See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/353914727
DEALING WITH THE IMPACT OF MONETARY POLICY IN THE US AFTER THE GFC:
CAPITAL FLOWS TO THE SEACEN ECONOMIES AND INDONESIA'S POLICY EXPERIENCE
Article · December 2020
CITATIONS
14
READS
122 2 authors, including:
Solikin M. Juhro Bank Indonesia
92PUBLICATIONS 806CITATIONS SEE PROFILE
All content following this page was uploaded by Solikin M. Juhro on 15 August 2021.
The user has requested enhancement of the downloaded file.
WORKING PAPER
DEALING WITH THE IMPACT OF MONETARY POLICY IN THE US AFTER THE GFC: CAPITAL FLOWS TO THE SEACEN ECONOMIES AND INDONESIA’S POLICY EXPERIENCE
Solikin M. Juhro Reza Anglingkusumo 2020
WP/ 5 /2020
This is a working paper, and hence it represents research in progress. This paper represents the opinions of the authors, and is the product of professional research.
It is not meant to represent the position or opinions of the Bank Indonesia. Any errors are the fault of the authors.
2
Dealing with the Impact of Monetary Policy in the US after the GFC:
Capital Flows to the SEACEN Economies and Indonesia’s Policy Experience
Solikin M Juhro1 and Reza Anglingkusumo1,2
Abstract
This paper empirical shows that unconventional monetary policy (UMP) in the US after the global financial crisis (GFC) affects capital inflows to SEACEN economies. For open middle income SEACEN economies, such as Indonesia, capital flows volatility induced by the UMP in the US adds to the complexity of managing monetary policy trilemma (MPT). A recent hypothesis states that in post GFC, it is possible for monetary authority in an open emerging market economy to retain monetary policy sovereignty (MPS) if and only if capital flows is managed, directly or indirectly, regardless the degree of exchange rate flexibility. This paper contends that for the case of Indonesia, MPS remains feasible even without a direct capital control. This supports the argument that MPS depends more on the strength of the policy framework to address domestic policy objectives. We argue that the implementation of central bank policy mix by Bank Indonesia provides such strength.
Keywords: capital inflows, unconventional monetary policy, monetary policy trilemma JEL classification: E22, F32, F36, F41
1 Bank Indonesia Institute
2 Views expressed in this paper are of the authors and do not in any way reflect the views of Bank Indonesia or its Board of Governors. The authors wish to thank Amelia Azjahra Hidayat and Syahid Izzulhaq for their excellent and valuable research assistance. All errors belong to the authors.
3 1 Introduction
This paper is set to accomplish three main tasks. First is to empirically study the effect of the changing size of the US Federal Reserve’s balance sheet, or the so called unconventional monetary policy (UMP) / quantitative easing (QE) during and after the global financial crisis (GFC), on capital flows to the SEACEN economies. Second is to look at the empirical configuration of monetary policy trilemma (MPT) management in Indonesia, a key member of SEACEN, during the decade after the GFC which was marked by UMP / QE in the US. Based on these two empirical studies, third, policy implications for central bank policy will be discussed.
Since the GFC, research on the spill-over effects of US monetary policy on capital flows to emerging markets has proliferated considerably. Among the key contributions are inter alia Joyce et al (2012), International Monetary Fund (2013), Rey (2013, 2016, 2018), Miranda and Rey (2015), Passari and Rey (2015), Georgiadis (2016), and Anaya et al (2017). These studies argue that the post GFC UMP, i.e. the quantitative easing (QE) policy, in advanced economies (AEs), the US in particular, influences global credit condition and cross-border financial flows, including capital flows to emerging market economies (EMEs). Rey (2013, 2016) further contends that amidst the global financial cycle, induced by UMP / QE in AE’s, the resulting cross border capital flows may have morphed the Mundell-Fleming MPT challenges traditionally faced by small and open EMEs into an “irreconcilable duo”, where sovereign monetary policy is possible if and only if the capital account is managed, regardless the exchange rate regime. Built upon Rey’s hypothesis that floating exchange rate regime has no insulation property under the global financial cycle, Han and Wei (2018) show the possible asymmetric effects of monetary policy shocks from advanced economies. Specifically, they argue that without capital controls, a flexible exchange regime offers some monetary policy autonomy when the center country tightens its monetary policy, yet when the center country enters the loosening cycle Rey’s irreconcilable duo may reappear.
4
Despite the proliferation of studies addressing the impact of UMP / QE after the GFC on capital flows to EMEs and the possible policy ramifications that follow, the empirical literatures are rather muted in terms of SEACEN economies as a group. Studies by Lim and Shrestha (2009) and Becker (2016) exhibits a series of discussions on individual country experiences with large and often volatile capital flows and their policy responses, including responses during the post GFC period. Siregar et al (2011) discuss capital flows to SEACEN countries after the GFC and their policy responses, including the management of MPT challenges. This paper therefore adds to the empirical literatures on the effect of the US monetary policy on capital flows to the SEACEN economies.
Our specific contributions are as follows. First, our study is the first to empirically test the impact of US monetary policy, measured by the changes in the US Fed balance sheet, on a select panel of SEACEN member countries, i.e. People’s Republic of China, Hong Kong SAR, India, Indonesia, South Korea, Malaysia, Philippines, Singapore, and Thailand. We use annual data covering the period from 2004 until 2018, hence effectively capture the influence of the US UMP / QE immediately prior, during, and a decade after the GFC. Figure B1 – B9 in Appendix B depict the dynamic of capital inflows to our sample economies. Second, the selected countries in the panel include countries at different stages of financial market development, which naturally add control to our empirical results in addition to other traditional pull factors. Using this empirical set up, we then derive policy implication for SEACEN member countries.
This paper also adds to the literatures on MPT configuration in SEACEN economies amidst volatile capital flows induced by UMP and QE in AEs, for the specific case of Indonesia -- the largest and most open middle income member of the SEACEN --. Capital flows volatility in post GFC refreshes the long-standing policy discussions on managing monetary policy trilemma (MPT) under capital mobility. The MPT, which is originated from the seminal works by Mundell (1962, 1963) and Fleming (1962), captures the policy trade-offs faced by macroeconomic policymakers in an open economy, wherein only two out of the following three
5
policy objectives can be consistently pursued simultaneously: (1) exchange rate stability (ERS), (2) full access to the global capital markets through financial openness (FO), and (3) ability to maintain monetary policy sovereignty (MPS) in pursuit of the domestic policy ends. For EMEs, the choices presented by the MPT are far from trivial. Securing financing access from the global capital markets through FO (defined as the implementation of open current (CA), financial (FA) and capital (KA) account regime), for instance, involves a difficult choice between maintaining MPS and ERS3.
After the GFC, the complexity of navigating the MPT for EMEs operating under FO has sharpened. In the past two decades, more EMEs have graduated from official development assistance (ODA), conducted reform to develop their financial markets for a better access to global financing, and borrowed from global financial markets in local currencies4. As such, the themes of policy discussions on trilemma management have also evolved. The period of early 2000s until the GFC highlighted lively discussions on ‘fear of floating’ to explain the fact that most EMEs do not converge to one of the MPT’s corner solution, i.e. opting for MPS and FO with floating exchange rate, and instead seek to strike an optimal balance amongst the MPT’s three objectives (Calvo and Reinhart, 2002) 5. Specifically, central banks in EMEs prefer to use active monetary policy to limit exchange rate fluctuations and accumulate foreign exchange reserves (‘a war chest’) as self-insurance against external shocks. Since the GFC, the policy discussion on MPT management has progressed with the expansion of global liquidity. Capital
3 A lesson learned from AEs after the break-down of the Bretton Woods system shows that by letting go of the latter, policymakers must learn to live with nominal volatility, for reverting back to a pegged/managed exchange rate regime is costly, or even ‘folly’ (Obstfeld and Rogoff, 1995).
4 In tandem, the last two decades is also characterized by the rise of global trade in value added (TiVA), as middle- income emerging markets/developing economies integrate into the global value chains (GVCs). Participation in GVCs promises supply of foreign exchange, sustained capital accumulation and expansion of productive employment, thereby lowering the probability of these economies falling into the so-called middle-income trap (OECD, 2015; World Bank, 2017). Since within the global TiVA, trade, production networks, and cross border investment, are tightly intertwined, integration into the global value chains as a development strategy naturally necessitates not only open CA regime but also a broader set of financial openness, to include FA and KA openness.
5 Inability to borrow from the global markets in local currency (original sin), liability dollarization that increases the risk of currency mismatch, as well as crisis experience, are among the reasons for the fear.
6
flows management measures enters the discussion immediately after the GFC as EMEs collective seek the IMF advice on how to deal with capital flows amidst ultra-accommodative monetary policy in AEs. Similar discussion is reflected within the SEACEN policy circles as shown in Lim and Shrestha (2009), Siregar et (2011) and Becker (2016).
UMP / QE and the ensuing global financial cycle has led Rey (2013, 2016, 2018) to argue that MPT may have morphed into a dilemma (an “irreconcilable duo”), where MPS is possible if and only if the potential adverse externalities of capital flows are managed directly, or indirectly via macroprudential policies, regardless of the exchange rate regime6. Of late, a growing discourse is emerging in international policy circle on MPT management after the GFC. Specifically, pursuant of multiple objectives by a central bank, as a strategy to maintain macroeconomic and financial stability amidst capital flows, is being viewed more favorably as long as there are matching instruments (in adherence to the Tinbergen Principle), that address specific (identifiable) market failures (Gopinath (2019), Tobias et al (2020)).
Notwithstanding the importance of MPT management in the overall construct of macroeconomic policy, there has never been an attempt to systematically measure MPT configuration in SEACEN economies after the GFC. For Indonesia, Ikhsan et al (2015) analyzed Indonesia’s MPT configuration prior to and immediately after the GFC, and assessed its impact on central bank balance sheet. Yet, the literature has paused afterwards. This paucity will be addressed by this present paper for the specific case of Indonesia’s MPT configuration a decade after the GFC.
Given the aforementioned backdrop, this paper will conclude the following. First, the expansion and contraction of the US Federal Reserve balance sheet, as a direct measure of UMP / QE in the US, positively affects capital flows to SEACEN economies. The effect is robust across countries in terms of portfolio investment flows. This lends support to the complication of
6 See also related discussions in Obstfeld (2015) and more recently in Nelson (2020).
7
macroeconomic policy management, particularly the management of MPT in the SEACEN economies, which comprise of open and internationally integrated emerging economies.
Second, in light of the above, a case study on the configuration of MPT in Indonesia suggests that in the decade after the global financial crisis (GFC) – a period of the implementation of UMP / QE in the US and resumption of global capital flows – the MPT management in Indonesia was characterized by the absence of direct capital control and a preference towards maintaining FO and MPS. Yet, despite such preference, Bank Indonesia (the central bank) tends to also avoid a strict corner solution for it allows room for exchange rate stabilization during episodes of large unexpected shocks in the global financial markets.
Third, MPT management in Indonesia is a reflection of a specific central bank policy framework that has been implemented by Bank Indonesia since 2010. This framework relies on the use of multiple instruments to ensure price and financial system stability. Specifically, this so-called Central Bank Policy Mix (CBPM) is a policy framework that is based on the implementation of inflation targeting framework (ITF) using interest rate as the main instrument, complemented by exchange rate policy, capital flows management and macroprudential measures. Since the GFC, the implementation of this framework has provided the monetary authority with a strong framework to achieve domestic policy objectives amidst capital flows volatility. Hence, the CBPM may provide lessons learned for other open lower middle income SEACEN economies.
For the ease of its presentation, this paper has been divided into several parts after this introduction. In Section II we will discuss the hypotheses development related to (H1) the impact of UMP / QE in the US on capital flows to the SEACEN economies and (H2) the configuration of MPT in Indonesia amidst UMP / QE in the decade following the GFC. Section III outlines the empirical design of empirical study for (H1) and the method to obtain the MPT configuration in (H2). Section IV presents the empirical results and discusses policy implications, as well as the CBPM framework in Indonesia. Section V concludes this paper.
8 2 Hypotheses Development
2.1 The Impact of US Unconventional Monetry Policy on Capital Flows to SEACEN Economies
As explained in Rudebusch (2018), unconventional monetary policy (UMP) in the US conducted by the US Federal Reserve is a non-standard policy to address severe economic downturn during and in the aftermath of GFC, when the ultralow interest rates were not enough to revive output and employment growth sufficiently. There are two types of UMP. First is the forward guidance, through which the US Federal Reserve communicates future short-term interest rates; and second is the purchase of long term government bonds or quantitative easing (QE). This paper analyses the impact of UMP in the US on capital flows to the SEACEN economies through the lens of QE. Despite some of the drawbacks to capture the US Fed’s QE policy as noted by Rudebusch (2018) and Gagnon and Sack (2018), following Gambacorta et al (2014) and Anaya et al (2017), this paper uses the total assets of the Federal Reserve balance sheet as a proximate for QE by the US Federal Reserve.
As discussed in Rudebusch (2018) and Bauer and Rudebusch (2014), through liquidity effect QE may directly reduce the term premium on long term US Government bond yields as well as its expectation component by way of the (indirect) signaling channel. This initial impact on returns of the US Government bonds (safe haven assets), may then affect the long term risk diversification benefits for global investors’ portfolio combination, trigger global portfolio rebalancing, and lead to capital inflows to EME. Mensi et al (2014, 2016) for instance show that capital markets in EMEs provide risk diversification benefits for international investors.
Accordingly, Anaya et al (2017) show that the US UMP / QE shock significantly increases portfolio outflows from the U.S. and equivalently associated with increase in portfolio inflows to and pro-cyclical interest rate response in EMEs. As further noted in Anaya et al (2017) these impacts through capital flows complement the earlier findings by Fratzscher et al (2016b) that UMP has a direct effect on portfolio reallocation between advanced economies and EMEs.
9
Differ from Anaya et al (2017) and Fratzscher et al (2016b), we also include foreign direct investment (FDI) flows and other investment flows (OI) in our analysis. The link between UMP and FDI is less well established. In fact, FDI has been the less volatile component of capital flows to EMEs. Nonetheless, by nature, UMP / QE is a permanent addition to global liquidity, by which one might argue that it will significantly reduce global interest rate and ease global credit conditions, allowing sovereign and corporate borrowers to engage in risk taking in more risky EMEs by financing their next best positive net present value investment projects that would have not been feasible had the global interest rate and credit condition been much tighter. Passari and Rey (2015) show that the US monetary policy indeed influences the global financial condition, while Borio et al (2011) and Brauning and Ivashina (2019) show that there is rather tight linkages between global and domestic financial cycle. As with other investment (OI) flows, Bruno and Shin (2015a, 2015b) and Azis and Shin (2015) suggest the presence of macro-financial linkages involving risk taking channel and cross-border banking which may amplify the impact of global financial condition on credit availability in the EMEs.
Accordingly, this paper will test a hypothesis that UMP / QE, represented by the expansion and contraction of the US Federal Reserve balance sheet, will have a positive impact on all types of capital inflows, i.e. portfolio, direct and other investment flows, to the SEACEN economies.
2.2 The Monetary Policy Trilemma Configuration in Indonesia after the GFC
The decade after the global financial crisis (GFC), a period roughly from 2009 until 2019, highlights a global economy that is marked by elevated risk and uncertainties in the global financial markets. Post GFC uneven recovery in AEs, prolonged consolidation in the Euro area, growth rebalancing in China, the United Kingdom (UK) exit from the European Union (EU), and rising trade tensions between the United States (US) and China, are among the notable features of the period7. However, for central banks, particularly in emerging market economies (EMEs),
7 See various issues of the International Monetary Fund (IMF) World Economic Outlook since the GFC.
10
including in Asia, an essential facet of this period has been the resumption of large and very often volatile capital flows, which can be attributed to the implementation of UMP, i.e. QE policy by AEs, particularly the US.
Azis and Shin (2015) discuss three phases of global liquidity expansion (and retrenchment) and its spill-over effects to emerging Asian economies. The first phase (the period leading up to the GFC in 2008/2009), was marked by banking led capital flows, intermediated by the global banking system. The second phase (began in 2010, after the QE in AEs), was precipitated by the global search for yield on the back of ample USD liquidity, and rapid growth of local currency (LCY) bond markets in Asia. The third phase started after the Fed announcement of US monetary policy normalization, which propelled capital flow reversals. This last phase accentuated the interconnectedness and high international integration of financial sector in Asia.
Recently, Rey (2016, 2018) argues that global financial cycles, i.e. the recurring episodes of increasing/declining risk tolerance and appetite for leverage, which is born by the rising financial globalization, has transformed the challenges about trilemma management in EMEs.
Specifically, financial shocks originating from the core AEs, i.e. the US, may spill across borders, making it harder for EMEs’ central banks to maintain monetary policy sovereignty (MPS). As also shown in Azis and Shin (2015) the episodes of large and volatile capital flows before, during and after the GFC demonstrate potent adverse implications of the global financial cycles on EMEs through the financial and welfare (real) channels.
A corollary to the above, in growingly interconnected and integrated financial markets, lax (tight) financial condition in the core AEs, may prompt lax (tight) global financial condition. It may further trigger an increase (a decrease) in risk tolerance in the global financial markets and induce large capital inflows (outflows) into (out of) EMEs and relax (tighten) borrowing constraints in the capital flow recipient countries. Hence, global financial cycles will determine domestic financial conditions regardless the degree of exchange rate stability (ERS), and may reduce monetary policy effectiveness in economies with a high degree of financial openness (FO),
11
leading into “an irreconcilable duo” where the trilemma morphs into a dilemma, such that MPS is possible if and only if capital mobility is restricted or a low degree of FO is preferred (Rey, 2016, 2018). Yet, Nelson (2020) shows that in a financially open economy with the floating exchange rate, MPS can still be achieved because authorities will have stronger control over domestic policy objectives. On this, a recent study by Eichengreen et (2020) concludes that, contrary to Obstfeld (2015), a flexible exchange rate has insulating properties against external shocks in EMEs. Additionally, if large exchange rate fluctuations remain of a concern for financial stability, maintaining a large international reserves buffers (“a war chest” in the sense of Calvo and Reinhart (2002)) may not always be sufficient, and additional tools, such as macroprudential policies, are needed, provided the country has initially low financial fragility (Aizenman, 2019).
In this paper we hypothesize that despite UMP / QE by AEs in post GFC and the volatility of global financial condition (and hence cross-border capital flows) that comes with it, as a middle income economy aspiring to sustain its transformation into a higher income group, Indonesia opts to have access to the global capital markets, i.e. implementing a financially open (FO) regime, while ensuring a strong degree of MPS to maintain its internal macroeconomic and financial system stability – the key domestic policy objectives --. This implies that the policy makers in Indonesia will allow the exchange rate to be flexible consistent with its underlying fundamental value. Accordingly, Indonesia’s MPT configuration will lean more towards the MPS and FO combination, vis a vis the MPS and ERS combination. In other words, we will examine Nelson (2020)’s contention that MPS can still be achieved under an FO regime (with no direct capital control), because the monetary authority can have a considerably good control over domestic policy objectives.
12 3 Empirical Methodology
3.1 Econometric Method
To address the first hypothesis and examine the impact of the US UMP / QE on capital flows to SEACEN economies, this paper will conduct empirical investigation by means of fixed effect panel data with country and time specific effect. This approach is in line with empirical studies on global push factors and domestic pull factors to capital flows, for instance as shown in Koepke (2015). The empirical relationship between capital inflows to the SEACEN economies in the sample and the corresponding push and pull factors can be expressed as follows:
𝑦, = 𝛽 + 𝛽 𝑙𝑏𝑠 + 𝛽 𝑖𝑥, + 𝛽 𝑔𝑥, + 𝛽 𝑚𝑠, + 𝛽 𝑡𝑜,
+ 𝛽 ℎ𝑑𝑖, +𝛿 + 𝜓 + 𝜀, (1)
where: 𝑦, denotes capital inflows (or each of the component thereof, i.e. portfolio, direct investment and other investment flow), 𝛽 is intercept and 𝑙𝑏𝑠 is the US Federal Reserve balance sheet (asset side) as the key determinant of capital flows to be examined. As control we include the following variables: 𝑖𝑥, and 𝑔𝑥, to capture nominal interest rate and real economic growth between the SEACEN countries in the sample and the US, respectively; 𝑚𝑠, to account for market size, 𝑡𝑜, for trade openness, and ℎ𝑑𝑖, for human development index. 𝜀, is the error terms, i denotes cross-section, t stands for time, and 𝛽 … 𝛽 are regression coefficients.
In estimating equation (1) we include country specific effect, 𝛿, to capture country specific heterogeneity, and time fixed effect, 𝜓 , to capture common trends across countries. The use of panel fixed effect estimator with country and time specific effect is motivated by the fact that despite its simplicity, this estimator serves our purpose well as it can efficiently summarize our results. The countries in our sample are Hong Kong SAR (HK), India (IND), Indonesia (IDN), Malaysia (MYS), Philippines (PHI), Singapore (SGP), and Thailand (THA), and People Republic of China (PRC), and the Republic of Korea (KOR). We use this sample set to sufficiently control for various stages of financial sector development within the SEACEN given data availability our
13
annual sample period (2004 – 2018). Using the recently launched Financial Development Index (IMF), the diversity of financial development in this sample of countries is reflected by the following values of individual country’s aggregate index in 2018 (ordered from the highest to the lowest): KOR (0.81), HK (0.78), SGP (0.75), THA (0.74), MYS (.66), PRC (0.65), India (0.44), the Philippines (0.37), and Indonesia (0.37). Using this empirical set up, we then derive policy implication for SEACEN member countries.
Our hypothesis dictates that 𝛼 and 𝛽 should be significantly > 0 for both the total and each of the component of capital inflows. The coefficients for the control variables 𝛽 … 𝛽 are also expected to be significantly greater than 0. 𝛽 > 0 and 𝛽 > 0 may reflect benefits from risk diversification by combining EMEs’ and AEs’ assets in a portfolio combination. Standard Capital Asset Pricing Model (CAPM) shows that a combination of risky assets (a portfolio) is less risky than any of its components (see inter alia seminal work by Sharpe (1964) and Merton (1973)). 𝛽 > 0, 𝛽 > 0 and 𝛽 > 0 suggest the importance of idiosyncratic / country specific fundamental pull factors, i.e. market size, trade openness, and level of human capital. Overall, we expect that both the push factor, i.e. UMP / QE in the US as common global shock and the diversification benefits, and the pull factors, i.e. the country specific fundamentals, will affect capital inflows to the SEACEN economies in our study during the sample period. Appendix A Table A1 describes the variable definition, the corresponding data and their sources; while Appendix A Table A2 summarizes the descriptive statistics of the data.
3.2 Constructing MPT Configuration
The empirical measurement of the trilemma hypothesis has been one of the challenging works in empirical open macroeconomics. Creating a systematic metrics that comprehensively measures the achievement in the three policy goals of the MPT is not trivial, as it must adequately capture the complexity of trilemma management faced by policymakers (Ito and Kawai 2014; IK hereafter). An essential scholarly contribution in this field is Aizenman, Chinn, and Ito (2008,
14
2010, 2013; ACI hereafter), who have introduced a method to develop trilemma indices that cover 172 countries from 1960 to 20178. This method amounts to measuring each of the trilemma component separately, i.e. Financial Openness (FO), Monetary Policy Sovereignty (MPS) and Exchange Rate Stability (ERS), before analyzing these components jointly.
Nevertheless, ACI’s methodology for the construction of MPT configuration consists of several drawbacks that are also recognized by ACI (2010) at the outset. First, the simple correlation-based calculation for measuring the MPS indicator is likely to be biased in particular cases, for instance, in the case where similar shocks expose both home and base countries, which produces co-movements of the interest rates in both countries. Second, the use of de jure measures of FO based on the method initially proposed by Chinn and Ito (2006) has its limitations. IK (2014) argue that legal aspects related to international capital flows might not reflect precisely the actual degree of FO, a point that is also recognized by ACI. Specifically, de jure measures of FO developed by Chinn and Ito (2006) tend to be biased against the specific arrangement since the IMF-based variables for de jure FO are too aggregated. Bush (2015) shows that the relation between legal openness and realized international financial flows is weak, especially in financially-vulnerable countries. Lastly, the simple approach by ACI may fail to depict the subtlety of the policy arrangements (IK, 2014).
Table 1
Key Methodological Differences between ACI and IK
ACI IK
Exchange Rate Stability (ERS)
The inverse of the standard deviation of percentage change of bilateral exchange rate.
Time-varying adjusted R- squared collected from an estimation based on a basket of currency.
8 It must be noted that initial results reported by ACI (2008, 2010) were until 2007, and have since been updated until 2017 and reported in Stojkov and Warin (2019).
15
ACI IK
Monetary Policy Sovereignty (MPS)
Correlation of money market interest rate between the home and base country.
The ratio of time-varying R- squared from Taylor-rule estimation and Synthetic Foreign Interest Rate (SFI) Taylor-rule.
Financial Openness (FO) De jure measurement. De facto measurement.
Accordingly, IK propose some refinements to the methodology developed by ACI to off- set the drawbacks mentioned above, but at the cost of fewer country observations (90 countries, including Indonesia, covering 1970 – 2009). Specifically, IK put forward the following improvements (see also Table 1 for a summary): (i) incorporating time-varying estimates on a basket of currency to measure the degree of ERS vis a vis the use of a simple pair-wise bilateral exchange rate measure; (ii) introducing the use of estimated synthetic foreign interests based on a basket of countries to capture the broad stance of ‘global/foreign’ monetary policy; (iii) utilizing the estimation of an augmented Taylor rule, for measuring the degree of MPS; and (iv) applying technical adjustments to ensure the theoretical validity of the measurement.
In this paper, we thus follow closely the methodology proposed by IK (2014) to construct Indonesia’s trilemma configuration from 2010 to 2019. We incorporate Indonesia-specific nuances to capture Indonesia specific idiosyncrasy and estimate several measurements for each of the trilemma indicator based on different data frequency and various empirical specifications.
As such, we will generate several combinations of Indonesian MPT configuration. In this regard, we slightly deviate from and improve on IK (2014), in a positive way, to ensure the sharpness of our results. We then examine the resulting MPT combinations by evaluating its consistency with the trilemma hypothesis and also by comparing our results with the specific results for Indonesia as reported in ACI based on their method using data until 2016. Next, we will be discussing the methodological aspects of our trilemma construction in more detail below.
16 3.2.1 Financial Openness (FO)
In measuring de facto Financial Openness (FO), there are at least two indicators that are proposed by Lane & Milesi-Ferretti (2001, 2007; L-MF hereafter) and IK (2014). In this paper we use these two measurements of de facto FO and compare these indicators to consider which indicator fits better in our case.
The L-MF approach measures FO based on the ratio of the sum of “total assets” and “total liabilities to gross domestic product (GDP), as expressed by the following equation:
𝐹𝑂 =(Total External Assets + Total External Liabilities )
GDP (2)
where 𝐹𝑂 denotes de facto financial openness and GDP represents real gross domestic product. The total external assets comprise of foreign direct investment (FDI) assets, portfolio equity assets, debt assets, financial derivatives assets, and FX reserves. For the total external liabilities, it includes FDI liabilities, portfolio equity liabilities, debt liabilities, and financial derivatives liabilities.
However, IK (2014) argued that L-MF approach possibly consists of crucial issue as it includes foreign exchange reserves (FX Reserves) from FO indicator, while investment made by the monetary authority should not be treated in the same way as a private investment. IK (2014) suggested that if massive foreign exchange reserves are included, FO may be biased upward and the country may appear “financially open”. IK (2014) then augments the L-MF measurement by both excluding foreign exchange reserves and weighting a ratio to total trade as in equation (3) below.
𝐹𝑂
= 1 2
(Total External Assets + Total External Liabilities − FX Reserves ) GDP
+(Total External Assets + Total External Liabilities − FX Reserves ) Export + Import
(3)
17 3.2.2 Exchange Rate Stability
For the measurement of Exchange Rate Stability (ERS), we utilize time-varying adjusted R-squared obtained by estimating the extent of influence of major currencies on Indonesian Rupiah. Specifically, we will empirically estimate the following equation:
𝑒 , = 𝛼 + 𝛽 ,
∈
𝑒 , + 𝜀 (4)
where 𝑒 is IDR per Special Drawing Rights (SDR); 𝐾 is a basket of major currencies.
For the estimation of equation (4), we define 𝐾 as selected-four major currencies per SDR that represent the currencies of Indonesia’s major trade and investment partners (MTIP).
Accordingly, this basket of currencies comprises of China Renminbi (CRM), Euro (EUR), Japanese Yen (JPY), and US Dollar (USD).
We will then proceed with obtaining the Adjusted R-squared using rolling regression technique. We also standardize each exchange rate before estimating equation (4). This standardization is essential for three main reasons: (i) it ensures that the estimated-parameters would not be explosive, (ii) it allows for direct comparison on each estimated-parameter, 𝛽 , and hence for pinpointing which currency primarily drives 𝑒 , and (iii) it gives guidance on which currency has a more significant contribution to the “synthetic” foreign interest rate, which will be used as inputs in the construction of the MPS indicator. ERS is therefore defined as the inverse value of time-varying adjusted R-squared (i.e., 𝐸𝑅𝑆 = 1 − 𝐴𝑅 ) where the higher value of ERS reflects a more stable exchange rate and lower 𝐴𝑅 indicates that foreign currencies affect IDR in a more limited way, vice versa.
3.2.3 Monetary Policy Sovereignty (MPS)
In measuring the Monetary Policy Sovereignty (MPS), IK (2014) employs interest rate reaction function estimation, i.e., Taylor rule. In this paper we also implement this approach. First, in estimating the interest rate reaction function, we utilize Generalized Method of Moment
18
(GMM) instead of OLS. Although Nechio, Carvalho, and Nechio (2019) argued that OLS could appropriately generate a consistent estimation, in this paper, we use Generalized Method of Moment (GMM) estimator as suggested by Caporale, Helmi, Çatık, Menla Ali, and Akdeniz (2018) and Clarida, Galí, and Gertler (2000). Following IK (2014), the following two specifications of Taylor rule will be estimated:
𝑖 = 𝛼 + 𝜌(𝑖 ) + 𝛿(𝑖∗) + 𝜀 (5)
𝑖 = 𝛼 + 𝜌(𝑖 ) + 𝛽(𝜋∗ ) + 𝛾(𝑦∗ ) + 𝜀 (6) where 𝑖, 𝑖∗, 𝜋∗, and 𝑦∗ denote BI policy rate, external factors such as the synthetic foreign interest rate (SFI) and real effective exchange rate (REER), inflation gap9, and output gap, respectively.
Equation (5) and (6) are used to measure the MPS by exploiting its time-varying adjusted R-squared value for 16-quarter rolling window. More specifically, equation (5) estimates the interest rate reaction function with respect to external factor only, while equation (6) approximates domestic Taylor rule function. Accordingly, MPS can be defined as the time-varying Adjusted R-squared (𝐴𝑅) ratio of the 𝐴𝑅 derived from 𝐴𝑅 of equation (6) divided by that of equation (5).
By this definition, the higher value implies higher MPS.
3.2.4 Technical Adjustment and Presentation in a Triangle Space
An approach proposed by IK (2014) to arrive at final MPT configuration is by applying technical adjustments to the three trilemma indicators to obtain a composite indicator that can be used to measure the share (contribution) of each of the trilemma objectives in the final MPT configuration. The adjustments involve constraining the sum of the average value of the three trilemma indicators to two, as expressed in the following equation.
𝐹𝑂 + 𝐸𝑅𝑆 + 𝑀𝑃𝑆 = 2𝐴 ⇒ 𝐴 =𝐹𝑂 + 𝐸𝑅𝑆 + 𝑀𝑃𝑆
2 (7)
9 For the inflation gap measurement, we will compare two indicators: (i) the detrended inflation gap estimated using Hodrick-Prescott Filter (HP Filter), and (ii) deviation between the actual inflation rate and the pre-announced / ex- ante official inflation target.
19
where 𝐴 denotes the adjustment coefficient, and the adjusted value of each trilemma indicator is obtained by dividing it to 𝐴, such that:
𝑋∗=𝑋
𝐴 (8)
with 𝑋 = {𝐹𝑂, 𝐸𝑅𝑆, 𝑀𝑃𝑆) and 𝑋∗ is the adjusted value for each trilemma indicator.
Figure 1
The MPT in a Triangle Space
Next, the MPT configuration can alternatively be presented in a triangular space as in Figure 1. Every axis of the triangle in Figure 1 represents the indicator’s percentage of the composite value. In extreme cases, the policymakers choose entirely two out of three policy objectives10. These choices are captured in point FO-ERS (Area A), MPS-ERS (Area B), and FO- MPS (Area C). The figure also informs a situation where the policymakers strive to address all aspects equally (“the middle ground”), as depicted by the middle point. Three areas in Figure 1, i.e., A, B and C, reflect the policy tendencies concerning the trilemma hypothesis. First, area A portrays a condition where the policymakers mainly emphasize FO and ERS at some expense of
10 As argued in IK (2014), the main proposition of the trilemma hypothesis is that monetary authorities are operating under a constraint of choosing only two out of the three policy choices if they aspire to implement each to the full extent.
20
MPS. Area B is where the policymakers focus on maintaining both MPS and ERS but at the cost of having less FO. The third area, C, represents policymakers who prefer MPS and FO, by allowing the exchange rate to be more flexible.
Appendix C Table C1 describes all the data involved in the construction of MPT in this sub- section and their respective sources.
4 Empirical Results
4.1 The US QE and Capital Inflows to the SEACEN Economies
To analyze the impact of the US Federal Reserve UMP / QE on capital inflows to the SEACEN economies in our sample, i.e. Hong Kong SAR (HK), India (IND), Indonesia (IDN), Malaysia (MYS), Philippines (PHI), Singapore (SGP), and Thailand (THA), and People Republic of China (PRC), and the Republic of Korea (KOR), we estimate panel fixed effect regressions with country and time specific dummies as in Equation 1.
Table 2a – 2d below report the baseline results with no control variables, for total investment, portfolio, direct investment, and other investment inflows, respectively. Table 3a – 3d report the results of estimations that include all the control variables. We summarize the results of the estimations of the four different models in each table. Newey – West robust standard errors are used in all estimations to circumvent the problem of serial autocorrelations and heteroscedasticity. Model I includes all the 9 SEACEN countries in the sample (SEACEN-9), Model II excludes PRC from the sample (SEACEN-8), Model III excludes SGP and HK (SEACEN-7), and Model IV excludes PRC, HK and SGP (SEACEN-6).
These variations in the number of countries serve as an additional way to check for the robustness of our results to some possible extreme country characteristics. We check for three possibilities. First is the exclusion of PRC due to the fact that this country is a dominant export or re-export and direct investment hub in Asia (“the Factory Asia”). Hence we exclude PRC in Model II. Second is the exclusion of both HK and SGP as these jurisdictions are the two largest
21
or dominant financial hubs in the region. This is represented by Model III. While Model IV combines the two exclusions.
Table 2 Baseline Estimations (without Control Variables)
2a. Dependent variable : laiit (all investment)
Variable I II III IV
lbst 0.7955***
(0.0469)
0.7468***
(0.0311)
0.7935***
(0.0610)
0.7282***
(0.0403)
Const 1.9624***
(0.6944)
2.6800***
(0.4525)
1.9836**
(0.9031)
2.9439***
(0.5882)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 851.48*** 1112.10*** 256.52*** 235.15***
R-squared 0.9882 0.9922 0.9832 0.9820
2b. Dependent variable : lfdiit (foreign direct investment)
Variable I II III IV
lbst 1.0339***
(0.0934)
1.0143***
(0.1039)
1.0409***
(0.1185)
1.0159***
(0.1379)
Const -3.2467**
(1.4240)
-2.9508*
(1.5791)
-3.3749*
(1.8081)
-3.0017 (2.0982)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 517.86*** 490.10*** 287.03*** 119.81***
R-squared 0.9801 0.9757 0.9723 0.9318
2c. Dependent variable : lpiit (portfolio investment)
Variable I II III IV
lbst 0.8990***
(0.0749)
0.8340***
(0.0648)
0.9543***
(0.0908)
0.8768***
(0.0830)
Const -0.1436
(1.0993)
0.8042 (0.9492)
-0.9476 (1.3327)
0.1819 (1.2140)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 261.54*** 317.77*** 158.83*** 162.36***
R-squared 0.9638 0.9657 0.9680 0.9676
2d. Dependent variable : loiit (other investment)
Variable I II III IV
lbst 0.5529***
(0.0806)
0.4591***
(0.0734)
0.4862***
(0.0995)
0.3952***
(0.0854)
Const 4.3908***
(1.1866)
5.3222***
(1.0827)
4.9234***
(1.4712)
6.2540***
(1.2714)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 816.24*** 1320.95*** 141.05*** 101.62***
R-squared 0.9748 0.9825 0.9569 0.9536
22
2a. Dependent variable : laiit (all investment)
Variable I II III IV
Notes: Robust standard errors are in parentheses ( ). *, **, *** significant at 10%, 5%, and 1% level.
The results reported in Table 2a – 2d show that, without additional control variables, the expansion and contraction of the US Federal Reserve’s balance sheet, lbst, as the proximate for QE policy in the US, significantly and positively affects capital inflows to SEACEN countries in all models (I – IV) both for the total measure of capital inflows (Table 2a) and all its components, i.e. portfolio inflows (Table 2b), foreign direct investment inflows (Table 2c), and other investment inflows (Table 2d). The robust standard error estimates show that the coefficients for lbst are significantly > 0 at 1%. The reported coefficients appear to be strongest for foreign direct investment inflows, followed by portfolio and other investment inflows, respectively.
A different picture emerges as we include the control variables in the regressions. lbst
remains significant across all models for the estimation of total measure of capital inflows, i.e.
laiit (all investment), as reported in Table 3a. However, when we consider the components of capital inflows in Table 3b – 3d, it appears portfolio inflows, lpiit , is the only measure of capital inflows that is consistently (and significantly) affected by lbst across the four different models.
As to the other measures of capital inflows, i.e. lfdiit (foreign direct investment) and loiit (others investment), the results are not consistent across Model I – IV. For lfdiit. , PRC tends to dominate the positive impact of lbst as shown by the results reported in Table 3b. In Model II and Model IV of Table 3 where PRC is excluded from the estimation, one can observe that the coefficient of lbst is not significantly different from 0, but significantly > 0 at 1% and 5% in Model I and III where PRC is included. A similar case is observed for loiit. Comparing the four models in Table 3d, PRC dominates the impact of lbst on loiit since the coefficient for lbst is found to be significantly > 0 in Model I and III, but = 0 in Model II and IV.
Next, we consider the contributions of other explanatory variables that were included as control. As can be observed in Table 3b, the coefficient for market size is significantly > 0 at 1
% across the four models. In Table 3d, level of human capital development, hdiit , is a strong
23
determinant of other investment flows as its coefficient is > 0 at 1% across Model I – IV. While, it is tempting to conclude that these two “pull factors” to capital inflows are robust indicators, we refrain from for taking such claim. Pull factors such as market size, trade openness, interest rate and growth differential might be positively affected by capital flows.
Table 3 Full Estimations (with Control Variable)
3a. Dependent variable : laiit (all investment)
Variable I II III IV
lbst 0.2265***
(0.0516)
0.40678***
(0.0681)
0.1515***
(0.0503)
0.3278***
(0.0731)
ixit 0.0011
(0.0041)
0.0018 (0.0044)
0.0051 (0.0046)
0.0067 (0.0052)
gxit -0.0066
(0.0056)
0.0000 (0.0063)
-0.0023 (0.0075)
0.0097 (0.0094)
msit 0.4087**
(0.1577)
-0.0596 (0.1587)
0.4827***
(0.1579)
0.0237 (0.1580)
toit 0.0005
(0.0004)
0.0004 (0.0004)
-0.0000 (0.0007)
0.0002 (0.0007)
hdiit 7.9919***
(1.5907)
7.9749***
(1.4062)
8.5921***
(1.6677)
8.7443***
(1.4158)
Const -8.3930**
(3.2861)
2.2255 (3.4668)
-9.9124***
(3.2226)
0.3194 (3.4768)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 1322.33*** 1309.88*** 1261.95*** 601.78***
R-squared 0.9951 0.9949 0.9954 0.9916
3b. Dependent variable : lfdiit (foreign direct investment)
Variable I II III IV
lbst 0.4281***
(0.1443)
0.2581 (0.1630)
0.3580**
(0.1693)
0.1711 (0.2119)
ixit -0.0012
(0.0063)
-0.0048 (0.0066)
0.0052 (0.0090)
0.0006 (0.0098)
gxit 0.0067
(0.0110)
0.0014 (0.0133)
0.0223 (0.0140)
0.0153 (0.0201)
msit 0.9928***
(0.2543)
1.4535***
(0.3087)
1.1456***
(0.2767)
1.6505***
(0.3858)
toit -0.0015**
(0.0007)
-0.0013*
(0.0007)
-0.0001 (0.0019)
-0.0003 (0.0020)
hdiit 3.5466
(2.6130)
3.8208 (2.502)
3.6862 (2.5461)
4.1265 (2.5223)
Const -25.4713***
(6.0618)
-36.2413***
(7.3440)
-29.0490***
(6.5551)
-40.9546***
(8.7818)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 703.48*** 545.34*** 518.87*** 136.50***
R-squared 0.9866 0.9830 0.9820 0.9543
24
3d. Dependent variable : lpiit (portfolio investment)
Variable I II III IV
lbst 0.4855***
(0.1097)
0.6883***
(0.1228)
0.4861***
(0.1248)
0.6720***
(0.1369)
ixit 0.0125*
(0.0073)
0.0163**
(0.0077)
0.0186*
(0.0096)
0.0245**
(0.0101)
gxit 0.0152
(0.0100)
0.0195 (0.0122)
0.0169 (0.0126)
0.0254*
(0.0151)
msit 0.8353***
(0.3025)
0.3036 (0.3535)
0.5138*
(0.3077)
0.0279 (0.3607)
toit 0.0009
(0.0007)
0.0006 (0.0006)
-0.0003 (0.0015)
-0.0002 (0.0015)
hdiit 1.8000
(2.9620)
1.3571 (2.6567)
4.8787 (3.1212)
4.3812 (2.6944)
Const -19.3815***
(6.5640)
-6.9250 (7.8560)
-12.9614*
(6.6097)
-1.5437 (8.0723)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 315.55*** 322.73*** 258.52*** 217.18***
R-squared 0.9730 0.9705 0.9769 0.9744
3d. Dependent variable : loiit (other investment)
Variable I II III IV
lbst -0.2534**
(0.1028)
0.0658 (0.1227)
-0.3845***
(0.1147)
-0.0848 (0.1400)
ixit -0.0158***
(0.0058)
-0.0146**
(0.0058)
-0.0099 (0.0077)
-0.0070 (0.0078)
gxit -0.0212***
(0.0081)
-0.0056 (0.0075)
-0.0227**
(0.0103)
0.0024 (0.0112)
msit -0.0618
(0.2738)
-0.9192***
(0.3042)
0.3301 (0.2435)
-0.4740*
(0.2705)
toit 0.0017***
(0.0005)
0.0014***
(0.0005)
0.0003 (0.0013)
0.0007 (0.0013)
hdiit 16.0470***
(2.5600)
16.3043***
(2.3520)
14.4958***
(2.4483)
14.9798***
(2.2891)
Const 3.1389
(5.8942)
22.4747***
(6.6735)
-4.5602 (5.2041)
13.2957**
(6.1548)
Country fixed effect YES YES YES YES
Year effect YES YES YES YES
Observations 135 120 105 90
F 540.33*** 884.29*** 281.40*** 254.36***
R-squared 0.9876 0.9896 0.9827 0.9716
Notes: Newey – West robust standard errors are in parentheses ( ). *, **, *** significant at 10%, 5%, and 1% level.
Accordingly, based on the above empirical results, we contend that the positive and significant impact of UMP / QE in the US on capital inflows to the SEACEN countries in our sample as a group is robust for portfolio inflows and less convincing for foreign direct investment
25
and other investment inflows11. For the other components of capital inflows, i.e. foreign direct investment and other investment, determinants which are more related to the “pull factors” of capital flows, are found to more relevant vis a vis QE policy in the US (a “push factor”), namely market size and the level of human capital development, respectively. But capital inflows may have some positive feed-back impact on these “pull factors”. These findings have some policy implications for the SEACEN as a platform for regional cooperation in the region which we will discuss further in Sub-Section 4.3.
4.2 MPT Configuration in Indonesia in the Decade after the GFC
This sub-section implements the methodology for construction MPT configuration as described in Sub-Section III.2, for the case of Indonesia, to examine the hypothesis that MPT configuration in Indonesia a decade after the GFC is characterized by (a) preference toward maintaining FO and MPS, and (b) domestic monetary policy that remains relevant despite the absence of direct capital control amidst episodes of volatile capital flows. These hypotheses correspond to Area C in Figure 1. Appendix C describes all the data involved in the construction of MPT in this sub-section and their respective sources. Below, we report the results for each of the component, i.e. FO, ERS and MPS, and analyze these parts jointly by the end of this sub- section.
4.2.1 Financial Openness (FO)
Using equation (2) and (3) we obtain two annual indicators for FO. First is the FO indicator calculated using equation (2) or the L-MF method; and second, using equation (3) or the IK method. Figure 2 depicts these two FO indicators. The gauges indicate that prior to GFC Indonesia had been increasingly less open. Yet, this trend was reversed during the GFC, and since then FO has been increasing. This steady increase in the de facto FO in Indonesia is in harmony with the post-GFC’s unconventional QE policy by AEs. As shown in our results in Sub-Section
11 Appendix B reports the expanded results of our estimations in Table 2 and 3.
26
IV.1. for the SEACEN economies and various other studies involving other EMEs, this QE policy prompted a considerable increase in capital inflows to EMEs, including to Asia and Indonesia (see inter alia Park et al (2014), Dahlhaus & Vasishtha, 2014; Lavigne, Sarker, & Vasishtha, 2014;
Ahmed Hannan, 2015; Azis & Shin, 2015; Basri (2016, 2017); Bevilaqua & Nechio, 2016;
Bhattarai, Chatterjee, & Park, 2018).
Figure 2
De Facto Financial Openness
Figure 2 also reveals that the ensuing policy normalization by the Fed in 2013 seems to have no significant effect in reversing the rising trend in Indonesia’s FO indicators. This suggests that despite the turbulence in the aftermath of “taper tantrum”, global investors continue to invest in IDR assets to take advantage of long run diversification benefits, and Indonesia seems to welcome such positive sentiment. Rating upgrades by rating agencies during the study period may also contribute to the sustained inflows amidst the elevated risk in the global markets.
4.2.2 Exchange Rate Stability (ERS)
Before estimating equation (4), we perform unit root tests to infer the generating process of our data and ensure appropriate empirical specification. We also test for structural breaks using the framework proposed by Narayan and Popp (2010). The results are reported in Table 4 and 5.
All currencies are found to be stationary in first-difference as shown by the ADF test statistics.
0.0000 0.5000 1.0000 1.5000 2.0000 2.5000
De Facto Financial Openness (Ito-Kawai) De Facto Financial Openness (LM-F)
27
Taking collectively in a panel setting, these exchange rate indicators in terms of SDR are stationary in first difference, despite sharing common unit root in level series, in the sense of Im, Pesaran, & Shin (2003) as shown by the W-statistic.
We also find structural breaks in the data using the Narayan-Popp unit root test with structural breaks. Looking at the suggested break dates in Table 5 one could argue that, broadly, these dates are heavily associated with the GFC event in 2008/2009 and the US monetary policy normalization which started in 2013.12 These results are also in line with (a) Laeven and Valencia (2013)’s findings that the GFC started in December 2007, then becoming a systemic crisis in October 2008 and lasted until 2011, and (b) the FOMC’s minutes of meeting concerning to the announcement of Fed’s monetary policy stance (see Appendix C Table C2).
We thus include GFC dummy and FOMC dummy based on the Fed’s minute of meetings to ensure that our main ERS estimation is consistent when we control for such events.13 For the FOMC dummy related to US monetary policy normalization, we mark two quarters forward for quarterly-based data. In addition to these events, we also include the COVID-19 pandemic dummy variable to control the period of 2019/Q4 to 2020/Q2. The results are reported in Table 6.
Table 4
Unit Root Test (Without Structural Breaks)
Variables
Quarterly-based Data
Level First
Difference
E_IDR -1.057 -3.383***
E_EUR -3.141 -2.542***
E_CRM -0.961 -3.367***
E_JPY -2.493 -3.836***
E_USD -2.349 -3.989***
Im, Pesaran & Shin (IPS) W-stat -0.976 -14.093***
Sample Period 2000Q1-2020Q1
Notes: The asterik denotes statistical significance *, **, *** at 10 percent, 5 percent, and 1 percent, respectively.
12 For Indonesian case, similar conclusions can be found in Sharma et al (2018) in their study on structural breaks in Indonesia’s macroeconomic time series data.
13 See Appendix D for the list of FOMC meeting.
28 Table 5
Unit Root Test with Structural Breaks
Variables In Level
t-value First Break Second Break
E_IDR -4.749*** 2008Q3 2013Q3
E_CRM -0.968 2005Q2 2008Q2
E_EUR -1.922 2008Q3 2014Q4
E_JPY -2.777 2008Q3 2012Q4
E_USD -1.192 2008Q3 2014Q4
Sample Period 2000Q1-2020Q1
Notes: The asterik denotes statistical significance *, **, *** at 10 percent, 5 percent, and 1 percent, respectively.
Table 6 ERS Estimation
(1) (2)
E_CRM -0.6808***
(0.1064)
-0.6861***
(0.1246)
E_EUR -1.3165***
(0.2653)
-1.1933***
(0.2853)
E_JPY 0.0541
(0.0755) 0.0745
(0.0889)
E_USD -1.0987***
(0.2794)
-0.9074***
(0.2951)
DUMMY_FOMC -0.0427
(0.1748)
DUMMY_GFC -0.2447
(0.2072)
DUMMY_COVID19 0.6851***
(0.2353)
C -9.96E-15
(0.0829) 0.0300
(0.1039)
R-squared 0.7976 0.8192
Adjusted R-squared 0.7871 0.8021
Observations 80 80
Notes: The asterik denotes statistical significance *, **, *** at 10 percent, 5 percent, and 1 percent, respectively. Number in parentheses ( ) represent the HAC-corrected standard error.
The results of the estimation are reported in Table 6 and show that CRM, EUR, and USD robustly affect the exchange rate movement of Indonesia. These results may be related to the fact that the currencies involved in the estimations belong to countries with strong trade and investment ties with Indonesia. As these currencies appreciated, Indonesia’s exchange rate would depreciate, and vice versa, in line with the expected signs.
