6th Global Conference on Business & Economics ISBN : 0-9742114-6-X
THE PERSISTENCE OF THE CURRENT ACCOUNT SURPLUS FOLLOWING AN EXTERNAL SECTOR CRISIS
Juan Manuel Jauregui[(]
Global Economics and Management
UCLA Anderson Graduate School of Management[*]
and
Business, Society and Economy
IAE Business School[(]
May 2006
Abstract
This paper addresses empirically the duration of the period of current account surplus that follow a current account reversal. Analyzing quarterly data from 1980 it identifies fast current account reversals –defined as episodes of a reduction of more than one percentage point of GDP in the current account deficit in just one quarter– effected simultaneously with a period of high distress in the foreign exchange market. Using a Cox proportional hazards model in which the episodes are stratified by country it finds that GDP growth and the change in international reserves both have an effect on the hazard rate. The effect is also of non trivial magnitude: the hazard rate is reduced by 5% per each additional percentage point of GDP growth and by 9% per each additional percentage point in the reserves to GDP ratio. These surplus periods that follow crises episodes reflect the rebalancing of the external position of countries. The amount of the correction and the length of the period of surplus have not been addressed thoroughly in the literature. The results of this paper suggest that countries are more likely to end the surplus period when the growth falters and when the reserve accumulation weakens. This finding is in line with episodes of recovery with no outside credit -called by Calvo et al (2006) Phoenix miracles- and presents the puzzle of why the regained access to external finance is not accompanied with stronger growth. Somewhat surprising is that it is not a surge in imports that drive the rebalancing to end.
1- Introduction
External sector crises are at the center of academic and policy debates. This area of research has been very fruitful and has helped policymakers to understand much better how to prevent crises (see for example Edwards and Frankel, 2002). Since their effects on the economy are remarkable in many cases, this proved to be a topic for research of great relevance for the welfare of the societies. Edwards (2005a, 2004a, b) show how the cost of crises is significant. Also Hong and Tornell (2005) find that after a crisis the level of GDP remains permanently below its trend. Cerra and Saxena (2005) find similarly that lost trend growth is not regained. Baldacci, de Mello and Inchauste also find that crises have negative consequences on poverty, and in certain cases they also increase inequality. They do so through a variety of channels: economic activity slowdown, relative price changes and fiscal retrenchment, just to name a few.
From the understanding of their immediate consequences, crises management can be improved. Much work has been done and continues to be done in this area. See for example Dooley and Frankel (2002). Moreover, the analysis of the recovery after a crisis is also being the focus of much attention. For example, Calvo et al (2006) recently showed the patterns for growth and credit during the period of recovery after crises. My paper aims at understanding better the period that follow a crisis.
In the center of the debate about crises are the current account reversals, and what drives them. Edwards (2004a, b, 2002) has studied them in depth the causes and consequences of them. Previously, a good analysis on their empirical regularities was done by Milesi-Ferreti and Razin (2000). Furthermore, the debate about crises is related to one of the most important political issues of these days: global imbalances. The world economy is facing an unprecedented situation of imbalances between countries with high current account deficits, the US mainly, and countries with high current account surpluses, mainly China. The risk of an abrupt correction, also called hard landing, posses a serious threat to the stability of the global economy. Understanding deeply the current account reversals is an issue of great importance. A thorough analysis of this issue can be found in Edwards (2005b, c, d)
This objective of this paper is to increase our understanding about the recoveries of the economies after the crises of the current account. Specifically, I study the persistence of the periods of current account surplus that follow them. During them, the countries undergo a process of rebalancing of the net international investment position. This process begins with a crisis, in a traumatic way, and very few things are known about how long they last. This paper shows evidence about the factors that make these periods last longer.
2- The Empirical Model
2-1- The beginning and end of the events
This paper analyzes the duration of the periods of rebalancing of the net international investment position (NIIP) after a period of extreme market pressure in the foreign currency market. The aim is to identify the factors that drive the length of the time during which a country that was originally in deficit, that is, that had a deteriorating NIIP, rebalances it, at least partially. The beginning of the time spell will be identified by three conditions: there has to be extreme market pressure in the currency market, the current account balance has to be at a deficit in the previous period and there has to be a reduction of the CAD to GDP ratio of at least one percentage point in just one quarter. The end of the spell is chosen as the moment in which one of two things happens. Either a current account deficit begins after having a surplus or, if the correction of the current account deficit did not ever reach the surplus, an increase of the deficit begins. Therefore, the episodes last from the moment of the crisis, when the country is not allowed to continue its deficit at the same level and is forced to an abrupt correction, to the moment in which it begins or increases its current account deficit (CAD).
2-2- The extreme market pressure index
The definition of market pressure is similar to the one used by Eichengreen, Rose and Wyplosz (1995)[1]. In their paper, they use a weighted index that combines the exchange rate changes, reserves changes and, and interest rate changes. For their index, they use the differentials with respect to the center that they define to be Germany. I just use the changes in the variables, not of the differentials, since the center may be different at different times. Their study covers a shorter span of time than mine, and consequently this was not a problem for them. As the volatilities of the three components of the index are very dissimilar, being highest for reserves and the lowest for interest rates, they weight the index so that the conditional volatility of each of them is equal. In this paper I follow Wyplosz (2001) by weighting the index by the inverse of the standard deviation of each component. Also, I define a crisis as a time in which the index is two standard deviations above its mean, as Eichengreen et al (1995) do.
2-3- The Cox proportional hazards model
This periods of rebalancing are analyzed using a Cox proportional hazards model (Cox 1972) in which the hazard rate is related to the explanatory variables xj in the following way:
The hazard rate is defined as:
Where S(t) is called the survival function and T is the duration of the event[2].
This model does not make any assumption about the baseline hazard h0(t), and this is a great advantage in this setting in which there is not any compelling evidence in the literature about how this hazard should evolve through time. Moreover, there is not any theoretical model widely accepted that we could use to assume a particular shape of h0(t). This model is very suited for the problem under study since, in fact, the baseline hazard is left unestimated[3]. Our model will include time-varying covariates[4].
The model was stratified by country, which means that it estimates a different baseline hazard for each country[5]. In this way, it is recognized that country specific effects make the hazard rates at each time different between countries even for the same set of variables. It is not that one country faces a higher probability or even that this is increasing, but that the whole shape can be different, because no restrictions are imposed on the face of the baseline hazard. Therefore, the baseline hazards for the episodes j are different for each country c:
The coefficients to be estimated are required to be the same for all events regardless of the country. This analysis does not include time varying-coefficients. This is feasible, but we saw no need to include this feature in our model[6].
In the likelihood calculations, the Breslow (1974) approximation has been used for the cases of tied failures. These are the cases in which two or more episodes fail at the same time according to the data. The exact calculation can be computational intensive, and consequently this approximation is typically used.
2-4- The economic factors
The literature about the determinants of the current account is very broad (for a general review see Obstfeld and Rogoff, 1997). Based on this and on the previous academic work about crises, the economic factors to be analyzed are the following:
a) GDP growth
b) Exports
c) Imports
d) Foreign direct investment (FDI)
e) Domestic credit
f) Fiscal surplus
g) International reserves
h) Interest rate
i) Exchange rate
j) Initial CA / GDP ratio
The exports and imports are obviously important variables to include in the model for their impact on the current account. Typically, the current account reversal occurs because of a sudden reduction in imports that is followed by a surge in exports. The relative size of the adjustment that comes from each of the variables allows a characterization that separates between countries whose exports make most of the correction and those whose imports are the responsible for the current account improvement (see Guidotti, Sturzenegger and Villar, 2004). In any case, one could expect that the component that changed most during the reversal could be reverting back and so bringing the current account surplus period to an end. Along this line, Desai and Mitra (2004) find that the pre-crisis difference in export sector strength between Argentina and Thailand provides a significant explanation for the post-crisis difference in the interest rate and exchange rate movements between the two countries. This points to the strength and flexibility of the exporting sectors, and consequently, to exports as an important factor.
The rate of growth of GDP is also important. Kaminsky, Reinhart and Vegh (2004) show that net capital flows are pro-cyclical. We can expect that the current account deficit will be higher in the presence of healthy growth. Moreover, since most of the external sector crises happen in emerging markets, this dynamic of the current account turning to deficit in good times and to surplus in bad times is of great importance for this study. One line of reasoning goes as follows: emerging countries are credit constrained and whenever they are favored by international capital inflows this eases this constrain and allows them to increase investment and consequently output. Good times come with the ability to borrow and propel GDP. Also, crises are periods of current account surplus and they are related to large falls in GDP. Whenever GDP recovers one can expect the times coming back to normal, with increased access to international financial markets.
However, there is also a view that is opposite to the one just mentioned. Based on the economic success experienced by some Asian countries, and on the fact that they are export oriented, there is the idea that growth can be stronger and sustained for longer periods of time when industrial exports are high. Healthy growth and current account surpluses come together in this theory. Furthermore, it has been recently observed that many countries that face sudden stops and crises have GDP growth recover without credit. Calvo (2005) discusses this issue extensively.
Foreign direct investment has been found to be the most stable type of capital inflow. It is long term, highly illiquid and also related to technological spillovers. It has been a good indicator of the quality of the capital inflows and there has been empirical evidence that the higher the levels of FDI, the lower are the chances of having a crisis and the less negative the impact of the crisis is. It is natural to expect that this component of the capital account will have an impact in the duration of the surplus period. It may be an indicator of increased foreign interest in the assets of the country in question and a predictor of a forthcoming deficit period.
Interest rates are also affected heavily by the capital inflows. The times of inflows are also times of low interest rates and we could expect that if the current account deficit is about to begin, the interest rates should be falling in the presence of increased inflows. Arbitration in the money market is very fast, and interest rates should be a great predictor of the foreign currency availability for the future times. Also, if capital inflows increase and the government decides to intervene to sustain a high exchange rate, it may decide to sterilize the purchases of foreign currency. By doing so, the interest rates show go up, and they should go down whenever the government changes its policy. Therefore, a fall in interest rates can be preceding the end of the surplus period.