The impact of foreign reserves accumulation on inflation and sterilization of the state bank of Viet Nam

study: +Space: The thesis examines the impact of foreign exchange reserves on inflation and sterilization in Vietnam. 5 +Time: Thesis studies in Vietnam from the first quarter of 2004 to the second quarter of 2017 1.5. Research methodology The thesis uses quantitative research method (ARDL Bound Test Model and 2SLS Estimation) in combination with synthesis, analysis, comparison and case study. 1.6. Contributions of the thesis The dissertation provides a comprehensive theoretical basis for the mechanism of impact of accumulating foreign exchange reserve effects on inflation and sterilization. The thesis first time examines these issues in the context of a dollarized economy. Moreover, the study first time analysis in deep whether dollarization and global financial crisis change the sterilization’s effectiveness in Vietnam. The dissertation proposes some policy recommendations and experience lessons for the SBV to neutralize unwanted effects on the economy as the SBV accumulates foreign exchange reserves. 1.7. Structure of the thesis The dissertation consists of five chapters: Chapter 1: Introduction of the research Chapter 2: Theoretical background and empirical evidence on the impact of foreign reserves accumulation on inflation and sterilization of the central bank. Chapter 3: Research methodology and data. Chapter 4: Research Results and Discussion Chapter 5: Conclusions and policy implications 6 CHAPTER 2 THEORETICAL BACKGROUND AND EMPIRICAL EVIDENCE ON THE IMPACT OF FOREIGN RESERVES ACCUMULATION ON INFLATION AND STERILIZATION OF THE CENTRAL BANK 2.1. Some concepts  Accumulation of foreign reserves Foreign reserves are those external assets that are readily available to and controlled by monetary authorities for meeting balance of payments financing needs, for intervention in exchange markets to affect the currency exchange rate, and for other related purposes such as maintaining confidence in the currency and the economy, and serving as a basis for foreign borrowing (IMF, 2009). Accumulation of a country's foreign reserves is an activity which gradually increases the country's foreign exchange reserves through a period of time. Inside, foreign reserves are foreign assets that are available and controlled by the monetary authority of the country in order to meet the balance of payments requirements, to intervene in the foreign exchange market for impacting exchange rates and other related purposes.  Sterilization of central bank According to Krugman et al (2012), sterilization occurs when the central bank performs simultaneous transactions on foreign assets and domestic assets for neutralizing the impact of central bank interventions in foreign markets on the domestic money supply. 7 2.2. Mechanism of the impact of foreign reserves accumulation on inflation and sterilization of the central bank 2.2.1. Mechanism of the impact of foreign reserves accumulation on inflation  The mechanism of impact of foreign reserves on inflation through monetary channel Accumulation of foreign exchange reserves affects inflation through the following mechanisms: The increase in foreign reserves led to a change in the money supply of a country, and the increase in money supply affected the country's inflation. This mechanism is explained in two phases: (i) Impact of accumulating foreign reserves on money supply (ii) Impact of money supply on inflation. In closing, the impact of foreign reserves on inflation can be summarized in the following diagram: Foreign reserves   Money Base   Money Supply   Inflation   The mechanism of impact of foreign reserves on inflation through channels of receiving allocated SDRs by IMF Receiving allocated SDR does not cause inflation but the use of SDR can lead to inflation. With increased international liquidity due to loosening foreign reserve limits, countries that receive allocated SDRs will pursue broader monetary policy. This in turn affects the money supply and causes inflation as above mechanism. 2.2.2. The mechanism of sterilization  Sterilization in the narrow sense Sterilization in the narrow sense reflects interventions by central banks to reduce the money base in the economy as the central bank purchases foreign currency in the foreign exchange market. When the central bank purchases foreign currency in the foreign exchange market without a neutral intervention, this result will increase Net Foreign Assets (NFA) and Base Money (MB). If the central bank sterilizes, instead of increasing MB, the Net Domestic 8 Assets (NDA) will decrease to the right amount by an increase in MB if there is no sterilization. Sterilization in the broad sense Sterilization in the broad sense reflects the central bank's intervention to reduce the money supply in the economy as the central bank purchases foreign currency in the foreign exchange market. In order to preform this matter, the central bank will use measures to affect the money multiplier to reduce the ability to make money of commercial banks. 2.3. Empirical evidences 2.3.1. Empirical studies of the impact of foreign reserves accumulation on inflation The previous papers estimate the effect of foreign reserves accumulation on inflation in two ways: though monetary channel and though receiving allocated SDRs by IMF channel. According to the monetary channel, the empirical studies approaches in three way: worldwide, group of countries and different countries. +The studies analyzing in the worldwide include: Heller (1979), Khan (1979), Rabin & Pratt (1981), Heller (1981), Steiner (2009), Steiner (2017). +The studies analyzing group of countries include: Lin & Wang (2005), Elhiraika & Ndikumana (2007), Borivoje & Tina (2015) + The studies analyzing different countries include: Chaudhry et al (2011), Chen & Huang (2012), Zhou et al (2013), Pham Thi Tuyet Trinh (2015). According to the receiving allocated SDRs by IMF channel, there are two representative papers: Neumamn (1973) and Chitu (2016). 2.3.2. Empirical studies of the efficiency of central banks sterilization The empirical studies of the efficiency of central banks sterilization in the world are divided into two main approaches. The first approach primarily examines the relationship between the NDA and the NFA in the monetary policy reaction function with two main estimation methods, the Ordinary Least Square method and the Vector Auto Regression model, includes the papers: Moreno (1996), Takagi & Esaka (2001), Cavoli & Rajan (2006), Aizenman & Glick (2009), 9 Glick & Hutchison (2009), Pham Thi Tuyet Trinh & Nguyen Thi Hong Vinh (2011), Dang Van Dan (2015), He et al (2005). The second approach used simultaneous equations to examine the relationship between NDA and NFA and applies the Two Stage Least Squares (Or Three Stage Least Squares if needing to fix autocorrelation between residuals), includes the papers: Brissimis et al (2002), Ouyang et al (2010), Wang (2010), Ljubaj et al (2010), Ouyang & Rajan (2011). 2.4. The research gaps 2.4.1. The research gap of the impact of foreign exchange reserve on inflation In Vietnam, the effect of accumulating foreign reserves on inflation through using SDRs channel can be considered as insignificant, so this channel is ignored and the thesis only analyzes the relationship between foreign reserves accumulation and inflation through monetary channel. According to the monetary channel, there are many studies established a research equation by examining the direct relationship between foreign reserves and inflation, or through the intermediate variable, money supply, such as Heller (1979), Khan (1979), Rabin & Pratt ( 1981), Abdullateef & Waheed (2010), Chaudhry et al (2011), Zhou et al (2013). Therefore, they do not clearly reflect the relationship between accumulation of foreign reserves and inflation. In addition, there is no study evaluating the effect of foreign reserves accumulation on inflation that examines the characteristics of the dollarized economy. So that, the thesis will focus on clarifying this problem. 2.4.2. The research gap of effectiveness of central bank’s sterilization In general, all previous studies aimed to estimate the sterilization coefficient and the offset coefficient to evaluate the effectiveness of central bank’s sterilization. But in the previous studies, there has been no study assessing the effectiveness of sterilization in the context of dollarized economy. In addition, the global financial crisis has had a significant impact on the performance of central banks and has not been mentioned in previous studies. Thus, the thesis will assess the impact of dollarization and the global financial crisis on the effectiveness of the SBV’s sterilization. 10 CHAPTER 3 RESEARCH METHODOLOGY AND DATA 3.1. Conceptual model 3.1.1. Conceptual model of impact of foreign reserves on inflation The model was originally derived from Steiner (2009). In addition, to consider the effect of accumulating foreign reserves on inflation in the context of a dollarized economy, the authors adds dollarization to the research model. Conceptual model is as follows: P = f (NFA, NDA, mm, V, Y, DL) (3.1) Where mm is the money multiplier; V is the velocity of money; DL is the rate of dollarization of the economy and Y is aggregate output. 3.1.2. Conceptual model of effectiveness of the SBV’s sterilization Based on the study by Brisimiss & ctg (2002), Ouyang & Rajan (2011) and combined with an analysis of the dollarized economy, the thesis uses the following simultaneous equations: NDA*t = 0 +1NFA*t + 2Dmmt +3DCPIt-1 + 4 Yt-1 + 5CAt-1 + 6(r*t+ Etet+1) + 7DDLt-1+8KH +9(d1-1) r,t-1+ 10ut (3.6) NFA*t = 0 +1NDA*t+ 2mmt +3CPIt-1+ 4Yt-1+ 5CAt-1 + 6(r*t+ Etet+1) + 7DLt-1+ 8KH + 9(d2-1) e,t-1 + 10vt (3.7) Where: CPI is the inflation; CA is the current account; r* + Ee is the foreign interest rate plus the expected nominal exchange rate; r is the interest rate volatility; e is the exchange rate volatility and KH is the financial crisis 3.2. Estimation method 3.2.1. Estimation method of the model which evaluates the impact of accumulating foreign reserves on inflation The study uses the ARDL Bounds Test model developed by Pesaran & ctg (2001) to examine cointegration of variables. We then use the ECM error correction model to 11 determine the rate of adjustment in the short term to return to the long-run equilibrium of inflation. Next, we assess the short-term impact and estimate the long-term effects of foreign reserves accumulation on inflation. The author approaches this model as it is an appropriate model for evaluating cointegration of variables in the small sample case. The ECM equation is as follows: ∆𝐶𝑃𝐼𝑡 = 0 + 𝜆𝐸𝐶𝑡−1 + ∑ 𝛼𝑗∆𝐶𝑃𝐼𝑡−𝑗 𝑞1−1 𝑗=1 + ∑ 𝛽1𝑁𝐹𝐴 ∗ 𝑞2−1 𝑗=0 𝑡−𝑗 + ∑ 𝛾𝑗∆𝑁𝐷𝐴 ∗𝑡−𝑗 𝑞3−1 𝑗=0 + ∑ 𝛿𝑗∆𝑚𝑚𝑡−𝑗 𝑞4−1 𝑗=0 + ∑ 𝜑𝑗∆𝑌𝑡−𝑗 𝑞5−1 𝑗=0 + ∑ 𝜇𝑗∆𝑉𝑡−𝑗 𝑞6−1 𝑗=0 + ∑ 𝜌𝑗∆𝐷𝐿𝑡−𝑗 𝑞7−1 𝑗=0 + 𝜀𝑡 (3.8) Where: λ is the speed of adjustment (the short-term adjustment rate of the CPI to return to the long-run equilibrium when independent variables change). q1, q2, q3, q4, q5, q6, q7 are the optimal lags length of variables’ difference in the model ECt-1 is the error in the CPIt-1 regression under independent variables. ECt-1 is defined as follows: 𝐸𝐶𝑡−1 = 0 + 𝐶𝑃𝐼𝑡−1 − 0 − 𝜃1𝑁𝐹𝐴𝑡−1 ∗ − 𝜃2𝑁𝐷𝐴𝑡−1 ∗ − 𝜃3𝑚𝑚𝑡−1 − 𝜃4𝑌𝑡−1 − 𝜃5𝑉𝑡−1 − 𝜃6𝐷𝐿𝑡−1 (3.9) Where: 0 is the intercept of the long-term equation. 1,2, 3, 4,5,6 are the regression coefficients of the long-term equation. Long-term equation are defined as follows: 𝐶𝑃𝐼𝑡 = 𝜃0 + 𝜃1𝑁𝐹𝐴𝑡 ∗ + 𝜃2𝑁𝐷𝐴𝑡 ∗ + 𝜃3𝑚𝑚𝑡 + 𝜃4𝑌𝑡 + 𝜃5𝑉𝑡 + 𝜃6𝐷𝐿𝑡 + 𝑢𝑡 (3.11) 3.2.2. Estimation method of the model which evaluates effectiveness of the SBV’s sterilization The study uses quantitative methods with 2SLS estimation to evaluate the effectiveness of sterilization in Vietnam by measuring sterilization coefficient and offset coefficient from the simultaneous equation system. 12 The estimation method was proposed by Theil (1953) and Basmann (1957). The procedure for estimating 2SLS with simultaneous systems (3.6) and (3.7) is as follows: + Stage 1: Estimation of short equations: Estimating NDA* and NFA* by all predetermined variables in equation (3.6) and (3.7). NDA*t = 0 + 1DMMt + 2DCPIt-1 + 3 Yt-1 + 4CAt-1 + 5(r*t+ Etet+1) + 6DDLt-1 +7KH +8(d1-1)r,t-1+ 9u1t (3.12) NFA*t = 0 + 1MMt +2CPIt-1+ 3Yt-1+ 4CAt-1 + 5(r*t+ Etet+1) + 6DLt-1+ 7KH + 8(d2-1)e,t-1 + v1t (3.13) + Stage 2: Estimate the derived model by substituting NDA * and NFA * in the right- hand side of the equations by ∆𝑁𝐷𝐴∗̂ và ∆𝑁𝐹𝐴∗̂ obtained in stage 1. NDA*t = 0 +1∆𝑁𝐹𝐴∗̂ + 2Dmmt +3DCPIt-1 + 4 Yt-1 + 5CAt-1 + 6(r*t+ Etet+1) + 7DDLt-1+8KH +9(d1-1)r,t-1+ 10u*t (3.13) NFA*t = 0 +1∆𝑁𝐷𝐴∗̂ + 2mmt +3CPIt-1+ 4Yt-1+ 5CAt-1 + 6(r*t+ Etet+1) + 7DLt- 1+ 8KH + 9(d2-1)e,t-1 + 10v*t (3.14) 3.3. Description of variables and data sources 3.3.1. Description of variable and data sources of the model which evaluates the impact of foreign reserves accumulation and inflation The thesis evaluates the effect of foreign reserves accumulation on inflation in Vietnam in the period from the first quarter of 2004 to the second quarter of 2017. 13 Table 3. 2. Variables and data sources of impact of foreign reserves accumulation and inflation Variable Symbol Calculating method Sources Adjusted NFA (1) NFA*t 𝑁𝐹𝐴𝑡 ∗ = 𝑁𝐹𝐴𝑡 − 𝑁𝐹𝐴𝑡−1 𝑒𝑡 − 𝑒𝑡−1 𝑒𝑡−1 𝐺𝐷𝑃𝑛𝑡 IFS 2018 Adjusted NDA (2) NDA*t NDA*t= (MBt /GDPnt) - NFA*t IFS 2018 Money multiplier mmt M2t/MBt IFS 2018 Inflation (3) CPIt CPI is seasonally adjusted by Cenxus X12 IFS 2018 Output Gap (4) Yt GDPrt – GDPpt Thomson Reuters Velocity of Money Vt GDPnt/M2t Dollarization DLt FDt /M2t FDt : Foreign Deposits M2t : Money Supply IFS 2018 Sources: Author’s synthesis (1) The research uses the NFA adjusted for subtracting the increase in NFA due to exchange rate fluctuations. As the exchange rate change, the NFA value in VND as the foreign exchange difference is recorded at the end of the accounting period, this change increases the accumulation of foreign exchange reserves. However, it is just the value on the book rather than the actual increased value of foreign exchange reserves. Accordingly, NFA adjustment has the formula as above. Where: et and et-1 are the VND / USD rates at the end of t and t-1 respectively; GDP is nominal GDP. 14 (2) Because the NDA is calculated in accordance with the NFA, the exchange rate difference is also adjusted as follows: NDA*t= (MBt /GDPnt) - NFA*t. Where: MBt is Money Base (3) The price level P is representative by Inflation CPI. (4) The research uses out put gap as the change of aggregate output of the economy. Output gap is calculated by the difference between the real output and the potential output. Where GDPrt is the real GDP; GDPpt: Potential GDP is calculated using the Hodrick-Prescott filter with a smoothing factor of 1600 in the 9.0 Eviews software. 3.3.2. Description of variable and Data Sources of the model which evaluates effectiveness of the SBV’s sterilization The thesis uses Vietnam's quarterly data from the first quarter of 2004 to the second quarter of 2017. Data sources are mainly from IFS 2018 (International Financial Statistic) and Thomson Reuter DataStream. The calculation of variables and data sources are shown in Table 3.3. Table 3.3. Variables and Data Sources of the model which evaluates effectiveness of the SBV’s sterilization Variable Symbol Calculating method Data Sources Money multiplier mmt Ln(M2t/MBt) IFS 2018 Inflation CPIt Ln(CPIt_sa) CPIt_sa: CPIt is seasonally adjusted by Cenxus X12 IFS 2018 Current account CAt CAt/GDPnt IFS 2018 Foreign interest rate plus the (r*t+Eet+1) r*t+ln(Eet+1) r*t: Three- Month US Treasury Bill rate. IFS 2018 15 expected nominal exchange rate. Eet+1: Average exchange rate VND / USD at time t + 1 Dollarization DLt Ln(FDt /M2t) IFS 2018 d1 and d2 Dummy variable d1 = 2 if NDAt<0; d1 = 0 if NDAt> 0. d2 = 2 if NFAt<0; d2 = 0 if NFAt> 0. Financial crisis KH KH = 1 : From Q1/2007 to Q4/2008 KH = 0 : Remain time Interest rate volatility (4) r,t σ𝑟,𝑡 = ( 1 5 ) √ ∑ (∆𝑟𝑡+𝑖 − ∆�̂�)2 2 𝑖=−2 Where : 𝑟 ̂ = ( 1 5 ) ∑ ∆𝑟𝑡+𝑖 2 𝑖=−2 IFS 2018 Exchange rate volatility (5) e,t σ𝑒,𝑡 = ( 1 5 ) √ ∑ (∆𝑒𝑡+𝑖 − ∆�̂�)2 2 𝑖=−2 Where : 𝑒 ̂ = ( 1 5 ) ∑ ∆𝑒𝑡+𝑖 2 𝑖=−2 IFS 2018 Sources: Author’s synthesis (4) & (5) According to Brissimis et al (2002), the iinterest rate volatility (r,t) and the eexchange rate volatility (e,t) are usually calculated as the standard deviation of the domestic interest rate (daily) and the standard deviation of the exchange rate (daily) within three months. However, due to inaccessibility of data by date, these variables are calculated as the moving average deviation of the interest rate from 5 quarters and the moving average deviation of the exchange rate from 5 quarters. 16 CHAPTER 4 RESEARCH RESULTS AND DISCUSSION 4.1. The status of foreign reserves, dollarization and sterilization in Vietnam 4.1.1. The status of foreign reserves in Vietnam From the first quarter of 2004 to the second quarter of 2017, foreign reserves tend to increase in Vietnam Considering all three indicators of safety threshold of foreign reserves: The ratio of foreign reserves to short-term debt, the ratio of foreign reserves to imports, the ratio of foreign reserves to money supply, although the scale of foreign reserves in Vietnam is not seriously warned but it is still quite low. 4.1.2. The status of dollarization in Vietnam The rate of dollarization has been continuously decreasing since 2004. According to the IMF, a highly dollarized economy when the ratio of foreign currency deposits and total means of payment is over 30%. Thus, during the study period, the rate of dollarization in Vietnam is not as high as the IMF warning, always below 25% and improved gradually. 4.1.3. The status of sterilizing instruments used in Vietnam The SBV mainly sterilizes the money base when intervening in the foreign exchange market using the Open market Operation (OMO). Over time, OMO has gradually been used by the SBV to regulate the monetary market. It helps the SBV stabilizes the money supply and controls inflation. During the period both against inflation and sterilization (2007 and 2008), reserves requirement and other instruments such as refinance rate, rediscount rate, government deposit have the effect of being sterilization instruments. However, these instruments are not used in time to sterilize, they just are used to implement national monetary policy. 17 4.2. Research results 4.2.1. Research results of the impact of foreign reserves accumulation on inflation in Vietnam After satisfying the diagnostic testing conditions of the ARDL Bound Test model, the results of long-term impact are shown as follows: Table 4.13. Estimated long run coefficients and speed of adjustment coefficient Variable Coefficient Standard Error t statistic Prob. NFAt * 0.38 (***) 0.11 3.23 0.0031 NDAt * 0.54(***) 0.15 3.43 0.0019 mmt 0.20(***) 0.02 7.11 0.0000 Yt 0.02(**) 0.00 2.15 0.0396 Vt 0.40(***) 0.12 3.33 0.0024 DLt -0.38(***) 0.05 -7.18 0.0000 ECt-1 -0.28 (***) 0.02 -10.25 0.0000 Sources: Author’s calculation In addition, to evaluate the short-term effects of accumulation of foreign reserves and dollarization to inflation, the author continues to test the Wald Test of the coefficients of NFA *t and DL variables in the difference equation (3.8). The estimation results are as follows: Table 4.14. Wald test results for ECM equations Variable Wald Test t- statistic Prob NFA* 5.49 0.000 DL 8.08 0.000 Sources: Author’s calculation 18 Thus, the results show that the null hypothesis H0, the coefficients of the variables NFA * and DL in the difference equation (3.8) are zero, are rejected. This shows that the coefficients are non-zero. Therefore, in the short run, both NFA *t and DL have an impact on CPI. 4.2.2. Research results of the effectiveness of the SBV’s sterilization The estimation results of the sterilization coefficient and offset coefficient are shown in Table 4.19 Table 4.19. The estimation results of the sterilization coefficient and offset coefficient Variable Equation (3.6) Equation (3.7) NDA*t - -0.880***(0.122) NFA*t -0.680***(0.149) - MMt -0.395***(0.045) -0.372***(0.037) CPIt-1 -0.064(0.153) -0.161(0.118) Yt-1 -0.245(0.332) -0.386(0.261) CAt-1 -0.046(0.073) 0.098**(0.032)  (r*t+Eet+1) 0.034(0.147) 0.172(0.123) DLt-1 -0.015(0.059) -0.145**(0.067) KH 0.046(0.283) 0.527***(0.135) (d1-1)r,t-1 -0.231*(0.132) - (d2-1)e,t-1 - -0.818(0.518) AR(1) - -0.305***(0.114) R2 0.920 0.933 SE 0.023 0.023 Correlation LM Test ( P_ value ) 0.714 0.172 Heteroskedasticity Test: ( P _ value ) 0.119 0.98 ADF Test of residual -7.808*** -7.24(***) Note: (***): Significance at 1%; (**) : Significance at 5% ; (*) : Significance at 10% Data in parentheses () is standard error 19 Sources: Author’s calculation In order to evaluate whether dollarization changes the effectiveness of the SBV's sterilization through offset coefficient, the authors continue to perform tests in Equation (3.7) with the interactive variable DDLt-1 * DNDA. The results showed that DDLt-1 * DNDA was not statistically significant (P value = 0.4135). Although dollarization has an impact on the NFA, dollarization has not changed the level of NDA impact on the NFA. Thus, dollarization does not directly affect the offset coefficient. Continuing the same test for KH, the results are as follows Table 4.20. The result of estimating the coefficient of offset with interactive variable Variable Coefficient Standard Error NDA*t - 0.733 *** 0.148 KH 0.311** 0.011 KH*NDA*t -0.343** 0.138 Note: (***): Significance at 1%; (**): Significance at 5%; (*): Significance at 10% Sources: Author’s calculation Thus, estimation coefficients are statistically significant. The author continues to perform Wald Test with KH variable and interactive variable KH*NDA*t. The result shows that the null hypothesis (these coefficients are zero) is rejected (P value = 0.0001). This mean that there is a difference in NFA variability in crisis condition and in no crisis condition. The results of rolling estimation of sterilization coefficient and offset coefficient are shown in figure 4.22 20 Figure 4.22. The results of rolling estimation of sterilization coefficient and offset coefficient Sources: Author’s calculation 4.3. The sustainability and financial costs of sterilization in Vietnam Pham Thi Tuyet Trinh (2015) studied sterilization in Vietnam with a view from the financial costs. Accordingly, the SBV has not bear the financial cost of the sterilization yet, remains at a surplus. However, this surplus is not sustainable because it depends on the fluctuation of the VND / USD exchange rate and the ratio of USD in Vietnam's total foreign reserves. Therefore, in the future, if the SBV continues to accumulate foreign reserves, the sustainability of the sterilization will be uncertain. 4.4. Sterilization experience of some central banks and lessons for Vietnam After studying the sterilization in China, India, South Korea, Thailand, Colombia and Slovenia, the author finds the following lessons: The purpose of sterilization is to balance liquidity in the domestic money market as the central banks accumulates foreign reserves. The main sterilization instrument used by all countries is OMO. In addition, some countries also use other sterilization tools such as reserves requirement and government deposits to stabilize the market. In addition to sterilization, other financial policy measures are also used to control capital inflows, such as in Korea and Thailand. -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 Hệ số can thiệp trung hòa +2se -2se -1.1 -1 -0.9 -0.8 -0.7 -0.6 -0.5 Hệ số bù đắp +2se -2se 21 In terms of efficiency, sustainability, and financial costs of sterilization, in general, with the efforts of the centr