The risk of a banking and currency crisis as a result of rapid credit expansion in Southeast Asia

Victor Sint Nicolaas

February 2016

The risk of a banking and currency crisis as a result of

rapid credit expansion in Southeast Asia

University of Amsterdam Faculty of Economics and Business

Supervisor: Egle Jakucionyte

2 Abstract

An ongoing credit expansion in emerging markets after the global financial crisis of 2008 has caused economists to voice their concerns about the region. Research has shown that major credit expansions correlate with crises. However it is unclear whether the levels of corporate debt in 2016 combined with other contextual factors will result in an actual crisis in any specific country. This paper tries to provide clarity on whether there is a risk of a banking or currency crisis in 2016 in Indonesia and the Philippines, two very similar neighboring countries in Southeast Asia. Two recent influential early warning systems provided eight potential banking- and four potential currency crisis indicators. Literature provided a rationale for which trends could signal a crisis. Subsequently, the timing and size of the indicator movements in the five-year period before 2016 were compared to the timing and size of the indicator movements in the five-year period before the Asian financial crisis. By comparing indicators over these periods, we hoped to find distinct similarities or differences that can give clarity on whether or not a crisis might emerge. We find that in both countries there is no indication of either a banking or currency crisis in the short run, but the lack of complete data for eight indicators highlight potential issues with this approach.

Statement of Originality

This document is written by Victor Sint Nicolaas who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3

The risk of a banking and currency crisis as a result of

rapid credit expansion in Southeast Asia

Introduction

A variety of institutions, newspapers and individuals voiced their concern about the economic situation in emerging market countries in late 2015 (Kynge and Wheatley, 2015; Blitz and Moore, 2015; The Economist, 2015). Three indicators were at the root of the speculation: high levels of borrowing, high bond finance exposure and high foreign currency finance exposure (IMF, 2015a). The IMF (2015a) names three global economic challenges that may reinforce the negative effects of the rapid credit expansion: emerging market vulnerabilities, legacy issues from the crisis in advanced economies, and weak systemic market liquidity. Additionally, the Federal Reserve Bank announced on December 16th _{that U.S. interest rates would start to increase again. This can put more pressure}

on capital to move away from emerging economies and in turn worsen their credit problems (Morris, 2015).

Research has shown that major credit expansions are correlated with currency, debt and banking crises (Mendoza, Enrique and Terrones, 2008; Elekdag, Selim and Wu, 2011). When such a crisis does emerge, economic growth can be severely disrupted (IMF, 2015a). However, it is unclear whether the levels of corporate debt in 2015, will result in an actual crisis when contrasted to other macro-economic indicators. The large amount of corporate debt opens up the possibility of more unfulfilled debt obligations when credit contracts again. Depending on which classification you use, there is a general consensus that an inability to fulfill debt obligations in the private sector can lead to a crisis (IMF, 2015a; 2015b; Radelet and Sachs, 1998). First, the ability of banks to repay their obligations are made more difficult if assets turn out to be worthless. This situation may evolve in a bail-out, bank-run restructuring of the bank, or any combination. Second, the lower value of assets can bring the governments balance sheet in trouble and start a debt crisis. and lead to a capital flight which deteriorates the balance of payments and depreciates the currency, leading to a currency crisis. Policy interventions could help, in theory, to modify the economy and prevent a crisis (IMF, 2015a). The choice and timing of policy measures will depend on the type and timing of the expected crisis. Given the levels of leverage in emerging countries in late 2015, it is a very relevant moment to investigate whether a crisis might emerge.

There is no academic agreement on the prediction of crises (Pilbeam, 2013). One method which might help quantify certain indicators and their thresholds is by using a so called Early Warning System (EWS). This type of system was first proposed after the Asian financial crisis of 1997. It aims to provide reliable indicators that signal when a crisis is about to start, and which do not signal when there is no risk of a crisis (Pilbeam, 2013). In order to aid the prediction of a crisis in late 2015 in emerging markets, twelve crisis indicators are analyzed for Indonesia and the Philippines. The indicators are chosen based on their proven significant relationship to the occurrence of currency and banking crises. The timing and size of the indicator movements in the five-year period before the Asian financial crisis are then compared to the timing and size of the indicator movements in the five-year period before 2016. By comparing indicators over these periods, we hope to find distinct similarities or differences that can give clarity on whether or not a crisis might emerge.

4

This paper can contribute to an understanding of the severity of the situation in 2015 in Indonesia and the Philippines by extending the analysis of an IMF (2015a) report. That report aggregated data on many countries for a limited number of indicators to see if a crisis was emerging. In this paper, an analysis of a larger number of indicators is done for only two countries. This gives an opportunity to deliver a more in depth analysis, and to clarify if the concerns for emerging markets are valid for Indonesia and the Philippines in particular.

The structure of this thesis is as follows: the next section contains a selection of influential literature on the Asian financial crisis. After a further exploration of the economic background of Indonesia and the Philippines, concerns about current developments are presented. The methodology is then explained, in which the choices of countries and indicators in the analysis are defended. The resulting indicators are plotted over 1993 to 1997 and 2011 to 2015, after which their levels and movements are compared. Finally, a conclusion is drawn on whether the chosen indicators show movements which could indicate a crisis in 2016 in Indonesia or the Philippines.

Literature review

According to many researchers, certain movements of indicators can predict crises (Candelon et. al., 2012). So called EWS indicators can be analyzed using a variety of econometric models. These models find indicator thresholds for which indicators will signal an upcoming crisis (Candelon et. al., 2012). Finding significant and optimal indicator thresholds to signal crises is the goal of an EWS model. Any analysis of such kind starts by correctly specifying the start of a crisis. If this was not done correctly, the relation between indicators and crises would be subject to a degree of endogeneity. Indicators would then measure the impact of an indicator on the chance of a crisis occurring while it is affected by the crisis. Both in- and outside the academic world, people use a variety of classifications for crises. One classification that divides crises in a similar pattern as literature on EWSs, is the classification of Reinhart and Rogoff (2009). They name the existence of six common types of crises.

First, currency crises are defined by the experience of a ‘large pressure’ on a currency to depreciate (Reinhart and Rogoff, 2009). It is also known as a balance of payments crisis. This is because the supply of a country’s currency exceeds the demand during large balance of payments deficits, which leads to depreciation if the central bank does not intervene (Krugman, 1979). A well-documented way in which a pressure can be created is through credit growth accompanied by monetary expansion, which increases import demand (Candelon et. al., 2012). More recently, problems in the financial sector have increasingly been quoted as another major contributing factor (Radelet & Sachs, 1998). The exact definition of ‘large pressure’ differs across analyses. Reinhart and Rogoff (2009) set the threshold for annual depreciation at 15 percent. If this threshold is exceeded in any period, they classify the period as experiencing a crisis. Depreciation only measures the effect of successful speculative attacks, in which large amounts of currency are sold by speculators (Krugman, 1979). Instead of using depreciation, a number of recent papers use a constructed parameter, the so called Exchange Market Pressure (EMP) (Bussiere and Fratzscher, 2006; Kaminsky et. al., 1998). It is defined as a weighted average of not only exchange rate changes, but also reserve losses and changes in the interest rate. In this way, unsuccessful speculative attacks which only lower reserves are also measured (Bussiere and Fratzscher, 2006).

Second, banking crises are defined by the occurrence of a closing, merger or take-over of a bank by the public sector after a bank run (Kaminsky and Reinhart, 1999). When there is no bank run, but a string of other financial institutions still experience a closing, merger or take-over, it is also

5

called a banking crisis. This definition has its shortcomings; any quantitative specification of a crisis comes with a degree of subjectivity due to indicator levels being affected by government intervention or deposit insurance (Davis and Karim, 2008). For example, the percentage of non-performing loans can be a good indicator, but its levels may be manipulated by an insurance or government intervention when the bank is unable to fulfill its obligations. This information is often not disclosed to the public in full detail in order to prevent crises to cause excessive damage (Davis and Karim, 2008). This can lead to a discrepancy between the moments a crisis starts and the moment it is observed. Furthermore, the start and end date of a banking crisis is hard to pin down due to the variety of possible classifications that different institutions have. Past studies have chosen to focus on multiple criteria to determine the occurrence of a crisis. Demirgüḉ-Kunt and Detragiache (1997) use four different criteria which can independently classify a crisis: more than 10 percent of loans are non-performing; the cost of a sovereign intervention are more than two percent of Gross Domestic Product (GDP); banks are nationalized; a bank run occurs or emergency measures such as deposit freezes are put into place to prevent an acute bank run.

Third and fourth, domestic and external debt crises are defined by the occurrence of debt restructuring (Marchesi, 2003). This may occur if a government has failed to fulfill her debt obligations to either domestic or foreign creditors. To prevent bankruptcy, agreements can be made to move or diminish interest payments. Domestic debt restructurings are often only partially documented at best (Reinhart and Rogoff, 2009). External debt restructuring are often better documented because of communication with international institutions. The IMF often provides programs to restructure debt and these are well documented (Marchesi, 2003).

Fifth and sixth, inflation crises and currency debasements are defined by a large increase in price levels and the lowering of the value of a currency (Reinhart and Rogoff, 2009). Reinhart and Rogoff (2009) set the crisis threshold of inflation at 20 percent. In modern times, some cases of hyperinflation are followed by debasement of the currency in order to stabilize the trust in the use of the national currency. However, currency debasements have been uncommon the past centuries. A very extreme and recent example occurred in Zimbabwe which saw its currency lower in value in 2008 by a factor ten billion (Reinhart and Rogoff, 2009).

One might assume that a sector is still missing: the private sector. Even though high levels of private debt can cause problems, there is no such thing as a private debt crisis. This is because the burden of large corporate defaults lays on the parties that supply funds, like the banking and public sector.

One of the most prominent collections of the crises defined above was the Asian financial crisis of 1997. Multiple countries unexpectedly experienced large crises, and as a result a large drop in GDP growth (Radelet and Sachs, 1998). Five countries were hit the hardest: Indonesia, Malaysia, South Korea, the Philippines and Thailand (Barro, 2001; Chang and Velasco, 1998). As opposed to other Asian countries, they experienced nominal currency depreciations of more than 50 percent from late 1997 to early 1998 and interest rates reached the 25 percent mark (Barro, 2001). The prices of stocks and real estate saw decreases of similar magnitude (Chang and Velasco, 1998). This was particularly worrisome since most bank lending was to corporate entities, which saw their stock and asset values drop, instead of other banks or the government. A large reversal of private capital inflows showed that confidence in the competitiveness of the area decreased. Those inflows decreased by 105 billion US dollar in the five countries named above, equal to 11 percent of their combined GDP (Radelet and Sachs, 1998). A drop in commercial bank lending accounted for 77 billion of the drop, whereas portfolio equity and non-bank lending lost the remaining 23 and 5 billion

6

respectively. This led to increased lending from foreign sources. In the second half of 1997, there was an outflow of 34 billion. The resulting current account deficit was not critically high in itself, but the fact that it was caused by a loss of competitiveness revealed a fundamental overvaluation (Chang and Velasco, 1998). The foreign capital outflow caused an exchange rate depreciation, whilst the disappearance of foreign credit increased the interest rate.

A combination of depreciations, high interest rates and real estate bankruptcies increased the amount of non-performing loans because of low asset values and difficulties to claim new credit (Barro, 2001; Chang and Velasco, 1998). The currency and banking crises were marked by the large depreciation and the amount of non-performing loans. Still, Chang and Velasco (1998) note two discrepancies in this line of reasoning. The large heterogeneity in movements of interest rates and asset prices between countries and the relatively small initial deterioration of competitiveness and the current account cannot fully explain the drops in GDP growth and the widespread capital outflows. It was the resulting composition of short term assets and liabilities which did the real damage. Chang and Velasco (1998) introduce the concept of international illiquidity: a situation in the financial system of a country in which short term obligations in foreign currency exceed the foreign currency at its disposal through convertible assets. Here the financial system of a country is defined as the central bank and all domestic institutions that perform bank-like operations. A proxy for International illiquidity is the ratio of short term loans from international banks to reserves. Chang and Velasco (1998) note that this ratio increased in four of the five worst hit Asian countries between 1994 and 1997. Only Indonesia kept a stable level. Indonesia experienced one of the largest disruptions of all countries even though its current account deficit was only 3.5 percent of GDP in 1996, its export grew fast, the stock market rose strongly and the government had been running a surplus. It was the historically high international illiquidity, combined with a very high ratio of M2 over reserves, made the country very vulnerable to capital outflows.

In Indonesia, Korea, and Thailand, the IMF put emergency programs in place to help undercapitalized banks (Lane et. al., 1999). Banks rushed to cut back lending in order to reach sustainable capital ratios in all countries, but especially in the countries where IMF exercised pressure on the banks (Radelet and Sachs, 1998). Interbank lending stopped short, leading to a downgrade of public debt by Moody’s. Partial corporate debt defaults became a reality in Indonesia, Korea, and Thailand (Radelet and Sachs, 1998). While asset prices dropped, the currency depreciation made imports very expensive. In Indonesia this also led to an inflation crisis in 1998, with inflation rates reaching over 60 percent (IMF, 2015c).

Three trends which the IMF (2015a) noted have sparked renewed attention for emerging markets. First, corporate debt has increased over the past years. As seen during the Asian financial crisis, credit can deteriorate bank and corporate balance sheets, inflate asset prices and overheat the economy. The large debt levels are difficult to uphold when capital flees out of a country (Elekdag and Wu, 2011). Second, debt has increasingly been issued in bonds. Even though bond finance often has a longer maturity than bank finance, it exposes firms to the volatility of financial markets. Third, debt has increasingly been issued in foreign currency. Foreign currency denoted debt can lead to international illiquidity, bringing further risk areas like depreciation to the table.

Ayala et. al. (2015) find three measurable components of corporate debt: outstanding stock of bonds, domestic bank loans and foreign bank loans. They add up to an underreported total but can still proxy movements. From these, they find that in 13 emerging markets debt ratio’s increased considerably, with a large role for foreign exchange debt. There is quite some variation between countries, with China showing the biggest increase of almost 25 percentage points of its GDP. While

7

corporate equity finance decreased in emerging markets from 1.7 percent of GDP in 2008 to 0.5 percent in 2014, bond issuance increased from 0.8 to 3.3 percent of GDP. The percentage is still relatively small compared to the 40.5 percent of GDP of domestic and foreign bank loans. Lastly, the share of foreign exchange bonds increased on average from 5.6 percent in 2008 to 8.0 percent in 2014 (Ayala et. al., 2015).

Country backgrounds

Both Indonesia and the Philippines experienced a banking crisis in the Asian financial crisis as some financial institutions were unable to fulfill their external short-term obligations (Chang and Velasco, 1998; Radelet and Sachs, 1998). They also both experienced a currency crisis, as noted by the massive depreciation they experienced (Radelet and Sachs, 1998). Table 1 shows which months marked the beginning of the banking and currency crises.

Table 1 Start of different crises during the Asian financial crisis

Banking crisis Currency crisis

Indonesia November 1997 August 1997

Philippenes July 1997 July 1997

Source: (Chang and Velasco, 1998; Radelet and Sachs, 1998)

The Philippines and Indonesia are both classified as emerging markets (IMF, 2015b). Before the crisis the Philippines had a floating exchange rate which behaved like a pegged rate because there was so little variation, whereas Indonesia had a crawling peg (Radelet and Sachs, 1998). Both countries let their currency float freely during the currency crisis, the Philippines from July 1997 to December 1997 and Indonesia from August 1997 to March 1999. After these periods, a managed float around the US dollar was introduced in both countries (Ilzetzki et al., 2008). Both economies are relatively open to trade which gives them a good incentive to keep a stable exchange rate. It can be advantageous to uphold for international trade, but it can also be difficult to maintain during speculative attacks (Wilson and Stupnytska, 2007; IMF, 2014).

Two macro-economic variables are shown to mark the economic history of the Indonesian and Philippine economies relative to the world economy: inflation and GDP growth. These indicators proxy financial stability and prosperity. The plots depict the variables starting from 1994, similar to Radelet and Sachs (1998) financial plots, in order to show their movements before the Asian financial crisis. Graph 1 in the appendix shows both countries have had similar historic developments when it comes to economic growth. The image does supports the idea of Radelet and Sachs (1998) that Indonesia has been hit harder by the Asian financial crisis. Since the year 2000 Indonesia seems to be less susceptible to cyclical movements of the world economy. Graph 2 shows that Indonesia has had consistently higher levels of inflation when compared to the Philippines, and was hit harder during the Asian financial crisis with regards to inflation.

In Indonesia and the Philippines, total corporate debt increased by nine percentage points of GDP over the period 2007 to 2014 (IMF, 2015a). Ayala et. al. (2015) report similar numbers: eight percentage points of GDP for Indonesia and ten percent point of GDP for the Philippines over the period 2008 to 2013. Ayala et. al. (2015) attribute the growth of debt in emerging markets primarily to the movements of the global financial cycle, not strong economic fundamentals. In times of financial distress in developed economies, firms in developing countries see more opportunities to receive and expand their financing. He and McCauley (2013) name three factors causing this: bond

8

yields in emerging markets are relatively higher; emerging market interest rates are set low in congruence with developed countries in order to prevent capital fleeing the country and currency depreciation; and funding abroad is cheaper due to an appreciated currency and low foreign interest rates. This results in investors being more willing to invest in developing economies, while domestic players can more easily accrue debt both at home and abroad. Graph 3 shows that historical stock data of Indonesia and the Philippines supports this view. The composite indexes of IDX , PSEi and S&P500 are used to represent the stock indexes of Indonesia, the Philippines and the U.S. respectively. First of all, the IDX and PSEi increased dramatically more than the S&P 500 in 2008, the year which marked the start of the financial crisis in the U.S. (Elekdag and Wu, 2011). Also in 2012, there was an increase in the stock of the PSEi, which was the height of the sovereign debt crisis in Europe. This supports the notion that investors moved away to developing countries during bad economic times in developed countries. The IDX did grow significantly less than the S&P500 in 2012. Indonesia might have already experienced the worst of its debt cycle without experiencing a crisis, since its composite stock is on its way up again (Elekdag and Wu, 2011).

As most emerging markets Indonesia and the Philippines saw an increase in bond financing over the period 2008 to 2013, equal to approximately 10 percentage points of their respective GDPs (Ayala et. al., 2015). The increase of foreign currency denoted debt over GDP in Indonesia and the Philippines did not resemble the emerging market average however, and was around five times lower than their increase in domestic currency denoted debt over GDP. That comes down to an increase of around 1 percentage point of GDP (Ayala et. al., 2015). Chang and Velasco’s (1998) idea of international illiquidity may not pose a problem for the Philippines, since their stock of foreign reserves over GDP increased by more than 50 percentage points in the same period (IMF, 2015c). Indonesia’s stock of foreign reserves decreased by 1.5 percentage points of GDP, and is in more danger.

A large body of literature has been written on which indicators can predict different kinds of crises. Currency crises have spawned the most EWSs, but banking and sovereign debt crises are also represented (Van den Berg et. al., 2008). Each type of crisis has its often used set of indicators and techniques; however most of the indicators are not unique to one type of crisis (Jacobs and Kuper, 2003). Econometric models to analyze indicators have changed over time from linear regressions to multinomial logit and Markov switching models, but the set of indicators have changed little (Candelon et. al., 2012). Kaminsky et al. (1998) are widely cited to have proposed the first EWS. They use 15 indicators for currency crises, and try to minimize the signal-to-noise ratio in predicting crisis thresholds. This ratio is the amount of correct signals divided by the amount of wrong signals that a certain indicator threshold leads to. They find that an effective EWS uses a broad variety of indicators, and the following are most significant for currency crises: foreign reserves, real exchange rate, credit growth, credit to the public sector and domestic inflation. Demirgüḉ-Kunt and Detragiache (1998) pioneered an EWS for banking crises. They use 12 indicators and run a multivariate logit model. They find that the following weak macro-economic fundamentals increased the likelihood of a crisis occurring: low GDP growth, high real interest rates, high inflation and adverse terms of trade shocks. Structural characteristics of the banking sector also play a role, like capital outflows, especially when the private sector bears the credit.

9

Methodology

The analysis of crisis indicators will be done for two countries: Indonesia and the Philippines. These neighboring countries have been chosen for two reasons. Firstly, they were both hit hard by the 1997 Asian financial crisis along with South Korea, Malaysia and Thailand (Elekdag & Wu, 2011). This was the only crisis in the past decades which coincided with a credit boom in those countries (Mendoza and Terrones, 2008). This facilitates the comparison of indicator movements in the situation before 2016 with the situation before the Asian financial crisis. Secondly, among the five countries named above Indonesia and the Philippines currently have the most similar and relatively worst levels of macro-economic fundamentals in the period 2011 to 2014. Yearly GDP growth has fluctuated around 5 percent for Indonesia, the Philippines and Malaysia except for a shock in 2009. Of those three, Indonesia and the Philippines are the only two countries between 2011 and 2014 with annual inflation levels over 2 percent, Purchasing Power Parity GDP per capita below 10.000 dollars, total trade below 65 percent of GDP (World Bank, 2016). In short: emerging market economies on the lower end of development. These similarities are important, they control for external factors that might cause deviations in the movements of indicators. Despite the similarities, it must be noted that the historic movements also have their differences. Philippine GDP growth has been more volatile, Indonesia experienced a much larger shock in the Asian financial crisis, and endless more specific variations exist. It should come to no surprise if differences emerge in the movements of indicators.

In order to assess if Indonesia or the Philippines are likely to experience a crisis in the near future, the levels, movements and trends of crisis indicators before 2016 will be compared to the levels and movements of those indicators before the Asian financial crisis. Indonesia experienced a domestic debt and inflation crisis in 1998, but only banking and currency crises are analyzed because EWSs exist for those crises.

EWSs can provide significant indicators and their signal thresholds. The disadvantage of using the EWS thresholds is that the exact model with the exact indicators will need to be used. One regression model can for example regress five indicators on the chance that a crisis occurs. The resulting regression coefficients of that model can work in any sample period, but only if the same econometric model is constructed. This is outside the scope of this paper, due to the complexity of transformations and econometric techniques. Instead of both supplying indicators and their thresholds, an analysis on the literature of EWSs can solely provide a list of the most significant indicators to predict a crisis. There are dozens of papers that construct EWSs. It is impossible to find objectively perfect indicators because different papers use different indicators, periods and samples, and econometric models. However, EWSs often build further on past papers and improve on the methodologies (Candelon et. al., 2012). Indicators will thus be taken from the most recent influential papers which cite past influential papers. Indicators which are not significant, not measurable, completely lack open-source data or which are not used in analyses for Indonesia and the Philippines not used. To summarize, indicators are found for each type of crisis using two recent influential papers.

The hypothesized relationships between an indicators and the chance of a crisis occurring are taken from Kaminsky et. al. (1998) and Demirgüḉ-Kunt and Detragiache (1997). Subsequently, the timing and size of the indicator movements in the five-year period before 2016 are compared to the timing and size of the indicator movements in the five-year period before the Asian financial crisis. By comparing indicators over these periods, we hoped to find distinct similarities or differences that can give clarity on whether or not a crisis might emerge. If crisis indicators show

10

different patterns across the past and current periods, external factors are sought which are able explain the difference. We can then explain whether the similarity of patterns, combined with an explanation of the divergence of patterns, may lead to a new banking or currency crisis. Data is taken from IMF International Finance Statistics (IFS) (2015), the Bank of International Settlements (BIS) (2015), the World Bank Development Indicators (WDI) (2015), the Asian Development Bank (ADB) (2016) and the Philippine Statistics Agency (PSA). Annual growth rates are computed for each period to avoid excessive volatility in the graphs. Even though the percentage of debt which is denoted in bonds and in foreign currency is what drove the worries in Indonesia and the Philippines, recent EWSs have not shown significant results for these two indicators. The possible existence is only mentioned briefly Kaminsky et al. (1998), but they decide not to include it in their final EWSs due to a lack of significant results.

Indicators of currency crises

The indicators from the pioneering work of Kaminsky et al. (1998) reappear in recent influential literature (Lestano et. al., 2003; Bussiere and fratzscher, 2006; Candelon et. al., 2012). The significance, definition, hypothesized effect on currency crisis occurrence (sign) and data source of the indicators are taken from Candelon et. al. (2012). In their research several universal techniques are proposed to evaluate EWSs. Time series and regional panel models are constructed to create an EWS which is effective both in- and outside the sample it is built on. When compared to other recent literature, their final model has the highest crisis prediction rates based on individual indicators. After a series of tests, they use a regional panel model approach for their final model which includes the indicators from table 2. They also exclude indicators with a Pearson correlation of more than 30 percent.

An empirical relationship has thus been proven between the level of four indicators and the likelihood of a banking crisis to occur. The analysis of Kaminsky et al. (1999) can shed light on the theoretical relationship by explaining the hypothesized effect of indicators on the occurrence of a banking crisis (sign). Balance of payments problems are resolved through a depreciation of the currency. If the government intervenes, the stock of foreign reserves is sacrificed in order to keep a stable exchange rate (Kaminsky et al., 1999). The reduction of foreign reserves can thus be a good proxy for speculative attacks. In some cases reserve growth was already down 20 percent when compared to the usual stock a year before a crisis emerged. Furthermore, exports are delayed if foreign buyers believe the currency is overvalued, because the nominal value of future exports is expected to decrease. At one-and-a half years before a crisis, exports are already lower than usual when accounting for cyclical effects (Kaminsky et al., 1999). The growth of exports is thus also a good proxy.

Large credit growth to the private sector is suspicious. If the new level of loans is not compensated by a sustained higher level of returns, it might be growing an asset price bubble. Loans are used to invest in property, which sees its value inflate. Once the bubble pops and people recognize property for its original value, investment decreases in value and fleeing capital can lead to a collapse of a fixed exchange rate (Kaminsky et al., 1998). As a proxy for total credit to the private sector, claims from depository corporations is used as opposed to claims from the financial corporations, because more data is available.

The yield spread is the difference in yield between long-term government bonds and short-term money interest rates. In an economy with stable growth people discount the value of future returns and yield spreads are positive. It signals the extra yields from delaying investments to the

11

future. The more attractive current returns are in comparison to future returns, the higher the discount rate, and the larger the yield spread. The expectation of future returns is based on the expectation of two variables: the future real interest rate and the future inflation rate. The first one is a proxy for the growth of investments, the latter one for monetary policy. Both credit growth accompanied by monetary expansion and low domestic returns can drive imbalances in the balance of payments (Candelon et. al., 2012). During the European sovereign debt crisis, yield spreads from risk countries like Greece, Spain and Portugal increased considerably (De Santis, 2012). So a large yield spread signals uncertainty of future returns, requiring a risk premium on normal long term bond yields. However, just before a crisis emerges, long term bonds may still be relatively attractive. The negative relationship shown between yield spreads and currency crises is thus very forward looking (Candelon et. al., 2012). Open source data on yield spreads is scarce. A choice has been made in favor of the spread between 2- and 10-year bond yields from the Asian Development Bank (2016a; 2016b) for the period 2011 to 2015. Using the spread between the short term money market and 10-year bond yields would have required the consolidation of multiple sources, while the movements are similar to that of the 2- and 10-year bond yield spreads.

Table 2 Summary of significant indicators for currency crises

Indicators (sign) Definition Data source and

Data frequency Growth of foreign reserves (-) reserves𝑡− reserves𝑡−1

reserves𝑡−1

IFS: RAFAFX_XDR Quarterly Growth of exports (-) exports𝑡− exports𝑡−1

exports𝑡−1

IFS: NX_XDC Quarterly Growth of Domestic credit

over GDP(+) 𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝑐𝑟𝑒𝑑𝑖𝑡 𝐺𝐷𝑃 _𝑡− 𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝑐𝑟𝑒𝑑𝑖𝑡 𝐺𝐷𝑃 _𝑡−1 𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝑐𝑟𝑒𝑑𝑖𝑡 𝐺𝐷𝑃 𝑡−1 IFS: FDSAOP_XDC IFS: NGDP_D_IX Quarterly Yield spread (-) 10-year government bond yield minus the

2-year bond yield

ADB: Interest Rate Spread - 2yrs vs 10yrs - LCY Bond

Quarterly

Sources: (Kaminsky et al., 1998; Kaminsky et al., 1999; Candelon et. al., 2012) Indicators of banking crises

Indicators from the pioneering work of Demirgüḉ-Kunt and Detragiache (1997; 1998; 2000; 2005) reappear in recent influential literature (Jacobs and Kuper, 2003; Davis and Karim, 2008; Davis et. al., 2011). The significance, definition, and data source of the indicators are taken from Davis et. al. (2011). In their research a logit and a binary recursive tree model is constructed to create EWSs. Only the binominal logit model is effective both in- and outside the sample it is built on. In comparison to other recent literature, their final model has the highest crisis prediction rates based on individual indicators. The total percentage of correct identifications of crisis and non-crisis periods is 84 percent. After a series of tests, they use a final binominal logit model which includes the indicators from table 3. It is possible that some of the indicators are correlated, leading to a less efficient analysis. Calculating these correlations is outside of the scope of this paper due to the need for time-series calculations. Luckily, this may not pose any problems because the combined effect of all indicators is tested, not the individual strength of an indicator.

12

An empirical relationship has thus been proven between the level of each indicator and the likelihood of a banking crisis to occur. The analysis of Demirgüḉ-Kunt and Detragiache (1997) can shed light on the theoretical relationship by explaining the hypothesized effect of indicators on the occurrence of a banking crisis (sign). Bad macro-economic developments can have a negative impact on bank performance. Low growth of real GDP means that domestic market players may have less income, which makes it harder to pay back debt. GDP per capita is a proxy for general development in a country. A low level of development signals weakness of financial institutions, which can prevent a crisis sooner. Low external terms of trade mean that domestic exporters may see their income decrease and domestic importers may see their expenses rise, which makes it harder to pay back debt. These factors can raise the level of non-performing loans in the banking sector, which in turn increases the chance of a banking crisis (Demirgüḉ-Kunt and Detragiache, 1997).

Certain characteristics of the banking system can also negatively impact bank performance. If the ratio of M2 to foreign exchange reserves is high, it signals a vulnerability to a sudden capital flight. This is because then the amount of money in circulation is larger than the amount of foreign exchange reserves. In case of large withdrawals of capital, it can potentially be more difficult to maintain the currency peg (Calvo and Mendoza, 1996). Debt repayments and new loans denominated in foreign currency can become more costly. Davis et. al. (2011) derive a negative relationship between M2 to foreign reserves and banking crises. The authors disagree with this sign, and argue that it might have occurred due to a sampling error. A positive sign is used in table 3.

As discussed, growth of real domestic credit to the private sector can increase the likelihood of the emergence of a currency crisis. Davis et. al. (2011) show on the other hand a negative relationship. This is because a decrease in real domestic credit growth can signal a contracting economy. A disproportionate increase in credit is bad, but a disproportionate withdrawal of credit can be even worse. This result, similar to the surprising sign accompanying the yield spread, comes from the fact that an EWS looks at indicator movements just before a crisis occurs.

Regardless of credit growth, the level of credit plays an important role as well. Increased domestic credit to the private sector is seen as dangerous, as this creates vulnerability through the possible size of non-performing loans. Calvo and Mendoza (1996) find that an increase in domestic credit can signal both an increase and a decrease in foreign reserves.

Two more indicators show the strength of the country’s monetary and fiscal policy. A high rate of depreciation of the exchange rate can make debt repayments and new loans denominated in foreign currency more costly. Whether it is mostly the private or banking sector, excessive debt from both groups can deteriorate banks’ balance sheets. A large government surplus as a percentage of GDP is related to a government’s need for financing. Government deficits in developing countries are often financed by loans from the banking system. A large deficit could signal large outstanding loans which, again, deteriorate banks’ balance sheets (Demirgüḉ-Kunt and Detragiache, 1997). Due to the scarcity of open source data, the cash surplus is used instead of the full budget surplus. Table 3 Summary of significant Indicators for banking crises

Indicators (sign) Definition Data source

Growth of Real GDP (-) Real GDP𝑡− Real GDP𝑡−1 Real GDP𝑡−1

IFS: NGDP_R_PT Quarterly/Yearly Real GDP per capita (-) Real GDP per capita in Dollars WDI: GDP per capita

(constant 2005 US$)

13

Yearly Growth of Terms of trade

(-)

ToT𝑡− ToT 𝑡−1 ToT𝑡−1

WDI: Terms of trade Yearly

M2 to foreign reserves (+) Ratio of M2 to foreign exchange reserves of the Central Bank in Dollars

IFS: 35L___XDC IFS: RAFAFX_XDR Monthly

Growth of real domestic credit (-)

Real Dom. credit𝑡− Real Dom. credit 𝑡−1 Real Dom. credit 𝑡−1

IFS: FDSAOP_XDC IFS: NGDP_D_IX Quarterly Domestic credit to GDP

(+)

Ratio of domestic credit to the private sector over GDP

IFS: FDSAOP_XDC IFS: NGDP_XDC Quarterly Depreciation (+) Rate of depreciation of the Real Effective

Exchange Rate

BIS: Effective exchange rate

indices Monthly Fiscal balance to GDP (-) Cash budget surplus scaled by nominal

GDP in home currency

IFS:GBXCCB_G01_CA_XDC IFS: NGDP_XDC

WDI: Cash surplus/deficit (% of GDP)

PSA: National government Cash budget

Quarterly/Yearly

Sources: (Demirgüḉ-Kunt and Detragiache, 1998; Calvo, 1996; Davis et. al., 2011)

Results

This section summarizes the movements of all twelve indicators and their relation to a possible crisis. Individually, each of these movements might occur on a regular basis. It is the simultaneous movements and interplay which might cause a crisis according to the Candelon et. al. (2012). Literature has predicted the direction an indicator must be going for an indication of a crisis. An examination of the indicators before Asian financial crisis provides information about the timing, size and duration of the hypothesized indicator movement. This is then looked for in the five-year period before 2016. Two subsections are presented in which information from all movements is consolidated, and evidence for the occurrence of a currency and banking crisis is interpreted.

Indication of a currency crisis

Graphs 4 and 5 show annual growth of foreign reserves. In Indonesia there was an upward trend of foreign reserve growth from the first quarter of 1994 until the second quarter of 1996, reaching a rate of around 40 percent. The Philippines experienced a large 20 percent point increase and decrease of the same indicator in 1994. After this anomaly the Philippines experienced an upward trend until the third quarter of 1996, reaching a rate of around 65 percent. After the respective peaks of both countries, the growth rates of both countries move to a negative level. This is in line with the negative sign found in the literature. It is noteworthy that while in Indonesia the growth rate had a first large decline in the quarter the crisis starts, while in the Philippines this happens a quarter earlier. At the moments the crises erupt, the growth rates are minus 4 and minus 22 percent, and the decrease in growth was 60 and 40 percentage points in Indonesia and the Philippines respectively. In the period 2011 to 2015, both countries experienced a downward trend until the first and second quarter of 2012, after which no clear trend is visible. A drop in growth rates

14

of the magnitude before the Asian financial crisis only occurs in 2011 for both countries. However, the Indonesian growth rate decreased by 25 percentage points in the last two quarters.

Graphs 6 and 7 show annual growth of exports. Indonesia saw its growth of exports decrease and increase 30 percentage points during 1993 and 1994. After this, a downward trend set in until the second quarter of 1997. The Philippines do not show a clear long term trend for the same indicator. The growth rates of Indonesia and the Philippines move up by 55 and 25 percentage points respectively. The larger increase in Indonesia can be explained by the larger depreciation of the Rupiah, see graph 24. Only the small downward trend of Indonesia is in line with the negative sign found in the literature. In the period 2011 to 2015 no general trend is visible for either country due to large fluctuations. A slow continuing drop in growth rates of the magnitude before the Asian financial crisis does not appear. However, the Philippine growth rate decreased by 12 percentage points in the last four quarters.

Graphs 8 and 9 showannual growth of domestic credit over GDP. Growth of domestic credit over GDP in the Philippines showed a small downward trend from 1993 to 1995, after which an increase set in. The same indicator does not show a clear trend in Philippines. A decrease of 18 percentage points occurs from 1993 to 1994, after which the growth rate increases 20 percentage points until 1996. The increase of the Philippine rate from 1994 to 1996 and the increases of the Indonesian rate from 1995 to 1997 are in line with the positive sign found in the literature. Just like the growth of foreign reserves, the Philippine indicator moves earlier. The increases end in 10 and 30 percent annual growth for Indonesia and the Philippines respectively. In the period 2011 to 2014 only the Philippines growth of domestic credit over GDP sees an upward trend, from 2012 to 2014, which culminates in a 9 percent annual growth rate. The Philippine growth of domestic credit over GDP moves in the same direction as before the Asian financial crisis, but is 20 percentage points lower at the moment.

Graph 10 shows the yield spread. As mentioned in the methodology, the scarcity of open source data on yield spreads has led to the creation of only a graph for 2011 to 2015. Both countries experienced a downward trend which is in line with the negative sign found in the literature. The Indonesian spread decreases from 175 basis points to 38, while the Philippine spread decreases from 260 to 72 basis points, both with occasional upward jumps in between. According to the literature, this would signal that future returns are increasingly preferred over current returns. The yield of the Philippines is higher than that of Indonesia, which corresponds to a higher level of risk for Philippine bonds.

In summary, all indicators show an alignment with the predicted crisis movement of Candelon et. al. (2012) in the period 1993 to 1997 in at least one country. Still, the indicators varied considerably in size and timing of these movements. For Indonesia, the decreasing yield spread signals that bond buyers have a decreased confidence in the performance of the government in the long run. In the short run, only the decrease in foreign reserves signals a weakness of the currency. Even though there was a decrease of domestic credit to GDP, which shows capital was leaving the country, graph 24 shows that the currency is still appreciating. This explains the decrease in foreign reserves. If pressure on the currency persists, and foreign reserves are unable to prevent a depreciation, a currency crisis might emerge. Even though growth of domestic credit has been higher over the course of 2011 to 2013 than 1993 to 1995, the level of credit over GDP in 2014 was still 20 percentage points below the level in 1996, as shown in graphs 22 and 23. Finally, increased growth of exports under an appreciating currency indicates confidence in a future increase of the future nominal export value. All information taken together, in the short run there is no risk of a

15

currency crisis in Indonesia. For the Philippines, again the decreasing yield spread signals a decreased confidence in the long run government performance. Exports are decreasing very strongly, and as mentioned before this can signal a crisis over a year in advance as it shows a structural weakness of the economy. Furthermore, growth of credit over GDP is increasing, but is still 20 percentage points below the level before the Asian financial crisis. Any existing asset bubble in 2015 cannot be as large as in 1997. All information taken together, there is no sign that a currency crisis will occur in the short run in the Philippines, but there are worries about current performance of its economy and the future performance of its government.

Indication of a banking crisis

Graphs 11 and 12 show annual growth of real GDP. The growth rate in Indonesia experienced an upward trend until 1995, reaching 8 percent at its peak, after which a decrease set in towards 5 percent in 1997. The growth rate in the Philippines also experienced an upward trend until mid-1996, reaching 6.5 percent at its peak, after which a decrease set in towards 5 percent in 1997. The decline in growth levels of late 1996 and the beginning of 1997 is in line with the negative sign found in the literature. Again, the Philippine indicator moved earlier. In the period 2011 to 2015 the Philippine growth rate shows an upward trend reaching 8 percent mid-2013, after which a decline sets in. The Indonesian growth rate shows a steady decline from 6.5 to 4.5 percent. Although both countries experience a decreasing trend at some point, the decline is very slow in comparison to before the Asian financial crisis.

Graphs 13 and 14 show real GDP per capita. Both countries show a consistent upward trend both from 1993 to 1997 as well as from 2011 to 2015. This increase is not in line with the negative sign found in the literature.

Graphs 15 and 16 show annual growth of the terms of trade. The growth of terms of trade shows an upward trend reaching 8 percent in 1997 for Indonesia, whereas the growth rate of the Philippines show a downward trend reaching minus 2.5 in 1997. Only the decrease of the Philippine terms of trade growth is in line with the negative sign found in the literature. In the period 2011 to 2015 neither country experiences a large decrease in terms of trade growth for more than a year.

Graphs 17 and 18 show M2 over foreign reserves for Indonesia, while graphs 19 and 20 show the same indicator for the Philippines. Quarterly data was not available for the period 2011 to 2015. The Indonesian ratio is more than a hundred times larger than the Philippine ratio. This is because foreign reserves are measured by the IMF (2015c) in foreign convertible currency. The size of foreign convertible currency vary with the exchange rates in each respective country. The large discrepancy between graph 16 and 18 is also caused by this: the Indonesian real effective exchange rate depreciated considerably between 1997 and 2010, leading to an increase in the ratio of M2 over foreign reserves. Trends can still be compared. Over the period 1993 to 1997 Indonesia saw a doubling of the M2 over foreign reserves, with an extra big increase in the fourth quarter of 1997. The Philippine M2 over foreign reserves experienced a decreasing trend from 1995 to 1996, and then an increasing trend. Both countries only show a very minor increase before the respective crises started. The increase is in line with the positive sign found in the literature, but the minor size is found many times in the graphs. In the period 2011 to 2015 M2 over foreign reserves in the Philippines show a positive trend until 2014 and in Indonesia until 2013. Both increases are larger than the increase before the Asian financial crisis. The level of M2 over foreign reserves in Indonesia is way higher, so there is more risk of depreciation.

16

Graph 21 shows annual growth of real domestic credit. Again the scarcity of open source data has led to the creation of only a graph for 2011 to 2015. Indonesia shows an upward trend reaching 21 percent in the third quarter of 2012, after which a downward trend sets in reaching 7 percent mid-2015. The Philippine growth of real domestic credit does not show a significant trend and fluctuates between 10 and 15 percent. The decrease of Indonesia is in line with the negative sign found in the literature.

Graphs 22 and 23 show domestic credit over GDP. In both countries there was an upward trend from 1993 to 1997. In 1993 there is a difference of 22 percentage points between Indonesia and the Philippines, but this gap almost disappeared by 1997 when both countries’ domestic credit over GDP was around 60 percent. The increase is in line with the positive sign found in the literature. In the period 2011 to 2014 only Indonesia’s domestic credit over GDP experiences an upward trend, from 2011 to 2013. The levels of both countries’ domestic credit over GDP are still 20 percent points lower in 2014 than they were in 1997.

Graphs 24 and 25 show annual appreciation of the Indonesian rupiah and the Philippine Peso. Both currencies experienced a decreasing trend of appreciation before the respective crises, and a large depreciation after the third quarter of 1997. When the crises erupted, appreciation was 8.5 percent in the Philippines and 1 percent in Indonesia. Surprisingly, the depreciation set in at the same time even though the start of the currency crisis was at different moments (Radelet and Sachs, 1998). The decrease is in line with the negative sign found in the literature. In the period 2011 to 2015 the Rupiah experiences a decreasing trend of appreciation from the third quarter of 2013 to mid-2014, and from 2015 onwards. The appreciation of the Peso experiences almost the same general movements, decreasing from mid-2013 to mid-2014, and from mid-2015 onwards. Nevertheless, the latest data points of the fourth quarter of 2015 show an appreciation of 1.5 percent for Indonesia and 5 percent for the Philippines.

Graphs 26 and 27 show the fiscal balance over GDP. The fiscal balance over GDP experienced a downward trend from 1994 onwards, reaching 1.5 percent in 1997. The fiscal balance over GDP in the Philippines improved until 1996, after which a decline set in reaching minus 3.5 percent in 1997. Whereas the Philippines ran a deficit except for in 1996, Indonesia had a surplus each year before the Asian financial crisis. The decrease both countries’ fiscal balance over GDP is in line with the negative sign found in the literature. It must be noted that Indonesia has a slow decrease over multiple years, while the Philippines a major one in the year they experienced a crisis. In the period 2011 to 2015 the Philippine fiscal balance over GDP shows an upward trend. The Indonesian balance is very volatile, but a downward trend can be seen starting in mid-2013. Now both countries mostly run a deficit. Only the declining trend of Indonesia shows that there is more risk for Indonesia, but the Philippines have been living with a larger deficit for longer.

In summary, all indicators show an alignment with the predicted pre-crisis movement of Davis et. al. (2011) in the period 1993 to 1997, except for real GDP per capita and the growth of real domestic credit. Still, the indicators varied considerably in size and timing of these movements. For Indonesia, even though the increased fiscal deficits of 2014 and 2015 have put an increased pressure on the banking sector, there was no increase in real domestic credit growth. Domestic credit as a percentage of GDP increased, but this can be attributed to lower GDP growth. The currency is still appreciating, and with a decreasing M2 over reserves, there is no substantial risk of depreciation when capital flees the country. All information taken together, the growth of Indonesia’s economy is slowing down, but there are no signs that a banking crisis will occur. For the Philippines, even though the fiscal deficits of 2011 to 2015 might have put pressure on the banking sector, both credit growth

17

and real credit over GDP do not signal a crisis. The current negative trend in currency appreciation does show a weakness, because the increased level of M2 over foreign reserves can make it harder to uphold the currency peg if capital flees the country. Given the projection of an increased U.S. interest rate (Morris, 2015), this is exactly what might happen. The question then is if a large amount of loans are denoted in foreign currency, which was not the case. The increase of foreign currency denoted debt over was around five times lower than the increase in domestic currency denoted debt over GDP (Ayala et. al., 2015). All information taken together, there is no sign that a banking crisis will occur.

Conclusion

In this thesis the movements of twelve macro-economic indicators have been examined for Indonesia and the Philippines in the year period before the Asian financial crisis and the five-year period before 2016. Many economic institutions and newspapers have expressed their concerns for the average economic situation in emerging markets, and this paper can clarify if these concerns are valid for Indonesia and the Philippines specifically.

Certain movements signaling long term risk were visible for a currency crisis: a decrease in export growth in the Philippines and decreasing yield spread for both countries. In the short run, only the decreasing growth of foreign reserves signals a vulnerability in Indonesia, but the increasing export growth and low level of credit over GDP tone down the perceived risk of large depreciations. In the Philippines growth of domestic credit over GDP has increased, but is still at a low level. Little risks are visible for a banking crisis. Increased fiscal deficits and a lower appreciation in both countries, and an increased M2 over foreign reserves in the Philippines specifically, are not reason enough to worry about a banking crisis, due to the low amount of foreign currency denoted debt, and the decreasing levels of credit growth in general. The results suggest that neither Indonesia nor the Philippines are at imminent risk of experiencing either a currency or banking crisis.

Four stages of analysis paved the way for this conclusion. First EWS literature was aggregated and analyzed to find a list of relevant crisis indicators and their hypothesized movements before a crisis. The correct timing and size of such indicatory movements where extracted from the period of 1993 to 1997. Having collected the movement, timing and size of indicators that may indicate a crisis, the period of 2011 to 2015 was analyzed for all indicators to see which indicators showed a risk of a crisis. Finally, the movements were put into perspective of one another to see how large the indicatory signals really were. For example, the increase in credit over GDP in Indonesia was accompanied by a decrease in GDP growth, which diminishes the credibility of the signal that credit over GDP gives.

Three main limitations can press on any conclusions that were drawn in this paper. First, no EWS that was used in this paper has been tested on its accuracy for the five years preceding 2016. Second, the EWS indicators were established by using large datasets spanning dozens of countries. The benefit of this is that results can be generalized over many situations, but accuracy would be higher in this analysis if indicators were established by only using Southeast Asian, or even Indonesian and Philippine specific indicators. Indicator relations in this analysis might deviate from the empirical relationships that literature suggests, either because the literature results cannot be generalized to the used sample countries or to the current economic environment. Third, a number of gaps were present in the data due to the lack of access to paid services. No graphs were available for the yield spread and growth of real domestic credit from 1993 to 1997, the yield spread used is not similar to the one used by Candelon et. al. (2012) and it was impossible to find all quarterly data

18

for real GDP growth, real GDP per capita, domestic credit over GDP, the terms of trade and the fiscal balance to GDP. This can be seen as an opportunity to continue answering the research question of this essay with a more complete dataset.

The results contain some minor implications for IMF research and bring forward some interesting suggestions for future crisis analyses. EWSs provide a rigid econometric model, with only one possible grouping of indicators and coefficients. The advantage of a qualitative analysis like this one is that it can take into account the interplay between indicators more effectively, as well as external factors that were not directly analyzed. It would thus be interesting to see EWSs taking into account the possibility of variable signs of indicators, depending on different contextual inputs. By tweaking a contextual input like GDP growth, a reader could personalize the model. Finally, a first glance at the semi-annual IMF (2015a) Global Financial Stability Report gives the impression that risk existed in many emerging markets, although a closer look does not implicate this at all. There is a general trend of credit growth, but maybe only a few problem cases. Instead of a clear nuance, the same argument is repeated three times in the report. This could be implicated more clearly in future publications.

Reference list

Asian Development Bank. (2016a). Interest Rate Spread - 2yrs vs 10yrs - LCY Bond. Data. Indonesia. Retrieved from: https://asianbondsonline.adb.org/indonesia/data/bondmarket.php?code= Int_rate_spread_2yrvs10yr

Asian Development Bank. (2016b). Interest Rate Spread - 2yrs vs 10yrs - LCY Bond. Data. Philippines. Retrieved from: https://asianbondsonline.adb.org/philippines/data/bondmarket.php?code= Int_rate_spread_2yrvs10yr

Barro, R. J. (2001). Economic growth in East Asia before and after the financial crisis (No. w8330). National Bureau of Economic Research.

Berg, A. and Pattillo, C. (1999). Predicting currency crises: the indicators approach and an alternative, Journal of International Money and Finance, 18(4), 561–586.

Blitz, R. and Moore, E. (2015). What has gone wrong for emerging markets? Financial Times. Retrieved from:

www.ft.com/intl/cms/s/0/821985d0-4695-11e5-af2f-4d6e0e5eda22.html#axzz3tRknkBdv

Bussiere, M., and Fratzscher, M. (2006). Towards a new early warning system of financial crises. Journal of International Money and Finance, 25(6), 953-973.

Candelon, B., Dumitrescu, E. I., and Hurlin, C. (2012). How to evaluate an early-warning system: Toward a unified statistical framework for assessing financial crises forecasting methods. IMF Economic Review, 60(1), 75-113.

Calvo, G. A., and Mendoza, E. G. (1996). Mexico's balance-of-payments crisis: a chronicle of a death foretold. Journal of International Economics, 41(3), 235-264.

Chang, R., and Velasco, A. (1998). The Asian liquidity crisis (No. w6796). National bureau of economic research.

Crisostomo, R., Padilla, S., and Visda, M. (2013). Philippine stock market in perspective. In Proc. 12th National Convention on Statistics, 1-2.

Davis, E. P., and Karim, D. (2008). Comparing early warning systems for banking crises. Journal of Financial stability, 4(2), 89-120.

19

Davis, E. P., Karim, D., and Liadze, I. (2011). Should multivariate early warning systems for banking crises pool across regions? Review of World Economics, 147(4), 693-716.

De Santis, R. A. (2012). The Euro Area Sovereign Debt Crisis: Safe Haven, Credit Rating Agencies and the Spread of the Fever from Greece, Ireland and Portugal. ECB working paper No. 1419 Demirgüḉ-Kunt, A., and Detragiache, E. (1998). The determinants of banking crises in developed and

developing countries. IMF Staff Paper, 45(1).

Demirgüḉ-Kunt, A. and Detragiache, E. (2000). Monitoring banking sector fragility: a multivariate logit approach, World Bank Economic Review, 14(2), 287–307.

Demirgüḉ-Kunt, A. and Detragiache, E. (2002). Does deposit insurance increase banking system stability? An empirical investigation. Journal of Monetary Economy, 49, 1373–1406. Demirgüḉ-Kunt, A. and Detragiache, E. (2005). Cross-country empirical studies of systemic bank

distress: a survey. IMF Working Papers, 96(5).

Edison, H. J. (2003). Do indicators of financial crises work? An evaluation of an early warning system, International Journal of Finance and Economics, 8(1), 11–53.

Eichengreen, B. and Arteta, C. (2000). Banking crises in emerging markets: presumptions and

evidence, Center for International and Development Economics Research, Working Papers 115. Elekdag, S. A., and Wu, Y. (2011). Rapid credit growth: boon or boom-bust? IMF Working Papers,

1-42.

Fuertes, A. M., and Kalotychou, E. (2007). Optimal design of early warning systems for sovereign debt crises. International Journal of Forecasting, 23(1), 85-100.

He, D., and McCauley, R. N. (2013). Transmitting global liquidity to East Asia: policy rates, bond yields, currencies and dollar credit. HKIMR Working Paper, 15

Ilzetzki, E., Reinhart, C., and Rogoff, K. (2008). The country chronologies and background material to exchange rate arrangements in the 21st century: which anchor will hold. Retrieved from: http://personal.lse.ac.uk/ilzetzki/data/ERA-Country_Chronologies_2011.pdf

International Monetary Fund (IMF). (2014). Annual Report on Exchange Arrangements and Exchange Restrictions 2014.

International Monetary Fund (IMF). (2015a). Global Financial Stability Report October 2015. Vulnerabilities, Legacies, and Policy Challenges Risks Rotating to Emerging Markets. Washington. Retrieved from: www.imf.org/External/Pubs/FT/GFSR/2015/02/pdf/text.pdf International Monetary Fund (IMF). (2015b). World Economic Outlook: Adjusting to Lower

Commodity Prices.

International Monetary fund (IMF). (2015c). International Finance statistics

Jacobs, J. P., & Kuper, G. H. (2003). Indicators of financial crises do work! An early-warning system for six Asian countries.

Kamin, S.B., Schindler, J.W., and Samuel, S.L. (2001). The contribution of domestic and external sector factors to emerging market devaluations crises: an early warning systems approach, International Finance Discussions Papers 711, Board of Governors of the Federal Reserve System, Washington, D.C.

Kaminsky, G., Lizondo, S., and Reinhart, C. M. (1998). Leading indicators of currency crises. Staff Papers-International Monetary Fund, 48(1)

Kaminsky, G. L., and Reinhart, C. M. (1999). The twin crises: the causes of banking and balance-of-payments problems. American economic review, 473-500.

Krugman, P. (1979). A model of balance-of-payments crises. Journal of Money, Credit, and

20

Kynge, J. and Wheatley, J. (2015). Is it time to declare an emerging Markets crisis? Financial Times. Retrieved from:

www.ft.com/intl/cms/s/3/9197da2c-5336-11e5-8642-453585f2cfcd.html#axzz3t0X6khuJ

Lane, T., Ghosh, A., Hamann, J., Phillips, S., Schulze-Ghattas, M., and Tsikata, T. (1999).

IMF-Supported Programs in Indonesia, Korea, and Thailand. A Preliminary Assessment. Occasional paper 178.

Marchesi, S. (2003). Adoption of an IMF Programme and Debt Rescheduling. An empirical analysis. Journal of Development Economics, 70(2), 403-423.

Mendoza, E. G., and Terrones, M. E. (2008). An anatomy of credit booms: evidence from macro aggregates and micro data (No. w14049). National Bureau of Economic Research. Morris, B. (2015). US rate rise: Why it matters. BBC Business News. Retrieved from:

http://www.bbc.com/news/business-35105299

Pilbeam, K. (2013). International finance, fourth edition. Palgrave Macmillan. ISBN: 9780230362895 Philippine Statistics authority (PSA). (2016). National Government Cash Budget. Public Finance.

Statistics. Retrieved from: www.nscb.gov.ph/secstat/d_finance.asp

Radelet, S., and Sachs, J. (1998). The onset of the East Asian financial crisis (No. w6680). National bureau of economic research.

Reinhart, C. M. and Rogoff, K. S. (2009). This Time Is Different: Eight Centuries of Financial Folly. Princeton, New Jersey: Princeton University Press.

The Economist. (2015). The never-ending story. The Economist. Retrieved from:

www.economist.com/news/leaders/21678220-first-america-then-europe-now-debt-crisis-has-reached-emerging-markets-never-ending

Van den Berg, J., Candelon, B., and Urbain, J. P. (2008). A cautious note on the use of panel models to predict financial crises. Economics Letters, 101(1), 80-83.

Wilson, D. and Stupnytska, A. (2007). The N-11: More Than an Acronym. Global Economics Paper No: 153. Goldman Sachs.

World Bank. (2015). World Bank Development Indicators

Zhuang, J., and Dowling, J. M. (2002). Causes of the 1997 Asian Financial Crisis: What can an early warning system model tell us? (No. 26). ERD working paper series.

21

Appendix

Source: (IMF, 2015)

Graph 2 Annual GDP deflator

Source: (IMF, 2015)

Graph 3 Annual growth of stock indices

22 Graph 4 Annual growth of foreign reserves

Source: (IMF, 2015c)

Graph 5 Annual growth of foreign reserves

Source: (IMF, 2015c)

Graph 6 Annual growth of exports

Source: (IMF, 2015c) -40 -20 0 20 40 60 80 Q1 1993 Q2 1993 Q3 1993 Q4 1993 Q1 1994 Q2 1994 Q3 1994 Q4 1994 Q1 1995 Q2 1995 Q3 1995 Q4 1995 Q1 1996 Q2 1996 Q3 1996 Q4 1996 Q1 1997 Q2 1997 Q3 1997 %

Annual growth of foreign reserves

Indonesia Philippines -20 -10 0 10 20 30 40 50 60 Q1 2011 Q2 2011 Q3 2011 Q4 2011 Q1 2012 Q2 2012 Q3 2012 Q4 2012 Q1 2013 Q2 2013 Q3 2013 Q4 2013 Q1 2014 Q2 2014 Q3 2014 Q4 2014 Q1 2015 Q2 2015 %

Annual growth of foreign reserves

Indonesia Philippines -20 0 20 40 60 80 %

Annual growth of exports

23 Graph 7 Annual growth of exports

Source: (IMF, 2015c)

Graph 8 Annual growth of domestic credit over GDP

Source: (IMF, 2015c)

Graph 9 Annual growth of domestic credit over GDP

Source: (IMF, 2015c) -10 0 10 20 30 40 Q1 2011 Q2 2011 Q3 2011 Q4 2011 Q1 2012 Q2 2012 Q3 2012 Q4 2012 Q1 2013 Q2 2013 Q3 2013 Q4 2013 Q1 2014 Q2 2014 Q3 2014 Q4 2014 Q1 2015 Q2 2015 %

Annual growth of exports