Macroeconomic Context and Fiscal Policy: Europe and Central Asia During 2000-2012

This paper examines the interaction between fiscal policy and the broader macroeconomic context in open economies. It asks two questions. First, what was the relationship between fiscal policy and current account balances in countries in Europe and Central Asia during the past dozen years? Second, how might changes in (a) output composition and (b) financial sector profitability affect revenues and thus, the assessment of the underlying structural fiscal balance? The study finds that, for flexible exchange rate countries, expansionary fiscal policy has been associated with wider current account deficits. Moreover, changes in net exports and in financial sector profitability may have significant impacts on fiscal balances because of changes in revenues from the value-added tax and the corporate profits tax as a share of gross domestic product. These findings suggest that the countries of Europe and Central Asia have reason to be prudent in terms of fiscal policy choices, even as gross domestic product rises.


Policy Research Working Paper 6621
This paper examines the interaction between fiscal policy and the broader macroeconomic context in open economies. It asks two questions. First, what was the relationship between fiscal policy and current account balances in countries in Europe and Central Asia during the past dozen years? Second, how might changes in (a) output composition and (b) financial sector profitability affect revenues and thus, the assessment of the underlying structural fiscal balance? The study finds that, for flexible exchange rate countries, expansionary fiscal policy has This paper is a product of the Poverty Reduction and Economic Management Unit, Europe and Central Asia Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank. org. The author may be contacted at rislam@worldbank.org.
been associated with wider current account deficits. Moreover, changes in net exports and in financial sector profitability may have significant impacts on fiscal balances because of changes in revenues from the valueadded tax and the corporate profits tax as a share of gross domestic product. These findings suggest that the countries of Europe and Central Asia have reason to be prudent in terms of fiscal policy choices, even as gross domestic product rises.

Introduction
This paper explores a number of interactions between fiscal and other macroeconomic variables in the Europe and Central Asia region in the period immediately preceding the 2008 financial crisis and its aftermath. Specifically, it asks two questions: (1) Did the fiscal stance have a destabilizing influence on external balances in the pre-crisis period; and (2) What are the implications of the new global macroeconomic conditions for fiscal outcomes and policy? Two main conclusions emerge from this analysis. First, that fiscal policy was indeed destabilizing for many ECA countries in the pre-2008 period in that structural fiscal deficits increased when external balances were increasingly in deficit. Second, with lower capital inflows and a changing composition of output, fiscal revenues may remain lower in terms of GDP for some countries, even as GDP recovers. Together, these factors imply that greater prudence in fiscal policy is desirable in the longer term, if it is to stabilize economic prospects. In view of existing revenue and current account constraints and weaker growth prospects, this translates into a policy that constrains real expenditure growth and government size in ECA countries. In essence, this will also require that greater attention be paid to the composition of government expenditures and revenues.
Of the last dozen years, the period since 2004 has been one of substantial volatility in global markets. The first part of this period is marked by a commodity and financial boom until 2008. For many non-commodity exporters, it also meant a period of rising current account deficits. These developments were followed by a collapse in financial markets and commodity prices by the end of 2008 and in 2009. The financial and economic crisis led to banking problems in many countries and a sovereign debt crisis in several markets. While commodity prices have recovered, financial markets are still wavering. Growth prospects are still weak in developed countries and weakening in emerging markets; public finances are less strong relative to other regions. 2 Policymakers in each country have struggled to find the appropriate balance between supporting longer term growth, protecting the vulnerable and stabilizing economies in response to short-term fluctuations. This paper examines the links between fiscal and other macroeconomic outcomes. It does not investigate the extent to which domestic policies in ECA countries affected the magnitude or type of capital inflows into each country, 3 but instead, asks whether fiscal policies might have had an impact on the magnitude and direction of the current account balance. The objective is to determine whether fiscal policy could have been more "stabilizing" given global market conditions characterized by low global interest rates and a general optimistic view of emerging market risk-reward trade-offs. 4 If one of the goals of fiscal policy is to moderate troughs in growth, then symmetrically, it should also be reining back economic activity during temporary booms. Failure to be sufficiently prudent in peak times makes it more difficult to adjust during troughs.
Another question this paper addresses, namely, the implications of the global macroeconomic scenario for fiscal policy focusses on the impact on fiscal revenues. It considers two linkages between macroeconomic outcomes and fiscal revenues that, if considered, would help better understand true structural balances in ECA countries. In conjunction with a period of slower growth, ECA countries are likely to continue collecting lower tax revenues as a share of GDP than in the pre-crisis period because of changes in output composition and financial sector profitability. Specifically, the large current account deficits of the pre-crisis period are unlikely to be sustained. Current accounts improved substantially in the crisis as domestic demand collapsed dramatically, offsetting the large decline in export demand faced by ECA countries. In some cases, current account balances have since widened, but countries are increasingly aware that high external deficits (and a corresponding accumulation of debt), may make their economies more vulnerable to capital outflows during periods of financial stress in global markets, and, accordingly, more volatile. If capital inflows and corresponding current account deficits continue to be lower than before the crisis, revenues from the VAT, which vary directly with the value of net exports, will decline as net exports to GDP decline.
Another source of revenue "weakness" may appear as a result of lower financial sector profitability. Financial sector conditions affect the profits of the financial sector (which in some countries is large), and, in addition, they affect the profits of non-financial firms. When financial sector conditions are good and credit is easily available at low interest rates, firms face lower interest payments and are able to expand operations easily (in good times, domestic and global lenders are less risk averse). The condition of domestic financial sectors in ECA is closely linked to global financial markets. In addition, financial sector booms are not perfectly correlated with GDP cycles. Thus tighter credit markets and lower financial sector profitability may be expected to reduce corporate profits as a share of GDP. Lower profits would be expected to reduce tax revenues from this sector relative to the boom period (ceteris paribus). 4 One way to think of the optimistic view of emerging market risk and investor confidence is to consider the spreads on credit default swaps for a number of ECA countries and their risk ratings. These three interactions between external markets and fiscal outcomes are chosen as the focus of this paper as these have been under-researched in the past and are of relevance to ECA countries. There are, however, differences among countries in the ECA region in terms of their initial conditions and outlook and some of these differences will be explored in the paper. For example, oil exporting countries have very different economic and tax structures from others in the region; fixed exchange rate countries face constraints on policies relative to flexible exchange rate countries; EU countries are perceived differently by financial markets or have different tax structures from the other countries in the group. Part I of the paper addresses the link between internal and external deficits. Part II of the paper examines the potential impact of changes in output composition and financial booms on tax revenues.

Part I: Twin Deficits
Several theoretical and empirical papers have highlighted the links between fiscal outcomes (public savings or dis-savings) and current account outcomes. Fiscal policy may have an effect on the current account through various channels. Most directly, changes in government demand can translate into changes in the trade balance. Fiscal policy, by changing demand can also affect the real exchange rate and/or interest rate. If it induces a real exchange rate appreciation for example, the current account could worsen. Fiscal expansion may raise risk premia and therefore external interest payments (it may also cause capital flight and a forced narrowing of the current account deficit). The actual effect varies between countries. This paper adds to the existing empirical literature on this topic by its coverage of countries and time period. It considers 28 countries in Eastern and Central Europe and Central Asia, of which some are high income, while others are classified as emerging or developing countries. The period of analysis is 2000-2011/12.
Among the most recent papers written about the "twin deficits" of the fiscal and current accounts, Atoyan, Jaeger, and Smith (2012) study the link between fiscal deficits and current account balances for emerging Europe during 2000-07. They find that push factors -low returns in flow-originating countriesrather than pull factors -high returns in flow destination countries-drove most of the private capital flows to emerging Europe. They also find that anti-cyclical fiscal policy acted as a brake on capital inflows, but conclude that fiscal policy alone is unlikely to be an effective tool to put an effective brake on sudden capital flow surges. They measure the stance of fiscal policy by adjusting for the automatic changes in fiscal outcomes which are the result of endogenous changes in the absorption gap. They define the absorption gap to be the sum of the output gap and the gap between the actual and the sustainable current account deficits. Katrakilidis and Trachanas (2013)  Leigh (2011), using a new dataset for 17 OECD countries that focusses only on fiscal policy changes whose motivation is deficit management, find that a 1 percentage point of GDP consolidation improves the current account balance to GDP ratio by 0.6 percentage points. 5 Among some older, but relevant papers, Kumhof and Laxton (2009) find that permanent fiscal deficits in large countries can significantly raise the world interest rate. In the short run they cause a short run current account deterioration equal to around 50 percent of the fiscal deficit deterioration. In the longer run, the current account deterioration equals almost 75% for a large economy such as the United States, and almost 100% for a small open economy. Chinn and Ito (2005) investigate the medium-term determinants of the current account and find that the government budget balance is an important determinant of the current account balance in industrial countries. They also find that countries with more developed equity markets are more likely to run current account deficits. Salvatore (2006) empirically confirms the relationship between the current account and fiscal deficits for the G7 countries for the 1973-2005periods. Corsettin and Muller (2006 find evidence supporting the twin deficits hypothesis for Australia, Canada, the UK and the US; the relationship is stronger when the country is more open. Kaminsky, Reinhart and Vegh (2004) examine 104 countries and find some interesting patterns regarding international financial flows, fiscal and monetary policies. They also find countries with (a) capital inflows are pro-cyclical and (b) fiscal policy is pro-cyclical for most developing countries.
This means that more expansionary fiscal policy is positively related to higher capital inflows (or larger current account deficits). For OECD countries, Piersanti (2000) finds that expectations of future budget deficits (such as those engendered by tax cuts) are associated with higher current account deficits. 6 Kearney and Monadjemi (1990) use VAR analysis on eight industrialized countries and confirm the twin deficits hypothesis for the period 1972-1987.

Empirical Estimation of the Twin Deficits Hypothesis
This section examines the relationship between fiscal and current account balances in the ECA countries. The main question of interest is whether fiscal deficits were destabilizing in the sense that they (or rather, the discretionary component of them) tended to be larger when current account deficits were larger. In order to investigate this relationship, panel regression analysis is used. The main regression of interest estimated in levels is of the following form: 7 (1) CAB/GDP is the current account balance as a ratio to GDP. In order to focus on discretionary elements of fiscal outcomes rather than those influenced by cyclical fluctuations and past actions, I use the cyclically adjusted primary fiscal balance as a ratio to GDP as a measure of the fiscal stance. The adjustment used focuses on revenues, rather than expenditures as described in Annex 1. 8 The list of countries included in the empirical estimation is in Annex 1.
Other explanatory variables included in various specifications, as prescribed by theory and related empirical papers, are the lagged current account balance to GDP ratio, the output gap relative to GDP, the terms of trade (TOT), a measure of trade openness (OPEN/GDP), a measure of financial indebtedness 6 Expectations of future budget deficits are instrumented with past values of the deficit. 7 In alternative specifications, a measure of the real exchange rate is also used as an explanatory variable. 8 This implies that all expenditures changes were discretionary. Clearly, in some of the ECA countries, automatic stabilisers on the expenditure side may be large (though revenue effects are larger). Thus, the adjustment done for this paper would overestimate the deficit in bad times and underestimate it in good times (when output is above potential).
(DEBT/GDP), 9 a measure of whether the exchange rate is fixed or not (FIX), and interactions of the primary fiscal balance to GDP ratio with these other variables. The lagged current account ratio is included to account for persistence in current account. An increase in the terms of trade might have a positive impact on the current account balance (depending on how demand for exportables versus other goods changes). Countries' external debts to GDP ratios are used as explanatory variables as higher debt ratios may signal greater access to finance in international markets, or, alternatively, they may signal greater risk to lenders. Greater trade integration may be associated with larger external imbalances, though the effect may vary across time depending on domestic and international market conditions and output responses. Greater trade integration is represented by the value of imports and exports as a share of GDP. 10 Cyclical output changes may also affect current account balances and leaving these effects out would tend to bias estimates on the public sector balance. However, the effect may be either to reduce or raise the external deficit. 11 For example, if output being above potential means that a higher proportion of domestic demand is met through imports, the current account balance would deteriorate as the output gap becomes positive, and the deficit accordingly smaller. However, if the output gap is driven by a boom in export demand, then the current account impact would be positive and the deficit accordingly smaller.
Thus a measure of the output gap is added as a regressor. Net foreign assets as a share of GDP was also used as a regressor to see if financial openness had an impact on current account balances. Table 1 below shows some summary statistics for the main variables of interest. The Table   shows the wide variation among outcomes in the dataset, particularly in the measures of trade integration and external debt. 9 A measure of financial openness was also used, but was not significant in most specifications. A measure of the real exchange rate was also used and found to be insignificant. 10 An interaction term of trade integration with the primary balance is also tried in some specifications. That is, structural deficits may have a stronger impact on current account balances in countries that are more open because part of the higher demand for goods would be met by imports.   Figure 1 shows a measure of fiscal policy behavior during the decade. It graphs growth in real expenditures against the output gap in ECA countries. The scatterplot indicates that real expenditure growth and the output gap have a positive association -real expenditures growing more when the output gap is larger (and positive). Figure 2 shows the output gap as a ratio to GDP graphed against the primary structural balance (oil exporters excluded). These figures suggest that public finances were not stabilizing output changes in the region during the period under consideration.  The following empirical estimation considers how this stance may have influenced external balances. Table 2 below shows some of the results from the first set of regressions. Almost all regressions exclude the oil exporters. 12 The first column is the simple OLS regression. In the fixed effects regression (3), ECA countries (excluding oil exporters) 13 do not exhibit a consistent pattern in terms of the relationship between external balances and fiscal balances. However, when the group of countries with fixed exchange rates is distinguished from those with flexible rates; the results show different and consistent patterns for the two groups during the period under consideration. In the first specification, there is an interaction term of the primary structural deficit with the exchange rate regime. 14 15 The interaction term between the exchange rate regime and the primary structural balance is constructed by multiplying the primary structural balance with a dummy variable that takes the value 1, if the country has a fixed exchange rate system. Changes in the structural primary deficit are always significantly and positively associated with changes in the current account balance, once the interaction term for countries with fixed exchange rate systems is included. Moreover, for almost all specifications, the magnitude is around 0.3. The countries in the sample that have fixed exchange rates show different results: their 12 However, the basic results hold and the effects are larger when oil exporters are included. 13 Including oil exporters makes the twin deficits stronger; even in this specification, when oil exporters are included, with country and time fixed effects, the primary structural deficit enters with a significant coefficient.
14 The countries that are denoted as fixed exchange rate countries are the Baltic countries, Bosnia-Herzegovina, Bulgaria, Kosovo, Montenegro, Slovenia and the Slovak Republic (see IMF Annual Report on Exchange Rate Arrangements, Oct, 2012). 15 Inclusion of oil exporters strengthens the relationship. A positive output gap has a negative effect on the current account balance, implying that when output is above potential, the current account deficit tends to be greater. This reflects the higher import demand (and possibly, real exchange rate appreciation), associated with higher output. The openness of a country is negatively associated with current account balance and the effect is found to be significant across all specifications. An increase in aggregate demand is more likely to be met through imports if the country is more open. The size of net financial assets as a ratio to GDP was added as an indicator of financial openness, but was not significantly associated with the current account balance. However, debt to GDP ratios (reflecting past current account deficits), are positively associated with current account balances. 16 A scatterplot of the current account balance and the primary structural deficit for fixed exchange rate countries (not shown) shows this pattern.

Part II: Fiscal revenues, output composition and financial cycles
The literature analyzing the relationship between tax receipts and output composition is limited.
Gabriela Dobrescu and Ferhan Salman (2011) discuss the relationship between a country's fiscal stance and domestic absorption. Specifically, they show that just as ignoring output gaps can bias assessments of a country's true fiscal stance, so can ignoring domestic absorption cycles -cycles during which net exports boom and fall-bias assessments. They find net exports to be anti-cyclical in many emerging and advanced economies. In tax systems that rely heavily on indirect taxes, periods of low net exports will be associated with high fiscal revenues, ceteris paribus. Thus a period of high GDP growth and low net exports will be a period during which indirect tax receipts will be higher relative to GDP than a period of high GDP growth associated with high net exports. If net exports change in an anti-cyclical fashion, then periods of recession will see tax revenues to GDP fall more than would be expected with net exports remaining constant. Standard measures of cyclically adjusted fiscal balances would not capture this additional impact of GDP composition, which may vary with the GDP cycle, on revenues. These authors conduct a cross country analysis covering 59 advanced and emerging economies during 1990-2009 and examine fiscal outcomes during domestic absorption changes. They find that absorption booms (periods when net exports are low because domestic absorption is higher than normal) are associated with procyclical fiscal policy. They conclude that pro-cyclicality arises because policymakers mistake absorption booms for permanent (structural) increases in revenues.
Li and Lopez-Murphy (2010) examine tax revenue downturn episodes, that is, episodes where tax revenue to GDP ratios decline sharply, and explore the relationship between tax revenues and imports.
Their analysis covers 63 countries during 1977-97 and 26 countries during 1995-2007. They find that tax revenue to GDP ratios in emerging and developing countries are almost twice as volatile as in advanced countries; import to GDP ratios are also much more volatile in emerging and developing countries. They also find that expenditure taxes account for the bulk of the downturn in emerging and developing countries, while income taxes account for the bulk in developed countries. Changes in the import to GDP ratio are a statistically significant determinant of changes in the tax revenue to GDP ratio even when controlling for changes in real output and the terms of trade. Rahman (2010)  In this section, I take a closer look at the revenue implications of potential changes in GDP composition; specifically changing net exports to GDP, given growth. As global capital markets have weakened leading to reduced external financing, many countries have had to contain their current account deficits. Another factor tending to reduce imports and increase net exports has been much weaker consumption demand in many ECA countries. As a result, consumption based tax revenues have declined. I explore how value added taxes are affected by changes in net exports. Value added taxes are likely to decline more than the amount estimated by declining GDP if net exports rise relative to GDP. Figure 3 shows how net exports tend to be anti-cyclical, being more negative when GDP growth is high and higher when GDP growth is low in ECA countries. In other words, net export patterns may worsen revenue declines during recessions and improve them during booms.  VAT revenues vary in importance between countries; the lowest ratios are found in the oil exporting countries. For these countries, oil revenues are the single most important source of fiscal revenue but these revenues do not accrue to the government through the VAT. Table 3 below shows VAT revenues as a share of GDP for the Balkan countries, the CIS and South Caucasus, the EU11, the Oil Exporters and the ECA average for 2007, 2009 and 2011. 18 They are the most important, as a share of GDP, in the Balkans. Figure 4 shows that the importance of VAT revenues varies with per capita income in ECA countries. VAT revenues as a share of overall revenues, as depicted along the Y axis, tend to decline with increases in per capita income.  Finally, Figure 5 below shows a scatterplot of the VAT revenue to GDP ratio graphed against the net export to GDP ratio. The graph indicates a strongly negative relationship. Figures 4 and 5 together imply that similar changes in net exports would have larger effects on countries with lower GDP. 18 These numbers are based on data received during 2012.  In order to do explore the relationship further, I use the basic regression below: (2) Where VAT/GDP is the ratio of VAT revenues to GDP, TOT is an index of the terms of trade, GDPGROWTH is the growth rate of real GDP, NETEXPORTS/GDP is the ratio of net exports to GDP and GDPPERCAPITA is constant per capita GDP. The lagged VAT share is included to account for persistence in revenues. GDP growth is expected to affect tax revenues: in good times tax collection improves. The output gap is added as a regressor to account for additional revenue effects related to boom and bust periods. Regression results are shown in Table 4. The results indicate that, as expected, net exports and revenue collection from the VAT as a share of GDP are negatively related and the relationship is always strongly significant. The first regression shows the OLS pooled simple regression, the second column shows the same regression augmented by country fixed effects. The magnitude and significance of the coefficient on the net export share increase in the second specification. Subsequent specifications add the variables that theory dictates would predict VAT revenues and include either time Relationship Between Net Exports and VAT Revenues and country fixed effects or both or are GMM estimations. In all these specifications, the coefficient on the net exports ratio is between -0.05 to -0.07.
The terms of trade and the VAT share do not show a significant association and there is (a statistically) significant and substantial persistence in the current account as shown by the large coefficient on the lagged value of the VAT share. Including both time and country fixed effects does not reduce the significance of the variable of interest. The coefficient on the TOT is not robust to alternative specifications. In particular, it is not significant when the sample is broken into non-EU11 countries.
Neither is it significant in the GMM specifications. The coefficient on log GDP per capita is usually significant and positive, though not always once other controls are added. A positive coefficient indicates that countries become better at collecting taxes as they become richer. Surprisingly, GDP growth is not a significant predictor of VAT revenues when the output gap is added as an explanatory variable. The output gap has a negative relationship with VAT revenues, even after controlling for net exports, when oil exporters are excluded from the sample. 19 In other words, it is probably capturing the effect of booms related to export earnings.
What do these regressions reveal about the magnitude of the impact of a change in net exports on the VAT revenue ratio? I take a value of 0.05 for the coefficient-looking at regressions (3)-(7). Taking the change in the net export to GDP ratio for Albania from 2008 to 2012, the immediate impact on the VAT share is predicted to be 0.33 percentage points of GDP (6.6x.05). The effect is even larger in the "long run". The "long run" coefficient on net exports (1-.4) is .05/.6 or .08. Using this coefficient implies a reduction of 3.5 percentage points in the VAT ratio! In other words, the longer the decline in net exports exists, the larger the effect on VAT revenues, ceteris paribus. A similar number is obtained for the change in revenue for the Kyrgyz Republic from 2008-09. Equation (9) in Table 5 shows the estimation for the EU11 countries only. Net exports have a much larger impact on VAT revenues as a share of GDP for this group, the coefficient being 0.09. For Bulgaria, where the net export ratio rose 18.92 points between 2008 and 2011, and using the coefficient in equation (9)  L is the lag operator and L (Xt) = Xt-1.
1/ Country and time fixed effects.

Empirical Estimation II: Revenues from Corporate Taxes and Financial Sector Profitability
Another change that is of relevance to ECA is the impact of the financial sector boom on corporate, and particularly, financial, sector profitability. The magnitude and pattern of the financial sector's booms and busts are not perfectly correlated with that of the output cycle. Yet, financial sector volatility may have substantial effects on revenues through its effect on the profits of financial and nonfinancial firms. A sustained reduction in financial sector profitability would be expected to have an impact on the share of the corporate income tax in GDP. This poses a potential concern for ECA countries-will tighter financial market conditions and lower returns in the financial sector have implications for tax revenues? Table 6 shows corporate income tax revenues as a share of GDP during 2007, 2009 and 2011. On average, CIT revenues tend to be a more important source of revenue for the higher income countries in ECA while VAT tends to be a relatively more important source of revenue for countries with lower income, as shown in Figure 6 below.  Source: World Bank staff estimates based on government accounts. Note: Oil Exporters are excluded from the graph In order to examine the effect of financial sector profits on corporate tax revenues, an exercise similar to the one conducted for VAT revenues, was done for the share of corporate income taxes in GDP.
The main regression of interest is shown below: Where CIT/GDP is the share of corporate tax revenue in GDP and ROE is the return on equity for financial sector firms. The return on equity is used as a proxy for financial sector profits. 21 The lagged share of CIT is included to account for persistence in the revenue share of GDP from one period to the next. As for the regressions concerning the VAT, GDP growth is included as an explanatory variable: tax revenues collected are expected to be higher in good times. The TOT are expected to affect profits tax revenues, particularly in cases where corporate profits are strongly linked to oil revenues; moreover, the country's (and financial system's) creditworthiness may be affected by the TOT. If so, movements in the TOT would be correlated with movements in the return on equity. GDP per capita is included as a control variable as richer countries are expected to have higher corporate/financial sector tax revenues relative to GDP. Financial openness is expected to affect the CIT/GDP ratio as it is likely to have an impact on financial flows. Lower net foreign assets relative to foreign debt, implies higher external borrowing; though higher NFA may indicate better ability to repay. The sign could be positive or negative.
In all the cases, the coefficient on return on equity has the expected sign and is significant, usually at the 1% level. The results are shown in Table 7  In terms of the other variables, GDP growth is not a significant factor in explaining the corporate tax share of GDP, once the TOT and return on equity are included, and if oil exporters are excluded from the sample. However, it is always significant (though not always with the expected sign) in explaining the CIT share once oil exporters are included in the sample. The results are similar for the TOT variable. 21 The regressions were also run using the return on assets and similar results were obtained. 22 The longer run coefficient is calculated by taking the difference between the current and lagged coefficients of the CIT ratio and dividing the coefficient on ROE by this number.
The NFA coefficient does not add to the explanatory power of the regression. Measures of the output gap are not significantly correlated with the CIT share of GDP when oil exporters are excluded. L is the lag operator and L (Xt) = Xt-1. ***, ** and * denote significance at 1%, 5% and 10% levels respectively. 1/ Country and year fixed effects L is the lag operator and L (Xt) = Xt-1. ***, ** and * denote significance at 1%, 5% and 10% levels respectively. 1/ Country and year fixed effects 2/ Regression exclude oil exporters.

Conclusion
Macroeconomic developments resulting from changes in global markets have had a significant impact on fiscal accounts in ECA. Fiscal policies have also had an impact on countries' external balances. This paper has discussed three important linkages between fiscal accounts and the macroeconomy (a) the relationship of external balances with fiscal balances; (b) the effect of output composition on fiscal revenues and (c) the effect of financial sector profitability on fiscal outcomes. By examining these particular macro-fiscal issues, a better understanding of ECA countries' macro-fiscal outcomes is developed.
Three main conclusions from this paper are that (i) Expansionary fiscal policy has been associated with periods of rising external deficits in flexible exchange rate countries (which constitute the majority of ECA countries). In the ECA region, countries with fixed exchange rates did not on average, have fiscal policy as a significant factor in worsening current account balances; (ii) Changes in net exports have affected fiscal revenues as a share of GDP; (iii) Declines in financial sector profitability may also have longer term consequences for profits taxes. The implication of (i) is that countries that are concerned with managing their external balances must adjust fiscal balances accordingly. Points (ii) and (iii) imply that movements in net exports and in financial markets may distort perceptions of revenue sustainability. In ECA countries, trade balances and international financial market conditions have impacts on fiscal revenues. Ignoring changes in these factors could lead to an overestimation of revenue sustainability and lead policymakers to mistake temporary revenue changes for permanent ones. Thus, structural deficits may be under-estimated. These analyses highlight that when adopting fiscal policies, it is important to take account of economy-wide factors, apart from growth, in affecting outcomes.

Structural/Cyclically Adjusted Fiscal Balances
When assessing the stance of fiscal policy, it is useful to think of structural or cyclically adjusted balances instead of actual fiscal balances which also reflect non-discretionary or one-off policy decisions.
Structural balances are those from which all transitory factors affecting the fiscal balance have been purged. A complete calculation of structural balances would need to account for the following factors: • Cyclical revenue and expenditure items: expenditure and revenue outcomes for government depend on both discretionary elements and on elements that increase/decrease with the economic (business cycle). For example, unemployment compensation payments and income tax revenues vary with GDP; • Items whose cyclical behavior may not be strongly correlated with domestic output but have important transitory effects on revenues or expenditures; and • Other transitory items that affect the budget only in a single year or a limited number of years such as a civil service reform measure that only induces expenditures in the current and next year, or a one-time fee levied on a transaction.
The first two elements affect the budget automatically as a result of the changes in economic variables. The third element is usually the result of a policy decision.
The concept of the structural fiscal balance used in this paper mostly corrects for the automatic effects of the business cycle (the deviations of actual and potential output) on fiscal balances. They focus on the revenue side. There is no correction for automatic stabilizer effects on the expenditure side (for example, an increase in unemployment compensation payments in recessionary times).
In adjusting for revenues, this study uses an elasticity of 1 with respect to output. Ideally, country specific elasticities should be used. However, there is empirical evidence that 1 is a good approximation of the weighted average of disaggregated elasticity estimates over a range of countries. In addition, for a regional study, this is an appropriate approximation. The elasticity of personal income taxes with respect to the output gap has been found to be approximately between 1-1.7, for corporate income taxes, between 1.2-1.8, for social security contributions, 0.5-0.9 and for indirect taxes, around 1. 23 Correcting for the behavior of items whose movements are not perfectly correlated with the business cycle, would provide a better estimate of structural balances. However, for the important items in this category, asset and commodity prices, and the composition of output, correction of fiscal balances becomes complex (both theoretically and empirically) even in a single country. Among the reasons are that long run equilibrium prices and cyclical behavior are not easy to establish or distinguish, and the equilibrium composition of output (or say current account balances and net exports) depends on a variety of other endogenous factors. The paper uses empirical relationships between VAT revenues and net exports to GDP to assess the direction of the impact of a contraction in net exports, given growth, on VAT revenues and therefore, to get a better idea of the underlying structural balance. Similarly, it looks at profitability in the financial sector as having a substantial impact on tax revenues.
In-depth knowledge of country fiscal policy decisions are needed to identify one-off or very short-lived expenditure and revenue changes. However, it is often hard to identify measures that are truly one-off. In practice, there are no rules that can be applied across the board. Also, items that seem transitory may be "judged" to be permanent. For example, the EC suggests that deficit increasing measures should not be regarded as one-off measures as expenditure increases intended to be temporary often become permanent. At the same time, fiscal stimulus measures should not be taken out of calculations, even if intended to be temporary. A large amount of judgment needs to enter into estimation of a particular countries' structural deficit. 24 In addition, for analytical purposes, if the goal is to examine discretionary fiscal measures, temporary elements should not be removed from the estimation of structural budgets.

Estimation of the output gap
The Hodrick-Prescott filter is used to decompose output into trend and cyclical components. Hodrick and Prescott (1980) propose the optimization procedure in Equation (1)

Rate of Return on Asset
Net income to yearly averaged total assets of commercial banks within each country listed in Bankscope*.

Rate of Return on Equity
Net income to yearly averaged equity of commercial banks within each country listed in Bankscope*.