Dealing with Dutch Disease:
The Case of Chile
Lola Isabel Damstra
Master’s Thesis
Student number 10444718 University of Amsterdam Faculty of Economics and Business
MSc. Economics Track International Economics
Date: 7 July 2016
Supervisor: Dr. K.B.T. Thio
1 Statement of Originality
This document is written by Student Lola Damstra, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
2 TABLE OF CONTENTS
Acronyms 3
I. Introduction 4
II. Literature review 7
2.1. Dutch disease mechanism 7
2.2 Empirical tests 10
2.3 Policy tools 11
III. Case Study 13
3.1 Presence of Dutch disease 13
3.1.1 Real exchange rate appreciation 13
3.1.2 Wage growth 20
3.1.3 Growth in the non-copper export sectors 22
3.2 Chile’s economy 28
3.2 Implemented policy tools and their effect on Dutch disease 30
3.2.1 Fiscal policy 31 3.2.2. Revenue management 33 3.2.3 Monetary policy 35 3.2.4 Macroeconomic measures 36 IV. Conclusion 40 Appendices 42 Bibliography 48
3 Acronyms
AMCHAM Chilean American Chamber of Commerce
CPI Consumer Price Index
COCHILCO Comisión Chilena Del Cobre CODELCO Corporación Del Cobre
DIRECON Dirección General de Relaciones Económicas Internacionales
EA Euro Area
EMU European Monetary Union
FDI Foreign direct investment
FEES Fondo de Estabilización Económica y Social FIC Fondo de Innovación para la Competitivida
FRL Fiscal Responsibility Law
FRP Fondo de Reserva de Pensiones
FTA Free Trade Agreement
GDP Gross Domestic Product
GNI Gross National Income
INE Instituto Nacional de Estadísticas
OEC Observatory of Economic Complexity
R&D Research and Development
RER Real Exchange Rate
REER Real Effective Exchange Rate
4 I. Introduction
The thesis that a resource boom can have a negative impact on growth was first conceived when the Economist (1977) coined the term “Dutch disease”. In using this term they referred to the government’s poor management of the revenues that were generated by the natural gas sector after the discovery of this natural resource in the Netherlands, leading to the appreciation of the national currency followed by negative economic growth.
Corden and Neary (1982) were the first to formalise the theory on Dutch disease and since then the term has been generalised to refer to the tendency of a boom in the natural resource sector to crowd out the tradables sector that is not related to the boom, by bidding up wages and through the appreciation of the real exchange rate. This leads to a deterioration of the international competitive position, thus lowering the rents of exporting sectors. For a small open economy that is dependent on trade this can have as a consequence that the net impact of the resource boom on the economy is negative.
Dutch disease has been analysed quite extensively over the past decades and a lot of attention has been given to the factors that cause it. This led to the findings that not only a commodity boom, but that all kinds of factors that result in a foreign currency influx, such as foreign aid inflows, can lead to the Dutch disease, which is not part of the topic of this thesis. A lot fewer studies have been devoted to the policies that can prevent the economy from experiencing the negative effects of a natural resource boom. In this view Chile is a remarkable case as it has often been cited as a counterexample of the resource curse.
The Chilean economy is hugely dependant on copper. The contribution of copper as a percentage of Chilean export revenue reached 57 percent in 2006 and the next year 21 percent of GDP was generated by the copper sector. Given the large share of copper in the export basket and the volatility of copper prices, which are determined on the world market, we can expect these price fluctuations to significantly influence the exchange rate and according to Dutch disease even lower GDP. This volatility can be seen in figure I, which shows that average yearly prices on the London Metal Exchange fluctuated in the range of 70.6 to 399.7 in the period 1999 to 2011.
5
Source: Graph constructed by author using data from the US Geological Survey and Cochilco
It appears that the macroeconomic policies carried out by the Chilean government have been crucial in coping with the volatile copper prices and the appreciation pressures on the exchange rate (Meller and Simpasa, 2011) and that Chile stands out as an exemplar of effective management of natural resource revenue and containing exchange rate appreciation.
Few case studies have looked at the role the government played in mitigating or preventing Dutch disease effect. This paper will contribute to the debate on Dutch disease by analysing how it affected the Chilean economy and how the government conducted its economic policy that is aimed at dealing with Dutch disease. The research question of this thesis is: “How did the measures implemented by the Chilean government affect the presence of Dutch disease effects?” To be able to do so it is necessary to first establish which Dutch disease effects did and did not occur by empirically testing the data from before, during and after the 2003-2007 copper boom. Next, a quick overview of the economic history of Chile will be presented. In the final chapter it will be examined which policy tools were employed, together with an analysis of their influence on the presence or absence of Dutch disease effects. This will be done by firstly looking at Chile’s fiscal policy, mainly its stance on fiscal expenditure and how this affected macroeconomic stability. The next part investigates how windfall revenue was managed by looking into the most important sovereign wealth funds,
6 namely a stabilisation fund and an infrastructure fund. Of particular interest will be how these funds impacted budget spending, external debt and non-copper economic activities. This is followed by an analysis of the Central Bank of Chile’s monetary policy with regard to the exchange rate and how it dealt with inflationary pressures. Lastly this paper will look at the macroeconomic measures that were undertaken. In particular, it will review Chile’s stance on trade policy and its effect on production and export growth in non-copper sectors. By assessing the effectiveness of Chile’s policy tools this paper can provide a lesson for similar countries in coping with Dutch disease.
7 II. Literature review
The first section of this literature review will consider the mechanism behind Dutch disease. It will explain the forces that drive it and the direct and indirect consequences, followed by a review on the empirical tests that are used to detect the materialisation of these consequences, or “symptoms”. The final section analyses the different policy tools that can be used to fight off the main symptoms of Dutch disease, and specifically those relevant to the case of Chile.
2.1. Dutch disease mechanism
Corden and Neary (1982) explain the Dutch disease phenomenon by dividing an open economy into three sectors. Two traded goods sectors of which one is the “lagging” sector trading in non-commodity goods, to which in Chile’s case traded services are added, and the other is the booming sector trading in commodity related goods, here copper. The third sector provides locally produced goods and services that are not tradable, such as construction, housing services, personal services and wholesale and retail. Crowding out of the tradable sector in non-commodity goods occurs through two channels, namely the
resource movement effect and the spending effect.
The first effect is related to the shift of production factors. Corden and Neary explain that, as a result of the boom, assuming elasticity of supply, the demand for production factors in the booming sector goes up. Whether it is because of a technological improvement, a discovery of new resources or an exogenous price increase of its export product relative to import prices, the increased demand for labour and capital in this sector leads to higher returns to the respective factors. Assuming that production factors are mobile between sectors this causes a movement of factors towards the booming sector, which provides higher returns. The decline in output in the non-tradables sector leads to excess demand, causing a price increase of non-tradables, making them relatively more expensive than traded goods. This represents a real appreciation of the exchange rate.
The second effect is induced by the increased returns on commodity production, causing extra inflow of foreign currency, and consequently an increase in the domestic supply of money, which represents an increase in aggregate income. Under fixed exchange rate this allows for a rise in the level of domestic spending, thereby increasing demand for non-tradables. It is assumed that tradables’ prices are determined on the world market, whereas the price of non-tradables is determined by the productivity levels. Therefore, only the price of non-tradables increases through the rise in spending making non-tradables relatively more
8 expensive. The rise in the price of non-tradables relative to tradables constitutes an appreciation of the real exchange rate. Under a floating regime the increase in money supply drives up the value of domestic money, thus representing a rise in the nominal exchange rate. In this case the spending effect also causes a real appreciation of the exchange rate.
The excess demand for non-tradables leads to an increase in the demand for labour in this sector. This will bid wages up further and, if labour is mobile between sectors, this will incite a shift of workers from the lagging export sector to the non-tradables sector. Now, exporting sectors are also forced to raise their wages, without being able to compensate by increasing their prices. Non-commodity export sectors see their costs rise, which reduces mark-ups and thus profits. In response they reduce production leading to unemployment (Oomes and Kalcheva, 2007). This effect is referred to as indirect de-industrialisation. Although the term would seem to indicate it only involves manufacturing goods, it clearly applies to all exporting sectors that are not related to the booming commodity, often the agricultural sector.
An important debate around Dutch disease is the appropriateness of the use of the word disease. Especially regarding the real exchange rate the term disease is not always fitting. As noted by Hevia and Nicolini (2015):
“Dutch disease is not really a disease, it is just the optimal response of prices and quantities to a relative price shock.” “Indeed, in the model, following a large increase in the price of the exportable, the nominal exchange rate - and the real, given price stability - suffers a strong appreciation, generating the traditional Dutch disease effect, which is the optimal response of prices and quantities to a relative price shock.”
The increase in copper prices triggers an instant (real) exchange rate appreciation, as it is generally defined as the price of non-tradables relative to tradables. An exogenous international price increase always has an instant effect on the terms of trade and this in itself should in the author’s opinion not be dubbed as having contracted Dutch disease. The instant response of the exchange rate to a change in prices is natural and in fact an adjustment towards equilibrium as the appreciation reflects a change in the underlying fundamentals. However, when the reaction to this change in fundamentals is based on incorrect expectations, for example when the permanence of a terms-of-trade gains is overestimated, the real exchange rate overshoots and a substantial overvaluation has a strong negative impact on growth (Brahmbhatt et al., 2010). Therefore an important distinction is to be made with regard to the real exchange rate is the appreciation that is due to a nominal increase, i.e. terms of trade gain, and the increase that stems from inflation, as it is the latter
9 that is of concern when using the term disease. It would therefore be better to limit Dutch disease to symptoms that do not inevitably follow from a price impulse but that are created within the economy. It is when the reaction to the change in relative prices and to the adjustments is not adequate that Dutch disease will come into play, leading to the contraction of non-copper export sectors.
In particular when the windfall from the copper boom ends up in the hands of the government, the economy can become very prone to Dutch disease. The boom generates significant surpluses for the government allowing it to increase spending, thus causing a level of inflation higher than the level directly resulting from the copper price impulse. This is a large concern in the case of Chile as the government receives large rents from copper production, via royalties, tax and profits on production. Therefore the impact of the copper boom on the economy depends to a large extent on government policies. In fact, the government can even negate some of the Dutch disease effects on inflation and the exchange with its surpluses.
Source: Table constructed by author based on studies by Oomes and Kalcheva (2007) and Algieri (2011)
As shown in Table I, if Dutch disease were to prevail we expect to detect at least some of the following effects; an increase in prices in the non-tradables sector leading to a real exchange rate appreciation; a rise in real wages in all sectors; and a drop in output and employment in the exporting sectors that are not related to copper. The net effect on output and employment in the non-tradables sector depends on whether the resource movement or the
10 spending effect is larger. If the spending effect dominates, the net effect will be an increase in output and employment in the non-tradables sector, leading to indirect de-industrialisation. The goal of the tests constructed in this thesis is to detect the symptoms above in order to establish the presence of the Dutch disease.
2.2 Empirical tests
The main symptom of Dutch disease is the appreciation of the real exchange rate, thus worsening international competitiveness of exporters through a strong local currency. Algieri (2011) tested for this symptom by looking at whether real oil prices are cointegrated with the real effective exchange rate (REER) while controlling for productivity growth, relative government expenditure and foreign reserves. Algieri used the Augmented Dickey Fuller (ADF) and Perron unit root tests to determine the trending behaviour of the series and then extracts the trend to transform them into stationary series. To test for a long-run relationship between the variables above she deployed a VECM-based cointegration analysis using the Johansen methodology, which simultaneously evaluates multivariate equations. In this paper Algieri’s approach will be adopted to determine the long run relationship between the REER and copper prices, provided that the variables are integrated of order one.
Another symptom that is predicted by the Dutch disease thesis is that an increase in commodity prices causes the real wage to rise through both the spending and the resource movement effect. Because of a lack of data on wage growth across sectors, this paper will compare the general real wage growth with copper price fluctuations and offer a descriptive analysis on the presence of the resource movement effect by looking at employment in the copper sector.
As seen in the previous section, Dutch disease leads to a de-industrialisation or crowding out effect, amounting to a reduction of output in the lagging tradables sector. Gylfason et al. (1999) used an export measure to identify the presence of this effect. The first symptom is skewness of the composition of exports towards the booming commodity; the second is a reduction in exports altogether. These indicators can be tested by looking at the commodity export ratio and at the ratio of exports to GDP, respectively. A rise in the share of copper exports and a decline in the total export to GDP ratio points to the presence of direct de-industrialisation. Oomes and Kalcheva (2007), in their study on the presence of Dutch disease in Russia, tested for the crowding out effect by estimating the effect of higher commodity prices on the output growth rate in the manufacturing sector. They tested the effect of higher oil prices on five different manufacturing sectors while controlling for changes caused by the change in demand from abroad by introducing a variable for EU import from Russia. A limitation of
11 introducing this variable is that it is likely to result in endogeneity. EU import from Russia is presumably correlated with Russian export and the exchange rate, which under the Dutch disease hypothesis is influenced by oil prices. If the export equation in turn is also dependent on copper prices and EU import from Russia, this leads to simultaneity. The test to be performed in this thesis will therefore include the trading partner’s total import demand of the product instead of import demand from Chile only. A reduction in the product’s import demand is likely to reduce Chile’s export volume of the product, but the reduction in total demand is unlikely to be caused by changes in Chile’s exchange rate. In this thesis the effect of a copper boom on the volume of other export products will thus be analysed through a cointegration of copper prices with each of the five largest export sectors other than copper. The main trading partners’ total import demand of these products will be used as a control variable.
To test for indirect de-industrialisation Ruehle and Kulkarni (2011) looked at the correlation between GDP in tradables, non-tradables and the copper sector. When indirect deindustrialisation is present it is expected that non-tradables have a negative linear relationship with copper, whereas tradables will have a positive correlation. However, in this approach it is important to use real GDP so as to be able to measure reductions in real output, reflecting the loss of production factors.
2.3 Policy tools
During a commodity boom government receipts increase through enlarged revenue from tax on commodity profits paid by private companies and through revenues received from commodity profits on production that is state-owned. Therefore the spending effect is not only caused by higher disposable income but also through larger government revenues allowing for higher government spending.
According to Bresser-Pereira (2008) the Dutch disease can be neutralised via two measures. The first is a tax on the sale of commodity goods. The second is the offshoring of these tax revenues through the creation of an international wealth fund. By not spending the increased revenues domestically, foreign currency needs not be converted into national currency, which would lead to an appreciation of the Chilean peso.
Davis et. al (2011) point out that a stabilisation fund can be used as a mechanism to shield the economy from the volatility of prices and revenue. The stabilisation fund is meant to insulate government expenditure from fluctuations in commodity prices, i.e. uncertainty and volatility is transferred from public expenditure onto the fund. This can be done by making the wealth fund contingent on commodity prices. Such a fiscal rule targets countercyclical
12 spending by increasing government spending when prices are low and by accumulating government resources when prices are high. This decoupling of fiscal revenue from windfall resources should help to minimise Dutch disease effects by smoothing consumption, thus curbing inflation and nominal exchange rate appreciation.
Other potential benefits of a fund are alluded to by Van der Ploeg (2011). These relate to the absorption constraint that arises from the assumption that supply curves are not perfectly elastic inasmuch as the expansion of production requires investment. These investment goods comprise both capital structures and human skills, which can’t be imported from abroad (or only partly) and thus accumulate only over time. However, the windfall revenues prompt an increase in demand for non-tradables, causing the economy to face an absorption problem, leading to inflation in non-tradables and putting appreciating pressures on the currency. Therefore Van der Ploeg considers it wise to build an investment fund to park some of the windfall revenue until absorption constraints are alleviated and the country is ready to absorb extra domestic investment.
As the economy tends to specialise in the commodity that uses abundant factors more intensely (Heckscher-Ohlin model) it will become more and more reliant on the commodity sector and its prices, making the economy still more vulnerable to Dutch disease effects. Governments can offset the Dutch disease effects of an exchange rate appreciation by temporarily subsidising agricultural and manufacturing sectors, thus promoting diversification (Ross, 1999). Strong advantages associated with diversification are also found by Balassa (1985). Diversification of the export sector by boosting its productivity helps improve the international competitive position. This is done through outward oriented policy responses to external shocks, which entails providing incentives to production for domestic and export markets (Balassa, 1980). In the face of Dutch disease these advantages of outward orientation are apparent from the positive correlation between the extent of export promotion to the terms of trade effects of external shocks and the rate of economic growth (Balassa, 1985).
13
III. Case Study
3.1 Presence of Dutch disease
The main symptoms of Dutch disease will be analysed in the next three sections, whereby the first section looks at real exchange rate appreciation caused by an increase in copper prices. The second section examines the presence of rapid wage growth and the resource movement effect. The last section looks at direct and indirect deindustrialisation, the former by testing the effect of copper prices on non-copper export sectors and the latter by analysing the linear relationship between copper production and non-tradables and tradables production.
3.1.1 Real exchange rate appreciation
In this paper the effect of copper prices on the Chilean peso is analysed using the effective exchange rate. As copper exports are priced in dollars, the direct effect on the dollar/peso exchange rate is most pronounced, confirmed by the clear downward trend of the exchange rate during the rise in copper prices, which represents an appreciation of the Chilean peso (appendix I). However, this measure is not the most interesting to use, as exports are not solely invoiced in dollar, though it is the most dominantly used currency. On that account the effective exchange rate is used, which is the value of a currency measured as a trade-weighted average of the bilateral RERs of the country and each of its trading partners (IMF, 2007). Therefore, this measure is a good indicator of the strength of a currency relative to its trading partners and will form a good indicator of how Chile’s overall competitiveness is affected by copper prices. Since the Dutch disease effect on the exchange rate not only comes from the instant terms of trade gain of higher copper prices, but also from inflation via the spending effect, the exchange rate measure used for the analysis should also account for the increase in relative consumer prices, which is done by using the real effective exchange rate (REER). From 2003 to 2006 the REER was subject to rapid appreciation after six years of constant depreciation. This appreciation coincides with a sudden surge in copper prices, suggesting the presence of Dutch disease. Growing copper prices have been accompanied by a significant appreciation of the peso from 2003 to 2006 as can be seen from figure II, which depicts the fluctuations in the REER and copper prices as measured by an index based on the year 2010. However, comparing these effects to the volatility of the peso in earlier years the fluctuations during the copper boom seem attenuated. In the previous copper boom, from 1994 to 1997 the REER appreciated 20.1 percent, followed by a depreciation of 20.8 percent. Comparing this to the appreciation and depreciation of respectively 10.1 percent during the boom of 2003-2006 and 3.6 percent during the “bust”
14 from 2007 to 2009, even though copper prices rose by over 350 percent during this boom compared to 30 percent in 1994-97, this would suggest Dutch disease effects were largely mitigated.
Source: Graph made by author using data from the World Bank and Central Bank of Chile
It should be noted that the depreciation from 1998 coincides with both the emergence of the Argentinian debt crisis, the effects of which are believed to have been transmitted to the Chilean exchange rate, and the transition of the exchange rate towards a free floating regime, which has officially been in place since 1999, both probable culprits for exchange rate volatility.
To test whether the rise in copper prices has been at the origin of real exchange rate appreciation, this paper performs a cointegration analysis. This type of analysis allows determination of the long run relationship among the variables. To determine the effect of copper prices on Chile’s real exchange rate this paper follows an analytical model that merges three strands of literature stressing different elements of influence on exchange rate movements. The first strand of literature explains the movements through the Balassa-Samuelson effect, according to which there will be real exchange rate appreciation when an economy experience stronger productivity growth in the tradables sector than its trading partner, represented by a positive productivity differential. The reason that productivity growth causes the appreciation of the Chilean peso is that, under the assumption of labour
15 mobility, it leads to wage increases in all sectors, including non-tradables, where it is followed by a price increase of non-tradables. The second strand points to the importance of including relative price changes between imports and exports, or in other words, terms of trade changes (De Gregorio and Wolf, 1994). The third strand emphasises the influence of fiscal and monetary policy changes on the exchange rate.
This has led to the following equations in which model (a) is based on the study by Algieri (2011) and equation (b) includes two additional variables that have been found to be important determinants of the exchange rate:
ln REER = α + β1 ln copper + β2 ln prod + β3 ln govt + β4 ln res + γ float (a)
ln REER = α + β1 ln copper + β2 ln prod + β3 ln govt + β4 ln res + γ float
+β5 ln spending + β6 ln assets (b)
REER represents the CPI-based real effect exchange rate obtained from the IMF, copper is
the price of copper (Central Bank of Chile), prod is productivity in the manufacturing sector, calculated using data from Oxford economics and the OECD, spending is government final consumption as a percentage of GDP as issued by the World Bank , govt represents the ratio of government expenditure to government revenue (DIPRES) which is a proxy for fiscal deficit, res denotes net foreign assets published by the World Bank and float is a time dummy which is equal to 1 when the exchange rate is under a free floating regime. All variables are yearly indices based on the year 2010 with the series going from 1988 to 2014 for model (a) and from 1991 to 2014 for model (b). In this model the second strand of literature is represented by copper, the copper price being a proxy for the terms of trade, as an increase in copper prices is assumed to constitute an increase in the price of export relative to import. The Balassa-Samuelson effect is usually represented by the productivity differential between the country concerned and its trading partner. This will not be included in this analysis, because Chile has a perfectly diversified trade portfolio, i.e. its trade value is perfectly dispersed across many markets (measured by the HHI, which is calculated to be 0.07 for Chile in 2006). As such, the trade-weighted productivity differential would need to be calculated, which is too complicated to be done here. Instead, productivity growth in the manufacturing sector will be used following Algieri (2011).
The third strand of literature is represented by several variables. The first is govt, the ratio of government expenditure to government revenue. A rise in government revenue relative to government spending causes the fiscal deficit to decrease. This leads to lower interest rates, causing capital outflow and results in the depreciation of the Chilean peso. Spending is the
16 second variable that represents Chile’s fiscal position. In the short term increased government spending might appreciate the real exchange rate by increasing relative prices, but in the long run this is likely to have a negative effect, by undermining confidence in a currency (Osbat and Schnatz, 2001). The ambiguous effect higher spending might have on the exchange rate, through higher interest rates, which causes appreciation, is accounted for by including the government ratio govt.
The variable included to represent monetary policy is that of international reserves (Algieri, 2011; Oomes and Kalcheva, 2007). The build-up of foreign exchange reserves is either the sale of domestic currency for foreign currency or the preserving of foreign currency state income in the bank, in both ways keeping the domestic currency value lower than it otherwise would have been (Ebrahim-Zadeh, 2003). The Central Bank’s accumulation of foreign currency is thus an important factor in avoiding exchange rate appreciation and is therefore included in the regression. In addition, net foreign assets are included as an explanatory variable as it considered a long-run determinant of the REER via balance of payment variations (MacDonald and Nagayasu, 2000). An increase in the outstanding stock of debt requires an increase in financing by foreign investors. For these investors to rebalance their portfolios towards Chilean assets they require a higher yield, which at given interest rates can only be obtained through the depreciation of a currency. The variable assets is thus expected to have a positive relationship with the REER.
Furthermore, a dummy variable is included that is equal to one when the Chilean exchange rate is under a free floating regime, officially put in place in 1999.
The methodology of this analysis starts with an ADF unit root test using MacKinnon critical values, to check whether series are stationary or not and whether they are when first differenced, i.e. when the trend is subtracted. Conventionally the Akaike information criterion is used for the determination of the lag length, but due to the small sample size only one lag will be included. The ADF unit root test at 5 percent critical values fails to reject the null hypothesis of non-stationarity for all of the variables, also when including one lag, for each of the following three models: (i) with intercept alone (ii) with intercept and trend, (iii) with no intercept and no trend. All tests reject the null at the 95 percent confidence level, meaning the variables are integrated of order one, except for reserves and the government ratio which are found to have a unit root for model (iii). Based on plots and the study by Oomes and Kalcheva (2007), it is assumed that the model with an intercept and trend is most realistic, for which the ADF-test results are displayed in appendix II. The test also finds that the coefficient on the lagged value of the variables are negative, another necessary condition for the model to be valid. This non-stationarity indicates that OLS regressions can lead to spurious results, or in other words, a seemingly significant relationship between
17 uncorrelated variables. The ADF test rejects the null hypothesis of a unit root for the variables when first differenced, also when including one lag (appendix II). This validates the use of a cointegration analysis, as it is found that there might exist a cointegrating vector of the variables’ coefficients, forming a linear combination of the variables that is stationary. In other words, it implies that a statistically significant, long run equilibrium relationship exists between the series (Stock and Watson, 2011).
To determine the number of cointegrating equations the Johansen test is used. The null hypothesis of a maximum number of cointegrating equations (CEs) is rejected when the statistic is larger than the critical value. The results of the Johansen test can be found in table II below. For model (a) the trace statistic shows there is at most one cointegrating equation, though the max statistic indicates there is no cointegration relationship. When including government spending and net foreign assets, model (b), the trace statistic indicates that the model has at most two cointegrating relations, whereas the maximum eigenvalue statistic suggests there is at most one cointegrating relation. The different statistics lead to different conclusions on the number of cointegration equations, but because the possibility of one cointegrating vector is not ruled out this paper will continue to examine these models with a cointegration analysis.
18 The vector error-correction model allows estimation of the long run effects and by imposing Johansen normalisation a unique cointegrating vector is identified. Table III below presents the estimated cointegrating vectors for both the models. Model (a) displays the opposite sign of what would be expected for the coefficient of copper and govt, though neither is statistically significant. Model (b) displays the expected sign for all of the coefficients of the cointegrating vectors and all are significant at the highest level. Because all of the variables, with exception of the dummy, are in logs, the coefficients of the variables can be interpreted as long run elasticities. This means that in model (b) a one percent increase in the price of copper will lead to an approximate 0.1 percent real appreciation of the exchange rate.
The VECM of equation (b) suggests that the accumulation of foreign exchange reserves and the implementation of the free floating regime have allowed Chile to prevent a large part of the appreciation. It estimates that a 1 percent increase in international reserves leads to a 0.2 percent depreciation of the exchange rate and that a 1 percent increase in net foreign assets leads to a 0.1 percent depreciation of the exchange rate, indicating that the Central Bank can play an important role in mitigating the Dutch disease effect.
19 Moreover, 1 percent productivity growth in tradables is found to cause a 1.5 percent exchange rate appreciation, an effect fifteen times stronger than that of copper prices. Fiscal policy has a delicate role to play. A rise in government spending causes the real exchange rate to depreciate because of loss of confidence, whereas the rise in fiscal deficit leads to the appreciation of the peso via higher interest rates. This means that when the ratio of spending to GDP as well as the ratio of spending to revenue falls, changes in fiscal expenditures have an ambiguous effect on the exchange rate. Inasmuch as Chile’s government expenditure is countercyclical to government revenue and therefore to copper price fluctuations, the government ratio goes down when copper prices rise and vice versa. Increased tax revenue normally leads to increases in government spending, but in the case of Chile the opposite happens. Government spending and copper prices are negatively correlated, which means that a rise in copper prices results in a reduction of the fiscal deficit, as well as a reduction of spending as a percentage of GDP, i.e. there is downward and upward pressure on the exchange rate. Yet, the coefficients on the cointegrating vectors indicate that the effect of a 1 percent increase of the government ratio leads to a 0.5 percent depreciation, while a 1 percent increase of government spending leads to a 1.3 percent depreciation, an effect 2.6 times stronger. This can be explained by the double message the increase in government spending could send in Chile’s case. In the first place, increased government spending could lead to expectations of distorting taxes and unbalanced growth (Osbat and Schnatz, 2001). Secondly, given Chile’s countercyclical fiscal policy, this means that the government expects copper prices to drop and its largest exporting sector to bring in less revenue.
Although model (a) does not find a statistically significant relationship between copper prices and the REER, empirical evidence from model (b) shows that there is a positive long run relationship between copper prices and the exchange rate in Chile. This is to be expected given the direct terms of trade gains.
One cannot speak of Dutch disease in the exchange rate in the literal sense of a disease, as research by the IMF on Chile’s macroeconomic policies suggests that the REER is largely in line with its fundamentals, meaning the appreciation of the exchange rate represents growth in productivity and macroeconomic factors. The exchange rate is very close to its equilibrium value and the appreciated currency thus is not overvalued (IMF, 2011). This means that the instant terms of trade gain has been reacted to justly, and the appreciation of the exchange rate reflects the adjustment toward the equilibrium value that is in line with the changed underlying fundamentals. This could indicate that the government has managed to contain agents’ reactions that lie at the heart of the disease.
20
3.1.2 Wage growth
The resource movement effect is directly linked to the rise in income in the booming sector. As demand for labour in the booming sector rises, the real remuneration of labour in this sector increases. Under the condition that labour is mobile between sectors, a rise in wages causes labour to move towards the booming sector. The movement of workers away from the lagging sector causes a shortage of labour in that sector followed by a rise in wages, causing a loss of international competitiveness. Though the hypothesis of wage equalisation in all sectors is far from reality, it is expected that the movements in mining wages are correlated with wages in other sectors, if labour can shift from one sector to another. The resulting higher wages in the non-tradables sector then in turn cause a price increase of the non-tradable goods, representing a real appreciation of the exchange rate. However, if the condition is not fulfilled and labour cannot move to the copper sector, the rise in wages in the copper sector does not affect wages in other sectors.
Because of a significant lack of data on wage growth in different sectors, this section will give a descriptive analysis of the situation. The mining sector usually forms an enclave within the economy, the production factor capital used for mining often being sector specific and the work representing few similarities to other sectors. Indeed, reports suggest that despite the very high wages, in fact the sixth highest in the world (Cochilco, 2008), the mining sector seems to have been lacking the capacity to attract the workers it needs.
One explanation of this is the infeasibility of labourers in other sectors to do the work in the mining sector. This is especially likely because the key positions in the mining sector require a high level of education and training due to the increasing sophistication of mining operations (Pellandra, 2015). It would therefore be likely that the rise in mining wages would cause a shift of skilled workers toward this sector. However, Fundación Chile (2011) indicates that there is a mismatch between supply of and demand for skilled labour. The report indicates that the mismatch will lead to a significant shortage in terms of quantity as well as quality from 2011 onwards. One reason is that the skills necessary to mining projects are not acquired through common university degrees, such as business law and medicine. With the mining sector becoming increasingly automated and mechanised there is more and more need for workers with (computer) engineering and information technology knowledge to fill positions such as equipment operators, maintenance supervisors, and processing and extraction professionals. Moreover, jobs in the mining sector are for numerous reasons considered an unattractive career option by graduates (Pellandra, 2015).
Figure III, which shows the monthly index of the price of copper and real wage from 1993 to 2009, gives an indication of the impact of the copper boom on the evolution of real wages. The real wage has steadily increased between 1993 and 2009 and has reacted little to
21 the important changes in copper prices. Remarkably, the real wage grew at its fastest rate in 13 years at the end of 2008, just as copper prices reached their lowest level.
Source: Graph made by author using data from the Central Bank of Chile and the INE
This section concludes that it cannot be inferred from the real wage growth shown in figure III that this particular Dutch disease effect has been of significance in the case of Chile. Moreover, taking into account that the copper sector forms an enclave and has trouble attracting workers from other sectors, the resource movement effect seems to have been of lesser importance in Chile. This means that if Dutch disease is present, the spending effect has probably prevailed over the resource movement effect, thus predicting that output and employment have decreased in the export sectors and increased in the non-tradables sector (table IV).
22
3.1.3 Growth in the non-copper export sectors
Under the Dutch disease hypothesis the rise in copper prices leads to an increase in copper exports combined with a decrease in non-copper exports. This crowding out effect of the non-copper export sectors is combined with indirect de-industrialisation if tradables production decreases relative to non-tradables production, as is expected under the assumption that the spending effect has prevailed over the resource movement effect (table IV, section 3.1.2). Indirect de-industrialisation can be absolute or relative, the former meaning there was negative growth in the non-copper export sector. If relative de-industrialisation has taken place, there will have been an increase in the share of non-tradables in the economy at the expense of the share of non-copper tradable sector in the economy.
The appreciation of a currency following the commodity’s price increase can have two consequences; either producers increase foreign currency prices to maintain profit margins or they keep prices at the same level and see export prices measured in national currency drop. In this paper it is assumed that foreign market prices are very rigid when invoiced in another currency and that exports are seldom invoiced in Chilean pesos and instead are invoiced in US dollars or the export destination’s currency; conform to the findings of Cravino (2014). This means that an appreciation of the peso leads to lower export prices when converted into pesos and that companies who have most costs incurred in pesos suffer the highest loss in profit margins. Hence, as a result of the lower prices the export volume is expected to drop. Thus, Dutch disease predicts a slowdown in the annual export volume as a result of the rise in copper prices.
To verify whether this crowding out effect has ensued in Chile, the export volume of five export products that are not related to copper will be investigated. The export sectors being considered are the ones that had the largest export value in 2002, right before the copper boom set in. These sectors were vegetable products, foodstuffs, animal products, paper goods and forestry. The test will inspect the effect of the copper boom on the export volume of the most important products of the aforementioned sectors, namely grapes, wine, fish fillets, chemical sulphate bleached wood pulp and wood chips.
The annual export index of wine, grapes, salmon and trout, woodchips, bleached pulp and copper in figure IV shows that the export volume in grew in all export sectors, apart from woodchips, from 1996 to 2003, the years that the Chilean peso depreciated considerably. This depreciation gave a strong boost to international competitiveness and helped Chilean exporters to access foreign markets. However, when the copper boom arrived in 2003 the export volume kept on rising, despite an appreciating peso, with wine exports that grew by 60 percent and bleached pulp which more than doubled from 2003 to 2008. Conform to the
23 finding that the resource movement effect was of lesser importance (section 3.2.3), copper production decreased during the boom, except for the year 2007.
Source: Graph constructed by author using data from the INE
This export growth lasted until 2008, when there was a significant reduction in export volume in the salmon and trout, grapes and wood chips sector, though wood chips export picked up again the year after. Bleached pulp exports plunged in 2009. However, this drop coincides with the global recession, which caused a huge slump in import demand in Chile’s main trading partners (appendix III). Given this slump, a possible explanation for the slowdown in the export volume of fish fillets and grapes could be the sheer reduction in total US and Euro Area (EA) import demand in 2008. In the same vein, the temporary drop in wood chips export in 2008 could have been due to lower demand in Japan. Moreover, the rapid decline in fish fillet export growth is explained by the outbreak of a deadly viral disease of marine-farmed Atlantic salmon that came to light in 2008, causing large production losses and loss of export markets (Alvial et al., 2012).
To determine to what degree this downturn is due to Dutch disease it is necessary to analyse the effect of copper prices on the export volume, as well as the effect of other factors that might have cause the export volume to decline, such as a decrease in foreign demand for the products.
In 2002 the US was the main export destination of Chilean wine, grapes and fish fillets; Japan of wood chips and the EA-12 the main export destination of Chilean pulp. An
24 overview of the relative size of these exports and the trade partners to which Chile exported more than 10 percent of a product's’ total export value can be seen in table IV below.
Source: Table constructed by author using data from The Observatory of Economic Complexity
Note: Import demand for bleached pulp is calculated using a weighted average of conifer and non-conifer bleached pulp.
The relationship between copper prices and export growth of these products is tested with a cointegration analysis based on the model used by Oomes and Kalcheva (2007). The model consists of the export volume for an individual sector as the dependant variable and copper price as the independent variable, while controlling for foreign demand. The import value of the four largest trade partners is used as a control variable for foreign demand. These import volumes isolate the change in export that is due to a change in the demand side. Given that export to the different countries is not equally important for each product, a country’s import demand of the relevant product will be included in the model, if it has at least a 15 percent contribution in Chile’s total export value of that product in 2002 or 2010 (before or after the boom). This 15 percent threshold is somewhat arbitrary, but is chosen so that the export destination constitutes a relatively important market. This gives the following log models for the different sectors:
ln fish = α + β1 ln copper + β2 ln US + β3 ln Japan (1)
ln pulp = α + β1 ln copper + β2 China + β3 ln EA12 (2)
ln grapes = α + β1 ln copper + β2 ln US (3)
ln wine = α +β1 ln copper + β2 ln US + β3 ln EA12 (4)
ln woodchips = α + β1 ln copper + β2 ln Japan (5)
ln manufacturing = α + β1 ln copper (6)
Copper denotes copper prices and once again represents the terms of trade, this time as part
25 for the product in the equation. EA12 denotes import demand from the twelve countries that were the first to be part of the euro area (EA). The reason EA12 is used instead of import demand from the whole of the EA, is because of a lack of data for the other countries. Manufacturing represents industrial production of manufactured goods in Chile. All import and export variables are obtained from the UN Comtrade database. Copper prices come from the Central Bank of Chile. All variables are yearly indices based on the year 2003 and the series go from 1993 to 2015 for model (1), (2) and (3), from 1992 to 2015 for (4) and (5) and from 1992 to 2013 for model (6).
The methodology used to test the effect of copper production on other export sectors is the same as the one set out to test the effect of copper on the REER, namely the Johansen cointegration analysis. The first step in this analysis is to test whether the variables are stationary through the ADF unit root test using MacKinnon critical values. The tests indicate that the variables are integrated of order 1, justifying the use of a cointegration analysis. The results of the VECM visible in table V indicate that copper production has not had a
statistically significant negative effect on either of the five sectors’ export volume. On the contrary, for the wood sector a one percent increase in copper prices is found to increase wood exports by almost 0.37 percent at the highest level of significance. The same positive
26 relationship holds between copper prices and grapes export, pulp export and manufacturing production. Only for the wine sector copper prices are found to have a negative relation with its export volume, however, this is not statistically significant. An explanation for the positive cointegrating vector of copper prices and other products’ export volume is that a rise in copper prices represents higher government revenue which is used to increase international competitiveness through innovation. This means that higher copper prices can benefit other export products by raising productivity.
The model also estimates a negative coefficient of 11.36 for EA-12 wine demand. The reason for the negative relation between EA-12 import and Chile export of wine is that Chilean wine export has been growing substantially, with annual production increases of over 11 percent on average, while EA-12 wine demand has averaged 2.11 percent from 1992 to 2015. The enormous rise in Chile’s export volume is not primarily due to increased demand for wine, but first and foremost due to Chilean wine producers being able to conquer the existing market. This is made possible by major technological advances that have lifted exporters’ production constraints as well as improve the quality of the wines. Another factor that might be interfering with the cointegrating equation is the earthquake Chile experienced in 2010, which led to a massive loss of wine production in which period wine export dropped by 10 percent down while EA12 demand went up.
Based on this cointegration analysis it seems that the rising copper prices have not resulted in a decrease in export volume of other products and it even appears to have had a positive impact on several of the sectors. This means that the crowding out effect has not ensued in Chile.
The construction of a correlation matrix of production of the largest economic sectors is used to analyse the occurrence of indirect de-industrialisation (Kulkarni and Ruehle, 2011). The pairwise correlation in table VI shows the linear relationship of quarterly real GDP of non-tradables (construction, housing personal services, public administration and wholesale and retail trade) and tradables (manufacturing, agriculture and forestry, finance and transport) with quarterly real GDP of the copper sector. The matrix distinguishes between correlation before the boom, using data from March 1996 to September 2002 and during the boom, lasting from September 2002 to March 2008.
27 It can be seen that before the boom all sectors except construction and agriculture had a positive relationship with copper production, statistically significant at least at the 5 percent level. The negative linear relationship between agriculture and copper was all but non-existent and statistically not significant. In contrast, during the copper boom all sectors apart from transport and wholesale became negatively correlated with copper production, although only the correlation of copper production with agriculture and with personal services is statistically significant. The negative relationship between agriculture and copper production would seem to point towards a case of indirect de-industrialisation, or more precisely, de-agriculturalisation.
The overview of percentage GDP growth per sector from 2003 to 2007 shown in table VII refutes this premise. Indeed, overall real copper production decreased during the boom, conform to conform to the Dutch disease hypothesis which postulates that when the spending effect is larger than the resource movement effect output in the copper sector decreases. Under the same assumption that the spending effect has dominated, it is expected that there has been growth in the non-tradables sector, which is confirmed in table VII.
28 Nevertheless, the decrease in copper production and increase in non-tradables should have been accompanied by a decrease in tradables production to speak of indirect deindustrialisation. Instead, it can be seen that the tradables sector experienced substantial real GDP growth. In the case of the agriculture and finance sector there has even been growth of over 47 and 60 percent, respectively.
Moreover, appendix IV shows that during the boom non-tradables and tradables were positively correlated, the relationship between personal services and agriculture excepted. The finding that growth in the non-tradables was not at the expense of growth in the tradables sector and that these sectors were in fact mostly positively correlated seems to indicate that there has not been a case of indirect de-industrialisation.
3.2 Chile’s economy
Chile is the largest exporter of copper in the world. In 2004 it reached a record high and produced near enough 37 of world copper production, up from 10.9 percent in 1970 (appendix IV). After the Coup d’état in 1973 many market reforms were put in place, most of them aimed at liberalisation of the market, which led to privatisation of many state-owned firms. Chile’s major copper mines that were nationalised in the years before however, remained in the hands of the public by being transferred into the national company Codelco, established in 1976 (Ruiz-Dana, 2007). Three years later Codelco produced over 86 percent of all domestic copper output, a record high (Cochilco, 2016).
29
Source: Graph constructed by author using data from Cochilco
Although the largest copper mines remained in state control, a law passed in 1992 allowed newly discovered copper deposits to be invested in by private foreign investors (IBP USA, 2009). Since then Codelco’s share has declined rapidly, as more and more foreign investors were attracted to the rising commodity prices, illustrated by figure VI which shows the volume of copper production by state and privately owned mines from 1970 to 2010. Nonetheless, during the copper boom of 2005 to 2008 Codelco was responsible for a third of total domestic copper production and controlled about 20 percent of world reserves
(Ruiz-Dana, 2007).
In the beginning of the new millennium the percentage contribution of copper to GDP rose to nearly 21 percent due to significant increases in copper prices led by increased demand. Given the size of the copper sector and the volatility of prices this can have important implications for economic activity one of such being the appreciating pressures on the exchange rate. Moreover, the copper boom generates significant surpluses for the government by means of tax on sales, royalties from private companies and profits as well as taxes from state-owned production. Figure VII shows the percentage contribution of copper in total government revenue from 1991 to 2014. The share of copper budget receipts in total government revenue fell from 10.2 percent in 1995 to 1.6 percent in 1999 to then reach an all-time high of 20.7 percent in 2006. Copper revenue reached its peak in 2006, which was not only due to soaring copper prices, but also to the implementation of a new royalty law which requires all foreign firms to pay a 0.5 to 5 percent royalty depending on annual sales.
30 Moreover, a lenient regime through which certain large mining operations qualified for significant tax deductions was ended.
Source: Graph made by author using data from the Central Bank of Chile
Over half of fiscal revenue from copper comes in the form of foreign currency, as copper exports are invoiced in foreign currency and two thirds of copper production is owned by foreign companies (Meller and Simpasa, 2011). Converting all of this foreign currency into national currency puts huge appreciation pressures on the Chilean peso. Seen the magnitude of government revenue generated by cupriferous activity, Chile’s economy is prone to Dutch disease. However, the empirical tests effectuated in the previous section have shown that Chile’s exchange rate underwent only mild appreciation as a result of the copper prices and did not experience the crowding out effect, nor indirect de-industrialisation. The following section will set out the policy tools that have played an important role in the mitigation of Dutch disease.
3.2 Implemented policy tools and their effect on Dutch disease
Chile’s policy framework with regard to the copper sector can be divided into four phases. The first phase is from before the military Coup d’état, when Chile’s industrial policy was mainly focused around import substitution, a policy that subsequently has been abandoned. From 1973 to around 1989 many reforms were put in place to achieve trade liberalisation
31 and an open market. A very important step in this was freeing the Chilean peso from its peg against the dollar, which started with a devaluation of the peso in 1982. The peso was officially allowed to float freely in 1999. The 1990s to the early years of the new millennium revolved around social and economic development, spurred by the return to democracy. These reforms constitute the baseline of Chile’s defence against Dutch disease. Since their implementation, Chile has centred its attention on legal reforms ensuring macroeconomic stability and good management of copper revenues, and on strategies aimed at innovation and competitiveness, the details of which will be discussed next.
3.2.1 Fiscal policy
A chief instrument the government can use to deal with the negative impacts of the copper boom is fiscal policy, which it can use to make the increase in wealth permanent, to limit the spending effect and to smooth expenditures, the latter of which’s strong fluctuations can impact the exchange rate’s volatility.
In 2001 the Ministry of Finance established the structural fiscal surplus rule, according to which the Treasury has to generate a structural surplus of one percent of GDP, using long term copper price as a point of reference for future revenue. The value of this parameter is established by a committee of experts who forecast the average copper price for the next ten years, which is then used to estimate the structural revenue from copper (De Gregorio and Labbé, 2011). This panel has resisted the urge to believe in the permanence of price increase and adhered to the idea that they revert back to their long run equilibrium (Frankel, 2011). The aim of this rule is to achieve greater independence of fiscal expenditure with regard to short term copper price fluctuations.
Figure VII below compares fluctuations in general government expenditure as a percentage of GDP with copper prices from 1989 to 2014. Looking at this figure, it can be inferred that the government has been relatively successful at maintaining a stable level of expenditure as it has been kept between 16.42 and 22.68 percent of GDP in over 25 years. Another important feature to be seen in figure VII is the countercyclical behaviour of Chile’s government spending, which can also be attributed to the structural surplus rule. Because the structural surplus is contingent on future copper prices, i.e. expected revenue, the required government savings are high when current copper prices are high so as to compensate for the expected decrease in revenue. Consequently, when copper prices had more than quadrupled by 2006, public spending reached its all-time low. This was quickly reversed when prices decreased and the global recession set in. Such countercyclical fiscal policy is unique for a country so dependent on natural resources.
32
Source: Graph made by author using data from Oxford Economics and US Minerals
The importance of implementing a countercyclical relationship between copper price and government expenditure is due to the consequences of a copper bust. As greater income due to higher copper prices relaxes the fiscal liquidity constraint, this can lead to an increase of spending through direct use of resources that are generated by the commodity, increased borrowing by using savings from resources as collateral and a reduction in taxes so that fiscal revenue is more dependent on the revenue from the boom. Once the bust sets in all of these policies have to be reversed, reinforcing the recession the economy is likely to dip into. In using the fiscal rule government expenditure is isolated from copper prices, which thus avoids expansionist policies, extravagant use of credit and reduced tax during a boom, all of which are policies that cannot be sustained once the boom is over (Meller and Simpasa, 2011).
In 2006 the Fiscal Responsibility Law (FRL) was enacted, requiring the government to state how its policy changes will affect the structural budget balance thus embedding the surplus rule into the law. Hence, the FRL allows the structural surplus rule to be regarded as an instrument for institutionalising fiscal prudence. Moreover it is noteworthy that right after the implementation of the FRL, the REER dipped after 3 years of appreciation.
An important determinant for volatility in the exchange rate is volatility in government consumption through the spending effect, which will be much higher if closely related to natural resource revenue (Van der Ploeg and Poelhekke, 2009). Therefore important benefit
33 of consumption smoothing is its ability to help curb exchange rate volatility. Chile managed to smooth fiscal expenditure through the structural surplus rule as it detaches government expenditure from copper revenue. If government spending is detached from copper revenue this can smooth expenditure and indirectly the exchange rate. Moreover, the fiscal rule helps take away the uncertainty regarding fiscal policy, by providing a predictable path of government expenditures. Thanks to the fiscal rule which is strongly countercyclical most of the windfall was saved. From figure VI in section 3.2.2 below it can be seen how Chile maintained its expenditure levels though copper revenue was soaring. It is impossible to know the counterfactual exchange rate, but from the size of the budget surplus it is clear that a very large part of the increased natural resource income was set aside thus inevitably attenuating the undesired currency appreciation.
3.2.2. Revenue management
The implementation of the fiscal structural surplus rule leads to a significant accumulation of wealth during the years that copper prices are high. For the rule to be effective, windfall revenue needs to properly managed. During a copper boom, the significant accumulation of wealth through the fiscal rule requires a special body to administer and allocate the resources. To that end Chile has put in place sovereign wealth funds that are designed to manage this revenue. The two main funds that have been created in Chile are the Economic and Social Stabilisation Fund (FEES) and the Pension Reserve Fund (FRP). Although the latter has been a successful and important fund for Chile’s economy it is not relevant for this paper and will not be discussed further.
The predecessor of the FEES was the Stabilisation Fund, which was created in 1987 to stabilise foreign currency income generated by Codelco’s copper exports (Ruiz-Dana, 2007). By investing the surplus foreign resources abroad instead of converting it into pesos to spend domestically the fund aimed to alleviate upward pressure on the exchange rate. The Stabilisation Fund was replaced by the FEES in 2007 with an initial contribution of US dollar 2.58 billion. The FEES, seems to have played a key role in preventing the economy from experiencing negative effects after a commodity boom. It was created as an instrument to smooth government expenditure, which was achieved mainly thanks to the FRL, which ties the fund to the structural surplus target. The fund provides stability to government expenditure by saving revenue windfalls during a boom and financing of fiscal deficit during economic downturn. The economic downturn was right around the corner and the FEES was used for fiscal easing so as to protect the economy from the global recession in 2008-09 (Frankel, 2011), which can also be seen in figure VIII (section 3.2.1.) from the steep increase in government expenditure in the same period.
34
Source: Graph made by author using data from the Central Bank of Chile
The graph in figure VI shows the fluctuations in central government budget spending and its copper receipts as a percentage of GDP from 1996 to 2014. It can be seen that government expenditure has fluctuated little since the end of the previous century and instead remained between 10 and 13 percent of GDP. Furthermore, government spending has moved in opposite direction of the copper receipts: When copper revenues increased, government spending declined. In 2006, when government revenues from copper reached its peak at 5.4 percent of GDP, public spending actually declined to its record low of 9.99 percent. This illustrates the functionality of the surplus rule. By being tied to the structural surplus target the fund thus operated under an expenditure rule, instead of being based on contributions as was the case for the Stabilisation Fund which was allocated revenues in function of copper prices. In 2006 savings reached a record height of US dollar 11.29 million, or 7.9 percent of GDP (Ruiz-Dana, 2007). The resources of the fund are mostly invested in high grade corporate and government securities abroad (Ministry of Finance, Chile). This has led to a substantial improvement of Chile’s sovereign debt which rated BBB in 1992 and was rated A+ in 2007 as measured by Standard & Poor’s.
The FEES’ two main objectives are thus to insulate government expenditures from external shocks and ensuring its stability by savings during booms that are to be used in times of economic downturn and to control the impact of a boom on the exchange rate.
