Fiscal Policy Formulation and Implementation in Oil-Producing Countries

2 An Alternative Interpretation of the “Resource Curse”: Theory and Policy Implications

Jeffrey Davis, Annalisa Fedelino, and Rolando Ossowski
Published Date:
August 2003
  • ShareShare
Show Summary Details
Rlcardo Hausmann and Roberto Rlgobon 

I. Introduction

It is often said that most people when reading about a theory wonder if it works in practice. Economists, when seeing things working in practice, wonder if they work in theory. The natural resource curse is a case in point. Countries highly dependent on oil or other natural resources have performed very poorly since 1980. Figure 2.1 shows GDP per capita at purchasing power parity for highly resource-intensive countries such as Saudi Arabia, Nigeria, Venezuela, and Zaire, and for less intensive countries such as Indonesia and Mexico. The pattern is clear: the more dependent performed remarkably poorly. The less oil-dependent did better.

Figure 2.1.GDP Per Capita at Purchasing Power Parity for Different Countries

(In constant U.S. dollar prices)

Sources: World Penn tables; and authors’ calculations.

The concern that natural resource wealth may somehow be inmiserat-ing is a recurring theme in both policy discussions and empirical analysis. The empirical regularity seems to be in the data,2 but understanding its causes has been a much harder task.3 Theorists have been hard at work to find a rationale. Is it a consequence of the Dutch disease? Is it caused by the volatility that characterizes resource-based commodity prices? Is it due to political economy forces unleashed by the presence of rents? And what are the policy implications of this problem? Is there such a thing as having too much oil for the country’s own good? Should oil income be saved in net terms? Or is the question mainly that of dealing with the volatility in the flows? Are other policies called for?

In this paper we will propose an alternative rationale for the resource curse and discuss some of its policy implications. The approach is based on the interaction between two building blocks: specialization of the domestic economy in the production of nontradables and financial market imperfections. We show that as an oil economy becomes more specialized in the production of nontradables, the real exchange rate becomes more volatile because shocks to the demand for nontradables—associated for example with the fiscal expenditure of shocks to resource income—will not be accommodated by movements in the allocation of capital and labor across sectors but instead by expenditure switching. This requires much larger relative price movements.

Financial frictions such as risk aversion or costly bankruptcy on corporate debt imply that the interest rate will be a function of the volatility in the economy. In fact, the volatility of profits in the nonresource tradable sector can be shown to be larger than in the nontradable sector. As volatility increases, sector-specific interest rates rise causing a decline in the output that is larger for the nonresource tradable sector. A multiplier process is set in motion, where an initial rise in interest rates causes the tradable sector to contract, further raising volatility and interest rates until the sector disappears. At that point, the economy will face an even higher interest rate and a lower level of capital and output in the nontradable sector. An increase in resource income that leads to specialization causes a large decline in welfare: thus the idea of the curse. This form of specialization is inefficient and is characterized by high volatility and interest rates, weak real exchange rates, and low wages and investment. Inefficient specialization is determined by the level and the volatility of resource income and by the international interest rate. The paper discusses the role of fiscal saving of oil revenues as well as stabilization of expenditures. In addition, independent policies that reduce country risk and that improve the functioning of financial markets are seen as being particularly important in this context. More interventionist policies to subsidize investment in the nonresource tradable sector may also have a role to play.

II. Previous Approaches to the Resource Curse

There have been several approaches in the literature to account for the resource curse. The first is associated with the notion of the Dutch disease. The second has to do with the rent-seeking activities generated around the presence of the associated tax revenue. The third approach has to do with the damaging effects of volatility. In this section we will discuss each of these theories and their limitations.

The Dutch Disease Approach

Increases in resource-based revenues, such as oil, generate a greater capacity to import tradables, but typically prompt a greater demand for all goods including nontradables, which cannot be imported but must be produced locally. This requires the economy to move resources out of the nonresource tradable sector—call it manufacturing—in order to expand the production of nontradables such as construction and services. An oil boom would lead to a contraction in manufacturing. A real appreciation is the mechanism that gets the job done (Corden, 1982; Corden and Neary, 1982). This is the Dutch disease.

This logic is compelling, but by itself it does not imply any inefficiency or welfare loss. It only states that booms in resource income would be associated with contractions in manufacturing, not in overall growth. It cannot explain why a country would grow more slowly, just because it has oil.

To get some mileage, one has to assume that nonresource tradables play a special role in the growth process. This is the tradition started by Matsuyama (1992) where he assumed that there are increasing returns to scale in manufacturing, but not in the resource sector. Hence, an abundance of the natural resource makes the economy specialize in the less dynamic sector,4 which may explain the curse.

One problem with this explanation is that one should expect that over the long run oil-exporting countries would be doing poorly when oil is doing well and that they would be improving when oil becomes less dynamic. However, over the 60-year period between the early 1920s and the early 1980s, Venezuela was the country with the highest growth in per capita income out of the 40 countries for which Maddi son (1995) has data. During the following 20 years, Venezuela is among the worst performers in the world, a period that coincides with the decline of its oil income (Hausmann, 2001). This is not just a coincidence. The same pattern can be observed if we look at the growth rate of a wider group of countries over a shorter period. As Table 2.1 shows, oil-exporting countries grew faster in the period of rising oil prices and volumes between 1960 and 1980 relative to other developing countries and collapsed when oil revenues declined after 1980. If the Dutch disease story was right, the post-1980 story should have been one of greater growth. Hence, the story as such does not fit the facts.

Table 2.1.Average Annual Growth in Per Capita GDP at Purchasing Power Parity
Country Group1960-19981960-19801980-1998Number of Countries
All developing countries1.73.00.2115
Oil exporters1.15.2-2.115
Sources: World Penn Tables; and authors’ calculations.
Sources: World Penn Tables; and authors’ calculations.

Moreover, it is not obvious that learning by doing or technological development is slower in resource-based industries relative to others. As de Ferranti and others (2001) have shown in a recent report, several industrial countries such as Australia, Canada, and Finland became rich by developing the technology and capital goods industries associated with their resource sectors. Moreover, whatever bad effects specializing in natural resources might generate, they have to be compared with the benefits of owning large natural resources.

For example, natural resource income may raise national savings and hence facilitate capital accumulation and growth. This is not a minor aspect as oil-exporting economies have averaged twice the savings rate of non-oil exporters (Table 2.2). Moreover, they were able to save more and grow more in the period when oil was growing fast (1960-1980) relative to the more recent episode (1980-1998).

Table 2.2.Average Domestic Savings Rate
Country Group1960-19981960-19801980-1998Number of Countries
All developing countries17.118.416.2111
Oil exporters33.237.930.115
Sources: World Penn Tables; and authors’ calculations.
Sources: World Penn Tables; and authors’ calculations.

The Rent-Seeking Story

An alternative story is that resource wealth such as oil somehow makes societies less entrepreneurial. There is so much wealth floating around the government that entrepreneurial persons find it much more profitable to engage in unproductive rent-seeking activities to appropriate that wealth rather than in creating more wealth. The presence of common-pool problems or uncertainty over property rights over the resource income may generate low growth by inefficiently focusing economies in fighting over existing resources.

The common-pool problem—caused by situations where costs are shared between many agents but benefits are private (e.g., as in fiscal policy)—may lead to overspending on average and to a distorted allocation of spending over time. Overspending is associated with the idea that different constituencies do not internalize the full cost of their spending requests, as they only pay a small fraction of the additional tax burden (Johnsen, Shepsle, and Weingast, 1981; von Hagen and Harden, 1997). This problem is not specific to resource-rich economies, but instead is present in all countries. However, in resource-rich economies, where nonresource taxes are typically low and resource rents are large, it could be argued that this force could in theory be more powerful.

In a dynamic setting this logic may lead to overborrowing and to a voracity effect (Hausmann, Powell, and Rigobon, 1993; Tornell and Ve-lasco, 1995; and Lane and Tornell, 1999b). Assume that it is best to save a temporary boom until some future time when lower resource income is expected. An individual would choose to smooth consumption. However, when there is a common-pool problem each constituency will ask for a larger share of the pie in good times, fearing that if it does not, other constituencies might take it away.

This story again does not explain why oil economies did so well when incomes were rising and why they have underperformed so strongly in the last two decades.

In yet a different setting, others have argued (Karl, 1997) that oil economies, by not developing the political compact that allows the state to tax its citizens, are poorly equipped to deal with collapses in oil revenues without leading to macroeconomic crises.

This logic may be present, but other factors may well overwhelm it. We already mentioned the fact that resource revenue may allow for higher savings. It may also allow the country to reduce taxation over more mobile factors and hence achieve a less distortionary overall taxation scheme, thus generating a more propitious economic environment for growth. Moreover, the political skills required to allocate rents among different groups may be useful in achieving the necessary reallocations when income declines.

The Volatility Story

An alternative explanation to the curse puts the emphasis on volatility. Volatility has been shown to be bad for growth, for investment, for income distribution, for poverty, and for educational attainment.5 Natural resource rents tend to be very volatile because the supply of natural resources exhibits low price elasticities of supply (at least in the short term). For example, the standard deviation of oil price changes has been about 30 to 35 percent per year. For a country where oil represents about 20 percent of GDP, a one-standard deviation shock to the price of oil represents an income shock equivalent to 6 percent of GDP. This is huge relative to total GDP volatilities in industrial countries (about 2 percent) or even developing countries (between 3 percent and 4 percent).

But how does volatility in the terms of trade damage the economy? Assume that resource revenues are distributed to the population as a whole, say through government transfers. This means that the fact that the revenue is volatile makes it less valuable to risk-averse consumers. Let us take a relatively severe example. Assume that in a given economy oil is 30 percent of national income and that it has a standard deviation of about 30 percent per year. Assume that utility can be described with a constant relative risk aversion (CRRA) utility function with a relatively high coefficient of risk aversion of 3. This implies that consumers would be willing to sacrifice 4.05 percent of national income in order to make oil revenues perfectly certain. They would be willing to spend that much money in hedging their resource risk: certainly more than a simple nuisance, but nothing that could reasonably be called a curse, when compared with the revenue it generates. This could reasonably justify organizing a seminar to discuss how governments should manage this risk, but is not an adequate explanation for the massive collapse in growth exhibited in Figure 2.1.

So the welfare losses associated with the consumption risk of the flow itself are not particularly large. To get bigger effects, the rest of the economy must somehow be disrupted by the volatility in oil. Interestingly, in a neoclassical setting, it is quite hard to make volatility matter. Imagine first a competitive economy in which capital is perfectly mobile internationally and labor is nationally fixed, but is perfectly mobile domestically across sectors. Assume that there are three sectors: a resource sector—which we will call oil, a tradable sector, and a nontrad-able sector. Oil is produced without either capital or labor: it is like manna from heaven. One can also think of it as aid. The two other sectors use capital and labor and exhibit constant returns to scale. In this case, so long as all goods are produced domestically, the volatility of the oil sector will not affect the value of non-oil output or any non-oil relative price. The income of workers and the rate of return to capital in the non-oil economy will be unaffected by oil volatility. Hence, the only problem will be the dislike for oil volatility itself and this we have already found not to be too serious.

To understand this surprising result it is important to remember that there are five prices in this economy. First, the price of oil is determined abroad. Second, the rate of return to capital will be set by the world interest rate, given perfect capital mobility. Third, the price of tradables will be set internationally through the law of one price. Fourth, the internationally determined price of tradables and cost of capital together with the zero-profit condition required by perfect competition will set the wage that the tradable sector can pay, which will determine, through the free movement of workers between sectors, the wage rate for the whole economy. Fifth, this will determine nontradable prices since wages and the cost of capital are already set and profits must be zero. Shocks to the demand for nontradables (induced possibly by oil shocks) will be adjusted through movements of labor between sectors and movements of capital in and out of the economy. The production possibility frontier will be completely flat. Therefore, an increase in demand driven, for example, by a positive oil income shock would result in an increase in production of nontradables (the shift from YNA to YNB in Figure 2.2), to be accommodated by a decrease in the domestic production of tradables, from YTA to YTB, without changes in relative prices. At the same time, the upward shift in the budget constraint due to higher income will result in higher consumption of both nontradables and tradables (the point Ctb in Figure 2.2).

Figure 2.2.The Benchmark Model: The Diversified Economy

This is a very important benchmark not because it is an adequate description of the world, but because it provides a point of departure to think of the possible characteristics of the world—not included in the benchmark—that may explain why oil and its volatility may become really problematic.

With full employment, and set wages and returns to capital, non-oil income will be stable. The only source of volatility will be the direct impact of oil revenues on household income. As we mentioned above, the welfare losses associated with oil revenue uncertainty are not huge—a nuisance more than a curse. To explain the curse, the non-oil economy must be more seriously disturbed by oil volatility.

What features of the world does the benchmark model not take into account? First, capital takes time to build and once invested is usually irreversible. It cannot move instantaneously between sectors. This is certainly a problem. It means that the production function will be convex and that the volatility in spending of oil income will cause shifts in relative prices. However, the consequences of this are not as serious as one might think. In a standard model, profits are a convex function of relative prices (Caballero, 1991). This means that the greater the volatility in relative prices the larger the expected profits and average investment will be!6 Moreover, if we assume full employment, the welfare of workers may actually go up and not down with volatility, since they would be working on average with more capital. Hence, irreversibility of investment cannot be the basis for a serious curse.

Second, consider the presence of price and wage rigidities that prevent the labor market from clearing. In this case, and assuming that capital is predetermined and irreversible,7 volatility in oil income and government spending will translate into changes in unemployment and output in the nontradable sector. Notice, however, that the tradable sector will remain unaffected: it will face constant prices, wages, and stock of capital. Therefore employment and output in the tradable sector will be constant. The nontradable sector will have constant capital and wages but volatile output. There will also be more capital invested in the nontradable sector than in the benchmark model, given the convexity of the profit function. Welfare losses caused by oil volatility are likely to be larger, because volatility is larger, but the expected average levels of output and consumption should not be much affected.

A New Approach

So where can the curse come from? We will argue that it will arise from an interaction between specialization and financial market imperfections. In the benchmark model we assumed that the economy was producing all goods. What happens if the non-oil economy stops producing non-oil tradables and becomes completely specialized in nontradables? Central to the results of the benchmark model is the requirement that there be a positive level of production of tradables: the non-oil economy must not be fully specialized in nontradables. This allows labor movements between sectors to absorb the shocks to nontradable demand. If the economy were fully specialized in nontradables, this result would disappear (Figure 2.3). Labor would now be fixed and fully employed in nontradables. The only way to expand supply would be by increasing the amount of capital per worker in the sector. But capital is required to get the international rate of return. However, with labor fixed, the productivity of each additional unit of capital would be falling. To avoid a fall in returns to capital, the price of nontradables must go up. Hence, the supply of nontradables will now be upward sloping. But the demand for nontradables must be downward sloping. An increase in the price of nontradables will cause expenditure-switching effects as consumers will substitute away from the now more expensive nontradables and into tradables. The relative price between these two goods—that is, the real exchange rate—will have to move in order to clear the market for nontradables. In the end, the economy will remain “stuck” in the corner solution shown in Figure 2.3, where no tradables are produced, and shifts in demand driven by income changes only affect consumption levels of tradables.

Figure 2.3.The Benchmark Model: The Specialized Economy

So, in the benchmark model, a specialized economy with volatile resource revenue will see a volatile real exchange rate, while a diversified economy will have a constant real exchange rate. In this setting, irreversibility in capital will make relative prices even more volatile, as now the supply of nontradables would be completely predetermined, given that both labor and capital will be fixed. Only expenditure-switching forces will be at play. This will make the real exchange rate even more volatile.

It will be shown below that this still is not enough to generate a real curse. In addition, it is necessary to assume some form of financial friction, either risk aversion or costly bankruptcy. This will make interest rates go up as the volatility of the real exchange rate increases, creating a vicious circle between greater volatility of the real exchange rate and higher interest rates on the one hand and lower investment in tradables (more specialization) on the other, which induces greater real exchange rate volatility. The interest rate faced by the tradable sector increases until the sector disappears and the economy fully specializes inefficiently in nontradables. The inefficiently specialized economy will exhibit higher interest rates on nontradables, lower capital and wages, and a more depreciated exchange rate.

III. Modeling the Curse

In this section we offer a formal model of inefficient specialization. The ingredients of the model are the following: assume there are three sectors in the economy—tradables, nontradables, and oil. Oil is assumed to consume no inputs and generates a stochastic stream of revenues denominated in tradables. We assume it is exogenous and denoted by gL. For simplicity, we have assumed that government expenditure is defined in per capita terms.

The tradable and nontradable sectors comprise a finite number of firms, each using capital and labor.

We assume that capital is owned by foreigners. This simplifies the analysis as it allows us to disregard the effect of changes in capital income. Nevertheless, the results are unaffected (qualitatively speaking) by assuming that domestic agents own some capital. We will assume that capital is irreversible and that it has to be decided one period before production and oil revenues are realized.

We assume that oil belongs to the government, which consumes it entirely in nontradable goods. If the government decides to save its oil revenue, it will do so in foreign assets. We will assume that households derive no utility out of government consumption.8 This means that the volatile government consumption will not enter directly into the utility of risk-averse households who might want to smooth it. This eliminates the standard justification for stabilization.9 As discussed above, this cannot possibly be the source of the curse if there is one. We assume that there are no taxes.

Finally, we assume that capital is fully depreciated in one period and that consumers cannot save. Thus, in this regard, the model is equivalent to a single-period model.


We assume that firms are small (price takers). We assume there are NN firms in the nontradable sector. Each firm requires an investment of one unit of capital to operate, and each production function is given by

where y refers to the output per firm, l refers to the labor input per firm, and the subscripts N and T refer to the nontradable and tradable sectors, respectively.

Optimal labor decisions conditional on the capital invested in the nontradable sector solves

where PN and w are the price of nontradable goods and wages in domestic currency, respectively. The solution implies that labor demand is

production is

and profits in the sector in domestic currency are

The tradable sector is similar. There are NT firms, where, again, each firm requires one unit of capital. We have assumed that the capital share is the same in both sectors. Hence,

where e is the exchange rate. Production is

and profits in the sector in domestic currency are


We assume that oil exports are in foreign currency. Hence, the total government consumption of nontradable goods is

In this simple setup we are assuming that the government does not face any financial frictions. In other words, the cost of financing is the same as the benefits of saving.


Households consume tradable and nontradable goods, but not oil. As mentioned above, there are no taxes. We assume that consumers’ utility can be represented by the standard Cobb-Douglas utility function. Furthermore, to simplify the number of parameters under study, we assume equal weights on tradable and nontradable goods. Assume the consumers solve

where CT, CN, and W are the consumption of tradables, consumption of nontradables, and wealth of consumers, respectively; (3 is the share of nontradables. The first-order conditions (FOC) of the maximization problem implies the standard solution. Given our assumption that the capital is owned by foreigners, then consumers’ income is given by total labor income:

where L is total labor employed in both tradable and nontradable sectors.


Conditional on the amount of capital in the nontradable sector, we compute the labor market equilibrium:

and the nontradable goods market equilibrium in our benchmark model:

where YN is the total nontradable output.

Substituting, the two equations solving for equilibrium are


where q is the relative price of nontradables in terms of wages and Q is the real exchange rate.

This implies that the two equations collapse to

Note that the total number of firms must be greater than zero. Note also that it is impossible to find a set of parameters in which the number of nontradable firms is zero. However, there exists a set of coefficients where the number of tradable firms indeed is zero. For those circumstances, the set of equations determining the equilibrium in the labor market and the nontradable goods market is identical to the previous ones but with NT equal to zero.

After some algebra these two equations collapse to the following relationship:

Assume the number of firms in the nontradable sector is given.10 Then this equation uniquely defines the number of tradable firms in equilibrium. The equilibrium is studied in Figure 2.4.

Figure 2.4.Inefficient Specialization

The downward sloping line corresponds to the left-hand side (LHS) of equation (20), and the increasing concave schedule is the right-hand side (RHS). Notice that all the uncertainty of government oil expenditures appears on the RHS. Hence, we have depicted three schedules reflecting the (supposedly) maximum, median, and minimum of the shocks to oil income.

As can be seen, the line and the schedule determine (uniquely) the real exchange rate at which both equilibrium conditions are satisfied. Note that for each realization of oil income there is a corresponding real exchange rate (RER).

In the figure, we have depicted an increase in the number of tradable firms. This makes the LHS line steeper. For a given level of uncertainty, this implies an increase in the expected value of ψ and a reduction in its variance. Since is an inverse function of the RER, we conclude that a larger tradable sector requires a more depreciated exchange rate, but delivers a more stable RER for the same degree of oil-related uncertainty.

This is an important characteristic of the model and it is useful to understand what drives it. When the number of tradable firms is large (small), reductions in the demand for nontradables can be accommodated through an expansion in the output of tradables with a relatively small (large) decline in the real wage, since the high (low) stock of capital per potential worker invested in tradable production implies that the marginal product of labor declines little (a lot) for every additional worker. In the limit, if there is no capital invested in the tradable sector, there will be no employment in the sector and the adjustment will take place exclusively through the expenditure-switching implications of real exchange rate movements. This fact will become important below when we endogenize the number of firms.

Irreversible Capital and Financial Frictions

The final ingredients of the model relate to the investment decision, which in this setup is equivalent to the number of firms in each sector, as each firm has a single unit of capital. As was mentioned before, all capital is foreign-owned. Irreversibility implies that investors have to decide the number of firms that will operate in each sector before the oil-related uncertainty—in this case, the government expenditure—is realized. If investors are risk neutral, the entry condition for the nontradable sector is

and for the tradable sector is

where E is the expectation operator. Substituting equations (18) and (19) we see that the stochastic profits in each of the sectors are

Now, we introduce financial frictions in the model by appealing to risk aversion. However, the results presented in the paper would be the same if instead of risk aversion we had assumed costly bankruptcy.

Firms require a rate of return that must compensate them for the volatility of profits. The rate of return will be proportional to their degree of risk aversion a. Assume that the entry condition is given by

where i stands for sectors N or T.

Risk aversion in this setup implies that the expected return has to exceed the risk-free rate by σ times the variance of profits. Notice that the profits of the tradable sector inherit the stochastic properties of ψ. Indeed, the variance is proportional to the variance of ψ. When the number of tradable firms increases, the value of ψ increases (diminishing the expected profit) but reducing its variance. Notice that this effect is different in the nontradable sector. An increase in the number of tradable firms increases the expected value of ψ, increasing the expected value of the profits, and reducing the volatility of ψ, reducing further the risk premium. Hence, in the nontradable sector the effect on expected profits and its variance go in the same direction, while they go in opposite directions in the tradable sector.


In this section we simulate numerically the model in order to understand the implications of changes in the size of the oil revenues, their volatility, and degree of risk aversion on the economy. In particular, we will study their effect on the number of firms in each sector, the utility of households, and the volatility of the real exchange rate. The parameters chosen for the simulation are as follows. The economy has a size of L = 1. We assume that the risk-neutral rate of return required is equal to 5 percent. We set a equal to 0.25 and not the more common 0.3-0.4 because in this model, capital is composed of tradable goods. In real life, capital also has nontradable components. However, assuming a demand for nontradables for investment purposes would have complicated the model unnecessarily.

We let the mean level of oil income move from 0.1 to 0.8. We assume that it is uniformly distributed with a coefficient of variation of zero, 0.25,0.75, and 0.875. We present our results for two cases: a risk-neutral case (σ = 0, which we interpret as no financial frictions) and an alternative case with a relatively large degree of risk aversion (σ = 15).

In Figures 2.5 and 2.6, the x-axis indicates the mean level of the oil income. Each figure has four panels: the top panel is the number of nontradable firms in equilibrium, the second one is the number of tradable firms, the third is the utility households derive from consumption, and the final panel is the volatility of the real exchange rate. In each panel we represent four curves representing the four degrees of volatility mentioned above, where the thicker lines represent greater volatilities.

Figure 2.5.Simulation: The Risk-Neutral Case

Source: Authors’ calculations.

Figure 2.6.Simulation: The Risk-Averse Case

Source: Authors’ calculations.

The risk-neutral case is studied first (Figure 2.5). As the mean level of oil income increases, the number of firms in the nontradable sector rises and the number in the tradable sector declines. The economy mo-notonically tends to specialize away from tradables, as would be predicted from the standard Dutch disease literature. Notice that utility is little affected by the level of oil income. Recall that we assume that the government spends the oil revenue in nontradable goods that households do not value and that capital is foreign owned. Hence, the utility of households is determined by the real wage and this does not change with oil income.

Volatility in this setup has surprising effects: it increases the number of firms in the tradable sector, lowers the number of firms in the nontradable sector, and increases utility! This contradicts the conventional wisdom that suggests that the increase in uncertainty should be bad for investment, especially if it is irreversible. This intuition does not hold in this model, a feature amply discussed in the neoclassical literature (see Hartman, 1972; Caballero, 1991; Caballero and Pindyck, 1996). The main reason for this effect is the fact that the profits of tradable firms are a convex function of ψ This means that greater volatility increases expected profits and investment in tradables. On the other hand, the profits in the nontradable sector can be concave or convex in ψ, and hence the volatility may have less salutary effects. For the parameters in this model the function is concave and the number of nontradable firms falls with volatility.11 Utility increases because wages are also convex in ψ. Interestingly, specialization in this model is not associated with volatility but, instead, solely with the average level of oil income. Hence, policies geared at stabilization are not welfare improving.

We now turn to the case with risk aversion (Figure 2.6). We assume a level of risk aversion of 15, which is high by conventional standards but significantly less than the coefficient of 40 required to explain the equity premium in developed countries. Remember that we take risk aversion to be a proxy for financial frictions in the economy.

Here we observe a similar initial impact of increases in the mean level of oil income on the number of firms in both sectors. However, notice that now volatility lowers the number of firms in both sectors, as investors demand a higher return to compensate for the greater variance in their returns, overwhelming the otherwise convex relationship of profits on ψ There is a point at which the tradable sector completely shuts down. We refer to this phenomenon as inefficient specialization. At that point, the number of firms in the nontradable sector also declines (and then rises very gradually), the volatility of the real exchange rate increases by a factor of more than 10, and utility collapses.

The mechanism that brings this about is a vicious circle between specialization and volatility. Remember that the volatility of the real exchange rate and ψ are inversely proportional to the number of firms in tradables. As the number declines profits become more volatile, but risk aversion now requires a higher risk premium, which lowers investment in tradables further and further increases volatility. The sector disappears because the cost of capital makes expected firm profits negative. This happens only in the tradable sector because its price is exogenously set. In the nontradable sector, the increase in risk premiums lowers investment, but this increases the price of nontradables, preserving profitability.

The decline in utility is related to the magnitude of the inefficiency of the specialization. A measure of this inefficiency is the difference between the number of tradable firms in the zero-volatility curve and the actual number of firms. Notice that as the average level of oil income increases, the economy would naturally specialize, and the gap between the efficient number of firms and the actual number declines. Hence, utility recovers, but does not reach the no-volatility level because the absence of the tradable sector makes the real exchange rate more volatile and lowers the investment in nontradables, where the number of firms also falls relative to the optimal level.

This result might help explain why countries that were very specialized in oil production such as Saudi Arabia, Nigeria, and Venezuela fared so poorly when oil income declined, while countries such as Indonesia, Mexico, and Norway were much less affected. The first group was specialized in oil and when oil income declined that specialization became much more inefficient, while the lack of a tradable sector created a level of volatility and risk premiums that did not allow for investment. Diversified countries could adjust with much smaller costs.

In conclusion, (i) specialization in the production of nontradables creates an economy with more volatile relative prices; (ii) financial frictions interact with this volatility, further specializing the economy as the stock of capital will respond to the greater macroeconomic volatility; (iii) this specialization may lead to the complete and inefficient disappearance of tradable production; and (iv) this specialization reduces the investment in nontradables—which will face a larger cost of capital—and lowers welfare.

Clearly, in this context, a higher level of oil revenues can become a curse if it leads the economy to inefficiently specialize. Moreover, stabilization policies can have large welfare implications.

IV. Policy Implications

The curse of natural resources has so far been explained as being caused by either rent seeking or diversification away from sectors enjoying increasing returns. Separately, expenditure stabilization policies have been advocated based on the welfare benefits of consumption smoothing. This paper has proposed an alternative mechanism for the curse that integrates it with the discussion of stabilization.

An economy that is diversified, in terms of having a significant non-oil tradable sector, will be much less affected by volatility in government domestic spending than an economy that is already fully specialized in nontradables. This is so because in a diversified economy shocks to nontradable demand can be accommodated through changes in the structure of production, while specialized economies have to rely on expenditure switching. We note that countries with more resource rents will naturally produce fewer tradables and hence are more likely to be naturally specialized.

However, the presence of quite standard financial frictions, such as costly bankruptcy, will make relative price volatility affect the cost of capital through risk premiums. This has three major consequences: first, it causes the economy to specialize further by making it harder and in some cases impossible for the tradable sector to access capital; second, it causes higher interest rates and less capital in the nontradable sector under specialization; third, it greatly increases the welfare losses caused by volatility.

The policy implications of this model are relatively straightforward. They have to do with avoiding inefficient specialization and reducing the costs of volatility. We separate our discussion into two parts. First we deal with first-best policies, which are based on reducing the distortions in the economy. Then we move to second-best policies, where we assume that the distortions are hard to remove and look at interventions that can improve welfare, given that the distortions make the market outcome inefficient. This is in general the spirit of the policies we are discussing below.

Fiscal Policy

To discuss fiscal policy it is important to distinguish three types of countries:

  • those that are naturally specialized, i.e., those that would specialize even in the absence of volatility;
  • those that are inefficiently specialized; and
  • those that are not specialized and would like to stay that way.

Naturally specialized countries

These countries have so much natural resource wealth that it does not make sense for them to engage in the production of other tradables, even in a first-best world. Some Gulf states might be in this category. However, in these countries, specialization makes relative prices very sensitive to the volatility in government spending. These countries would benefit from policies that stabilize government expenditures in order not to transmit volatility to the domestic market.

Inefficiently specialized countries

These countries are suffering from major welfare losses associated with the inability to develop the tradable sector, which makes the cost of capital high, even for the nontradable sector. Countries such as Venezuela and Nigeria may be in this category. Inefficient specialization is the product of a combination of factors: the level of government spending, the volatility of that spending, the “commercial-risk-free” interest rate to which firms in the economy have access, and the magnitude of financial inefficiencies.

These countries need to make a big effort to change the structure of the economy sufficiently to get it over the specialization frontier. Incremental changes may not be enough to improve matters significantly. There are three fiscal margins in which these economies could work: the average level of government spending, its volatility, and the “commercial-risk-free” interest rate.

With respect to the first point, it is useful to be precise in what we call “government spending” in the context of our model. We assumed that there was no domestic taxation. In a world with taxes, the relevant policy variable is the non-oil domestic primary deficit, i.e., the primary deficit excluding oil revenues and external spending by the government. This variable must be credibly lowered and stabilized in order to cross the specialization frontier. There is some trade-off between lowering the average value of this deficit and lowering its volatility. The more credible the reduction in volatility, the less important the required reduction in average spending will be. However, our simulations suggest that, for the parameters we studied, feasible cuts in average spending seem quantitatively more effective than feasible reductions in volatility, although the latter brings greater improvements in welfare. Hence, an ideal policy would rely on both: stabilization and cuts in the non-oil domestic primary deficit.

Furthermore, fiscal policy and debt management policies have an important role to play in affecting the interest rate. In our model we assumed a riskless international interest rate to which we added the commercial risks faced by each sector. In real life, on top of this commercial risk, there may be country risk associated with fears about the sustainability of public debt and of the government’s fiscal position. Factors that increase country risk will have the effect of moving the specialization frontier inward, causing the economy to specialize at lower average levels of spending and at lower volatilities.

Country risk often arises because of concerns about willingness to pay, or because the budget institutions are perceived as not capable of imposing an effective intertemporal budget constraint. Also, poor debt management can inefficiently expose a country to rollover or other financial risks and thus increase country risk. These problems should be avoided by any country. However, resource-rich countries are at risk of suffering heavily because the higher interest rate these problems cause may inefficiently keep them specialized in a low income-high volatility environment.

It is important to stress that countries in this category do not become specialized because wages in dollars are too high. In fact, if they were able to move to a diversified equilibrium, capital intensity would rise and wages would increase. Moving to a diversified economy is welfare enhancing; it does not imply real wage cuts.

Diversified economies

Economies in this category are characterized by having large nonresource tradable sectors. Examples in this category are countries such as Ecuador, Mexico, and Indonesia. In these countries, volatility in oil revenues will have smaller effects on relative prices, provided that they have relatively flexible domestic markets.12 Hence, the benefits of stabilization, other things being equal, are likely to be smaller than in the other two cases. However, these economies run the risk of becoming specialized if they increase the average non-oil primary deficit, its volatility, or have high country risk. For example, Indonesia has seen a big increase in country risk, while Ecuador is undergoing a big expansion in oil production in the context of very high country risk. Here the role of fiscal policy is to keep the economy diversified. It is mainly a preventive strategy that is called for.

First-Best Financial Policies

In our framework, the inefficient disappearance of the non-oil tradable sector is a consequence of financial frictions. Policies that minimize these frictions will allow the financial market to better manage the risks faced by the tradable sector and hence to displace the specialization frontier and to offer financing at lower cost, when the sector exists. We shall start with a list of first-best policies. Later, we will discuss some more interventionist policies.

Policies that complete financial markets by expanding the space for credible contracts will have particularly powerful effects on countries that are inefficiently specialized. These include

  • policies that make contract enforcement less costly, through effective judicial enforcement and extrajudicial conflict resolution;
  • policies that contain willingness-to-pay problems in financial markets such as facilitating the use and effectiveness of collateral; and
  • policies that efficiently reduce the cost of bankruptcy.

In addition, the high volatility that characterizes resource-rich economies is bound to make equity particularly valuable, as it allows better risk sharing between firms and investors. This agenda calls for

  • policies that facilitate direct investment, especially in the nonresource sector; and
  • policies to improve corporate governance so as to make equity claims more credible.

Second-Best Policies

While first-best policies are good for all countries, they are particularly valuable for inefficiently specialized economies. However, this inefficiency might also be addressed through second-best policies, that is, policies that assume that it is hard to credibly avoid fiscal spending volatility or financial frictions. What policies could improve welfare in such a context?

The central problem with the inefficiency described here is that the tradable sector is starved out of capital because it faces too high a real exchange rate volatility for debt markets to manage. Second-best policies involve in one way or the other the need to stabilize the profits of the tradable sector. Two main forms of intervention are discussed: trade policies and financial policies.

Trade policies

If the tradable sector disappears because of unstable expected profits, what role could trade policy play? Let us consider state-contingent protection composed of export subsidies and import tariffs that would go up in times of real appreciation and would be lowered in periods of real depreciation. The idea is that with a more stable expected profit, the sector could attract more capital. In “good times” when oil is high and the real exchange rate is appreciated, export subsidies and import tariffs would kick in and keep profits from collapsing. In “bad times” when the real exchange rate weakens, this extra support could be taken away. The policy is not unrelated to the “price bands” that protects sensitive agricultural products in many countries.

This logic has a strong partial equilibrium flavor. Does it survive the general equilibrium logic? After all, changes in relative prices in this model are equilibrium movements. In the logic of our model, these are the required prices needed to clear both the market for nontradables and the labor market. Interfering with these relative price changes would likely cause even larger changes in underlying prices in order to achieve the requisite reallocation.

However, the policy may survive the general equilibrium counter-forces provided it can assure investors that ex post returns will be sufficiently stable. The existence of more capital in the tradable sector does not prevent the sector from shedding labor in “good times” while it is there to absorb it in “bad times.” So, labor can still move between sectors, and since it will be working with more capital, it will therefore be more productive. Moreover, if the tradable sector is substantial, its ability to absorb and shed labor will lower the volatility of the economy, reduce interest rates, and support even more capital in the economy

Furthermore, while in our model movements in relative prices are caused by fundamental forces, in real life they may reflect other shocks and distortions that are less stabilizing. Protecting the tradable sector from their consequences may also be welfare enhancing.

Thus, we find some rationale for state-contingent protection. Obviously, one important drawback is that political economy forces may prevent the government from lowering tariffs and export subsidies when the real exchange rate is weak or may lead to excessive protectionism. However, it is not obvious that trade policy is the most efficient instrument. Further analysis is surely called for.

Financial support for tradables

Trade policy operates by changing the relative prices for goods faced by the whole economy. An alternative approach would be to offer some form of financial assistance to the tradable sector, without getting involved directly in the price-setting process in the goods market. A financial intervention that increases the stock of capital in the tradable sector can help push the economy out of specialization. One instrument to consider is some form of financial guarantee. This guarantee can be made contingent not on the idiosyncratic risks of a project but instead on the real exchange rate, so as to limit moral hazard. Through this mechanism, the probability of having to incur costly bankruptcies would be reduced, thus calling for lower interest rates. If this helps increase the size of the tradable sector, overall volatility might be reduced. However, as these guarantees might be exercised with significant frequency, at least in the transition to a more diversified economy, the potential fiscal liabilities must be considered. It is important to remember that, at least in the typical case, the real exchange rate appreciates in good times, when oil revenues are high and the government spends too much. Having this financial guarantee in place might constitute a mechanism that “punishes the government” for overspending booms and thus might have the right incentive properties from a political economy perspective: it gives the tradable sector a contingent claim on future oil booms that might otherwise lead to their demise.

Having said this, it is important to remember that in practice, money is fungible and that the tradable sector is not precisely defined in real life. Hence, narrowing the definition to include only exportables may be valuable for this and other reasons. After all, exportables are subject to international competition and have a much larger market into which they can potentially expand. However, these policies may collide with current multilateral trade agreements that prohibit export subsidies. Moreover, any initiative involving public monies will be subject to political economy distortions. Assuring that rent seeking does not distort these interventions is critical to their effectiveness.


    Aizenman, Joshua, and NancyMarion, 1999, “Volatility and Investment: Interpreting Evidence from Developing Countries,”Economica, Vol. 66, pp. 157–79.

    • Search Google Scholar
    • Export Citation

    Auty, Richard M., 1990, Resource-Based Industrialization: Sowing the Oil in Eight Developing Countries (Oxford: Clarendon Press).

    Caballero, Ricardo J., 1991, “On the Sign of the Investment-Uncertainty Relationship,”American Economic Review, Vol. 81, No. 1 (March), pp. 279–88.

    • Search Google Scholar
    • Export Citation

    Caballero, Ricardo J., 2000, “Macroeconomic Volatility in Latin America: A View and Three Case Studies,”NBER Working Paper No. 7782 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    Caballero, Ricardo J., and RobertS. Pyndick, 1996, “Uncertainty, Investment, and Industry Evolution,”International Economic Review, Vol. 37 (August), pp. 641–62.

    • Search Google Scholar
    • Export Citation

    Corden, W.M., 1982, “Exchange Rate Policy and the Resources Boom,”Economic Record (Australia), Vol. 58, No. 160 (March), pp. 18–31.

    • Search Google Scholar
    • Export Citation

    Corden, W.M., and J.P.Neary, 1982, “Booming Sector and Deindustrialization in a Small Open Economy,”Economic Journal, Vol. 92, pp. 825–48;reprinted inW.M.Corden, 1985, Protection, Growth and Trade: Essays in International Economics (Oxford: Basil Blackwell); and inW.M.Corden, 1992, International Trade Theory and Policy: Selected Essays of W. Max Corden (Aldershot: Edward Elgar).

    • Search Google Scholar
    • Export Citation

    de Ferranti, David, GuillermoE. Perry, DanielLederman, and WilliamF. Maloney, 2001, From Natural Resources to the Knowledge Economy—Trade and Job Quality (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    Duryea, Suzanne, 1998, “Children’s Advancement Through School in Brazil: The Role of Transitory Shocks to Household Income,”IADB Working Paper No. 376 (Washington: Inter-American Development Bank).

    • Search Google Scholar
    • Export Citation

    Flug, Karnit, AntonioSpilimbergo, and ErikWachtenheim, 1998, “Investment in Education: Do Economic Volatility and Credit Constraints Matter?”Journal of Development Economics (Netherlands), Vol. 55, No. 2 (April), pp. 465–81.

    • Search Google Scholar
    • Export Citation

    Gavin, Michael, and RicardoHausmann, 1996, “Securing Stability and Growth in a Shock Prone Region: The Policy Challenge for Latin America,”IADB Working Paper No. 315 (Washington: Inter-American Development Bank).

    • Search Google Scholar
    • Export Citation

    Gavin, Michael, and RicardoHausmann, 1998, “Nature, Development, and Distribution in Latin America: Evidence on the Role of Geography, Climate, and Natural Resources,”IADB Working Paper No. 378 (Washington: Inter-American Development Bank).

    • Search Google Scholar
    • Export Citation

    Hartman, R., 1972, “The Effects of Price and Cost Uncertainty on Investment,”Journal of Economic Theory, pp. 5258–66.

    Hausmann, Ricardo, 2001, “Venezuela’s Growth Implosion: A Neo-Classical Story?,”Kennedy School of Government Working Paper (August) (Cambridge, Massachusetts: Harvard University).

    • Search Google Scholar
    • Export Citation

    Hausmann, Ricardo, AndrewPowell, and RobertoRigobon, 1993, “An Optimal Spending Rule Facing Oil Income Uncertainty (Venezuela),” in External Shocks and Stabilization Mechanisms,ed. byE.Engel and P.Meller (Washington: Inter-American Development Bank and Johns Hopkins University Press), pp. 113–71.

    • Search Google Scholar
    • Export Citation

    Higgins, M., and J.G.Williamson, 1999, “Explaining Inequality the World Round: Cohort Size, Kuznets Curves, and Openness,”NBER Working Paper No. 7224 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    Inter-American Development Bank, 1995, Economic and Social Progress in Eatin America,1995 Report (Washington).

    Johnsen, Christopher, KennethA. Shepsle, and BarryR. Weingast, 1981, “The Political Economy of Benefits and Costs: A Neoclassical Approach to Distributive Politics,”Journal of Political Economy, Vol. 89, No. 4.

    • Search Google Scholar
    • Export Citation

    Karl, Terry Lynn, 1997, The Paradox of Plenty: Oil Booms and Petro-States (Berkeley: University of California Press).

    Lane, Philip R., and AaronTornell, 1999a, “Voracity and Growth in Discrete Time,”Economics Tetters, Vol. 62, No. 1 (January), pp. 139–45.

    • Search Google Scholar
    • Export Citation

    Lane, Philip R., and AaronTornell, 1999b, “The Voracity Effect,”American Economic Review, Vol. 89 (March), pp. 22–46.

    Maddison, Angus, 1995, “Monitoring the World Economy: 1820-1992,” in Development Centre Studies (Paris and Washington: Organization for Economic Cooperation and Development).

    • Search Google Scholar
    • Export Citation

    Manzano, Osmel, and RobertoRigobon, 2001, “Resource Curse or Debt Overhang?,”NBER Working Paper No. 8390 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    Matsuyama, Kiminori, 1992, “Agricultural Productivity, Comparative Advantage and Economic Growth,”Journal of Economic Theory, Vol. 58 (December).

    • Search Google Scholar
    • Export Citation

    Ramey, Garey, and ValerieA. Ramey, 1995, “Cross-Country Evidence on the Link Between Volatility and Growth,”American Economic Review, Vol. 85, No. 5, pp. 1138–51.

    • Search Google Scholar
    • Export Citation

    Sachs, Jeffrey, and AndrewWarner, 1995a, “Economic Reform and the Process of Global Integration,”Brookings Papers on Economic Activity: 1, Brookings Institution, pp. 1–118.

    • Search Google Scholar
    • Export Citation

    Sachs, Jeffrey, and AndrewWarner, 1995b, “Natural Resource Abundance and Economic Growth”, Harvard Institute of Economic Research Discussion Paper, No. 517 (Cambridge, Massachusetts: Harvard Institute for International Development).

    • Search Google Scholar
    • Export Citation

    Tornell, Aaron, and AndresVelasco, 1995, “Fiscal Discipline and the Choice of the Exchange Rate Regime,”European Economic Review (Nethelands), Vol. 39, No. 3/4, pp. 759–70.

    • Search Google Scholar
    • Export Citation

    von Hagen, Jiirgen, and IanJ. Harden, 1997, “National Budget Processes and Fiscal Performance,”European Economy, Reports and Studies: Towards Greater Fiscal Discipline, No. 3 (Brussels: European Commission, Directorate-General for Economic and Financial Affairs), pp. 311–408.

    • Search Google Scholar
    • Export Citation

We are indebted to Dani Rodrik and Andres Velasco for useful comments.


For example, Sachs and Warner (1995b) estimated that countries fully dependent on the export of primary products grew about 2.5 percent per year more slowly in the 1970–1989 period. Gavin and Hausmann (1998) and Higgins and Williamson (1999) find a strong relationship between resource intensity and inequality.


For example, Manzano and Rigobon (2001) attribute the low growth to a debt overhang associated with overborrowing during the boom of the 1970s.


Alternatively, it may want to diversify into resource-intensive industrialization, which has failed in many countries (Auty, 1990).


Whether volatility on a variable causes the expected value of some other variable to increase or decrease depends on the concavity or convexity of the relationship between the two. Concavity lowers the expected value while convexity increases it. Hence, if the profit function is convex, then volatility is good for expected profits and leads to more investment.


It makes sense to assume that if prices cannot be readjusted, capital should be even harder to adjust.


Whether government spending involves directly hiring workers or buying services from the nontradable sector is immaterial for the analysis. Workers are fully employed and indifferent as to the sector of employment.


Volatile government spending will affect welfare through its general equilibrium effects.


We will relax this assumption in the next section.


In fact, for most reasonable sets of parameters the schedule is concave. The relationship becomes convex when the number of nontradable firms approaches zero.


After all, in our model we assumed sufficient labor mobility to clear the labor market.

    Other Resources Citing This Publication