Abstract
Peru: Selected Issues Paper
Drivers of Peru’s Equilibrium Real Exchange Rate: Is the Nuevo Sol A Commodity Currency?1
This chapter tests the hypothesis of ‘commodity currency’ on the nuevo sol and identifies the drivers of Peru’s equilibrium real exchange rate using cointegration analysis. The results show that export commodity prices do not have a statistically significant impact on Peru’s real effective exchange rate, suggesting that the nuevo sol is not a commodity currency. Large profit repatriation and foreign exchange intervention have effectively insulated Peru’s real exchange rate from the impact of commodity price shocks. The results suggest that Peru’s real exchange rate is broadly in line with fundamentals.
A. Introduction
1. Understanding whether the real exchange rate is in line with its equilibrium is important for the efficient allocation of resources in an economy.2 The real exchange rate is the relative price of tradable and nontradable goods. A misaligned real exchange rate, i.e. a real exchange rate that deviates substantially from equilibrium levels, could create macroeconomic imbalances and distort incentives and the allocation of resources.
2. While the equilibrium real exchange rate is an unobservable variable, economic theory suggests that it is driven by observable economic fundamentals. The fundamentals that underlie the equilibrium real exchange include the terms of trade (or the real prices of key export commodities for commodity dependent economies), the relative productivity of tradables to nontradables, government consumption, and the net foreign liability position. For commodity dependent economies like Peru, the equilibrium real exchange rate is conjectured to be primarily determined by the real prices of export commodities so much that their currencies are commonly referred to as ‘commodity currencies’ (Chen and Rogoff, 2003; Cashin et al, 2004; Bodart et al, 2012).
3. The objective of this chapter is to establish an econometric relationship between the real exchange rate and economic fundamentals. In particular, the study aims to test if Peru’s real exchange rate is primarily determined by the real prices of key export commodities as the ‘commodity currency’ hypothesis would suggest. To achieve this objective, the study employs the Johansen cointegration method. The robustness of the results is tested with various specifications, including with varying definitions of the real exchange rate and real commodity prices, sample sizes, and methodologies.
4. The study also attempts to estimate the path of the notional equilibrium real exchange rate. The equilibrium real exchange rate estimated in this study, however, does not have a normative implication as it does not necessarily imply optimality from a welfare perspective. A normative assessment of the equilibrium real exchange rate requires judgment on the optimality of the values of the fundamentals, which is beyond the scope of this study.
5. The study is organized as follows. The analytical framework is presented in section B, followed by a description of the estimation results in section C. Sections D and E discuss the drivers of the equilibrium real exchange rate. Section F concludes.
B. Theoretical and Empirical Framework
6. Attempts to model the equilibrium real exchange rate goes back to the Purchasing Power Parity (PPP) theory. In its absolute form, the PPP theory states that the exchange rate between the currencies of two countries is simply given by the relative price levels expressed in the same currency (i.e., generalization of the law of one price); in its relative form, the theory asserts that the percentage change in the exchange rate between two currencies is determined by the inflation differential between the corresponding countries. In its relative form, the PPP hypothesis requires deviations from the PPP real exchange rate to die out eventually and the real exchange rate to be stable, exhibiting a stationery or mean reverting property in the long run (Rogoff, 1996; Astorga, 2012). Then, the equilibrium real exchange rate would be constant and could be represented by the long-run or PPP real exchange rate. However, the PPP hypothesis received very little empirical support, especially in the short run, as most studies show that real exchange rate deviations are persistent and that the real exchange rate exhibits a unit root process (Meese and Rogoff, 1983; Rogoff, 1996; Engel, 2000; Astorga, 2012).
7. The empirical failure of the PPP theory led to the hypothesis that the equilibrium real exchange rate could be time varying driven by real factors or fundamentals. In a seminal paper on the PPP puzzle, Rogoff (1996) argues that the high short-term volatility of the real exchange rate and the very slow adjustment of shocks to PPP are so irreconcilable that the deviations from PPP must be accounted for by real factors. Real factors that are hypothesized to drive the equilibrium real exchange rate include the terms of trade (or real prices of commodities for commodity dependent economies), the relative productivity of tradables to nontradables, government consumption, and the net foreign liability position (Froot and Rogoff, 1995; Rogoff, 1996; Montiel, 2007; Ricci et al, 2013).
Real price of commodities: While the terms of trade are generally used in real exchange rate models, for commodity dependent small open economies the real price index of key export commodities is a more relevant variable. As Chen and Rogoff (2003) indicate, aggregate export and import price indices used to construct the terms of trade include goods with sluggish nominal price adjustments and incomplete pass-through, leading to identification problems in econometric estimations. On the contrary, world commodity prices are purely exogenous for small exporting economies as they are determined in world markets. An increase in commodity prices can lead to wage increases in the commodity sector, and across the economy since labor is assumed to be mobile, leading to an increase in the relative price of nontradables as the price of tradables is determined in the world market and, therefore, to a real exchange rate appreciation (Chen and Rogoff, 2003; Cashin et al, 2004).
Relative productivity of tradables to nontradables: According to the Balassa-Samuelson hypothesis (Balassa, 1964; Samuelson, 1964), an increase in the relative productivity of tradables to nontradables will drive up economy-wide wages, and assuming labor is mobile between the two sectors, will result in a higher relative price of nontradables (i.e., a real appreciation).
Net foreign liability position: An increase in net foreign liabilities will require a more depreciated real exchange rate to generate the trade surplus necessary to service the external debt (Rogoff, 1996; Ricci et al, 2013).
Government consumption: Higher government consumption is likely to lead to an appreciation of the equilibrium real exchange rate as government consumption tends to fall more on nontradables than tradables (Froot and Rogoff, 1995; Rogoff, 1996; Ricci et al, 2013).
8. To test if the nuevo sol is a commodity currency, this chapter follows Chen and Rogoff (2003) and Cashin et al (2004). They specify the real effective exchange rate as a function only of the real price of commodities. Given Peru’s reliance on commodity exports, in particular metals such as copper and gold,3 the hypothesis of commodity currency expects Peru’s real effective exchange to be driven primarily by the real price of export commodities. Hence, the regression model takes the following log-linear form:
Where,
REER = the real effective exchange rate index, which is a trade-weighted and exchange-rate-adjusted ratio of domestic to foreign prices; an increase in the REER is an appreciation. For the robustness exercise, the bilateral real exchange rate index (RER) vis-à-vis the US dollar is also used. The source of REER data is IMF’s Information Notice System (INS) database and the RER is constructed using data on the bilateral exchange rate and prices from the IMF’s International Financial Statistics (IFS) database.
RP_COM = the real price of export commodities, constructed as the weighted average world price indices of copper, gold, lead, and zinc (Peru’s major export metals) deflated by the manufacturing export unit value index (MUVI) of advanced economies. Metal price indices are obtained from the IFS database and the MUVI is from the IMF’s World Economic Outlook (WEO) database.
μ = stochastic error term; L = Natural logarithm transformation operator; and t = time index.
The nuevo sol would be regarded as a commodity currency if α1 is positive and statistically significant.
9. Equation (1) is modified by including the remaining fundamentals as:
Where,
PROD = relative productivity. The economy-wide labor productivity of Peru relative to a trade-weighted average labor productivity of trading partner countries is used since data on sectoral productivity is not available. The implicit assumption is that productivity growth is likely to be biased in favor of the tradable sector, meaning that a country with high growth of overall productivity will also exhibit higher productivity growth in the tradable sector relative to that of the nontradable sector. Source of data is Haver.
GCN = the primary current public sector consumption (spending on wages and salaries and goods and services) as a ratio of GDP of Peru relative to that of trading partner countries. Only U.S. data is used in the denominator as consistent time series data is not available for most other trading partner countries. Sources of data are the Central Reserve Bank of Peru (BCRP) and the U.S. Bureau of Economic Analysis (BEA).
NFL = the stock of net foreign liabilities at the end of the previous period as a ratio of previous period’s total external trade in goods and services. As alternatives, NFL as a ratio of GDP and the cumulative current account balance (as a ratio to trade and GDP) are explored. Source of data is the BCRP.
ε = stochastic error term.
All other terms are as defined above.
10. The regression sample covers quarterly data for the period 1992–2013. The year 1992 was chosen as the beginning of the sample period to avoid potential structural shifts in the real exchange rate data due to changes in currency prior to 1992 and major stabilization efforts realized since then. Peru’s current currency, the nuevo sol, was introduced and has been in use since July 1991. For robustness exercise, however, annual data for the sample period 1970–2013 and monthly data for the sample period 1992–2013 were also used.
11. Descriptive analysis of the data shows that Peru’s real effective exchange rate is strongly correlated with relative productivity and relative government consumption. On the other hand, the real effective exchange rate does not seem to have a discernible correlation with the real commodity price index and its correlation with the net foreign liability appears to shift from positive prior to 2007 to negative since 2007 (Figure 1).
Peru: Real Effective Exchange Rate and the Fundamentals
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: Central Reserve Bank of Peru; US Bureau of Economic Analysis; Haver; IMF (IFS, INS, and WEO); and author’s calculations.Peru: Real Effective Exchange Rate and the Fundamentals
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: Central Reserve Bank of Peru; US Bureau of Economic Analysis; Haver; IMF (IFS, INS, and WEO); and author’s calculations.Peru: Real Effective Exchange Rate and the Fundamentals
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: Central Reserve Bank of Peru; US Bureau of Economic Analysis; Haver; IMF (IFS, INS, and WEO); and author’s calculations.12. The real effective exchange rate does not seem to exhibit a stationary process (Figure 2a). The first difference of the real exchange rate, however, clearly portrays a stationary process (Figure 2b). This observation is supported by the results of formal unit root tests, which show that Peru’s real effective exchange rate follows an I (1) process (Appendix Table 1). Unit root tests for the fundamentals also shows that they are all integrated of order one (Appendix Table 1), implying that the right approach for estimating the real effective exchange rate equation is a cointegration analysis. Hence, the Johansen cointegration method is used to test and estimate cointegration relationships between the REER and the fundamentals. Alternative estimation methods, including Dynamic OLS (DOLS), Fully Modified OLS (FMOLS), and Two-Stage Least Squares (2SLS) are also explored to test the robustness of the results.
Peru: Real Effective Exchange Rate
(In logarithm)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: IMF and Author’s calculations.Peru: Real Effective Exchange Rate
(In logarithm)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: IMF and Author’s calculations.Peru: Real Effective Exchange Rate
(In logarithm)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Source: IMF and Author’s calculations.C. Is the Nuevo Sol a Commodity Currency?
13. The results suggest that the nuevo sol is not a commodity currency. The estimates below show that the real price index of commodities does not explain the behavior of the REER (the number in parenthesis is the t-value).
Although Johansen’s Trace and Maximum Eigenvalue tests indicate the presence of cointegration at 10 the percent level (Appendix Table 2a), the estimated coefficient on LRP_COM is very small and not statistically significant, ruling out the hypothesis of a commodity currency. The result is robust to changes in the definition of the real exchange rate (using the RER instead of the REER) and the RP_COM (using the real price of copper and the terms of trade in place of RP_COM), data frequency (using monthly and annual data),4 estimation method, and sample coverage. In all cases, the coefficients are positive as expected, but not statistically significant.
Peru: The Real Exchange Rate and Commodity Prices: Alternative Specifications
Peru: The Real Exchange Rate and Commodity Prices: Alternative Specifications
| Alternative specification | Coefficient | T-value |
|---|---|---|
| Dynamic OLS | 0.03 | 1.21 |
| Fully Modified OLS | 0.02 | 0.43 |
| RER as dependent variable | 0.05 | 0.87 |
| Real price of copper | 0.02 | 0.73 |
| Terms of trade | 0.04 | 0.41 |
| Monthly data: 1992-2013 | 0.03 | 1.17 |
| Annual data: 1970-2013 | 0.01 | 0.11 |
Peru: The Real Exchange Rate and Commodity Prices: Alternative Specifications
| Alternative specification | Coefficient | T-value |
|---|---|---|
| Dynamic OLS | 0.03 | 1.21 |
| Fully Modified OLS | 0.02 | 0.43 |
| RER as dependent variable | 0.05 | 0.87 |
| Real price of copper | 0.02 | 0.73 |
| Terms of trade | 0.04 | 0.41 |
| Monthly data: 1992-2013 | 0.03 | 1.17 |
| Annual data: 1970-2013 | 0.01 | 0.11 |
14. The absence of a statistically significant long run relationship between export commodity prices and the real exchange rate in Peru is somewhat puzzling. While similar studies on other commodity dependent economies generally find evidence of commodity currency, Peru was one of the few countries with no such evidence in Cashin et al (2004) as well (Appendix Table 5). Potential factors that could have weakened the statistical relationship between commodity prices and the real effective exchange rate may include large profit repatriation and active foreign exchange intervention.
15. Despite the commodity price boom, Peru has run current account deficits during most of the past decade as large profit repatriations more than offset trade surpluses. The mining sector in Peru is operated by the private sector, mostly owned by non-residents. As a result, most of the profit from the sector is repatriated. During 2003–13, the time identified by Adler and Magud (2013) as the commodity income windfall period, profit repatriation from Peru amounted to about 6 percent of GDP a year on average. This might have weakened the statistical relationship between commodity prices and the real effective exchange rate since a large part of the commodity price shock may have been reflected in profit repatriation without having a significant impact on domestic demand. It is true that a large part of the repatriated profit has been reinvested in Peru in the mining sector, but the investments rely mostly on imported machineries with limited impact on domestic demand.
Peru: Real Price of Commodities, Profit Repatriation, and Current Account Balance
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: Central Reserve Bank of Peru and author’s estimations.Peru: Real Price of Commodities, Profit Repatriation, and Current Account Balance
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: Central Reserve Bank of Peru and author’s estimations.Peru: Real Price of Commodities, Profit Repatriation, and Current Account Balance
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: Central Reserve Bank of Peru and author’s estimations.
Real Effective Exchange Rate Indices in Selected Latin American Economies 1/
(2010=100)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: IMF and author’s caculations.1/ Numbers in parenthesis infron of country names refer to standard deviations of the REER.Real Effective Exchange Rate Indices in Selected Latin American Economies 1/
(2010=100)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: IMF and author’s caculations.1/ Numbers in parenthesis infron of country names refer to standard deviations of the REER.Real Effective Exchange Rate Indices in Selected Latin American Economies 1/
(2010=100)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Sources: IMF and author’s caculations.1/ Numbers in parenthesis infron of country names refer to standard deviations of the REER.16. Peru’s central bank intervenes actively in the FX market with a stated objective of limiting exchange rate volatility to contain the risks of financial dollarization. Empirical evidence shows that the BCRP’s FX interventions are successful in containing exchange rate volatility (Tashu, 2014). On the other hand, Peru has one of the lowest and most stable rates of inflation in the region, thanks to an inflation targeting framework that has successfully anchored inflation expectations (Armas and Grippa, 2005; Armas et al 2014).5 As a result, Peru’s real exchange rate is the most stable among financially open large Latin American economies.
17. A sustained sterilized FX intervention in an inflation targeting regime appears to have weakened the impact of commodity prices on the real exchange rate.6 To illustrate this, consider a positive commodity price shock. In an inflation targeting regime, the central bank could prevent the inflationary pressure from the commodity windfall income by increasing its policy rate, which in turn can lead to an increase in capital inflows. In a freely floating exchange rate regime, the capital inflows would have appreciated the nominal, and hence the real, exchange rate. The BCRP’s sterilized FX intervention has, however, limited the impacts of capital inflows on the exchange rate, effectively insulating the real exchange rate from the impact of commodity price shocks.
18. The results support the hypothesis that the commodity price shock has been absorbed mostly by large profit repatriations and a sustained FX intervention. To test the hypothesis that large profit repatriations and the central bank’s FX interventions could have insulated the REER from the impact of commodity prices, alternative specifications were estimated where the REER depends on the commodity prices, profit repatriation in percent of GDP (PREP), and net international reserves in percent of GDP (NIR) as a proxy for FX intervention.7
Profit repatriation should lead to a depreciation of the nominal, and hence the real, exchange rate because it increases demand for foreign exchange. As a result, θ2 < 0. The NIR is also expected to have a negative relationship with the real exchange rate, as an increase in the NIR (FX purchases by the central bank) and a decrease in NIR (FX sales by the central bank) should lead to a depreciation and appreciation of the national currency, respectively, if successful. Hence, θ3 < 0.
Equations (5) and (6) aim to evaluate if changes in commodity prices can also affect profit repatriation and net international reserves. From (4), the impact of commodity prices on the REER if we were to hold PREP and NIR constant is θ1. In reality, however, both PREP and NIR change when commodity prices change. Firms’ profit increases as commodity prices increase, implying γ1 > 0, and a positive commodity price shock prompts central bank intervention in the FX market and hence an increae in the NIR, implying δ1 > 0. As a result, the net impact of commodity prices on the REER is given by (θ1 + θ2 * γ1 + θ3 * δ1), and could be zero, negative or positive depending on the relative size of the individual coefficients.
19. The results show that all of the coefficients have the expected sign and are statistically significant at standard levels of significance. Equations (4)–(6) were estimated using the Johansen cointegration method and the results are shown below: 8
The estimate for the net impact of the commodity prices (θ1 + θ2 * γ1 + θ3 * δ1) equals 0.01, which is very low and virtually the same as the estimated coefficient obtained when the real effective exchange rate is regressed only on the commodity prices (equation (3)).
The impact of commodity prices on the real effective exchange rate, if we were to hold profit repatriation constant and assume no FX intervention, would have been statistically significant with an estimated elasticity of about 0.5. In reality, however, changes in commodity prices have statistically significant positive impact on profit repatriation and central bank intervention, which in turn affect the real effective exchange rate negatively, neutralizing the initial impact of the commodity prices on the real effective exchange rate.
D. Identifying the Drivers of the Equilibrium Real Exchange Rate
20. Relative productivity and government consumption are the main drivers of the equilibrium real effective exchange rate in Peru. The search for a cointegrating vector between the REER and fundamentals involved an algorithm, which: (i) discards models that do not have a statistically significant vector; (ii) eliminates variables which do not have coefficients with the theoretically expected sign or whose inclusion changes the signs of other variables; (iii) discards models which do not have a statistically significant error correction term with negative sign; and (iv) maximizes the R-square of the ECM. The net foreign liability was dropped from the chosen model, following this algorithm, similar to the results of other studies, including Montiel (2007) and Coudart et al (2011). The test for cointegration among the remaining variables shows a single cointegrating vector at 10 percent significant level (Appendix Table 2b), which after normalizing for the coefficient of LREER, takes the following form:
Where numbers in parenthesis refer to t-values.
While all of the fundamentals in equation (10) have the expected signs on their coefficients, the real price of commodities is not statistically significant as is the case in equation (3). Tests for cointegration restrictions show that LRP_COM is not important for the cointegrating vector (Appendix Table 2c).
As a result, equation (10) is re-estimated without LRP_COM and the resulting cointegration vector, which becomes statistically significant at 1 percent level (Appendix Table 2d), and the short-run dynamic equation are shown in equations (11) and (12), respectively:
Where, D-stands for the first difference, the subscript (-1) refers to the first lag, and ECM stands for the error correction term, which is the error term of equation (11). Numbers in parenthesis are t-values.9
21. The results are robust to changes in specifications. The exception is when annual data used, which show a statistically significant RP_COM, but the elasticity remains very small (0.03).10
Peru: The Real Exchange Rate and Fundamentals: Alternative Specifications 1/
Numbers in parenthesis are t-values
RP_COM is dropped from the RER model as it carries a theoretically-wrong sign.
Net foreign liabilities becomes significant with a theoretically-expected negative sign and elasticity of 0.06. The trade openness index, which was included in the quarterly data since Peru liberalized its external trade regime in 1991, is also included in the annual sample (Appendix Tables 4b and dc).
Peru: The Real Exchange Rate and Fundamentals: Alternative Specifications 1/
| Alternative specification | RP_COM | LPROD | LGCN |
|---|---|---|---|
| Two-stage Least Squares | 0.01 | 0.43 | 0.14 |
| (using first lags as instruments) | (0.52) | (8.59) | (2.09) |
| Dynamic OLS | 0.02 | 0.36 | 0.15 |
| (0.95) | (3.63) | (1.95) | |
| Fully Modified OLS | 0.01 | 0.36 | 0.11 |
| (0.83) | (4.20) | (1.71) | |
| RER as dependent variable 2/ | … | 0.92 | 0.41 |
| (3.32) | (2.87) | ||
| Real price of copper | 0.02 | 0.40 | 0.41 |
| (1.05) | (3.08) | (4.38) | |
| Terms of trade | 0.05 | 0.43 | 0.41 |
| (0.65) | (3.12) | (4.22) | |
| Annual data: 1970-2013 3/ | 0.03 | 0.19 | 0.19 |
| (2.95) | (9.26) | (8.62) |
Numbers in parenthesis are t-values
RP_COM is dropped from the RER model as it carries a theoretically-wrong sign.
Net foreign liabilities becomes significant with a theoretically-expected negative sign and elasticity of 0.06. The trade openness index, which was included in the quarterly data since Peru liberalized its external trade regime in 1991, is also included in the annual sample (Appendix Tables 4b and dc).
Peru: The Real Exchange Rate and Fundamentals: Alternative Specifications 1/
| Alternative specification | RP_COM | LPROD | LGCN |
|---|---|---|---|
| Two-stage Least Squares | 0.01 | 0.43 | 0.14 |
| (using first lags as instruments) | (0.52) | (8.59) | (2.09) |
| Dynamic OLS | 0.02 | 0.36 | 0.15 |
| (0.95) | (3.63) | (1.95) | |
| Fully Modified OLS | 0.01 | 0.36 | 0.11 |
| (0.83) | (4.20) | (1.71) | |
| RER as dependent variable 2/ | … | 0.92 | 0.41 |
| (3.32) | (2.87) | ||
| Real price of copper | 0.02 | 0.40 | 0.41 |
| (1.05) | (3.08) | (4.38) | |
| Terms of trade | 0.05 | 0.43 | 0.41 |
| (0.65) | (3.12) | (4.22) | |
| Annual data: 1970-2013 3/ | 0.03 | 0.19 | 0.19 |
| (2.95) | (9.26) | (8.62) |
Numbers in parenthesis are t-values
RP_COM is dropped from the RER model as it carries a theoretically-wrong sign.
Net foreign liabilities becomes significant with a theoretically-expected negative sign and elasticity of 0.06. The trade openness index, which was included in the quarterly data since Peru liberalized its external trade regime in 1991, is also included in the annual sample (Appendix Tables 4b and dc).
E. Is the Real Effective Exchange Rate Misaligned?
22. The equilibrium real exchange rate is estimated using the cointegration relationship between the REER and fundamentals. While a proper estimation of the equilibrium real exchange rate requires a multi-country panel regression analysis similar to the IMF’s external balance assessment (Phillips et al, 2013), the estimated long-run relationship between the REER and statistically significant fundamentals is used to estimate the notional path of the equilibrium REER. In theory, the equilibrium real effective exchange rate is the value of the real effective exchange rate predicted by the ‘sustainable’ or ‘steady state’ values of the fundamentals (Montiel, 2007). Hence, the fundamentals are filtered by the Hodrick-Prescott (HP) filter to remove cyclical components and estimate their sustainable components.
23. Based on this study, the real exchange rate appears to be broadly in line with the fundamentals. The estimated results show that, over the past decade, Peru’s real effective exchange rate appears to have been broadly in line with the fundamentals with the exception of mild misalignments in some years. In particular, the REER was:
Mildly undervalued during 2004-07 by 2¼ percent on average: the REER depreciated by about 4 percent during this period, while the equilibrium REER depreciated by about 2 percent as the impact of large retrenchments in government consumption (relative to the U.S.) more than offset the impact of improvements in relative productivity.
Consistent with the equilibrium REER in 2008.
Mildly overvalued during 2009-13 by about 4¾ percent on average: possibly because the massive capital inflow, which caused a significant REER appreciation (14 percent), was driven not only by Peru’s fundamentals, which justified only 9 percent equilibrium REER appreciation, but also by global push factors. However, a large part of the misalignment, which peaked in the 1st quarter of 2013 at 8¾ percent, was corrected in the second half of 2013, as the nuevo sol depreciated following the U.S. Fed Reserve’s announcement of monetary policy tapering.
Broadly in line with fundamentals in 2014. The correction in the second half of 2013 continued through 2014, when the real exchange rate was overvalued only by about ½ a percentage point.
24. The REER assessment of this study does not necessarily have a normative value. A REER close to its equilibrium level may still reflect distortions in the fundamentals (Phillips et al, 2013). A normative assessment of the equilibrium REER requires making judgments on the ‘appropriateness’ of the fundamentals from a welfare perspective, which is beyond the scope of this study.
Peru: Actual and Estimated Equilibrium Real Effective Exchange Rate
Peru: Actual and Estimated Equilibrium Real Effective Exchange Rate
| Year | Actual | Equilibrium | Misalignment |
|---|---|---|---|
| 2004 | 100.8 | 102.3 | −1.4 |
| 2005 | 100.0 | 101.4 | −1.4 |
| 2006 | 98.2 | 100.7 | −2.4 |
| 2007 | 96.5 | 99.9 | −3.5 |
| 2008 | 100.8 | 99.6 | 1.2 |
| 2009 | 104.1 | 99.8 | 4.3 |
| 2010 | 106.6 | 100.9 | 5.6 |
| 2011 | 105.5 | 102.9 | 2.5 |
| 2012 | 114.5 | 105.8 | 8.2 |
| 2013 | 115.0 | 109.3 | 5.2 |
| 2014 | 114.0 | 113.3 | 0.6 |
Peru: Actual and Estimated Equilibrium Real Effective Exchange Rate
| Year | Actual | Equilibrium | Misalignment |
|---|---|---|---|
| 2004 | 100.8 | 102.3 | −1.4 |
| 2005 | 100.0 | 101.4 | −1.4 |
| 2006 | 98.2 | 100.7 | −2.4 |
| 2007 | 96.5 | 99.9 | −3.5 |
| 2008 | 100.8 | 99.6 | 1.2 |
| 2009 | 104.1 | 99.8 | 4.3 |
| 2010 | 106.6 | 100.9 | 5.6 |
| 2011 | 105.5 | 102.9 | 2.5 |
| 2012 | 114.5 | 105.8 | 8.2 |
| 2013 | 115.0 | 109.3 | 5.2 |
| 2014 | 114.0 | 113.3 | 0.6 |
Peru: Contributions of Fundamentals to Changes in EREER
(Percentage change)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Peru: Contributions of Fundamentals to Changes in EREER
(Percentage change)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
Peru: Contributions of Fundamentals to Changes in EREER
(Percentage change)
Citation: IMF Staff Country Reports 2015, 134; 10.5089/9781513560410.002.A003
F. Concluding Remarks
25. The results of this study suggest that the nuevo sol is not a commodity currency. This appears puzzling for a country that relies heavily on metal commodities for its exports. This study provides empirical evidence that large profit repatriation and the BCRP’s active FX intervention could have mitigated the impact of commodity prices on the real effective exchange rate.
26. Relative productivity and government consumption are the main drivers of the equilibrium real exchange rate in Peru. The empirical analysis identifies the main drivers of the equilibrium real exchange rate from a pool of economic fundamentals that include the real price of commodities, Peru’s productivity relative to that of trading partners, Peru’s government consumption relative that of trading partners, and net foreign liabilities. The results show that only productivity and government consumption, both relative to that of trading partners, have statistically significant relationships with the real effective exchange rate, suggesting that the equilibrium REER is driven by these two fundamentals.
27. Peru’s real effective exchange rate appears to be broadly in line with the notional equilibrium level predicted by the ‘sustainable’ values of the fundamentals. The equilibrium real effective exchange rate is estimated based on the cointegrating relationship between the real effective exchange rate and the statistically significant fundamentals. The REER was mildly overvalued in the years following the 2008 global financial crisis, which is not surprising given the surge in capital inflows triggered mostly by easy monetary policy in advanced economies. But the depreciation of the REER following the U.S. Fed Reserve announcement of unconventional monetary policy tapering in May 2013 appears to have broadly corrected the overvaluation.
28. The results of the study on the equilibrium real exchange rate need to be interpreted only as indicative. A proper exchange rate assessment requires a panel data based analysis, in line with the IMF’s EBA assessment, to deal with technical problems associated with small sample size and potential structural breaks. Also, a normative assessment of the real exchange rate requires determining the optimal levels of the fundamentals and desirable policy settings, as conducted in the Article IV staff report.
Appendix. Tables
Null Hypothesis is unit root in all cases. The Null Hypothesis is accepted for t-statistics greater than corresponding critical values.
All variables are expressed in natural logarithmic form.
As a ratio of previous period’s total external trade in goods and services.
In percent of GDP.
| Variable | ADF t-statistic | DF-GLS t-statistic | Remarks | |||||
|---|---|---|---|---|---|---|---|---|
| None | Constant | Contant and trend | Constant | Contant and trend | ||||
| Real effective exchange rate | Level | −0.24 | −2.59 | −2.44 | −1.36 | −1.71 | I(1) | |
| Difference (1st) | −7.51 | −7.46 | −7.53 | −7.01 | −7.58 | |||
| Real bilateral exchange rate | Level | −0.20 | −2.37 | −1.70 | −1.24 | −1.37 | I(1) | |
| Difference (1st) | −6.93 | −6.89 | −6.96 | −6.73 | −7.01 | |||
| Real price index of export commodities | Level | 0.84 | −0.73 | −2.13 | −0.41 | −1.76 | I(1) | |
| Difference (1st) | −6.75 | −6.80 | −6.79 | −6.84 | −6.84 | |||
| Real price of copper | Level | 0.57 | −0.89 | −2.54 | −0.65 | −1.87 | I(1) | |
| Difference (1st) | −7.01 | −7.02 | −7.01 | −7.06 | −7.07 | |||
| Terms of trade | Level | 0.06 | −1.78 | −2.41 | −1.78 | −2.19 | I(1) | |
| Difference (1st) | −6.38 | −6.34 | −6.29 | −6.21 | −6.24 | |||
| Relative productivity | Level | −0.57 | −0.93 | −0.54 | −0.91 | −0.73 | I(1) | |
| Difference (1st) | −8.01 | −7.98 | −7.99 | −2.50 | −6.19 | |||
| Relative government consumption | Level | −1.05 | −2.44 | −2.51 | −0.48 | −1.11 | I(1) | |
| Difference (1st) | −15.08 | −15.03 | −17.08 | −1.67 | −3.46 | |||
| Net foreign liability 3/ | Level | −0.95 | −0.65 | −1.44 | 0.53 | −1.43 | I(1) | |
| Difference (1st) | −6.97 | −7.36 | −7.32 | −7.38 | −7.33 | |||
| Net international reserves 4/ | Level | 2.50 | −4.46 | −4.35 | 0.94 | −1.03 | I(1) | |
| Difference (1st) | −6.34 | −6.81 | −7.14 | −4.53 | −5.60 | |||
| Profit repatriation 4/ | Level | −0.58 | −1.50 | −2.38 | −0.90 | −2.51 | I(1) | |
| Difference (1st) | −12.16 | −12.12 | −12.10 | −11.93 | −11.63 | |||
| Critical Values | ||||||||
| 1% | −2.59 | −3.51 | −4.07 | −2.59 | −3.63 | |||
| 5% | −1.95 | −2.90 | −3.46 | −1.95 | −3.07 | |||
| 10% | −1.61 | −2.59 | −3.16 | −1.61 | −2.78 | |||
Null Hypothesis is unit root in all cases. The Null Hypothesis is accepted for t-statistics greater than corresponding critical values.
All variables are expressed in natural logarithmic form.
As a ratio of previous period’s total external trade in goods and services.
In percent of GDP.
| Variable | ADF t-statistic | DF-GLS t-statistic | Remarks | |||||
|---|---|---|---|---|---|---|---|---|
| None | Constant | Contant and trend | Constant | Contant and trend | ||||
| Real effective exchange rate | Level | −0.24 | −2.59 | −2.44 | −1.36 | −1.71 | I(1) | |
| Difference (1st) | −7.51 | −7.46 | −7.53 | −7.01 | −7.58 | |||
| Real bilateral exchange rate | Level | −0.20 | −2.37 | −1.70 | −1.24 | −1.37 | I(1) | |
| Difference (1st) | −6.93 | −6.89 | −6.96 | −6.73 | −7.01 | |||
| Real price index of export commodities | Level | 0.84 | −0.73 | −2.13 | −0.41 | −1.76 | I(1) | |
| Difference (1st) | −6.75 | −6.80 | −6.79 | −6.84 | −6.84 | |||
| Real price of copper | Level | 0.57 | −0.89 | −2.54 | −0.65 | −1.87 | I(1) | |
| Difference (1st) | −7.01 | −7.02 | −7.01 | −7.06 | −7.07 | |||
| Terms of trade | Level | 0.06 | −1.78 | −2.41 | −1.78 | −2.19 | I(1) | |
| Difference (1st) | −6.38 | −6.34 | −6.29 | −6.21 | −6.24 | |||
| Relative productivity | Level | −0.57 | −0.93 | −0.54 | −0.91 | −0.73 | I(1) | |
| Difference (1st) | −8.01 | −7.98 | −7.99 | −2.50 | −6.19 | |||
| Relative government consumption | Level | −1.05 | −2.44 | −2.51 | −0.48 | −1.11 | I(1) | |
| Difference (1st) | −15.08 | −15.03 | −17.08 | −1.67 | −3.46 | |||
| Net foreign liability 3/ | Level | −0.95 | −0.65 | −1.44 | 0.53 | −1.43 | I(1) | |
| Difference (1st) | −6.97 | −7.36 | −7.32 | −7.38 | −7.33 | |||
| Net international reserves 4/ | Level | 2.50 | −4.46 | −4.35 | 0.94 | −1.03 | I(1) | |
| Difference (1st) | −6.34 | −6.81 | −7.14 | −4.53 | −5.60 | |||
| Profit repatriation 4/ | Level | −0.58 | −1.50 | −2.38 | −0.90 | −2.51 | I(1) | |
| Difference (1st) | −12.16 | −12.12 | −12.10 | −11.93 | −11.63 | |||
| Critical Values | ||||||||
| 1% | −2.59 | −3.51 | −4.07 | −2.59 | −3.63 | |||
| 5% | −1.95 | −2.90 | −3.46 | −1.95 | −3.07 | |||
| 10% | −1.61 | −2.59 | −3.16 | −1.61 | −2.78 | |||
Null Hypothesis is unit root in all cases. The Null Hypothesis is accepted for t-statistics greater than corresponding critical values.
All variables are expressed in natural logarithmic form.
As a ratio of previous period’s total external trade in goods and services.
In percent of GDP.
Johansen Cointegration Tests between the Real Effective Exchange Rate and the Fundamentals
Johansen Cointegration Tests between the Real Effective Exchange Rate and the Fundamentals
| (a) Cointegration between LREER and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None* | 0.147 | 14.960 | 15.495 | 0.060 |
| At most 1 | 0.015 | 1.304 | 3.841 | 0.254 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.146826 | 13.6561 | 14.2646 | 0.0622 |
| At most 1 | 0.015048 | 1.30398 | 3.841466 | 0.2535 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *Rejection of the hypothesis at 10% level. | ||||
| (b) Cointegration among LREER, LRP_COM, LGCN, and LPROD_M | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None* | 0.266075 | 46.37389 | 47.85613 | 0.0684 |
| At most 1 | 0.120272 | 19.76994 | 29.79707 | 0.4385 |
| At most 2 | 0.071095 | 8.749647 | 15.49471 | 0.3891 |
| At most 3 | 0.027603 | 2.407229 | 3.841466 | 0.1208 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.266075 | 26.60395 | 27.58434 | 0.0663 |
| At most 1 | 0.120272 | 11.02029 | 21.13162 | 0.6453 |
| At most 2 | 0.071095 | 6.342418 | 14.2646 | 0.5697 |
| At most 3 | 0.027603 | 2.407229 | 3.841466 | 0.1208 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *Rejection of the hypothesis at 10% level. | ||||
| (c) Cointegration restriction tests | ||||
| Null hypothesis | Restricted log-likehood | LR Statistic | Degrees of Freedom | Probability |
| Coefficient on LRP_COM is zero | 650.3344 | 1.920112 | 1 | 0.1658 |
| Coefficient on LPROD is zero ** | 649.2462 | 4.096616 | 1 | 0.0430 |
| Coefficient on LGCN is zero*** | 645.7832 | 11.02264 | 1 | 0.0009 |
| ** Rejection of the hypothesis at 5% level. | ||||
| *** Rejection of the hypothesis at 1% level. | ||||
| (d) Cointegration among LREER, LGCN, and LPROD_M | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None *** | 0.236197 | 37.57606 | 29.79707 | 0.0052 |
| At most 1 | 0.103207 | 14.40381 | 15.49471 | 0.0725 |
| At most 2 ** | 0.056874 | 5.035806 | 3.841466 | 0.0248 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None ** | 0.236197 | 23.17225 | 21.13162 | 0.0255 |
| At most 1 | 0.103207 | 9.368004 | 14.2646 | 0.2568 |
| At most 2 ** | 0.056874 | 5.035806 | 3.841466 | 0.0248 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| ** Rejection of the hypothesis at 5% level. | ||||
| *** Rejection of the hypothesis at 1% level. | ||||
| (e) LREER, LRP_COM, LPROFIT, and LNIR | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None ** | 0.3317 | 54.1336 | 47.8561 | 0.0115 |
| At most 1 | 0.1118 | 20.2804 | 29.7971 | 0.4040 |
| At most 2 | 0.0752 | 10.3257 | 15.4947 | 0.2565 |
| At most 3* | 0.0438 | 3.7582 | 3.8415 | 0.0525 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None *** | 0.331698 | 33.85326 | 27.58434 | 0.0069 |
| At most 1 | 0.111755 | 9.954619 | 21.13162 | 0.7489 |
| At most 2 | 0.075207 | 6.567558 | 14.2646 | 0.5415 |
| At most 3* | 0.043754 | 3.758178 | 3.841466 | 0.0525 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| * Denotes rejection of the hypothesis at 10% level. | ||||
| * * Denotes rejection of the hypothesis at 5% level. | ||||
| *** Denotes rejection of the hypothesis at 1% level. | ||||
| (f) LNIR and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None *** | 0.247573 | 24.18814 | 15.49471 | 0.0019 |
| At most 1 | 0.003496 | 0.294206 | 3.841466 | 0.5875 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None *** | 0.247573 | 23.89393 | 14.2646 | 0.0011 |
| At most 1 | 0.003496 | 0.294206 | 3.841466 | 0.5875 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *** Denotes rejection of the hypothesis at 1% level. | ||||
| (g) LPROFIT and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None ** | 0.155422 | 16.85422 | 15.49471 | 0.031 |
| At most 1 | 0.031229 | 2.665094 | 3.841466 | 0.1026 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.155422 | 14.18913 | 14.2646 | 0.0514 |
| At most 1 | 0.031229 | 2.665094 | 3.841466 | 0.1026 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| * Denotes rejection of the hypothesis at 10 level. | ||||
| * * Denotes rejection of the hypothesis at 5 level. | ||||
Johansen Cointegration Tests between the Real Effective Exchange Rate and the Fundamentals
| (a) Cointegration between LREER and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None* | 0.147 | 14.960 | 15.495 | 0.060 |
| At most 1 | 0.015 | 1.304 | 3.841 | 0.254 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.146826 | 13.6561 | 14.2646 | 0.0622 |
| At most 1 | 0.015048 | 1.30398 | 3.841466 | 0.2535 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *Rejection of the hypothesis at 10% level. | ||||
| (b) Cointegration among LREER, LRP_COM, LGCN, and LPROD_M | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None* | 0.266075 | 46.37389 | 47.85613 | 0.0684 |
| At most 1 | 0.120272 | 19.76994 | 29.79707 | 0.4385 |
| At most 2 | 0.071095 | 8.749647 | 15.49471 | 0.3891 |
| At most 3 | 0.027603 | 2.407229 | 3.841466 | 0.1208 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.266075 | 26.60395 | 27.58434 | 0.0663 |
| At most 1 | 0.120272 | 11.02029 | 21.13162 | 0.6453 |
| At most 2 | 0.071095 | 6.342418 | 14.2646 | 0.5697 |
| At most 3 | 0.027603 | 2.407229 | 3.841466 | 0.1208 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *Rejection of the hypothesis at 10% level. | ||||
| (c) Cointegration restriction tests | ||||
| Null hypothesis | Restricted log-likehood | LR Statistic | Degrees of Freedom | Probability |
| Coefficient on LRP_COM is zero | 650.3344 | 1.920112 | 1 | 0.1658 |
| Coefficient on LPROD is zero ** | 649.2462 | 4.096616 | 1 | 0.0430 |
| Coefficient on LGCN is zero*** | 645.7832 | 11.02264 | 1 | 0.0009 |
| ** Rejection of the hypothesis at 5% level. | ||||
| *** Rejection of the hypothesis at 1% level. | ||||
| (d) Cointegration among LREER, LGCN, and LPROD_M | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None *** | 0.236197 | 37.57606 | 29.79707 | 0.0052 |
| At most 1 | 0.103207 | 14.40381 | 15.49471 | 0.0725 |
| At most 2 ** | 0.056874 | 5.035806 | 3.841466 | 0.0248 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None ** | 0.236197 | 23.17225 | 21.13162 | 0.0255 |
| At most 1 | 0.103207 | 9.368004 | 14.2646 | 0.2568 |
| At most 2 ** | 0.056874 | 5.035806 | 3.841466 | 0.0248 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| ** Rejection of the hypothesis at 5% level. | ||||
| *** Rejection of the hypothesis at 1% level. | ||||
| (e) LREER, LRP_COM, LPROFIT, and LNIR | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None ** | 0.3317 | 54.1336 | 47.8561 | 0.0115 |
| At most 1 | 0.1118 | 20.2804 | 29.7971 | 0.4040 |
| At most 2 | 0.0752 | 10.3257 | 15.4947 | 0.2565 |
| At most 3* | 0.0438 | 3.7582 | 3.8415 | 0.0525 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None *** | 0.331698 | 33.85326 | 27.58434 | 0.0069 |
| At most 1 | 0.111755 | 9.954619 | 21.13162 | 0.7489 |
| At most 2 | 0.075207 | 6.567558 | 14.2646 | 0.5415 |
| At most 3* | 0.043754 | 3.758178 | 3.841466 | 0.0525 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| * Denotes rejection of the hypothesis at 10% level. | ||||
| * * Denotes rejection of the hypothesis at 5% level. | ||||
| *** Denotes rejection of the hypothesis at 1% level. | ||||
| (f) LNIR and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None *** | 0.247573 | 24.18814 | 15.49471 | 0.0019 |
| At most 1 | 0.003496 | 0.294206 | 3.841466 | 0.5875 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None *** | 0.247573 | 23.89393 | 14.2646 | 0.0011 |
| At most 1 | 0.003496 | 0.294206 | 3.841466 | 0.5875 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| *** Denotes rejection of the hypothesis at 1% level. | ||||
| (g) LPROFIT and LRP_COM | ||||
| Unrestricted Cointegration Rank Test (Trace) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Trace Statistic | Critical Value | Prob. 1/ |
| None ** | 0.155422 | 16.85422 | 15.49471 | 0.031 |
| At most 1 | 0.031229 | 2.665094 | 3.841466 | 0.1026 |
| Unrestricted Cointegration Rank Test (Maximum Eigenvalue) | ||||
| Hypothesized no. of CE(s) | Eigenvalue | Maximum-Eigen Statistic | Critical Value | Prob. 1/ |
| None* | 0.155422 | 14.18913 | 14.2646 | 0.0514 |
| At most 1 | 0.031229 | 2.665094 | 3.841466 | 0.1026 |
| 1/ MacKinnon-Haug-Michelis (1999) p-values | ||||
| * Denotes rejection of the hypothesis at 10 level. | ||||
| * * Denotes rejection of the hypothesis at 5 level. | ||||
Gregory-Hansen Test for Cointegration with Regime Shift: Annual sample (1970—2013)1/
The null hypothesis is ‘no cointegration’.
Includes LRP_COM, LPROD, LGCN and LNFL.
Null hypothesis regected at 10% significance level.
Null hypothesis regected at 1% significance level.
Gregory-Hansen Test for Cointegration with Regime Shift: Annual sample (1970—2013)1/
| (a) LREER and LRP_COM | |||||
| Test | Asymptotic critical values | ||||
| statistic | Shift year | 1% | 5% | 10% | |
| ADF | −4.70* | 1987 | −5.47 | −4.95 | −4.68 |
| Zt | −4.75* | 1987 | −5.47 | −4.95 | −4.68 |
| Za | −30.34 | 1987 | −57.17 | −47.04 | −41.85 |
| (b) LREER and All Fundamentals 2/ | |||||
| Test | Asymptotic critical values | ||||
| statistic | Shift year | 1% | 5% | 10% | |
| ADF | −7.08*** | 1988 | −6.92 | −6.41 | −6.17 |
| Zt | −6.24* | 1989 | −6.92 | −6.41 | −6.17 |
| Za | −30.34 | 1989 | −90.35 | −78.52 | −75.56 |
The null hypothesis is ‘no cointegration’.
Includes LRP_COM, LPROD, LGCN and LNFL.
Null hypothesis regected at 10% significance level.
Null hypothesis regected at 1% significance level.
Gregory-Hansen Test for Cointegration with Regime Shift: Annual sample (1970—2013)1/
| (a) LREER and LRP_COM | |||||
| Test | Asymptotic critical values | ||||
| statistic | Shift year | 1% | 5% | 10% | |
| ADF | −4.70* | 1987 | −5.47 | −4.95 | −4.68 |
| Zt | −4.75* | 1987 | −5.47 | −4.95 | −4.68 |
| Za | −30.34 | 1987 | −57.17 | −47.04 | −41.85 |
| (b) LREER and All Fundamentals 2/ | |||||
| Test | Asymptotic critical values | ||||
| statistic | Shift year | 1% | 5% | 10% | |
| ADF | −7.08*** | 1988 | −6.92 | −6.41 | −6.17 |
| Zt | −6.24* | 1989 | −6.92 | −6.41 | −6.17 |
| Za | −30.34 | 1989 | −90.35 | −78.52 | −75.56 |
The null hypothesis is ‘no cointegration’.
Includes LRP_COM, LPROD, LGCN and LNFL.
Null hypothesis regected at 10% significance level.
Null hypothesis regected at 1% significance level.
Estimating Non-linear Cointegration using the FMOLS Method: Annual Sample (1970–2013)
RS1987 refers to dummy for regime shift in 1987, identified by the Gregory-Hansen test (Appendix Table 3a).
RS1988 refers to dummy for regime shift in 1988, identified by the Gregory-Hansen test (Appendix Table 3b).
LNFL and TRADE_OPEN (dummy for trade openness) show no change in the sign of their coefficients when interacted with RS1988. As a result, they are included without interactions.
TRADE_OPEN was not included in the cointegration test in Appendix Table 3b since the Gregory-Hansen test does not allow for more than four right hand side variables and dummy variables.
Estimating Non-linear Cointegration using the FMOLS Method: Annual Sample (1970–2013)
| (a) LREER and LRP_COM | |||
| LREER= a(1)*LRP_COM+a(2)*LRP_COMx RS1987+a(3) | |||
| Coefficient 1/ | Coefficient | Standard error | Probability |
| a(1) | −0.26 | 0.12 | 0.0349 |
| a(2) | 0.27 | 0.02 | 0.0000 |
| a(3) | 4.59 | 0.54 | 0.0000 |
| (b) LREER and All Fundamentals | |||
| LREER=b(1)*LRP_COM+b(2)*LRP_COM×RS1988+b(3)*LGCN+b(4)*LGCN×RS1988 + b(5)*LPROD + b(6)*LPRODx RS1988 + b(7)*LNFL + b(8)*TRADE_OPEN + b(9) | |||
| Variable 2/ | Coefficient | Standard error | Probability |
| b(1) | −0.16 | 0.01 | 0.0000 |
| b(2) | 0.19 | 0.00 | 0.0000 |
| b(3) | −0.44 | 0.03 | 0.0000 |
| b(4) | 0.63 | 0.04 | 0.0000 |
| b(5) | −0.76 | 0.04 | 0.0000 |
| b(6) | 0.95 | 0.04 | 0.0000 |
| b(7) 3/ | −0.06 | 0.01 | 0.0000 |
| b(8) 3/4/ | 0.52 | 0.01 | 0.0000 |
| b(9) | 4.01 | 0.04 | 0.0000 |
| (c) Wald coefficient restriction tests | |||
| Null hypothesis | value | t-statistic | Probability |
| a(1)+a(2)=0 | 0.01 | 0.11 | 0.9100 |
| b(1)+b(2)=0 | 0.03 | 2.95 | 0.0057 |
| b(3)+b(4)=0 | 0.19 | 9.26 | 0.0000 |
| b(5)+b(6)=0 | 0.19 | 8.62 | 0.0000 |
RS1987 refers to dummy for regime shift in 1987, identified by the Gregory-Hansen test (Appendix Table 3a).
RS1988 refers to dummy for regime shift in 1988, identified by the Gregory-Hansen test (Appendix Table 3b).
LNFL and TRADE_OPEN (dummy for trade openness) show no change in the sign of their coefficients when interacted with RS1988. As a result, they are included without interactions.
TRADE_OPEN was not included in the cointegration test in Appendix Table 3b since the Gregory-Hansen test does not allow for more than four right hand side variables and dummy variables.
Estimating Non-linear Cointegration using the FMOLS Method: Annual Sample (1970–2013)
| (a) LREER and LRP_COM | |||
| LREER= a(1)*LRP_COM+a(2)*LRP_COMx RS1987+a(3) | |||
| Coefficient 1/ | Coefficient | Standard error | Probability |
| a(1) | −0.26 | 0.12 | 0.0349 |
| a(2) | 0.27 | 0.02 | 0.0000 |
| a(3) | 4.59 | 0.54 | 0.0000 |
| (b) LREER and All Fundamentals | |||
| LREER=b(1)*LRP_COM+b(2)*LRP_COM×RS1988+b(3)*LGCN+b(4)*LGCN×RS1988 + b(5)*LPROD + b(6)*LPRODx RS1988 + b(7)*LNFL + b(8)*TRADE_OPEN + b(9) | |||
| Variable 2/ | Coefficient | Standard error | Probability |
| b(1) | −0.16 | 0.01 | 0.0000 |
| b(2) | 0.19 | 0.00 | 0.0000 |
| b(3) | −0.44 | 0.03 | 0.0000 |
| b(4) | 0.63 | 0.04 | 0.0000 |
| b(5) | −0.76 | 0.04 | 0.0000 |
| b(6) | 0.95 | 0.04 | 0.0000 |
| b(7) 3/ | −0.06 | 0.01 | 0.0000 |
| b(8) 3/4/ | 0.52 | 0.01 | 0.0000 |
| b(9) | 4.01 | 0.04 | 0.0000 |
| (c) Wald coefficient restriction tests | |||
| Null hypothesis | value | t-statistic | Probability |
| a(1)+a(2)=0 | 0.01 | 0.11 | 0.9100 |
| b(1)+b(2)=0 | 0.03 | 2.95 | 0.0057 |
| b(3)+b(4)=0 | 0.19 | 9.26 | 0.0000 |
| b(5)+b(6)=0 | 0.19 | 8.62 | 0.0000 |
RS1987 refers to dummy for regime shift in 1987, identified by the Gregory-Hansen test (Appendix Table 3a).
RS1988 refers to dummy for regime shift in 1988, identified by the Gregory-Hansen test (Appendix Table 3b).
LNFL and TRADE_OPEN (dummy for trade openness) show no change in the sign of their coefficients when interacted with RS1988. As a result, they are included without interactions.
TRADE_OPEN was not included in the cointegration test in Appendix Table 3b since the Gregory-Hansen test does not allow for more than four right hand side variables and dummy variables.
Empirical Evidence on Commodity Currency
Empirical Evidence on Commodity Currency
| Author/s (year) | Country/ies | Sample | Method | Elasticity on commodity prices | Definition of commodity prices |
|---|---|---|---|---|---|
| Chen and Rogoff (2003) | Australia, Canada, New Zealand | Quarterly: year varies | Time Series cointegration | Australia (0.4), Canada (0.4), and New Zealand (0.6) | Real commodity prices |
| Cashin et al (2004) | 58 commodity exporting countries, including Peru | Monthly: 1980-2002 | Time Series cointegration | Median=0.4. TOT not important for Peru. | Real commodity prices |
| Ferreyra and Salas (2006) | Peru | Quarterly: 1980-2005 | Time series cointegration | 0.3 | TOT |
| Montiel (2007) | Argentina, Bolivia, Brazil, Chile, Paraguay, Uruguay | Annual: 1969-2005 | Time series cointegration | TOT important only for Argentina (1.7), Bolivia (0.6), and Uruguay(0.6) | TOT |
| Iossifov and Loukoianova (2007) | Ghana | Quarterly: 1984-2006 | Time series cointegration | 0.4 | Real commodity prices |
| Astorga (2012) | Argentina, Brazil, Chile, Colombia, Mexico, Venezuela | Annual: 1900-2000 | Time series cointegration | Argentina (0.4), Brazil (0.2), Chile (0.1), Colombia (0.4), Mexico (not significant), Venezuela (0.1) | TOT |
| Coudert et al (2011) | 52 commodity exporters | Annual: 1980-2007 | Panel cointegration | 0.4 | Real commodity prices |
| Boudart (2012) | 42 commodity dependent countries | Monthly: 1980-2009 | Panel cointegration | 0.2 | Real commodity prices |
| Ricci et al (2013) | 48 industrial and emerging countries | Annual: 1980-2004 | Panel cointergration | Advanced countries (0.8) Emerging markets (0.5) | Real commodity prices |
| Phillips et al (2013) | 40 advanced and emerging countries | Annual: 1990-2010 | Panel OLS(fixed effect) | 0.1 | Real commodity prices |
Empirical Evidence on Commodity Currency
| Author/s (year) | Country/ies | Sample | Method | Elasticity on commodity prices | Definition of commodity prices |
|---|---|---|---|---|---|
| Chen and Rogoff (2003) | Australia, Canada, New Zealand | Quarterly: year varies | Time Series cointegration | Australia (0.4), Canada (0.4), and New Zealand (0.6) | Real commodity prices |
| Cashin et al (2004) | 58 commodity exporting countries, including Peru | Monthly: 1980-2002 | Time Series cointegration | Median=0.4. TOT not important for Peru. | Real commodity prices |
| Ferreyra and Salas (2006) | Peru | Quarterly: 1980-2005 | Time series cointegration | 0.3 | TOT |
| Montiel (2007) | Argentina, Bolivia, Brazil, Chile, Paraguay, Uruguay | Annual: 1969-2005 | Time series cointegration | TOT important only for Argentina (1.7), Bolivia (0.6), and Uruguay(0.6) | TOT |
| Iossifov and Loukoianova (2007) | Ghana | Quarterly: 1984-2006 | Time series cointegration | 0.4 | Real commodity prices |
| Astorga (2012) | Argentina, Brazil, Chile, Colombia, Mexico, Venezuela | Annual: 1900-2000 | Time series cointegration | Argentina (0.4), Brazil (0.2), Chile (0.1), Colombia (0.4), Mexico (not significant), Venezuela (0.1) | TOT |
| Coudert et al (2011) | 52 commodity exporters | Annual: 1980-2007 | Panel cointegration | 0.4 | Real commodity prices |
| Boudart (2012) | 42 commodity dependent countries | Monthly: 1980-2009 | Panel cointegration | 0.2 | Real commodity prices |
| Ricci et al (2013) | 48 industrial and emerging countries | Annual: 1980-2004 | Panel cointergration | Advanced countries (0.8) Emerging markets (0.5) | Real commodity prices |
| Phillips et al (2013) | 40 advanced and emerging countries | Annual: 1990-2010 | Panel OLS(fixed effect) | 0.1 | Real commodity prices |
References
Adler, Gustavo and Nicolas E. Magud, 2013, “Four Decades of Terms-of-Trade Booms: Saving-Investment Patterns and a New Metric of Income Windfall,” IMF Working Paper No. 13/103 (Washington: International Monetary Fund).
Aizenman, Joshua, Sebastian Edwards, and Daniel Riera-Crichton, 2011, “Adjustment Patterns to Commodity Terms of Trade Shocks: the Role of Exchange Rate and International Reserves Policies,” NBER Working Paper No. 17692 (Cambridge, Massachusetts: National Bureau of Economic Research).
Astorga, Pablo, 2012, “Mean Reversion in Long-Horizon Real Exchange Rates: Evidence from Latin America,” Journal of International Money and Finance, Vol. 31, pp. 1529–50.
Armas, Adrian and Francisco, Grippa, 2005, “Targeting Inflation in a Dollarized Economy: The Peruvian Experience,” IADB Working Paper 538 (Washington: Inter-American Development Bank).
Armas, Adrian, Paul Castillo and Marco Vega, 2014, “Inflation Targeting and Quantitative Tightening: Effects of Reserve Requirements in Peru,” Banco Central de Reserva del Peru Working Paper, No. 2014–003.
Balassa, Bela, 1964, “The Purchasing Parity Doctrine: A Reappraisal.” Journal of Political Economics, Vol. 72, No. 6, pp. 584–96.
Bodart, V., B. Cabdelon, and J.F. Carpantier, 2012, “Real Exchange Rates in Commodity Producing Countries: A reappraisal,” Journal of International Money and Finance, Vol. 31, pp. 1485–1502.
Cashin, Paul, Luis Felipe Cespedes, and Ratna Sahay, 2004, “Commodity Currencies and the real Exchange Rate.” Journal of Development Economics, Vol. 75, No. 1, pp. 239–268.
Chen, Yu-chin, and Kenneth S. Rogoff, 2003, “Commodity Currencies,” Journal of International Economics, Vol. 60, No. 1, Special Issue, pp. 133–160.
Coudert, Virginie, Cecile Couharde, and Valerie Mignon, 2011, “Does Euro or Dollar Pegging Impact the real Exchange Rate? The Case of Oil and Commodity Currencies,” World Economy, Vol. 34, No. 9, pp.1557–1592.
Engel, Charles, 2000, “Long-run PPP may not hold After all,” Journal of International Economics Vol. 57, pp. 243–73.
Ferreyra, Jesus, and Jorge Salas, 2006, “Tipo de cambio real de equilibrio en el Peru: modelos BEER y construccion de bandas de confianza,” Banco Central de Reserva del Peru, Working Paper, No. 2006-006.
Froot, Kenneth and Kenneth Rogoff, 1995, “Perspectives on PPP and Long-run Real Exchange Rates.” in Gene Grossman and Kenneth Rogoff, eds. (The Handbook of International Economics).
Gregory, Allan W., and Bruce E. Hansen, 1996, “Resdiual-Based Tests for Cointegration in Models with Regime Shifts,” Journal of Econometrics, Vol. 70, pp. 99–126.
Hinkle, Lawrence E. and Peter J., Montiel, 1999, “Exchange Rate Misalignment and Measurement for Developing Countries,” A World Bank Research Publication. (New York: Oxford University Press.)
Iossifov, Plamen and Elena Loukoianova, 2007, “Estimation of a Behavioral Equilibrium Exchange Rate Model for Ghana,” IMF Working Paper 07/155 (Washington: International Monetary Fund).
Meese, Richard A. and Rogoff, Kenneth, 1983, “Empirical Exchange rate Models of the Seventies: do they fit out of Sample,” Journal of International Economics Vol. 14, pp. 3–24.
Montiel, Peter J., 2007, “Equilibrium Real Exchange Rates, Misalignment and Competitiveness in the Southern Cone,” CEPAL macroeconomia del desarrollo, Serie 62.
Phillips, Steven, Luis Catão, Luca Ricci, Rudolfs Bems, Mitali Das, Julian Di Giovanni, D. Filiz Unsal, Marola Castillo, Jungjin Lee, Jair Rodriguez and Mauricio Vargas Ricci, 2013, “The External Balance Assessment (EBA) Methodology,” IMF Working Paper 13/272 (Washington; International Monetary Fund).
Ricci, Luca Antonio, Gian Maria Milesi-Ferretti, and Jaewoo Lee, 2013, “Real Exchange Rates and Fundamentals: a Cross-Country Perspective,” Journal of Money, Credit and Banking, Vol. 45, No. 5, pp. 845–65.
Rogoff, Kenneth, 1996, “The Purchasing Power Parity Puzzle.” Journal of Economic Literature, Vol. 34, pp. 647–68.
Rossini, Renzo, Adrian Armas, and Zenon Quispe, 2014, “Global Policy Spillovers and Peru’s Monetary Policy: Inflation Targeting, Foreign Exchange Intervention and Reserve Requirements,” BIS Papers, No. 78.
Samuelson, Paul, 1964, “A Theoretical Note on Trade Problems,” Review of Economics and Statistics, Vol. 46, No. 2, pp. 145–54.
Tashu, Melesse, 2014, “Motives and Effectiveness of Forex Interventions: Evidence from Peru,” IMF Working Paper 14/217 (Washington: International Monetary Fund).
Prepared by M. Tashu.
The terminologies ‘real exchange rate’ and ‘real effective exchange rate’, both of which refer to the exchange rate of the nuevo sol against a basket of currencies of major trading partner countries adjusted for price differentials between Peru and trading partner countries, are used interchangeable in this study.
Metal exports represent about 55 percent of Peru’s total export receipts.
The test for linear cointegration in the annual sample yielded no cointegration with coefficients sensitive to changes in specification. The Gregory-Hansen cointegration test with a regime shift shows evidence of non-linear cointegration with a regime shift in 1987 at the 10 percent level (Appendix Table 3a). Following this result, a dummy was created for this structural shift and the non-linear cointegration relationship was estimated using FMOLS with LRP_COM and LRP_COM interacted with a dummy for a structural shift on the right hand side. The Wald restriction test for the sum of the coefficients equals zero could not be rejected at any level of significance (Appendix Table 4c).
While the inflation targeting framework was introduced in 2002, the monetary targeting framework, which was in place prior to 2002, is also credited to have reduced and stabilized inflation from the 1980s hyperinflation.
Complementary fiscal policy and the use of reserve requirements have helped the BCRP sustain its sterilized FX interventions without compromising the health of its balance sheet. For instances, about 37½ percent and 34½ percent of the FX intervention in 2013 was sterilized by public sector deposits and reserve requirements, respectively, and only about 11½ percent of the intervention was sterilized through central bank instruments (Rossini et al, 2014). In this regard, the positive commodity price shock, which increased tax revenues from the mineral sector, has helped the Treasury to provide support to the central bank’s sterilization effort.
The NIR used here excludes valuation effects so that changes in NIR reflect mostly of FX interventions and other measures aimed at containing exchange rate volatility such as changes in reserve requirements on foreign currency liabilities.
All of the variables have unit root (Appendix Table 1). The Augmented-Dickey-Fuller (ADF) test seems to suggest that LNIR is I(0) when constant or constant and trend are added. But the ADF test is known to have low power; i.e., has the tendency to reject the null hypothesis of I(1) too often when it is true. The more efficient unit root test, the Dickey-Fuller GLS (DF-GLS) test, however, accepts the null hypothesis at all levels of significance, suggesting that the NIR is I(1). Johansen’s Trace and Maximum Eigenvalue cointegration tests show the presence of a statistically significant cointegration vector among the variables in each of the three equations.
The half-life of a shock to the REER is estimated at about 5 quarters. The coefficient on the error correction term in the dynamic equation is -0.13 and is statistically significant at 1 percent, implying that about 13 percent of deviations of the real exchange rate from the long run equilibrium would be corrected after one quarter. Both productivity and government consumption are also significant in the short run dynamic model (equation (12)), the latter with an unexpected negative sign. The half-life of a shock to the REER is calculated as log(0.5)/log(1-0.13).
The results for the annual data are obtained following the procedure described above; i.e. testing for cointegration with regime shift using Gregory-Hansen’s test and estimating the long-run relationship using non-linear FMOLS (Appendix Tables 3b, 4b and 4c). In this case, the break was identified as 1988/89.
