I. Introduction

Sound budget institutions are vital for a country’s ability to design and implement effective fiscal policies.² Such institutions help ensure government accountability and prevent the leakage of public funds; increase efficiency of scarce public resources; and improve the prospects of maintaining fiscal stability and meeting social development needs. Stability is more fragile, resource constraints are more binding, and social needs remain more pressing in low-income countries than in developed and emerging market countries.³ As a consequence, efforts to strengthen budget institutions have featured prominently in IMF-supported programs and donor technical assistance in the past two decades.

The current global financial crisis and its impact on low-income countries have reinforced the importance of budget institutions in enhancing the effectiveness of fiscal policy as a stabilization tool. While policymakers across different regions have sought to respond to the crisis in a counter-cyclical manner, there is evidence that certain institutional arrangements enhance the scope for countercyclical policy. As countries exit from the crisis, sound budget institutions can support the consolidation process and help safeguard fiscal sustainability.

A sizeable literature has attempted to measure the quality of budget institutions by defining quantitative indices and examining their effect on fiscal performance. The present study is the first, however, to focus attention primarily on low-income countries. It develops a composite index of the quality of budget institutions for 72 low-income and middle-income countries drawing upon empirical studies, budget survey databases and assessment reports, supplemented by case studies and other reports and data from the IMF, the World Bank, and donors engaged in capacity building in low-income countries.⁴

The index breaks new ground by recognizing the multi-faceted nature of budget institutions and is broader in scope than other available indicators. It records the quality of budget institutions along two dimensions. The first dimension covers the various stages in the budget process (planning and negotiation, approval, and implementation). The second dimension reflects various characteristics of the budget process: the degree of centralization of budgetary decision-making; the existence and effectiveness of rules and controls; the sustainability and credibility of the budget as a key policy instrument; and its comprehensiveness and transparency. The choice of criteria used in constructing the index reflects specific characteristics of low-income countries. The index allows for benchmarking against the performance of middle-income countries, across regions, and according to different institutional arrangements that deliver good fiscal performance.⁵

Using the constructed indices, the paper provides some preliminary econometric evidence of the relationship between budget institutions and fiscal performance in low-income countries. In particular, the paper addresses two main questions: are strong budget institutions associated with greater fiscal discipline (lower deficits and debt); and do countries with stronger fiscal institutions have better scope to conduct countercyclical policies? Answers to these questions can help guide a more systematic and meaningful prioritization and sequencing of the reforms of budget institutions in low-income countries.

The paper is structured as follows. Section II summarizes the literature; Section III describes the proposed index of budget institutions, how it has been constructed, its statistical properties, and a comparison with other available indicators; Section IV presents the results of the econometric analysis of the relationship between budget institutions and fiscal outcomes; Section V presents some econometric evidence on whether countries with better institutions have been able to adopt more counter-cyclical policies, including during the current crisis; finally, Section VI draws conclusions, and discusses the policy implications of the empirical work presented in the paper.

II. Literature Survey

A well established body of theoretical and empirical evidence suggests that institutional arrangements that govern the budget process influence fiscal outcomes. Budget institutions shape and regulate the policy and process of generating and allocating public resources for carrying out government functions.⁶ They incorporate the formal and informal rules that govern the budget process, and the relative division of roles and responsibilities among the various actors at each stage of the process (Figure 1).

View Full Size

The Budget Process

Citation: IMF Working Papers 2010, 080; 10.5089/9781451982237.001.A001

Source: Modified from Scartascini and Filc (2007).

• Download Figure
• Download figure as PowerPoint slide

The following section summarizes the literature that relates the quality of budget institutions to fiscal outcomes. We then provide a selective literature review of the link between budget institutions and fiscal cyclicality. Finally, we discuss the special characteristics of budget processes in low-income countries that have been reflected in the choice of criteria used in constructing the index.

III. Budget Institutions in Low-Income Countries

The academic literature on budget institutions in low-income countries is nascent, and originates to a large extent from the field work of donors and development agencies. In general, budget institutions in low-income countries are much less developed than in developed and emerging market countries, and display widely different characteristics that reflect country-specific factors, such as colonial heritage, and a variety of cultural and administrative traditions and practices. Evidence suggests that weak capacity, ineffectiveinstitutions of civil society, and political economy factors act as a severe constraint on the progress of modernizing budget institutions (Allen, 2009).

In low-income countries, numerical targets and formal constraints on spending and fiscal deficit that exist on paper, may not be binding in practice. This is because mechanisms that make adherence to budget rules and procedures transparent, and hold government ministers and officials accountable for their decisions, are usually not well established. In this context, enhancing the transparency and comprehensiveness of the budget process, and public dissemination of budget documents, even in the absence of formal rules, may be particularly important (Campos and Pradhan, 1996).¹² This can be combined with a top-down approach to budgeting in which the finance ministry, supported by the cabinet, plays a prominent role in decision-making (Ljungman, 2009).¹³

Other features of budget institutions in low-income countries distinguish them from more developed and emerging market countries. For example, in many such countries, the central budget authority―usually the ministry of finance― is relatively weak in the ministerial hierarchy, and the budget is subordinate to the national plan as a policy-making instrument. Moreover, the existence of a powerful planning ministry that is responsible for preparing and executing the budget for capital expenditures fragments the budget process (“dual budgeting”). While extrabudgetary funds and accounts appear to be no more prevalent (as a percentage of government expenditure) than in developed countries, their integration with the budget process is often weaker (Allen and Radev, 2006). Moreover, many low-income countries are heavily dependent on flows of donor aid that are often not fully integrated with the budget (Gupta et al., 2008). These factors reduce transparency and accountability, render the allocation of resources according to priorities more difficult, and create parallel bargaining arenas separate from the budget process itself.

Weaknesses in financial reporting and audit systems in many low-income countries frequently imply that mechanisms to track whether actual expenditures are within aggregate limits are inefficient (Prakash and Cabezon, 2008; Gupta et al., 2008). Controls of spending and commitments are also characteristically weak, and arrears of spending and revenues are a familiar problem. Moreover, the absence of a sound accounting framework increases the scope for creative accounting to avoid or manipulate the government’s financial regulations and other controls.

IV. Index of Budget Institutions

Most quantitative indices systemize available information regarding the characteristics and functioning of the budget process, practices, and rules. Expanding on this line of research, we construct indices that are relevant to analyzing the overall quality of budget institutions in low-income countries. Our indices, however, do not attempt to provide an exhaustive catalogue of all aspects of budget institutions. The data sources that have been used―which constitute the bulk of publicly available information―focus primarily on the expenditure side of the budget, and fall short of the “ideal” index that could have been constructed if existing data sources were no impediment. For example, there are as yet few reliable indicators that measure institutional capacity in areas such as the delivery of core public services, the tracking of poverty-related spending, or revenue administration.

In constructing the index, we use several sources of data (see Appendix 1 for details). The assessments largely cover the 2006–2008 periods, and include 72 countries across different regions (Table 1). Both published and unpublished PEFA reports are the primary source of data for criteria related to budgetary practices and performance.¹⁴ However, the PEFA framework is not comprehensive: it focuses on the operational performance of the key elements of the PFM systems, rather than the legal framework and the strategic interactions between the various actors at different stages of the budget process. Moreover, it takes no account of the influence of political economy factors on the budget. Other published indicators are also selective in their focus. Information on legal regulations and numerical and procedural rules at various stages of the budget process is mostly obtained from the OECD database. Questions on the transparency and comprehensiveness of fiscal information are based on the IBP Open Budget Index, and IMF Reports on the Observance of Standards and Codes, fiscal transparency module (“fiscal ROSC”). These data sources were complemented by a survey of IMF country teams and fiscal economists. By combining these elements, our index provides a more complete picture of the budget process, and a strengthened framework for analyzing the various components.

Table 1.

Country Coverage

Table 1.

Country Coverage

Low-income Countries	Low-income Countries	Middle-income Countries
Afghanistan	Nepal	Barbados
Albania	Niger	Botswana
Armenia	Pakistan	Colombia
Azerbaijan	Papua New Guinea	Costa Rica
Benin	Rwanda	Dominican Republic
Burkina Faso	Samoa	Ecuador
Burundi	São Tomé & Príncipe	Egypt
Cameroon	Senegal	Fiji
Cape Verde	Sierra Leone	Indonesia
Chad	Sudan	Jordan
Congo, DR	Tajikistan	Kazakhstan
Côte d’Ivoire	Tanzania	Macedonia, FYR
Ethiopia	Togo	Mauritius
Gambia, The	Uganda	Morocco
Georgia	Yemen	Namibia
Ghana	Zambia	Paraguay
Guinea		Peru
Honduras		Philippines
Kenya		Russia
Kyrgyz Republic		Serbia
Liberia		Seychelles
Madagascar		South Africa
Malawi		Swaziland
Maldives		Thailand
Mali		Trinidad & Tobago
Moldova		Turkey
Mongolia		Ukraine
Mozambique		West Bank & Gaza

View Table

Table 1.

Country Coverage

Low-income Countries	Low-income Countries	Middle-income Countries
Afghanistan	Nepal	Barbados
Albania	Niger	Botswana
Armenia	Pakistan	Colombia
Azerbaijan	Papua New Guinea	Costa Rica
Benin	Rwanda	Dominican Republic
Burkina Faso	Samoa	Ecuador
Burundi	São Tomé & Príncipe	Egypt
Cameroon	Senegal	Fiji
Cape Verde	Sierra Leone	Indonesia
Chad	Sudan	Jordan
Congo, DR	Tajikistan	Kazakhstan
Côte d’Ivoire	Tanzania	Macedonia, FYR
Ethiopia	Togo	Mauritius
Gambia, The	Uganda	Morocco
Georgia	Yemen	Namibia
Ghana	Zambia	Paraguay
Guinea		Peru
Honduras		Philippines
Kenya		Russia
Kyrgyz Republic		Serbia
Liberia		Seychelles
Madagascar		South Africa
Malawi		Swaziland
Maldives		Thailand
Mali		Trinidad & Tobago
Moldova		Turkey
Mongolia		Ukraine
Mozambique		West Bank & Gaza

B. Index Construction

To capture the quality of budget institutions along each of the budget stages and across each of the categories discussed above, we developed and scored countries based on different sub-indices, which were combined to construct the overall index. The “stage index” is based on the aggregation of three sub-indices for each stage of the budget process (planning and negotiation, approval, and implementation). The “category index” is based on five sub-indices for each of the categories considered above (top-down procedures, rules and controls, sustainability and credibility, comprehensiveness, and transparency). As discussed below, the two indices are highly correlated, but their two dimensions (budget stages versus category) allow us to explore the impact and importance of their specific components on fiscal outcomes.²²

Most of the data used in the construction of the indices is qualitative in nature. For each question, a scale between 0 and 4 was used, with a higher score reflecting better performance. In answering the questions described in Box 1 and in assigning scores, it is inevitable that some degree of judgment was exercised. To minimize the degree of discretion, a set of coding rules was used, which can be found in Appendix II. The coding depended on the nature of the question. For some factual questions the coding was binary (0 or 4 score). Other questions allowed for a more-detailed scale for their answers, and hence greater differentiation across countries in terms of the various dimensions.²³

For each country, the stage index is a simple average of the sub-indices constructed for each of the stages: (i) planning and negotiation, (ii) approval, and (iii) implementation.

Each sub-index, in turn, is a simple average of the number of questions at each stage. That is, for the overall stage index we have:

$\mathit{\text{STAGE INDEX}}=\frac{1}{3}{\displaystyle \sum _{i=1}^3}{S}_i,$

where the S_i’s correspond to the value of the sub-indices of the stages

$\begin{array}{c}{S}_1=\mathit{\text{PLANNING}}=\frac{1}{14}{\displaystyle \sum _{i=1}^{13}}{q}_i,S_2=\mathit{\text{APPROVAL}}=\frac{1}{6}{\displaystyle \sum _{i=1}^6}{q}_i,\\ S_3=\mathit{\text{IMPLEMENTATION}}=\frac{1}{13}{\displaystyle \sum _{i=1}^{13}}{q}_i,\end{array}$

and the q_i’s are the scores for each of the questions, weighted according to the number of questions in each stage. For example, since the budget planning and negotiation stage has 14 questions, the score on each question receives a weight of 1/14 (see Box 1).²⁴

Similarly, the category index is a simple average of the sub-indices constructed for each of the categories: (i) top-down procedures, (ii) rules and control, (iii) sustainability and credibility, (iv) comprehensiveness, and (v) transparency across all stages of the budget process. That is,

$\mathit{\text{CATEGORY INDEX}}=\frac{1}{5}{\displaystyle \sum _{i=1}^5}{C}_i,$

where the C_i’s correspond to the value of the category sub-indices

$\begin{array}{c}{C}_1=\mathit{\text{TOP}}-\mathit{\text{DOWN}}=\frac{1}{5}{\displaystyle \sum _{i=1}^5}{q}_i,C_2=\mathit{\text{RULES}}=\frac{1}{7}{\displaystyle \sum _{i=1}^7}{q}_i,\\ C_3=\mathit{\text{SUSTAINABILITY}}=\frac{1}{7}{\displaystyle \sum _{i=1}^7}{q}_i,C_4=\mathit{\text{COMPREHENSIVENESS}}=\frac{1}{6}{\displaystyle \sum _{i=1}^5}{q}_i,\end{array}$

$C_5=\mathit{\text{TRANSPARENCY}}=\frac{1}{8}{\displaystyle \sum _{i=1}^8}{q}_i,$

and the q_i’s are the scores for each of the questions, weighted according to the number of questions in each category.

Table 2 reports the summary statistics for the two overall indices and the sub-indices. A simple inspection reveals that the two overall indices show similar properties. The stage index is highly correlated (correlation of 0.97) with the category index. The sub-indices for stages and categories, however, show some variation. In particular, across budget stages and categories, budget planning and negotiation and comprehensiveness, respectively, have the highest standard deviations.

Table 2.

Summary of Statistics for Budget Institutions Dimensions and Indices

Source: Author’s calculations.

Table 2.

Summary of Statistics for Budget Institutions Dimensions and Indices

		Mean	Std. Deviation	Min	Max	25th Percentile	Median	75th Percentile
	Planning (S1)	2.078	0.705	0.286	3.476	1.667	2.119	2.536
	Approval (S2)	1.998	0.663	0.778	3.333	1.445	2.112	2.445
	Implementation (S3)	1.843	0.583	0.487	3.231	1.423	1.898	2.205
Stage Index		1.973	0.513	0.554	3.127	1.691	1.990	2.321
	Top Down Procedures (C1)	2.367	0.691	0.400	4.000	2.000	2.400	2.934
	Rules and Control (C2)	1.953	0.532	0.286	3.239	1.619	1.904	2.310
	Sustainability and Credibility (C3)	1.780	0.724	0.190	3.333	1.309	1.762	2.309
	Comprenhensiveness (C4)	2.040	0.829	0.333	3.667	1.333	2.000	2.695
	Transparency (C5)	1.854	0.639	0.375	3.250	1.500	1.750	2.281
Category Index		1.999	0.532	0.469	3.123	1.641	2.004	2.329

Source: Author’s calculations.

View Table

Table 2.

Summary of Statistics for Budget Institutions Dimensions and Indices

		Mean	Std. Deviation	Min	Max	25th Percentile	Median	75th Percentile
	Planning (S1)	2.078	0.705	0.286	3.476	1.667	2.119	2.536
	Approval (S2)	1.998	0.663	0.778	3.333	1.445	2.112	2.445
	Implementation (S3)	1.843	0.583	0.487	3.231	1.423	1.898	2.205
Stage Index		1.973	0.513	0.554	3.127	1.691	1.990	2.321
	Top Down Procedures (C1)	2.367	0.691	0.400	4.000	2.000	2.400	2.934
	Rules and Control (C2)	1.953	0.532	0.286	3.239	1.619	1.904	2.310
	Sustainability and Credibility (C3)	1.780	0.724	0.190	3.333	1.309	1.762	2.309
	Comprenhensiveness (C4)	2.040	0.829	0.333	3.667	1.333	2.000	2.695
	Transparency (C5)	1.854	0.639	0.375	3.250	1.500	1.750	2.281
Category Index		1.999	0.532	0.469	3.123	1.641	2.004	2.329

Source: Author’s calculations.

As is common practice in the literature, we start by looking at the Spearman rank-order correlations. Tables 3a and 3b report these correlations among the three budget stages and 5 categories. The sub-indices for planning, approval, and implementation are significantly correlated with the overall stage index (Table 3a). However, there are some interesting differences across budget stages. In particular, budget planning and negotiation appears to be correlated with approval and implementation, while the budget approval stage has a very low correlation with implementation. Spearman rank correlations for the category index and its components show that the top-down procedures sub-index has the lowest correlation with the overall index, while the sustainability and credibility sub-index shows the highest correlation (Table 3b). In general, the sub-index for top-down procedures has low, albeit significant, correlations with the other dimensions of budget institutions. In contrast, the sub-indices for rules and control, sustainability and comprehensiveness, and transparency are highly correlated.

Table 3a.

Spearman Rank Correlations among Budget Stages

Note: *and” indicate that the correlation coefficient is significantly different from zero at the one percent and ten percent, respectively.

Table 3a.

Spearman Rank Correlations among Budget Stages

	Planning (S1)	Approval (S2)	Implementation (S3)
Approval (S2)	0.4357*
Implementation (S3)	0.5822*	0.2077”
Stage Index	0.8493*	0.7097*	0.7334*

Note: *and” indicate that the correlation coefficient is significantly different from zero at the one percent and ten percent, respectively.

View Table

Table 3a.

Spearman Rank Correlations among Budget Stages

	Planning (S1)	Approval (S2)	Implementation (S3)
Approval (S2)	0.4357*
Implementation (S3)	0.5822*	0.2077”
Stage Index	0.8493*	0.7097*	0.7334*

Note: *and” indicate that the correlation coefficient is significantly different from zero at the one percent and ten percent, respectively.

Table 3b.

Spearman Rank Correlations among Categories

Note: * indicates that the correlation coefficient is significantly different from zero at the one percent level.

Table 3b.

Spearman Rank Correlations among Categories

		Top Down Procedures (C1)	Rules and Controls (C2)	Sustainability and Credibility (C3)	Comprehensivenes s (C4)	Transparency (C5)
	Rules and Controls (C2)	0.3290*
	Sustainability and Credibility (C3)	0.3779*	0.5694*
	Comprehensiveness (C4)	0.3954*	0.5596*	0.6203*
	Transparency (C5)	0.2590*	0.5608*	0.6191*	0.6000*
Category Index		0.5975*	0.7391*	0.8448*	0.8234*	0.7744*

Note: * indicates that the correlation coefficient is significantly different from zero at the one percent level.

View Table

Table 3b.

Spearman Rank Correlations among Categories

		Top Down Procedures (C1)	Rules and Controls (C2)	Sustainability and Credibility (C3)	Comprehensivenes s (C4)	Transparency (C5)
	Rules and Controls (C2)	0.3290*
	Sustainability and Credibility (C3)	0.3779*	0.5694*
	Comprehensiveness (C4)	0.3954*	0.5596*	0.6203*
	Transparency (C5)	0.2590*	0.5608*	0.6191*	0.6000*
Category Index		0.5975*	0.7391*	0.8448*	0.8234*	0.7744*

Note: * indicates that the correlation coefficient is significantly different from zero at the one percent level.

Sensitivity analysis was carried out to check the robustness of our indices to alternative aggregating and weighting procedures (Appendix III). In particular, different weights and assumptions about the degree of substitutability and complementarity of components were considered. The rank order correlations between the different approaches are high and significant, suggesting that the additive aggregation procedure described above is robust to alternative specifications.

We also compare our indices with existing indicators on the quality of budget institutions and governance and accountability. Table 4 shows pair-wise correlations of our indices with PEFA indicators, a sub-index of the World Bank’s Country Policy and Institutional Assessment (CPIA), and the Kaufmann and Kraay governance indicators.²⁵ The correlations of the aggregate indices with the sub-CPIA index and the PEFA indicators are high. The aggregate indices are also positively and significantly correlated with the government effectiveness and regulatory quality components of the governance indicators.²⁶ Overall, high correlations with other indices instill confidence that the new indices capture meaningful information.

Table 4.

Pair-wise Correlations with other Indicators

Table 4.

Pair-wise Correlations with other Indicators

	PEFA	CPIA	Sub-CPIA	Governance Indicators
				Governance Effectiveness	Regulatory Quality
Stage Index	0.70	0.60	0.70	0.55	0.63
Category Index	0.76	0.65	0.73	0.62	0.69
Number of Observations	69	43	43	72	72

View Table

Table 4.

Pair-wise Correlations with other Indicators

	PEFA	CPIA	Sub-CPIA	Governance Indicators
				Governance Effectiveness	Regulatory Quality
Stage Index	0.70	0.60	0.70	0.55	0.63
Category Index	0.76	0.65	0.73	0.62	0.69
Number of Observations	69	43	43	72	72

Given the high correlation between the overall stage and the category indices, as well as the results from the previous sensitivity analysis, in what follows we focus on the overall stage index.

C. Descriptive Statistics

Some basic descriptive analysis of the data set we compiled is suggestive of relative strengths and weaknesses in budget institutions across different country groups. Across regions, transition economies followed by countries in Latin America and the Caribbean, have relatively more developed budget institutions (Tables 5–6). Countries in Sub-Saharan Africa (SSA) are characterized by weaknesses in all the budget stages, as well as in the sustainability and credibility of their budget institutions. The transparency of the budget process is a weak area across all regions included in the sample.

Table 5.

Budget Institutions Index, by Country Groups and Stages

^1/Includes Albania

Table 5.

Budget Institutions Index, by Country Groups and Stages

Regions	Overall	Budget Planning & Negotiation	Budget Approval	Budget Implementation
Sub-Saharan Africa	1.78	1.86	1.77	1.70
Asia	1.90	1.94	1.92	1.84
Middle East & North Africa	2.14	2.29	2.33	1.79
Latin America & Caribbean	2.14	2.15	2.11	2.17
Transition Economies1	2.30	2.45	2.47	1.97
Low Income Countries	1.89	1.95	1.99	1.73
Middle Income Countries	2.10	2.28	2.01	2.02

^1/Includes Albania

View Table

Table 5.

Budget Institutions Index, by Country Groups and Stages

Regions	Overall	Budget Planning & Negotiation	Budget Approval	Budget Implementation
Sub-Saharan Africa	1.78	1.86	1.77	1.70
Asia	1.90	1.94	1.92	1.84
Middle East & North Africa	2.14	2.29	2.33	1.79
Latin America & Caribbean	2.14	2.15	2.11	2.17
Transition Economies1	2.30	2.45	2.47	1.97
Low Income Countries	1.89	1.95	1.99	1.73
Middle Income Countries	2.10	2.28	2.01	2.02

^1/Includes Albania

Table 6.

Budget Institutions Index, by Country Groups and Categories

^1/Includes Albania

Table 6.

Budget Institutions Index, by Country Groups and Categories

Regions	Overall	Top down Procedures	Rules & Controls	Sustainability & Credibility	Comprehensiveness	Transparency
Sub-Saharan Africa	1.81	2.22	1.74	1.62	1.79	1.69
Asia	1.92	2.29	1.89	1.58	1.94	1.91
Middle East & North Africa	2.14	2.69	2.03	1.86	2.17	1.96
Latin America & Caribbean	2.18	2.49	2.25	1.98	2.20	1.96
Transition Economies1	2.29	2.56	2.27	2.04	2.54	2.07
Low Income Countries	1.91	2.35	1.87	1.69	1.92	1.70
Middle Income Countries	2.14	2.40	2.09	1.92	2.23	2.09

^1/Includes Albania

View Table

Table 6.

Budget Institutions Index, by Country Groups and Categories

Regions	Overall	Top down Procedures	Rules & Controls	Sustainability & Credibility	Comprehensiveness	Transparency
Sub-Saharan Africa	1.81	2.22	1.74	1.62	1.79	1.69
Asia	1.92	2.29	1.89	1.58	1.94	1.91
Middle East & North Africa	2.14	2.69	2.03	1.86	2.17	1.96
Latin America & Caribbean	2.18	2.49	2.25	1.98	2.20	1.96
Transition Economies1	2.29	2.56	2.27	2.04	2.54	2.07
Low Income Countries	1.91	2.35	1.87	1.69	1.92	1.70
Middle Income Countries	2.14	2.40	2.09	1.92	2.23	2.09

^1/Includes Albania

In general, low-income countries have weaker budget institutions than middle-income countries (Figure 2 and Tables 5–6).²⁷ Across budget stages, while both groups of countries on average exhibit similar low scores for the budget approval stage, budget planning and implementation are considerably weaker in low-income countries. Across categories, both groups of countries are characterized by relatively high scores for top-down procedures. However, low-income countries are significantly weaker along all the other dimensions.

View Full Size

Budget Institutions, Country Characteristics, and Debt Vulnerabilities in LICs and MICs

Citation: IMF Working Papers 2010, 080; 10.5089/9781451982237.001.A001

Sources: International Monetary Fund; and staff estimates.Notes: IDA-SSA: IDA countries in SubSaharan Africa; IDA-Other: IDA countries in other regions; Blend: Countries eligible for both IDA and IBRD lending; Fuel -exporting LICs: Cameroon, Chad, Congo DR, São Tomé & Príncipe, and Sudan.

• Download Figure
• Download figure as PowerPoint slide

There is considerable diversity within low-income countries with respect to the quality of their budget institutions (Figure 2, Panels 2 and 3). Blend countries (under the World Bank’s classification) have stronger planning and approval processes and score consistently higher than other country groups in all categories. Resource rich countries (oil exporters) have the weakest budget institutions along all dimensions. Low-income countries in SSA, on average, have weaker institutional capacity than other countries, particularly in the areas of sustainability and credibility, comprehensiveness, and transparency of the budget process.

The overall scores, however, mask important differences across countries.²⁸ Countries with weaker budget institutions also tend to be in higher levels of debt distress (Figure 2 and Panel 3).

V. Budget Institutions and Fiscal Discipline

A steady improvement in fiscal performance has been a key feature of many low-income countries since the early 2000s. Better policy and economic management, coupled with a favorable external environment, especially terms of trade improvements, and official debt relief contributed to the stronger fiscal positions (IMF, 2007). While a large number of countries achieved primary fiscal surpluses, supported by debt relief, and reduced external-debt burdens, cross-country variations in fiscal performance remain. In this section, we empirically investigate whether budget institutions play a role in explaining cross-country differences in fiscal outcomes, controlling for other factors.

We examine the relationship between budget institutions and fiscal discipline during 2003– 2007.²⁹ Consistent with the literature, we consider two measures of fiscal discipline: the average primary balance of the central government and the average public external debt owed to official creditors, both expressed as ratios to GDP. We focus on the average primary balance as a preferred measure of the government’s fiscal stance because it abstracts from the effect of inflation on interest payments, and the fact that interest payments are a function of accumulated debt and not the present fiscal stance. The reason to focus on debt is that primary deficits in some countries may not be driven by a systematic bias but rather reflect temporary effects. We use official public external debt as the data on total government debt was limited for a large number of countries in the sample.

In conducting the empirical analysis, we estimated a cross-section OLS regression with the selection of control variables being dictated by the existing literature.³⁰ The index and sub-indices were added to the regressions to test if budget institutions are related to the variable of interest after accounting for the selected control variables. The empirical analysis is cross-sectional in nature because the budget institutions indices lack time variation. We exclude outliers and consider only those countries for which all variables are available.³¹

The selection of control variables relies on the earlier literature, which serves as a benchmark to compare our results. The control variables include real economic growth (Growth) to control for economic circumstances, the log of initial GDP per capita in 2002 (Initial GDP per capita) to control for differences in economic and overall institutional development, a dummy for HIPC post-completion point countries (HIPC), a dummy for oil-exporting countries (Oil), and the initial debt-to-GDP ratio (Initial_debt). Following Alesina et al. (1999), changes in the terms of trade (Trade) are scaled by the degree of openness of the economy, measured as the sum of exports and imports to GDP. Since in some countries tax revenues are heavily linked to export activities, we expect improvements in the terms of trade to be associated with lower deficits and debt levels, and these effects to be more important for economies that are more open to international trade. Initial debt, proxied by external debt in the year prior to the beginning of the sample (2002), is included to focus on the effect of budget institutions on recent fiscal policy settings. The HIPC dummy controls for low-income countries that have benefited from official debt relief and, as a result, are expected to have stronger fiscal positions. Growth, terms of trade changes, and openness are measured as annual averages for the period 2003–2007 to control for cyclical effects. Appendix IV contains details on the variables and data sources.

Table 7 shows the relationship between budget institutions and primary balances and external debt including all control variables. The overall stage index of budget institutions is positively and significantly associated with the primary fiscal balance, suggesting that, consistent with the literature, countries with better budget institutions are likely to have higher fiscal balances (Column 2).³² Similar results are obtained when a sample excluding oil-exporting countries and a sample of non-oil exporting low-income countries is considered (Columns 3–4).³³ To give a sense of the magnitude of the effect of budget institutions on primary balances, consider that the difference between the index value of the average “low quality” budget institutions country and the average “high quality” country is 0.8 in the full sample. Therefore, using the estimate from column 2 of Table 7, an average high quality budget institutions country is predicted to have a primary balance that is around 1 percentage points higher than a low quality country.³⁴

Table 7.

Budget Institutions, Primary Balance, and External Debt

Notes: *: Significant at 10% level, **: Significant at5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

Table 7.

Budget Institutions, Primary Balance, and External Debt

Dependent Variable:	Central Government Primary Balance to GDP ratio				Official Public External Debt to GDP ratio
	All		Non-oil Exporting only	Low-income only (excl. oil- Exporting)	All		Non-oil Exporting only	Low-income only (excl. oil- exporting)
	1	2	3	4	5	6	7	8
Intitial_debt	0.0035 (0.0112)	0.0053 (0.0111)	0.0065 (0.0129)	0.0092 (0.0150)	0.5340*** (0.0807)	0.5177*** (0.0802)	0.5712*** (0.0851)	0.5677*** (0.1104)
Growth	-0.2242** (0.1090)	-0.2797** (0.1083)	-0.2609* (0.1519)	-0.1996 (0.2727)	0.1392 (0.5988)	0.6558 (0.6189)	0.6287 (1.2536)	1.7735 (1.9136)
Trade	0.1277* (0.0674)	0.1365** (0.0633)	0.1424** (0.0672)	0.1699 (0.1099)	-0.0198 (0.3733)	-0.1020 (0.3090)	0.1089 (0.2878)	0.0680 (0.3752)
Initial GDP per capita	0.0024 (0.0042)	0.0008 (0.0044)	-0.0008 (0.0045)	-0.0032 (0.0061)	-0.0552 (0.0439)	-0.0404 (0.0394)	-0.0466 (0.0408)	-0.0428 (0.0460)
HIPC	0.0047 (0.0099)	0.0025 (0.0101)	0.0023 (0.0093)	0.0016 (0.0174)	-0.0801 (0.0604)	-0.0595 (0.0579)	-0.0707 (0.0592)	-0.0246 (0.0938)
Oil	0.0229 (0.0143)	0.0269* (0.0141)			-0.1134** (0.0455)	-0.1506*** (0.0460)
Stage Index		0.0098* (0.0057)	0.0107* (0.0055)	0.0171* (0.0090)		-0.0909*** (0.0303)	-0.0770** (0.0329)	-0.0984** (0.0457)
N (Observations)	65	65	56	34	65	65	56	34
R Square	0.2101	0.2312	0.1439	0.1561	0.7626	0.7883	0.7914	0.8220

Notes: *: Significant at 10% level, **: Significant at5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

View Table

Table 7.

Budget Institutions, Primary Balance, and External Debt

Dependent Variable:	Central Government Primary Balance to GDP ratio				Official Public External Debt to GDP ratio
	All		Non-oil Exporting only	Low-income only (excl. oil- Exporting)	All		Non-oil Exporting only	Low-income only (excl. oil- exporting)
	1	2	3	4	5	6	7	8
Intitial_debt	0.0035 (0.0112)	0.0053 (0.0111)	0.0065 (0.0129)	0.0092 (0.0150)	0.5340*** (0.0807)	0.5177*** (0.0802)	0.5712*** (0.0851)	0.5677*** (0.1104)
Growth	-0.2242** (0.1090)	-0.2797** (0.1083)	-0.2609* (0.1519)	-0.1996 (0.2727)	0.1392 (0.5988)	0.6558 (0.6189)	0.6287 (1.2536)	1.7735 (1.9136)
Trade	0.1277* (0.0674)	0.1365** (0.0633)	0.1424** (0.0672)	0.1699 (0.1099)	-0.0198 (0.3733)	-0.1020 (0.3090)	0.1089 (0.2878)	0.0680 (0.3752)
Initial GDP per capita	0.0024 (0.0042)	0.0008 (0.0044)	-0.0008 (0.0045)	-0.0032 (0.0061)	-0.0552 (0.0439)	-0.0404 (0.0394)	-0.0466 (0.0408)	-0.0428 (0.0460)
HIPC	0.0047 (0.0099)	0.0025 (0.0101)	0.0023 (0.0093)	0.0016 (0.0174)	-0.0801 (0.0604)	-0.0595 (0.0579)	-0.0707 (0.0592)	-0.0246 (0.0938)
Oil	0.0229 (0.0143)	0.0269* (0.0141)			-0.1134** (0.0455)	-0.1506*** (0.0460)
Stage Index		0.0098* (0.0057)	0.0107* (0.0055)	0.0171* (0.0090)		-0.0909*** (0.0303)	-0.0770** (0.0329)	-0.0984** (0.0457)
N (Observations)	65	65	56	34	65	65	56	34
R Square	0.2101	0.2312	0.1439	0.1561	0.7626	0.7883	0.7914	0.8220

Notes: *: Significant at 10% level, **: Significant at5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

Columns 6–8 of Table 7 show that stronger budget institutions are also associated with lower debt. The coefficient on the overall index is statistically significant and negative at the 5 percent level, suggesting that, consistent with the findings from previous studies, countries with better budget institutions tend to have lower debt. The strong association between budget institutions and debt suggests that while the cross-country variation in external debt levels reflects past policies, there could be hysteresis in budget institutions since they are slow to evolve and difficult to change. Overall, our results seem to suggest a positive relationship between the strength of budget institutions in a country and their fiscal performance in the period leading up to the current crisis.

Disaggregating the overall index into different dimensions allows us to shed some light on which components of budget institutions are most relevant. Table 8 reports cross-country regressions for primary balance and public external debt in which the sub-indices for stages and categories are entered separately. Across budget stages, planning and negotiation appears to be driving the results with significant estimated coefficients. Across categories, one clear result is that more comprehensive and transparent budget processes have the most pronounced influence on both primary balances (Panel A). Moreover, both robust budget planning and implementation are also associated with lower debt (Panel B). In addition, stronger rules and controls and a more sustainable and credible budget process have a significant influence on debt levels.³⁵ The differences in the relative importance of the various sub-indices, reflect, in part, the larger cross-country variation in external debt levels relative to primary balances in the period under consideration. Overall, our results suggest that, with the exception of top-down procedures, the various components of the budget institutions have an impact on performance.

Table 8.

Disaggregated Budget Institutions and Fiscal Performance

Notes: *: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level, Robust standard errors are in parentheses. Controls (omitted from table) include the full set of controls shown in Table 7. Source: IMF staff estimates.

Table 8.

Disaggregated Budget Institutions and Fiscal Performance

	1	2	3	4	5	6	7	8
Planning	0.0098** (0.0039)
Approval		0.0007 (0.0053)
Implementation			0.0069 (0.0052)
Top Down Procedures				-0.0033 (0.0047)
Rules and Controls					0.0062 (0.0061)
Sustainability and Credibility						0.0058 (0.0043)
Comprehensiveness							0.0082** (0.0035)
Transparency								0.0131** (0.0056)
N (Observations)	65	65	65	65	65	65	65	65
R Square	0.2521	0.2103	0.2227	0.2159	0.2209	0.2234	0.2509	0.2704
Panel B: Dependent Variable: Official Public External Debt (ratio to GDP)
Planning	-0.0717*** (0.0259)
Approval		-0.0192 (0.0249)
Implementation			-0.0739** (0.0298)
Top Down Procedure				-0.0011 (0.0250)
Rides aid Controls					-0.0831*** (0.0277)
Sustainability and Credibility						-0.0841*** (0.0284)
Comprehensiveness							-0.04O2** (0.0190)
Transparency								-0.0740*** (0.0275)
N (Observation)	65	65	65	65	65	65	65	65
R Squats	0.7626	0.7903	0.8014	0.7763	0.7896	0.7967	0.7691	0.7852

Notes: *: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level, Robust standard errors are in parentheses. Controls (omitted from table) include the full set of controls shown in Table 7. Source: IMF staff estimates.

View Table

Table 8.

Disaggregated Budget Institutions and Fiscal Performance

	1	2	3	4	5	6	7	8
Planning	0.0098** (0.0039)
Approval		0.0007 (0.0053)
Implementation			0.0069 (0.0052)
Top Down Procedures				-0.0033 (0.0047)
Rules and Controls					0.0062 (0.0061)
Sustainability and Credibility						0.0058 (0.0043)
Comprehensiveness							0.0082** (0.0035)
Transparency								0.0131** (0.0056)
N (Observations)	65	65	65	65	65	65	65	65
R Square	0.2521	0.2103	0.2227	0.2159	0.2209	0.2234	0.2509	0.2704
Panel B: Dependent Variable: Official Public External Debt (ratio to GDP)
Planning	-0.0717*** (0.0259)
Approval		-0.0192 (0.0249)
Implementation			-0.0739** (0.0298)
Top Down Procedure				-0.0011 (0.0250)
Rides aid Controls					-0.0831*** (0.0277)
Sustainability and Credibility						-0.0841*** (0.0284)
Comprehensiveness							-0.04O2** (0.0190)
Transparency								-0.0740*** (0.0275)
N (Observation)	65	65	65	65	65	65	65	65
R Squats	0.7626	0.7903	0.8014	0.7763	0.7896	0.7967	0.7691	0.7852

Notes: *: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level, Robust standard errors are in parentheses. Controls (omitted from table) include the full set of controls shown in Table 7. Source: IMF staff estimates.

To determine whether the results are driven by special cases, we ran a number of robustness checks, estimating our baseline regressions with additional variables and different subset of countries. To control for the possibility that lower primary deficits in low-income countries are driven by the inability to borrow domestically on account of less-developed financial institutions, we included credit extended to the private sector from banks and other financial institutions (as share of GDP) as a proxy for financial development. Table 9 (Columns 1 and 3) shows that financial development does not have an important independent effect on primary balances or debt. External factors (e.g. an IMF program) may also play a significant role in some countries. To control for the possibility of de facto fiscal controls, we use a dummy variable that takes a value of 1 if the country has an IMF program in the period under consideration. The results for budget institutions are robust to the inclusion of this variable (Columns 2 and 4). Our results also remain significant if we substitute average yearly growth in external debt as the dependent variable and if non-oil exporting countries and low-income countries are considered separately. We also considered different time periods for the analysis and a sample excluding HIPC countries, but the results remained largely unchanged.³⁶

Table 9.

Budget Institutions and Fiscal Discipline: Robustness Checks

Notes:*: Significant at 10% level,**: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

Table 9.

Budget Institutions and Fiscal Discipline: Robustness Checks

Dependent variable	Primary Balance		Official Public External Debt		Growth in Official Public External Debt (average annual)
	All	All	All	All	All	Non-oil exporting only	Low-income only
	OLS	OLS	OLS	OLS	OLS	OLS	OLS
Initial_debt	0.0054 (0.0112)	0.0076 (0.0115)	0.5183*** (0.0807)	0.5120*** (0.0830)	-0.4270** (0.2034)	-0.5189** (0.2324)	-0.6809** (0.2836)
Growth	-0.2778** (0.1095)	-0.2516** (0.1187)	0.6650 (0.6229)	0.5886 (0.6184)	0.5554 (1.9334)	2.3054 (3.4186)	2.2687 (3.9329)
Trade	0.1311** (0.0647)	0.1225* (0.0638)	-0.1278 (0.3285)	-0.0685 (0.3160)	0.5372 (0.5268)	0.4719 (0.6018)	1.0876 (0.8757)
Initial GDP per capita	0.0023 (0.0056)	-0.0002 (0.0046)	-0.0332 (0.0455)	-0.0380 (0.0392)	-0.2770** (0.1376)	-0.2731** (0.1314)	-0.3545*** (0.1401)
HIPC	0.0031 (0.0103)	0.0045 (0.0102)	-0.0564 (0.0602)	-0.0645 (0.0577)	-0.2804** (0.1398)	-0.2632* (0.1349)	-0.2177 (0.1811)
Oil	0.0255* (0.0142)	0.0275* (0.0141)	-0.1568*** (0.0494)	-0.1520*** (0.0467)	-0.1396* (0.0813)		-0.2939* (0.1574)
Private Credit	-0.0001 (0.0002)		-0.0004 (0.0007)
IMF program		-0.0089 (0.0079)		0.0214 (0.0309)
Stage Index	0.0098* (0.0057)	0.0111* (0.0063)	-0.0909*** (0.0305)	-0.0941*** (0.0320)	-0.1795** (0.0800)	-0.2047** (0.0877)	-0.2135** (0.0931)
N (Observations)	65	65	65	65	65	56	39
R Square	0.2347	0.2440	0.7894	0.7893	0.4139	0.4372	0.5530

Notes:*: Significant at 10% level,**: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

View Table

Table 9.

Budget Institutions and Fiscal Discipline: Robustness Checks

Dependent variable	Primary Balance		Official Public External Debt		Growth in Official Public External Debt (average annual)
	All	All	All	All	All	Non-oil exporting only	Low-income only
	OLS	OLS	OLS	OLS	OLS	OLS	OLS
Initial_debt	0.0054 (0.0112)	0.0076 (0.0115)	0.5183*** (0.0807)	0.5120*** (0.0830)	-0.4270** (0.2034)	-0.5189** (0.2324)	-0.6809** (0.2836)
Growth	-0.2778** (0.1095)	-0.2516** (0.1187)	0.6650 (0.6229)	0.5886 (0.6184)	0.5554 (1.9334)	2.3054 (3.4186)	2.2687 (3.9329)
Trade	0.1311** (0.0647)	0.1225* (0.0638)	-0.1278 (0.3285)	-0.0685 (0.3160)	0.5372 (0.5268)	0.4719 (0.6018)	1.0876 (0.8757)
Initial GDP per capita	0.0023 (0.0056)	-0.0002 (0.0046)	-0.0332 (0.0455)	-0.0380 (0.0392)	-0.2770** (0.1376)	-0.2731** (0.1314)	-0.3545*** (0.1401)
HIPC	0.0031 (0.0103)	0.0045 (0.0102)	-0.0564 (0.0602)	-0.0645 (0.0577)	-0.2804** (0.1398)	-0.2632* (0.1349)	-0.2177 (0.1811)
Oil	0.0255* (0.0142)	0.0275* (0.0141)	-0.1568*** (0.0494)	-0.1520*** (0.0467)	-0.1396* (0.0813)		-0.2939* (0.1574)
Private Credit	-0.0001 (0.0002)		-0.0004 (0.0007)
IMF program		-0.0089 (0.0079)		0.0214 (0.0309)
Stage Index	0.0098* (0.0057)	0.0111* (0.0063)	-0.0909*** (0.0305)	-0.0941*** (0.0320)	-0.1795** (0.0800)	-0.2047** (0.0877)	-0.2135** (0.0931)
N (Observations)	65	65	65	65	65	56	39
R Square	0.2347	0.2440	0.7894	0.7893	0.4139	0.4372	0.5530

Notes:*: Significant at 10% level,**: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported. Source: IMF staff estimates.

These results presented here should be considered as preliminary, since there are a number of econometric problems that need to be addressed. First, is the problem of endogeneity – the possibility that fiscal outcomes influence the evolution of budget institutions, rather than the other way around. In many low-income countries, enhancements to budget institutions are often part of a larger package of fiscal consolidation under IMF-supported programs. In this sense, budget institutions can be viewed, to some extent, as endogenous to current and past fiscal outcomes. This makes it difficult to claim a strong causality between budget institutions and fiscal outcomes, but correspondence between them is a significant result.³⁷ The working assumption in earlier papers, which we maintain, is that budget institutions are relatively costly to change and are stable over at least the short to medium-term as changes in fiscal performance cannot quickly feedback into altering institutions.

A second problem is that of omitted variables, which arises if variables omitted from the regression are those really driving the relationship between budget institutions and fiscal performance. For example, it could be that countries with a strong overall institutional infrastructure exhibit greater fiscal discipline. Good economy-wide institutions are correlated with the quality of budget institutions, but it may be that it is other institutions excluded from the regression that really drive fiscal performance. Failure to control for this could overstate the effect of budget institutions on primary balances and debt. To partially tackle this problem, we included broad indicators of institutional quality such as government effectiveness and the control of corruption in the baseline regression (Table 10).³⁸ The effects of budget institutions on fiscal discipline are very close to, although slightly smaller for primary balance as a percent of GDP than, the estimates obtained in the baseline regression. While these robustness checks provide some assurance, the lack of time dimension in our indices renders it difficult to adequately control for other unobservable factors.

Table 10.

Robustness Checks: Role of Economy-wide Institutions

Notes:*: Significant at 10% level, **: Significant at 5% level, *** :Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported.

Table 10.

Robustness Checks: Role of Economy-wide Institutions

Dependent variable	Primary Balance		External debt		External Debt Growth
Initial_debt	0.0044 (0.0105)	0.0042 (0.0104)	0.5100*** (0.0837)	0.5074*** (0.0819)	-0.4713** (0.2184)	-0.4702** (0.2179)
Growth	-0.2698** (0.1074)	-0.2630** (0.1061)	0.5149 (0.6780)	0.7498 (0.6613)	0.8618 (1.8984)	0.6881* (1.9412)
Trade	0.1429** (0.0651)	0.1420** (0.0630)	-0.1076 (0.3374)	-0.1399 (0.3398)	0.4363 (0.4875)	0.4602 (0.4893)
Initial GDP per capita	0.0027 (0.0054)	0.0039 (0.0064)	-0.0193 (0.0376)	0.0005 (0.0484)	-0.2790** (0.1274)	-0.2802 (0.1411)
HIPC	0.0100 (0.0123)	0.0110 (0.0119)	-0.0466 (0.0544)	-0.0323 (0.0587)	-0.2875** (0.1231)	-0.2852** (0.1349)
Oil	0.0262* (0.0147)	0.0255* (0.0151)	-0.1784*** (0.0531)	-0.1862*** (0.0547)	-0.1832* (0.1038)	-0.1863* (0.0980)
Control of Corruption	-0.0011 (0.0081)		-0.0494 (0.0419)		0.0435 (0.0599)
Government Effectiveness		-0.0032 (0.0095)		-0.0766 (0.0587)		0.0366 (0.0832)
Stage Index	0.0099* (0.0059)	0.0105* (0.0061)	-0.0785** (0.0313)	-0.0712** (0.0347)	-0.2124** (0.0858)	-0.2095** (0.0924)
N (Observations)	65	65	65	65	65	65
R Square	0.2428	0.2438	0.7987	0.8022	0.4208	0.4192

Notes:*: Significant at 10% level, **: Significant at 5% level, *** :Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported.

View Table

Table 10.

Robustness Checks: Role of Economy-wide Institutions

Dependent variable	Primary Balance		External debt		External Debt Growth
Initial_debt	0.0044 (0.0105)	0.0042 (0.0104)	0.5100*** (0.0837)	0.5074*** (0.0819)	-0.4713** (0.2184)	-0.4702** (0.2179)
Growth	-0.2698** (0.1074)	-0.2630** (0.1061)	0.5149 (0.6780)	0.7498 (0.6613)	0.8618 (1.8984)	0.6881* (1.9412)
Trade	0.1429** (0.0651)	0.1420** (0.0630)	-0.1076 (0.3374)	-0.1399 (0.3398)	0.4363 (0.4875)	0.4602 (0.4893)
Initial GDP per capita	0.0027 (0.0054)	0.0039 (0.0064)	-0.0193 (0.0376)	0.0005 (0.0484)	-0.2790** (0.1274)	-0.2802 (0.1411)
HIPC	0.0100 (0.0123)	0.0110 (0.0119)	-0.0466 (0.0544)	-0.0323 (0.0587)	-0.2875** (0.1231)	-0.2852** (0.1349)
Oil	0.0262* (0.0147)	0.0255* (0.0151)	-0.1784*** (0.0531)	-0.1862*** (0.0547)	-0.1832* (0.1038)	-0.1863* (0.0980)
Control of Corruption	-0.0011 (0.0081)		-0.0494 (0.0419)		0.0435 (0.0599)
Government Effectiveness		-0.0032 (0.0095)		-0.0766 (0.0587)		0.0366 (0.0832)
Stage Index	0.0099* (0.0059)	0.0105* (0.0061)	-0.0785** (0.0313)	-0.0712** (0.0347)	-0.2124** (0.0858)	-0.2095** (0.0924)
N (Observations)	65	65	65	65	65	65
R Square	0.2428	0.2438	0.7987	0.8022	0.4208	0.4192

Notes:*: Significant at 10% level, **: Significant at 5% level, *** :Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions, but is not reported.

Finally, a related problem is that of measurement error–given subjectivity involved in their construction, our indices may not be an accurate representation of the actual quality of budget institutions in countries. If this is the case, then using OLS as an estimation method can give biased results. To tackle this problem, we adopt an instrumental variable strategy that uses as an instrument a second index that is a non-linear transformation of the original index.³⁹ In our analysis, a natural candidate corresponds to the index with different assumptions about the substitutability of its components (α = 2; see Appendix III for details). The results for the IV 2SLS regressions for external debt, and the average yearly growth of external debt, which are available upon request, show that the effects of budget institutions on fiscal discipline are very close to, although slightly smaller than, the estimates obtained by OLS.

VI. Budget Institutions and Fiscal Cyclicality

A large number of low-income countries have used fiscal policy to counter the impact of the global economic slowdown (IMF, 2009b). This apparent shift toward countercyclical fiscal policies in response to the crisis can be largely attributed to the steady improvement in macroeconomic performance and structural reforms over the last three decades.⁴⁰ Among the minority of countries that instead tightened their fiscal policy, initial debt levels were significantly higher. In this section, we ask whether countries with stronger budget institutions have adopted less procyclical fiscal policies in response to the global crisis and also over the past decade.

Why should budget institutions play a role in shaping the fiscal response to the cycle? As discussed earlier, budgetary spending tends to expand during booms, often resulting in increased spending commitments which are difficult to rescind. Well-designed and efficiently managed budget institutions can enable policymakers to adopt a countercyclical policy stance by reducing the deficit bias noted in the literature, raising awareness about the medium term implications of policy actions, and by highlighting the need for sustainable policies.

As a first step, we examine whether the quality of budget institutions in low-income countries played a role in shaping fiscal policy responses to the recent global crisis. We consider two measures of fiscal accommodation in 2009: real growth in central government expenditures and the change in the primary fiscal balance (a negative sign implies higher fiscal accommodation). We estimate a cross-section OLS regression with a parsimonious set of variables to control for cross-country variation and differences in key indicators at the outset of the crisis. The control variables include primary balance in 2008 (Primary Balance lagged), external debt to GDP ratio in 2006 (Debt lagged), a dummy variable for oil-exporting countries (Oil), real GDP growth, and the log of GDP per capita in 2006 to control for differences in economic and institutional development.

Table 11 details the results of the cross-section estimation. It shows that controlling for initial debt and primary balances, fiscal accommodation in 2009 was higher for countries with stronger budget institutions. The coefficient of the overall stages index is significant at the 5–10 percent level regardless of whether the change in the primary balance or real growth in central government spending is considered as the relevant dependent variable. This finding provides some preliminary support for the idea that countries with stronger budget institutions were more likely to implement a countercyclical policy response during the Crisis—or at least be less procyclical.⁴¹

Table 11.

Budget Institutions, Fiscal Accommodation, and the Global Crisis, 2008–09

Notes:*: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions,

Table 11.

Budget Institutions, Fiscal Accommodation, and the Global Crisis, 2008–09

Dependent Variable:	Change in Central Government Primary Balance		Change in Central Government Expenditure
	1	2	3	4
Primary Balance (lagged)	-5.0120 (8.1787)	-5.2138 (8.4038)	0.6002* (0.3138)	0.6073* (0.3222)
Oil	0.4113 (1.1898)	0.1642 (1.0113)	-0.1868** (0.0696)	-0.1782** (0.0638)
External Debt (lagged)	-0.5585 (1.8197)	-0.9132 (1.9283)	0.0456 (0.0757)	0.0580 (0.0777)
Growth	-0.2487** (0.1207)	-0.2443** (0.1219)	0.0018 (0.0038)	0.0017 (0.0035)
Log GDP per Capita	-1.6209 (1.0706)	-1.5517 (1.0231)	-0.0085 (0.0216)	-0.0109 (0.0220)
Overall Stages Index		-1.5629* (0.8907)		0.0545** (0.0269)
N (Observations)	62	62	62	62
R Square	0.1023	0.1337	0.2378	0.2842

Notes:*: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions,

View Table

Table 11.

Budget Institutions, Fiscal Accommodation, and the Global Crisis, 2008–09

Dependent Variable:	Change in Central Government Primary Balance		Change in Central Government Expenditure
	1	2	3	4
Primary Balance (lagged)	-5.0120 (8.1787)	-5.2138 (8.4038)	0.6002* (0.3138)	0.6073* (0.3222)
Oil	0.4113 (1.1898)	0.1642 (1.0113)	-0.1868** (0.0696)	-0.1782** (0.0638)
External Debt (lagged)	-0.5585 (1.8197)	-0.9132 (1.9283)	0.0456 (0.0757)	0.0580 (0.0777)
Growth	-0.2487** (0.1207)	-0.2443** (0.1219)	0.0018 (0.0038)	0.0017 (0.0035)
Log GDP per Capita	-1.6209 (1.0706)	-1.5517 (1.0231)	-0.0085 (0.0216)	-0.0109 (0.0220)
Overall Stages Index		-1.5629* (0.8907)		0.0545** (0.0269)
N (Observations)	62	62	62	62
R Square	0.1023	0.1337	0.2378	0.2842

Notes:*: Significant at 10% level, **: Significant at 5% level, ***: Significant at 1% level. Robust standard errors are in parentheses. A constant term was included in all regressions,

To examine whether developing countries with better budget institutions have adopted less procyclical policies over the past decade, we consider the following model:

where F and Y measure the real expenditure growth and output, respectively; i denotes the country and t the time period; Z is a set of control variables, and μ_i is a country level fixed effect. The (pro)cyclicality of fiscal policy is determined by looking at the sign and size of coefficient β: if β <0, fiscal policy is countercyclical; if β =0 it is acyclical; and if β >0 it is procyclical.

Since our budget institutions indices lack time variation, following Alt and Lassen (2006), we adopt an indirect approach to examine the link between the quality of budget institutions and the extent of fiscal procyclicality across countries. We replicate the analysis in Lledo et al. (2009), using annual data for 70 countries in the period 2000–2009 for which we have constructed indices. In particular, we estimate equation (1) separately for countries with weak and strong budget institutions using the median sample value of the overall budget institution index and each of the sub-indices as alternative thresholds.⁴² Differences in the magnitude and significance of the procyclicality parameters for the two groups of countries provide evidence of the relative importance of budget institutions.

In line with the recent empirical literature, we use a Difference GMM approach to address the reverse causality between output growth and fiscal policy as well as the correlation between output growth and country fixed effects.⁴³ We use growth in real central government spending as the dependent variable, consistent with the argument developed by Kaminsky et al. (2004) that policy instrument variables under the direct control of fiscal authorities rather than outcome or target variables are a more appropriate way to measure the cyclicality of fiscal policy.⁴⁴ Our choice of real GDP growth to proxy changes in the business cycle rather than output gap reflects recent critiques on the appropriateness of detrending output in developing countries (see Aguiar and Gopinath, 2007). See Appendix IV for details on variables and data sources used.

We find preliminary evidence that fiscal policies have on average been less procyclical over the last decade in countries with strong budget institutions (Table 12). The estimated fiscal procyclicality coefficient is statistically significant in countries with below median values for the composite index of budget quality. Fiscal policy is also procyclical in a statistically significant way for countries with index values above the sample median. However, the procyclicality coefficients in this group were about 2 ½ times smaller than in the group with index values below the median.⁴⁵ Overall, this provides support for the hypothesis that high quality budget institutions are associated with less procyclical fiscal policies. However, these results should be treated with caution in light of the econometric problems outlined in the previous section.

Table 12.

Budget Institutions, Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM estimates

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.

Table 12.

Budget Institutions, Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM estimates

	Above Median Group	Observations		Below Median Group	Observations	Median Threshold
Overall Index	0.55*	263	1.47**		237	2.01
	(0.29)		(0.72)
Planning	0.60*	263	1.40*		237	2.14
	(0.35)		(0.79)
Approval	0.40	263	0.89**		237	2.11
	(0.51)		(0.43)
Implementation	0.62	191	1.05		253	2.11
	(0.52)		(0.64)
Top-Down Procedures	0.52	207	0.76		261	2.4
	(0.45)		(0.50)
Rubs and Control	0.69	247	0.90		205	1.90
	(0.50)		(0.72)
Sustainability and Credibility	0.49	255	0.53		245	1.80
	(0.35)		(0.42)
Comprehensiveness	0.73**	244	0.52		192	2.00
	(0.35)		(0.70)
Transparency	0.84*	256	1.78		188	1.75
	(0.45)		(1.16)

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.

View Table

Table 12.

Budget Institutions, Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM estimates

	Above Median Group	Observations		Below Median Group	Observations	Median Threshold
Overall Index	0.55*	263	1.47**		237	2.01
	(0.29)		(0.72)
Planning	0.60*	263	1.40*		237	2.14
	(0.35)		(0.79)
Approval	0.40	263	0.89**		237	2.11
	(0.51)		(0.43)
Implementation	0.62	191	1.05		253	2.11
	(0.52)		(0.64)
Top-Down Procedures	0.52	207	0.76		261	2.4
	(0.45)		(0.50)
Rubs and Control	0.69	247	0.90		205	1.90
	(0.50)		(0.72)
Sustainability and Credibility	0.49	255	0.53		245	1.80
	(0.35)		(0.42)
Comprehensiveness	0.73**	244	0.52		192	2.00
	(0.35)		(0.70)
Transparency	0.84*	256	1.78		188	1.75
	(0.45)		(1.16)

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.

We also find some evidence that fiscal procyclicality appears to be mostly driven by weaknesses in the initial stages of the budget process. Table 12 shows that fiscal procyclicality coefficients are statistically significant for countries with planning and approval sub indices below median values and more than twice the level of countries with higher quality budget institutions countries. By contrast, fiscal procyclicality in countries with higher scores on the implementation sub-index, although larger than in below-median score countries, is not significant at conventional levels. Our preliminary results also seem to suggest that a more transparent budget process plays a more important role in shaping fiscal cycles than other characteristics. Unlike other budget characteristics, fiscal procyclicality coefficients in countries with stronger budgetary transparency are nearly twice as small as those with less transparent budgets.

Since financing restrictions and improvements in fiscal discipline seem to play an important role in shaping past fiscal responses to the cycle among low-income countries, we next interact the composite indexes and their subcomponents separately with a low-income country dummy and with the level of external debt. The pattern of results remains qualitatively the same in both cases (Table 13). We find that fiscal policy is strongly procyclical in low-income countries, with elasticities of government spending to output growth above those displayed by countries in the full sample. Moreover, income levels appear to magnify the effect of budget institutions on fiscal procyclicality. Differences in fiscal procyclicality estimates between low-income countries with below and above median values for the composite index is about four fold.⁴⁶ Similar results also apply to more indebted countries, where weak budget transparency once again stands out as detrimental to countercyclical fiscal policies.

Table 13.

Budget Institutions, Income and Debt, Combined Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM Estimates

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.

Table 13.

Budget Institutions, Income and Debt, Combined Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM Estimates

Overall Indexes		Low-Income Countries		Debt
		Above Median Group	Below Median Group	Above Median Group	Below Median Group
Overall Index		1.10** (0.48)	3.85* (2.16)	0.29 (0.61)	1.61** (0.79)
	Planning	0.91** (0.44)	3.27** (1.37)	0.52 (1.26)	1.53 (1.02)
	Approval	0.68 (0.79)	5.27** (3.09)	0.33 (0.47)	1.18 (0.82)
	Implementation	1.15 (1.02)	2.86** (1.34)	1.05 (0.89)	2.09 (1.29)
	Top-Down Procedures	1.29** (0.60)	0.77 (0.62)	-0.19 (0.78)	1.59 (1.11)
	Rules and Control	1.13 (0.76)	2.89 (2.9)	0.87 (1.47)	0.68 (0.78)
	Sustainability and Credibility	0.83 (0.61)	1.60 (2.09)	0.61 (0.97)	0.53 (0.68)
	Comprehensiveness	1.20** (0.49)	2.31*** (0.82)	0.34 (0.98)	0.82 (0.83)
	Transparency	1.29** (0.51)	2.44 (2.50)	0.56 (0.70)	2.41* (1.13)

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.

View Table

Table 13.

Budget Institutions, Income and Debt, Combined Impact on Procyclicality Dependent Variable: Growth in central government expenditures Two-step Difference-GMM Estimates

Overall Indexes		Low-Income Countries		Debt
		Above Median Group	Below Median Group	Above Median Group	Below Median Group
Overall Index		1.10** (0.48)	3.85* (2.16)	0.29 (0.61)	1.61** (0.79)
	Planning	0.91** (0.44)	3.27** (1.37)	0.52 (1.26)	1.53 (1.02)
	Approval	0.68 (0.79)	5.27** (3.09)	0.33 (0.47)	1.18 (0.82)
	Implementation	1.15 (1.02)	2.86** (1.34)	1.05 (0.89)	2.09 (1.29)
	Top-Down Procedures	1.29** (0.60)	0.77 (0.62)	-0.19 (0.78)	1.59 (1.11)
	Rules and Control	1.13 (0.76)	2.89 (2.9)	0.87 (1.47)	0.68 (0.78)
	Sustainability and Credibility	0.83 (0.61)	1.60 (2.09)	0.61 (0.97)	0.53 (0.68)
	Comprehensiveness	1.20** (0.49)	2.31*** (0.82)	0.34 (0.98)	0.82 (0.83)
	Transparency	1.29** (0.51)	2.44 (2.50)	0.56 (0.70)	2.41* (1.13)

Notes: *: Significant at 10% level; **: Significant at 5% level; ***: Significant at 1% level. The reported estimated standard errors, using Windmeijer (2005)’s finite sample correction, are in parentheses. All regressions include a control for terms of trade growth and lagged growth in central government spending. GDP growth and the lagged dependent variable are instrumented for using lags. See appendix for variable description.