The experience of developing countries over 1990-2010 indicates that commodity prices have a significant impact on fiscal outcomes. Both revenue and expenditure rise in response to commodity (import or export) price increases; the response of the fiscal deficit is ambiguous. A floating exchange rate regime only partially offsets the impact; foreign-exchange reserves do not dampen the effects. Hence, there is a strong case for fiscal hedging against commodity price shocks. Hedging instruments based on a limited set of benchmark world prices for a narrow set of commodities may suffice to realize most of the potential benefits.

Abstract

The experience of developing countries over 1990-2010 indicates that commodity prices have a significant impact on fiscal outcomes. Both revenue and expenditure rise in response to commodity (import or export) price increases; the response of the fiscal deficit is ambiguous. A floating exchange rate regime only partially offsets the impact; foreign-exchange reserves do not dampen the effects. Hence, there is a strong case for fiscal hedging against commodity price shocks. Hedging instruments based on a limited set of benchmark world prices for a narrow set of commodities may suffice to realize most of the potential benefits.

I. Introduction

The impact of commodity price shocks on fiscal outcomes remains a subject of considerable controversy in both academic and policy circles. The 2007–08 boom in food and fuel prices, current indications that a second global food-price shock may be underway, and the observed volatility in commodity prices have all greatly intensified this interest. In particular, they have led to significant concerns that commodity price shocks may complicate the management of fiscal and debt policy, by increasing budget uncertainty, encouraging a pro-cyclical fiscal policy, and threatening debt sustainability. Such concerns are especially acute in the case of low-income countries (LICs), which are relatively more exposed to commodity price shocks, and may be expected to rise further as LICs continue to integrate into international markets. As a result, there is renewed debate on whether hedging commodity shocks, through either market-based instruments or contingent official financing, would be beneficial and feasible.1 This paper pushes the debate forward by analyzing empirically two broad, related questions.

First, is there evidence that commodity price shocks significantly influence fiscal outcomes, inducing fiscal uncertainty? Put differently, is there a prima facie case for hedging against commodity price shocks? To this end, the paper assesses the impact of commodity export and import price shocks on fiscal revenue, expenditure, social expenditure, and public debt. It performs the analysis for several different economic groupings, including LICs, middle-income countries (MICs), commodity exporters, and commodity importers.

Second, is commodity price hedging, based on derivative instruments, likely to yield significant benefits in practice? This question, in turn, raises at least five separate issues. To start, can most of the adverse impacts of commodity price shocks be mitigated by traditional policy buffers, including floating exchange rate regimes and reserve assets?

Next, effective hedging instruments will, in the foreseeable future, likely be available at a reasonable cost only for a narrow set of commodities. Will such a limited set of instruments suffice to realize most of the potential benefits from hedging?

In addition, hedging instruments will likely be tied to a limited set of benchmark world commodity prices, rather than to country-specific commodity prices. Given the implied lack of precision in insuring against country-specific shocks (“basis risk”), will it still be possible to realize significant benefits from hedging?

Also important is that commodity exports and imports are influenced by shocks to not just prices, but also volumes. Are price shocks sufficiently dominant that price hedging will suffice to stabilize export revenue? Finally, commodity export and import prices may move together over time. Will such co-movements act to stabilize fiscal outcomes, reducing the importance of additional hedging?

Overall, the paper assesses the extent of fiscal exposure to commodity price shocks, and makes a case for financial hedging by the public sector. The bulk of the existing literature on commodity price volatility focuses on its growth impact, and on the pass-through of international prices into domestic prices. A few papers discuss the fiscal impact of commodity price shocks. In particular, Kaminsky (2010) documents that terms-of-trade booms are not necessarily associated with large fiscal surpluses in developing countries, reflecting the pro-cyclicality of government spending. In the same vein, Medina (2010) and Villafuerte et al. (2010) find a strong response of fiscal revenue and expenditure to commodity prices in Latin America and the Caribbean, with significant differences across countries, and Arze del Granado et al. (2010) find evidence of pro-cyclicality in social spending in developing countries. However, these analyses only covers a limited set of fiscal variables and countries, and fail to distinguish between commodity import and commodity export price shocks. Again, Cespedes and Velasco (2011) show that fiscal policy in commodity-rich nations was historically quite pro-cyclical, with the fiscal balance often deteriorating as commodity prices increased, but find evidence of reduced pro-cyclicality in the 2000s. However, their analysis only focuses on large, sustained commodity booms, rather than on commodity price changes more generally.

As for the role of hedging, the limited literature focuses on private, micro-level hedging rather than public, macro-level hedging. Among the exceptions, Borensztein et al. (2009) demonstrate the welfare gains associated with hedging against commodity price risks for commodity-exporting countries. In particular, they show that introducing hedging financing enhances domestic welfare by reducing both export income volatility, and the need to hold foreign assets as precautionary saving. Likewise, Daniel (2001) argues that many governments could benefit substantially from hedging against oil-price risk.

A related strand of the literature argues that commodity exporters or importers can insure against volatility in commodity prices not just through financial hedging, but also through policy buffers and non-financial hedging (for instance, by accumulating foreign assets, diversifying exports, and employing conservative price assumptions in the budget). Indeed, the public sector has typically relied on nonfinancial hedging. Many governments have strived to build up policy buffers, including creating fiscal space through fiscal consolidation and public debt payment. Other buffers, such as commodity stabilization fund scheme, are generally set up to deal both with the expected depletion of commodity resources and the volatility of commodity-related income. There are trade-offs between financial and nonfinancial hedging. In particular, the potential limitations of nonfinancial hedging include the need for strong institutions and efficient policy coordination (see Ossowski et al., 2008).

Further, as stressed by Borensztein et al. (2009), building up financial asset for precautionary motives comes at the cost of reducing consumption and welfare. The costs of financial hedging include, for instance, ill-conceived contract negotiations which lock in commodity prices lower than market trends. More fundamentally, opportunities for financial hedging are incomplete, especially over longer horizons (Becker et al., 2007).

The rest of this paper is organized as follows. Section II sets out some key stylized facts, focusing on the importance of commodity trade, and some prima facie evidence that it may influence fiscal outcomes. Section III sets out a formal empirical methodology for examining the cross-country link between fiscal outcomes and commodity prices. Section IV discusses the results. Section V examines some of the above-mentioned potential problems with price hedging. Section VI concludes.

II. Stylized Facts

A. The Importance of Commodity Trade

We set the stage by briefly illustrating the importance of commodity exports and imports, and showing that this has not diminished over time. We abstract from developments over the past couple of years, since these are dominated by the possibly temporary response to the financial crisis. Since 1980, for developing countries as a whole, exports and imports have grown, not just in absolute terms, but also relative to GDP (Figure 1). Middle-income countries (MICs) are broadly more open than LICs. In much of the subsequent analysis, we focus on “commodity exporters” or “commodity importers”, defined as those countries where commodity exports or imports account for at least 20 percent of GDP.2 Largely by construction, these countries are significantly more open than the average developing country. Commodity trade has broadly followed the same trend as overall trade. Both commodity exports and imports have generally grown relative to GDP (Figure 2).

Figure 1.
Figure 1.

Trade / GDP, 1981–2008

(Mean Values)

Citation: IMF Working Papers 2012, 112; 10.5089/9781475503333.001.A001

Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE/WITS.
Figure 2.
Figure 2.

Commodity Trade / GDP, 1988–2008

(Mean Values).

Citation: IMF Working Papers 2012, 112; 10.5089/9781475503333.001.A001

Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE/WITS.

Both commodity exports and imports are significantly concentrated, with no clear trend toward an increase or decrease in specialization over time. The share of the 3 most important commodities in total commodity exports or imports has remained roughly stable since the mid-1990’s (Figure 3). Interestingly, the mean degree of specialization appears greatest in LICs. Even for developing countries as a whole, the top 3 commodities account on average for over 40 percent of total commodity exports. This suggests that, at least for some countries, hedging on a limited number of markets might yield significant insurance against aggregate commodity-revenue fluctuations. In the aggregate, crude oil dominates developing countries’ commodity exports and imports (Table 1 and Table 2). Other key commodities include copper, fish, coal, and iron.

Figure 3.
Figure 3.

Share of Top 3 Commodities in Total Commodity Trade, 1988–2008

(Mean Values).

Citation: IMF Working Papers 2012, 112; 10.5089/9781475503333.001.A001

Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE/WITS.
Table 1.

Top 10 Commodities: Exports as a Share of Total Commodity Exports, 2000–08

(Aggregate Values).

article image
Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE/WITS.
Table 2.

Top 10 Commodities: Imports as a Share of Total Commodity Imports, 2000–2008

(Aggregate Values).

article image
Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE / WITS.

Finally, commodity export and import prices are both very volatile. For developing countries as a group, the late 1990’s saw a sharp decrease in commodity prices, followed by a boom for most of the 2000’s, and a collapse in 2008–09 (Figure 4). Quantitatively, the developing-country average annual growth rate of commodity prices exhibits over most time periods a standard deviation on the order of at least 2–3 percentage points. For individual countries, fluctuations can be much more severe: commodity export and import prices both display an overall panel standard deviation on the order of 10 percentage points.

Figure 4.
Figure 4.

Growth Rates of Commodity Export and Import Prices, 1989–2010

(Income Effect, Percent of GDP, Mean Values).

Citation: IMF Working Papers 2012, 112; 10.5089/9781475503333.001.A001

Source: IFS; WEO; World Bank; Commodity Prices Database; and COMTRADE/WITS.Notes: Income effect is calculated as the growth rate of commodity export (or import) prices, multiplied by the share of commodity exports (or imports) in GDP.

B. Commodity Price Shocks and Fiscal Exposure

This section employs simple statistics, based on correlation analysis, to illustrate to what extent movements in commodity prices are associated with changes in fiscal variables, including revenue, expenditure, deficits, and public debt. Overall, the data suggest strong correlations between world commodity prices and fiscal outcomes (Table 3). For instance, commodity export prices have positive, relatively large correlations with revenue / GDP, and negative correlations with debt / GDP. The correlations are even larger with respect to the prices of the 3 most important export commodities.3 These preliminary results suggest a prima facie case that hedging against commodity price volatility may smooth fiscal adjustment.

Table 3.

Correlation Analysis: Fiscal Outcomes and Commodity Prices, 1990–2010

article image

III. Methodology

The empirical strategy involves reduced-form cross-country panel regressions. The dataset is an unbalanced annual panel, covering the period 1990–2010 and (depending on the precise variable) up to 116 countries. We focus on the extent of fiscal exposure to commodity price shocks, and adopt the following benchmark fiscal exposure equation:

ΔYit=αi+βt+γΔYit1+(δx+δxer.ERit+δxres.RESit)Δpx,itxit1+(δm+δmer.ERit+δmres.RESit)Δpm,itmit1+δerERit+δresRESit+wit(1)

The subscripts i and t denote, respectively, the country and the time period, while the subscripts x and m denote, respectively, commodity exports and imports.

ΔYit denotes the percentage point change in each of the following dependent variables, in turn: (i) total revenue / GDP; (ii) total expenditure / GDP; (iii) social expenditure (on education and health) / GDP; (iv) fiscal balance (surplus) / GDP; and (v) public debt / GDP.4

px, it and pm, it are the country-specific, time-varying commodity (spot) export and import price indices. They are constructed based on the actual weight of each commodity in the country’s export or import basket, and on the world price of that commodity.

xit=xitGDPitandmit=MitGDPit are country-specific, time-varying weights set equal to, respectively, the share of commodity exports and imports in GDP. That is, commodity price indices are weighted by the country’s total commodity exports or imports, relative to GDP. The weights are lagged to reduce endogeneity concerns.

In commodity exporters, we expect commodity export prices to be positively associated with revenue, through their impact on income taxes (and in particular profit taxes) and non-tax revenue (including royalties and production sharing agreements). The direct impact of commodity export prices on trade taxes is likely less significant, given that export taxes have been widely removed in most developing countries since the 1980s. We also expect commodity export prices to be positively associated with expenditure, including social expenditure. The magnitude of the response would optimally depend on the extent to which commodity price changes are seen as permanent, as well as whether public investment is required to take full advantage of the increased export prices. We have less definite priors on the response of the fiscal balance (surplus) and debt, although the general normative presumption is that, to the extent that the commodity price changes are seen as temporary, the fiscal balance should increase.

In commodity importers, we expect commodity import prices to be positively associated with revenue, through their direct impact on trade taxes. We also expect commodity import prices to be positively associated with expenditure, and in particular increased spending on social safety nets, or food and fuel subsidies. Again, we have less definite priors on the response of the fiscal balance and debt.

We also interact commodity prices with the following variables, to examine how they affect the impact of commodity price shocks:

  1. ERit, which denotes a country’s de facto exchange rate regime. ERit = 1 for fixed exchange rate regimes, and 0 otherwise (the classification is based on the llzetzki et al., 2009, approach).5 The hypothesis is that floating exchange rates may weaken the impact of commodity-price changes on revenue, for two reasons. First, floating exchange rates may dampen any impact on output. Second, in response to, say, a reduction in commodity export prices, a depreciation would act to increase revenue from trade taxes (which tend to be significant in LICs). On the other hand, floating exchange rates may magnify the impact of commodity-price changes on external debt service and the debt burden, as long as debt is denominated in a foreign currency.

  2. RESit, which denotes a country’s reserves (relative to imports). There are (at least) two competing hypotheses. First, greater reserves may allow for a smaller response of exchange rates to changes in commodity prices (and in particular to negative price shocks), reducing any dampening effect from exchange rate movements. Second, greater reserves may allow governments to smooth consumption in the face of negative shocks.

More precisely, we estimate four different specifications of equation (1) above: (i) with neither exchange-rate regime nor reserve interactions; (ii) with exchange-rate regime interactions; (iii) with reserve interactions; (iv) with both exchange-rate regime and reserve interactions. We also test for asymmetric fiscal responses to positive versus negative commodity-price shocks. The dynamic model is estimated using the Arellano - Bond difference GMM.6

For each specification, we estimate the response of each of the five fiscal outcomes for several different groups: (i) LICs; (ii) LIC commodity exporters; (iii) LIC commodity importers; (iv) MICs; (v) MIC commodity exporters; (vi) MIC commodity importers; (vii) LICs and MICs; (viii) LIC and MIC commodity exporters; and (ix) LIC and MIC commodity importers.7 The focus on commodity exporters and importers is motivated by the hypothesis that any impact of increases in commodity export and import prices will be easier to observe in countries that are heavily reliant on commodity trade.

Further, we examine to what extent fiscal outcomes depend on price fluctuations for a narrowly defined set of commodities, and by extension whether hedging strategies based on a narrow set of hedging instruments might prove useful. To do so, we re-estimate the above regression by replacing the aggregate commodity price indices with price sub-indices for the three most important commodities (the “top-3 commodities”), and for all other commodities (the “non-top-3 commodities”). These sub-indices are again based on the weight of each commodity in the country’s export or import basket.

In addition, we present both the short-run impact and the long-run effect of a change in commodity prices on the fiscal variable of interest, where the long-run effect = short-run impact / (1 - γ).

The forecast error variance decomposition is obtained by taking the variance of both sides of equation 1 above (averaged over time and across countries). The resulting terms on the RHS will include the fiscal variable’s own effect, a pure commodity export price effect, a pure commodity import price effect, and various interactions.

Appendix I describes in greater detail the above variables and their sources. Appendix II lists the countries and country groupings. Appendix III provides summary statistics for the key variables.

IV. Results

Overall, cross-country panel regressions suggest a large fiscal exposure to commodity-price shocks, stemming from automatic stabilizers on the revenue side, and a positive and significant response of expenditure, including social spending, to commodity prices. In LICs, and particularly in commodity exporters and importers, the magnitude of the responses is relatively high. Further, in these countries, expenditure tends to respond more strongly than revenue. As a result, we typically observe a significant, positive response of public deficits and debt to commodity import prices in LIC commodity importers, and even to commodity export prices in LIC commodity exporters. The effects are larger under fixed exchange-rate regimes, and persist over time. We now discuss these results in greater detail.

A. Commodity Export Prices and Fiscal Outcomes

All statistically significant fiscal responses to a change in commodity export prices are summarized in Table 4 and Figure 5. The full underlying baseline regressions are reported in Table 6Table 10.8,9

Table 4.

Response of Fiscal Outcomes to a 10 Percent Increase in Commodity Export Prices

(Percentage Points of GDP).

article image
Source: Authors’ calculations, based on regression results (Tables 611). The short-term response of a fiscal variable is computed as a linear combination of the statistically significant coefficients, excluding that on the lagged dependent variable.
Figure 5.
Figure 5.

Short-Run Response of Fiscal Outcomes to a 10 Percent Increase in Commodity Export Prices

(Percentage Points of GDP).

Citation: IMF Working Papers 2012, 112; 10.5089/9781475503333.001.A001

Source: IFS; WEO; World Bank; Commodity Prices Database; COMTRADE/WITS; and IMF staff estimates.Notes: Depicts Arellano-Bond estimates of the average short run effects over 1990-2010 of a 10 percent increase in commodity export prices in a benchmark economy, where commodity exports / GDP equal 20 percent.
Table 5.

Response of Fiscal Outcomes to a 10 Percent Increase in Commodity Import Prices

(Percentage Points of GDP)

article image
Source: Authors’ calculations, based on regression results (Table 6Table 10). The responses are computed as linear combination of the statistically significant coefficients.
Table 6.

Panel Regressions. Dependent Variable: Change in Central Government Total Revenue / GDP

article image
Notes: All regressions include country- and year-effects. Robust standard errors in parentheses. *, **, and *** denote significance at the 10, 5, and 1 percent level.
Table 7.

Panel Regressions. Dependent Variable: Change in Central Government Total Expenditure / GDP

article image
Notes: All regressions include country- and year-effects. Robust standard errors in parentheses. *, **, and *** denote significance at the 10, 5, and 1 percent level.
Table 8.

Panel Regressions. Dependent Variable: Change in Central Government Social Expenditure / GDP

article image
Notes: All regressions include country- and year-effects. Robust standard errors in parentheses. *, **, and *** denote significance at the 10, 5, and 1 percent level.
Table 9.

Panel Regressions. Dependent Variable: Change in Central Government Fiscal Balance / GDP

article image
Notes: All regressions include country- and year-effects. Robust standard errors in parentheses. *, **, and *** denote significance at the 10, 5, and 1 percent level.
Table 10.

Cross-Country Panel Regressions. Dependent Variable: Change in Central Government Gross Debt

article image
Notes: All regressions include country- and year-effects. Robust standard errors in parentheses. *, **, and *** denote significance at the 10, 5, and 1 percent level.