The Renminbi Equilibrium Exchange Rate 2005-2015:

(1)

Faculty of Economics and Business

The Renminbi Equilibrium Exchange Rate 2005-2015:

based on the Behavioral Equilibrium Exchange Rate approach

Master thesis

by Jin Huang

(S2858959)

(2)

The exchange rate misalignment of the Renminbi (RMB) has been discussed over the last decades. Many economists have determined a different equilibrium exchange rate for Renminbi during different sample periods, thus the conclusions of whether it is misaligned continue to remain as heated debates. This paper aims to find the equilibrium exchange rate of the RMB against the USD from 2005-2015 using monthly data. To shed light on this problem, we applied the behavioral equilibrium exchange rate approach suggested by Clark and MacDonald. Several tests have been carried out in our research. Finally, our empirical results show that the misalignment problem is not as significant as before, and we conclude that the Renminbi exchange rate is moving to an equilibrium level.

(4)

1. Introduction

(5)

2000. After 2010, the Chinese government allowed the exchange rate against the USD to appreciate to 6.12, and this appreciation process continued over the following years. According to research by The Economist, in 2013 the exchange rate was 35% stronger than at the same time in 2003. By using a different approach which they designed, they concluded that the Renminbi has strengthened over 50% in more than ten years. Diana Choyleva from Lombard Street Research also came to a similar conclusion in one of her pieces of research for the Wall Street Journal.

This paper aims to investigate the Renminbi exchange rate misalignment problem during the last decade. To probe this issue, we follow MacDonald and Clark to establish a reduced-form OLS model which was designed to identify the relationship between the real effective exchange rate and the economic fundamentals. Then we use the HP-filter technique to estimate and calculate the current and total equilibrium exchange rate to detect the degrees of misalignment between the real and equilibrium exchange rate.

There are three main research objectives that we pursued in this paper. First, what is the equilibrium exchange rate of the Renminbi from 2005 till 2015? Secondly, can we conclude that the Renminbi is undervalued/overvalued or at the correct value during the last decade? Lastly, what conclusion and future policy implications we can conduct from our empirical results.

This paper is different from earlier research because previous literatures has rarely studied this period before. What makes this period interesting is the global financial crisis which occurred between 2007 and 2009, and the economic transformation in terms of economic policy and financial environment in the last ten years.

(6)

2. Literature review

2.1 A brief history of Renminbi exchange rate development

On 1st December 1948, People’s Bank of China was established and for the first time issued its currency the “Renminbi”. In the beginning, the policies were mainly to incent and help exports, and a few years later, policies for helping imports were also established. The exchange rate went from a crawling peg1 to the single currency USD to a crawling peg currency basket and then back to a crawling peg to the USD (1940s – 1980s)2.

Figure 1 RMB exchange rate against USD from 1985 to 2014

source: National Bureau of Statistics of China database

Between 1993 and 1994, we can observe a significant devaluation of the RMB (from 576.2 to 861.87) because at that time the inflation rate was very high in China, and the purchasing power of RMB was also weaker than ever. Between 1994 and 2005, the exchange rate regime was still a crawling peg to the USD in a managed floating system. The Renminbi was a crawling peg to a currency basket from the end of 2005 till now. In order to maintain the fixed exchange rate against the USD, the PBC had to intervene in the foreign exchange market to peg it at 8.28RMB/USD for almost ten

1_{which means it is not completely fixed but float within a small band} 2_{The history of this period refers to Sina Finance:}

(7)

years (John B. Taylor, 2003). The RMB exchange rate appreciated from 819.17RMB/USD in 2005 to 760.4RMB/USD in 2007, and in the same year, PBC announced that the crawling peg exchange rate floating range against the USD increased from 3% to 5%. The exchange rate appreciation stopped until 2008 financial crisis; it was fixed to 694.51RMB/USD at that time to against the financial shock. The RMB returned to a crawling pegged exchange rate until June 2010, but the middle rate decreased by 1.9% in August 2015, which was about 622RMB/USD. This depreciation caused a big shock in the Chinese stock market.

2.2 The equilibrium exchange rate

The concept of the equilibrium exchange rate was first proposed by Nurkse (1945) in his work “Conditions of international monetary equilibrium.” He defines this concept as the exchange rate that can balance the current account.

In order to find out the equilibrium exchange rate, the first thing we need to know is what kind of exchange rate we should use when estimating the equilibrium exchange rate in econometric models. The nominal exchange rate is the most familiar one. By definition, the nominal exchange rate is the price at which one can exchange one currency to another. However, this “nominal” exchange rate is not comprehensive enough to explain the movement in the real economy since it only compares the prices between two products.

The second option is the real exchange rate (RER). As a contrast, the RER measures the value of a country’s goods against those of another country (Catão, 2007). One of the ways to determine the RER is the “Big Mac Index”. This index indicates how much a Big Mac costs in, say China, compared to the price of an identical Big Mac in the US. The purchasing power parity is the mechanism behind this index. For example, in 2015, the annual average nominal exchange rate of the RMB/USD is 6.21, and if a Big Mac also costs 6.21 yuan in China, then we can say the purchasing power parity holds. According to data from The Economist3, from 2000 untill 2015, the

(8)

Renminbi shows a great undervaluation. This index and PPP approach are widely known, but they have many shortcomings. Some academic research criticizes the PPP approach, so we do not go deeper here.4

The last option, the Real Effective Exchange Rate (REER), is a better indicator because the nominal exchange rates do not reflect the competitiveness of a country. The REER has a real effect on economic movement. In addition, it is weighted by trade share or other (a basket of) currencies compared to the RER. So the REER provides more specific information about the exchange rate. Economists usually use the REER as a fundamental variable to assess an equilibrium exchange rate (Catão, 2007). Thus, we can say a currency is not misaligned when its real effective exchange rate equals to its equilibrium exchange rate. In conclusion, we choose the REER as the dependent variable to study the exchange rate misalignment problem.

In this paper, the real effective exchange rate we use is based on the consumer price index (CPI), which takes 2010 as its base year. In other words, the 2010 average annual exchange rate index equals 100. The State Administration of Foreign Exchange in China, the OECD and the World Bank database all use 2010 as the base year. In 2010, the annual average exchange rate against the USD was 6.8275, and it only appreciated 0.91% compared to 2010, which is a very small increase. Therefore, we believe that 2010 is a stable year and is the right one to use as the base year for the real effective exchange rate index.

Based on Nurkse’s theory, we assume that the real effective exchange rate of 100 in 2010 is the equilibrium exchange rates. Hence, ideally we will observe a horizontal line over time if the real effective exchange rate stays stable. Whenever a price goes up, the nominal exchange rate depreciates in order to compensate for the changes in price and maintain the equilibrium exchange rate. If we draw a graph to show this, we should have a basic idea of whether the exchange rate is overvalued or undervalued by looking at whether the real rate moves upwards or downwards from the horizontal line. However, this approach does not take into account several economic fundamentals (for

(9)

example, monetary policy, oil price, terms of trade index, etc.) that can affect the movement of the exchange rate. Some of the economic fundamentals have a short-term effect on the exchange rate, some of them have a long-term effect, so we cannot simply diagnose whether a currency is misaligned or not by looking at the graph of the movement of the real effective exchange rate. We need a more sophisticated approach to discover the equilibrium exchange rate in both the short- and the long-term. Our main purpose in this paper is to locate the proper equilibrium exchange rate, and then compare the equilibrium rates to the real effective exchange rate to see the difference. For now, we use a simple equation to show our idea:

∆ misalignment = 𝑟𝑒𝑒𝑟 − 𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚 𝑟𝑒𝑒𝑟

𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚 𝑟𝑒𝑒𝑟 ∗ 100%

By using this equation, we can calculate the misalignment percentage of the Renminbi exchange rate and answer our research questions. The next section introduces the approach “behavioral equilibrium exchange rate” to show how we estimate the equilibrium exchange rate.

2.3 The Behavioral Equilibrium Exchange Rate approach

In the last few decades, economists have invented several different econometric techniques to research the exchange rate misalignment problem. These techniques range from the easiest; the purchasing power parity approach, to the more sophisticated approach of the behavioral equilibrium exchange rate (BEER), and in addition, the permanent equilibrium exchange rate approach (PEER). Based on availability and the pros and cons of different approaches, we choose the behavioral equilibrium exchange rate to conduct our research. But first, we have to clarify what the BEER approach is and why we choose it instead of the other econometric approaches.

(10)

characterize this approach as a “current and cyclical equilibrium exchange rate”, because this model also considers cyclical factors that affects the exchange rate. There are two important characteristics we need to know about this approach.

First of all, the BEER approach has a predecessor called the fundamental equilibrium exchange rate (FEER) approach which was invented by Williamson (1994). FEER focuses on the relationship between the real effective exchange rate and the economic fundamentals. However, it has very strict presumptions. In this approach, there must be full employment, and the current account must equal the capital account. As we assumed in the previous section, under these presumptions the equilibrium exchange rate will deliver a horizontal line over time. FEER assumes that when internal (productivity, full employment, etc.) and external (a balanced current account and capital account, etc.) equilibrium stay constant, the equilibrium exchange rate will stay constant as well. However, this is not realistic5. MacDonald and Clark invented the behavioral model to conquer the flaws in the FEER approach. As its name tells, the BEER model intended to find out the “behavior” of the equilibrium exchange rate under the influences of the economic fundamentals.

Second of all, BEER has many advantages. One of the them is that this approach only uses a reduced-form model (MacDonald, Clark, 1997). In other words, this approach was built on an OLS model, between the dependent variable (the real effective exchange rate) and the independent variables (economic fundamentals). Thereby, constructing a simple econometric model makes this approach easier to understand and to use. In addition, MacDonald (1997) argues that the core of this approach is the real effective exchange rate, which has a long-term effect on the economy. In this model, the right-hand side includes short-term, long-term, and random effect variables. They emphasize that this approach reveals the fact that the systematic fluctuations of the exchange rate can be explained by economic fundamentals. MacDonald (1997) applied the Johansen co-integration technique to identify the co-integration relationships between the real effective exchange rate and economic fundamentals. Moreover, this

(11)

behavioral approach is very useful not only in studies of developed countries but also in those of developing countries (Montiel, 1991). It is worth pointing out that the economic fundamentals, which usually have less and less effect on the exchange rate in the long run, have a perpetual effect on the exchange rate in this approach. Therefore, we find this approach is the best option to conduct our study.

The alternative approaches have been widely discussed in academic studies already. Hence, we think it is redundant to discuss them again.6

2.4 Related literatures

The approaches most commonly used in academic research are as follows: purchasing power parity approach (PPP), the behavioral effective exchange rate approach (BEER), and the fundamental equilibrium exchange rate approach (FEER). Our reviews of previous studies are based on two broad categories, the PPP approach, and the economic fundamentals approaches (BEER and FEER).

By using the PPP approach, Chou and Shih (1997) show that the Renminbi was overvalued between 1978 and 1989, but it was undervalued after 1990. Frankel (2004) concludes that the real effective exchange rate was undervalued by about 36% at that time, but Yu Qiao (2000), Yang et al. (2004), and Cheung, Chinn, and Fujii (2008) argue that the Renminbi has not been undervalued. Zhang (2001), Wang (2004), Funke and Rahn (2005) neither found any significant undervaluation based on both the PPP and economic fundamentals approaches (BEER and FEER). Yang and Zhong (2011) use price level and the PENN effect to conduct their research. They conclude that up until 2005, there is not much evidence to show that the Renminbi is undervalued, and on the contrary, they even find evidence of overvaluation of the Renminbi. They also argue that the consumer price level in certain Chinese cities has already surpassed that of New York and London.

By using the BEER and FEER approaches, Coudert and Couharder (2007) found a 63.5% undervaluation of Renminbi between 2000 and 2004 in a worldwide dataset.

(12)

When they only use the dataset from developing and emerging countries, the estimation decreases to 52.4% of undervaluation. After removing all the unique factors (outliers) that can influence the result, the overall undervaluation is between 43.9% and 45.8%. Wang, Hui, and Abdol (2007) used the same technique to determine the long-run equilibrium rate, to which the Renminbi fluctuates on a small scale. They also argue that Chinese monetary policy may have an insignificant impact on its trade surpluses. Zhang (1999) studied the equilibrium exchange rate from 1984Q1 to 1999Q1, and he found that the Renminbi real effective exchange rate fits with the equilibrium exchange rate. He argued that the three devaluations from the authorities in 1986, 1989, and 1990 were very useful in adjusting the misalignment of the exchange rate.

3. Model and data selection

3.1 Theoretical framework and a reduced-form model

Before turning to our own model, we lay out the theoretical framework for how the model was built. MacDonald and Clark (1997) proposed the reduced-form expression of the behavioral equilibrium exchange rate approach. Such an expression shows the following:

𝑞_: = 𝛿_<=_Ζ

<: + 𝛿@=Ζ@:+ 𝜏=Τ:+ 𝜀: (a)

In this equation (a), the real effective exchange rate 𝑞_: is affected by a set of fundamental variables Ζ<: 𝑎𝑛𝑑 Ζ@:, and variables Τ:, which only affects the real effective exchange rate 𝑞: in the short-term, and an error term 𝜀:. MacDonald and Clark (2004) argue that these economic fundamentals variables affect the exchange rate movement, and this approach can check whether the exchange rates are misaligned or not.

(13)

𝑞_:= _{= 𝛿}

<=Ζ<:+ 𝛿@=Ζ@:

Now we have both the real effective exchange rate 𝑞_:, which is observable in real life, and the current equilibrium exchange rate 𝑞_:=_{. Hence, we can define and calculate} the current misalignment of the exchange rate 𝑐𝑚_: in a more sophisticated way:

𝑐𝑚_: = 𝑞_:− 𝑞_:= _{= 𝑞}

:− 𝛿<=Ζ<:− 𝛿@=Ζ@: = 𝜏=Τ:+ 𝜀:

MacDonald and Clark (1998) further distinguish current misalignment from total misalignment by arguing that economic fundamentals might depart from sustainable and desirable levels. Therefore, it is necessary to estimate the total misalignment 𝑡𝑚_: by using the following equation:

𝑡𝑚_: = 𝑞_:− 𝛿_<=_Ζ

<:− 𝛿@=Ζ@:

In this case, Ζ_<: 𝑎𝑛𝑑 Ζ_@: are denoted as “the difference between the actual real rate and the real rate given by long-run values of economic fundamentals” (MacDonald, 1998). Then they add and subtract 𝑞_:=_{from the right-hand side of this equation, and} decompose the total misalignment into two parts:

𝑡𝑚_:= (𝑞_:− 𝑞_:=_{) + [𝛿}

<= Ζ<:− Ζ<: + 𝛿@=(Ζ@:− Ζ@:)] Since (𝑞_:− 𝑞_:=_{) = 𝜏}=_Τ

:+ 𝜀:, we replace this term from 𝑡𝑚:: 𝑡𝑚_: = 𝜏=_Τ

:+ [𝛿<= Ζ<:− Ζ<: + 𝛿@=(Ζ@:− Ζ@:)] + 𝜀:

In conclusion, in this new equation, we can see that total misalignment is affected by the short-term factor 𝜏=_Τ

:, the long-term factors 𝛿<= Ζ<:− Ζ<: + 𝛿@= Ζ@:− Ζ_@: and the error term 𝜀_: overtime.

Based on the availability and operability, we follow the suggestions from R. MacDonald (1998), Wang et al. (2007), and Tang et al. (2009) to build our model based on the reduced-form equation (a):

REER = 𝜹0 + 𝜹1M2 + 𝜹2res + 𝜹3tot + 𝜹4tnt + 𝝁 (1)

where7_{: REER is the real effective exchange rate;}

M2 is the money supply as a proxy for monetary policy; res is the foreign exchange reserve;

tot is the terms of trade;

(14)

tnt is the relative price of non-tradable goods to tradable goods as a proxy for “Balassa-Samuelson” effect;

𝝁 is the error term.

We expect to see these economic fundamentals affect the behavior of the real effective exchange rate overtime.

3.2 Motivation and definition of variables

In this section we lay out our motivation and definition of variables in our main model. Regarding to the variable selection, the fundamental research paper by MacDonald (1998)8 found different economic fundamentals that have a long-term effect on the real effective exchange rate (for example the “Balassa-Samuelson” effect, the real oil price, the savings from private sector, etc.). MacDonald also found several factors that have a short-term effect on the real effective exchange rate by using the uncovered interest rate parity theory. In the other paper by MacDonald and Clark (1998), they also use more or less similar variables to estimate the equilibrium exchange rate, such as the real effective exchange rate, terms of trade, the relative price of non-tradable to tradable goods, net foreign assets, the relative stock of government debt, and the real interest rate. In more recent Chinese studies, Tang et al. include some variables that are more closely related to the economic situation in China, for instance, trade liberalization as an indicator variable to capture the openness of the Chinese market, net foreign asset to capture the situation of Chinese current account situation, and government expenditure. Based on data availability we selected the following five variables: a. Real effective exchange rate (REER)

The core of the behavioral equilibrium exchange rate approach is REER. The World Bank defines REER as “the nominal effective exchange rate (a measure of the value of a currency against a weighted average of several foreign currencies) divided by a price deflator or index of costs”. Compared to the nominal exchange rate, REER measures not only the currency price in terms of other currencies but also how much

(15)

one can buy with this currency (Catão, 2007).

The real effective exchange rate of the RMB against the USD monthly data we use is based on the consumer price index (CPI) (2010=100). The RMB will appreciate if REER goes up.

Data source: State Administration of Foreign Exchange database

b. Money Supply (M2↑àequilibrium exchange rate↓)

Economists often use M2 to measure money supply, as it includes cash and checking deposits and near money in a broader way. According to the Macroeconomics Theory, an expansionary monetary policy means increasing the money supply to the market, which in turns increases the aggregate demand in the economy. Since the production output level stays constant, an increasing aggregate demand will push up the price level. In order to balance the current account, the equilibrium exchange rate needs to depreciate. If the purchasing power parity theory holds, the domestic price level will move in the opposite direction to the exchange rate (the price goes up then the exchange rate goes down). In short, a growth in money supply increases a country’s inflation rate, when its current account deteriorates. Hence, the equilibrium exchange rate should depreciate. We expect a negative relationship between the money supply and the real exchange rate in the estimation.

Data source: National Bureau of Statistics of China database

c. Foreign exchange reserve (res↑àequilibrium exchange rate↑)

(16)

effective exchange rate.

Data source: State Administration of Foreign Exchange database

d. Relative price ratio of non-tradable to tradable goods (tnt ↑àequilibrium

exchange rate↑)

This variable intends to capture the “Balassa-Samuelson effect” that was proposed in 1964 which says that in a country where economic growth is high, the increase in the real wage and the real exchange rate would be higher as well. This phenomenon would eventually lead to growth in the relative price of non-tradable goods to tradable goods. We follow the suggestion of Clark and MacDonald (1998) that non-tradable goods are measured by the consumer price index (CPI) and tradable goods are measured by the producer or wholesale price index (PPI or WPI). This leads to a better current account situation, therefore, the equilibrium exchange rate will also appreciate (Zhang, 1999). We expect a positive relationship between this and the real effective exchange rate.

e. Terms of trade (tot↑àequilibrium exchange rate?)

Terms of trade are calculated by the export price to the import price index ratio. It measures the relationship between the export volume and import volume of a country. If the total is less than 1, this means that the country is importing more than it is exporting to another country. If the value is more than 1, then it means that the country is running a trade surplus, which is a very important bilateral trade measurement.

(17)

in order to keep the current account balance. If terms of trade fall, then the current account should deteriorate, and the equilibrium exchange rate should also depreciate. On the other hand, in a lot of empirical evidence shows that whenever terms of trade go up, the exchange rate for exporting goods always depreciates. Thus, some economists claim that the improvement of "terms of trade" in one country often increases the demand for non-tradable goods and further leads to an appreciation of the exchange rate of importing goods. In this case, we cannot expect that terms of trade have a positive or negative relationship with the real effective exchange rate of the RMB.

4. Empirical Analysis

Before considering the calculation of current and total misalignment, we have to produce a co-integration equation based on the equation (1) in this section. The idea behind the empirical analysis is that, in the economy, there are non-stationary time series sequences, for example, which might have seasonal factors or trends. However, these variables can have a long-term stable relationship(s). In other words, there are certain equilibrium relationships exist amongst them. In time series analysis, we can call this relationship a co-integration relationship. Therefore, co-integration analysis is a strong tool to discover the equilibrium between the real effective exchange rate and economic fundamentals, and eventually provide us with a way to estimate the equilibrium exchange rate and calculate the misalignment (Zhang, 1999).

According to Econometrics theory, the stationary test (Augmented Dickey-Fuller test) is essential for any time series analysis. We will follow these steps in MacDonald and Clark’s studies.

(18)

4.1 The Augmented Dickey-Fuller Test

The “Augmented Dickey-Fuller unit root test” is an essential pre-examination in any time series analyses. This is because we need to know if our time series sequences are stationary or integrated or not before carrying out regressions. If they are not stationary at level or integrated at the same order and we run a regression on a non-stationary time series, there would be a spurious regression, which can lead us to draw a wrong conclusion in our empirical analysis. Technically, to avoid this problem, ADF-test can be used to find out if the time series variables are stationary at order zero 𝐈 (0), which means they are stationary; or if they are integrated at order one 𝐈 (1), which suggests that the time series which are not stationary wander widely.9

When we follow the rule and run the test throughout all of the variables, the results are as follows:

Table 1 ADF Unit Root Test on time sequence (2005M01-2015M12)

* C denotes the constant, T denotes the trend and 0 means no constant or trend has been selected. L denotes the lag numbers.

9_{more detail see appendix}

(19)

Figure 2 First difference of time series variables

In Table 1, we can see that all five variables are non-stationary at the level. However, after taking their first difference, they are all stationary and integrated at their first order. Among all the variables, dlnreer, dlnm2, dlntot, and dlntnt can reject the null hypothesis of having a unit root at the 1% confidential level and dlnres can reject the null hypothesis at the 5% confidential level. The results of our test are shown in Figure 2 as well, in which the first difference of all the variables fluctuates around zero.10 Based on the fact that all the variables are non-stationary at the level, although they are stationary at the first order, we can now do the Johansen co-integration test to see if there are co-integration relationships amongst variables.

4.2 The Johansen co-integration test

One of the contributions of this behavioral econometric model is that it can detect a significant and sensible long-term currency relationship (MacDonald, 1997), which has to be done by running a Johansen co-integration test. MacDonald (1999) also acknowledges the advantage of Johansen co-integration models in a way that this model can not only test for the co-integration relationship, but can also provide the number of co-integration relationships.

The Johansen test is designed to analyze several variables, for example, k variables, which allows more than one co-integration relationship exist in a regression equation.

10_{graphs of each variable can be found in appendix, produced by Stata.}

(20)

This test has three advantages. First, it allows endogeneity to exist in a regressor. Second, this test implies that all the variables can be endogenous, and there is no specifically fixed dependent variable. Third, since there is no fixed dependent variable, more than one co-integration can exist in the model. Hence, here we follow MacDonald to carry out the Johansen co-integration test. The results are in Table 2.

Table 2 Johansen test for co-integration

Trend: constant Lag: 1 Maximum

rank

parms LL eigenvalue Trace-statistic 5% critical value 0 5 1798.1787 . 167.7939 68.52 1 14 1841.7434 0.48578 80.6644 47.21 2 21 1861.5545 0.26100 41.0422 29.68 3 26 1875.21 0.18818 13.7312* 15.41 4 29 1880.6629 0.07988 2.8255 3.76 5 30 1882.0756 0.02134

* asterisk indicates that we reject the null hypothesis and conclude the number of co-integrations.

Based on the Johansen co-integration tests results, we can conclude that there are at most three co-integration relationships among these variables.

Next, we run the regression of our main equation (1), the results are in Table 3. Table 3 Regression on equation (1)

(1) lnreer

variables coefficient standard error t-value p-value

lnres 0.042* 0.023 1.86 0.066

lnm2 -0.291*** 0.025 -11.53 0.000

lntot -0.281*** 0.068 -4.16 0.000

lntnt 0.181 0.172 1.05 0.294

constant -1.835*** 0.100 -18.23 0.000

(21)

The co-integration equation (3) is formed from Table 3 and we will need to use this equation to estimate the equilibrium exchange rate.

𝐥𝐧𝐑𝐄𝐄𝐑 = −𝟏. 𝟖𝟒 − 𝟎. 𝟐𝟗𝐥𝐧𝐦𝟐 + 𝟎. 𝟎𝟒𝐥𝐧𝐫𝐞𝐬 − 𝟎. 𝟐𝟖𝐥𝐧𝐭𝐨𝐭 + 𝟎. 𝟏𝟖𝐥𝐧𝐭𝐧𝐭 (3)

(18.23) (11.53) (-1.86) (4.16) (-1.05)

From Table 3 we can see that money supply (m2) and terms of trade (tot) are significant on a 1% confidential level, foreign exchange reserve (res) is significant on a 10% confidential level, and only the relative price ratio of non-tradable to tradable goods (tnt) is not significant at all. There are two possible reasons that the Balassa-Samuelson effect is not significant in China. On one hand, the household registration policy “Hukou” in China majorly restricts the free labor movement, which is one of the most important premises of the Balassa-Samuelson effect. On the other hand, the price control and labor restriction among diverse industries are still significant countrywide. Moreover, different policy system, legal system, and market composition might also contribute to this insignificant issue.

(22)

5. Renminbi current and total misalignments

5.1 Current and total misalignments of the exchange rate

We produced the co-integration equation in Section 4. Now we will use this equation to estimate the equilibrium exchange rate in this section. First, we need to derive the current misalignment from the co-integration equation (3):

𝐥𝐧𝐑𝐄𝐄𝐑 = −𝟏. 𝟖𝟒 − 𝟎. 𝟐𝟗𝐥𝐧𝐦𝟐 + 𝟎. 𝟎𝟒𝐥𝐧𝐫𝐞𝐬 − 𝟎. 𝟐𝟖𝐥𝐧𝐭𝐨𝐭 + 𝟎. 𝟏𝟖𝐥𝐧𝐭𝐧𝐭. As its name indicates, current misalignment measures the difference between the actual value of the real effective exchange rate and the estimated value of the equilibrium exchange rate given by the current value of economic fundamentals (MacDonald, Clark, 2004). By putting four fundamental economic variables (lnres, lnm2, lntot, and lntnt) back into this equation, we can calculate the current equilibrium exchange rate. We recall the simple misalignment equation we had in Section 2.2, and make a few changes to it, then we apply the following equation to calculate the current Renminbi misalignment:

current misalignment =𝑙𝑛𝑟𝑒𝑒𝑟 − 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝐸𝐸𝑅

𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝐸𝐸𝑅 ∗ 100%

The results are transformed into graphs. As we can see from the right panel in Figure 4, current misalignment varies from about -1.8% to 2.3%. Compared to the estimations from previous literature, our results fluctuate in a relatively small band.

Figure 3 Current BEER and current misalignment of Renminbi (%)

(23)

exchange rate and that given by the sustainable or long-run values of the economic fundamentals. What we have done so far is to calculate the current misalignment. However, this current equilibrium rate can be biased because cyclical factors such as the business cycle can affect the estimation outcome. Therefore, further treatment needs to be carried out. The Hodrick-Prescott filtering technique (J. Money, Credit, Banking, 1997) is a useful tool for achieving the smoothed time series sequences, which have been widely used on non-stationary time series sequences of financial or macroeconomics literature. The underlying mechanism of the HP-filter is to solve the minimization problem. This filter can easily remove the cyclical components from the time series variable. There are two parts to this process. First, filter all the time series variables except lnreer. Second, put these filtered values back into the co-integration equation (3). By carrying out these two steps, we now have a sequence of long-term REER. In the same way as we calculate the current misalignment, we can now calculate the total misalignments. The results are in the right panel in Figure 5. We can see there are not many changes compared to the current misalignment. We will discuss this further in the next section.

Figure 4 Long-term REER and total misalignment of Renminbi (%)

5.2 Discussion of current and total misalignment

(24)

results show that the maximum overvaluation is about 11.4% and the maximum undervaluation is about 9.6%. We also find that the volatility of the current misalignment is slightly higher than the total misalignment. Surprisingly, the misalignment range is not as large as we expected. Previous studies such as those of Tang and Chen (2010) and Qin (2004) suggest that around 10% misalignment means that there are no severe misalignment problems exist between the real effective exchange rate and long-term equilibrium exchange rate.

The whole time line can be roughly divided into four time periods. The first period is from January 2005 until November 2007; the second period is from December 2007 until June 2009; the third period is from July 2009 until December 2014, and the last period is the whole year of 2015.

(25)

6. Conclusions and policy implications

In this paper, we have discussed the exchange rate misalignment problem of the Renminbi between 2005-2015. To shed the light on this, we use the behavioral equilibrium exchange rate approach (BEER) suggested by Clark and MacDonald (1997) to find out the impact of economic fundamentals on the real effective exchange rate (REER). We use money supply, foreign reserves, terms of trade and relative price ratio of non-trade goods to trade goods as our independent variables. The estimated coefficients of independent variables accord with previous studies and economic theory. They are -0.29, 0.04, -0.28 and 0.18, respectively. On one hand, the money supply and relative price ratio (tnt) have a negative effect on REER. For example, REER depreciates 0.29% if the money supply increases by 1%. On the other hand, terms of trade and foreign reserve have a positive effect on REER. Our empirical analysis also found co-integration relationships (up to three at most) between the equilibrium exchange rate and these economic fundamentals. The estimation of current and total misalignment showed a relatively small variation band, but the trends mostly confirm to reality and recent studies from the International Monetary Fund (IMF). In our last observation period of 2015, the Renminbi showed an overvaluation due to consistently revaluation of the exchange rate.

(26)

Renminbi exchange rate almost reached equilibrium level (showing a slightly undervaluation at the end of 2014). Our result gives a similar conclusion.

Overall the relatively small band of total misalignment shows that the Renminbi exchange rate is moving in the right direction. The capital account (free capital flow) has been liberalized gradually in recent year in China11_{. As the second economic entity}

in the world, China doubtless needs to sustain the sovereign monetary policy. Hence, a floating exchange rate regime must be carried out in the future, which should be done in the next two or three years otherwise the development space of the Chinese financial market will shrink (IMF, 2015). The economy has slowed down over the past few years in China, which definitely makes things difficult for the authorities. Due to productivity and other factors, the competitiveness of Chinese economy also decreased a lot. The unexpected devaluation of the exchange rate at the end of 2015 might be a way to restore the competiveness of the economy (The Economist), but a continued devaluation cannot always be a magic bullet to rectify the economy in a long-term. We conclude that stabilize the exchange rate is still very important since the Renminbi will be joining the SDR in 2016. Maintaining a relatively stable exchange rate but still moving on the path of achieving the goal of a floating exchange rate regime should be the primary target for the authorities over the next few years, especially after the last unexpected devaluation. Moreover, stabilizing the exchange rate can reduce the capital outflow (capital flight), which can further benefit the country. After all, the fact that the exchange rate was approaching a stable equilibrium level is still good news. China has already started along the path, but there are, in the upcoming years, still many miles to go.

(27)

Bibliography

Changjiang Yang, Ninghua Zhong, Purchasing power parity and equilibrium exchange rate of Renminbi, Journal of Financial Research, Vol. 379 (2012), pp. 36-50. Clark, P.B., MacDonald, R. Exchange Rates and Economic Fundamentals: A

Methodological Comparison of BEERs and FEERs[R]. IMF Working paper. 1998, No.67

Erlat, Guzin and Arslaner, Ferhat (December 1997). "Measuring Annual Real Exchange Rate Series for Turkey". Yapi Kredi Economic Review, vol. 2 (8), pp.35–61.

Johansen, Søren (1991). "Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models". Econometrica 59 (6): 1551–1580. Kenneth Rogoff, The Purchasing Power Parity Puzzle, Journal of Economic Literature,

Vol. 34, No. 2. (Jun., 1996), pp. 647-668.

MacDonald, R. Exchange rate behaviour: are fundamentals important? The Economic

Journal, 109 (November) 1999 F, 673-F691

Michael Funke and Jörg Rahn: Just How Undervalued is the Chinese Renminbi?, the

world economy, 2005, vol.28, pp.465-489.

Peter B. Clark, Ronald MacDonald, Filtering the BEER: A permanent and transitory decomposition, Global Finance Journal, 15 (2004) 29-56.

Peter B. Clark: Comparison between FEER and BEER, IMF working paper, 1998, pp.1-38.

R. MacDonald. What determines real exchange rates? The long and the short of it.

Journal of International Financial Markets, Institutions and Money, vol. 8 (1998),

pp.117–153.

Robert F. Engle and C. W. J. Granger, Co-Integration and Error Correction: Representation, Estimation, and Testing, Econometrica Vol. 55, No. 2 (Mar., 1987), pp. 251-276

W. L. Chou and Y. C. Shih: The Equilibrium Exchange Rate of the Chinese Renminbi,

(28)

Wang Yajie, Hui Xiaofeng, Abdol S. Soofi: Estimating renminbi (RMB) equilibrium exchange rate, Journal of Policy Modeling, 29 (2007), pp.417-429.

Xiaopu Zhang: The theory and models of equilibrium exchange rate of Renminbi, Economic Research Journal, Vol. 12 (1999), pp.70-77.

Xiliang Zhao, Jingwen Zhao: Renminbi Equilibrium exchange rate analysis: BEER),

The Journal of Quantitative & Technical Economics, 2006, vol.12, pp.33-42

Yahui Tang, Shoudong Chen: Estimation of the equilibrium exchange rate and the exchange rate misalignment of Renminbi based on BEER model: 1994Q1~2009Q4, Studies of International Finance, 2012, vol.12, pp.29-37. Yin-Wong Cheung, Menzie D. Chinn, Eiji Fujii: The overvaluation of Renminbi

undervaluation, Journal of International Money and Finance, Elsevier, vol. 26(5),

pages 762-785

Diana Choyleva, China's Dangerous Overvaluation, Wall Street Journal, 25th February 2014

Intelligence Unit, Renminbi secures elite status, The Economist, 4th_{December 2015}

Peter Isard: Equilibrium Exchange Rates: Assessment Methodologies, IMF working paper, WP/07/296

(29)

Appendix

Augmented Dicky-Fuller test

ADF-test can solve the autocorrelation problem and it has three different models: 1. ∆𝑌_: = 𝐵_<+ 𝑑𝑌_:o<+ 𝑎_p + 𝑒_:

2. ∆𝑌_: = 𝐵_<+ 𝐵_@: + 𝑑𝑌_:o<+ 𝑎_p+ 𝑒_: 3. ∆𝑌_: = 𝑑𝑌_:o<+ 𝑎_p + 𝑒_:

By looking at the composition of these three, it is not hard to see that there is only intercept in the first equation, there are both trend and intercept in equation two, and there is neither trend nor intercept in the last equation. To decide which model should be chosen, we can start making graph of each variable to see whether there is a trend overtime. If it is the case, then a trend term has to be included in the model. The next step is to see whether the data are fluctuating around zero, in other words, has an average of zero. If it is the case again then we should not include a constant in the model. Based on the nature of these models, a trend cannot exist solely in any models, therefore, when doing the ADF–test, if the constant term is not significant but trend term is significant, and then the second equation should still be selected (both trend and constant).

(30)

(31)

Stability tests

We use “Lagrange multiplier test for autocorrelation” and “Inverse Roots of AR Characteristic Polynomial” to exam the soundness of our model. The results of “Lagrange multiplier test for autocorrelation” in Table 4 suggests that there is no autocorrelation problem in our model. And “Inverse Roots of AR Characteristic Polynomial” figure shows that the the unit roots lie within the unit circle. This test confirms the stability of our model.

Table 4 Lagrange multiplier test for autocorrelation

Lag Chi 2 Degree of freedom P-value

1 2 3 30.9823 22.3534 29.6614 25 25 25 0.18961 0.61526 0.23722 Figure 4 -. 0 5 0 .05 .1 lntnt 2005m1 2010m1 2015m1 month -. 0 4 -. 0 2 0 .02 .04 dlntnt 2005m1 2010m1 2015m1 month -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

The Renminbi Equilibrium Exchange Rate 2005-2015:

Faculty of Economics and Business

The Renminbi Equilibrium Exchange Rate 2005-2015:

based on the Behavioral Equilibrium Exchange Rate approach

Table of contents