The Macro-Management of Commodity Booms: Africa and Latin America’s responses to Asian Demand
Rolando Avendaño, Helmut Reisen* and Javier Santiso OECD Development Centre
Abstract
Strong growth in the Asian Drivers’ has benefited the terms of trade of raw-material exporting countries and attracted complementary finance. However, the long-term challenge for countries with fragile institutions is to avoid the resource curse. The Asian drivers are also contributing to heightened volatility of both commodity prices and proceeds, thus raising a particular challenge to macro management. The paper aims at informing policy choices in both Africa and Latin America. First, it highlights global macroeconomic links. Second, it provides brief literature summaries on the optimal policy response from a macroeconomic and development perspective. Third, empirical evidence is presented as for broad macroeconomic and fiscal policy responses to Dutch disease and specialization effects caused by Asian Drivers’ demand. Fourth, benefits and challenges due to the rise of the Asian Drivers are summarized from a macro perspective.
JEL Classification: F00; O11; 057; E62.
Keywords: commodity booms; Asian Drivers; developing countries, Dutch disease.
* Corresponding author: helmut.reisen@oecd.org
1
The Macro-Management of Commodity Booms: Africa and Latin America’s responses to Asian Demand
1. Introduction
China and India are rapidly integrating their huge labour forces into the world economy and are growing swiftly. Each year since 2001, their combined contribution to global output growth has been more than a third. Moreover, this contribution has helped to hold world growth above the 4 per cent threshold which is critical for improving the terms of trade for primary commodity producers. By investing huge foreign exchange reserves in US securities, Asian investors have also contributed both to low US interest rates and to higher raw material prices. This in turn helped to push down the spreads of most developing countries contributing to improve their access to private capital and lower the cost of capital for their firms.
All in all, commodity producers benefited from a higher global demand for their exports and from improved terms of trade. The Asian drivers’ desire for secure and ample access to natural resources necessary for their continued growth has attracted complementary finance, both from the Asian Drivers directly (e.g. infrastructure loans) and from global investors. All these factors have fuelled growth performances in Africa and Latin America. In parallel, China’s and India’s growing demand for commodities contributes to diversify export client portfolio, away from OECD countries. In a sense, therefore, African and Latin American commodity exporters have been enjoying an important windfall profit as a result of the Asian Driver related commodity booms.
Windfall profits imply challenges, not least for macroeconomic policy. By diverting resources from nonresource sectors and contributing to real exchange appreciation (the so-called Dutch Disease), the boom will tend to lock developing-country commodity exporters into the raw material corner, with little scope for industrial learning by doing and limited absorption of low-skilled labour. In order to avoid the Dutch Disease, resource-rich Africa and Latin America need to find ways to capitalize on windfall gains arising from resource extraction and promote job-rich sectors. Managed currency floats, lowering vulnerability through reduced short-term debt or higher foreign-exchange reserves and, above all, a countercyclical fiscal stance are generally held to constitute the basic tenets of an appropriate development-friendly macroeconomic response to raw material booms.
This papers aims at informing macroeconomic policy how to deal appropriately with Asian Driver (AD) related commodity induced booms by tracing the macro responses of most AD affected countries in Africa
2
and Latin America against a control group. The indicators observed are consumer price inflation indexes, real effective exchange rates, official reserves and short-term debt, fiscal response functions, and Dutch disease indicators.
2. The macroeconomic links
The sheer size of the Asian Drivers, their phenomenal rates of growth, their demand for natural resources, and their growing economic and political power re-shape the world economy. They provide both competition and opportunities across the board to major trading partners in OECD countries, to developing countries and to other emerging economies (see for Africa Goldstein et al., 2006, and for Latin America Santiso, ed., 2007 ). This section identifies the important conduits through which the rise of the Asian Drivers impacts on the macroeconomies of commodity-oriented countries in Africa and Latin America. These conduits are quite different from those that affect economies with important manufacturing sectors that are subject to increased competition from the East. The integration of the Asian giants into the world economy has dramatically changed the nature of global macroeconomic and financial interdependence (Reisen et al. 2004) which in turn, shapes primary commodity markets:
• Global output growth is a major determinant for primary commodity prices. A recent estimate finds that world commodity prices move pro-cyclically with the growth rate of world industrial production. This is recorded around 1.5 per cent for every one per cent increase in world industrial output, with a one-quarter lag at the most (Bloch et al. 2004). • If world industrial growth exceeds 4 per cent, the barter terms of trade of primary commodity to finished goods rise (Bloch et al. 2004). High global growth has recently halted and reversed the secular decline of raw commodity prices since World War II. At the same time, Kaplinsky (2006) shows that the greater China’s participation in global product markets, the more likely prices will fall. • Lower US interest rates (which closely govern variations in global key interest rates) have a generally positive impact, as higher output prospects and lower storage costs lead to higher raw material prices. • Likewise, a weakening of the US dollar will raise raw material prices, partly for the same reasons which were evoked for US interest rates, partly as a result of the fact that most commodity markets are dollar denominated.
3
Table 1: The Asian Drivers’ Contribution to Global Growth, 2000-2007
2000
2001
2002
2003
2004
2005
2006
2007
Global growth, % y/y
7.0
5.0
4.8
6.1
8.0
7.1
6.7
6.7
China
18.1
27.2
30.0
27.8
23.8
27.2
28.1
27.9
India
5.9
7.0
7.5
8.9
7.3
8.2
7.7
7.9
Source: Own calculation based on the IMF World Economic Outlook Database.
γc × Note: The contribution to world growth is calculated as
Yc Yw
Y Y γc × c +γ r × r Yw Yw India),
γ w the growth rate of world output, and Yc and Y
w
, where
γc
is the growth rate of the country (China,
are the country and the world output, respectively.
i.e. China’s (India’s) growth rate times China’s (India’s) percentage share in world output divided by the sum of China’s growth rate plus the growth rate of the rest of the world, each weighted by their respective share in world output. Calculations are done on a PPP basis.
What impact do the Asian Drivers have on these macroeconomic determinants of the price of raw materials? China’s and India’s contribution to global output growth is substantial. Since 2001, their combined part to global output growth has been around 35 per cent. This helped maintain global output growth far above the 4 per cent threshold which is necessary for improving the terms of trade for primary commodity producers (Goldstein et al. 2006).
The accumulation of foreign exchange reserves by the Asian Drivers and their investment in US Treasuries have contributed to lower US rates (the Asian bid). By end 2005, China and Hong Kong had accumulated more than one trillion dollars in foreign exchange reserves, of which 30 per cent were invested in US Treasury Bills. This in turn constituted more than an eighth of all outstanding US Treasury Bills. Since 2006, China has started to lower the share of reserves invested in US treasuries. India’s foreign exchange reserves were relatively lower at the end of 2005 and were more widely invested. Closely related to the Asian Driver induced commodity boom is the recycling of Petrodollars, estimated in early 2007 to exceed 1 trillion US dollars. Oil exporters became also net exporters of capital, contributing to lowering the cost of capital for developing countries. Investors looking to recycle their petrodollars or 4
commodity related dollars have also been engaged over the past years in the purchase of developing countries assets (see Lubin, 2007).
Africa – still connected to the world economy chiefly through raw material exports – and Latin America are benefiting from the China-driven ‘super cycle’ – a significant rise in real commodity prices,driven by the urbanization and industrialization of a major country- reinforced by India’s emergence. That said, commodity prices passed historical lows by the beginning of the decade to unusual figures in 2007. Figure 1 documents the price development for oil, industrial and precious metals as well as for soft commodities since 2001. These are the commodities that weigh most importantly in African and Latin American export baskets and impact positively on their overall growth performance. As estimated by Collier (2007), in both 2005 and 2006, the boom added nearly 2.5 percentage points to the growth of the typical African economy. As for the long-run growth effects of raw materials and the fiscal decision rules in the case of price booms, Collier and Goderis (2007) show the importance of distinguishing between non-agricultural (depletable, location bound) and agricultural (replenishable) resources.
Figure 1: Price index for selected commodities, January 2001 = 100
Source: OECD Development Centre and African Development Bank (2007), African Economic Outlook 2007, Paris, OECD
The benefits of China’s and India’s rising global demand (net imports) for commodities relevant to Africa and most of the Latin American countries may be attenuated by the volatility of demand on the part of the Asian giants. This is caused partly due to cyclical variations and to arbitrage between domestic production 5
and imports. Moreover, as a large share of manufacturing exports from China are produced by multinational corporations, high demand for raw materials partially reflects relocation of raw material demand from production sites elsewhere. Such adjustments do not occur without friction, which in turn could have fuelled demand volatility.
Goldstein et al. (2006) compare the volatility (measured as standard deviation around the trend) in commodities relevant to Africa for two time periods. Volatility rose for oil, cotton and industrial minerals except copper. Although it is difficult to separate the relative contribution of different factors, increased volatility between 2000 and 2004 may have been partly due to the fact that China and India are swing producers – exporting when prices are high and stockpiling when (be it for cyclical or exceptional reasons) they are not as attractive. Given their large economies, any behavioural change is likely to translate into volatility in world prices.
Another source of volatile foreign exchange inflows that macro management has to handle are the commodity-related capital inflows. FDI, portfolio equity flows and project loans have been on the rise for developing countries in recent years. Since the early 2000s, that is since the global re-emergence of China, foreign direct investments in Africa, in particular, have been on the rise, multiplied by three over the period. According to the African Economic Outlook 2007, the inward flows of FDI in Africa jumped from less than 10 billion of dollars in 2000 to more than 30 billion in 2005 (the total cumulated inflows over the period pass the threshold of 100 billion of dollars for all the continent). The annual flow of FDI going from China to Africa has tripled since the beginning of the 2000s (Mßhlberger, 2007). The Asian Drivers have been new actors, entering with capitals into African markets. Not only China increased its foreign direct investments in Africa but it also multiplied loans and trade credit lines all around the continent (Goldstein et al., 2006). These are particularly pronounced for countries for which oil and gas reserves provide an important share of export proceeds. Natural resources, in particular oil, and the required network infrastructure, are the destinations for Chinese project lending. Annual statistics published by China’s Ministry of Commerce indicate that new contractual commitments to Sub-Saharan Africa tripled from just under US$2 billion to just over US$6 billion in 2005. In November 2006, during the first SinoAfrican Summit, China strengthened its cooperation framework with the continent. The event came one year after China announced 10 billion in concessional loans to Africa for the period 2006-2008.
6
3. The macroeconomic policy challenge
The macroeconomic policy challenges posed by the Asian drivers relate to monetary policy including the choice of the currency regime, the counter-cyclical stance of fiscal policy (both overall budget deficits and government spending) as well as inter-temporal public spending decisions including reserve and asset management. Since the government is the conduit for the mineral revenues to the rest of the economy, fiscal policy is the key to managing booms (Gottschalk and Prates, 2007). Informed judgments about the persistence and shape of the Asian Driver induced commodity boom will be fundamental to macroeconomic policy formulation. Goldstein et al. (2006) argue that prospective growth patterns of the Asian Drivers, in particular of China, can be expected to move from industrialization, exports and urbanization towards domestic consumption, globalise eating habits and services, implying a shift in their raw commodity demand patterns, away from iron ore, copper and other industrial minerals toward soft commodities.
Sub-Saharan African countries, given their dependence on commodity exports, account for half the commodity-currency countries. Moreover, for these 22 countries, over 80 percent of the variation in the real exchange rate can, on average, be accounted for by movements in real commodity prices alone—a surprisingly strong result (Cashin et al., 2003). While the variability of the real effective exchange rate is similar across the various nominal exchange rate regimes, for countries with pegged nominal regimes, a larger relative share of real exchange rate variability is driven by the variability of relative prices. Theory tells us that, for economies prone to frequent real external shocks, flexible nominal exchange rates facilitate the smoothing of real output to such shocks, especially when domestic wages and prices are slow to change (e.g., Chen and Rogoff, 2003). Following a positive real shock (such as a rise in the world price of a key export), a nominal appreciation lowers the domestic price of exported goods (to partially offset the rise in the international price) and reduces real wages in line with reduced labour demand. In contrast, in countries with inflexible nominal exchange rates experiencing such positive real shocks, prices and wages need to rise to ensure that employment and output are in equilibrium.
Most low-income countries have preferred managed floats (unsterilized intervention on the foreign exchange markets to target the real appreciation needed to accommodate the commodity boom) to either a pure float or nominal exchange-rate pegs, for good reason (Buffie et al., 2006): •
With a commodity boom (just as with a surge in aid), a pure float results in nominal appreciation and, with sticky prices and wages as well as with complementary capital flows, in exchange-rate
7
overshooting up to a point where output contracts in the nontradables sector and where substitution effects drive the economy in a recession. •
A currency peg allows a short-run spike in inflation; while bond sterilization can dampen the spike, it does so at the cost of rising real interest rates, inciting further capital inflows. The sterilization may also require bond sales on a scale that exceed the absorptive capacity of the shallow domestic financial markets.
With countries’ increasing international financial integration, considerations regarding reserve adequacy have shifted from an emphasis on trade (traditionally associated with the “three-months-of-imports” rule) to the financial account and balance sheet fragilities (associated with the “Greenspan-Guidotti” rule, according to which reserves should cover short-term debt). The rise of international reserves in developing countries opened a new issue related to the optimal level and management of this commodity related windfall. Since the classical Baumol-Tobin inventory model with fixed costs of depleting and replenishing reserves (see Flood and Marion, 2002, for a review), the literature has focused on the optimal level that developing countries should have (Jeanne and Rancière, 2006). Reserves tend to allow a country to smooth domestic absorption in response to sudden stops but they also yield a lower return than the interest rate on the country’s long term debt. The optimal answer is not however evident as there is not only the related financial costs of holding reserves but also social costs (Rodrik, 2006).
Whatever the exchange-rate regime, the basic requirement for avoiding macroeconomic complications with free capital flows is fiscal control. Since in general the government is the conduit for the mineral revenues to the rest of the economy, fiscal policy is the key to managing booms. Unless the government commands fiscal control for stabilization purposes, it has to violate the Mundell assignment and use monetary policy for internal balance. According to Mundell (1962), however, once the capital account is open, even imperfectly, monetary policy acquires a comparative advantage in dealing with external imbalances, while fiscal policy is assigned to maintaining internal balance.
Households where goverments are fiscally weak, and in view of a commodity boom, force an unstable assignment upon the authorities. If fiscal policy does not tighten sufficiently to cope with the boom, monetary policy is left to hold down aggregate demand. This in turn would widen the interest differential and thus reinforce the balance of payments surplus. Confronted with a commodity boom, fiscal policy should be used to reduce demand for non-tradables, hence limiting unwarranted exchange-rate appreciation. The aim is to eliminate instability in aggregate demand, and consequently the real exchange rate, by smoothing expenditure over time, which implies self-insuring against revenue downfalls. Next to
8
expenditure restraint, revenue management is important, consisting of self-insurance and asset diversification. The ability to maintain expenditure during the bust depends on having been prudent during the boom.
Table 2: Managing Public Sector Commodity Booms
Decision
Rule
How much to deplete? Arbitrage: The country should be indifferent
1. Hotelling/Solow Rule. This requires that the
between keeping the natural resource under the
price of natural resource should grow at the
ground in which case the return is the capital gain
world rate of interest and that under some
on the reserves compared to selling it and getting
conditions the rate of depletion should equal
a market rate of return on it.
the demand elasticity times the world rate of interest. 2.
The steady-state depletion rate is capital return minus population growth, so that societies with fast growing populations should deplete their natural resources less rapidly.
How much to save? To maximise intergenerational HH utility, which
1. Hartwick Rule. If there is no population
saving rate sustains a stable consumption per
growth, to sustain a constant income per capita
capita.
all resource rents must be invested in capital,
Consuming
rents
from
exhaustible
including education. If consumption per head
resources is literally consuming capital.
were rising (falling) over time, social welfare The mid-term saving decision is ruled by
could be increased if earlier (later) generations
stabilisation and diversification concerns. Fiscal
saved and invested less or consume capital at
policy is superior to monetary policy to deal with
the expense of later (earlier) generations.
the first, active diversification involves use of
2. Commodity price smoothing rule. Unlike the
funds for new activities (see Norway & Chile).
savings generated by the Hartwick rule, these
9
savings are intended to finance subsequent consumption during periods when the oil price is below its long run path. There is thus a strong case for holding these assets in liquid form, which implies the acquisition of financial assets abroad.
How much to invest at home? 1. Excess return of home investment 2. Construction price smoothing rule
How much to invest abroad vs. retire public 1. Excess cost of public debt over global return
debt?
Source: Based on discussion in van der Ploeg (2006) and Collier (2007).
Economic theory offers useful insights into the optimal management of natural resources. One strand of literature focuses on the Hotelling rule. The rationale behind this rule is arbitrage. The country should be indifferent between keeping the natural resource under the ground in which case the return is the capital gain on the reserves and selling it and getting a market rate of return on it. This requires that the price of natural resource should grow at the world rate of interest and that under some conditions the rate of depletion should equal the demand elasticity times the world rate of interest (van der Ploeg, 2006). Another line of theoretical research investigates the optimality of the Hartwick rule, which demands that the proceeds of natural resource revenues are reinvested in productive assets. The World Bank has calculated that many resource abundant economies do not follow the Hartwick rule. In fact, many of these countries have negative genuine saving rates and become poorer each year (World Bank, 2006). Asset accumulation should not be confined to the central bank, but should cover the public sector as well. There are good reasons to save part of the commodity bonanza. How much? In the case of depleteable resources, the long run saving rule can be applied: the expected rate of increase in the commodity price, divided by the extraction rate. If, for purposes of illustration, the expected price increase is one percent annually and the extraction rate is three percent (sufficient reserves for 33 years of extraction at the present rate), then one third of extraction proceeds can be consumed. However, if the judgment is that the current 10
commodity price is way above the long run level, then the above-normal proceeds should be saved entirely.
There are also good theoretical reasons for investing a substantial part of the windfall initially abroad: the return to investment would fall below the world interest rate if the windfall were to be used entirely for domestic investment. Investing abroad offers an escape from diminishing returns: foreign assets can be repatriated gradually and used for domestic investment. The construction price smoothing rule can be employed to dampen rising capital cost, such as typically occur in a construction boom, by deferring domestic investment until the construction boom abates. However, in practice the efficient balance between domestic and foreign assets is politically difficult to sustain. Domestic debt repayment may solve this dilemma, and is lucrative as long as domestic debt cost exceeds expected foreign returns. It has the added advantage of making foreign asset accumulation difficult to reverse by future predator governments.
To be sure, the management of commodity price booms in low-income countries goes beyond the standard textbook prescription that might hold for OECD minerals-dependent countries such as Australia or Canada. In all commodity exporters, the short-run growth effect of higher commodity prices is positive. In lowincome countries, the long-run effect of higher non-agricultural commodity prices has been shown to be negative, while it has been positive for agricultural exporters, in a study covering the global experience of primary commodity exporting countries over the period 1960-2004 (Collier and Goderis, 2007). This could suggest a fundamental difference on the macroeconomic policies regarding hard and soft commodity exporters. Three factors, which should be integrated in any macroeconomic policy prescription, have been observed to generate the adverse long-run effect:
(a) Dutch Disease
A surge in resource exports leads to a real appreciation of the county’s exchange rate and this hurts other exporters and producers in import-competing sectors. This phenomenon is variously known as the “Dutch disease” (Corden and Neary, 1982), “the Gregory effect” and “de-industrialization”. A resource boom affects the economy through the resource movement effect and through the spending effect. For Dutch Disease to arise and become a serious policy issue, there must be other sectors for which the rise in the real exchange rate would create problems relating to competitiveness. Torvik (2001) has shown that the conventional Dutch disease effects may be overturned if there are productivity spillovers in both tradable
11
and non-tradable sectors. Adam and Bevan (2006) examine the case where public infrastructure investment generates an inter-temporal productivity spillover for both tradable and non-tradable production, but in a potentially unbalanced manner. For example public investment in rural roads is likely to impact more on the production of (non-tradable) food crops than on urban-based (tradable) manufactures and vice versa for, say, telecommunications infrastructure. Collier and Goderis (2007) find that Dutch disease, although significant, can only explain a minor part of the long-run negative growth effect of higher non-agricultural commodity prices on long-run growth. (b) The Leamer Triangle
The concept of the Leamer Triangle may therefore be more relevant when it comes to studying the impacts of both the resource boom and commodity price volatility that the Asian Drivers tend to exert on resourcerich economies. Leamer (1987) has shown using a three-factor multi-good model that resource-rich countries can take a development path very different from resource-poor countries. The corners of the Leamer Triangle represent three factors of production: labour, natural resources, and physical as well as human capital. A natural resource discovery, for instance, swings a country’s endowment point directly toward the resource corner. This can be used to contrast the development path taken by resource poor countries with the development path taken by resource rich countries. The Leamer analysis points to four problems connected to a resource boom in raw-material rich economies (see also Leamer et al.,1999; Alvarez and Fuentes, 2006): i) The absorption of low-skilled labour that goes along with the development of manufactured goods is foregone, hence inequality is deepened; ii) Those manufacturing activities that do emerge are capital intensive and skill intensive; iii) Human capital accumulation may be impeded, as skills in the resource sector are very specific and spillovers limited; iv) Volatility in the prices of raw commodities may raise capital risk in resource-dependent, undiversified countries, which might deter investment and make it more difficult for other tradable activities to emerge.
(c) Volatility
Volatility is an important dimension when explaining growth paths in developing countries. Those where fundamentals have experienced significant variations (e.g. Brazil or Mexico in the 1990s) have experienced strong growth fluctuations, and the negative correlation between both variables has been stressed elsewhere (see Hnatkovska and Loayza 2004). Aghion and Marinescu (2006) emphasise the effect of fiscal policy, where a counter cyclical budgetary policy fosters innovation and growth by easing the
12
impact of a shock on innovating-firms. Consistently, they find that a procyclical public debt growth is negatively correlated with growth. Therefore, a more countercyclical public debt growth tends to enhance growth, even more when there is low financial development. In practice, in many developing countries fiscal policy tends to display the opposite properties: it is procyclical (Hausmann and Gavin, 1996). In particular, government spending as a share of GDP goes up during booms and down in recessions, while deficits increase in booms and decrease in recessions. In some commodity-exporting countries, not all, the government acts as “trustee” of the resources for the country and is an important recipient of the mineral rents. A political explanation particularly relevant for commodity-dependent countries has been advanced by Tornell and Lane (1999): when more resources are available (i.e. in booms), the common pool problem is more severe and the fight over common resources intensifies, leading to budget deficits. Again, this problem seems less relevant for agriculture-based economies as these do not normally produce rents (the surplus of export revenue over the cost of production); by contrast, non-agricultural commodities provide persistent location-specific rents, the bulk of which accrues to the government (Collier, 2007).
3. Some recent policy evidence
This section will focus on the period 2000-2005, where we examine to what extent authorities in Africa and Latin America followed the policy script outlined in section 3 and how this helped commodity-boom economies to contain inflationary pressures, escape excessive specialisation effects, avoid being ‘Leamercornered’ and to reduce vulnerabilities to possible future currency attacks.
An approach for assessing the impact of the Asian Drivers on the regions studied consists on decomposing the demand effect and the price effect on different commodities markets. On this basis, a selection and control groups are defined in each region, categorizing countries by these two impacts. Regarding the first effect, we consider for the selection group those countries where the Asian Drivers weigh particularly strong in terms of both a country’s share in export receipts and Gross Domestic Product (GDP). Those countries were chosen that displayed above-region median values for both the AD share in exports.
For the second effect, we propose a different technique, regarding at the increase on export commodity prices on both Latin America and Africa. Our hypothesis is that the rise on the Asian Drivers’ demand for commodities has boosted prices, creating a heterogeneous effect, depending on both trade structure and commercial partners for each country. We follow a similar methodology to the one by Kamin et al.
13
(2006)1. In this case, we assess the impact of commodity imports into global prices, to estimate the individual contribution of the AD boom. Data is obtained from the UN/Comtrade database, SITC Revision 3 classification. Initially the estimated equation is as follows:
ΔPi , t , main _ com = α 0 + α1 * import _ ADt , main _ com + α 2 * import _ ADt − n , main _ com + υit where ΔPt , main _ com represents the percentage change on the price of the main commodity export for country i to the rest of the world, import _ ADt ,main _ com is the import share for the commodity by the Asian Drivers at time t and import _ ADt −n ,main _ com is their initial import share. To determine the price effect, we regard the magnitude on estimator α1 .2 Results from the sampling are summarized in Table 3.
We run the discussion along the rising degree of endogeneity of macroeconomic variables. This means that we start out with estimating fiscal response functions for the various country groups, both for government spending and government deficits. Then we look at the Greenspan-Guidotti indicators as an indicator for the effective currency regimes employed and lower exposure to short-term debt achieved. This will lead to the discussion of inflation trends and developments in the real effective exchange rates.
1
See Kamin, S. Marazzi, M. and John W. Schindler. “The impact of Chinese Exports on Global Import Prices”. Review of International Economics, 14(2), 179-201, 2006. 2
The dominant commodities per country in Africa were mainly gold (Burundi, Tanzania), textile fibres (Benin), petroleum (Egypt, Cameroon, Sudan, Senegal), non-ferrous metals (South Africa, Zambia), coffee (Ethiopia), apparel/clothing (Tunisia, Morocco) and tobacco (Zimbawe). For Latin America the predominant commodities were petroleum (Venezuela, Colombia, Ecuador), copper (Chile), gas (Bolivia), metalliferous ores and metal scrap (Peru), and coffee, tea, cocoa and other grains (Honduras, Nicaragua, El Salvador).
14
Table 3. Selection and Control Groups (Latin America and Africa) - each region’s median values in bold Latin America Exports to Asian Drivers/Total Exports (Avg. 2003-05)
Chile Peru Argentina Brazil Uruguay Paraguay Costa Rica Panama
0.115 0.099 0.097 0.068 0.041 0.029 0.027 0.024
0.040 0.021 0.012 0.010 0.006 0.006 0.009 0.002
L. America
0.024
Colombia Bolivia Mexico Honduras Venezuela Guatemala Ecuador Nicaragua
0.009 0.009 0.007 0.007 0.006 0.006 0.005 0.004
Country
Selection Groups
Control Groups
Africa Memo: Exports to Asian Drivers/GDP (Avg 200305)
Exports to Asian Drivers/Total Exports (Avg. 2003-05)
Memo: Exports to Asian Drivers/GDP (Avg 200305)
Benin Gabon Senegal Nigeria Tanzania Egypt South Africa Mali Morocco Zambia Cameroon Madagascar
0.382 0.150 0.141 0.105 0.098 0.078 0.045 0.042 0.042 0.039 0.037 0.030
0.041 0.019 0.035 0.017 0.011 0.003 0.012 0.009 0.010 0.014 0.008 0.002
0.002
Africa
0.030
0.007
0.002 0.002 0.002 0.002 0.002 0.001 0.002 0.001
Cote d'Ivoire Mozambique Kenya Ghana Tunisia Malawi Mauritius Uganda Algeria Niger Burkina Faso Botswana
0.027 0.026 0.020 0.016 0.010 0.010 0.009 0.008 0.007 0.003 0.000 0.000
0.017 0.007 0.002 0.006 0.004 0.003 0.003 0.001 0.002 0.000 0.000 0.000
Country
4.1. Government Budget Response Function
Fiscal response functions are estimated for both government spending (as percentage of GDP) and for government budget deficits or surpluses (as percentage of GDP). We distinguish two different estimation periods: First, we will assess the degree of procyclicality for a set of Latin American and African countries during the period 1987-1999. Second, we will focus on the recent upsurge of commodity prices and the terms of trade for these countries, and assess their impact on public revenue management during 2000 2005. The equation is estimated for the whole sample of countries, and for the selection and control 15
groups. The fixed-effects estimation is supported by a Hausman test on each sample. The estimation procedure follows Alesina and Tabellini (2005) and Jimenez and Tromben (2006).
The first panel regression is defined in the following equation:
ΔFit = β 0 + β1 * output _ gapit + β 2 * Fit −1 + β 3 * TOTit + β 4 * Z it + ε it
where Fit is an indicator of fiscal policy (in this case government expenditure, expressed as a percentage of GDP). The output gap is a measure of the business cycle, TOTit is are the terms of trade, and Z it is a set of variable controls. Information on government expenditure and some controls for this period is obtained on the World Development Indicators (World Bank) and the International Financial Statistics (IMF). The output gap is calculated as the log deviation of real GDP from its Hodrik-Prescott trend. Identically,
TOTit is a deviation between the terms of trade and its Hodrik-Prescott filtered trend, also obtained from the World Development Indicators. In this equation, a positive value for the output gap coefficient ( β1 ) is assumed to denote a procyclical behaviour. The measure compares the actual GDP (output) of an economy and the potential GDP (efficient output). When the economy is running an output gap, either positive or negative, it is thought to be running at an inefficient rate as the economy is either overworking or underworking its resources. Economic theory suggests that positive output gap will lead to inflation as production and labour costs rise.
For countries with full fiscal control, the null hypothesis is that government spending will neither be affected by varying output gaps, terms of trade or AD shares in exports; the only objective, to smooth government spending over different states of nature, is fully respected. Countries without fiscal control and hence pro-cyclical (due to rent seeking, fragile institutions, etc.), by contrast, should be expected to show significant positive correlation coefficients for government spending and varying output gaps as well as terms of trade. It would be unsurprising to find such a positive association as well for government expenditures and AD export shares as many would argue that the Asian Drivers tend to engage particularly in countries with weak governance scores.
16
Table 4a. Government Expenditure Response - All countries -
Regression Government Expenditure All Countries 1987-1999 Latin America Africa OECD output_gap 4.7818e-11* 5.03E-11 2.53E-13 [1.92] [0.27] [0.17] lag_gov_exp 0.7562*** 0.4753*** 0.7842*** [16.73] [9.46] [16.10] terms_trade 8.0764e-13* -2.7668e-12*** 2.48E-13 [1.79] [2.70] [1.36] Observations 207 318 195 Number of id_gen 16 25 15 R-squared 0.6 0.28 0.65 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
2000-2005 Latin America Africa OECD 2.53E-12 -5.8470e-10*** -3.02E-12 [0.09] [2.98] [1.46] 0.3450* 0.5717*** 0.6558*** [1.93] [6.75] [6.39] -1.85E-13 2.05E-13 -4.69E-14 [0.89] [0.32] [0.23] 89 144 72 16 25 15 0.07 0.31 0.47
In this estimation a positive coefficient for the output gap denotes procyclicality. The regression shows that Latin America had a significant procyclical behaviour (at 10% of significance) during the period 19871999, which was not the case during the period 2000 – 2005, indicating increased fiscal control. For the case of Africa results are not conclusive, but seems to suggest a high degree of volatility in public spending. In the OECD group, results indicate fiscal control for both sub-periods as the output gap does not exert a significant impact on government expenditure. The following regressions takes into account the selection (4b) and control (4c) groups in each region in order to find further differences. Table 4b confirms the existence of procyclicality for the Latin American selection group during the period 1987-1999. During the second period the null hypothesis of nonsignificance for the output gap cannot be rejected, and therefore it is assumed that no-cyclicality prevailed in the region in after 2000. While the signs of the output gap turn from positive to negative in the African selection group (indicating a move toward an anti-cyclical stance), they are not significant in both subperiods. By contrast, the African control group displays a surprisingly significant anti-cyclical response of public spending in both sub-periods and for both explanatory variables, i.e. the output gap and the terms of trade.
17
Table 4b. Government Expenditure Response - Selection Group -
Regression Government Expenditure Selection Group 1987-1999 2000-2005 Latin America Africa Latin America Africa output_gap 3.8198e-11** 2.05E-10 1.0716E-11 -2.14E-10 [2.25] [1.06] [0.24] [1.29] lag_gov_exp 0.7746*** 0.4904*** 0.6225* 0.2270*** [13.30] [7.33] [1.79] [2.71] terms_trade -4.2171E-13 -2.6502E-12 -1.8463E-13 3.3567E-12 [0.51] [1.02] [0.41] [0.78] Observations 103 151 45 66 Number of id_gen 8 12 8 12 R-squared 0.68 0.3 0.1 0.14 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Table 4c. Government Expenditure Response - Control Group -
Regression Government Expenditure Control Group 1987-1999 Latin America Africa OECD output_gap 7.2943E-11 -1.1254e-09** 2.53E-13 [1.18] [2.07] [0.17] lag_gov_exp 0.7533*** 0.4353*** 0.7842*** [11.48] [5.76] [16.10] terms_trade 9.5359E-13 -2.9224e-12** 2.48E-13 [1.55] [2.54] [1.36] Observations 104 167 195 Number of id_gen 8 13 15 R-squared 0.59 0.29 0.65 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
2000-2005 Latin America Africa -1.0326E-11 -1.5932e-09*** [0.43] [4.00] 0.11 1.0084*** [0.83] [8.04] -2.0351E-13 9E-14 [1.40] [0.15] 44 78 8 13 0.12 0.58
OECD -3.02E-12 [1.46] 0.6558*** [6.39] -4.69E-14 [0.23] 72 15 0.47
To confirm the robustness of results, a second set of regression was estimated for the different samples, using the growth rate of government expenditure. As they neither added nor modified the results reported here, they have not been reproduced here.
Next we took government budget balances as a proxy (dependent variable) for fiscal policy . Data on government budget balances, expressed as a percentage of GDP, were obtained from The Economist Intelligence Unit, the OECD African Economic Outlook and Jimenez and Tromben (2006). Results are reported in the following tables.
18
Table 5a. Government Budget Balance Response - All Countries -
Regression Budget Balance All Countries 1987-1999 Latin America Africa OECD output_gap 5.45E-11 -3.62E-10 9.14E-13 [1.49] [0.60] [0.22] lag_budg_bal 0.28*** 0.1232 0.85*** [3.71] [1.31] [16.75] terms_trade 1.00e-12** 4.70e-11*** 8.02E-13 [2.42] [3.23] [1.30] Observations 146 140 167 Number of id_gen 15 23 15 R-squared 0.19 0.13 0.66 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Latin America 2.76E-11 [1.10] 0.45*** [4.07] 2.84E-13 [1.39] 90 15 0.27
2000-2005 Africa 1.05e-09*** [3.57] 0.08 [0.97] 2.23E-14 [0.02] 138 23 0.11
Table 5b. Government Budget Balance Response - Selection Group -
Regression Budget Balance Selection Group 1987-1999 2000-2005 Latin America Africa Latin America Africa output_gap 5.76E-11 -1.16E-09 1.65E-11 1.09e-09*** [1.19] [1.40] [0.60] [3.63] lag_budg_bal 0.24** 0.1 0.42*** 0.11 [2.12] [0.67] [3.06] [1.05] terms_trade 2.08e-12** 4.45e-11** 9.95e-13*** -1.95e-11*** [2.34] [2.48] [2.90] [2.87] Observations 75 58 48 72 Number of id_gen 8 12 8 12 R-squared 0.15 0.21 0.42 0.27 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
19
OECD 1.70e-11** [2.33] 0.33** [2.49] 1.19E-12 [1.37] 84 14 0.23
Table 5c. Government Budget Balance Response - Control Group -
Regression Budget Balance Control Group 1987-1999 Latin America Africa OECD output_gap 5.11E-11 2.2330e-09** 9.14E-13 [0.94] [2.28] [0.22] lag_budg_bal 0.29*** 0.08 0.85*** [2.78] [0.73] [16.75] terms_trade 8.51E-13 1.7810e-10*** 8.02E-13 [1.55] [2.75] [1.30] Observations 71 82 167 Number of id_gen 7 11 15 R-squared 0.21 0.15 0.66 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Latin America 4.93E-11 [1.04] 0.44** [2.55] 1.53E-14 [0.06] 42 7 0.22
2000-2005 Africa 1.20e-09* [1.79] 0.15 [1.16] 2.73E-13 [0.26] 66 11 0.08
OECD 1.70e-11** [2.33] 0.33** [2.49] 1.19E-12 [1.37] 84 14 0.23
The null hypothesis as for government budget balances is for countries with fiscal control to find a significant positive correlation coefficient between government budget balances and varying output gaps as well as terms of trade. To save for bad days in good times is a precondition for governments to be able to smooth public spending over different states of natures. Absence of a significant positive correlation would indicate pro-cyclical fiscal stance and hence more macroeconomic complication as a result of the AD induced commodity booms, such as higher price levels for non-tradables, thus higher inflation levels, and higher real currency appreciation than warranted by the fundamental equilibrium exchange rate.
During the period 1987-99, the terms of trade explain significantly in both Africa and Latin America the government budget balance, which rise when the terms of trade do so. In the second sub-period the termsof-trade effect on budget balances (Table 5a) has turned insignificant in both regions, suggesting less direct dependence of government finance on raw materials. However, in the AD selection group (Table 5b), the terms-of-trade effect continues to impact significantly on government budget balances during 2000 – 2005.
In both Africa and the OECD sample the output gap starts to significantly explain budget balances since 2000, suggesting an increased anti-cyclical fiscal stance. In Latin America, no statistical significance can be detected for the output gap variable.
These findings may, perhaps surprisingly, suggest that in the recent years when the AD driven commodity boom gained full traction, macroeconomic policy challenges have been better coped with and monetary
20
policy better assisted by an anti-cyclical fiscal stance in Africa than this has been the case on average in Latin America.
4.2. Respecting the Guidotti-Greenspan Rule: Higher Reserves, Lower Debt
A rise in commodity proceeds should neither affect the level of foreign exchange reserves nor the level of short-term debt if this rise was fully accommodated by a pure float in the exchange rate. Just the nominal exchange rate would appreciate, immediately on impact. Any other currency regime, from a hard peg to dirty float, will show up in a rise of the Guidotti-Greenspan indicator, the level of official foreign exchange reserves as a fraction of the country’s short-term foreign debt. For those countries where this indicator had been below one, any rise above one can be interpreted as a step toward lowered exposure to a speculative currency attack by foreign and domestic investors as the assets easily cancelled fall short of the reserves to defend the exchange rate.
Figure 2, based on data from the World Bank Global Development Finance Database, related official foreign exchange reserves and short-term external debt. (Ideally, the indicator could be broadened to include all liquid assets held abroad instead of official reserves only and all liquid liabilities, both external and domestic; however, data availability precludes to establish time series as presented in Figure 2). The Guidotti-Greenspan rule indicator improved and now exceeds one in all sample countries. This leads to the observation that the Asian driver induced commodity boom served to reduce vulnerability to future speculative attacks. Especially Argentina, Cameroon and Gabon managed to climb out of the danger zone, from very low levels below one, to close to one or above while the rest of the selection group countries shows even more comfortable values.
21
Figure 2: Ratio Foreign Exchange Reserves to Short Term Debt – Selection Group
100
100
10
10
2005
2004
1995
2005
2004
2003
2002
2001
2000
1999
1998
0.001 1997
0.001 1996
0.01
2003
Zambia
0.01
1995
Tanzania
2002
Madagascar
0.1
2001
Egypt
South Africa
2000
0.1
Gabon
1999
Mali
1
1998
Benin
1997
1
Nigeria
1996
Cameroon
% (Logarithmic Scale)
% (Logarithmic Scale)
Africa
Latin America
10
Argentina Brazil 1
Chile Costa Rica
Panama Peru 1
Uruguay Paraguay
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
0.1 1995
0.1
% (Logarithmic Scale)
% (Logarithmic Scale).
10
Source: OECD Development Centre, computed on the basis of World Bank Global Development Finance Database, 2007.
Changes in debt composition, maturities and structure have importantly contributed to improved Greenspan-Guidotti Rule indicators. Figure 3 displays some evidence for the reduced percentage share of
22
short-term domestic debt in total domestic debt. The risk debt management to the current commodity boom, partly driven by increasing Asian demand, has been therefore every different from the previous commodity shocks, in particular the one of the 1970s that ended in Latin America with the debt crisis of the 1980s (Blommestein and Santiso, 2007). Since the global re-emergence of China, exchange rateindexed debt also has been reduced all around in Latin America, the most impressive case being Brazil where the share of such indexed debt in total public debt fell from 37 per cent in 2002, the year of the crisis, to 2.3 per cent at the start of 2006. However, the reallocation towards more local currency debt is also inducing a change in the risk profile of sovereign issuers. Foreign currency debt is decreasing, although this meant in some countries that debt maturity became shorter (even if things are changing quickly as some other emerging bond issuers are starting to be able to issue bonds in local currencies with maturities now over ten years as for example Mexico).
Figure 3: Short Term Domestic Debt in Emerging Markets 70
% Domestic Debt .
60
1996
50
2004
40 30 20 10 0 Brazil
Mexico
Chile
Venezuela
India
Colombia
China
Russia
Source: Blommestein and Santiso, 2007.
At the same time, during the 2000s, current account surpluses in most emerging markets enabled several countries to reduce their external debt. They have also helped reduce one major source of vulnerability (to liquidity crisis in particular), the net open forward positions in hard currencies taken by central banks of some emerging countries which played a key role in the collapse of the Thai baht in 1997 and was regarded as a major source of vulnerability for South Africa until the trimming down of its forward book in 2003.
23
In some cases external debt levels have been reduced drawing on these foreign reserves. In 2005, Brazil repaid the IMF, the Paris Club creditor countries and in 2006 it paid off all its remaining Brady bonds ($ 6.6 billion), the securities that kick-started the emerging market bonds boom in the 1990s (albeit partly funded by new external debt), officially ending the debt restructuring process of the 1980s. Argentina followed also the Brazilian example, repaying its outstanding debt to the international financial institutions. In 2006, Nigeria became the first African country to cancel its Paris Club debt (totalling $ 30 billion; one third being repaid and the remaining being forgiven). These mechanisms of self-insurance through increased levels of reserves continue to be pursued even after repayments as underlined by the Brazilian and Argentinean examples.
4.3. Inflation and Real Effective Exchange rates
Figure 4 concentrates for each region – Africa and Latin America – on those selection countries for the period 2000 - 2006, where the commodity import demand from China is most felt. In general, the picture that emerges is one of macroeconomic stability, with inflation and real effective appreciation well contained.
Strong appreciation of the real effective exchange (REER) – exceeding 50 percent during that period - can be noted for Zambia in particular, partly reflecting not just higher commodity prices but also simultaneous debt relief and renewed capital inflows. According to Collier (2007), dramatic contrasting examples of the consequences of different public savings strategies for the real exchange rate are Chile and Zambia during 2005, a period when the world price of their common commodity export, copper, was exceptionally high. The Government of Chile followed a savings rule such that all the incremental revenue was saved, whereas the government of Zambia continued to run a fiscal deficit. During 2005, the real exchange rate mildly depreciated in Chile despite the boom, whereas in Zambia it appreciated by around 80%, causing intense problems for non-copper exports. Note, however, that the equilibrium exchange rate may have appreciated more in Zambia than in Chile as the former benefited from important debt relief measures. Therefore, the fundamental equilibrium exchange rate may have appreciated correspondingly; an appreciation (REER) should not be confounded with real effective overvaluation.
24
Figure 4: Real Effective Exchange Rates and CPI Inflation
Africa – Selected countries
Cameroon
4 3 2 1
Tanzania
140
30
160
120
25
100
20
140 120 100
20
40 20 0
10
25
2005
2003
0 2001
0 2005
0
2003
20
5
30
80 60
1999
10
40
40
1997
15
60
Inflation (right axis)
50
1995
80
2005
2003
2001
6
60
4
40
2
20
0
0
2005
2005
2003
Zambia REER
Inflation (right axis)
2001
2001
85
80
2003
90
8
100
2001
95
10
120
1999
100
Inflation (right axis)
140
1995
105
2005
2003
2001
0
REER
9 8 7 6 5 4 3 2 1 0 -1 1999
50
South Africa Inflation (right axis)
110
1997
100
1999
1997
1995
0
1997
REER
1995
150
2005
1995 -5
5
Senegal
200
1999
2005
100 0
90 88 86
2005
0
80 70 60 50 40 30 20 10 0
1999
2003
200
2003
94 92
2001
5
1999
6
1997
10
500 400 300
1995
600
Inflation (right axis)
250
Inflation (right axis)
40
2005
2003
2001
1999
1997
REER
Inflation (right axis)
7
Nigeria
1997
0
100 98 96
-10
1997
20
50
0 1995
40
102
10
1995
60
15
20
1995
80
800 700
30
REER
100
Morocco
REER
60
10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0
Inflation (right axis)
18 16 14 12 10 8 6 4 2 0
Mali
Inflation (right axis)
REER
120
0
Madagascar REER
8
2 1995
2005
2003
2001
1999
1997
1995
74 72 70
140
4
2001
76
REER
10
6
1999
82 80 78
1997
84
114 112 110 108 106 104 102 100 98 96 94
2003
16 14 12 10 8 6 4 2 0
86
Egypt
Inflation (right axis)
2001
REER
1997
Inflation (right axis)
1999
Benin REER
Latin America – Selected Countries
Chile
Brazil REER
Inflation (right axis)
120
30
120
70
120
100
25
100
60
100
50
80 60
0
0
2005
2003
2001
1999
1997
1995
Panama REER
Inflation (right axis)
20
115
1.6
120
110
1.4 1.2
100
105
15
100
10
95
1 0.8 0.6
90
Peru 120
12
100
12 10
60
8 6
40
2005
2003
80
20
4 2
0
0
6 4
85
2005
2003
2001
80
60 40
2
20
0
0
2005
90
80
2003
8
1999
14
45 40 35 30 25 20 15 10 5 0
2001
10
95
1997
16
Inflation (right axis)
1999
100
14
1997
105
1995
Inflation (right axis)
Uruguay REER
Inflation (right axis)
1995
REER
2001
0
1999
80
1997
2005
2003
2001
1999
1997
1995
0
85
0.4 0.2
1995
5
REER
1995
25
106 104 102 100 98 96 94 92 90 88 86 84
Paraguay
Inflation (right axis)
1997
Costa Rica REER
1995
0
2005
-5
2003
20
0
2001
10
1999
20
1997
0
1995
20
20
2005
40
40
2003
30
5
2005
10
40
40
60
2001
60
80
2003
15
9 8 7 6 5 4 3 2 1 0
2001
20
80
Inflation (right axis)
1999
REER
1997
Inflation (right axis)
1999
Argentina REER
Source: Authors, 2007; based on Economist Intelligence Unit and IMF Statistical Yearbook, 2007. Data on inflation for Africa from World Economic Outlook and Penn World Tables, 2007.
In none of the sample countries has inflation risen during the Asian Driver induced boom period. These findings suggest some degree of sterilized foreign exchange intervention and the absence of hard nominal exchange rate pegs, which would have had to accommodate the commodity-induced appreciation pressures through a rise in inflation. In Zambia, inflation had been exceeding 20 percent until 2004. This uncomfortably high level of inflation was also brought down through some exchange-rate based stabilization.
26
A priori, the AD export share can be expected to lower inflation as a result of nominal appreciation. This expected response will depend very much on the degree to which nominal currency appreciation is allowed to operate. To estimate the impact of the Asian Drivers on inflation, we estimate the following equation:
dev _ inf lat it = α 0 + α1 * exp ort _ ADit + β 2 * gov _ exp_ growth + ε it where the inflation deviation is defined as the difference between the average CPI-inflation during the period 1987-1999 and the CPI-inflation in year t, and export_AD represents the export share of country i to the Asian Drivers. Data on inflation was obtained from the Economist Intelligence Unit (based on the IFS Database in Datastream). Information on the export shares was collected through the World Integrated Trade Statistics (WITS) Database. The equation is estimated over the period 2000-2005, using a Hausman tested fixed-effect estimator for both samples. Results are as follows:
Table 6a. Inflation Deviation and Exports to Asian Drivers - All countries -
Inflation Deviation vs Export AD 2000-2005 All countries 2000-2005 Latin America Africa -4.91E+01 -3.19E+00 [0.55] [0.23] gov_exp_growth 0.21 0.06** [0.81] [2.19] Observations 9.10E+01 1.23E+02 Number of id_gen 16 23 R-squared 0.01 0.05 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1% export_ad
For the total country sample, there is a negative relationship between inflation deviation and the export share to the Asian drivers, although these results are not significant. However, they suggest that the recent increase of AD exports could have had the expected impact on the gap between the observed inflation and the average inflation before the Asian boom. The second explanatory variable, the growth of government expenditure, seems to be positively correlated with this deviation, being significant (at 5%) for the African case. This finding suggests that Africa would have experienced higher inflation in the absence of its general anti-cyclical fiscal stance observed in section 4.1.
27
Table 6b. Inflation Deviation and Exports to Asian Drivers - Selection Group -
Inflation Deviation vs Export AD 2000-2005 Selection Group 2000-2005 Latin America Africa export_ad -2.90E+01 -2.96E+00 [0.67] [0.29] gov_exp_growth 0.2 0.05** [1.07] [2.40] Observations 4.60E+01 6.40E+01 Number of id_gen 8 12 R-squared 0.04 0.1 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Results for the selection group are quite similar to the previous sample. Both coefficients for the export share are negative, but statistically non-significant. However, the government expenditure growth coefficient for Africa is consistently positive and significant, which suggests again an important role of fiscal policy in explaining inflation deviations. Interestingly, for the control group, the coefficient for the export shares in Africa turns positive (although not-significant). As in the previous cases, the government spending coefficient results positive and significant for the African case.
Table 6c. Inflation Deviation and Exports to Asian Drivers - Control Group -
Inflation Deviation vs Export AD 2000-2005 Control Group 2000-2005 Latin America Africa export_ad -4.98E+02 2.05E+01 [0.85] [0.21] gov_exp_growth 0.12 0.31* [0.25] [1.94] Observations 4.50E+01 5.90E+01 Number of id_gen 8 11 R-squared 0.03 0.08 Absolute value of t statistics in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
28
5. Export diversification
As shown previously debt management and fiscal responses have been in general more prudent that during previous episodes of bonanzas. The effects on the real economy and the management of trade impacts have also changed. In this section we concentrate on the trade impacts looking through the lens of different indexes (concentration, competition and specialisation).
In order to analyze the degree of diversification (or concentration) in developing countries, two samples are studied, Latin America and Africa. We used a classic Herfindahl-Hirschmann index for that purpose, calculated for the years 2000 and 2005, which corresponds to the period of Asian Drivers emergence. This concentration measure takes into account the weighted average of each good and country, so where values exported values are low (high), the influence on the indicator is reduced (increased). The index is calculated as follows:
⎛ n 2 1⎞ ⎜∑ p j − ⎟ ⎜ n ⎟⎠ j =1 HH = ⎝ 1 1− n where p j = xij / X i represents the market share of country j on the exports of country i in its total exports ( X i ). The squared-sum of all shares in also known as the Herfindahl-Hirschmann Index. Given that shares are weighted by the number of observations, the Herfindahl-Hirschmann index is adopted, allowing to compare different sets of goods and destinations.
29
Figure 5a. Herfindahl-Hirschmann Index for destinations Africa – selected countries
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
2000
South Africa
Uganda
Tanzania
Malawi
Cote d'Ivoire
Cameroon
Senegal
Ghana
Morocco
Benin
Mauritius
Niger
Tunisia
Zambia
2005
Mozambique
HH Index
Africa Herfindahl-Hirschmann Index by Destination
In the case of Africa, two major groups of countries showing different patterns can be discerned. On the one side, in countries like Mozambique, Zambia, Tunisia, Benin, Ghana, Cote d’Ivoire and Malawi, experienced deterioration between 2005 and 2005 with an increased of the HH indexes. Mozambique and Zambia experienced the sharpest increases of the concentration indexes by destination. In the case of Mozambique the major explanation lies in a boom of aluminium exports to one destination that is the trading platform of this commodity located in Switzerland (China and India are among the major purchasers). In the case of Zambia, increased exports of copper are behind this further specialisation and here again China became a major buyer. The others on the contrary experienced more trade diversification by destination, in particular countries like Niger, Tanzania, Cameroon or Uganda, large soft commodities exporters. Also South Africa managed to diversify its client portfolio.
30
Figure 5b. Herfindahl-Hirschmann Index for Destinations Latin America – selected countries
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
2000
Brazil
Argentina
Chile
Uruguay
Dominica
Guyana
Peru
Paraguay
Nicaragua
Bolivia
Colombia
Costa Rica
LAC avg
Panama
Ecuador
Guatemala
Honduras
Venezuela
Belize
2005
Mexico
HH Index
Latin America Herfindahl-Hirschmann Index by Destination
Source: OECD Development Centre, 2007.
In the case of Latin America, there is a considerable number of countries with high degrees of client concentration, especially Mexico, Venezuela, Honduras, Guatemala and Ecuador. Other countries like Colombia, Bolivia, and Peru display a moderate level of concentration, while Chile, Argentina and Brazil present low concentration levels. However, some countries have experienced more important changes in the last five years than others. While client concentration has increased importantly in Guatemala, Ecuador, Peru and Bolivia, all the countries experienced an increasing trade client diversification: China and the Asian drivers have been helping hands, inducing a trade diversification of Latin American exports in terms of destination.
The picture of the HH indexes by product confirm the increasing impact of the Asian Drivers in nearly all the African continent (Figure 6a). As shown in the graph, almost all African countries monitored by the African Economic Outlook experienced an increasing specialization of their trade profiles in terms of products. The sharpest movements between 2001 and 2005 are registered in soft commodities exporters like Mali and Niger. Hard commodities exporters like Angola, Chad, Nigeria, Congo, Mozambique or Mozambique also experienced sharp concentration movements. Interestingly, manufacturing exporters like South Africa, and to a lesser extend Morocco experienced, a trade deterioration, impacted both by foreign exchange appreciations and increased competition from the Asian Divers (Kaplinsky and Morris, 2007; in this Special issue). Note, however, that the results are biased by changes in commodity prices as they are 31
not computed at constant prices. So a fall of commodity prices would show fewer tendencies for product specialization. Figure 6a: Herfindahl-Hirschmann Index for Sectors Africa – selected countries
Export Concentration in Products for Africa Herfindahl Hirschman Index 1.0 0.9
2000
0.8
2005
0.7 0.6 0.5 0.4 0.3 0.2 0.1
Tunisia
Morocco
South Africa
Kenya
Senegal
Zimbabwe
Côte d'Ivoire
Namibia
Gambia
Ghana
Cameroon
Zambia
Algeria
Mozambique
Niger
Mali
Congo
Chad
Nigeria
Angola
0.0
Source: African Economic Outlook 2005-2006, OECD Development Centre, 2007.
Figure 6b:. Herfindahl-Hirschmann Index by Product Latin America – Selected countries Export Concentration in Products for Latin America Herfindahl Hirschman Index 0.9 0.8
2001
0.7 2006
0.6 0.5 0.4 0.3 0.2 0.1
Brazil
Guatemala
Mexico
Costa Rica
Colombia
Uruguay
Guyana
Honduras
Paraguay
Peru
Bolivia
Panama
Chile
Ecuador
Venezuela
0.0
Source: OECD Development Centre. Based on World Integrated Trade Statistics (WITS). Nomenclature: SITC Revision 3 (4digit).
32
Latin America does not seem to display major movements in product specialization nor diversification, once the HH-Indexes are corrected for price effects. The only exception of importance is Mexico, as a result of resource depletion rather than diversification success.
In terms of trade diversification, it seems that, for the majority of countries studied, the Asian Drivers helped to diversify the destinations of exports both for Latin American and for African countries, even if in the case of Africa the picture is more nuanced. When we focus on trade diversification by products, the picture is also more mixed, with some African and Latin American countries in particular experiencing a deepening of resource dependence.
Conclusive remarks: an opportunity for better management
This paper arrives at an overall optimistic assessment about the net macroeconomic policy responses to the impact of Asian Driver induced commodity booms in Africa and Latin America.
Commodity-exporting countries have accrued clearshort-term benefits from the current boom accrued by higher prices and export proceeds and greater profits. The commodity boom has also had a positive impact in terms of broadening exporters’ client base, enabling them to retire costly debt and improve credit profile, increase foreign exchange reserves so as to reduce vulnerability to future speculative attacks, finance infrastructure for future growth and to build nest eggs abroad and at home for leaner times. Meanwhile, the negative Dutch disease effects and increased ‘Leamer corner’ solutions and lower nonresource exports that might have resulted from the commodity boom have been milder than is often argued. The prospective sources of commodity demand from the Asian Drivers could well migrate from mineral/depletable to agricultural/replenishable commodities which recent research (Collier and Goderis, 2007) has shown to exert more positive long-run benefits for development.
Much of the beneficial effects observed so far are due to policy performance. The monetary policy choices have in general tried to target both inflation and real effective exchange rates, with some success (with exceptions such as Zambia). These findings, unexpectedly, suggest that over recent years, when the AD
33
driven commodity boom gained full traction, macroeconomic policy challenges have been better coped with and monetary policy better managed in Africa than has been the case ,on average, in Latin America.
The rapid growth of Asian emerging economies is raising demand for Africa's and Latin America’s commodities (oil, metals and precious stones) and has resulted in improved terms of trade as well as rapid growth in primary commodity exports. Yet, deficient infrastructure, insufficient investment in human capital, and inadequate policies limit the ability of African and Latin American economies to fully seize the opportunities opened up by global markets. The increased global demand for agricultural and food products gives hope for a more even development pattern than that which can be achieved under a solely extractive model, and therefore resources need to be dedicated, and productivity enhanced in these sectors.
For exporters of natural resources, the challenge is to capitalise on mineral windfalls to ensure that a large proportion of the proceeds from the minerals sector are invested in infrastructure and human-capital development. Diversification remains an imperative, it is a medium-term objective that will depend, among other things, on the capacity to promote private-sector development, particularly in Africa, and to expand markets. Commodity exporters are aware of their potential to consolidate export proceeds in sovereign wealth funds (SWF). These government controlled investment vehicles have stimulated protectionist sentiments in some OECD countries. Their asset size, so as their owners (governments) create fertile ground for suspicion. Nonetheless, several motives can be mentioned for commodity-oriented countries to run such funds: First, foreign exchange reserves – mostly held in US treasury bonds – have grown excessively large: interest rate and currency risk argue in favour of portfolio diversification, and central banks cannot control monetary aggregates anymore. Second, reducing resource dependence through vertical and horizontal sector diversification. Third, responding to expected demographic pressures, while smoothing consumption levels of future generations when resources are exhausted. Fourth, raising production efficiency as a future driver of growth. As discussed earlier, commodity exporters in both Africa and Latin America should be indifferent to whether keeping resources unextracted (in which case the return is the expected rise in future prices) and getting a market rate of return on its sale (Hotelling Rule for efficient depletion). Extracting and selling raw-commodity amounts to running down capital, unless the receipts are fully reinvested in financial, physical or human capital (Hartwick Rule for intergenerational equity); ‘genuine’savings are negative if the Hartwick rule is not respected. Funds can also be helpful for stabilising notoriously volatile raw
34
material prices. The law of diminishing returns forces commodity exporters to invest a large share of the savings abroad. They would be forced to disregard both the Hotelling and the Hartwick rules, if SWFs could not invest in OECD countries. The Hotelling rule warns that lowering the returns on investment from commodity receipts, by preventing investments by SWFs from commodity-rich countries, would lead to a reduced supply and higher commodity prices. This would be counter-productive for resource-abundant economies. If they do not follow the Hartwick rule, they have negative ‘genuine’ savings rates and become poorer each year3. This highlights the important policy question of what resource-rich economies can do to avoid the resource curse. Funds are an option, to the extent that commodity receipts are eventually transformed into other forms of wealth. Next to shifting out of excessive reserves and saving for future generations and old age, economic diversification and efficiency gains are major economic motives for establishing SWFs. Some countries (i.e. United Arab Emirates) are using their fund for rapid diversification of their economies away from oil toward tourism, aerospace and finance. Such a diversification motive is as legitimate as the desire to raise the efficiency of their economy through acquiring stakes in leading global companies. The insertion of commodity-based funds into the financial architecture could guarantee a more efficient management of proceeds and a sustainable instrument for enhancing growth.
3
The World Bank, Where Is the Wealth of Nations?, Washington, 2006.
35
References Adam, C.S. and D. L. Bevan (2006), "Aid and the Supply Side: Public Investment, Export Performance, and Dutch Disease in Low-Income Countries," World Bank Economic Review, Vol. 20 (2), pp. 261-290. Aghion, P. and I. Marinescu (2006), “Cyclical Budgetary Policy and Economic Growth: What Do We Learn from OECD Panel Data?”, mimeo, Cambridge (Ma.), Harvard University. Alesina, A. Tabellini, G. 2005. Why is Fiscal Policy often procyclical? NBER Working Paper 11600, Cambridge (Ma.), National Bureau of Economic Research. Alvarez, R. Fuentes, R. (2006), “Paths of Development, Specialization and Natural Resources Abundance”, Working Paper, Banco de Chile, 2006. Buffie, E., C. Adam, S. O’Connell, and C. Pattillo (2006), “Riding the Wave: Monetary Responses to Aid Surges in Low-Income Countries, CSAE WPS/2006-04, Oxford University. Blázquez, J., J. Rodríguez, and J. Santiso (December 2006), “Angel or Devil? China's Trade Impact on Latin American countries”, ECLAC Review, 90, pp. 15-41. Bloch, H., A.M Dockery, C. Wyn Morgan and D. Sapsford (2004), “Growth, Commodity Prices, Inflation and the Distribution of Income”, Research Papers, No. 0404, The University of Liverpool School of Management, Liverpool. Blommestein, H. and J. Santiso (April 2007), “New Strategies for Emerging Domestic Sovereign Bond Markets in the Global Financial Landscape”, OECD Development Centre Working Paper No. 260. Cashin, P., L. Céspedes, and R. Sahay (March 2003), “Commodity Currencies”, Finance & Development, Volume 40 (1). Chen, Y. C. and K. Rogoff (2003), “Commodity currencies”, Journal of International Economics, 60 (1), pp. 133 - 160. Collier, P. (January 2007), “Managing commodity booms: lessons of international experience”, Oxford University, Centre for the Study of African economies, Department of Economics, paper prepared for the African Economic Research Consortium; See http://users.ox.ac.uk/~econpco/research/africa.htm Collier, P. and B. Goderis (January 2007), “Commodity prices and growth: reconciling a conundrum”, University of Oxford, Department of Economics (unpublished). See http://users.ox.ac.uk/~econpco/research/africa.htm Collier, P. and J.W. Gunning (2005), “Asset Policies During an Oil Windfall: Some Simple Analytics”, The World Economy, Vol. 28 (10), 1401–1415. Corden, M.W. and P.J. Neary (1982), “Booming Sector and De-Industrialization in a Small Open Economy,” Economic Journal, Vol. 92, No. 368, pp. 825-48. Goldstein, A., N. Pinaud, H. Reisen, and X. Chen (2006), The Rise of China and India: What’s In It for Africa?, Paris, OECD Development Centre Studies. Gottschalk, R. and D. Prates (2007), “Asia’s Growing Demand for Primary Commodities: Macroeconomic Policy Responses from Latin America”, Brighton (UK), Institute of Development Studies at Sussex University. Hausmann, R. and M. Gavin (1996), “Securing Stability in a Shock-prone region: The Policy Challenge for Latin America”, in R. Hausmann and H. Reisen (eds.), Securing Stability and Growth in Latin America, Paris, OECD Development Centre, pp. 23 – 64.
36
Hnatkovska, V. and N. Loayza (2004), “Volatility and Growth”. Policy Research Working Paper Series 3184, The World Bank. Jeanne, O. and R. Rancière (October 2006), “The optimal level of international reserves for emerging market countries: formulas and applications”, IMF Working Paper, 229. Jimenez, J.P. and V. Tromben (December 2006), “Fiscal policy and the commodities boom: the impact of higher prices for nonrenewables in Latin America and the Caribbean”, ECLAC Review, 90, pp. 59-84. Kaplinsky, R. (2006), “Revisiting the revisited terms of trade: will China make a difference?” World Development, 34 (6). pp. 981995. Kamin, S. Marazzi, M. and John W. Schindler. “The impact of Chinese Exports on Global Import Prices”. Review of International Economics, 14(2), 179-201, 2006. Leamer, E. (1987), “Paths of Development in the Three-Factor, N-Good General Equilibrium Model”, Journal of Political Economy, Vol. 95, No. 5, pp. 961-99. Leamer, E., H. Maul, S. Rodriguez and P.K. Schott (1999), “Does Natural Resource Abundance Cause Latin American Income Inequality?”, Journal of Development Economics, Vol. 59, No. 1, pp. 3-42. Lubin, D. (2007), “Petrodollars, emerging markets and vulnerability”, Citigroup Global Markets, London, February (mimeo). Mühlberger, M. (2007), “Africa’s natural resources in the spotlight again”, Frankfurt, Deutsche Bank Research. Mundell, R. (March 1962), “The Appropriate Use of Monetary and Fiscal Policy for Internal and External Stability,” Staff Papers, International Monetary Fund, Vol. 9, pp. 70–77. OECD Development Centre and African Development Bank (2007), African Economic Outlook 2007, Paris, OECD. Reisen, H., M. Grandes and N. Pinaud (2004), “New Issues in Macroeconomic Interdependence”, OECD Development Centre Working Papers, No. 241, OECD, Paris. Santiso, J., ed. (2007), The visible hand of China in Latin America, Paris, OECD Development Centre Series. Tornell, A. and P. Lane (1999), “Voracity and Growth", American Economic Review, Vol. 89.1., pp. 22-46. Torvik, R. (2001), “Learning by doing and the Dutch disease”, European Economic Review, Vol. 45.2., pp. 285-306. Van der Ploeg, Rick (2006), Challenges and opportunities for resource rich economies, CEPR DP 5688 (2006). World Bank (2006), “Where is the Wealth of Nations?”, Washington DC.
37