Too Sunny In Latin America?
The IMF’s Overly Optimistic Growth Projections and Their Consequences

Dean Baker and David Rosnick[1]

September 16, 2003

 

This paper examines the track record of the IMF’s growth projections for Latin America over the last two decades and its implications for policy. The paper finds that:

Given the evidence of a systematic upward bias in IMF growth projections for Latin America, the paper recommends that governments adjust IMF annual growth projections downward by 1.6 percentage points in planning policy. This should provide a more accurate basis for designing macroeconomic policy.    


Introduction

            In order to help governments, businesses, and other economic actors plan for the future, the International Monetary Fund (IMF) regularly produces projections of economic growth for its member countries. In principle, policy makers could treat the IMF’s numbers as unbiased projections for output growth. Such projections could be enormously helpful to governments when setting their tax and spending policies, and also for private firms and individuals making investment and saving decisions.

            However, the evidence of the last two decades suggests that there is reason to question whether the IMF’s growth projections for Latin America can be viewed as unbiased. In the seventeen comparisons available for the years from 1986 to 2002, the IMF projections from the prior year have overstated growth thirteen times and understated growth only four times, as shown in Table 1.[2]

Table 1

IMF Year Ahead Growth Projections For the Western Hemisphere
and Actual GDP Growth

Year

Projected

Actual

Difference

 

Growth

Growth

(percentage points)

1986

4.2%

4.3

-0.1

1987

3.5

2.2

1.3

1988

4.7

0.4

4.3

1989

3.5

1.0

2.5

1990

3.2

-0.1

3.3

1991

4.2

2.9

1.3

1992

3.3

2.8

0.5

1993

4.2

4.0

0.2

1994

2.6

5.0

-2.4

1995

3.4

1.8

1.6

1996

3.7

3.6

0.1

1997

4.8

5.2

-0.4

1998

5.1

2.3

2.8

1999

4.3

0.2

4.1

2000

3.5

4.0

-0.5

2001

4.7

0.6

4.1

2002

4.4

-0.1

4.5

All Years

4.0

2.4

1.6

                                Source: World Economic Outlook, various years (April projections).[3]

           These data suggest a strong tendency for the IMF’s projections to overstate actual growth. On average, the projected growth rate has exceeded the average growth rate by 1.6 percentage points as shown below:[4] Both the frequency and  size of the overstatement are highly significant, providing solid evidence of an optimistic bias in IMF projections.[5]

            The existence of a statistically significant bias in IMF growth projections for Latin America can lead to serious policy mistakes, with both short-run and long-run consequences. The short-run issue is fairly straightforward. When the IMF makes policy recommendations for an economy, it weighs the risks from higher inflation against the costs of excessive unemployment. If it systematically overestimates the economy’s growth rate, then it will exaggerate the risks of inflation – which is associated with more rapid growth – and underestimate the prospect of higher unemployment. This error will lead the IMF to recommend policies that are more contractionary than would be appropriate, if it had an accurate assessment of a country’s economic prospects.

            In other words, if a Latin American country can actually expect to see a growth rate of 2.0 percent, then the IMF’s forecast bias would lead it to project a 3.6 percent growth rate. Since the IMF is expecting more growth than will actually materialize, it will recommend tighter fiscal policy (higher taxes and/or spending cuts) than would be appropriate given the country’s actual growth rate. It may also recommend higher interest rates than would be appropriate, because the Fund wrongly sees a need to slow the growth of the economy. If Latin American countries follow recommendations based on the IMF’s overly-optimistic growth projections, then they are likely to experience slower growth than necessary. These policies – fiscal austerity and high interest rates – have the effect of slowing growth. Therefore, the IMF’s exaggerated growth projections may be a direct cause of slower growth in Latin American countries.

            There can also be a long-term cost to the IMF’s overly optimistic growth projections. Such projections may lead countries into debt traps, which they could avoid if they relied on more accurate projections of growth. This can be seen with a simple example:

Suppose a country like Brazil tries to chart a fiscal course that is consistent with meeting its debt service obligations. Brazil has a debt-to-GDP ratio of approximately 60 percent and faces a real interest rate of approximately 12 percent.[6] If the IMF projects that Brazil’s economy will grow 3.5 percent annually, then this means that Brazil can keep its debt to GDP ratio constant if its government runs a primary budget surplus (net of interest payments) equal to 5.1 percent of GDP.

            However, if Brazil’s economy only grows by 2.0 percent, which would be expected given the bias in IMF projections, then a primary budget surplus of 5.1 percent of GDP would be insufficient to keep the debt-to-GDP ratio constant. In this scenario, the debt to GDP ratio would continue to rise, even if Brazil ran a primary budget surplus equal to 5.1 percent of GDP. After one year, the debt-to-GDP ratio would have risen to 60.9 percent of GDP. This rise in the debt-to-GDP ratio would require an even larger primary budget surplus the following year. However, if the target is again based on an overly optimistic projection from the IMF, then the surplus would still be insufficient to stabilize the debt-to-GDP ratio. If Brazil continued to set surplus targets based on overly optimistic growth projections, then each year its debt to GDP ratio would rise, as would its primary surplus target. After ten years, its primary surplus target would reach 5.9 percent of GDP, and after twenty years its surplus target would hit 6.7 percent of GDP, as shown in figure 1.This surplus would be the equivalent of the U.S. government running an annual primary budget surplus of $737 billion.

Figure 1

            Source: Authors’ calculations, see text.

            A surplus of this magnitude represents an enormous drain on an economy. It means that the tax burden has exceed the level of expenditures necessary to maintain government services by an amount equal to nearly 7.0 percent of GDP. While the original burden resulting from a primary budget surplus of 5.1 percent of GDP is already enormous, the future burden in this scenario is considerably larger. Even if the current burden is viewed as acceptable, compared to the available alternatives, a government that recognized that this burden will only grow through time might choose different options.

            In this sense, the overly optimistic projections from the IMF may lead a country to follow a long-term path that is unsustainable, and one that it would not choose if it had more accurate projections. In the case of Brazil outlined above, the cost of a primary budget surplus equal to 5.1 percent of GDP might be viewed as acceptable, if it were associated with a stable debt-to-GDP ratio. However, if the government recognized that this primary surplus would be associated with a rising debt to GDP ratio, requiring even larger primary budget surpluses for future years, then it may opt for a different long-term path. The IMF’s biased projections may lead governments to choose paths that they never would have followed, had they possessed more accurate growth projections.


Conclusion

            There has been a systematic bias in the IMF’s growth projections for Latin America over the last seventeen years. On average, this bias has led the IMF to overestimate actual growth by 1.6 percent annually. If government policy has been based on these biased projections, then there may be large costs in both the short-run and long run. In the short-run, governments may adopt policies that are more contractionary than is appropriate. This means that there are higher taxes and more spending cuts than necessary and possibly higher interest rates as well. The result of such contractionary policies would be slower growth.

            In the long run, the IMF’s biased projections may lead countries to fall into a debt trap, where they try to continue to service an unmanageable debt, rather than making an earlier attempt at restructuring. Overly optimistic growth projections will make a debt burden appear more manageable than it actually is. Based on overly optimistic projections, countries may try to service debt levels that they would recognize as unmanageable if they possessed unbiased growth projections. In this sense, the IMF’s biased growth projections may lead countries to follow a long-term path that they would quickly reject if they possessed more accurate information.

            Insofar as IMF growth projections provide a basis for national policy, there are likely to be serious costs to their upward bias. Countries should recognize this bias and adjust for it when determining national economic policy.

 

 

[1] Dean Baker and David Rosnick are, respectively the co-director and a research associate at the Center for Economic and Policy Research in Washington, D.C. John Schmitt and Mark Weisbrot gave helpful comments on earlier versions of this paper.

[2] The data are taken from the IMF’s World Economic Outlook (various years). The table compares the spring projections from the prior year with the most recent available data for growth in the subsequent year.

[3] The 1993 growth projection was made for growth calculated on an exchange-rate GDP basis. The actual growth measure uses a purchasing power parity measure of GDP.

[4] The IMF has noted some tendency for its projections to overstate actual growth. For example, the December, 2001 World Economic Outlook includes a discussion of the accuracy of IMF’s projections in the nineties (pp 37-39). This discussion notes that annual regional growth projections for the decade have been on average 0.4 percentage points above actual growth in the subsequent year. It notes that this error, as well as regional forecasting errors, are not statistically significant, except in the case of Africa. More recently, the IMF’s Independent Evaluation Office recognized that the IMF’s growth projections have tended to be overly optimistic for countries for which it designs programs (International Monetary Fund, Independent Evaluation Office, 2003. Fiscal Adjustment in IMF-Supported Programs, pp 73-75.)

[5] The probability of randomly overstating growth in thirteen of seventeen years is less than 2.5 percent. The estimate of a 1.6 percent upward bias in the IMF projections has a t-statistic of 3.2, making it significant at the1 percent level.  

[6] These numbers are probably somewhat more optimistic than the situation Brazil actually faces. Presently its debt to GDP ratio is somewhat below 60 percent, although this figure fluctuates with Brazil’s currency, since approximately one-third of its debt is denominated in dollars. However, on average Brazil has paid a real interest rate of more than 16 percent on its debt over the last decade (see Weisbrot, M. and Baker, D, 2002."Paying the Bills In Brazil: Does the IMF’s Math Add Up?”,  Washington, D.C.: Center for Economic and Policy Research).