Foreign Exchange Market Intervention, Inflation and. Export Competitiveness

Transcription

1 Foreign Exchange Market Intervention, Inflation and Export Competitiveness Wukuang Cun Rutgers University Jie Li Central University of Finance and Economics March

2 Abstract During recent years, the central banks of emerging market economies, in particular the BRICs, have been actively intervening foreign exchange markets in order to prevent their currencies from sharp appreciation, which greatly increased pressure on domestic inflation. The central banks of these economies are reluctant to allow appreciation of their currencies, since it could erode their export competitiveness. However, inflation also erodes a country s competition position by raising the relative price of its exports. We present a two-country DSGE model with a banking sector which produces liquidity used in transaction by combining monitoring efforts and collateral goods. When there is a positive shock in international demand for domestic assets, the results show that it is better for the central bank to allow a moderate appreciation of its currency, rather than conduct currency intervention and fix exchange rate at its target. This not only brings down the inflation, but also helps maintain export competitiveness. In addition, sterilization has very limited effectiveness in terms of taming inflation. The domestic assets sold by the central bank as sterilization instruments can also be used to produce liquidity, adding inflation pressure to the economy, which substantially compromises the effects of monetary sterilization. JEL Classification: E51, E52, E58 Keywords: Currency Intervention; Export Competitiveness; Inflation; Sterilization 2

3 1 Introduction During recent years, growing balance of payments surpluses via current and financial accounts put upward pressure on the value of domestic currencies for many emerging market economies (EMEs), in particular the BRICs. Immediate appreciation of domestic currencies may not top the EMEs policy response list, due to either the worries of losing export competitiveness 1 or eagerness for international reserve hoarding (Ghosh et al.2012). To continue their export oriented growth strategy and prevent domestic currencies from sharp appreciation, central banks of EMEs, particularly BRICs countries, have been actively intervening foreign exchange markets and hence accumulated massive amounts of international reserves, which leads to greatly increased pressure on domestic inflation. Figure 1 shows the foreign reserve holdings and the inflation rates in the BRICs during the last ten years. The blue lines represent the inflation rates and the orange bars represent the foreign reserve holdings of the BRICs. The shaded area highlights the recession periods that are documented by the OECD Recession Indicator. Table 1 gives a summary of Figure 1. We divided the last ten years into two periods: 2003Q1-2007Q4, the five years before the recent crisis and 2008Q1-2012Q4, the five years after its breaking out. Obviously, the foreign reserve holdings of the BRICs have greatly increased during the last ten years. Inflation rates have also increased in the BRICs, except for Brazil, which historically experienced high inflation. The nominal exchange rate is not the only thing that matters to a country s competitiveness position. The domestic inflation relative to foreign inflation can also change the relative prices of the exports. A tight management of nominal exchange rate might erode the country s export competitiveness, rather than improve it, due to its inability to control domestic inflation. Actually, it could be better for the central banks of the BRICs to allow 1 There is a well-documented phenomenon, fear of floating, by Calvo and Reinhart,

4 their exchange rates more flexibility, which not only helps tame domestic inflations but also helps maintain export competitiveness. In addition, taming domestic inflations seems more urgent today, under the context of recent recession. After breaking out of the recent global crisis at the end of 2007, the world economy went into recession. As shown in Table 1, the US inflation rate, as a benchmark, actually dropped from 2.9 (the average during ) to 2.1 (the average during ). Under the context of the current global recession, recent high inflations in the BRICs could be even more disturbing than usual, since it narrows the space for economic stimulus in these economies. Only when their inflations are under control, the central banks of the BRICs can have more space for economic stimulus that helps lift their economies out of recession. To tie down domestic inflations, the central banks of these economies sterilize the increased liquidity from foreign exchange market intervention by selling out government bonds or central bank bills in domestic financial markets, in order to keep monetary base stable. However, monetary sterilization may not be really effective in the sense of controlling domestic inflation. The domestic assets (government bonds or central bank bills) are also high liquid assets which can be easily converted to cash and add inflation pressure to the domestic economy. This may largely compromise the effects of monetary sterilization. To examine these issues, we build a two-country DSGE model with a banking sector which produces liquidity used in transaction by combining collateral and monitoring labour (Goodfriend and McCallum, 2007). Consumption of households is constrained by the liquidity they have (cash in advance constraint). The government bonds sold as sterilization instruments can be used as collateral and facilitate the production of liquidity. We add to the literature by showing that, sterilized intervention, which keeps monetary base constant, however, increases the supply of government bonds, another type of high liquid asset, and thus adds pressure to domestic inflation. In addition, we compare two different foreign ex- 4

5 change rate policy rules 2, under the same monetary policy rule. The first exchange rate policy is referred as Type A policy and its only objective is to stabilize nominal exchange rates, not allowing domestic currency appreciation. This type of exchange rate policy, consistent with many EMEs export oriented growth strategy, was widely adopted by central banks in EMEs in 1990s, intending to preserve export competitiveness and boost growth. The second exchange rate policy is referred as Type B policy, which is aimed to stabilize both domestic inflation and exchange rates. Central bank which adopts Type B policy allows its currency to appreciate in response to domestic inflation. After a positive capital inflow shock, the results show that Type B exchange rate policy is more effective not only in terms of stabilizing inflation, but also in terms of maintaining the exports competitiveness. The Type A policy is ineffective even in the sense of protecting exports due to its inability to control domestic inflation which erodes the export competitiveness by raising the real exchange rate, even if nominal exchange rate is kept constant. The rest of this paper is structured as follows. Section 2 presents some stylized facts in China. Section 3 introduces the model, followed by section 4, the calibration of the model. Section 5 analyse the dynamics of the model and compares the two types of foreign exchange rate polices. Section 6 concludes the paper. 2 Some Stylized Facts in China During the last ten years, China has accumulated large stockpile of foreign reserves via both current and capital account surpluses, which has put great pressure on domestic inflation. Recent inflation narrows the space for the monetary authority of China to stimulate economy 2 Deliberately altering exchange rates to influence the macro-economic environment can also be regarded as a special type of monetary policy. 5

6 under the context of global crisis. Many suggest that in order to obtain more control on inflation and open the space for economic stimulus, China should allow its exchange rate more flexibility. In mid 2005, the People s Bank of China loosened its control on exchange rate and allows its currency to appreciate gradually. But the People s Bank of China never wants rapid or aggressive revaluation of its currency, since fast appreciation of Yuan would harm the competitiveness of export oriented sectors which most emerging market economies heavily depend on. However, tight control on Yuan/USD exchange rate may also erode export competitiveness. Nominal exchange rate are not the only thing that determines a country s competitive position. Inflation rate in China relative to inflation in the US also affect the real exchange rate. As shown in Figure 2, since 2008, the real exchange rate actually appreciates faster than the nominal exchange rate of USD to Chinese Yuan due to higher inflation in China. In addition, high inflation also pushes up labour costs and deteriorate the profitability of the export oriented sector in China which heavily depends on labour intensive production. In fact, the People s Bank of China allows the appreciation speed of Yuan deviate from its target level temporarily and moderately when domestic inflation is extremely high. As shown in Figure 2, the appreciation speeds of nominal Yuan/USD exchange rate are slightly faster in the periods of high inflation. But this may not be enough to tame the inflation in China, which rose to 7% in 2011 again after it topped 8% at the beginning of To get better control on inflation, the People s Bank of China should allow more flexibility of Chinese Yuan in the future. 6

7 2.1 Inflation v.s. Appreciation To describe how the People s Bank of China balances the trade-off between stabilizing nominal exchange rate and controlling inflation, we introduce the exchange market pressure index (EMP), which is originated from Girton and Roper (1977) and developed by Weymark (1995), Jie Li (2012). EMP index use a weighted average of decrease in foreign reserves and depreciation of local currency to capture foreign exchange market pressure. Let ˆf t and ê t be the percentage growth rate of foreign reserves and nominal exchange rate (as domestic currency cost of per unit of foreign currency). The EMP index is defined as, EMP t = ê t φ ˆf t EMP index measures the total excess demand for a currency in foreign exchange market as the exchange rate change that would have been required to remove this excess demand in the absence of currency intervention. For example, when EMP t < 0, to stop the domestic currency from appreciating by one percent, the central bank has to increase its foreign reserve holdings by φ percent. The term, φ, represents the effectiveness of central bank s foreign exchange market intervention. For model independent EMP index, φ is often set to σ e /σ f, where σ e and σ f are standard deviation of ê t and ˆf t respectively. Originally, EMP was proposed and used to describe the pressure faced by a central bank which committed to defend its currency. In an opposite way, it can also be used to describe the pressure faced by a central bank which intervenes foreign exchange market to avoid sharp appreciation of domestic currency (as shown in Figure 3). Here, we use EMP t to describe the appreciation pressure on domestic currency. Figure 4 shows the appreciation pressure of Chinese Yuan and foreign exchange market intervention activities during last 17 years. The blue line (solid) shows the total appreciation 7

8 pressure on Chinese Yuan, i.e., EMP t. The orange line (dashed) shows the fraction of appreciation pressure removed by currency intervention, that is, φ ˆf t. The difference between the two lines is thus the actual change of exchange rate. The periods highlighted are those in which inflation (annualized) rate is higher than 5% 3. As shown in Figure 4, when inflation rate is lower than 5%, the most part of the market pressure on appreciation was removed by the central bank s intervention and the excess foreign demands for Chinese Yuan turned into foreign reserves of the central bank. In those periods with high inflation (> 5%), Chinese Yuan is allowed to appreciate moderately to remove part of the market pressure. In fact, the People s Bank of China allows its currency to appreciate faster when inflation is high. However, the current management of Yuan/USD exchange rate still seems not flexible enough to tame the inflation in China, which rose to 7% in mid 2011 again after it topped 8% in To have better control on inflation, the monetary authority of China may need to allow more flexibility of Yuan in the future. 2.2 Other Elements to the Story There are, however, other elements to the story. As in other emerging market economies, the sequencing of capital account liberalization in China follows the rules of inflow first and outflow second and long term flow first and short term flow second. In the early 1990s, China s capital control was aimed to encourage foreign direct investment and discourage capital outflow, which reflects the need to develop manufacturing sector. From the beginning of the 2000s, this policy was changed in response to external imbalance. The government not only encourage FDI inflows but also encourage FDI outflows, which was referred as Go- 3 The Chinese government prefers to keep the annual inflation rate under 4%, which they think consistent with healthy economic growth. 8

9 out policy. For the short term flows, after WTO entry in the year 2001, foreign investors are allowed to hold a wider range of domestic financial assets through, for example, the Qualified Foreign Institutional Investor (QFII) scheme introduced in However, at that time, special policies towards exporters and restrictions on domestic holdings of foreign financial assets were mostly kept the same as before. Exporters have to swap most part of their export revenue with central bank at par values and domestic investors face relatively heavy restrictions when investing on foreign financial assets. At the end of 2006, the Qualified Domestic Institutional Investor (QFII) scheme was launched. After that, Chinese institutions and residents can entrust Chinese commercial banks to invest in financial products overseas. This sequencing of capital account liberalization gives foreign investors more access to domestic financial assets, while domestic households have relatively limited access to foreign financial assets and their potential demand for foreign assets cannot be satisfied. This partial and asymmetric liberalization of capital account is likely to lead to capital account surplus and force the People s Bank of China to intervene in order to keep Yuan from appreciation. In addition, intervention may become more difficult when the openness of capital account is increased. Ouyang, Rajan and Willet (2010) find that the indicator of openness of Chinese capital account is growing and the efficiency of currency intervention of Chinese central bank is decreasing over time. To prevent nominal exchange rate from appreciating by one percent, Chinese central bank will need to sell more domestic currency in the foreign exchange market. Moreover, after the breaking out of recent crisis, the US government has started two rounds of Quantitative Easing and announced the third round. This decreases interest rates in the US, reversed the spread between US and China s interest rates and may lead to more capital flows towards China. In order to control inflation, the People s Bank of China sterilize its currency intervention 9

10 by selling central bank bills 4 to financial institutions. Ouyang et al. (2010) and He (2005) find that, until recently, China had been conducting complete sterilization and every unit of increased net foreign assets of the central bank is accompanied by one unit reduction of net domestic assets, keeping monetary base constant. Goodfriend and Prasad (2007) finds that the share of total net foreign exchange inflows sterilized by Chinese central bank is between 20 to 48 percent for Burdekin and Siklos (2005), however, document that the People s Bank of China has not done so sufficiently to prevent M2 from increasing, even with complete sterilization. However, sterilization can be costly. The central bank may face a quasi-fiscal cost when it conducts sterilized intervention, especially when domestic interest rates are higher than foreign interest rates (Chang, Liu and Spiegel 2012), which makes sterilization unsustainable. 3 The Model We consider a global economy with two countries home and foreign countries. We focus on the home country and assume that the foreign country is passive. There are four types of goods: final goods, bundled intermediate goods, retail intermediate goods and foreign goods. The first three types of goods are domestically produced while the foreign good is produced in the foreign country. The final good is produced by combining bundled intermediate goods and foreign goods, while the bundled intermediate good is a composite of differentiated retail intermediate goods. The retail intermediate good is produced using labour and capital. The final goods can be consumed or exported to foreign economy. All the markets are competitive except the market for retail intermediate goods. 4 Central bank bill is a type of short term debt issued by central bank to drain excess liquidity in banking system 10

11 There are a continuum of households living in the home country. Following Goodfriend and McCallum (2007), we assume that the households not only consume, supply labour and save, but also own and operate monopolistic firms which produce retail intermediate goods and banks which produce liquidity assets used in transaction. Thus, the optimizing analysis can be done in one step. In this model, there are three types of assets, domestic high powered money, domestic government bond, and foreign government bond. The capital account of domestic economy is partially and asymmetrically opened. The domestic households are not allowed to hold any foreign assets, while foreign households can hold both domestic and foreign government bond. This is consistent with current institutional features in most emerging market economies, whose capital account liberalization process follows the rule of inflow first and outflow second. The domestic exporters have to swap all their revenue in foreign currency with the central bank at market exchange rate. To manage their oversea assets, foreign households entrust a bank operated in domestic country, which sell and buy domestic assets on behalf of them. However, any change in foreign households holdings of oversea assets will induce an adjustment cost. The central bank intervenes foreign exchange market and sterilizes currency intervention by exchanging domestic government bonds for domestic currency. There is no offshore market for domestic assets and all the domestic assets are transacted in domestic markets and circulate in domestic banking system. 11

12 3.1 The Aggregation Sector Final Goods Producers. The final goods producers produce final goods y t by combining bundled intermediate goods y d t and foreign goods y f t. The production technology is given as: y t = α(y d t ) α (y f t ) 1 α where α = (α) α (1 α) (1 α). Let P d t be the price of the bundled intermediate good, P f t be the price of foreign goods in foreign currency and e t be the nominal exchange rate (price of foreign currency in terms of domestic currency). The demand functions derived from the optimization are given as: y d t = α ( P d t /P t ) 1 yt (1) y f t = (1 α)(e t P f t /P t ) 1 y t (2) Combing the demand functions and the production function, the pricing rule of final goods is thus given by: P t = (P d t ) α (e t P f t ) 1 α (3) Bundled Intermediate Goods Producers. Bundled intermediate goods y d t are produced by combing a continuum of retail intermediate goods y d t (j) for j [0, 1]: ( 1 yt d = 0 ) θ yt d (j) θ 1 θ 1 θ dj where θ > 1 is the elasticity of substitution between differentiated retail intermediate goods. 12

13 The demand function for retail intermediate goods j is thus given by, where P d t intermediate goods P d t (j) by P t = [ P yt d d (j) = t (j) P d t ] θ y d t (4) is the price of bundled intermediate goods, which is related to the prices of retail [ ] 1 1 P d 0 t (j) 1 θ 1 θ dj. 3.2 The Households As in Goodfriend and McCallum (2007), a typical household not only consume, supply labour and save, but also own and operate a monopolistic firm which produces retail intermediate goods and a bank which produces liquid assets used in transaction. The household take as given all the prices except the price of its own retail intermediate good. We assume that there are many households which are similar and focus only on symmetric equilibrium. Thus, the index of household j is temporarily dropped. The households lifetime expected utility function is given as: E 0 t=0 β t [ψ log(c t ) + (1 ψ) log(1 n s t m s t)] (5) where c t denotes the consumption, n s t and m s t are labour supplied to firms and banks which are operated by other households and ψ is the weight of consumption in the utility function. 13

14 The budget constraint of the household is given as: q t (1 δ)k t + B h,t 1 R t 1 + H [ t 1 P + w t (n s t + m s P t P t) + qt d yt d d t q t k t+1 B h,t P t H t P t w t (n t + m t ) µ 2 t (j) P d t ] 1 θ φ [ P d t (j) π d P d t 1(j) 1 ] 2 qt d yt d c t γ t = 0 (6) Here, k t is the quantity of capital at start of time period t. Households can sell or buy capital on market and q t is the price of capital in terms of consumption goods. For simplicity, we assume that the aggregate quantity of capital equals to its steady state value k in each period. In each period, δ k units of final goods are transformed into capital. One can think that the capital adjustment cost is so large that it is always optimal for the household to keep the capital stock at its steady state level. The household hold government bonds which can be sold and bought on the bond market. Here, B h,t is the nominal government bond held by domestic households at the end of period t and R t 1 is the nominal interest rate on government bond held from period t 1 to period t. Households also hold high powered money H t, which is interpreted as bank reserves (the household also operate a bank). Finally, φ is the fixed cost of operating retail intermediate goods firm, γ t is a lump-sum transfer (e.g. tax/subsidy), n t and m t are the labour services bought from other households which are used in production of retail intermediate goods and loan monitoring, and q d t = P d t /P t is the price of bundled intermediate goods in terms of final goods. Retail Sector. Each household operates a monopolistic firm that produces a differentiated retail intermediate good y d t (j) using a Cobb-Douglas production technology which 14

15 combines labour services n t and capital k t : y d t (j) = k η t n 1 η t In addition, to operate the retail intermediate goods firm, households have to pay a fixed cost φ. Combining the above equation with the demand function for retail intermediate goods (4), we can have the second constraint faced by households which says the sales should be equal to production of retail intermediate goods, k η t n 1 η t y d t [P d t (j)/p d t ] θ = 0 (7) Let mc t denote the marginal cost for the firms in terms of final goods. Let λ t and ξ t be the Lagrangian multipliers of constraints (6) and (7). Thus, the marginal cost of firms is defined as: mc t = ξ t /λ t (8) The households choose the labour service n t and the cost minimization implies: ξ t /λ t = w tn t (1 η)y d t (9) where w t is the real wage. Since all the households are similar and hold the same amount of capital, thus the marginal cost of production is identical for all the households. In addition, households are also suppliers of labour services, which choose labour supplies n s t and m s t to maximize its lifetime expected utility. The optimizing decision is given as: 1 ψ 1 n s t m s t = w t λ t (10) 15

16 Households take as given the final good price P t, the bundled intermediate good price P d t, the real wage w t and the demand function for its product (4) as given, and set price P d t (j) for their own product. Following Rotemberg (1982), households face a quadratic price adjustment cost which is given as: [ ] µ P d 2 t (j) 2 π d Pt 1(j) 1 q d d t yt d where π d is the steady state inflation of bundled intermediate goods price P d, q d t = P d t /P t is the relative price of bundled intermediate goods in terms of final goods. Since all households are the same, in a symmetric equilibrium with P d t (j) = P d t, the optimal price setting decision is given as: mc t q d t = θ 1 θ + µ θ [( ) π d t π d π 1 t d π βe d t ( λt+1 λ t ) ( ) ( ) ( ) ] y d t+1 q d t+1 π d t+1 π d 1 t+1 π d π d y d t q d t Note that without adjustment cost, that is, µ = 0, the marginal cost in terms of bundled intermediate goods mc t /q d t equals to the inverse of mark-up (θ 1)/θ. The price setting rule in log-linearised form is thus given by: ˆπ d t = βe t (ˆπ d t+1) + κ( ˆmc t ˆq d t ) (11) where κ = (θ 1)/µ. Baking Sector. The household are subject to a liquidity constraint and has to pay for their consumption in time period t using deposit D t held in that period. To get liquidity used in transaction, households can borrow from banks operated by other households using government bonds and capital goods as collateral. Let v be the velocity of money. Thus, 16

17 the liquidity (cash in advance) constraint is given as: c t vd t P t (12) We assume that liquidity constraint (12) always binds. The balance sheet of bank is given as: H t + L t = D t (13) where H t is the high power money (bank reserve), L t is loan to other households, and D t is the nominal deposit. Let rr be the required ratio of bank reserve H t to deposit D t, thus we have, H t D t rr (14) We assume that the reserve-deposit constraint always binds. Loan is produced by combing collateral and loan monitoring efforts. The collateral is the sum of value of government bond and capital. The monitoring is performed by workers. Let B p,t be the quantity of domestic government bond circulating in the private sector of home country. Let B h,t be the amount of government bonds held by domestic households and B f,t be the amount held by foreign households. Thus, B p,t = B h,t + B f,t. As mentioned at the beginning of this section, for the foreign households to manage their oversea assets, they entrust a domestic bank, which sell and buy domestic asset on behalf of them. There is no offshore market for domestic assets and currency. All the domestic assets and currency are circulating in the domestic financial system. Domestic households choose B h,t while taking B f,t as given. Define b p,t = B p,t /P t to be the real amount government bonds circulating in domestic banking system. The loan production function is given as, L t /P t = χ(b p,t + τq t k t+1 ) ω (m t ) 1 ω 0 < ω < 1 (15) 17

18 where χ is the parameter for efficiency of loan production and τ, with 0 < τ < 1 denotes the inferiority of capital goods for collateral purpose. As we can see from the loan production function, even if the government sterilizes its currency intervention by exchanging domestic government bonds for domestic currency, the total value of liquid assets excluding cash, i.e. b p,t + τq t k t+1, available in the economy increases, which can lead to inflation by facilitating loan production. By combing the liquidity constraint of household (12), balance sheet of commercial bank (13), reserve-deposit ratio constraint (14) and loan production function (15), we have, c t = vχ(b p,t + τq t k t+1 ) ω (m t ) 1 ω (1 rr) (16) By differentiating above equation with respect to value of collateral, we have, Ω t = ωc t /(b p,t + τq t k t+1 ) (17) Here, Ω t is defined to be the partial derivative of the loan production function with respect to collateral value. For perfectly competitive markets, the equilibria of markets replicate planner s solution. The financial market is perfectly competitive in this model and one can think the optimization problem faced by household banker as a planner s problem: households choose capital k t+1, government bond b h,t and labour force used in loan production m t (m s t) to maximize its lifetime expected utility (5) subject to its budget constraint (6) and constraint (16). For simplicity, we can just substitute the equation (16) into (5) and (6). The optimal decision 18

19 rules of households are given by: ( ) [ ( )] ψ y d k t+1 : q t λ t = λ t Ω t τq t + β(1 δ)e t (λ t+1 q t+1 ) + βηe t ξ t+1 t+1 (18) c t k ( ) ( ) t+1 ψ R t b h,t : λ t = λ t Ω t + βe t λ t+1 (19) c t π ( ) t+1 ψ (1 ω)ct m t : w t λ t = λ t (20) c t m t Here, the term ( ψ c t λ t ) represents the price of liquidity service. Note that ψ c t is the marginal utility of consumption while λ t is the price of final goods. Since consumption is constrained by the availability of liquidity, thus the difference between is the price of liquidity service. The left side of equation (18) is the cost of holding capital, while the three terms on the right side are the benefits: (i)the first term is the liquidity service produced by the next unit of capital; (ii) the second term is the resale value of capital tomorrow; (iii) the third term represents the marginal productivity of capital. For the equation (19), the left side is the cost of holding government bonds and the two terms on the right are the liquidity service and interest revenue one can get by holding government bonds. 3.3 The Public Sector The central bank holds government bonds and foreign reserves as assets while issuing high power money as liability. In this model, we assume that bank reserves is the only type of base money and that government does not pay interest to its debt held by central bank. Let B g,t be the domestic government bonds held by central bank and F t be foreign reserves. Central bank holds foreign government bonds B g,t as foreign reserves and thus F t = e t B g,t. For simplicity, we assume that central bank does not receive any interest rate payment by holding B g,t and B g,t. Let K C t be the capital of central bank and let K C t = i=t (e i e i 1 )B g,i 1. Thus, K C t 19

20 is compromised solely of capital gain on foreign reserves due to fluctuation of exchange rate. Central Bank Asset Government Bond B g,t Foreign Reserve F t Liability Bank Reserves H t Central Bank Capital K C t The central bank s balance sheet (in difference) is given as, H t B g,t = F t where F t is total value of foreign reserves (in domestic currency), B g,t the government bonds held by central bank, H t is the base money (bank reserves). Let B t denote the total amount of bonds issued by the domestic government. Thus, we have B t = B h,t + B f,t + B g,t, where B h,t, B f,t and B g,t represent the amounts of domestic government bonds held by domestic households, foreign households and the central bank of domestic country. In addition, B p,t = B h,t + B f,t represents the amount of government debt circulating in the private sector and banking system of domestic country. The domestic government has liabilities in the form of outstanding debt B t and assets in the form of government spending and subsidy(tax deduction). Government Asset Subsidy(Tax Deduction) i=0 R i T t+i Government Spending i=0 R i G t+i Liability Government Bond B t 20

21 Here, R denotes the interest rate of government debt. Government raises money by issuing government bond B and spend money on subsidy ( T ) and public goods G. Note that tax is considered as negative subsidy here. The government s balance sheet (in difference) is given by: B t B t 1 (R t 1 1) = G t T t For simplicity, we assume that the tax policy is given as T t = B t 1 (R t 1 1). Tax is exactly equal to the interest payment in each period. We use lower case to denote the real terms of corresponding variables. For example, b t B t /P t and g t = G t /P t are the increase in the government debt and government expenditure in real term. By combing tax policy rule and the balance sheet of government, then we have b t = g t. The government expenditure g t is completely financed by issuing government bonds b t. Government bonds are held by the central bank B g,t and the private sector B p,t. The total outstanding debt of government equals to the sum of B g,t and B p,t, i.e. b t = b p,t + b g,t. Note that b t = g t, thus we have: b p,t + b g,t = g t By combining the balance sheets of central bank and government, we can obtain the balance sheet of the public sector: H t + B p,t = F t + G t (21) which can be re-written as: h t + b p,t = f t + g t (22) 21

22 where h t = h t h t 1 /π t, b p,t = b p,t b p,t 1 /π t and f t = f t f t 1 /π t. Here, the left side of above equation represents the liability of the public sector and the right side is its assets (or expenditure). Open Market Operation. Let b g,t = B g,t /P t be the government bonds held by central banks in the real term. Define b g,t = b g,t b g,t 1 /π t Thus, b g,t denotes the quantity of government bonds sold to the private sector by the central bank in time period t. We assume that the central bank conduct open market operation according to the following rule: b g,t = b 1 (π t π) + b 2 (y t y) b 3 (b p,t 1 b p ) + b 4 f t (23) where b 1 governs the extent to which open market operation responses to inflation, b 2 reflects the consideration of output gap in the conduct of monetary policy, b 4 represents the extent to which foreign exchange market intervention is sterilized. The parameter b 4 is the measure of intervention intensity. For each unit of increase in foreign reserve f t, b 4 units of government bonds is sold out to sterilize the intervention. Foreign Exchange Market Intervention. The central bank buys and sells domestic currency on the foreign exchange market in order to keep exchange rate at its target level. In particular, the central bank can be thought as a market maker in the foreign exchange market, who buys and sells any amount of domestic currency at the exchange rate it announces. We assume that the capital account is not completely open and domestic economy is large, thus central bank has enough power to affect the market exchange rate. As discussed in the introduction, we compare two types of foreign exchange rate policy rules. For the central bank which adopts Type A policy, the only objective of intervention is to stabilize nominal exchange rate. It intervenes the foreign exchange market in order to fix the nominal exchange rate at the target level which is previously set. For the central bank which adopted 22

23 this type of intervention strategy, its exchange rate policy rule can be expressed as: ê t = 0 (24) For the central bank which adopts Type B exchange rate policy, the objectives of intervention are not only stabilizing nominal exchange rate but also controlling inflation and stabilizing output. Their policy rule can be expressed as: ê t = e 1ˆπ t e 2 ŷ t + e 3 ê t 1 (25) where e 1 and e 2 govern the extent to which exchange rate response to domestic inflation and output gap, and e 3 reflects represents the commitment to a crawling peg. 3.4 The External Sector Foreign Demand for Domestic Assets. As mentioned in the introduction, we assume an asymmetrically and partially opened capital account of domestic country. In particular, households of domestic country are not allowed to hold any foreign assets and exporters have to swap their foreign currency revenue with the central bank. Foreign households are allowed to hold both domestic and foreign assets. Foreign households can open an account in a bank operated in domestic country and entrust the bank to manage their oversea assets. To incorporate capital control into the model, we assume that for the foreign households (1)it is costly to adjust portfolio of their oversea assets and (2)any interest revenue from investing on oversea assets is subjected to a fractional loss. Let B f,t (j) denote the one period bonds issued by the government of domestic country 23

24 and held by foreign household j. Let B f,t 1 0 B f,t(j)dj denote the aggregate quantity of oversea assets held by foreign households. In a symmetric equilibrium, B f,t (j) = B f,t. Let B t (j) denote the one period bond issued by the government of foreign country. Let R t and R t be the domestic and foreign interest rates. We use variables with to denote the corresponding variables in the foreign economy. Thus, the simplified budget constraint for a typical foreign household is given by: w t n t + R t 1B t 1(j) + (1 τ t )R t 1 B f,t 1 (j)/e t P t B t (j) + B f,t (j)/e t P t + Π t Γ ((B f,t(j) B f,t 1 )/P t u t )P t e t P t c t δq t k 0 (26) where c t denotes the consumption of foreign household in real term, w t n t denotes the wage income of foreign household and Π t denotes all the profits delivered from the household owned firms. Similar as in domestic country, capital goods in foreign country is fixed at its steady state level and owned by firms. Thus, capital revenue is part of total profits Π t. Here, B f,t (j) is the quantity of bonds issued by the government of domestic country and held by foreign households j and B f,t is its aggregate quantity. In addition, (B f,t (j) B f,t 1 )/P t represents the change of foreign household s holdings of oversea assets (in real term) and u t is a shock in foreign demand for domestic assets. Thus, Γ represents the cost of portfolio adjustment of foreign households. When there is a positive shock on u t, assets B f,t become more preferable to foreign households, which lead to capital inflow to domestic country. Finally, τ t represents the fractional loss of oversea investment. 24

25 The first order condition of foreign households optimization problem are given by: B t : 1 = β E t λ t+1 λ t R t π t+1 (27) B f,t : 1 + Γ ( B f,t /P t u t ) = β E t λ t+1 λ t R t π t+1 e t e t+1 (1 τ t+1) (28) Assume that the additional cost for foreign households to invest oversea assets τ t is increasing in the spread between interest rates of the two countries, R t 1 R t 1. In particular, we assume that τ t (1 λ m )( ˆR t 1 ˆR t 1). Linearising both sides of first order conditions and combing the two equations leads to the equilibrium condition of foreign exchange market: λ m ( ˆR t ˆR t ) = E t ê t+1 + Γ ( b f,t u t ) (29) where b f,t = B f,t /P t. The above equation is actually a modified uncovered interest rate parity (UIP) condition. The left side of the above equation represents the modified interest rates spread and 1 λ m represents the fractional loss for oversea investment. In addition, the two assets B t and B f,t are not perfect substitutes for foreign households. The change of foreign households oversea asset holdings b f,t will lead to change in the UIP premium. Since domestic households are assumed to be not allowed to hold foreign assets, the trade-off between domestic and foreign assets is actually faced by foreign households. Intuitively, an increase in holdings of one assets indicate a higher return premium of that asset. However, empirical studies showed that increases in the supply of domestic assets actually indicate a decrease in the premium of domestic assets(e.g. Siklos and Weymark 2006). Thus, to be consistent with empirical findings, we assume that the marginal cost of adjustment, that is, Γ ( b f,t u t ) is locally decreasing. In particular, we assume that Γ ( b f,t u t ) = λ + λ s ( b f,t u t ) λ c 2 ( b f,t u t ) 2 25

26 As mentioned above, domestic households are not allowed to hold any oversea asset, while foreign households are allowed to hold assets issued by governments of both countries. Thus, the change in the quantity of domestic government bonds held by foreign households, i.e. b f,t is equal to the capital account surplus(deficit). Let ka t denote capital account surplus of domestic economy, thus, ka t = b f,t. Thus, the modified UIP condition can be rewritten as: E t ê t+1 = λ m ( ˆR t ˆR t ) λ s + λ c (ka t u t ) (30) where λ c represent the level of capital control. International Trade. The foreign goods y f t are imported into home country as inputs of final goods. As discussed at beginning of this section, the domestic demand for foreign goods y f t is determined by its relative price and total output of domestic economy y t (2): y f t = (1 α)(e t P f t /P t ) 1 y t In addition, final goods y t are in turn exported to foreign economy. Let x t denote the amount of final goods that are exported to foreign economy. We assume that foreign goods are imperfect substitute of domestic final goods for foreign households, and the foreign demand for exports is given by: x t = ( Pt e t P f t ) ɛ x (31) where x denotes the foreign demand for domestic final goods at steady state and ɛ is the elasticity of foreign demand for domestic goods. Balance of Payment. The current account surplus expressed in real term is given as: ca t = x t e tp f t P t y f t + e tb g,t 1R t 1 B f,t 1 R t 1 P t + γ t (32) 26

27 where ca t is the current account surplus denoted in real term, x t is the total amount of exported goods, (e t P f t /P t )y f t is the total value of imported goods in terms of domestic final goods, the third term is net financial investment revenue and γ t denotes a lump-sum transfer from foreign country to domestic government. When foreign interest rates are higher than domestic interest rates, sterilized intervention may lead to a quasi fiscal cost of government, which may decreases central bank s incentive to intervene. However, in this paper, we assume that central bank only care about fluctuations of inflation, output gap, exchange rate and exports and focus only on the trade off between stabilizing inflation and stabilizing exchange rate. Thus, the net interest revenue or cost of the central bank s foreign reserve holdings will not be taken into account. For simplicity, we set the lump-sum transfer of foreign country to domestic central bank to γ t = (e t Bg,t 1R t 1 B f,t 1 R t 1 )/P t. Finally, the balance of payment of domestic economy is given as: f t = ca t + ka t (33) where ca t and ka t are current and capital account surpluses. Note that ca t is endogenously determined by domestic economic activities and capital account ka t is determined by the modified UIP condition given the path of ê t, which is set by the central bank s exchange rate policy, and then f t is passively determined after them two. 3.5 Equilibrium To close this model, we still need to specify the path of foreign goods price P f t. we assume foreign goods exporters set price of their products according to a pricing to market scheme (PTM, Krugman 1986). In particular, we assume that foreign exporters adjust prices of its 27

28 produce P f t according to domestic inflation and exchange rate changes but with an adjustment lag. Let π f t = P f t /P f t 1 and π t = P t /P t 1 be the inflation rates of foreign goods and final goods prices. Thus, the price setting process of foreign exporters is given by: ˆπ f t = ˆπ t 1 ê t 1 (34) Given the policy-set time paths for b p, e, h, g, time paths for foreign good price P f, R and shock processes, the model determines values of twenty endogenous variables: n, m, w, λ, ξ, mc, Ω, P d, P, q d, q, R, y d, y f, y, x, c, ca, ka, f. 5 4 Calibration The results of calibration are presented in Table 2. All the parameters are divided into three groups. The first group of parameters are those related to steady state of the model, which are calibrated based on long term characteristics of Chinese economy. The subject discount rate β is set to in order to get an annual interest rate roughly around 2%. Note that government bonds can also be used as collateral in loan production and provide collateral service, thus households discount rate is lower than 1 R. The weight of consumption in utility function is calibrated as ψ = 0.4, to ensure that the sum working time in both production sector n and banking sector m is roughly equal to 1/3 of total time endowment. To ensure a high investment rate in steady state, which is a typical characteristics of Chinese economy, the capital share η is set to 0.5, the quarterly depreciation rate δ is set to and the fixed cost of production φ is set to 0.3. Thus, the saving rate at steady state is around 45%. The elasticity of substitution between differentiated goods θ is set to 10, which 5 The equilibrium of this model is described in Appendix 1 28

29 yields a mark-up around For the banking sector, the reserve-deposit ratio is set at rr = 0.15, which is the level of reserve-deposit ratio of Chinese banks at the end of The average velocity of money v is set to 0.15, which roughly equals to the average ratio of Chinese GDP to M2 in the last three years. Following Goodfriend and McCallum (2007), we set τ = 0.2, which represents the inferiority of capital for the purpose of collateral. The collateral share in loan production ω is set to The efficiency of loan production χ is set to 11.8 to ensure that wage income of financial sector is 6.67% of the total wage income of the economy as reported by National Bureau of Statistics of China. For the external sector, the share of domestic bundled intermediate goods in the final goods production function, α, is set to 0.8, which implies a import to GDP ratio of 25% at steady state. The second group of parameters include those which are related only with the dynamics of the model. Following Chang, Liu and Spiegel (2013), the elasticity of foreign demand for domestic final goods, ɛ is set to 1.4. Following Goodfriend and McCallum (2007), κ in the price adjustment process of retail intermediate goods producers (11) is set to We estimated the modified UIP condition (30) using Chinese data. The simple empirical model is given by: E t ê t+1 =λ m ( ˆR t ˆR t ) λ s + λ c ka t u t (35) with u t = ρu t 1 + ε e t where ˆR t ˆR t is the spread between Chinese and US policy interest rates, e t is the exchange rate of Chinese Yuan to US dollar, u t is the shock in foreign demand for domestic assets or capital account shock and ε e t is the innovation of capital account shock. In addition, ka t is 6 The reserve-deposit ratio has been greatly raised since In 2003, the reserve-deposit ratio is about 7%; however, it increased to 20 % in

30 calculated by: ka t = KA te t e tr t Y tr t where KA t is the capital surplus in US dollar and e t is the exchange rate of Yuan to US dollar. Thus, KA t e t is the nominal capital surplus in the Chinese Yuan, which is normalized by the product of the trend components of exchange rate e tr t and the trend components of nominal output Y tr t. The point estimates of λ m and λ s are around 0.11 and 0 respectively. The point estimate of λ c is around Note that the steady state GDP is around 1.4 in our calibrated model. Thus, we set λ c to 0.11 for our model 7. Finally, the point estimate of ρ is about 0.8 and the standard deviation of residual is about The third groups of parameters include those in policy functions. We estimate the exchange rate policy rule (25) using Chinese data during 2005Q1-2012Q4 8. The estimates for e 1, e 2 and e 3 are around 0.63, 0 and For the parameters in open market operation rule, we set b 1 = 0.5(b p /π) and b 2 = 0.5(b p /y), where b p, π and y are steady state values of government debt in the private sector, inflation and output. This implies that monetary policy s responses to inflation and output gap are equally strong. The b 3 is set to 0.2. For the sterilization intensity, Ouyang et al.(2010) and He(2005) find that China had been conducting complete sterilization. However, Goodfriend and Prasad (2007) finds that the share of total net foreign exchange inflows sterilized by the People s Bank of China is between 20 to 48 percent for For the benchmark, we set the intensity of sterilization b 4 to 0.5, which implies that for each unit increase in base money due to foreign exchange intervention, central bank sells 0.5 units of government bonds to sterilize. 7 In the equation estimated, ka is normalized by output trend while in our model ka is level. This implies that λ c in our model should be 0.15/1.4 correspondingly. 8 The exchange rate regime reform of China took place at the beginning of

31 5 Dynamic Analysis In the first exercise, we simulated our calibrated model. The innovations of capital account shock {ε e t} are drawn from an i.i.d normal distribution with standard deviation Then, we calculated the simulated appreciation pressure, EMP t = φ f t ê t and the fraction of pressure removed by intervention, φ f t, where φ = σ e / σ f and σ e and σ f are standard deviation of simulated ê t and f t. The blue line (solid) shows the simulated appreciation pressure and the orange line (dashed) shows the fraction of pressure removed by intervention. The periods highlighted in grey are those in which the annualized inflation rates are higher than 5% and the periods highlighted in pink are those in which the annualized inflation rates are lower than 5%. As shown in Figure 5, the simulated dynamics of currency intervention activities have a pattern which is similar to that observed in China (Figure 4). When inflation is high, domestic currency is allowed to appreciate moderately. When the inflation is low, most part of appreciation pressure are removed by currency intervention. For the second exercise, the innovations of capital account shock {ε e t} are obtained by estimating the modified UIP condition (35). We use the estimation errors as the innovation process. Similar to the first experiment, the appreciation pressure EMP t = φ f t ê t is calculated using model generated ê t and f t. For both graphs, the blue line (solid) shows the appreciation pressure and the orange line (dashed) shows the fraction of appreciation pressure removed by intervention. The periods highlighted in grey are those in which annualized inflation rates are higher than 5%. The first graph in Figure 6 shows the dynamics of appreciation pressure and intervention activities, which are generated by our calibrated model. They are similar to our observations, which are shown in the second graph of Figure The second graph in Figure 6 is actually a part of Figure 4 31

32 In the third exercise, we examine the effectiveness of different sterilization rules after an one percent positive capital account shock, under the benchmark foreign exchange rate policy rule. Let us refer the monetary policy rule (23) with b 4 = 0.4 as weak sterilization rule and the policy rule with b 4 = 1 as full sterilization rule. Figure 7 reports the impulse responses of key macro variables to a positive capital account shock. After a positive capital account shock hits the economy, the central bank intervenes the foreign exchange market in order to stabilize its nominal exchange rate. This increases the total amount of liquid assets (base money and/or government bonds) that are circulating in the domestic banking system, which facilitates the production of liquidity used in transaction and adds pressure to inflation. Even if the currency intervention is fully sterilized, it increases the amount of government bonds that circulates in private sector, which can also be used as collateral in liquidity production. As shown in Figure 7, the price of liquidity service drops significantly after the shock due to increased supply of base money and government bonds which are used in liquidity production. The full sterilization policy stabilizes the economy better than the weak sterilization policy only in the first few periods. However, as shown in the last graph of the second column, when full sterilization policy is adopted, the private sector s holdings of government bonds greatly increases, which may induce significant increase in quasi fiscal costs of government, especially when domestic interest rates are lower than foreign interest rates. Thus, the benefit of full sterilization could be small compared with its cost. In the fourth exercise, we use the estimated foreign exchange rate policy (25) as benchmark and compare two alternative exchange rate policy rules: type A exchange rate policy, which fixes the nominal exchange rate at its target, and type B exchange rate policy, which allow a strong and immediate appreciation in response to inflation. For the benchmark rule, e 1, e 2 and e 3 are 0.63,0 and For type A exchange rate policy, e 1, e 2 and e 3 are set to 0, 0 and 0. For type B exchange rate policy, e 1, e 2 and e 3 are set to 1, 0 and 0.1. Note that in 32

33 type B policy rule, the response of exchange rate to inflation is both stronger (e 1 = 1) and quicker (e 3 = 0.1) than in benchmark rule. Figure 8 displays impulses responses of the key economic variables to a positive capital account shock. Comparing with type A exchange rate policy, type B policy is more successful not only in terms of taming inflation, but also in terms of protecting exports. The inflation rate rises less due to appreciation of nominal exchange rate and exports decreases less when domestic currency appreciates in response to inflation. The type A exchange rate policy is not successful even in the sense of protecting exports due to its inability of controlling domestic inflation. In the fifth exercise, we assume a higher openness of capital account and decrease λ c to 0.07 from its benchmark value Figure 9 displays impulses responses of the key economic variables after a positive capital account shock. As capital account openness increases, it becomes more important for the central bank to take domestic inflation into account when they make decisions on foreign exchange market intervention. Comparing Figure 8 and 9, we can see that if the central bank switches to type B exchange rate policy from type A policy, the inflation rate roughly decreases by 1% in the benchmark case, but 1.5% when the openness of capital account is increased. When the capital account becomes more open, the efficiency of foreign exchange market intervention becomes low and the central bank can benefit more by allowing the nominal exchange rate more flexibility. 6 Conclusion During recent years, the emerging market economies (EMEs), in particular, the BRICs have been actively intervening their foreign exchange markets in order to prevent their currencies from sharp appreciation, which greatly increased domestic inflation pressure. The central 33

34 banks of these economies never want to see sharp appreciation of their currencies since it may harm their export competitiveness. However, domestic inflation relative to foreign inflation also erodes a country s competitiveness position by raising the relative price of its exports. In addition, under the context of current global recession, taming inflation becomes more important and urgent than usual, since high inflation narrows space for economic stimulus in these economies. To examine this issue, we build a two country DSGE model with a banking sector which produces liquidity that is used in transaction by combing collateral and monitoring labour. In addition, we incorporate into the features of emerging economic markets including capital account controls, foreign exchange market intervention, and sterilization. We calibrated our model to Chinese data and simulated the model. The simulated dynamics of appreciation pressure and intervention activities has a similar pattern to that observed in China. When the inflation is high, domestic currency is allowed to appreciate moderately. When inflation is low, most part of appreciation pressure is removed by foreign exchange market intervention. We compared two different exchange rate policy rules with the actual policy rule: type A policy rule which fixes the exchange rate at its target and type B policy rule which allows a strong and immediate appreciation of domestic currency in response to inflation. When there is a positive capital account shock, type B policy rule is more successful not only in terms of taming domestic inflation, but also in terms of protecting export competitiveness as well. Type A policy is not successful even in the sense of protecting exports, due to its inability of controlling domestic inflation. In addition, when the capital account becomes more open, the trade off between stabilizing domestic inflation and stabilizing exchange rate becomes more important. The central bank can benefit more by switching to type B policy from type A policy when the capital account openness increases. 34

35 In addition, we examined the effectiveness of sterilization in terms of controlling domestic inflation. We found that sterilization can help control inflation but have only limited effectiveness. Moreover, full sterilization greatly increases the holdings of government bonds by the private sector, which leads to quasi fiscal cost of government, especially when domestic interest rates are lower than foreign interest rates. The benefit of sterilization could be small compared with its cost. 35

36 Appendix1. Equilibrium Conditions (1 ψ)/(1 n t m t ) = w t λ t (36) k η t n 1 η t y d t = 0 (37) mc t = ξ t /λ t (38) mc t = w t n t /((1 η)y d t ) (39) ˆπ d t = βe t (ˆπ d t+1) + κ( ˆmc t ˆq d t ) (40) c t = vh t /rr (41) ( ) [ ( )]} ψ y d q t λ t = λ t Ω t τq t + β {(1 δ)e t (λ t+1 q t+1 ) + ηe t ξ t+1 t+1 (42) c t k ( ) ( ) ψ R t λ t = λ t Ω t + βe t λ t+1 (43) c t π ( ) t+1 ψ (1 ω)ct w t λ t = λ t (44) c t m t x t = (P t /(e t P f t )) ɛ x (45) y d t = α ( p d t /p t ) 1 yt (46) y f t = (1 α)(e t p f t /p t ) 1 y t (47) y t = α(y t d ) α (y f t ) 1 α (48) y t = c t + x t + g t + φ + δ k + µ ( ) π d 2 t 2 π 1 q d d t yt d (49) ca t = x t (e t P f t /P t )y f t (50) (1 µ m )( ˆR t ˆR t ) = E t ê t+1 + λ s λ c (ka t u t ) (51) f t = ca t + ka t q d t = P d t /P t (52) (53) Ω t = ωc t /(b p,t + τq t k t+1 ) (54) c t = vχ(b p,t + τq t k t+1 ) ω (m t ) 1 ω (1 rr) (55) 36

37 The policy rules are given by: h t + b p,t = f t + g t (56) b g,t = b 1 (π t π) + b 2 (y t y) b 3 (b p,t b p ) + b 4 f t (57) b g,t = b p,t g (58) ê t = 0 or ê t = e 1ˆπ t e 2 ŷ t + e 3 ê t 1 (59) Given the policy-set time path for b p, e, h, g, time path for foreign good price P f, R and shock processes, equations (36)-(59) determines values of twenty endogenous variables: n, m, w, λ, ξ, mc, Ω, P d, P, q d, q, R, y d, y f, y, x, c, ca, ka, f. 37

38 References J. Aizenman and J. Lee. Financial versus monetary mercantilism: Long-run view of large international reserves hoarding. The World Economy, 31(5): , R. C. Burdekin and P. L. Siklos. What has driven chinese monetary policy since 1990? investigating the people s bank s policy rule. Journal of International Money and Finance, 27(5): , G. A. Calvo and C. M. Reinhart. Fear of floating. The Quarterly Journal of Economics, 117 (2): , C. Chang, Z. Liu, and M. M. Spiegel. Monetary policy in a dsge model with chinese characteristics. Federal Reserve Band of San Francisco, working paper, A. R. Ghosh, J. D. Ostry, and C. G. Tsangarides. Shifting motives: Explaining the buildup in official reserves in emerging markets since the 1980s (pdf download) M. Goodfriend and B. McCallum. Banking and interest rates in monetary policy analysis: A quantitative exploration. Journal of Monetary Economics, 54(5): , M. Goodfriend and E. Prasad. A framework for independent monetary policy in china D. He, C. Chu, C. Shu, and A. Wong. Monetary management in mainland china in the face of large capital inflows. Research Memorandum 07/2005, Hong Kong Monetary Authority (HKMA). J. Li. A monetary approach to the exchange market pressure index under capital control. Applied Economics Letters, 19(13): , A. Ouyang, R. Rajan, and T. Willett. China as a reserve sink: The evidence from offset 38

39 and sterilization coefficients. Journal of International Money and Finance, 29(5): , D. Weymark. Estimating exchange market pres-sure and the degree of exchangemarket intervention for canada. Journal of International Economy, 39(3 4): ,

40 Table 1: Foreign Reserve Accumulation and Inflation in the BRICs Time Periods 2003Q1-2012Q4 Foreign Reserves (Trillion USD) Inflation (Percentage) 2003Q1-2007Q4 2008Q1-2012Q4 2003Q1-2007Q4 2008Q1-2012Q4 Brazil Russia India China USA Data Source: St Louis Fed. Table 1 shows the foreign reserve holdings and inflation rates in the BRICs during the last ten years, The first two columns give the average foreign reserve holdings during and The last two columns are the average inflation rates during the two periods. Inflation rates significantly increase in all countries expect for Brazil, which historically experienced high inflation. Note that the recent crisis started at the end of 2007 and the world economy went into recession after that. The US inflation rate actually went down to 2.1 from 2.9. The high inflation narrowed space for economic stimulus in the BRICs. 40

41 Table 2: Parameter Calibration Parameters Description Value steady state parameters β subjective discount rate ψ weight of consumption in utility function η capital share in production function δ quarterly capital depreciation rate θ elasticity of substitution between differentiated goods φ fixed cost of production rr bank reserve-deposit ratio ω share of collateral in loan production function χ efficiency of loan production τ inferiority of capital as collateral v money velocity α cost share of domestic goods in final goods production non-steady state parameters κ price adjustment parameter ɛ elasticity of foreign demand for domestic goods λ m parameters in modified UIP condition λ s parameters in modified UIP condition λ c parameters in modified UIP condition ρ persistence of foreign exchange market shock monetary policy rule b 1 response to inflation 0.5(b p /π) b 2 response to output gap 0.5(b p /y) b 3 policy persistence b 4 sterilization intensity foreign exchange rate policy rule e 1 response to inflation e 2 response to output gap e 3 crawling peg

42 Figure 1: Foreign Reserve Accumulation and Inflation 42

43 Data Source: St Louis Fed. 43

44 Figure 2: Nominal and Real Exchange Rate of USD to Chinese Yuan Data Source: St Louis Fed. In the first graph, the blue line (solid) shows the nominal exchange rate of USD to Chinese Yuan and the orange line (dashed) shows the real exchange rate, which is normalized to the same initial level. The shaded area are the periods in which annualized inflation is higher than 5%. After 2007, the appreciation speed of real exchange rate is faster than that of nominal exchange rate due to higher inflation in China. In the second graph, the red line (dashed) shows the difference between nominal and real exchange rate. When it is increasing, real exchange rate appreciates faster than nominal exchange rate; when it is constant, real exchange rate appreciates as fast as nominal exchange rate. The difference between nominal and real exchange rates increases sharply during high inflation periods. 44

45 Figure 3: Foreign Exchange Market Intervention 45

46 Figure 4: Appreciation Pressure and Currency Intervention Data Source: St Louis Fed. All variables are seasonally adjusted using X12ARIMA and expressed as percentage change. The blue line shows the appreciation pressure on Yuan (i.e. -EMP). The orange line shows the fraction of pressure on appreciation removed by intervention (i.e. φ ˆf t). The difference between the two lines is the actual change of exchange rate. The periods highlighted are those in which annualized inflation rate is higher than 5%. In those periods with high inflation, the People s Bank of China allows its exchange rate to appreciate moderately. When inflation is low, the most part of appreciation pressure is removed by intervention. 46

47 Figure 5: Foreign Exchange Market Pressure and Currency Intervention (Simulated) The foreign innovations of capital account shock, {ε e t }, are drawn from an i.i.d normal distribution with standard error The blue line shows the pressure of foreign exchange market (i.e.-emp), which is calculated using the model simulated data. The yellow line shows the fraction of pressure (on appreciation/depreciation) removed by currency intervention. The periods highlighted in grey are those in which annul inflation rate > 5%. The periods highlighted in pink are those in which annual inflation rate < 5%. The simulated dynamics of currency intervention activity have a pattern which is similar to that observed in China (Figure 4). When inflation is high, the central bank allows its exchange rate to appreciate moderately. When inflation is low, the most part of appreciation pressure is removed by currency intervention. 47

48 Figure 6: Foreign Exchange Market Pressure and Currency Intervention A. Model Generated B. Data For both graphs, the blue line shows the pressure of appreciation (i.e.-emp) and the yellow line shows the fraction of pressure removed by currency intervention. The first graph is calculated based on our model. For the innovations of capital account shock, {ε e t }, we use the estimation error obtained by estimating the modified UIP equation (35). The second graph is calculated using the data, which is actually a part of Figure 4. 48