Accepted Manuscript Title: Is the yen misaligned more during the Abenomics period? Author: SaangJoon Baak PII: DOI: Reference: S0922-1425(17)30056-7 https://doi.org/10.1016/j.japwor.2017.10.004 JAPWOR 904 To appear in: JAPWOR Received date: Revised date: Accepted date: 18-6-2017 23-10-2017 23-10-2017 Please cite this article as: Baak, SaangJoon, Is the yen misaligned more during the Abenomics period?.Japan and the World Economy https://doi.org/10.1016/j.japwor.2017.10.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Is the yen misaligned more during the Abenomics period? SaangJoon Baak Professor School of International Liberal Studies Waseda University, 1-6-1 Nishi-Waseda, Shinjuku-ku, Tokyo 169-8050, Japan Email: [email protected] Oct 23, 2017 Highlights Is the yen misaligned more during the Abenomics period? We estimate the BEER model of the Japanese real effective exchange rate (REER). The misalignments of the Japanese REER are computed based on the estimation results. The yen was substantially overvalued during the global financial crisis period. The yen was undervalued during the Abenomics period. The yen was much less misaligned during the Abenomics period than in pre-Abenomics periods. Abstract This paper measures the extent to which the real effective exchange rate of the Japanese yen is misaligned from its equilibrium value. The equilibrium value is estimated using the behavioral equilibrium exchange rate (BEER) approach to determine whether the yen is more misaligned during the Abenomics period. Economic fundamentals such as terms of trade, relative price of nontraded to traded goods, net foreign asset ratio over trade volume, and real interest rate differentials are used to assess the equilibrium exchange rate. Because parameter instability is detected for the whole period (1993 Q1 to 2016 Q2), stable estimation results for the subsample period (2003 Q1 to 2016 Q2) are used to compute misalignments. The results indicate that the yen was substantially overvalued during the global financial crisis period (2008 Q4 to 2012 Q4). In contrast, it was undervalued during the Abenomics period 0 (2013 Q1 to 2016 Q2) but was much less misaligned than in pre-Abenomics periods. JEL Classification: C22, F31 Keywords: Japanese yen, misalignment, behavioral equilibrium exchange rate, Abenomics 1 1. Introduction After the Japanese government launched an aggressive quantitative easing (QE) policy as part of Abenomics in 2013, the value of the yen against the US dollar depreciated by about 56% within only three years. The quarterly average exchange rate of the Japanese yen changed from 78.6 yen per dollar in the third quarter of 2012 to 122.3 yen in the third quarter of 2015. Because the currency values of Japan’s major trading partners changed less drastically during the same period, the real effective exchange rate of the yen also depreciated sharply. The sharp depreciation of the yen has provoked concerns among neighboring countries that compete with Japan in the global market. Their major concern is that cheaper Japanese products might crowd out their own products in the global market.1 The recent depreciation of the yen, however, does not necessarily imply that the Japanese yen is substantially undervalued compared to the country’s economic fundamentals. Instead, it may be returning to its equilibrium value following a sharp appreciation during the global financial crisis (2008–2011). Figure 1-1 shows that the yen depreciated against the US dollar from 2012 to 2015 almost as much as it appreciated from 2007 to 2011. It is true that the yen depreciated drastically beginning in 2012 while the currencies of its neighboring countries remained relatively steady. However, the figure also shows that the yen appreciated much faster than the Chinese yuan or the Korean won between 2007 and 2011. Because the value of the yen shows much greater volatility than Japan’s economic fundamentals from 2008 to 2016, it is highly probable that the yen is misaligned from the country’s economic fundamentals. Even so, the extent to which the yen is under- or overvalued has rarely been quantified in literature. Against this background, this paper aims to examine whether and to what extent the real effective exchange rate of the Japanese yen is under- or overvalued compared to Japan’s economic fundamentals. A special focus is given placed on the recent period to determine whether the yen is more misaligned since the implementation of Abenomics. Toward this end, this paper estimates the equilibrium value of the Japanese yen by using the 1 Bahmani-Oskooee and Hegerty (2009) reported that depreciation of the yen improved Japan’s trade balance for about one-third of Japanese industries both in the short-term and in the long-term. In addition, Baak (2008, 2014) showed that depreciation of an Asian currency negatively affected the exports of its neighbors. 2 behavioral equilibrium exchange rate (BEER) approach proposed by Clark and MacDonald (1998, 1999). The BEER approach explains the dynamics of an exchange rate by economic fundamentals related to the exchange rate and interprets unexplained movements of the exchange rate as misalignments. Although it is generally accepted that predicting the future values of an exchange rate is impossible, a rich body of research papers—whether employing the BEER method or not—has shown that fundamental variables such as terms of trade, real interest rate differentials, and net foreign assets are robustly linked to exchange rate dynamics.2 The findings of these papers have enabled researchers to apply the BEER method to various countries and regions by adopting appropriate economic fundamentals as explanatory variables. For example, Coudert et al. (2013) and El-Shagi et al. (2016) applied the BEER model to Euro member countries. Coudert et al. (2013) analyzed the period from 1980 to 2010 and reported that the exchange rates of peripheral member countries had become seriously overvalued since the mid-2000s. Similarly, El-Shagi et al. (2016), who analyzed the period from 1980 to 2012, reported that the exchange rates of peripheral member countries were significantly overvalued when the global financial crisis occurred in 2008. A substantial number of papers have adopted the BEER approach to measure the misalignments of Asian currencies. Examples include Funke and Rahn (2005), Wang et al. (2007), and Zhang and Chen (2014) for the Chinese yuan; Koske (2008) for the Malaysian ringgit; and Kinkyo (2008) and Baak (2012) for the Korean won. Despite its use worldwide, the BEER model has been used very limitedly for the Japanese yen. Although the yen was included in research by Bénassy-Quéré and LahrècheRévil (2008), Ricci et al. (2013), and Adler et al. (2017), these papers focus mainly on determining economic fundamentals that explain exchange rate dynamics rather than quantifying misalignments of the yen. Of interest is that in the first BEER paper by Clark and MacDonald (1998), the authors applied their model to Japan’s time series data for the period of 1960 to 1996 and claimed that their model explained the dynamics of the yen fairly well. To the best of our knowledge, however, no paper has analyzed Japanese time series data for the 2000s using the BEER model. In fact, misalignments of the yen are rarely quantified in literature, despite the hot debates surrounding the appropriate value of 2 See Engle et al. (2007), Ricci et al. (2013), and Adler et al. (2017), for example. 3 the currency.3 One conjecture regarding such limited use of the BEER model for the yen is that the model’s performance may not be satisfactory when applied to the post-bubble period data of the yen. Some preliminary estimation experiments in this paper confirmed this speculation. As will be shown later in the paper, if the typical BEER model is employed for the yen for the period of 1993 Q1 to 2016 Q2, the model does not perform as well as in Clark and MacDonald (1998). However, by considering a structural change, and by adopting the suggestion of Bleaney and Tian (2014) to modify the BEER equation, this paper shows that the BEER model applied to the yen’s time series data produces estimation results that are quite consistent with economic theories. Since the Hansen (1992) stability test indicates parameter stability of the BEER model for the period of 2003 Q1 to 2016 Q2, the misalignments of the yen are quantified for that period. Results show that the yen was undervalued from 2005 Q4 to 2008 Q3, overvalued from 2008 Q4 to 2012 Q4 (global financial crisis period), and undervalued again from 2013 Q1 to 2016 Q2 (Abenomics period). However, the misalignments of the yen since 2013 have not been as severe as those in the periods of 2005 Q4 to 2008 Q3 and 2008 Q4 to 2012 Q4. The following section describes the BEER approach, the variables employed in the model, how the variables are computed, and how the model differs from typical BEER models. Section 3 reports the estimation results and shows how the performance of the model is improved by the modification explained in section 2 and structural break tests. Section 4 measures misalignments of the yen using the estimation results. Section 5 concludes the paper. 2. The model Based on the interest parity condition, the BEER approach proposed by Clark and MacDonald (1998) derives the following reduced form equation: q t = β′ Z t 3 (1) According to Aminian et al. (2012), the Japanese yen was the first Asian currency whose appropriate value provoked an international debate. 4 where q t is the real equilibrium exchange rate expressed as the foreign currency price of a unit of domestic currency4, Zt is a vector of economic fundamentals, and β is a vector of coefficients. Based on theoretical and empirical developments in the field, Clark and MacDonald (1998) proposed using the following five variables in Zt : relative terms of trade (TOT), a variable to capture the Balassa-Samuelson effect (TNT), ratio of net foreign assets (NFA) over GDP, real interest rate differential (IRD), and country risk premium. However, because literature shows that no economic variable is strongly supported as a proxy for the country risk premium (Chionis and MacDonald, 2002), this paper does not include the country risk premium as an explanatory variable.5 The four variables mentioned above (TOT, TNT, NFA, and IRD) are widely supported by literature as economic fundamentals related to exchange rate dynamics. For example, Bussière et al. (2010), Ricci et al. (2013), and Adler et al. (2017) confirmed in their panel data analyses that TOT, TNT, and NFA are significant and robust explanatory variables. IRD is more often included in time series data analysis than in panel data analysis (Kinkyo, 2008; Baak, 2012). Kitamura and Akiba (2006) showed from their analysis of exchange rates and interest rates from 1999 to 2002 that the yen-dollar exchange rate is affected by interest differentials between the countries. Among the variables, TNT is used to capture the Balassa-Samuelson effect. Typically used proxies for the effect are relative price of nontraded to traded goods (Clark and MacDonald, 1998; Kinkyo, 2008; Baak, 2012), GDP per capita (Koske, 2008; Bussière et al., 2010; Adler et al., 2017), or labor productivity in nontraded and traded goods sectors (Ricci et al., 2013). This paper adopts the relative price of nontraded to traded goods as the proxy, in line with Clark and MacDonald (1998), for the following two reasons. First, the analysis of Ricci et al. (2013) implies that GDP per capita may not be an appropriate proxy for the Balassa-Samuelson effect. Second, this paper analyzes quarterly data to examine the misalignment of the yen since Abenomics, but cross-country labor productivity data are available only on an annual basis. This paper also includes the Euro area as one of Japan’s major trading partners, and the labor productivity data for the Euro area are not available.6 4 Therefore, a decrease in the exchange rate means a depreciation of the Japanese yen. 5 In fact, the risk premium is rarely included in Zt in the recent BEER literature. 6 By including the Euro area, the major trading partners of Japan involved in the computation of variables occupy more than 65% of Japan’s trade. 5 According to Balassa (1964) and Samuelson (1964), the real exchange rate should be negatively related to the relative productivity of the non-tradable goods sector to the tradable goods sector. Since the relative productivity between the two sectors is negatively connected to their relative price, the relative price of nontraded to traded goods is expected to have a positive relationship with the real exchange rate. In the meantime, different from Clark and MacDonald (1998), this paper replaces the ratio of NFA over GDP with the ratio of NFA over trade volume following the arguments of Bleaney and Tian (2014). Bleaney and Tian (2014) showed that if the NFA is scaled by the GDP, the valuation effects generate biases in the estimation because the measure of the NFA/GDP ratio is affected by the real exchange rate. To resolve this problem, they suggested that the NFA should be scaled by the trade volume. Following their suggestion, this paper employs the ratio of NFA to Japan’s trade volume (NFAT). As shown in the following section, this change substantially improves the model’s estimation performance. Accordingly, the specific form of the BEER equation estimated in this paper is as follows: LQ = β0 + β1 LTNT + β2 LTOT + β3 IRD + β4 NFAT + ε (2) where LQ is the log value of Japan’s real effective exchange rate, LTOT is the log value of TOT, and LTNT is the log value of TNT. IRD is the interest rate differential between Japan and the US, and NFAT is NFA over trade volume.7 The residuals of the equation are interpreted as misalignments from the equilibrium values. In literature, this deviation from the fitted values is called the “current misalignment.” In addition, literature on BEER distinguishes the current equilibrium exchange rate (q t ) determined from the current values of economic fundamentals (Zt ), from the long-run equilibrium exchange rate (q̅t ) determined from the long-run values of economic fundamentals (Z̅t ). Practically, the long-run values of economic fundamentals in the BEER literature are obtained using the Hodrick-Prescott filter.8 The deviation of the actual exchange rate from the long-run equilibrium exchange rate is called the “total 7 More detailed definitions of the variables are written later. 8 Clark and MacDonald (1998, 1999), Kinkyo (2008), and Baak (2012), among others, used the Hodrick- Prescott filter to obtain the long-run equilibrium exchange rates. 6 misalignment.” Both the current and the total misalignment are calculated in section 4. Variables and data9 This paper estimates the BEER equation using quarterly data from the period 1993 Q1 to 2016 Q2. The starting point, 1993 Q1, was selected considering data availability and the fact that the Japanese bubble burst in the early 1990s. The real effective exchange rate is CPI based and is obtained from the weighted geometric average of the indices of the bilateral real exchange rates between Japan and its twelve major trade partners. These twelve partners—Australia, Canada, China, Hong Kong, Indonesia, South Korea, Malaysia, Singapore, Thailand, the UK, the US, and the Euro area—were selected based on their shares in Japanese trade. Their total share in Japanese trade never decreased below 65% during the period covered in this research. Figure 1-2 illustrates the Japanese real effective exchange rates computed in the paper along with those reported by the IMF and the BIS, confirming that the three series are almost identical. TOT is the terms of trade of Japan divided by the weighted average of those of the twelve partner countries. A country’s terms of trade consist of the ratio of its export unit value to its import unit value. The terms of trade are known to have two contradicting effects on the equilibrium exchange rate.10 Therefore, β1 may be either positive or negative. TNT is calculated by dividing the Japanese ratio of CPI over PPP by the weighted average of the same ratios of the twelve countries. As explained previously, this variable is included to capture the Balassa-Samuelson effect. Therefore, β2 is expected to be positive. IRD is the differential of the real interest rate between Japan and the US. The real interest rate is defined as the lending rate minus the CPI-based inflation rate. Because no interest rate data are available for all the countries involved, the US rate is used as a representative foreign yield, considering the impact of the country on international financial 9 10 The data used in the paper can be obtained from the authors upon request. See Kinkyo (2008). 7 transactions.11 An increase in the real interest rate differential (domestic minus foreign rate) induces currency appreciation. Therefore, β3 is expected to be positive. Finally, NFAT is the ratio of Japan’s net foreign asset to its trade volume (export plus import). If the net foreign asset decreases, the real exchange rate should depreciate to generate a trade surplus; this is needed to finance more interest payments induced by the decline in net foreign assets. Therefore, β4 is expected to be positive. Data sources Most data were collected from the International Financial Statistics (IFS) of the IMF. The Euro exchange rates were collected from both the IFS and the Euro Stat. The unit value of exports and the unit value of imports of the Euro Area, Indonesia, and Malaysia were obtained from the Data Stream. The unit value of exports and the unit value of imports of Korea were obtained from the Bank of Korea. 3. Estimation results Unit root tests Because conventional unit root tests such as the ADF test may fail to detect nonstationarity when a non-stationary series has a structural break, as Perron (2006) discusses, and because Japan’s economic variables are often suspected to have structural breaks, this paper performs the S-L unit root test suggested by Saikkonen and Lutkepohl (2002), which is robust in the presence of a structural break. As reported in Table 1, the null hypothesis of a unit root is accepted at the 5% significance level for the levels of all the variables. In addition, it should be noted that the S-L tests with the first differences, which are not reported in the paper, strongly indicate stationarity for all the variables involved.12 11 In preliminary estimations, the weighted average of the foreign lending rates was calculated using available data. When the weighted average rate replaces the US rate, the estimation coefficient values of equation (2) were altered only marginally. However, the coefficient of IRD becomes insignificant. 12 The S-L unit root test here and the S-L cointegration test below are implemented by the software JMulTI. 8 <Insert Table 1> Cointegration tests Considering the possibility of any structural changes in the relationship among the variables in equation (2), this paper performs the S-L cointegration test (Saikkonen and Lutkepohl, 2000a, 2000b, 2000c), which is robust to a structural break in the long-term relationship. The test results reported in Table 2 indicate the presence of a long-term relationship among the variables at the 5% significance level. <Insert Table 2> Estimation results Because the cointegration test indicates the presence of a cointegrating vector among the variables in equation (2), the cointegrating vector is estimated by the fully modified OLS (FMOLS) of Phillips and Hansen (1990) and the canonical cointegrating regression (CCR) of Park (1992). Tables 3-1 and 3-2 list the estimation results. For a comparison, NFA over trade volume (NFAT) is used as an explanatory variable in Table3a, following Bleaney and Tian (2014) by the logic explained in section 2. In contrast, NFA over GDP (NFAG) replaces NFAT in Table3b. <Insert Table 3a and 3b> In Table3a, when equation (2) is estimated for the whole period (1993 Q1 to 2016 Q2), NFAT has significant and negative coefficients, thus contradicting economic theories, while the signs of other explanatory variables are consistent with what is expected. LTNT is not significant at the 5% significance level, but the other variables are all significant. In the meantime, the null hypothesis of parameter stability is rejected at the 5% significance level by the Hansen (1992) stability test. In addition, the null hypothesis of no cointegration of the Phillips and Ouliaris (1990) test is accepted even at the 10% significance level. These test results strongly indicate that a structural change has occurred. Accordingly, estimations and stability tests were performed using various sub-sample period data. Specifically, we 9 shortened the sample period by eliminating earlier years one by one until the p-value of the Hansen stability test was higher than 10 percent. Table 4 reports the Hansen stability test results for various sub-sample periods, showing the stability test is accepted when the sample period starts from 2001. Because the Hansen stability test statistic is lowest when the sample period is restricted to 2003 Q1~ 2016 Q2, Table3a reports the estimation results for that subsample period. However, it should be reported that moving the starting point of the sub-sample period from 2003 Q1 by a few quarters back and forth changes the estimation results only marginally. The present paper does not explore why there was a structural break in the early 2000s. However, it should be noted that Fukuda (2016) also reported a change in the dynamics of the yen between the 1990s and the 2000s. In addition, the volatility of inward and outward portfolio investment of Japan drastically increased from 2003 and 2004, respectively. Considering that the effects of NFAT and IRD on a country’s currency value will get stronger when the country is more open to capital flow, it does not seem surprising that the break point is found in the early 2000s in Japan. When the period is restricted to 2003 Q1 to 2016 Q2, all coefficients including NFAT show the expected signs and are significant even at the 1% significance level. Furthermore, the Phillips and Ouliaris (1990) test reconfirms the presence of cointegration, and the null hypothesis of parameter stability is accepted because the p-value of the Hansen (1992) stability test is higher than 0.2. These results indicate that the sub-sample period of 2003 Q1 to 2016 Q2 does not suffer a structural change in the regression equation. The high R-square, 0.85, implies that the BEER model explains a fairly large part of the variation of the Japanese real effective exchange rate. On the other hand, when NFAT is replaced by NFAG in Table3b, the estimated coefficient values have unexpected signs or are insignificant except for IRD for both the whole sample period and the sub-sample period. Different from NFAT, the sign of NFAG is still negative in the sub-sample period, even though overall performance of the model seems improved in the sub-sample period. In fact, IRD is the only variable that has the right sign and is significant. The poor performance of equation (2) when net foreign asset is scaled by GDP can be regarded as empirical evidence that supports the argument of Bleaney and Tian (2014). In summary, the BEER model performs much better when net foreign asset is scaled by trade volume rather than GDP. Even so, the regression results of the whole sample 10 period do not pass stability tests. The regression results of the sub-sample period do pass the stability test, however. In addition, estimated coefficient values are all significant and consistent with the prediction of economic theories. Therefore, the misalignments of the Japanese yen will be quantified by plugging the estimated parameter values for the period from 2003 Q1 to 2016 Q2, as reported in Table3a, into the BEER model equation (2). Specifically, the parameter values estimated by the FMOLS (column 3 in Table3a) are plugged into equation (2) in the following section. Because the CCR estimation results (column 4 in Table3a) are very similar to those of FMOLS, the misalignments computed by both illustrate only negligible differences. 4. Measuring misalignments As explained in section 2, the behavioral equilibrium exchange rate (BEER) is calculated by equation (2) using estimated coefficient values and actual data of the four explanatory variables. On the other hand, when the long-run behavioral equilibrium exchange rate (LBEER) is calculated, actual data of the four variables are replaced by their long-run values that can be obtained by the Hodrick-Prescott filter. Figures 2a through 2d illustrate the four explanatory variables in equation (2) along with their long-run trend obtained by the Hodrick-Prescott filter. As the figures show, actual values deviated the most from their long-run values in general during the global financial crisis. The great volatility of the explanatory variables implies that they are very much influenced by temporary shocks. These facts confirm the importance of calculating the LBEER because the impacts of temporary shocks are eliminated from the data used to calculate it. <Insert Figures 2a through 2d> Figure 3 shows the BEER and the LBEER along with the actual real effective exchange rate (REER). The BEER is moving around the REER without a big margin, implying minor misalignments. Even during the global financial crisis in which the yen maintained relatively high values, the actual exchange rate (REER) was lower than the equilibrium exchange rate (BEER) from time to time, implying that the yen was not continually overvalued in that time period. Also, during the Abenomics period in which the 11 value of the yen was relatively lower, the actual exchange rate (REER) was sometimes higher than the equilibrium exchange rate (BEER), implying that the yen was not continually undervalued in that time period. The over- and undervaluation of the yen should not be determined only by the BEER, however, because the equilibrium values are not free from unexpected temporary shocks of economic fundamentals. Therefore, the actual exchange rate should be compared to the long-run equilibrium exchange rate (LBEER), which is unaffected by those temporary shocks. Figure 3 shows that the actual exchange rate was higher than the LBEER during the post-Lehman shock and pre-Abenomics period (2008 Q4 to 2012 Q4), and that it was lower than the LBEER during the Abenomics period (2013 Q1 to 2016 Q2). This implies that the yen was overvalued during the global financial crisis period and undervalued during the Abenomics period. Considering that the yen was believed to be a relatively safe asset during the global financial crisis, and that the Bank of Japan did not implement aggressive monetary policies in that period while the Fed, the European Central Bank, and the Bank of England did, it is not surprising that the yen was overvalued in that period. On the other hand, as Fukuda (2015) shows, foreign investors responded to Abenomics by aggressively selling the Japanese yen in the foreign currency market. Their behaviors should be one of the main forces that lead to the undervaluation of the yen during the Abenomics period. While the Japanese yen was overvalued during the global financial crisis, according to Baak (2012), who examined the misalignments of the Korean won from 1982 to 2009, the currency was substantially undervalued in 2008–2009. Zhang and Chen (2014) analyzed the misalignments of the Chinese renminbi from 1980 to 2012, and concluded that the currency was overvalued by around 10 percent in 2011–2012. However, according to them, the renminbi was not substantially overvalued before 2011. Figure 4 illustrates the specific magnitudes of the misalignments. The current misalignments are computed by dividing the difference between the REER and the BEER by the BEER—that is, (REER – BEER)/BEER. The misalignments are then transformed into percentage terms. The total misalignments are computed in the same way by replacing the BEER with the long-run BEER. That is, (REER – LBEER)/LBEER. <Insert Figures 3 and 4> 12 As could be predicted from Figure 3, the current misalignments measure much smaller than the total misalignments. According to the current misalignments, the actual exchange rate of the Japanese yen was overvalued by 1.9% on average for the period of 2008 Q4 to 2012 Q4 (the global financial crisis period). It was near-equilibrium (or overvalued by 0.1%) from 2013 Q1 to 2016 Q2 (the Abenomics period). It was the most overvalued in 2008 Q4 (20.8%), when the Lehman shock hit the world, and the most undervalued in 2013 Q1 (7.1%), right before implementation of the QQE (Qualitative and Quantitative Monetary Easing) of Bank of Japan.13 However, according to the total misalignment, the Japanese yen was overvalued by 6.9% on average from 2008 Q4 to 2012 Q4 and was undervalued by 3.0% on average from 2013 Q1 to 2016 Q2. It was the most overvalued in 2012 Q3 (13.2%) and the most undervalued in 2014 Q4 (7.8%). The yen was undervalued in the pre-Lehman shock period (2005 Q4 to 2008 Q3) as well. In addition, the undervaluation in that period was much more substantial than in the Abenomics period, as the yen was undervalued by 8.1% on average in the pre-Lehman shock period. Of particular interest is that the yen was overvalued in 2016. As seen in Figure 3, the real effective exchange rate of the yen began to appreciate in 2016; the appreciation can be explained by the movements in economic fundamentals depicted in Figures 2a through 2d. However, because the long-run values (the Hodrick-Prescott trend) of the economic fundamentals are lower than the actual values, the long-run equilibrium exchange rate was measured to be much lower than the actual exchange rate. This, then, is why the yen is estimated to be overvalued in 2016. Overall, whether according to the current misalignments or the total misalignments, the yen was misaligned from its equilibrium value with a much smaller margin during the Abenomics period than during the pre-Abenomics period. 4. Conclusions This paper measures the extent to which the real effective exchange rate of the Japanese yen was misaligned from its equilibrium value. The equilibrium value was estimated using the BEER approach to determine whether the yen is more misaligned since 13 The second Abe administration started in December 2012, and the QQE of Bank of Japan started in April 2013. 13 the implementation of Abenomics than before. Twelve countries and regions were chosen as Japan’s major trading partners. Then, the real effective exchange rate of Japan was computed using the nominal exchange rates of the countries involved and their consumer price indices. The economic fundamentals, which are used as explanatory variables in the equation in which the real effective exchange rate is the dependent variable, are the terms of trade, the relative price of nontraded to traded goods, the real interest rate differential between Japan and the US, and the net foreign assets of Japan relative to its trade volume. While the estimated coefficient values for the period of 1993 Q1 to 2016 Q2 were unstable, those estimated for the period of 2003 Q1 to 2016 Q2 were found to be stable and consistent with economic theories. The total and current misalignments calculated for 2003 Q1 to 2016 Q2 indicate that the yen was undervalued with relatively small deviation from its equilibrium value during the Abenomics period. On the contrary, it was substantially undervalued in the pre-Lehman shock period (2005 Q4 to 2008 Q3) and substantially overvalued during the global financial crisis period (2008 Q4 to 2012 Q4). According to the total misalignment measures, the yen was overvalued by 6.9% on average from 2008 Q4 to 2012 Q4 and was undervalued by 8.1% on average from 2005 Q4 to 2008 Q3. In contrast, it was undervalued only by 3.0% on average from 2013 Q1 to 2016 Q2. These results indicate that the yen was misaligned from its equilibrium value with a much smaller margin during the Abenomics period than in the pre-Abenomics period. Funding: This work was supported by JSPS KAKENHI, grant number 25380337. Acknowledgements: I would like to thank two anonymous reviewers, participants of the 2015 KAEA annual meeting and the 2015 World Congress of Comparative Economics for their invaluable comments and suggestions for the previous version of this paper. Any remaining errors are mine. 14 References 1. Adler, K., Grisse, C., 2017. Thousands of BEERs: Take your pick. Review of International Economics, 1-27. 2. Aminian, A., Fung, K.C., Garcia-Herrero, A., Lin, C., 2012. The political economy of exchange rates: The case of the Japanese yen. Japan and the World Economy, 24, 193196. 3. Baak, S., 2008. The bilateral real exchange rates and trade between China and the US, 4. 5. 6. 7. 8. 9. 10. 11. China Economic Review, 19, 117-127. Baak, S., 2012. Measuring misalignments in the Korean exchange rate. Japan and World Economy, 24, 227-234. Baak, S., 2014. Do Chinese and Korean products compete in the Japanese market? An investigation of machinery exports. Journal of the Japanese and International Economies, 34, 256-271. Bahmani-Oskooee, M., S., Hegerty, 2009. The Japanese-US trade balance and the yen: Evidence from industry data. Japan and the World Economy, 21, 161-171. Balassa, B., 1964. The purchasing power parity doctrine: A reappraisal. Journal of Political Economy 72, 584-596. Bénassy-Quéré, A., Lahrèche-Révil, A., 2008. Is Asia responsible for exchange rate misalignments within the G20? Pacific Economic Review, 13(1), 46-61. Bleaney, M., Tian., M., 2014 Net foreign assets and real exchange rates revisited. Oxford Economic Papers, 66, 1145-1158. Bussière, M., Ca’ Zorzi, M., Chudik, A., and Dieppe, A. (2010). Methodological advances in the assessment of equilibrium exchange rates (ECB Working Paper No. 1151). Frankfurt: European Central Bank. Chionis, D., MacDonald, R., 2002. Aggregate and disaggregate measures of the foreign exchange risk premium. International Review of Economics and Finance, 11, 57-84. 12. Clark, P., MacDonald, R., 1998. Exchange rates and economic fundamentals: A methodological comparison of BEERs and FEERs. IMF working paper, WP/98/67, International Monetary Fund. 13. Clark, P., MacDonald, R., 1999. Exchange rates and economic fundamentals: A methodological comparison of BEERs and FEERs. In: Stein, J., MacDonald, R. (Eds.). Equilibrium Exchange Rates. Boston: Kluwer. 15 14. Coudert, V., Couharde, C., Mignon, V., 2013. On currency misalignments within the euro area. Review of International Economics, 21(1), 35-48. 15. El-Shagi, M., Lindner, A., Schweinitz, G., 2016. Real effective exchange rate misalignment in the euro area: A counterfactual analysis. Review of International Economics, 24(1), 37-66. 16. Engel, C., Mark, N., West, K., 2007. Exchange rate models are not as bad as you think. NBER working paper 13318, National Bureau of Economic Research. 17. Fukuda, S., 2015. Abenomics: Why was it so successful in changing market 18. 19. 20. 21. expectations? Journal of the Japanese and International Economies, 37, 1-20. Fukuda, S., 2016. On the predictability of daytime and night-time yen/dollar exchange rates. Applied Economic Letters, 23, 618-622. Funke, M., Rahn, J., 2005. Just how undervalued is the Chinese renminbi? World Economy, 465-631. Hansen, B.E., 1992. Tests for parameter instability in regression with I(1) processes. Journal of Business and Economic Statistics, 10, 321-335. Kinkyo, T., 2008. Disorderly adjustments to the misalignments in the Korean won. Cambridge Journal of Economics, 32, 111-124. 22. Kitamura, Y., Akiba, C., 2006. Information arrival, interest rate differentials, and yen/dollar exchange rate. Japan and the World Economy, 18, 108-119. 23. Koske, I., 2008. Assessing the equilibrium exchange rate of the Malaysian ringgit: A comparison of alternative approaches. Asian Economic Journal, vol. 22, no. 2, 179-208. 24. Park, J. Y., 1992. Canonical cointegrating regressions. Econometrica, 60, 119-143. 25. Perron, P., 2006. Dealing with structural breaks. In: Patterson, K, Mills, T.C. (Eds.). Palgrave Handbook of Econometrics, Vol. 1: Econometric Theory, Palgrave Macmillan, 278-352. 26. Phillips, P.C.B., Hansen, B.E., 1990. Statistical inference in instrumental variables regression with I(1) processes. Review of Economic Studies, 57, pp. 99-125. 27. Phillips, P., Ouliaris, S., 1990. Asymptotic properties of residual based tests for cointegration. Econometrica, 58, 165-193. 28. Ricci, L., Milesi-Ferretti, G., Lee, J., 2013. Real exchange rates and fundamentals: A cross-country perspective. Journal of Money, Credit and Banking, 45 (5), 845-865. 29. Saikkonen, P., Lutkepohl, H., 2000a. Testing for the cointegrating rank of a VAR process with an intercept. Econometric Theory 16(3), 373-406. 16 30. Saikkonen, P., Lutkepohl, H., 2000b. Testing for the cointegrating rank of a VAR process with structural shifts. Journal of Business & Economic Statistics, Vol. 18, Iss. 4, p. 451-464. 31. Saikkonen, P., Lutkepohl, H., 2000c. Trend adjustment prior to testing for the cointegrating rank of a vector autoregressive process. Journal of Time Series Analysis 21(4): 435-456. 32. Saikkonen, P., Lutkepohl, H., 2002. Testing for a unit root in a time series with a level shift at unknown time. Econometric Theory 18(2), 313-348. 33. Samuelson, P., 1964. Theoretical notes on trade problems. Review of Economics and Statistics 46, 145-54. 34. Wang, Y., Hui, X., Soofi, A., 2007. Estimating renminbi (RMB) equilibrium exchange rate. Journal of Policy Modeling 29, 417-429. 35. Zhang, Z., Chen, L., 2014. A new assessment of the Chinese RMB exchange rate. China Economic Review 30, 113-122. 17 <Table 1> SL Unit Root Test for the Levels Variable SL Statistic Lag(1) Suggested Break(2) LQ -1.228 -0.867 -1.199 -1.209 -2.657 3 6 5 5 5 2008Q4 2003Q3 2008Q4 2009Q1 1997Q2 LTNT LTOT NFA IRD Notes: (1) The lags were determined by the four criteria used in JMulTI. (2) The breaks reported in the table are those suggested by JMulTI. (3) The 1%, 5%, and 10% critical values are -3.48, -2.88, and -2.58, respectively. <Table 2> Cointegration Tests with a Structural Break r0 r 1 r 1 r2 r2 S-L Statistic 70.14* (p-value) 0.005 Statistic H0: HA: r4 r3 r3 r4 35.12 15.05 4.34 0.34 0.149 0.459 0.664 0.620 r 5 Notes: (1) r denotes the number of cointegrating vectors. (2) The lag length included in the test equation is set to 1 based on the Schwarz criterion. (3) The asterisk (*) indicates the rejection of the null hypothesis of no cointegration at the 5% significance level. 18 <Table 3a> Estimation Results when NFAT is Used Variables Constant LTNT LTOT IRD NFAT R squared Hansen test 1) P-O test2) 1993 Q1 to 2016 Q2 2003 Q1 to 2016 Q2 (1) (2) (3) (4) FMOLS CCR FMOLS CCR 4.767 4.773 4.271 4.272 (0.000) (0.000) (0.000) (0.000) 0.959 0.992 4.366 4.391 (0.108) (0.084) (0.000) (0.000) 0.475 0.467 0.270 0.264 (0.000) (0.000) (0.000) (0.000) 0.025 0.024 0.020 0.020 (0.001) (0.001) (0.001) (0.001) -0.027 -0.028 0.041 0.040 (0.010) (0.009) (0.000) (0.000) 0.779 0.778 0.848 0.849 (0.017) (0.019) (> 0.2) (> 0.2) (0.294) (0.294) (0.040) (0.040) Notes: (1) The null hypothesis of the Hansen (1992) test is that the parameters are stable. (2) The null hypothesis of the P-O test (Phillips and Ouliaris, 1990) is that the variables are not cointegrated. (3) The numbers in parentheses are p-values. 19 <Table 3b> Estimation Results when NFAG is used 1993 Q1 to 2016 Q2 2003 Q1 to 2016 Q2 Variables FMOLS CCR FMOLS CCR Constant 4.935 4.934 4.712 4.642 (0.000) (0.000) (0.000) (0.000) -0.069 -0.037 1.322 1.906 (0.912) (0.950) (0.264) (0.170) 0.003 0.003 0.052 0.124 (0.985) (0.984) (0.763) (0.508) 0.026 0.026 0.043 0.040 (0.000) (0.000) (0.000) (0.000) -0.188 -0.188 -0.067 -0.032 (0.000) (0.000) (0.397) (0.724) R squared 0.823 0.823 0.795 0.794 Hansen test1) (0.027) (0.027) (> 0.22) (> 0.22) P-O test2) (0.146) (0.146) (0.228) (0.227) LTNT LTOT IRD NFAG Notes: See Table 3a. <Table 4> Hansen Stability Test Results Period 1998Q1~ 1999Q1~ 2000Q1~ 2001Q1~ 2002Q1~ 2003Q1~ 2004Q1~ 2016Q2 2016Q2 2016Q2 2016Q2 2016Q2 2016Q2 2016Q2 Lc statistic 1.611 1.648 1.594 0.577 0.589 0.441 0.562 p-value <0.01 <0.01 <0.01 >0.2 >0.2 >0.2 >0.2 20 <Figure 1-1> Exchange Rates against the US Dollar (Index, 2005=100) 140 130 120 110 100 90 80 70 2000 Q1 2000 Q3 2001 Q1 2001 Q3 2002 Q1 2002 Q3 2003 Q1 2003 Q3 2004 Q1 2004 Q3 2005 Q1 2005 Q3 2006 Q1 2006 Q3 2007 Q1 2007 Q3 2008 Q1 2008 Q3 2009 Q1 2009 Q3 2010 Q1 2010 Q3 2011 Q1 2011 Q3 2012 Q1 2012 Q3 2013 Q1 2013 Q3 2014 Q1 2014 Q3 2015 Q1 2015 Q3 2016 Q1 60 China, P.R.: Mainland Japan Korea, Republic of Note: This graph illustrates the indices (2005=100) computed from quarterly average exchange rates against the US dollar. Data Source: IFS <Figure 1-2> Japanese Real Effective Exchange Rates (base year=2010) 160 140 120 100 80 60 1994 Q1 1994 Q4 1995 Q3 1996 Q2 1997 Q1 1997 Q4 1998 Q3 1999 Q2 2000 Q1 2000 Q4 2001 Q3 2002 Q2 2003 Q1 2003 Q4 2004 Q3 2005 Q2 2006 Q1 2006 Q4 2007 Q3 2008 Q2 2009 Q1 2009 Q4 2010 Q3 2011 Q2 2012 Q1 2012 Q4 2013 Q3 2014 Q2 2015Q1 2015Q4 40 REER IMF BIS Note: The REER is the real effective exchange rate computed in the paper. Since the BIS data are available from 1994, the graphs are illustrated from that year, too. Data Sources: IFS and BIS 21 <Figure 2-1> Relative Price of Nontraded Goods to Traded Goods (Log Value, LTNT) 0.04 0.02 0 -0.02 -0.04 -0.06 -0.08 -0.1 -0.12 LTNT HP_LTNT <Figure 2-2> Terms of Trade (Log Value, LTOT) 0.4 0.3 0.2 0.1 0 -0.1 -0.2 LTOT HP_LTOT 22 <Figure 2-3> Net Foreign Assets over Trade Volume (NFTA) 14 12 10 8 6 4 2 0 NFAT HP_NFAT <Figure 2-4> Real Interest Rate Differential (IRD) 3 2 1 0 -1 -2 -3 -4 -5 -6 IRD HP_IRD 23 <Figure 3> REER, BEER, and Long-Run BEER 120 110 100 90 80 70 60 REER BEER Long-run BEER REER = real effective exchange rate; BEER = qt = behavioral equilibrium exchange rate; long-run BEER = q̅t <Figure 4> Misalignments of the Japanese real effective exchange rate (%) 25.0 20.0 15.0 10.0 5.0 0.0 -5.0 -10.0 compared with BEER compared with Long-run BEER 24 2016 Q1 2015 Q3 2015 Q1 2014 Q3 2014 Q1 2013 Q3 2013 Q1 2012 Q3 2012 Q1 2011 Q3 2011 Q1 2010 Q3 2010 Q1 2009 Q3 2009 Q1 2008 Q3 2008 Q1 2007 Q3 2007 Q1 2006 Q3 2006 Q1 2005 Q3 2005 Q1 2004 Q3 2004 Q1 2003 Q3 2003 Q1 -15.0 no misalignment

1/--страниц