- Inflation Targeting

Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order). The degree of pass-through is defined as the ratio of the cumulative impulse response in prices to one Cholesky standard deviation shock on the exchange rate.

8. Our findings are consistent with studies for other inflation targeting small open economies and the earlier analysis for Romania. Using a GARCH framework to study the link between the exchange rate regime, inflation targeting and exchange rate pass-through for emerging markets and advanced countries, Floerkemeier (2013) found that the introduction of inflation targeting tends to reduce pass-through from the exchange rate to domestic prices, and reduce both exchange rate and interest rate volatility. On Romania, Cozmanca and Manea (2010) showed that the degree of exchange rate pass-through into producer and consumer prices display a decline in magnitude in more recent times, and Gueorguiev (2003) found that the pass-through to producer prices tend to be larger than that to consumer prices.

C. What Explains Inflation Dynamics in Romania?

9. To examine the main drivers of inflation dynamics in Romania, we study the variance decomposition of the structural VAR framework used in the pass-through analysis. The variance decomposition captures the percentage of the variance of the error made in forecasting inflation due to a specific shock, for example, shock or innovation in the exchange rate equation at a specific time horizon (two years in our analysis). The variance decomposition therefore reflects the amount of information each variable contributes to innovations in the inflation variable.

10. The role of the exchange rate in explaining fluctuations in consumer prices has declined (Figure 4). Prior to August 2005, the exchange rate (in particular, the dollar-lei exchange rate) is one of the three key factors in driving CPI inflation, in addition to innovations in producer prices and CPI inflation’s own innovation. Since August 2005, the share of innovation explained by exchange rate has fallen from around 30 percent to 2 percent for the dollar-lei exchange rate, and from around 10 to 7 percent for the euro-lei exchange rate (Tables 4 and 5). This decline in the role of exchange rate is consistent with our finding of a reduced pass-through to consumer prices after the adoption of inflation targeting. Similarly, wage developments also account for a smaller share of consumer price fluctuations. Instead, inflation appears to have become more persistent.

Table 3.Variance Decomposition of D_logCPI with Lei Euro Exchange Rate
Pre-IT
MonthsDLOG_CPIDLOG_E_EURODLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
164.13.40.09.220.41.01.9
352.310.90.46.922.65.21.6
1245.09.82.26.226.88.11.9
2443.010.02.76.027.58.52.3
Post-IT
MonthsDLOG_CPIDLOG_E_EURODLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
183.60.40.03.69.30.03.1
375.46.60.05.010.00.32.7
1275.16.80.05.010.00.32.8
2475.06.80.05.010.00.32.8
Note: Results are based on a structural VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).
Table 4.Variance Decomposition of D_logCPI with Lei Dollar Exchange Rate
Pre-IT
MonthsDLOG_CPIDLOG_E_DOLLARDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
164.37.20.08.418.21.50.3
338.426.00.05.222.37.11.0
1234.129.40.04.622.97.41.6
2434.129.40.04.622.97.41.6
Post-IT
MonthsDLOG_CPIDLOG_E_DOLLARDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
182.30.00.03.513.60.30.3
368.52.00.49.515.40.93.4
1267.42.40.79.615.50.93.4
2467.42.40.79.615.50.93.4
Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).
Table 5.Variance Decomposition of D_logCPI with NEER
Pre-IT
MonthsDLOG_CPIDLOG_NEERDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
158.15.80.08.422.64.40.7
338.317.41.55.622.713.41.0
1234.417.84.75.122.214.61.2
2434.417.84.75.122.214.61.2
Post-IT
MonthsDLOG_CPIDLOG_NEERDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
181.71.00.02.512.70.02.0
376.53.90.03.513.90.41.8
1276.34.00.03.513.90.41.9
2476.34.00.03.513.90.41.9
Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).

Figure 4.Variance Decomposition of CPI inflation

Note: Results are based on a structural VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).

11. The exchange rate is more important in driving producer price fluctuations compared with consumer prices (Figure 5). Up to 35 percent of the innovations in producer prices can be explained by the exchange rate in the pre-inflation targeting period, with other important factors being the commodity prices and PPI’s own innovations. The importance of the exchange rate in explaining producer price inflation could reflect the relatively high import content of exports in Romania, where a high proportion of intermediate inputs are imported and therefore priced in foreign currencies.

Figure 5.Variance Decomposition of PPI Inflation

Note: Results are based on a structural VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).

12. Commodity prices have overtaken the exchange rate as a more important factor in explaining PPI inflation in recent years. Commodity prices now account for almost 30 percent of fluctuations in PPI inflation. The share of innovations explained by the dollar-lei exchange rate has declined from around 35 percent before August 2005, to around 9 percent thereafter. Interestingly, the importance of the lei-euro exchange rate has been stable, possibly reflecting that an increasing share of Romania’s foreign trade is conducted in euro, and the importance of the European Union and some euro zone economies as Romania’s trading partners.

Romania: Share of Imports

(Percent)

Source: IMF Direction of Trade Statistics.

D. Conclusions and Policy Implications

13. The empirical analysis suggests that the pass-through is typically higher for PPI than for CPI, but both reduced sharply since the adoption of inflation targeting. The introduction of inflation targeting has led to a decline in consumer and producer price inflation, likely better anchored inflation expectations, and reduced the role of the exchange rate in inflation. The degree of pass-through reduced to below 10 percent for CPI inflation, and to below 25 percent for PPI inflation, compared with a maximum of 45 percent prior to introducing inflation targeting. The reduction in pass-through is typically less prevalent in the lei-euro exchange rate, compared with the lei-dollar exchange rate.

14. Furthermore, the exchange rate plays a less important role in explaining innovations in CPI inflation in recent years. The exchange rate accounts for about 5 percent of fluctuations in CPI inflation since the adoption of inflation targeting, while this number was higher at around 10 to 30 percent prior. For PPI inflation, the share of innovations explained by dollar-lei exchange rate has declined to about 10 percent, while the importance of lei-euro exchange rate has been stable. World commodity prices have overtaken exchange rate as the most important factor in explaining PPI inflation in recent years, after its own innovations, accounting for almost 30 percent of the fluctuations in PPI inflation.

15. These empirical results have a number of policy implications. First, the impact of exchange rate pass-through appears to depend on the monetary policy regime and the level of inflation in the economy. The degree of pass-through tends to decline in an environment of low inflation and low volatility of inflation. Second, the credibility of a low inflation regime, and clear communications from the central bank could help anchor inflation expectations in the economy and reduce in part vulnerabilities of the economy to external shocks. Third, the reduction in the degree of pass-through suggests that Romania can afford to allow more flexibility in its exchange rate. As shown in the literature, a rise in exchange rate volatility would in turn lead to a reduction in pass-through (Corsetti, Dedola, and Leduc, 2008). Finally, the decline in exchange rate pass-through, if permanent, is a favorable development for sustainable inflation convergence and for meeting the convergence criteria for joining the euro zone in the medium- and long-term horizon (Beirne and Bijsterbosch, 2011).

Table 6.Variance Decomposition of D_logPPI with Lei Euro Exchange Rate
Pre-IT
MonthsDLOG_CPIDLOG_E_EURODLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.06.90.01.691.30.20.0
38.120.33.57.755.44.90.1
1213.516.33.66.150.27.72.6
2414.215.94.15.948.78.23.1
Post-IT
MonthsDLOG_CPIDLOG_E_EURODLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.04.70.018.473.53.30.1
31.815.80.025.951.62.32.5
121.815.50.126.649.52.34.2
241.815.50.126.649.52.34.2
Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).
Table 7.Variance Decomposition of D_logPPI with Lei Dollar Exchange Rate
Pre-IT
MonthsDLOG_CPIDLOG_E_DOLLARDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.06.90.01.686.42.42.6
31.632.00.94.553.54.62.8
121.934.70.94.149.85.63.0
241.934.70.94.149.85.63.0
Post-IT
MonthsDLOG_CPIDLOG_E_DOLLARDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.010.30.016.469.53.30.5
31.09.10.423.457.73.45.0
121.18.70.825.754.93.35.5
241.18.70.825.854.93.35.5
Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).
Table 8.Variance Decomposition of D_logPPI with NEER
Pre-IT
MonthsDLOG_CPIDLOG_NEERDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.034.90.00.956.65.81.7
32.132.95.84.342.210.72.0
121.930.89.83.939.212.42.0
241.930.79.83.939.212.42.0
Post-IT
MonthsDLOG_CPIDLOG_NEERDLOG_IRDLOG_P_COMMDDLOG_PPIDLOG_WAGEGAP_LOG_IP
10.020.20.013.664.61.60.0
31.122.50.022.250.21.22.7
121.221.80.123.448.21.34.1
241.221.80.123.448.11.34.1
Note: Results are based on a structrual VAR with the following variables: commodity prices, wages, output gap, exchange rate, PPI, CPI and short term interest rate (in this order).
Annex. Data and Methodology

Data

1. The main data sources are the National Institute of Statistics (NIS), the National Bank of Romania (NBR) and the IMF, through Haver Analytics. The CPI and PPI series are monthly, seasonally adjusted with base year at 2010 from the NIS. The euro-lei and dollar-lei exchange rates are monthly series from the NBR. The world commodity price index from the IMF is used to capture commodity prices in the model. The output gap measure is computed as the deviation of seasonally adjusted industrial production (IP) series from trend. The IP series is preferred over the GDP series since it is of higher frequency and requires no approximation or interpolation to construct the gap at monthly frequency. Labor cost is based on the gross average monthly wage and salary earning series from the NIS. Finally, the short-term interest rate variable is constructed using two series, the first one being the monetary policy rate series from the NBR from January 2003 to August 2014, and this series is backdated using the growth rate of the central bank reference rate series (the second series) from the NBR.

2. The Augmented Dickey-Fuller test suggests that a unit root is present in the all variables except for the output gap variable. Given that most of the variables are non-stationary in levels but stationary in first differences, the I(1) variables are then differenced and incorporated in the structural VAR analysis in first difference.

Unit Root Tests based on Augmented Dickey-Fuller Test
t-StatisticProb.t-StatisticProb.
LOG_CPI-1.560.8DLOG_CPI-5.980
LOG_PPI-1.590.8DLOG_PPI-6.690
LOG_E_DOLLAR-1.70.75DLOG_E_DOLLAR-11.20
LOG_E_EURO-1.70.75DLOG_E_EURO-10.970
LOG_IR*-2.980.14DLOG_IR-12.890
LOG_P_COMMD-2.540.31DLOG_P_COMMD-10.790
LOG_WAGE-1.50.83DLOG_WAGE-19.670
GAP_LOG_IP-3.560.04DGAP_LOG_IP-18.280
Note: The ADF test statistics for all level variables are based on regressions including an intercept and a linear trend; except for GAP_LOG_IP. The ADF tests for all variables in first differences are based on regressions including an intercept. *LOG_IR=1/12*ln(1+IR/100).

Methodology

3. To investigate the degree of exchange rate pass-through, we consider a structural VAR model, represented below.

where Xt is a 7 × 1 vector containing CPI, PPI, exchange rate, commodity prices, output gap, and total labor cost, A0 describes the contemporaneous relations between the variables. A(L)=ΣAiLi, where L is the lag operator. εt is the vector of structural shocks, the impact of which is captured by the matrix B. The lag order of the structural VAR model is chosen according to the Akaike Information Criteria. The above equation can be expressed in reduced form as follows,

where et=A01Bɛt, the reduced form shocks, with the following variance-covariance matrix,

4. For identification, we follow the Cholesky decomposition originally proposed by Sims (1980) and assume a recursive ordering for the variables. In particular, we order the most exogenous variable first, the commodity prices (also proxy for supply shock). Wages and output gap are ordered next, as macroeconomic variables are considered slow-moving in comparison to price variables. Exchange rate follows, assuming that a contemporaneous impact of demand shock (as proxied by output gap) on the exchange rate, while imposing a time lag on the response of output to exchange rate. PPI and CPI prices are included next, being contemporaneously influenced by all the above shocks. Finally, we include interest rate, permitting monetary policy to react simultaneously to all variables in the model. Our modeling strategy is in line with the earlier literature on exchange rate pass-through in Romania and other economies, see for example McCarthy (2000), Gueorguiev (2003), Billmeier & Bonato (2004) and Cozmanca & Manea (2010).

5. The pass-through coefficient is defined as the ratio of the respective cumulative impulse response to one standard deviation shock on the exchange rate.

which accounts for the total impact of exchange rate changes on prices in a given time horizon, capturing potential second round impact.

References

Prepared by TengTeng Xu.

Detailed data and methodology can be found in the annex.

Other Resources Citing This Publication