Effect of International Reserves on Crisis Severity & Recovery

(1)

1

Effect of International Reserves on Crisis Severity & Recovery

University of Groningen Faculty of Economics and Business

MSc Economic Development & Globalization 2019 – 2020

Student: Jan Meijer Master’s Thesis

Student number: S2845202 Supervisor: Prof. Dr. de Haan Email: j.meijer.32@student.rug.nl Co-assessor: Dr. Steiner

Abstract: This study investigates the effect of reserve accumulation on reducing the impact of the 2008 economic crises. The paper examines the consequence of reserve accumulation on the time taken to recover to pre-crisis GDP growth levels as well as on the cumulative output lost as a result of the economic crisis. A cross sectional regression analysis was performed on 43 Asian developing and emerging countries. The results indicate that the accumulation of reserves is far more important for reducing the time taken to recover GDP growth levels as opposed to lowering the post-crisis output lost. This indicates that Asian countries should look towards reserves for reducing the time taken to reach pre-crisis growth levels but must examine other tools for minimizing the loss resulting from economic crises.

(2)

2

I. Introduction

Countries’ emphasis on the accumulation of reserves has become ever increasing. After times of crisis, countries gather additional reserves as a safety measure against future instability. Many authors have confirmed the positive effect of accumulating reserves on crisis severity, but there is a growing literature on the topic of reserves as the relative benefits likely depend on a variety of factors in different countries.

This thesis will examine whether outcomes such as the severity of the 2008 economic crisis and the time taken to recover from it are affected by the amount of reserves taken on by 43 Asian developing and emerging economies. This paper will determine whether there are differences in the benefits of reserves for different Asian countries. It will show that although reserves are an effective tool, other factors also come into play that influence the effectiveness of reserve accumulation.

In times of economic distress such as a financial crisis, countries’ economies suffer. There are many variables that influence the effect of these crises. For instance, Wilms, Swank & de Haan (2018) examine cross country differences in the costs of economic crises. They determined that investment, financial openness, capital account balance, liquidity support and independent monetary policy all affected the severity of economic crises in terms of output loss. Overall, there are many factors which can reduce or increase the severity of economic crises; however, many countries still place substantial importance on the accumulation of reserves. In fact, after the Asian financial crisis in 1997, Asian countries built up international reserves as a buffer to reduce risks of later crises and thereby lowering their dependency on the IMF (Bird & Mandilaras, 2005).

International reserves serve many purposes such as ensuring the central government has a form of security in case their national currency devalues. On the one hand, reserves can be sold to take local currency out of the world market, thereby raising the value of the domestic country’s currency (Klitgard & Higgins, 2004). On the other hand, accumulating reserves has been claimed to give a comparative advantage by maintaining an undervalued currency, thereby increasing the attractiveness of the country’s exports (Bird & Mandilaras, 2005). Furthermore, reserves can also be used to aid local industries during periods of restricted foreign capital access, in order for them to continue to pay foreign suppliers and ensure sustained economic operations.

(3)

3

firms that are permitted to continue operations in times of crisis increases, since foreign currency becomes more accessible due to smaller liquidity constraints.

Although hoarding international reserves provides a form of safety, it has also been criticized by some scholars. If countries borrow to accumulate these reserves, interest rates must be paid and thus additional costs arise (Mills, 1986). Rodrik (2006) calculated the approximated cost of hoarding and found it to be a substantial portion of a country’s income. Although many authors agree that the accumulation of reserves is beneficial, the relative advantages may differ depending on how these reserves are gathered. Furthermore, there are many other tools that could aid countries without the entailed costs of accumulating reserves (Wilms, Swank & de Haan, 2018).

In an attempt to determine whether increased reserves benefited Asian countries during the 2008 economic crisis, this paper determines two measures of crisis performance, namely crisis severity and crisis recovery and conducts a cross sectional analysis on international reserves of the selected Asian countries. These reserves will be examined as a in relation to each country’s GDP, imports and short-term debt. The goal of using different measures of crisis effects and reserves is to ensure robust results. The controls included in the analysis include the interest rate, current account position, imports, exports, and the exchange rate.

This paper aims to answer the following question: Does the accumulation of reserves truly benefit Asian countries with regard to lower crisis severity and improved recovery speed during the 2008 economic crisis? Moreover, does the addition of the aforementioned control variables influence the strength of this relationship? To my knowledge, there are no other papers which directly analyze these questions side by side.

The thesis proceeds as follows: section II performs a literature review, section III outlines methodology, section IV elucidates the main results of the study along with robustness checks, section V discusses the findings, section VI concludes, the bibliography is listed in section VII. and finally, the appendix can be found in section VIII

II. Literature review

Importance of foreign currency reserves

(4)

4

In today’s world supply chains have become globally integrated and countries have become increasingly reliant on intermediate products from around the world. Thus, countries are dramatically more dependent on foreign capital, for both imports and exports (Timmer, Miroudot & de Vries, 2019). Therefore, access to foreign currency has become significantly more important. In times of economic crisis, this dependence increases further when this access becomes more limited. During periods of economic distress, when inflow of foreign capital is often restricted or halted (Forbes & Warnock, 2012), operations may be forced to shut down due to an inability to pay foreign suppliers. In this case, countries must find alternative ways to fund their operations. An increasingly popular method is to accumulate reserve assets, which provides a buffer in times of crisis (Aizenman & Lee, 2006) and lowers exchange rates. This consequently increases the current account balance due to its stimulating effects on exports (Klitgard & Higgins, 2004). Tools used to acquire foreign capital include: Foreign exchange (FX) dealers, central bank swap lines, government borrowing taxation and exchange rate manipulation, as will be explained in the subsequent sections.

Acquiring FX reserves through dealers and banks

During a crisis, FX dealers may have to take on a lot of currency from less developed, more volatile countries, and supply non-central banks with the international currencies they require (Banti & Phylaktis, 2012). This results in dealers having a large amount of assets in risky foreign currency and an increasing amount of liabilities in stronger currencies. The more burdened by weaker currencies they become and the fewer strong currencies they possess, the more cautious the dealers become. These dealers take on the liquidity risk of not have enough strong currencies (e.g. dollars) for international payments. This is called the survival constraint (Minsky, 1957) and is a measure of liquidity. This constraint is not just a phenomenon for foreign exchange traders but also for countries. It essentially refers to the payment system and states that one requires enough liquidity to meet payment obligations. Being further away from one’s survival constraint generally increases safety in times of crisis, as financial institutions possess more liquid capital, which can be supplied to the domestic market in order to continue operations by paying foreign suppliers. Since the international monetary system is asymmetric, hierarchical, and centered around the USD reserve currency, countries must have liquid access to dollar funding to finance international obligations. Countries should at least match their dollar-denominated expenditure and cash inflows in such a way that it permits them to meet payment obligations.

(5)

5 Acquiring FX through sovereign borrowing

When financial institutions’ access to foreign capital is restricted due to dealers’ inability to take on more weak currency, the central bank is forced to act. If local banks no longer can provide foreign currency, domestic industries which are reliant on it will suffer. Here the central bank can act as a lender of last resort providing funding to banks and by proxy to vital industries. To do so, central banks must be able to access foreign currency. Foreign exchange reserves can be acquired through government borrowing; however, additional costs arise when acquiring this form of foreign exchange reserves (Mills, 1986; Rodrik, 2006). These authors find that since countries must pay interest rates on their international borrowing, reserves become costly to hold and are vulnerable to fluctuations in international financial markets. Generally, less creditworthy countries face higher borrowing costs when interest rates increase. Thus, countries which are most sensitive to borrowing costs adjust their reserve holdings the quickest. Quick adjustment through borrowing can be necessary for vulnerable countries but can entail high costs. Therefore, acquiring reserves through government borrowing can be a last resort option. Although many thought that short-term debt was a cause of economic crisis, some authors have found this to be untrue (Diamond & Rajan, 2001). It has been shown that increasing short-term debt is a consequence of crises, and not a cause, demonstrating the necessity of government borrowing in times of crisis. Diamond & Rajan (2001) found that banks and countries that need to finance illiquid investments are forced to increase short-term debt. On top of this they concluded that banning short-term debt would likely increase the severity of a crisis. However, Corneli & Tarantino established that raising debt increases sovereign exposure to liquidity crises, yet the simultaneous gathering of reserves helps mitigate these negative effects. Although borrowing to acquire reserves can be seen as counter intuitive, it has been shown to lower the severity of crises, since the benefit of having additional reserves outweighs the costs of acquiring them.

Acquiring FX by increasing the current account

(6)

6

amount of foreign currency flowing into the country. With a large enough current account surplus, borrowing to acquire reserves becomes less important.

Reserves and crisis recovery

With greater reserves, the severity of an economic crisis can be reduced will be discussed below. However, does the accumulation of reserves also decrease the time required to recover from a crisis? Dominguez, Hashimoto & Ito (2011) found that real GDP growth recovery was stronger for countries which accumulated increased foreign reserves pre-crisis, however, in their analysis the effect of reserve depletion and currency depreciation during the crisis had differential effects depending on the country. Furthermore, Hong & Tornell (2005) state that economies with high liquidity prior to the crisis tended to maintain relatively high growth through the post-crisis period. In essence, countries with inadequate reserves have a slower post-crisis growth rate than those with larger reserves, however, there are clear differentials depending on other macroeconomic factors within each country. Thus, the first hypothesis examined in this paper is H1: Reserve accumulation reduces the time taken to recover to pre-crisis GDP growth levels.

Reserves and crisis severity

The hoarding of international reserves was determined to be a method of self-insurance by Aizenman & Lee (2006). The authors stated that the act of hoarding is a countermeasure against capital flight or sudden stop risks, which they defined as the “precautionary approach”. They compared this to the “mercantilist approach”, which states that reserve accumulation is a residual of policies that impose negative externalities on trading partners. Their evidence was in line with the “precautionary approach”, demonstrating that reserve accumulation is indeed a mechanism used to safeguard against economic volatility. Furthermore, Bussiere, Cheng, Chinn & Lisack (2014) used data on 112 emerging and developing economies to determine whether reserves protected countries during the 2008 crisis. In their paper they found that countries with higher reserves relative to short-term debt were less affected by the crises. In line with these findings, they also found that countries which used greater amounts of their reserves to offset the negative effects of the crisis also rebuilt them quicker afterwards. Dominigues, Hasimoto & Ito (2011) confirm this in their analysis, where they determined that countries which were reserve accumulators before a crisis utilized their reserves, which allowed them to depreciate their currency. It was also found that many of the examined emerging market countries reverted to pre-crisis reserve accumulation trends post-crisis. This demonstrates not only the benefits of reserves in times of crisis but also the relative importance that countries place on them. This leads us to our second hypothesis H2: Reserve accumulation reduces the GDP output lost after an economic crisis.

(7)

7

effectiveness of reserve accumulation for reducing crisis severity likely also depends on each country’s exchange rate regime (Gordon, 2005). For instance, with a flexible exchange rate, intervention is possible. Thus, buying reserves to push local currency into the foreign market to manipulate the exchange rate would be effective for a flexible exchange rate regime but not for a fixed one.

Additional benefits of holding reserves

Reserves have been shown to be a beneficial strategy in times of economic crisis; however, the accumulation thereof seems to also benefit countries’ general growth and current accounts. Benigno & Fornaro (2012) examine which countries tend to accumulate more reserves, and what the consequences are. They found that faster growing countries tend to accumulate larger amounts of international reserves. Hoarding reserves consequently was found to result in a real exchange rate depreciation, making exports more attractive, thereby moving production towards the tradable sector. In their paper the authors state that policy intervention in the form of reserve accumulation is beneficial and a seemingly fast rate of reserve accumulation also leads to higher growth and current account surpluses. Thus, FX reserves appear not only be beneficial in times of crisis, but aid countries in making their exports more attractive through lowering the exchange rate, and thus pushing people to operate in the tradable sector. However, this only works because not all countries accumulate vast amounts of reserves. In essence, it is a “Beggar thy neighbour” policy, whereby a country attempts to solve their economic issues by means that tend to worsen the economic issues of other countries. When operating internationally, firms are forced to become more productive and competitive due to international competition (Caves, 1974). Generally, reserves appear to benefit countries whether in times of crisis or not, however, their accumulation does entail additional costs and is not infinitely sustainable if everyone relied on reserves so heavily.

Costs of holding reserves

(8)

8

a dramatic increase, likely prompted by the heightened demand for reserves post Asian financial crisis. This increased demand was confirmed by the UN (2001) who found there was indeed a 60% increase reserves in developing countries since the Asian crisis. Although reserves act as a buffer in times of crisis, the excessive accumulation thereof needs to be considered, especially due to the ever-increasing costs of hoarding.

Investments into reserve accumulation has further negative connotations. An observation by Rodrik & Velasco (2000) was that short-term debt exposure has continued to climb in many countries over time. Countries kept investing precious resources into increasing the reserve asset base. As a result, many developing countries had higher short-term debt – reserves ratio in 2004 compared to 1990. In fact, the ratio grew from 6.5% to 8.4%, on average. This increased reliance on debt can have detrimental effects on countries, as was shown in the 2010 European sovereign debt crisis. Although short-term debt has been shown by some to be a consequence and not a cause of crises, it increases the severity of a crisis (Diamond & Rajan, 2001), although this can be offset through reserve accumulation (Bussiere, Cheng, Chinn & Lisack, 2014) . When countries become too reliant on debt due to expenditures exceeding savings, they become increasingly vulnerable to economic instability.

A further cost was found by Fukuda & Kon (2012), who show evidence that increased reserves diminish consumption. They state that this occurs because increased foreign exchange reserves cause permanent declines in consumption, as the country’s currency devalues, and imports become more expensive. As a result of the country’s exports becoming cheaper for foreign consumers. This increases foreign demand for your products and domestic labour starts to move towards the tradable sector from the non-tradable sector, therefore, firms become increasingly reliant on demand from abroad. In doing so countries can increase economic growth and investment, at the cost of domestic consumption. Thus, in the shortterm this accumulation can help competitiveness and the current account. However, in a Keynesian demand side model, this outcome will not be relevant in the long-term. In the long-run, generally the current account must be balanced such that the real exchange rates are adjusted to equilibrium values. However, a notable exception is the U.S., who have sustained a huge current account deficit due to their global reserve currency status. In other words, Fukuda & Kon (2012) found that in the short-term, increased reserves lower liquidity risk, and in the long-term they decrease consumption and expand the share of the tradable sector. Therefore, domestic consumption would fall, and exports would increase as a consequence of this insurance mechanism. Benigno & Fornaro (2012) and Abdullateef & Waheed (2010) reinforce the theory of Fukuda & Kon (2012), as they found that increased reserves decrease the exchange rates, resulting in higher demand from foreign consumers, yet imports become relatively more expensive for the domestic population. In this sense the domestic consumers suffer at the expense of raising exports.

(9)

9

result in shifting the income distribution against labour and reducing employment and output. They then established that it cannot be expected that devaluations result in international trade stabilization methods, seeing as in the short-run, the balance of payments deficit is structural, i.e. imports and exports are not very price sensitive for a certain domestic output level. Finally, Oskooee & Miteza (2006) used several models to determine the relationship between exchange rates and domestic output. In their paper they found that in the long-run, devaluations are contractionary for all non-OECD and OECD countries, regardless of the model specifications they use.

The accumulation of reserves clearly has its benefits, however, like most macroeconomic tools, it can result in negative consequences and can entail more costs than previously imagined. Countries must determine whether the benefits of accumulation outweigh costs.

Global effect of accumulating reserves

(10)

10

III. Methodology:

Data collection

Data was collected from both the IMF and The World Bank. The sample selected was collected from 42 developing and emerging Asian countries. A list of these countries can be seen in the Table 32 in the appendix. This paper selected Asian countries as many of them are still marked as developing, and the role of reserves likely plays a heightened role in these more unstable economies as opposed to developed countries. For instance, the European monetary union (EMU), provides additional safety nets and support for its member states. For instance, the EU supplied Greece with a $80 billion bailout package during the 2010 Euro crisis (Papadimas & Strupczewski, 2010). Asian countries, however, must rely more heavily on their own economy to mitigate the effects of economic crises.

This paper will conduct a cross sectional analysis to determine the effect on the output loss and years taken to recover GDP growth levels post-crisis. It will test the effects of reserve accumulation, measured by: Reserves/GDP, Reserves/Imports, and Reserves/Short-term debt and uses the control variables: Interest rate, Exchange rate, CA balance/GDP, Imports/GDP and Exports/GDP.

Dependent variables

Crisis recovery: Measured as the years taken to recover to average pre-crisis GDP growth levels. This variable was determined by first calculating an 8-year average pre-crisis GDP growth level. From this it could be seen how many years it took countries to re-reach these levels (1).

When: 𝑅𝑒𝑎𝑙 𝐺𝐷𝑃 𝑔𝑟𝑜𝑤𝑡ℎ(𝑡) = ∑ 𝐺𝐷𝑃 𝑔𝑟𝑜𝑤𝑡ℎ

2000 𝑡=2008

8 (1)

t equals to years to recover to pre-crisis average GDP growth levels

Where Real GDP growth refers to real annual GDP growth rate, t equals the years taken to recover and and GDP growth refers the 8-year pre-crisis growth levels used to calculate the pre-crisis average GDP growth.

(11)

11 Multiplier:

Multiplier (average) to calculate expected growth = λ = ∑ 𝐺𝐷𝑃 𝑔𝑟𝑜𝑤𝑡ℎ

2000 𝑡=2008

8 (2)

Expected growth year 𝑡 = 𝜑 = 𝜆 ∗ 𝐺𝐷𝑃 (𝑡 − 1) (3)

Output loss = 𝜑 (𝑡) – 𝑅𝑒𝑎𝑙 𝐺𝐷𝑃(𝑡) (4)

Expected growth in year t is defined as φ, which is equal to λ, the pre-crisis 8-year average GDP growth, multiplied by the GDP(t-1) which is defined as the GDP level on year prior to the expected growth in year t. Lastly to determine the output loss: Real GDP for year (t) is subtracted from φ for year (t).

Independent variables:

All independent variables were taken from a single year: 2007. This was done since this paper seeks to examine the effect of the amount of reserves accumulated pre-crisis against crisis severity and recovery. The independent variables were all scaled to normalize the data. The following variables were chosen to determine the scaled effects of reserves on the dependent variables.

Reserves/Imports: For this variable, the Reserve level of 2007 was divided by imports for each country. This variable shows how changes in the ratio between reserves and imports affects the variables of interest. If the value is larger than 1: there are more reserves than imports, if the value is less than 1: there are more imports as opposed to reserves.

Reserves/GDP: For this variable, the reserve level of 2007 was divided by real GDP for each country. This was done to take away the effect of country size. Countries with greater GDP levels could be influenced differently than smaller countries with lower GDP. Dividing reserves by GDP shows how many reserves a country has per unit of GDP. If the value is larger than 1: there are more reserves than GDP, if the value is less than 1: GDP is larger than the reserve stock.

Reserves/Short-term debt: For this variable, the reserve level of 2007 was divided by short-term debt for each country. This shows how many reserves a country has in comparison to their short-term debt. This variable was transformed into its natural log variant to reduce skewness and normalize the value. If the value is larger than 1: there are more reserves than short-term debt, if the value is less than 1: there is more short-term debt as opposed to reserves.

Control variables:

(12)

12

Interest rates: Changing interest rates alter people’s propensity to save or consume (Elmendorf, 1996). It is also a precursor to changing exchange rates as the demand of your currency changes. It was expected that lower interest rates would minimize negative crisis outcomes as it encourages consumption, which in turn brings growth. This variable was measured in percentage terms.

Exchange rate: Changing exchange rates determines how much a country exports and imports as your local currency becomes relatively more expensive/cheap (Sekkat & Varoudakis, 2000). With a low exchange rate exports are more attractive, which is expected to improve the crisis impact. This variable was measured by Local currency/ USD.

Current account balance/ GDP: The current account (CA) balance demonstrates whether a county has a current account surplus or deficit. If a country has a CA surplus, their net financial inflows are greater than their outflows. This takes trade, dividends, interest, and other transfers such as aid into account. It was expected that countries with a larger CA surplus would have to rely less on reserves as the inflow of foreign capital can aid in supporting the economy in times of crisis. This variable was also scaled to GDP to normalize the results, it was measured in USD/GDP

Imports/GDP: This variable was included to determine whether the imports would affect the dependent variables. Theory states that increased imports may increase the impact of the crisis as more capital flows out of the domestic economy. This variable was also scaled to GDP to normalize the results and was calculated as imports in USD/GDP

Export/GDP: This variable was included to determine whether the extent to which a country exports would affect the dependent variables. It was expected that increased exports may decrease the impact of the crisis as more capital flows into the domestic economy. This variable was also scaled to GDP to normalize the results and was calculated as imports in USD/GDP.

Regression models:

Years taken to recover:

Years to recover(i) = Reserves/Imports(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Years to recover(i) = Reserves/GDP(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

(13)

13

Output loss 2 years:

Output loss 2 years(i) = Reserves/Imports(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Output loss 2 years (i) = Reserves/GDP(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Output loss 2 years (i) = Reserves/Short-term debt(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Output loss 2 years:

Output loss 3 years(i) = Reserves/Imports(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Output loss 3 years (i) = Reserves/GDP(i) + Interest rate(i,t) + Exchange rate (I,t) + CA/GDP(I,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

Output loss 3 years (i) = Reserves/Short-term debt(i) + Interest rate(i,t) + Exchange rate (i,t) + CA/GDP(i,t) + Imports/GDP(i,t) + Exports/GDP(i,t) + ε

IV. Results

The results section will proceed as follows: First the descriptive statistics will be discussed. After that graphs regarding the independent variables will be examined. Following this are main regression results for each dependent variable and subsequent robustness checks.

(14)

14

Table 1: Descriptive statistics stats years Output loss

3 Output loss 2 Reserves/GD P Reserves / Imports Reserves / Short-term debt Interest rate Exchang e rate CA/GDP Imports/GD P Exports/GD P N 42 43 43 41 39 25 277 339 355 313 320 mean 3.047619 8.33E+09 -1.65E+10 0.271307 0.59957 1.529769

3.76435

8 1340.859 0.021475 0.482437 0.467804 min 1 -2.69E+11 -3.90E+11 0.022911 0.027029 0.062633 -20.1293 0.268828 -0.43771 0 0.013241 max 8 1.09E+12 6.53E+11 0.918309 1.629439 3.65935

48.0000 3 20828 0.482099 2.083387 2.290001 sd 3.011785 1.79E+11 1.29E+11 0.204949 0.364675 0.941113 9.53213 9 3555.731 0.144641 0.304558 0.346969 skewness 1.027274 5.120439 2.553406 1.595545 1.203709 0.75477 1.42503 7 3.242941 0.897294 2.349222 2.671695 kurtosis 2.129982 32.30261 19.52269 5.132901 3.846193 3.161322 8.10962 2 13.58401 4.774053 11.21261 13.50634

Graphs of dependent variables:

The graphs below have been split into 2 groups. One group which recovered to average pre-crisis GDP growth levels quickly (in one year or less) and the other group of countries which took longer. Outliers were removed from these graphs. Furthermore, any missing values were removed from all graphs, so as to keep the countries compared the same. Additional graphs can be seen in the appendix (graph 7 and 8) which show the average time taken to recover pre-crisis growth levels, as well as the 3-year cumulative output loss.

Reserves/GDP comparison: Graph 1 and 2 below

Quick recovery time

- Mean Reserves/GDP: 0.306503

- Mean Reserves/GDP before crisis 2007: 0.309756 - Mean Reserves/GDP during crisis 2008: 0.307549 - Mean Reserves/GDP after crisis 2009: 0.36183

Slow recovery time

- Mean Reserves/GDP: 0.171264

- Mean Reserves/GDP before crisis2007: 0.210867 - Mean Reserves/GDP during crisis 2008: 0.147819 - Mean Reserves/GDP after the crisis 2009: 0.184901 Reserves/GDP analysis

On average, countries which recovered fast had a Reserves/GDP ratio of 0.306. When examining the average, pre (2007), during (2008) and post (2009) crisis values, countries which recovered quickly kept a much more steady and higher level of reserves, relative to their GDP. Of particular note is that the 2008 crisis value did not shift much, however, they likely do not show the true magnitude of country’s lost reserves, seeing as during this time the GDP levels of countries also dropped.

(15)

15

their GDP. During the 2008 economic crisis, these countries dropped their reserve ratio dramatically, likely to aid in reducing the effects of the crisis.

After the crisis, both country groups attempted to rebuild reserve levels. However, the countries which recovered quickly, built up their reserves to levels even higher than their pre-crisis level. Although the group which recovered slower attempted to build up reserves, it proved more difficult and did not reach the same pre-crisis ratio.

(16)

16 0 0.2 0.4 0.6 0.8 1 1.2 1.4 2005 2006 2007 2008 2009 2010 2011 2012

Graph 1: Reserves/GDP - Fast recovery

China India Indonesia Philippines

Turkey Thailand Myanmar Yemen

Nepal Sri Lanka Laos Lebanon

Mongolia 0 0.1 0.2 0.3 0.4 0.5 0.6 1 2 3 4 5 6 7 8

Graph 2: Reserves/GDP - slow recovery

Pakistan Bangladesh Vietnam Kazakhstan

(17)

17 Reserves/imports comparison: Graph 3 and 4 below

Quick recovery time:

- Mean Reserves/Imports: 0.649355

- Mean Reserves/Imports before crisis 2007: 0.69557 - Mean Reserves/Imports during crisis 2008: 0.636351 - Mean Reserves/Imports after crisis 2009: 0.878963

Slow recovery time

- Mean Reserves/Imports: 0.409564

- Mean Reserves/Imports before crisis 2007: 0.454104 - Mean Reserves/Imports during crisis 2008: 0.375633 - Mean Reserves/Imports after crisis 2009: 0.517289

Reserves/Imports analysis:

Similar trends can be observed within both country groups when examining Reserves/ Imports. However, countries which recovered faster, generally had much higher average of reserves compared to imports, during the crisis the reserves ratio dropped for both groups as they were used for to aid the economy.

Both country groups had increased their reserve levels relative to their imports post-crisis to levels dramatically higher than pre-crisis levels. However, the fast post-crisis increases in Reserves/Imports ratio may also be explained by drops in imports, due to the crisis. However, when examining the next series of graphs, which reassure these results, it is likely that this was caused by decreased reserve levels.

(18)

18 0 0.5 1 1.5 2 2.5 3 2005 2006 2007 2008 2009 2010 2011 2012

Graph 3: Reserves/Imports - fast recovery

Mongolia 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 2005 2006 2007 2008 2009 2010 2011 2012

Graph 4: Reserves/Imports - slow recovery

(19)

19 Reserves/Short-term debt comparison: Graph 5 and 6 below

Quick recovery

- Mean Reserves/Short-term debt: 9.829794

- Mean Reserves/Short-term debt before crisis 2007: 9.035575 - Mean Reserves/Short-term debt during crisis 2008: 9.559515 - Mean Reserves/Short-term debt after crisis 2009: 11.04964

Slow recovery

- Mean Reserves/Short-term debt: 5.207202

- Mean Reserves/Short-term debt before crisis 2007: 3.522918 - Mean Reserves/Short-term debt during crisis 2008: 3.490479 - Mean Reserves/Short-term debt after crisis 2009: 4.625812 Reserves/Short-term debt analysis:

When examining both country groups reserves in relation to their short-term debt, it is interesting to note that fast growing countries had dramatically higher levels of reserves relative to short-term debt under all scenarios. During the crisis this groups reserves increased relative to short-term debt and raised further after the crisis. This could demonstrate that these countries made an increasing effort to acquire more reserves during the crisis but did this without increasing their short-term debt as much.

The ratio between reserves and short-term debt is much lower for the slowly recovering country group. This indicated that their reserves in comparison to the first group, are lower relative to their short-term debt.

(20)

20 0 10 20 30 40 50 60 70 2005 2006 2007 2008 2009 2010 2011 2012

Graph 5: Reserves/ Short-term debt - fast recovery

Mongolia 0 2 4 6 8 10 12 2005 2006 2007 2008 2009 2010 2011 2012

Graph 6: Reserves/ Short-term debt - slow recovery

(21)

21

Overall graph analysis:

The graphs in general illustrate that countries which recovered quicker, tended to have higher reserve ratios. Of further notice is that, consistent with Aizeman & Lee (2006), countries tend to increase their reserves relative to GDP, imports, and short-term debt after the economic crisis. This demonstrates that out of fear countries tend to accumulate greater amounts of reserves. The graphs indicate that countries with higher levels of reserves tend to regain pre-crisis growth levels quicker.

Next the regression results will be analyzed for all dependent and independent variables. The first set of regressions will examine the years taken to recover followed by robustness checks the next set will look into the output loss, followed by robustness checks for the output loss variable.

Regression results: Years taken to recover - sequentially added control variables

First the regression results for the years taken to recover to pre-crisis average GDP growth levels will be examined, with control variables added sequentially. Each model adds one additional control.

Table 2: Reserves/GDP

Years to recover

growth model1 model2 model3 model4 model5 model6

Reserves/GDP -4.115 -5.147 -5.209 -5.601 -9.469** -8.772** [2.646] [3.228] [3.228] [3.282] [3.615] [3.652] Interest rate 0 0.019 0.029 0.047 0.001 [0.109] [0.111] [0.112] [0.112] [0.119] Exchange rate 0 0 0 0 [0.000] [0.000] [0.000] [0.000] CA/GDP -2.956 1.401 5.788 [3.557] [4.090] [5.676] Imports/GDP 6.130* 9.833** [3.150] [4.581] Exports/GDP -4.29 [3.869] Constant 3.915*** 4.264*** 4.403*** 4.585*** 2.427 2.493 [0.813] [1.032] [1.041] [1.070] [1.657] [1.650] R-squared 0.065 0.087 0.121 0.145 0.283 0.321 Observations 37 30 30 30 29 29

(22)

22

negative too. Implying that greater levels of reserves relative to GDP, the less time taken to recover.

Of further notice is that imports relative to GDP dramatically increase the years taken to recover. This is significant at the 10% level in model 5 and at the 5% level in model 6. A 1 increase in imports relative to GDP, results in a 9.8-year increase in time taken to recover. It is unreasonable to assume that imports will be anywhere near a countries GDP level, hence an increase in the ratio by 1 is dramatic, but it tells us that increased imports during an economic crisis likely increases the time taken to recover.

When all variables are added, a 1 increase in the reserves relative to GDP ratio in 2007, reduces the time taken to recover by around 8.7 years. A 1 increase in Reserves/GDP, however, is a huge amount of additional reserves. This 1 increase in the reserve’s ratio is equivalent to a -287% decrease in time taken to recover.

Table 3: Reserves/ Imports

Years to recover

Reserves/Imports -3.763** -4.056** -4.161** -4.095** -4.044** -3.771** [1.551] [1.614] [1.603] [1.652] [1.740] [1.736] Interest rate -0.037 -0.016 -0.014 -0.012 -0.058 [0.107] [0.107] [0.110] [0.113] [0.118] Exchange rate 0 0 0 0 [0.000] [0.000] [0.000] [0.000] CA/GDP -0.924 -0.66 4.41 [3.387] [4.127] [5.817] Imports 0.338 4.93 [2.884] [4.714] Exports -4.795 [3.919] Constant 5.032*** 5.495*** 5.699*** 5.693*** 5.483** 5.350** [0.963] [1.142] [1.146] [1.168] [2.156] [2.136] R-squared 0.151 0.2 0.243 0.245 0.246 0.294 Observations 35 29 29 29 29 29

(23)

23

Table 4: Reserves/Short-term debt

Years to recover

Reserves/Short-term debt -1.458** -2.016** -2.060** -2.028** -2.088** -2.129** [0.582] [0.781] [0.773] [0.928] [0.947] [0.976] Interest rate 0.004 0.027 0.025 0.033 0.041 [0.120] [0.120] [0.126] [0.128] [0.133] Exchange rate 0 0 0 0 [0.000] [0.000] [0.000] [0.000] CA/GDP -0.402 2.956 -1.66 [5.986] [7.738] [12.294] Imports 2.334 -0.662 [3.325] [6.983] Exports 3.116 [6.335] Constant 5.111*** 6.165*** 6.439*** 6.388*** 5.338** 5.453** [1.039] [1.277] [1.283] [1.525] [2.154] [2.223] R-squared 0.215 0.294 0.348 0.348 0.369 0.38 Observations 25 21 21 21 21 21

Finally, the table 4 above examines the effect of Reserves/Short-term debt regarding the years taken to recover to pre-crisis average GDP growth levels. Here too, all models demonstrate a significant independent variable at 5%. The coefficient is even lower than the previous output, however, it still demonstrates that a 1% increase in the Reserves/Short-term debt ratio decreases the time taken to recover by 0.0212 years, in model 6. Since this variable was transformed into a log, it must be interpreted differently. Thus a 100% increase in reserves relative to short-term debt would result in 2.129 less years taken to recover.

The results of these outputs will be verified in the next series of regressions and the appendix (table 20 – 22), where the regressions were run with adding control variables one at a time. The results from these regressions very much support the prior results, the only major difference is that the coefficient for the ratio of Reserve/ Short-term debt is smaller for all models. In these models, a 1% increase in Reserves to short-term debt ratio, decreases the time taken to recover by approximately 0.015 years (100% = 1.5 years decrease).

(24)

24 Robustness check: years taken to recover

These robustness checks all use the same control variables. The 2 robustness checks performed in this section remove certain outliers from the regression but keep all control variables the same as the main regression. The control variables were not included if they did not show significant results. Finally, a last robustness check can be seen in the appendix (table 20-22), where the analysis is the same as in the main regressions, with the same countries but with individually added control variables. These results show the similar results the main regressions.

Regression results: Robustness, with countries removed (+ output loss)

In tables 5-7 below are regression performed with: Indonesia, Israel, Jordan, Laos, Lebanon, Myanmar, Nepal, Sri Lanka and Uzbekistan removed. These countries were removed since they demonstrated positive output gains post-crisis. All else remains the same as the previous regressions

Table 5: Reserves/GDP- removed countries with positive output

Years to recover

Reserves/GDP -7.063** -10.536** -10.413** -10.410** -11.485** -10.546*

[3.285] [4.301] [4.434] [4.533] [4.874] [5.271]

Table 6: Reserves/Imports - removed countries with positive output

Years to recover

growth model1 model2 model3 model4 model5 model6 Reserves/Imports -1.696 -1.958 -2.082 -2.064 -2.792 -2.421

[1.660] [1.761] [1.793] [1.859] [2.043] [2.084]

Table 7: Reserves/Short-term debt – removed countries with positive output

Years to recover

growth model1 model2 model3 model4 model5 model6 Reserves/ Short-term

debt -1.435* -2.089** -2.034* -2.300* -2.304 -2.352

[0.740] [0.951] [0.993] [1.236] [1.295] [1.349]

(25)

25

independent variables and that the results were influenced by the inclusion of these removed countries.

Regression results: Robustness, countries with pegged and fixed exchange rate removed For these regressions (table 8-10), Oman, Qatar, Saudi Arabia, Lebanon, Jordan and Bahrain were removed since they had a fixed or pegged exchange rate, in addition to the countries which had a positive output gain were also removed. Countries with a less flexible exchange rate cannot manipulate exchange rates through buying and selling reserves to stimulate their imports or exports. These countries were removed from the next regressions to see whether the results would differ. This was done to test further robustness of these results. This test demonstrates similar results as the previous robustness check. Reserves/Imports (table 9) is insignificant, and the other 2 variables are remained significant and reduce the years taken to recover (tables 8 & 10).

Table 8: Reserves/GDP – Removed countries with positive output + fixed/ pegged exchange

Years to recover

Reserves/GDP -8.874** -10.870** -10.727** -10.728** -10.895** -9.804*

[4.042] [4.212] [4.350] [4.449] [4.837] [5.238]

Table 9: Reserves/Imports – Removed countries with positive output + fixed/ pegged exchange

Years to recover

growth model1 model2 model3 model4 model5 model6 Reserves/Imports -1.487 -1.766 -1.9 -1.839 -2.702 -2.279

[1.652] [1.763] [1.802] [1.877] [2.032] [2.077]

Table 10: Reserves/Short-term debt – Removed countries with positive output + fixed/ pegged exchange

Years to recover

growth model1 model2 model3 model4 model5 model6 Reserves/ Short-term

debt -1.435* -2.089** -2.034* -2.300* -2.304 -2.352

[0.740] [0.951] [0.993] [1.236] [1.295] [1.349]

(26)

26 Regression results: Output loss 3 years - sequentially added control variables

In the next series of regressions (table 11-13), the cumulative output loss over 3 years are examined. These results are backed up in the appendix (tables 23-31) , where the same regression were run with the cumulative output loss 2 years after the crisis, as well as both sets of regressions being run again with individually added control variables. The tables below demonstrate the regressions run for all 3 independent variables: Reserves/ GDP (table 11), Reserves/Imports (table 12) and Reserves/ Short-term debt (table 13). All variables in these regressions are insignificant, thus we cannot make any inferences about reserves affecting the amount of GDP lost during the 2008 economic crisis. Possible reasons for these insignificant results may be that it takes more than 3 years for the benefits of reserve accumulation to be realized in the output loss. Reserves increase growth, but it this growth takes time before it is transferred into real increased GDP levels. Further reasons can include that due to the developing nature of these countries there tends to be more corruption (Olken & Pande, 2012). This can lead to issues, such as banks using reserves provided by central banks to bolster their own balance sheets. An additional issue could result from inefficient allocation of reserves to unsuitable economic sectors. Finally, reducing the value of the currency through reserve accumulation can lead to increased lending to finance imports, which may increase inflation, and thus harm the economy. This, however, must be examined in future research.

Table 11: Reserves/GDP

Output loss3 model1 model2 model3 model4 model5 model6 Reserves/GDP 1.50E+10 2.48E+10 2.57E+10 2.43E+10 4.74E+10 5.49E+10

[5.81e+10] [6.39e+10] [6.44e+10] [6.64e+10] [7.94e+10] [8.19e+10] Interest rate -4.38E+08 -7.16E+08 -6.79E+08 -1.27E+09 -1.77E+09 [2.16e+09] [2.22e+09] [2.27e+09] [2.47e+09] [2.67e+09]

Exchange rate 2.29E+06 2.22E+06 2.26E+06 2.52E+06

[3.14e+06] [3.24e+06] [3.34e+06] [3.43e+06]

CA/GDP -1.05E+10 -4.54E+10 1.92E+09

[7.19e+10] [8.98e+10] [1.27e+11]

Imports -4.57E+10 -5.75E+09

[6.92e+10] [1.03e+11]

Exports -4.63E+10

[8.67e+10] Constant -2.14E+10 -1.20E+10 -1.40E+10 -1.34E+10 8.10E+09 8.80E+09

[1.81e+10] [2.04e+10] [2.08e+10] [2.16e+10] [3.64e+10] [3.70e+10]

R-squared 0.002 0.006 0.026 0.027 0.046 0.058

(27)

27

Table 12: Reserves/Imports

Output loss3 model1 model2 model3 model4 model5 model6 Reserves/Imports 3.23E+10 3.13E+10 3.27E+10 3.45E+10 3.28E+10 3.55E+10

[3.39e+10] [3.37e+10] [3.41e+10] [3.51e+10] [3.69e+10] [3.78e+10] Interest rate -7.08E+08 -9.88E+08 -9.23E+08 -9.78E+08 -1.43E+09 [2.23e+09] [2.28e+09] [2.33e+09] [2.40e+09] [2.57e+09]

[3.15e+06] [3.24e+06] [3.30e+06] [3.39e+06]

CA/GDP -2.56E+10 -3.46E+10 1.46E+10

[7.19e+10] [8.76e+10] [1.27e+11]

Imports -1.16E+10 3.30E+10

[6.12e+10] [1.03e+11]

Exports -4.66E+10

[8.53e+10] Constant -3.35E+10 -2.10E+10 -2.38E+10 -2.40E+10 -1.68E+10 -1.81E+10 [2.18e+10] [2.39e+10] [2.43e+10] [2.48e+10] [4.57e+10] [4.65e+10]

R-squared 0.026 0.035 0.057 0.062 0.063 0.076

Observations 36 29 29 29 29 29

Table 13: Reserves/Short-term debt

outputloss3 model1 model2 model3 model4 model5 model6

Reserves/Short-term debt 1.26E+10 3.50E+09 3.91E+09 7.89E+08 2.05E+09 2.31E+09 [1.21e+10] [9.21e+09] [9.27e+09] [1.10e+10] [1.08e+10] [1.12e+10] Interest rate -1.46E+09 -1.67E+09 -1.53E+09 -1.69E+09 -1.74E+09

[1.42e+09] [1.44e+09] [1.50e+09] [1.47e+09] [1.53e+09]

[1.75e+06] [1.79e+06] [1.76e+06] [1.83e+06]

CA/GDP 3.92E+10 -3.16E+10 -2.49E+09

[7.11e+10] [8.86e+10] [1.42e+11]

Imports -4.92E+10 -3.03E+10

[3.81e+10] [8.04e+10]

Exports -1.96E+10

[7.30e+10] Constant -2.52E+10 9.49E+09 6.96E+09 1.19E+10 3.41E+10 3.33E+10

[2.17e+10] [1.51e+10] [1.54e+10] [1.81e+10] [2.47e+10] [2.56e+10]

R-squared 0.045 0.056 0.1 0.117 0.205 0.209

(28)

28 Robustness checks: Output loss

These robustness checks all use the same control variables. The 2 robustness checks performed in this section remove certain outliers from the regression but keep all control variables the same as the main regression. They were not displayed as they did not show significant results. Finally, additiona robustness checks can be seen in the appendix in tables 23-31, where the first analysis is the same as in the main regressions, with the same countries but with individually added control variables. These results show the same as in the main regression. The second check runs the regression against the cumulative output loss for 2 years with both sequentially and individually added control variables. These results are also consistent with the main regression.

Regression results: Robustness, with countries removed (+ output loss)

In tables 14-16 the regressions are were performed with Indonesia, Israel, Jordan, Laos, Lebanon, Myanmar, Nepal, Sri Lanka and Uzbekistan removed. This was done because some countries demonstrated a large positive output gains after the crisis. To determine whether this influenced the results, these countries were removed for the robustness checks in this subsection.

All else remains the same as the previous regressions. The control variables were not significant in all cases. Only the regression for Reserves/Imports (table 15) became statistically significant. The coefficients for all the below regressions, however, remain positive. Which may indicate that higher reserves increase the output lost during a crisis, however, no conclusions can be made due to the insignificance.

Table 14: Reserves/GDP – removed countries with positive output

Output loss3 model1 model2 model3 model4 model5 model6

Reserves/GDP

-1.26E+10

4.00E+10 5.20E+10 5.21E+10 2.44E+11 2.88E+11

[2.28e+11

]

[3.36e+11] [3.46e+11] [3.54e+11] [3.64e+11] [3.95e+11]

Table 15: Reserves/Imports – removed countries with positive output

Reserves/Imports 2.49e+11** 2.79e+11** 2.79e+11** 2.87e+11** 2.40e+11* 2.51e+11*

[9.73e+10] [1.12e+11] [1.16e+11] [1.19e+11] [1.31e+11] [1.36e+11]

(29)

29

enters the world economy. This raises the prices of imports which could increases the output loss if the demand for imports is relatively inelastic.

Table 16: Reserves/Short-term debt – removed countries with positive output

Reserves/Short-term debt 2.12E+10 8.86E+09 9.74E+09 1.04E+10 1.06E+10 1.08E+10 [1.52e+10] [9.83e+09] [1.02e+10] [1.27e+10] [1.30e+10]

]

[1.37e+10 ]

Regression results: Robustness, countries with pegged and fixed exchange rate removed For these regressions (table 17 – 19), Oman, Qatar, Saudi Arabia, Lebanon, Jordan and Bahrain were removed due to having fixed or pegged exchange rates, in addition to the countries which had a positive output gain. This was done to test further robustness of the results. The results remain consistent with the previous robustness check. All reserve measures remain insignificant, except the Reserves/Import’s ratio (table18). This variable is significant at the 5% level in model 1-4. This, however, is the only variable that is significant and appears to be increasing the output loss. This is consistent with the previous robustness check. Since the majority of the variables regressed against the output loss are not significant, and those which are, appear to increase the output loss. Thus hypothesis H2 cannot be accepted. Reserves do not appear decrease the output lost during an economic crisis.

Table 17: Reserves/GDP – removed countries with positive output + fixed/ pegged exchange

Output loss 3 model1 model2 model3 model4 model5 model6 Reserves/GDP 5.19E+10 2.61E+10 4.15E+10 4.15E+10 2.52E+11 2.91E+11

[2.96e+11] [3.60e+11] [3.71e+11] [3.82e+11] [3.84e+11] [4.20e+11]

Table 18: Reserves/Imports – removed countries with positive output + fixed/ pegged exchange

Output loss 3 model1 model2 model3 model4 model5 model6 Reserves/Imports 2.50e+11*

*

2.78e+11** 2.77e+11* *

2.85e+11** 2.38E+11 2.50E+11

[1.01e+11 ] [1.18e+11] [1.22e+11 ] [1.27e+11] [1.39e+1 1] [1.46e+1 1]

Table 19: Reserves/Short-term debt – removed countries with positive output + fixed/ pegged exchange

Output loss 3 model1 model2 model3 model4 model5 model6 Reserves/Short-

(30)

30

V. Discussion

In this section the results will be discussed as well as their potential practical implications. Regarding output loss, the accumulation of reserves appears to have no effect on the GDP losses that resulted from the economic crisis. The regression was run excluding certain countries, as well as with a 2-year vs 3-year of accumulated loss from the crisis to test robustness. In most instances results were insignificant, except for Reserves/Imports, which were significant and increased the output loss, however, only when countries with post-crisis output gains as well as when countries with fixed and pegged exchange rates were removed. This variable, however, needs to be examined further over longer periods to test the accuracy of this outcome. During times of crisis imports often decrease. Fukuda & Kon (2012) determine that increased reserves levels decrease consumption, due to the change of exchange rates. When accumulating reserves, more domestic currency enters the world market and thus devalues the currency. This in turn decreases imports as foreign products are relatively more expensive. This finding can be backed up by the regressions performed on Reserves/Imports excluding the aforementioned countries, where the results were mostly significant and were shown to increase the output lost during the crisis. Considering the other reserves ratios remained insignificant, a portion of the additional loss likely comes from decreased imports resulting from weakening currencies and a propensity to save post-crisis. Esfahani (1991) and Awokuse (2007) determined that import shortages restrict growth of country output dramatically in semi industrialized countries. Furthermore, Guisan (2004) examined the bilateral causality between private consumption and GDP in the US and Mexico. In this paper it was determined that there was a high degree of causal dependence between the two variables. Thus, the lack of consumption, imports and lower GDP may in turn be restricting the gain that can come from accumulating reserves.

(31)

31

own balance sheets. An additional issue could result from inefficient allocation of reserves to unsuitable economic sectors. Regardless if countries accumulate reserves or not, there will be significant economic losses from crises, but countries which accumulate reserves return quicker to pre-crisis economic growth levels. The effect of this growth, however, has little impact on minimizing the output loss resulting in 2 – 3 years of output loss. Nonetheless, in the longer term, these countries are more likely to improve their output loss quicker than those which do not accumulate reserves. In other words, the accumulation of reserves, therefore, seems to have little effect on the short-term detriment of the output loss from the 2008 economic crisis on Asian countries, but has longer term consequences on the recovery resulting from the growth implications.

The time taken to recover from an economic crisis, however, is reduced dramatically by accumulating reserves pre-crisis, in particular Reserves/GDP. This is in line with Dominguez, Hashimoto & Ito (2011), who determined that real GDP growth recovery was improved for countries which accumulated reserves pre-crisis. Furthermore, the results of this paper are reassured by the findings of Hong & Tornell (2005), who stated that economies which had more liquidity prior to a crisis usually maintained relatively high levels of growth during the crisis. In this paper, it was shown that countries that recover faster tended to have higher Reserves/GDP, Reserves/Imports and Reserves/Short-term debt (graphs 1-6). The regression models (table 2-10) further back up these graphs, demonstrating significant negative relationships between the years taken to recover and the reserve ratios, except for Reserves/Imports which became insignificant in the robustness checks. These results mean that the increased liquidity gained through accumulating reserves pre-crisis are utilized during and after the crisis to help rebuild the economy. Considering today’s globalized supply chain and the importance of foreign capital (Timmer, Miroudot & de Vries, 2019), in times of crisis, when the inflow of foreign capital is restricted (Forbes & Warnock, 2012), operations are often forced to shut down. With reserves, these industries can be assisted, which can allow them to sustain operations and regrow faster than without support.

(32)

32

after financial crises, and therefore, suffer (Mendoza, 2004). On top of this, if every country attempted to increased reserves excessively, there would be an unsustainable pressure on the global reserve currency and prices would rise beyond that which many developing countries can afford.

VI. Conclusion

This paper examined the effect of reserve accumulation on crisis severity and recovery for the 2008 economic crisis. The topic of reserves has been heavily discussed but remains relevant due to the sustained importance countries place upon them, regardless of their entailed costs. This paper adds to the current literature by comparing the different effects of reserves on GDP lost as a result of the 2008 economic crisis and the time taken to recover growth rates. Based on the results of this paper, countries should consider the role of reserves on crisis severity more closely. This study ran a cross sectional regression on emerging and developing Asian countries to analyze the effects of Reserves/GDP, Reserves/Imports and Reserves/Short-term debt on the years taken to recover to pre-crisis GDP growth levels and the cumulative output loss that occurred due to the 2008 economic crisis.

The main findings of this paper are that the pre-crisis accumulation of reserves dramatically helps countries improve the years taken to recover to pre-crisis GDP growth levels. However, this accumulation did not significantly reduce the cumulative output lost as a result of the 2008 economic crisis. The findings on recovery time are in line with prior authors discussed in this paper, however, a surprising result of the analysis is the little impact reserves have on minimizing the cumulative output lost. Accumulating reserves allows countries to support economies, and although consumption declines during a crisis, which may increase the output loss. Possessing additional reserves allows for industries to regain their business performances quicker post-crisis, which results in quicker growth recovery, and perhaps reduce the longer-term output loss.

Countries should also examine the extent to which they accumulate additional reserves after a crisis. The post-crisis reactionary response is usually to accumulate excess reserves. However, due to the additional costs of acquiring reserves, this post-crisis accumulation could harm their economies more than it aids them. Asian countries should certainly keep accumulating and maintaining reserves as a buffer against future crisis, however, they should examine to alternate factors to reduce the short-term output lost as a result of economic crises.

Future research & Limitations:

(33)

33

over time this would be reflected in the output loss. Additionally, it should be examined how these results would differ if countries are export-based economies vs import-based. This could influence the results as export-based economies may have to rely less on reserves as they have a steadier stream of foreign capital flowing into their economy.

A limitation of this paper that should be examined in the future is to include more control variables, since none of them were significant in this analysis. Some variables that should be further examined are unemployment, inflation, and GDP per capita, as all of these variables are likely to influence crisis severity and recovery. Furthermore, many countries the data required was not available, thereby lowering the accuracy of the results and the databases examined were not updated past 2018, this resulted in a cap for the analyzed years taken to recover. With more years available, perhaps more countries would have recovered growth levels. Finally, a longer period of time should have been analyzed for the cumulative output loss, to determine whether reserves truly don’t decrease this loss.

VII. Bibiography

Abdullateef, U. and Waheed, I., 2010. External reserve holdings in Nigeria: implications for investment, inflation and exchange rate. Journal of Economics and International finance, 2(9), p.183.

Aizenman, J. and Marion, N., 2003. The high demand for international reserves in the Far East: What is going on?. Journal of the Japanese and international Economies, 17(3), pp.370-400. Aizenman, J. and Lee, J., 2007. International reserves: precautionary versus mercantilist views, theory and evidence. Open Economies Review, 18(2), pp.191-214.

Awokuse, T.O., 2007. Causality between exports, imports, and economic growth: Evidence from transition economies. Economics letters, 94(3), pp.389-395.

Bahmani-Oskooee, M. and Miteza, I., 2006. Are devaluations contractionary? Evidence from panel cointegration. Economic Issues, 11(1), pp.49-64.

Banti, C., Phylaktis, K. and Sarno, L., 2012. Global liquidity risk in the foreign exchange market. Journal of International Money and Finance, 31(2), pp.267-291.

Benigno, G. and Fornaro, L., 2012. Reserve accumulation, growth and financial crises.

Bird, G. and Rajan, R., 2003. Too much of a good thing? The adequacy of international reserves in the aftermath of crises. World Economy, 26(6), pp.873-891.

Bird, G. and Mandilaras, A., 2005. Reserve accumulation in Asia. World Economics, 6(1), pp.85-99.

(34)

34

Bussière, M., Cheng, G., Chinn, M.D. and Lisack, N., 2015. For a few dollars more: Reserves and growth in times of crises. Journal of International Money and Finance, 52, pp.127-145. Bussière Matthieu et al. (2015) “For a Few Dollars More: Reserves and Growth in Times of Crises,” 52, pp. 127–145. doi: 10.1016/j.jimonfin.2014.11.016.

Caves, R.E., 1974. Multinational firms, competition, and productivity in host-country markets. Economica, 41(162), pp.176-193.

Cruz, M. and Walters, B., 2008. Is the accumulation of international reserves good for development?. Cambridge Journal of Economics, 32(5), pp.665-681.

Diamond, D.W. and Rajan, R.G., 2001, June. Banks, short-term debt and financial crises: theory, policy implications and applications. In Carnegie-Rochester conference series on public policy (Vol. 54, No. 1, pp. 37-71). North-Holland.

Dominguez, K.M., Hashimoto, Y. and Ito, T., 2011. International reserves and the global financial crisis (No. w17362). National Bureau of Economic Research.

Elmendorf, D.W., 1996. The effect of interest-rate changes on household saving and consumption: a survey (pp. 96-27). Division of Research and Statistics, Division of Monetary Affairs, Federal Reserve Board.

Esfahani, H.S., 1991. Exports, imports, and economic growth in semi-industrialized countries. Journal of Development Economics, 35(1), pp.93-116.

Forbes, K.J. and Warnock, F.E., 2012. Capital flow waves: Surges, stops, flight, and retrenchment. Journal of international economics, 88(2), pp.235-251.

Fukuda, S.I. and Kon, Y., 2012. 5. Macroeconomic impacts of foreign exchange reserve accumulation: theory and international evidence. Monetary and Currency Policy Management in Asia, 1500, p.120.

Gordon, M., 2005. Foreign reserves for crisis management. Reserve Bank of New Zealand Bulletin, 68(1), pp.4-11.

Guisan, M.C., 2004. A Comparison of Causality Tests Applied to the Bilateral Relationship between Consumption and GDP in the USA and Mexico. International Journal of Applied Econometrics and Quantitative Studies, 1(1), pp.115-130.

Hong, K. and Tornell, A., 2005. Recovery from a currency crisis: some stylized facts. Journal of Development Economics, 76(1), pp.71-96.

Klitgaard, T. and Higgins, M., 2004. Reserve accumulation: Implications for global capital flows and financial markets. Current Issues in Economics and Finance, 10(10).

Krugman, P. and Taylor, L., 1978. Contractionary effects of devaluation. Journal of international economics, 8(3), pp.445-456.

(35)

35

Mendoza, R.U., 2004. International reserve-holding in the developing world: self insurance in a crisis-prone era?. Emerging Markets Review, 5(1), pp.61-82.

Minsky, H.P., 1957. Central banking and money market changes. The Quarterly Journal of Economics, 71(2), pp.171-187.

Olken, B.A. and Pande, R., 2012. Corruption in developing countries. Annu. Rev. Econ., 4(1), pp.479-509.

Papadimas, L., Strupczewski, J., 2010. EU, IMF agree $147 billion bailout for Greece: From Reuters. Available at: https://www.reuters.com/article/us-eurozone/eu-imf-agree-147-billion-

bailout-for-greece-idUSTRE6400PJ20100502?fbclid=IwAR3SsY6XTqn7lyjbtp4eK7eLoBMChhSduiri9VvVE4wxIqTwUSMep ZP_pI8

Park, D. and Shin, K., 2009. Saving, investment, and current account surplus in developing Asia. Asian Development Bank Economics Working Paper Series, (158).

Pettinger, T., 2019. Economic effect of a devaluation of the currency. Available at:

https://www.economicshelp.org/macroeconomics/exchangerate/effects-devaluation/#:~:text=Elasticity%20of%20demand%20for%20exports%20and%20imports.&text=The %20impact%20of%20a%20devaluation,and%20have%20a%20bigger%20effect.

Reinhart, C.M. and Rogoff, K., 2020. The coronavirus debt threat. The Wall Street Journal, 26.

Rodrik, D. and Velasco, A., 1999. Short-term capital flows (No. w7364). National bureau of economic research.

Rodrik, D., 2006. The social cost of foreign exchange reserves. International Economic Journal, 20(3), pp.253-266.

Sekkat, K. and Varoudakis, A., 2000. Exchange rate management and manufactured exports in Sub-Saharan Africa. Journal of Development Economics, 61(1), pp.237-253.

Steiner, A., 2014. Reserve accumulation and financial crises: From individual protection to systemic risk. European Economic Review, 70, pp.126-144.

Timmer, M.P., Miroudot, S. and de Vries, G.J., 2019. Functional specialisation in trade. Journal of Economic Geography, 19(1), pp.1-30.

UEI., 2012. International reserves: IMF concerns and country perspectives. IN: IEO Report. International Monetary Fund.

(36)

36

VIII. Appendix

Additional graphs

The graphs below demonstrate how long it took countries to reach average pre-crisis GDP growth levels. The countries with a value of 8, indicate that by the year 2016, they still had not recovered these levels.

The cumulative output loss graph demonstrates how much countries lost in the 3 years after the 2008 economic crisis.

Graph 7: Years taken to recover

0 1 2 3 4 5 6 7 8 9

(37)

37

Graph 8: 3-year cumulative output loss

-3E+11 -2.5E+11 -2E+11 -1.5E+11 -1E+11 -5E+10 0 5E+10 1E+11 1.5E+11

(38)

38 Additional Robustness checks:

Regression results: Years taken to recover individually added control variables

Table 20: Reserves/GDP- individually added control variables

years model1 model2 model3 model4 model5 model6

Reserves/GDP -4.115 -5.147 -4.128 -4.087 -8.119** -5.140* [2.646] [3.228] [2.644] [2.672] [3.322] [2.984] Interest rate 0 [0.109] Exchange rate 0 [0.000] CA/GDP -1.839 [3.151] Imports/GDP 5.234** [2.369] Exports/GDP 1.36 [2.264] Constant 3.915*** 4.264*** 4.093*** 3.956*** 2.450** 3.624*** [0.813] [1.032] [0.831] [0.823] [1.111] [1.097] R-squared 0.065 0.087 0.093 0.074 0.186 0.083 Observations 37 30 37 37 35 36

Table 21: Reserves/Imports – individually added control variables

(39)

39

Table 22: Reserves/ Short-term debt – individually added control variables

Reserves/Short-term debt -1.458** -2.016** -1.459** -1.365* -1.307** -1.448** [0.582] [0.781] [0.581] [0.666] [0.598] [0.589] interest 0.004 [0.120] exchange 0 [0.000] CAgdp -1.679 [5.469] imports 2.391 [2.260] exports 1.725 [2.503] Constant 5.111*** 6.165*** 5.332*** 4.923*** 3.726** 4.398*** [1.039] [1.277] [1.060] [1.224] [1.669] [1.474] R-squared 0.215 0.294 0.251 0.218 0.253 0.231 Observations 25 21 25 25 25 25