Every time there is a debt crisis, policymakers, investors, and the international community ask the same question: “How did debt in country X get to be so high?” The purpose of this chapter is to answer that question by describing and quantifying some of the factors that lead to debt accumulation in emerging economies. Essentially there are two factors that determine debt growth. The first factor is the budget deficit, and the second is an unexplained residual entity called “stock-flow reconciliation.”
These two components of debt growth display some characteristics that are surprising, indeed. Specifically, quantification of the stock-flow reconciliation shows that, contrary to what is commonly thought, this is not a residual entity of limited importance, but often a key determinant of debt explosions. The discussion of stock-flow reconciliation is somewhat technical, but its key message should be clear: although most of the policy debate focuses on measured deficits, there is a large share of change in debt that cannot be explained by the deficit, and a better understanding of this “unexplained part of debt” is key to preventing debt crises.
With respect to the behavior of the component of debt accumulation, the chapter investigates the determinants of cross-country differences in the cyclicality of the budget deficit. The findings in regard to this cast doubt on the conventional wisdom. In particular, the chapter shows that the use of appropriate statistical techniques challenges the standard finding that fiscal policies are countercyclical in developed countries and procyclical in developing countries. Again, this seemingly technical discussion has important policy implications, as it may lead to the devising of policies that could reduce the high income and consumption volatility that characterizes most developing countries.
Some Simple Debt Arithmetic
The answer to the question “How do countries get into debt?” may seem trivial.[1] Anyone who has taken even the most basic economics course knows that countries accumulate debt whenever they run a budget deficit (i.e., whenever public expenditures are greater than revenues) and reduce their debt when they run a budget surplus. In fact, the standard Economics 101 textbook debt accumulation equation states that the change in the stock of debt is equal to the budget deficit (for those who like equations, this can be expressed as DEBTt – DEBTt–1 = DEFICITt) and that the stock of debt is equal to the sum of past budget deficits.
However, anyone who has worked with actual debt and deficit data knows that the equation presented above rarely holds and that debt accumulation can be better described as the sum of deficit plus an unexplained residual. Formally, this can be written as
DEBTt – DEBTt–1 = DEFICITt + SFt ,
where SFt measures the stock-flow reconciliation, a cumbersome name that comes from the fact that this residual entity reconciles the deficit, which is a variable measured over a period of time (i.e., a “flow” variable), with debt, which is a variable measured at a given moment (i.e., a “stock” variable).
Clearly, the textbook equation is a good approximation for debt accumulation only if one assumes that SFt is not very large. In fact, the stock-flow reconciliation is often considered to be a residual of little importance. Is it really the case that the stock-flow reconciliation doesn’t play a major role? Should policymakers not worry about stock-flow reconciliations and just focus on the deficit? One of the main findings of this chapter is that the stock-flow reconciliation does matter and that policymakers do need to take account of it.
Before moving to a systematic analysis of the stock-flow reconciliation, it is useful to consider three examples. In December 1998, Brazil’s net debt-to-GDP ratio stood at approximately 42 percent of GDP, but by January 1999 this ratio had surpassed 51 percent of GDP. Could the Brazilian government have run a deficit of almost 10 percent of GDP in just one month? This seems highly improbable.
Likewise, in 2001 Argentina’s debt-to-GDP ratio stood at just above 50 percent of GDP, and by 2002 the country’s debt was well above 130 percent of GDP. Conversely, in 2004 Argentine debt totaled 140 percent of GDP, but by the end of 2005 the country’s debt had fallen to 80 percent of GDP. Was it truly possible for the Argentine government to run a deficit of 80 percent of GDP in one year and a surplus of 60 percent of GDP less than two years later?
Uruguay presents a third case that is puzzling at first glance. In March 2002, Uruguay’s debt-to-GDP ratio was 55 percent, yet by the end of 2003 the country’s debt had soared to 110 percent of GDP. Could the Uruguayan authorities have run a deficit of 55 percent of GDP in less than two years?
These jumps in debt were clearly not due to standard budget deficits. In the case of Brazil, the sudden jump in debt resulted from the currency devaluation that followed the abandonment of the Real Plan. In the case of Uruguay, debt surged because of both a currency devaluation (which led to an increase in the debt-to-GDP ratio of approximately 40 percentage points) and the resolution of a banking crisis (which had a cost of approximately 18 percent of GDP). In the case of Argentina, the causes are similar but even more complex (see Box 3.1).
Campos, Jaimovich, and Panizza (2006) express the stock-flow reconciliation in terms of GDP and show that, on average, the change in debt explained by stock-flow reconciliation is 5 percent of GDP (Figure 3.1), clearly a residual of no small importance! The highest values for this reconciliation are in Sub-Saharan Africa (almost 9 percent of GDP), Latin America and the Caribbean (above 7 percent of GDP), and the Middle East and North Africa (7 percent of GDP). These high values may be driven by a few episodes (due to either exceptional events or measurement errors) with very large values for stock-flow reconciliation. There are in fact some observations in which this residual entity is well above 200 percent of GDP. The green bars in Figure 3.1 report average values of the stock-flow reconciliation obtained by dropping the top and bottom 2 percent of the distribution of this variable. As the figure shows, extreme values are irrelevant for the advanced economies and East Asia and the Pacific but are important for other regions. Excluding outliers substantially lowers the averages for the Middle East and North Africa, Latin America and the Caribbean, and Sub-Saharan Africa, but the last two regions remain the ones with the highest average stock-flow reconciliations (4 and 6 percent of GDP, respectively). Considering the whole sample of countries, the figure shows that excluding extreme values brings the average stock-flow reconciliation to 3 percent of GDP. This is much lower than in the sample with outliers but still a substantial figure indicating that, in the average country year, debt grows three percentage points of GDP faster than is implied by the budget deficit.[2]
Another way to assess the importance of the stock-flow reconciliation is to divide both sides of the equation discussed at the beginning of this chapter by GDP and use it to estimate the following statistical model:
di,t = β * deft,i + αi + εt,i ,
where di,t is the change in debt divided by GDP,[3] deft,i is deficit over GDP, αi is a country-specific parameter (this parameter controls for the fact that the data come from different sources, that countries have different levels of debt, and that they use different methodologies for computing debt and deficit), and εt,i is the error term of this statistical model, which should be interpreted as the stock-flow reconciliation. If the stock-flow reconciliation is unimportant, the estimation of the above equation should fit the data well 
and yield a value of β close to one. Figure 3.2 shows the results obtained estimating the above equation using the sample without outliers. The diagonal line in the figure indicates the value of β and shows that this parameter takes a value slightly greater than one. The position of the points gives a graphical representation of the “goodness of fit” of the statistical model. Points that are close to the line indicate observations for which the data fit the model well, and points that are far away from the line indicate observations for which the data fit the model poorly. As the figure shows, there are large cross-country differences. In the case of advanced economies, the points tend to be close to the line, indicating a relatively good fit. However, in Latin America and the other developing countries, the points are far away from the line, indicating that in these countries deficits do not do a good job of explaining the change in debt.
A more precise measure of goodness of fit is the statistical model’s R2. This statistic measures the share of variance of the dependent variable (di,t) which is “explained” by the independent variables (in this case defi,t). An R2 value of one indicates a perfect fit with the independent variables, explaining the totality of the variance of the dependent variable, whereas an R2 value of zero indicates that there is no relationship between the dependent and the explanatory variables. Figure 3.3 displays the R2 values obtained by estimating the equation described above for different subsamples of countries. It shows that when all countries are pooled together, the R2 value is just above 0.07, indicating that deficits explain less than 8 percent of the change in debt (and the stock-flow reconciliation more than 90 percent)—a very poor fit for an equation which is often considered to be an identity.[4]
As the figure shows, the region with the poorest fit, according to the model, is Sub-Saharan Africa. In this group of 29 countries, the deficit explains only 3 percent of the variance of the change in debt. In the cases of Latin America and the Caribbean (25 countries) and South Asia (5 countries), the deficit explains between 5 and 6 percent of the variance of the change in debt. The developing region with the best fit is Eastern Europe and Central Asia (15 countries), for which the deficit explains 23 percent of the variance of the change in debt. Only in the advanced economies (24 countries) does the deficit explain more than one-quarter of the within-country variation in the change in debt, but even in this case, the regression can explain only half of the variance of the dependent variable, suggesting that the stock-flow reconciliation is as important as the deficit in explaining changes in debt.[5]
It is also interesting to explore whether the difference between deficit and change in debt is associated with debt growth. In other words, is the stock-flow reconciliation one of the main determinants of debt explosions? A look at the relationship between the growth rate of debt over GDP and the ratio of deficit to change in debt shows that for countries with relatively low levels of debt growth (below 5 percent per year), the deficit explains between 70 and 80 percent of the change in debt. However, when debt starts growing at a faster rate, the share of debt explained by the deficit drops dramatically. When annual debt growth reaches 10 percent of GDP, the deficit explains less than 40 percent of the change in debt.
Most of the preceding discussion has focused on the change in debt divided by GDP rather than on the growth of the debt-to-GDP ratio. The first concept focuses on changes in debt without considering the effect of nominal GDP growth, while the second focuses on the change in debt relative to the change in GDP.[6] While the difference between these two measures may seem to be a technical one, both are useful concepts. The first makes it possible to estimate precisely the difference between deficit and change in debt without the need to isolate the effects of GDP growth and inflation. The second allows debt growth to be decomposed and the relative contributions of each of its determinants to be evaluated. Furthermore, it is the variable commonly used to assess fiscal sustainability.
Figure 3.4 focuses on the second measure and decomposes the growth of the debt-to-GDP ratio into five components: inflation, real GDP growth, stock-flow reconciliation, interest expenditure, and primary deficit (the last two components add up to the total deficit).[7] Inflation and GDP growth are the main mechanisms of debt reduction (there is also a small positive effect of primary surpluses in Latin America, the advanced economies, and East Asia, and a larger effect of this variable in the Caribbean), and the effect of inflation dominates that of real GDP growth in every region of the world.[8] The effect of inflation is particularly large in Sub-Saharan Africa, Eastern Europe and Central Asia, the Middle East and North Africa, and Latin America. In the advanced economies and the Caribbean, interest payments are the main determinant of debt accumulation, and in South Asia, the budget deficit (primary balance plus interest payment) is the main determinant. In all other regions of the world, the stock-flow reconciliation is the key determinant of debt accumulation. In Latin America, for instance, the total deficit adds up to 2.4 percent of GDP, wheareas the stock-flow reconciliation equals 5.5 percent of GDP.
Figure 3.5 decomposes debt growth for Mexico and six South American countries and shows that in four of these countries, the stock-flow reconciliation is the main determinant of debt growth. The exceptions are Brazil, Colombia, and Mexico, where the main determinant of debt growth is interest payments (in Mexico, the amounts for interest payments and the stock-flow reconciliation are basically the same). All the countries, with the exception of Colombia, have primary surpluses, which are a substantial source of debt reduction in Brazil and Chile. Only in Chile is real GDP growth a substantial source of debt reduction, and in fact inflation is the main source of debt reduction in all seven countries. 
Figure 3.6 repeats the experiment for five countries located in Central America and the Caribbean. Although in three countries (The Bahamas, Costa Rica, and Guatemala) the deficit (again, primary balance plus interest payment) is the main determinant of debt growth, in two of them (Guatemala and Costa Rica), the stock-flow reconciliation is nevertheless an important determinant of debt growth (representing 30 and 90 percent of the deficit, respectively). Panama has a primary surplus but large interest payments, which dominate the stock-flow reconciliation as a main determinant of debt growth (which, however, remains an important factor). 
In El Salvador, the stock-flow reconciliation is the main determinant of debt growth. Focusing on the factors that contribute to debt reduction, inflation is the main determinant of debt reduction in Costa Rica, El Salvador, Guatemala, The Bahamas, and Panama. Figure 3.7 decomposes debt growth year by year by aggregating data for the seven largest Latin American economies (Argentina, Brazil, Chile, Colombia, Peru, Mexico, and Venezuela). As expected, the stock-flow reconciliation is shown to have a tendency to be very large at the time of crisis or just after a crisis. In particular, it was very high in the two years that followed the Tequila crisis (1995–1996), the year of the Russian crisis (1998), and the year of the Brazilian devaluation (1999) and reached epic levels at the time of the Argentine crisis (2002–2004). Interestingly, the stock-flow reconciliation was basically zero (or even negative) in tranquil years like 1997 or 2005.[9]
Is It Possible to Explain What Drives the Unexplained Part of Debt?
Having documented that there are large differences between deficits and changes in debt, it is interesting to explore the determinants of these differences. Campos, Jaimovich, and Panizza (2006) use a statistical model that tries to explain the determinants of the stock-flow reconciliation using three groups of variables.[10]
The first set of variables aims at capturing balance sheet effects due to the interaction of currency depreciations and the presence of foreign currency debt. The idea is that currency devaluations should lead to large stock-flow reconciliations in countries with high levels of foreign currency debt. Figure 3.8 plots the main results and shows that this prediction is supported by the data. The figure shows that, assuming a real depreciation of 30 percent (not an uncommon event in some developing countries), in countries with no foreign currency debt, the depreciation has basically no effect on the stock-flow reconciliation (less than 1 percent of GDP and not statistically significant). In countries with moderate levels of foreign currency debt, a similar devaluation leads to a difference between deficit and debt of approximately 3 percent of GDP. Finally, in countries with high levels of foreign currency debt (i.e., the top third of the distribution), a 30 percent depreciation is associated with a stock-flow reconciliation equal to 10 percent of GDP.[11]
The second set of variables attempts to capture the effect of the resolution of sovereign default episodes. As default episodes result in partial debt cancellation (e.g., Sturzenegger and Zettelmeyer, 2005a, show that recent defaults implied “haircuts” that ranged from 13 to 73 percent of outstanding bonded debt), they should be associated with negative stock-flow reconciliations. In fact, Figure 3.8 shows that defaults are associated with a negative stock-flow reconciliation of approximately 2 percent of GDP.
The last explanatory variable explores the role of banking crises. These are important events, because they generate a series of contingent liabilities and other off-balance-sheet activities that can translate into debt explosions (see Box 3.1). In fact, the statistical model of Campos and her colleagues shows that the average banking crisis is associated with a stock-flow reconciliation of almost 3 percent of GDP.
While these are interesting results that suggest that building a safer debt structure and implementing policies aimed at limiting the creation of contingent liabilities are key to avoiding debt explosions, it is important to note that the variables discussed explain only 20 percent of the variance in the stock-flow reconciliation (country-specific factors explain another 30 percent of this variance).[12]
There are two possible reasons why the statistical model described above does such a poor job of uncovering the determinants of the unexplained part of debt. The first has to do with the fact that measurement errors that lead to an underestimation of the deficit are more important in some countries than in others. This is probably related to the fact that developing countries have less transparent accounting and budgeting systems, which make it possible to hide some liabilities. This is consistent with the findings of Aizenman and Powell (1998), who suggest that governments have incentives to misreport public expenditure and that this comes back to haunt them as debt is subsequently reassessed.
The second possible reason for the limitations of the statistical model is that the importance of contingent liabilities that lead to debt explosions varies across countries and that the controls included in the statistical exercise described above do not capture all the possible sources of contingent liabilities. One variable that is likely to be important, for example, but that is not included in the analysis is the effect of court decisions that force a government to make payments that it has not budgeted for (see Box 3.1 for the role of courts in Argentina and Chapter 9 for a discussion of how courts may affect the budget).[13]
How Should Deficits Move, and How Do They Move in Reality?
As noted in the first chapter of this report, most economists agree that a sound fiscal policy should exhibit countercyclical behavior. By running deficits in bad times and surpluses in good times, countries can smooth consumption, reduce the volatility of output, and minimize tax distortions. But the benefits of countercyclical policies are not limited to their welfare effects in terms of stabilization of the business cycle. Such policies can also be an effective strategy for limiting the growth of public debt.[14] This is because in the presence of procyclical fiscal policies, a stable debt ratio would require expenditure cuts (or tax increases) during recessions, and such adjustments are extremely difficult to implement. As a consequence, procyclical fiscal policies may contribute to snowballing budget deficits and debt levels that are eventually resolved with debt crises, high inflation, or outright default.[15]
If procyclical policies are so bad and countercyclical policies so good, one would expect all countries to adopt countercyclical policies. However, this does not seem to be what actually happens. Gavin and Perotti (1997) compare the main characteristics of fiscal policy in Latin America and the advanced economies and find that, while in the latter group of countries, policies tend be countercyclical, Latin America is characterized by procyclical fiscal policies. In particular, Gavin and Perotti use a statistical model that estimates how GDP growth affects a country’s fiscal balance and find that in advanced economies, when a country’s GDP grows by 1 percent, its budget surplus grows by approximately 0.4 percent. In Latin America, in contrast, they find that there is basically no correlation between GDP growth and changes in the budget balance. They argue that the lack of a positive relationship between growth and a country’s fiscal balance suggests that discretionary fiscal policies are procyclical because, in the absence of such a procyclical response, a country’s fiscal balance would automatically be positively correlated with growth.
Figure 3.9 updates the estimations of Gavin and Perotti (1997), with the blocks showing the point estimates and the vertical bars the respective 95 percent confidence intervals. The first block shows that in advanced economies, a 1 percentage point increase in output growth is associated with an increase in the fiscal surplus of 0.2 percentage points (an effect smaller than what Gavin and Perotti found, but still large and statistically significant). The second block focuses on developing countries and shows that, while the relationship between GDP growth and fiscal balance is still positive and significant, the point estimate indicates a much lower elasticity than that of industrial countries. In this case, a 1 percentage point increase in output growth is associated with an increase in the fiscal surplus of 0.08 percentage points. The next three vertical bars split the sample of developing countries into three subgroups and show that developing countries are far from uniform in regard to the way the fiscal balance responds to GDP growth.
The first group (middle-high-income economies) consists of 18 emerging markets.[16] In this group of countries, there is no significant correlation between GDP growth and fiscal balance. In fact, this is the group of countries with the lowest level of countercyclicality. The second group includes 25 middle-low-income countries (see Kaminsky, Reinhart, and Végh, 2005, for a full list of countries); in these countries the correlation between GDP growth and the budget balance is about three-quarters of the level found in the advanced economies but still large, positive, and statistically significant. The third group focuses on low-income economies; here the correlation between GDP growth and fiscal balance is lower, but still larger than in the emerging market countries and significantly greater than zero.
The last two vertical bars in Figure 3.9 focus on Latin America and the Caribbean. The first of these two vertical bars uses the largest possible sample of countries and, contrary to the findings of Gavin and Perotti (1997), shows a positive and statistically significant correlation between output growth and changes in the budget balance. The last vertical bar restricts the sample to the 13 countries (mostly emerging markets) used by Gavin and Perotti and confirms their result of a low and not statistically significant correlation between output growth and changes in the budget balance. This suggests that Gavin and Perotti’s finding was driven by the behavior of Latin American emerging markets and that Latin America is not different from other developing regions of the world, where procyclicality is higher in emerging market countries.
While the foregoing discussion has confirmed that there are large differences between the degree of fiscal procyclicality in developing and advanced economies, it has also shown that there are large differences within the sample of developing countries and that one should allow for heterogeneous effects when trying to estimate the degree of fiscal cyclicality in this group of countries. But lumping together different types of developing countries is not the only problem with standard analyses of the difference in procyclicality between developing and advanced economies. There is also a problem with the variable that is usually used to measure the cyclicality of fiscal policy. Kaminsky, Reinhart, and Végh (2005) criticize the use of the budget balance to measure cyclicality and argue that procyclicality should be studied by looking at the behavior of public expenditure.[17] According to their definition, countercyclical policies would be associated with a negative correlation between GDP growth and the growth rate of government expenditure, while procyclical policies would be associated with a positive correlation between these two variables.[18]
Figure 3.10 focuses on the cyclicality of public expenditure and shows that in advanced economies, there is no correlation between output growth and expenditure growth (an observation consistent with an acyclical policy) and that in developing countries there is a strong and statistically significant correlation between output growth and expenditure growth (an observation consistent with procyclical policies).[19] As in Figure 3.9, the group of middle-low-income countries is the one with the lowest procyclicality, but in Figure 3.10 the coefficient remains high and statistically significant for all subgroups of developing countries. In fact, Figure 3.10 shows that the various groups of developing countries have similar levels of procyclicality and that not only are the coefficients statistically significant, but they are also large. The point estimates are close to one, suggesting that a 1 percent increase in output growth almost fully translates into a 1 percent increase in government spending (in other words, the share of government expenditure in GDP remains constant).
Why Procyclicality?
So, everyone seems to agree that, when measured in the proper way, fiscal policies are countercyclical (or, at worst, acyclical) in the advanced economies and procyclical in developing economies. By why is this so? Do policymakers in the advanced economies know something that policymakers in developing countries do not know? In other words, is this different behavior on the part of policymakers due to incompetence, or does it reflect deeper economic problems? The literature has suggested two classes of explanations for this situation. The first is based on capital market imperfections and borrowing constraints and the second on voracity effects and political distortions.[20]
Gavin and Perotti (1997) argue that developing countries find it hard to follow a countercyclical policy because, more often than not, they lack access to credit during recessions.[21] Consider, for instance, the case of a commodity exporter (which many developing countries are). If commodity exports are part of the collateral backing up a country’s sovereign debt, the value of the collateral moves together with the price of the commodity, and when the price of the commodity falls, the risk of default increases. As a consequence, the interest rate increases as well and, in some cases, it becomes so high that the country will be virtually prevented from accessing international capital markets. The opposite occurs when the commodity price increases. In such a situation, conducting a countercyclical policy would require the country to issue debt when it is expensive to do so and to retire debt at times when it is cheap to borrow (Rigobón, 2006). The bottom line is that policymakers located in developing countries would like to implement countercyclical fiscal policies, but they cannot do so because they cannot finance fiscal deficits during difficult economic times.
There are three questions that arise from this explanation. The first is: why is this not a problem for the advanced economies? The standard answer is that these countries do not face this sort of problem because they have small country risk premiums. As a consequence, the procyclicality of their interest rates is negligible. This suggests that any explanation of the procyclical behavior documented above needs to take into account the precarious creditworthiness of developing countries. Clearly, this gives rise to another question: why is precarious creditworthiness a problem for developing countries and not the advanced economies, when the latter often have much higher debt ratios? Chapter 12 focuses on this issue and shows that precarious creditworthiness is a greater problem for developing countries partly because they have smaller governments, more volatile sources of revenues, and a more dangerous debt structure. The third question has to do with lack of self-insurance: why do developing countries not avoid borrowing in bad times by saving in good times and creating a stabilization fund? The answer is that they often try to do so, but stabilization funds are problematic because they tend to be very expensive (see Chapter 14 for a discussion of this problem) and they can be easily expropriated by politicians. This problem is related to the second class of explanations for procyclical policies, one based on political rather than market failures.[22]
Tornell and Lane (1999) describe voracity effects that arise in the presence of various interest groups that compete for a share of tax revenues and treat the country’s resources as a common pool. The presence of such groups will generate procyclicality because when there is a positive shock to the country’s resources, no group will be willing to moderate its claims on the increased resources, as it knows that the saved resources will be appropriated by another group. Talvi and Végh (2005) use a model that assumes that fiscal surpluses will generate political pressures for wasteful public spending. In order to avoid this wasteful public expenditure, a benevolent social planner will adopt a procyclical fiscal policy by decreasing taxes during booms (and hence avoiding the accumulation of surpluses) and increasing taxes during recessions. Alesina and Tabellini (2005) show that the political pressure for higher spending assumed by Talvi and Végh (2005) represents optimal behavior in the presence of a situation that combines voters with imperfect information on the level of government borrowing and corrupt politicians who can appropriate part of tax revenues for their own consumption. Alesina and Tabellini’s empirical analysis is consistent with the main predictions of their model and shows that procyclicality is positively correlated with corruption.
Is It Procyclicality or Reverse Causality?
None of the explanations previously discussed takes into account one of the first things that one learns in economics: correlation does not imply causation.[23] While it is uncontroversial to state that the correlation between GDP growth and either the budget balance or government expenditure is consistent with procyclicality in developing countries and countercyclicality in the advanced economies, these correlations do not prove that policymakers located in developing countries do adopt procyclical policies. This could be a case of what economists call “reverse causality.”[24]
Box 3.2 discusses this problem in greater detail and shows that, if shocks to the growth rate of government expenditure are larger than shocks to GDP growth, any attempt to estimate the effect of GDP growth on expenditure growth may end up capturing the opposite relationship (i.e., the effect of expenditure growth on GDP growth). Therefore, the standard finding of procyclical policy in developing countries and countercyclical policy in the advanced economies could be due simply to the fact that in the advanced economies, GDP growth shocks dominate shocks to expenditure growth (a situation like the one depicted in panel B of the figure in Box 3.2), and in developing countries, expenditure growth shocks dominate GDP growth shocks (a situation like the one depicted in panel C of that figure).[25]
While reverse causality is a serious problem, if it were possible to find a variable that has a direct effect on GDP growth and no direct effect on the fiscal account, then it would still be possible to estimate the cyclicality of fiscal policy even in the presence of reverse causality (Box 3.2). Jaimovich and Panizza (2006a) argue that the average growth rate of a country’s trading partners has these properties and reproduce standard estimations of the relationship between fiscal policy and growth using this variable as an “instrument” for GDP growth.[26] Figure 3.11 reports the results. It shows that once reverse causality is controlled for, fiscal policy in the advanced economies becomes clearly countercyclical (the coefficient is negative and statistically significant). In developing economies, instead, the coefficients are often negative (the exception is the middle-high-income countries, for which the coefficient is close to zero) but not statistically significant, indicating that fiscal policy is either countercyclical or acyclical. This is an intriguing result suggesting that procyclical policies capture only part of the story in regard to the factors that lead to high volatility in emerging market countries.
Summing Up
While the fiscal policy debate focuses on deficits, most debt explosions have little to do with measured deficits but arise from contingent liabilities often associated with past policies or with inherent vulnerabilities in a country’s debt structure. While this finding has several important policy implications, it is important to start with what it does not imply. It does not imply that politicians should not worry about deficits. The statement above emphasizes measured deficits because debt explosions are often associated with past deficits which were not appropriately accounted for (see Box 3.1) as a result of extrabudgetary activities. So, a first policy suggestion is to build better accounting systems that make it possible to keep track of liabilities as soon as they are incurred.[27] But the findings of this chapter cannot be due only to measurement error associated with bad public accounting. If they could, positive and negative error would wash out, and there would be no evidence that a country’s change in debt is systematically higher than its deficit. Hence, there is something that induces politicians and bureaucrats to hide actual deficits and create “skeletons in the closet” which will then be associated with successive debt explosions (Aizenman and Powell, 1998). Hence, another policy implication is to expand the definition of budget tracked by the authorities and explicitly include in a country’s budget several of the items that are now off-budget. The market seems to know that these are important issues, and there is evidence that countries with better and more transparent accounting tend to have not only better fiscal results (Wallack, 2004), but also lower financing costs (Wallack, 2005; Cady and Pellechio, 2006).
However, poor accounting and implicit liabilities are not everything. The chapter shows that debt structure is extremely important. The usual arrangement, in which deficits are decided in the political arena and debt management is left to technocrats who often have the explicit objective of minimizing the cost of borrowing, may generate perverse incentives towards issuing too much low-cost, high-risk debt. Policymakers should be aware of the cost-safety trade-off and, by recognizing that more costly debt may have a desirable insurance component, internalize this trade-off in their decision on the costs of financing a given deficit (this would lead to setting technocrats’ incentives in terms of both the cost and risk of debt). It is a welcome development that several emerging market countries are indeed moving in this direction.[28]
What should one make of the results suggesting that the difference in fiscal procyclicality between developing countries and the advanced economies is not as strong as previously thought? Again, one should start by highlighting what the findings of this chapter do not imply. They do not imply that the previous findings that the correlations between fiscal outcomes and GDP are different in developing countries and the advanced economies are wrong. In fact, the chapter presents strong evidence in support of those findings. What the chapter questions is the mechanism that drives this difference in correlations. While the previous literature has suggested that this difference was driven by differences in fiscal policy (possibly due to different constraints faced by policymakers in developing and advanced economies), this chapter offers a potential alternative explanation: that part of the difference might be due to differences in the exogenous shocks faced by the two groups of countries. Understanding more on the causes of these different correlations is extremely important, because if they are due to differences in policies, then any solution to the procyclicality problem should focus on removing the constraints (due to either political or market imperfections) that lead policymakers to adopt procyclical policies. However, if they are due to the different nature of the shocks faced by developing countries, then any effort aimed at reducing procyclicality should be aimed at determining the main drivers of these different shocks.
