Raising Tax Revenue: How to Get More from Tax Administrations?

Eui Soon Chang; Elizabeth Gavin; Nikolay Gueorguiev; Jiro Honda

doi:10.5089/9781513550831.001.A001

Jump to Content Jump to Main Navigation

Raising Tax Revenue: How to Get More from Tax Administrations?

Author: Eui Soon Chang¹, Elizabeth Gavin, Nikolay Gueorguiev², and Mr. Jiro Honda

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ² 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: EGavin@imf.org, NGueorguiev@imf.org, JHonda@imf.org, esoon.chang@gmail.com

Publication Date:: 24 Jul 2020

eISBN:: 9781513550831

Language:: English

Keywords:: WP; reform effort; administration staff; Tax administration; tax administration official; estimated tax; tax administration practice; tax administration strength

Download PDF (5.4 MB)

Can we empirically show the benefits of improving the practices and characteristics of tax administration agencies for revenue collection? While many country experiences suggest a strong link, there is little systematic empirical evidence on the matter. This paper analyzes the association between tax collections and tax administrations, using the novel dataset (ISORA). We find that tax performance is positively and strongly associated with the operational strength of tax administrations. Among emerging and low-income economies, countries at the top 25 percent (in terms of the operational strength) collect substantially larger tax revenues (by 3¼ percent of GDP) than countries at the lowest 25 percent, assuming other conditions are equal. Our results also suggest that adopting key administrative practices such as compliance risk management and use of third-party data is associated with stronger tax collections. Furthermore, larger staffing of a tax agency improves tax revenue up to a point. These findings have important policy implications, particularly during the unprecedented global pandemic situation.

Abstract

I. Introduction

1. Mobilizing tax revenue is essential for a country’s development. Achieving stable domestic tax revenue allows budget resources to finance their essential spending needs (e.g., healthcare, education, and critical infrastructure). As noted in Besley and Persson (2009) and Gaspar et al. (2016), increasing tax collection is important for economic growth and development. For many developing countries, enhancing tax revenues is critical for the attainment of the Sustainable Development Goals (IMF, 2018). In order to build up tax capacity, country experiences suggest the importance of tax policy and administration reforms.²

2. The recent Covid-19 pandemic also highlights the importance of establishing sound administrative features in tax administration. With the pandemic, many countries are experiencing erosion of tax collection, reflecting the decline in economic activities and possible deterioration in tax compliance. Most tax administrations have had to shift to working remotely, with little or no face-to-face contact with taxpayers. Tax administrations’ practices will have an impact on their ability to adapt and respond to changing circumstances. The situation underscores the importance of good tax administrative features. After the pandemic recedes, strengthening tax administration would be even more important because of the critical need to generate fiscal resources for addressing development needs and safeguarding debt sustainability through enhancing revenue mobilization.

3. Despite the importance of tax administration to revenue collection, empirical literature on this subject has been limited. Many country experiences suggest the benefits of tax administration reforms, in terms of increased taxpayer compliance, increased transparency, and perceptions of fairness of the tax system, among other outcomes. Various reform efforts have been made in many developing countries (often assisted by IMF and other international partners), however, there has been limited empirical literature on this matter, largely owing to the lack of data on practices and features of tax administration. This poses some important questions. Can the beneficial outcomes of better tax administration be empirically supported? If so, what specific practices and characteristics of tax administration would be associated with better tax performance?

4. This paper empirically examines the benefits of specific tax administration practices and characteristics, using the novel dataset of the International Survey on Revenue Administration (ISORA), based on the two-step approach. The ISORA dataset results from the completion of a standardized questionnaire designed jointly by the ISORA partners,³ based on responses from 135 administrations in 2016, and 159 administrations in 2018. It covers a wide spectrum of data on institutional characteristics, practices, and characteristics of tax administration. Given the limited availability of the historical data from the ISORA, the analysis in this paper is based on the two-step approach, as proposed by Hsiao (2003). This would allow us to focus on the effects of tax administration while controlling for other factors (e.g., macroeconomic developments and tax policy changes).

5. The paper confirms the importance of sound design of practices and characteristics of tax administration agencies to mobilize domestic tax revenue (Box 1).⁴ The empirical results reveal that the operational strength of the agency is positively and strongly associated with tax revenue. Among emerging market and low-income economies, countries at the top 25 percent (in terms of operational strength) collect more tax revenues (by 3.25 percent of GDP) than countries at the lowest 25 percent, assuming other conditions are equal. Our results also highlight the importance of adopting key tax administration practices and characteristics—such as compliance risk management (CRM), use of third-party data, degree of digitalization, service orientation, and public accountability.

^{Box 1.}

Key Findings of This Paper

The practices and characteristics of tax administration agencies matter significantly for tax performance. The operational strength of the agency, comprising compliance risk management (CRM) practices, the use of third-party data, degree of digitalization of services, service orientation, public accountability, and autonomy, is tightly associated with tax collections.
Amongst sound practices of tax administration, CRM and the use of third-party data are particularly important. To enhance tax collection, tax administration reform efforts should prioritize: (i) strengthening CRM by adopting automated risk profiling and electronic audits; and (ii) utilizing third party data by adopting computer systems for processing the data and prefilling returns.
Many of the above factors are highly correlated. This aligns with advice to countries that tax administrative reform benefits derive from the strategic integration of multiple components, rather the implementation of individual measures. For example, it is the effective operation of an office or program for large taxpayers that matters, rather than the establishment of such an office or program.
Furthermore, adequately staffing tax administration agencies is important. Increased staffing of a tax administration agency improves revenue performance up to a threshold of 0.25 percent of the labor force. The staff complement of many LIDC tax administrations is below this level. Broadening the taxpayer base is important. The number of active taxpayers (in relation to the labor force) matters for tax collection.

6. The rest of the paper is organized as follows. Section II presents a brief review on related literature. Section III describes the ISORA dataset and key stylized facts on tax collection and design of tax administration. Section IV describes the empirical methodology. Section V discusses empirical findings. Section VI concludes with policy implications.

II. Literature Review

7. A number of studies examine various determinants of tax revenue performance. The impact of the structure of the economy, economic dynamics, socio-political factors, and tax policy have been examined through cross-sectional studies, the analysis of panel data, and case studies. Several studies point out the complex interrelationship of factors impacting revenue outcomes, and the difficulties in isolating relevant factors (Kidd, 2006; Dom, 2017).

A stream of literature investigates how the level of economic development and structure affect revenue collected. There is broad consensus that per capita GDP is positively correlated with revenue (Gupta, 2007; IMF, 2018). The contribution of agriculture to the economy is generally found to be negatively correlated with central government revenue (Yohou, 2017), although not in all studies (Boukbech, 2018; IMF, 2018). Gupta (2007) notes that per capita GDP and agriculture’s value add as a proportion of GDP are strongly correlated. Revenue collections are also correlated with trade openness (Gupta, 2007).
Some studies explore the impact of changes in tax policy on revenue collection, though proxies for measuring the impact of tax policy are not well-established yet. Akitoby et al. (2018) found that episodes of revenue growth in emerging markets and low-income countries were frequently associated with tax policy reforms such as the reduction of exemptions, and/or rate increases in indirect taxes. Dabla-Norris et al. (2017) investigated how the level of tax compliance burden faced by firms affects firms’ performance in 21 countries and considered the corporate income tax (CIT) and value-added tax (VAT) rates to control for tax policy. The impact of the introduction of a VAT has been examined in various studies, with some finding a growth in revenue in Sub-Saharan Africa (Ebeke, 2014) and the Caribbean (Schlotterbeck, 2017), while in others no significant impact was identified (Ngoma, 2017). Ngoma finds that when controlling for other reforms on non-resource tax revenue, there is no evidence that the introduction of a VAT increased tax collection performance in Sub-Saharan Africa. Productivity of CIT and VAT is found to be lower in countries with a higher number of exemptions (IMF, 2018).
Recent literature also recognizes the role of socio-political factors in determining the level of revenue relative to GDP. The Gini coefficient, a measure of income inequality, appears to be negatively correlated with revenue collection, while the spending on education within a country correlates positively with performance (Fenochietto, 2013). Amongst countries with similar incomes, those with the lowest levels of corruption collect four more percentage points of GDP in tax revenues, than those with the highest corruption levels (IMF, 2019).

8. There is, however, far less empirical literature on the effect of tax administration practices and characteristics on tax performance. As one of these few pieces of literature, Crivelli (2018) conducted cross-sectional correlation analyses between ‘tax administration strength,’ comprising a range of tax administration characteristics and tax efficiencies (i.e., VAT, PIT, and CIT) in Europe using data collected by the OECD.⁵ This study—while being novel in comprehensively exploring the effects of specific administration features—does not control for other factors (such as macroeconomic environments and tax policy).

Other studies examine the effects of a Large Taxpayer Office (LTO) or program. Baum (2017) finds that LTOs enhance revenue collections through dampening the perception of corruption, while Ngoma (2017) does not identify any positive change in revenue associated with LTO establishment. Ebeke (2016) does not find the positive impact of a LTO but notes that the establishment thereof is often a precursor for the introduction of VAT or establishment of a semi-autonomous revenue authority (and thus its impacts).
In addition to these empirical studies, there are some country case studies on tax administration: Pomeranz (2015), for instance, finds that the use of third-party data impacts VAT collections positively, while Almunia and Lopez-Rodriquez (2018) finds that the increased monitoring and auditing of large firms through an LTO enhanced CIT collected from large companies. Both studies suggest that a combination of factors—use of third-party data together with enforcement capacity is more effective than either factor alone.

9. A lack of comparable data on the practices and characteristics of tax administrations hindered comprehensive empirical studies. While Crivelli (2018) conducted cross-country analyses on tax administration in Europe using data collected by the OECD, such data was not available for countries in other regions. Using the novel ISORA dataset, this paper fills the gap by comprehensively studying various practices and characteristics of tax administration across countries and their impacts on tax performance while controlling for other factors (such as macroeconomic environments, tax policy, and socio-political factors).

III. Data on Tax Administration Practices and Characteristics

10. The ISORA dataset provides a source of comparable data on the characteristics and practices of revenue administrations across income groups in all regions. The most recent data (for 2017) covers 37 Advanced Economies (AEs), 76 Emerging Markets (EMs), and 46 Low-Income Developing Countries (LIDCs). All regions are represented in the data collected. For the ISORA dataset, tax administrators of each country provide information on their operational practices and other characteristics based on questions and definitions agreed on by four partner international organizations. As the data is not in the public domain, inaccuracies may remain despite checking for inconsistencies in the data provided.⁶

11. Based on the ISORA data, the operational strength index of tax administration is constructed with seven sub-indices on specific tax administration practices and characteristics. The ISORA dataset consists of responses to both numerical and categorical questions, with 982 data points. We categorize these responses into seven different sub-indices and compile them into an overall operational strength index. While ISORA questions were not designed with the creation of indices in mind, the broad set of questions lends itself to thematic grouping.⁷ Features associated with facets of the administration’s operations and generally regarded as indicators of good practice are grouped to create each sub-index.⁸ The seven sub-indices are: (i) degree of autonomy; (ii) establishment of LTO or large taxpayer program; (iii) public accountability; (iv) service orientation; (v) CRM approach; (vi) use of third party data; and (vii) digitalization of services (depicted in Figure 1, and further detailed in Appendix 1). CRM, the use of third party data and digitalization of services are advocated to enhance compliance (IMF, 2015) and all these features have been recommended to reduce corruption potential with the administration (World Bank, 2019).

12. Closely looking at these indices, AEs tend to have higher values across them (Figure 2). The radar diagram points the strength of tax administration in AEs (relative to EMs and LIDCs), for all but one index on the establishment of an LTO.⁹ Almost all EMs and LIDCs have LTOs, while just under 80 percent of AEs indicate an LTO or program for large taxpayers.¹⁰ The box and whiskers plots of the operational strength index show a spread in value within each country grouping, with the mean and median value for AEs greater than that of EMs and LIDCs.

13. Two further descriptors of tax administrations have been derived from ISORA responses—the ratio of full-time-equivalents¹¹ to labor force, and the ratio of active taxpayers to the labor force. The first provides an indication of the scale of the tax administration. For the second, active taxpayers are taxpayers that interact with the tax administration, providing a proxy for the tax administration’s workload. It is worth noting that the number of active taxpayers is a product of tax policy, societal attitudes and norms, as well as efforts by the administration to ensure that all taxpayers comply with their tax obligations. The spread in these ratios by AE, EM, and LIDC groupings is illustrated in Figure 3 below.

14. The tax-to-GDP ratio is positively correlated with the operational strength index, as well as the ratios of staff to labor force and active taxpayers to labor force. As Figure 4 shows, the ratios of staff to labor force¹² and active taxpayers to labor force vary by one and two orders of magnitude respectively, exhibiting far smaller values in general for LIDCs than for AE or EM countries. In countries with a higher active taxpayer to labor force ratio, tax administrations are likely to have higher compliance risk management and third-party data indices.

15. There are high correlations between sub-indices of tax administrations. Table 1 on the correlation between the sub-indices generally indicates that countries with a stronger index on one aspect tend to be stronger on other indices. The service orientation index, for instance, correlates strongly with all other indices (LTO establishment aside). Given common underlying information management capability prerequisites, the correlations between the use of third-party data and both compliance risk management and the degree of digitalization are also high. Tax administration reform experience demonstrates that the major benefits of administrative reform arise from the strategic integration of multiple components, rather than the implementation of individual measures (IMF, 2015). These correlations, while not unexpected, present challenges in isolating exactly which characteristics or practices are associated with enhanced revenue performance.

^{Table 1.}

Correlations Among ISORA Sub-Indices

^{Table 1.}

Correlations Among ISORA Sub-Indices

	Compliance risk management	Third party data	Digitization	Service Orientation	Public Accountability	Autonomy	Staff/Labor Force	Active Taxpayers /Labor Force
Compliance Risk Management	1
Third party data	0.47	1
Digitalization	0.39	0.46	1
Service Orientation	0.47	0.51	0.45	1
Public Accountability	0.41	0.49	0.39	062	1
Autonomy	0.32	0.36	0.35	0.57	0.38	1
Staff/Labor Force	0.31	0.33	0.26	0.14	0.13	0.13	1
Active Taxpayer/Labor Force	0.48	0.61	0.37	0.33	0.36	0.22	0.55	1

^{Table 1.}

Correlations Among ISORA Sub-Indices

	Compliance risk management	Third party data	Digitization	Service Orientation	Public Accountability	Autonomy	Staff/Labor Force	Active Taxpayers /Labor Force
Compliance Risk Management	1
Third party data	0.47	1
Digitalization	0.39	0.46	1
Service Orientation	0.47	0.51	0.45	1
Public Accountability	0.41	0.49	0.39	062	1
Autonomy	0.32	0.36	0.35	0.57	0.38	1
Staff/Labor Force	0.31	0.33	0.26	0.14	0.13	0.13	1
Active Taxpayer/Labor Force	0.48	0.61	0.37	0.33	0.36	0.22	0.55	1

16. The recent Covid-19 pandemic is affecting tax administration (Box 2). With the pandemic, many countries are experiencing erosion of tax collection, reflecting the decline in economic activities and likely deterioration in taxpayer compliance. Thus, securing revenue streams and bringing taxpayers’ compliance behavior back to pre-crisis levels and beyond, will be a key priority for revenue administrations in the months and years ahead.

^{Box 2.}

Potential Impact of Covid-19 on Tax Administration

During the pandemic, tax administration is likely to be adversely impacted. Containment efforts for Covid-19 can slow down economic activities, while the easing of taxpayer obligations, such as extensions to filing and payment deadlines, together with the administration of crisis relief measures, adds complexity to the tasks of compliance monitoring and non-compliance response. As discussed in IMF (2020), during the crisis, filing, declaration, and payment compliance may deteriorate due to extended deadlines, limited availability of staff and taxpayers weakened financial positions. Face-to-face interactions of tax administration officials and taxpayers are limited, in the realm of both services and enforcement (IMF, 2020).¹ Most tax administrations have had to shift from face-to-face audits to desk audits.² For administrations with limited digital service offerings and/or remote access to data, these challenges are more severe. Furthermore, for the tax administration agencies with lower levels of staffing, any further reduction due to illness in staff may further undermine tax collection.

Strong tax administrative operations will be particularly important in the post Covid-19 recovery period. After the pandemic recedes, strengthening tax administration will be even more important, because of the critical needs to generate fiscal resources for addressing development needs and safeguarding debt sustainability through enhancing revenue mobilization. Enhanced efforts for strengthening tax administration should be sought, accompanied by appropriate reform efforts for tax policy. In the case of staff shortages, the focus should turn to prioritizing staffing for “critical” areas, based on a risk management approach. The expansion of digital services should be explored, where possible.

LIDCs may experience greater challenges in maintaining compliance and revenue collections during the pandemic and in the recovery period. They may be less able to work remotely effectively, due to a lower degree of digitalization. As shown in Figure 5, LIDCs are in general less equipped with CRM and digitalization. The figure further illustrates LIDCs’ lower level of tax administration staff that could pose challenges for business continuity during the pandemic.

¹ FAD Notes on ‘Tax and Customs Administration Responses’ and ‘Revenue Administration: Reinvigorating Operations to Safeguard Collection and Compliance,’ available from https://www.imf.org/en/Publications/SPROLLs/covid19-special-notes#fiscal.² In a survey of 89 tax and customs administrations (conducted by the IMF in April 2020), over 70 percent indicated that measures easing taxpayer obligations had been implemented and 64 percent had made adjustments to their audit program, with over half of the administrations noting a shift from face-to-face to desk audits.

IV. Empirical Methodolgy

17. To empirically explore the impact of practices and characteristics of tax administrations, the two-step approach proposed by Hsiao (2003) is used. The ISORA dataset is only available for 2014–17, and there is little variation in tax administration features within such a short period. Thus, the typical panel estimation with fixed effects cannot be applied.¹³ So, we employ the “two-step approach” proposed by Hsiao (2003). Several studies have adopted this approach to address the problem posed by time-invariant variables in the panel dataset.¹⁴

First Step

18. At the first step, we estimate the coefficients for the time-varying control variables, including tax policy changes, using a fixed effect panel regression. The model for the first stage regression can be written as follows:

y_i,t = β₀X_i,t+u_i+γ_t+ε_it

where _y is the tax to GDP ratio excluding trade taxes and social security contributions, X is a set of time variant control variables, u_i is a set of estimated tax collection after controlling other factors (i.e., country fixed effects), γ_t is a set of time fixed effects, and ε_it is the residual.¹⁵ As time varying control variables, macro-economic data such as GDP, CPI, trade openness, and external debt are included. We also control for the structure of economy by adding the share of oil export and agriculture out of GDP.

19. At this first step, we also employ a novel approach to control for tax policy changes. In examining tax performance, controlling for tax policy changes is technically challenging, as there is no convenient variable representing tax policy of each country. In the absence of information on effective tax rates across countries, we use “projected” tax revenues as a proxy for the revenues reflecting all anticipated changes of tax policy. As the projected revenues may reflect not only tax policy changes but some gains from improving the tax administration, we also test other indicators as robustness check.¹⁶ These proxies have their own deficiencies in measuring tax policy. Despite the limitations, however, our consistent findings suggest robustness.

The difference between the budgeted revenue of this year as share of projected GDP and the actual revenue of last year as share of actual GDP:¹⁷ The budgeted revenue reflects planned tax policies and the economic forecast when the budget is prepared. Since macroeconomic changes are controlled separately in the regressions and within-year tax policy changes are rare, this gap can reflect the effect from changes in tax policy.¹⁸
The difference between the budgeted revenue of this year as share of projected GDP and the budgeted revenue in last year as share of projected GDP: While less accurate a proxy for the tax policy changes in year t, this gap can reduce possible systematic biases in budget process such as continuous over or underperformance of actual revenue vs. the revenue projections in the budget.
Top tax rates: Changes in tax rates may lead to change in tax collections. Since our tax rate data from the IMF Tax Rate Database (DART) record only the top tax rates for tax types, any tax policy changes such as tax cuts, changing the tax bracket, and exemptions which does not affect top rate cannot be captured.

Second Step

20. At the second step, the estimated fixed effect values (from the first step) are regressed on the time-invariant tax administration variables. This step looks at the part of tax collection (as share of GDP) for each country, i, that cannot be explained by the time varying variables in the first stage, u_i and estimates the relationship between this and the practices and characteristics of countries’ tax administration using ISORA data. The model for the second stage regression can be written as follows:

{\hat{u}}_{ι} = β_{1} I_{i} + h_{i}

where ${\hat{u}}_{ι}$ is the estimated tax collection after controlling for other factors from the first stage, I_i is the set of time invariant tax administration’ practices and characteristics, and h_i is the unobserved heterogeneity.¹⁹

21. The role of the practices and characteristics of tax administration agencies is investigated at three different levels. First, we develop a compound index measuring the operational strength of the agency by averaging the individual sub-indices. Second, we run regressions with the individual sub-indices to test which practices and characteristics are positively correlated with tax outcomes. This approach reduces the possible multicollinearity problem given that the good practices and characteristics are correlated with each other. Lastly, we try to investigate the impact of multiple features of tax administration by including all tax administration sub-indices in the regression and distinguishing the significant ones that are tightly associated with tax revenue performance. This complements the second approach, as the inclusion of all variables may lead to multicollinearity, while regressions with one variable only may lead to omitted variable problem.

V. Empirical Findings

First-Stage Regression Results

22. As expected, macro-economic and economic structure variables are significantly associated with tax collections, so controlling for their effect from the data is necessary (Table 2).²⁰ Regardless of the choice of proxies for tax policy, per capita GDP is positively associated with tax performance as it is expected to be a good indicator of the overall level of economic development (Gupta, 2007). The results also find that the coefficient for trade openness (measured by the combined share of exports and import) is positive and statistically significant, consistent with the literature (Leuthold, 1991; Ghura,1998). The effect of inflation on tax revenue is negative, possibly reflecting the Olivera-Tanzi effect (the lags in revenue collection reduce revenue relative to GDP). The degree of external indebtedness of a country turns out to be insignificant suggesting its mixed effects on tax collections.²¹ To control for the structure of the economy, we also include variables for oil export and agriculture as a share of GDP. The coefficient of oil exports is positive and statistically significant (likely reflecting larger tax collection from the sector in larger oil exporters). The share of the agricultural sector in GDP turns out to be insignificant (likely reflecting possible multicollinearity between the share of agriculture and per capita GDP). The coefficients of all proxies of tax policy are positive and statistically significant, suggesting a robust effect on collections, as expected.

^{Table 2.}

First Stage Regression Outcomes

Note: Estimation of equation by panel fixed effect. Standard Errors in Parentheses. *, **, *** denote statistical significance at the 1, 5, and 10 percent levels, respectively. Bold italics means statistically significant at least 10 percent level.

^{Table 2.}

First Stage Regression Outcomes

Dependent variable: Tax revenue/GDP	(1) Tax Rate	(2) Bdgt-Act	(3) Bdgt-Bdgt
*Log (PerCapitaGDP_PPP), lagged*	0.1810*** (0.0402)	0.1374*** (0.0406)	0.1417*** (0.0419)
*Square log (PerCapitaGDP_PPP), lagged*	-0.0110*** (0.0024)	-0.0084*** (0.0024)	-0.0088*** (0.0025)
*Trade openness (out of GDP), lagged*	0.0228*** (0.0058)	0.0236*** (0.0066)	0.0251*** (0.0067)
External debt (out of GDP), lagged	-0.0005 (0.0033)	-0.0038 (0.0046)	-0.0063 (0.0051)
*CPI, lagged*	-0.0094 (0.0119)	-0.0348*** (0.0131)	-0.0308** (0.0128)
Terms of Trade (2000=1), lagged	0.0020 (0.0038)	0.0021 (0.0039)	0.0038 (0.0039)
*Oil exports (as a % of GDP), lagged*	0.1414*** (0.0254)	0.1570*** (0.0224)	0.1452*** (0.0225)
Log (Agri, %GDP), lagged	0.0029 (0.0046)	-0.0032 (0.0054)	-0.0006 (0.0056)
Control Corruption, lagged	0.0056 (0.0044)	0.0027 (0.0047)	0.0047 (0.0049)
Top Combined CIT Rate (%), lagged	0.0002 (0.0003)
Top Combined PIT Rate (%), lagged	0.0002 (0.0002)
*Standard VAT Rate (%), lagged*	0.0030*** (0.0006)
*Budgeted (t) minus Actual (t-1) tax rev.*		0.1273*** (0.0407)
*Budgeted (t) minus Budgeted (t-1) rev.*			0.1953*** (0.0323)
Constant	-0.6443*** (0.1655)	-0.4239** (0.1707)	-0.4323** (0.1744)
Observations	675	894	830
R-squared	0.2301	0.1984	0.2441
Number of countries	105	127	125
Country FE	YES	YES	YES
Year FE	YES	YES	YES

^{Table 2.}

First Stage Regression Outcomes

Dependent variable: Tax revenue/GDP	(1) Tax Rate	(2) Bdgt-Act	(3) Bdgt-Bdgt
*Log (PerCapitaGDP_PPP), lagged*	0.1810*** (0.0402)	0.1374*** (0.0406)	0.1417*** (0.0419)
*Square log (PerCapitaGDP_PPP), lagged*	-0.0110*** (0.0024)	-0.0084*** (0.0024)	-0.0088*** (0.0025)
*Trade openness (out of GDP), lagged*	0.0228*** (0.0058)	0.0236*** (0.0066)	0.0251*** (0.0067)
External debt (out of GDP), lagged	-0.0005 (0.0033)	-0.0038 (0.0046)	-0.0063 (0.0051)
*CPI, lagged*	-0.0094 (0.0119)	-0.0348*** (0.0131)	-0.0308** (0.0128)
Terms of Trade (2000=1), lagged	0.0020 (0.0038)	0.0021 (0.0039)	0.0038 (0.0039)
*Oil exports (as a % of GDP), lagged*	0.1414*** (0.0254)	0.1570*** (0.0224)	0.1452*** (0.0225)
Log (Agri, %GDP), lagged	0.0029 (0.0046)	-0.0032 (0.0054)	-0.0006 (0.0056)
Control Corruption, lagged	0.0056 (0.0044)	0.0027 (0.0047)	0.0047 (0.0049)
Top Combined CIT Rate (%), lagged	0.0002 (0.0003)
Top Combined PIT Rate (%), lagged	0.0002 (0.0002)
*Standard VAT Rate (%), lagged*	0.0030*** (0.0006)
*Budgeted (t) minus Actual (t-1) tax rev.*		0.1273*** (0.0407)
*Budgeted (t) minus Budgeted (t-1) rev.*			0.1953*** (0.0323)
Constant	-0.6443*** (0.1655)	-0.4239** (0.1707)	-0.4323** (0.1744)
Observations	675	894	830
R-squared	0.2301	0.1984	0.2441
Number of countries	105	127	125
Country FE	YES	YES	YES
Year FE	YES	YES	YES

Second-Stage Regression Results

23. The fixed-effect values estimated in the first stage are regressed against: (i) an operational strength index (aggregated ISORA sub-indices); (ii) each sub-index of ISORA data separately; and (iii) all sub-indices in the regression at the same time.

24. The regression results on the operational strength confirm that the strength of tax administration, as measured by the index compiled from ISORA data, is highly relevant for tax performance (Table 3). ²² The estimated tax collection after controlling for the influence of other relevant factors from the first stage—the estimated country fixed effect—was regressed on the operational strength index of tax administration, which is the average score of seven sub-indices ranging from 0 and 1. The coefficients of the index are positive and statistically significant, and the finding also holds when the sample is restricted to emerging and low income countries. Variables considering the relative size of the administration such as the share of staff in tax administration agency in a country’s total labor force and the share of active taxpayers in the labor force are included to control for features of the agencies that might not be captured by the operational strength index.

^{Table 3.}

Second Stage Regression Outcomes (Panel Between Estimation) ²³

Note 1: Above second stage regression outcomes are based on the first stage regression using the difference between the budgeted and actual tax revenue as a proxy for the tax policy.²⁴Note 2: Estimation of equation by panel between effect with least Square Estimation. Standard Errors in Parentheses. *, **, *** denote statistical significance at the 1, 5, and 10 percent levels, respectively. Bold italics means statistically significant; at least 10 percent level.

^{Table 3.}

Second Stage Regression Outcomes (Panel Between Estimation) ²³

Dependent variable: Estimated	All Countries			EMs and LIDCs
Country Fixed Effects	(1)	(2)	(3)	(4)	(5)	(6)
*Operational Strength*	0.2226*** (0.0349)	0.1767*** (0.0358)	0.1228*** (0.0417)	0.1546*** (0.0357)	0.1485*** (0.0377)	0.1277*** (0.0438)
*#Tax Staff/LaborForce*		0.0802*** (0.0235)	0.0535** (0.0250)		0.0472* (0.0278)	0.0487 (0.0366)
*Sq(#Tax Staff/LaborForce)*		-0.0157* (0.0083)	-0.0130 (0.0087)		-0.0058 (0.0112)	-0.0110 (0.0171)
*ActiveTaxpayer/LaborForce*			0.0443*** (0.0122)			0.0346* (0.0196)
Constant	-0.1187*** (0.0222)	-0.1444*** (0.0235)	-0.1122*** (0.0273)	-0.0729*** (0.0215)	-0.1027*** (0.0245)	-0.0965*** (0.0297)
R-squared	0.2911	0.4937	0.5753	0.1935	0.3672	0.3941
Number of countries	101	85	70	80	64	54

^{Table 3.}

Second Stage Regression Outcomes (Panel Between Estimation) ²³

Dependent variable: Estimated	All Countries			EMs and LIDCs
Country Fixed Effects	(1)	(2)	(3)	(4)	(5)	(6)
*Operational Strength*	0.2226*** (0.0349)	0.1767*** (0.0358)	0.1228*** (0.0417)	0.1546*** (0.0357)	0.1485*** (0.0377)	0.1277*** (0.0438)
*#Tax Staff/LaborForce*		0.0802*** (0.0235)	0.0535** (0.0250)		0.0472* (0.0278)	0.0487 (0.0366)
*Sq(#Tax Staff/LaborForce)*		-0.0157* (0.0083)	-0.0130 (0.0087)		-0.0058 (0.0112)	-0.0110 (0.0171)
*ActiveTaxpayer/LaborForce*			0.0443*** (0.0122)			0.0346* (0.0196)
Constant	-0.1187*** (0.0222)	-0.1444*** (0.0235)	-0.1122*** (0.0273)	-0.0729*** (0.0215)	-0.1027*** (0.0245)	-0.0965*** (0.0297)
R-squared	0.2911	0.4937	0.5753	0.1935	0.3672	0.3941
Number of countries	101	85	70	80	64	54

25. The square term of tax administration staff controls for the non-linear relationship between the size of staff and tax performance. The coefficients for this term—marginally significant for all countries and insignificant for EMs and LIDCs—are negative. This suggests the decreasing marginal tax gains as the staff level incrementally increases among all countries, leading to a threshold where additional staffing does not increase revenue. However, EMs and LIDCs do not reach this threshold yet, implying potential for ramping up revenue by hiring staff (see Appendix 4 for details).

26. Based on the results, we further estimate the possible magnitude of tax gains by improving the operational strength of the tax administration agency. Among EMs and LIDCs, a country at the top 25 percentile threshold in terms of operational strength collects larger tax revenues by 3.25 percent of GDP than a country at the lowest 25 percent threshold, assuming other conditions are equal (we estimate tax collections at both quartiles using the estimates from the column (5) in Table 3).²⁵ Though our analysis does not prove strict causality between operational strength and tax outcomes, the estimated results suggest that a country with stronger tax operational capacity collects significantly more tax revenues in EMs and LIDCs.

27. The results on each ISORA sub-index suggest that most indices on practices and characteristics of tax administration are positive and statistically significant. Here we regress the country fixed effects from the first stage on these sub-indices separately to identify which features of tax administration play an important role in explaining the tax collection. The regressions found that six sub-indices out of seven are significant. Interestingly, the goodness of fit measures (R-squares) are quite varying among sub-indices, which provides a rough guidance to the relative importance of specific features in explaining tax performances. The establishment of the LTO was not found to be significant. Most countries indicated the existence of an LTO, but the nature of program for large taxpayers varies considerably, according to data provided in ISORA. While many countries segmented taxpayers by size for ease of collections, including programs for LTOs, the impact of the establishment of an LTO is not confirmed. This suggests that the LTO needs to be effectively operationalized (see Appendix 8 for details).

Table 4.

Second Stage Regression Outcomes with Sub-Indices – Single Variate²⁶

Note: Estimation of equation by panel between effect with least Square Estimation. Standard Errors in Parentheses. *, **, *** denote statistical significance at the 1, 5, and 10 percent levels, respectively. Bold italics means statistically significant at least 10 percent level.

Table 4.

Second Stage Regression Outcomes with Sub-Indices – Single Variate²⁶

Dependent var: Estimated Country Fixed Effects	(1)	(2)	(3)	(4)	(5)	(6)	(7)
*Comp. Risk Management*	0.1154*** (0.0184)
*Third Party Data*		0.1136*** (0.0181)
*Digitalization*			0.0830*** (0.0197)
*Service Orientation*				0.1175*** (0.0348)
*Public Accountability*					0.0874*** (0.0286)
*Autonomy*						0.0809** (0.0322)
Est. LTO							-0.0245 (0.0325)
Constant	-0.0457*** (0.0117)	-0.0240*** (0.0089)	-0.0273** (0.0125)	-0.0592** (0.0238)	-0.0245 (0.0155)	-0.0408* (0.0245)	0.0416 (0.0317)
R-squared	0.2835	0.2843	0.1515	0.1034	0.0860	0.0600	0.0057
Number of countries	101	101	101	101	101	101	101

Table 4.

Second Stage Regression Outcomes with Sub-Indices – Single Variate²⁶

Dependent var: Estimated Country Fixed Effects	(1)	(2)	(3)	(4)	(5)	(6)	(7)
*Comp. Risk Management*	0.1154*** (0.0184)
*Third Party Data*		0.1136*** (0.0181)
*Digitalization*			0.0830*** (0.0197)
*Service Orientation*				0.1175*** (0.0348)
*Public Accountability*					0.0874*** (0.0286)
*Autonomy*						0.0809** (0.0322)
Est. LTO							-0.0245 (0.0325)
Constant	-0.0457*** (0.0117)	-0.0240*** (0.0089)	-0.0273** (0.0125)	-0.0592** (0.0238)	-0.0245 (0.0155)	-0.0408* (0.0245)	0.0416 (0.0317)
R-squared	0.2835	0.2843	0.1515	0.1034	0.0860	0.0600	0.0057
Number of countries	101	101	101	101	101	101	101

28. Among other variables, enhancing CRM and actively using third-party data show stronger correlation with tax collections. When all tax administration sub-indices together with quantitative variables are included in the second stage regressions, CRM and the use of third-party data remain significant, while other variables become insignificant (Table 5). It implies that given the tax administrations’ limited resources, priority should be given to the enhancement of the tax administration’s ability to identify taxpayers’ risk profiles, apply different treatments to different risk groups, and establish computer-based information system enabling data from third parties. This is not to say that the other activities do not matter—but, as a matter of priority, removing bottlenecks in compliance risk management and use of third-party data seems to deliver stronger tax collection.

^{Table 5.}

Second Stage Regression Outcomes – Multivariate

^{Table 5.}

Second Stage Regression Outcomes – Multivariate

Dependent variable: Estimated Country Fixed Effects	(1)	(2)	(3)
*Compliance Risk Management*	0.0830*** (0.0212)	0.0772*** (0.0226)	0.0873*** (0.0247)
*Third Party Data*	0.0763*** (0.0199)	0.0607*** (0.0206)	0.0612** (0.0237)
Digitalization	0.0228 (0.0215)	0.0098 (0.0220)	0.0016 (0.0225)
Service Orientation	-0.0297 (0.0438)	-0.0295 (0.0432)	-0.0425 (0.0472)
Est. LTO	-0.0087 (0.0271)	-0.0162 (0.0271)	0.0123 (0.0295)
Public Accountability	-0.0033 (0.0314)	0.0112 (0.0318)	-0.0180 (0.0340)
Autonomy	0.0160 (0.0315)	0.0140 (0.0319)	0.0103 (0.0331)
*#Tax Staff/LaborForce*		0.0725*** (0.0228)	0.0641** (0.0242)
*Sq (#Tax Staff/LaborForce)*		-0.0153* (0.0081)	-0.0177** (0.0086)
*ActiveTaxpayer/LaborForce*			0.0301** (0.0132)
Constant	-0.0508 (0.0332)	-0.0809** (0.0350)	-0.0891** (0.0375)
R-squared	0.4210	0.5755	0.6492
Number of countries	101	85	70

^{Table 5.}

Second Stage Regression Outcomes – Multivariate

Dependent variable: Estimated Country Fixed Effects	(1)	(2)	(3)
*Compliance Risk Management*	0.0830*** (0.0212)	0.0772*** (0.0226)	0.0873*** (0.0247)
*Third Party Data*	0.0763*** (0.0199)	0.0607*** (0.0206)	0.0612** (0.0237)
Digitalization	0.0228 (0.0215)	0.0098 (0.0220)	0.0016 (0.0225)
Service Orientation	-0.0297 (0.0438)	-0.0295 (0.0432)	-0.0425 (0.0472)
Est. LTO	-0.0087 (0.0271)	-0.0162 (0.0271)	0.0123 (0.0295)
Public Accountability	-0.0033 (0.0314)	0.0112 (0.0318)	-0.0180 (0.0340)
Autonomy	0.0160 (0.0315)	0.0140 (0.0319)	0.0103 (0.0331)
*#Tax Staff/LaborForce*		0.0725*** (0.0228)	0.0641** (0.0242)
*Sq (#Tax Staff/LaborForce)*		-0.0153* (0.0081)	-0.0177** (0.0086)
*ActiveTaxpayer/LaborForce*			0.0301** (0.0132)
Constant	-0.0508 (0.0332)	-0.0809** (0.0350)	-0.0891** (0.0375)
R-squared	0.4210	0.5755	0.6492
Number of countries	101	85	70

29. The size of staff in tax administration is also significantly associated with tax collections, up to a point. The share of tax staff out of labor force is positively associated with tax collection while its quadratic term is negatively associated with tax collections. It suggests that tax performance could improve with increasing size of staff in tax agencies, but there is an optimal size beyond which this effect is no longer present.

Robustness Checks

30. These results are robust in several directions. We conduct several robustness checks:²⁷

Control variables: We also consider selecting control variables with a low level of correlation in order to reduce multicollinearity problems given tax administration sub-indices are interlinked to each other (as depicted in Table 1). To test the significance of each of the sub-indices (e.g., degree of digitalization, service orientation), we drop the variables with correlation higher than 0.4 and conduct panel between regression analysis across all countries.²⁸ Our main results broadly remain unchanged with the regressions using these selection processes of control variables. The variables significant in our baseline models stay significant while insignificant variables in our baseline models remain insignificant. Some variables showing negative signs (for instance service orientation and public accountability) in case all sub-indices are included as regressors now have expected positive signs though still insignificant.
Limiting the sample to EMs and LIDCs: One may argue that our findings are driven largely by the strong tax revenue performances and good practices and characteristics of AE tax administrations. We tested our models only to EMs and LIDCs and the findings remain broadly unchanged except that the quadratic term of staffing loses its significance, implying that these tax administration agencies are mostly understaffed. It again supports our main theme: enhancing key practices and characteristics of tax administration could reap significant gains in tax collections among the countries with strong needs for mobilizing domestic revenues.
Keeping the same sample size across the first stage regressions: Given that we are using three proxies separately to control for the effects of changing tax policies in the first stage, the sizes of sample in the first stage are different across the selection of tax policy proxies (Table 2). While two proxies (i.e., the difference between the budgeted revenue and tax revenue, and the difference between the budgeted revenue of this year and the last year) broadly capture the changes in tax bases, the tax rates proxy looks narrowly at the changes in top tax rates in CIT, PIT, and VAT. As robustness checks, we ran regressions keeping the same sample size across the three regressions and the outcomes are quite similar to the ones presented in Table 2, which suggests that including the tax base does not have any significant impact and hence we use the full sample as in the presentation.
Calculating standard errors with different econometric methods: Our baseline regressions adopt the weighted least square method to correct for heteroskedasticity. Different techniques for estimating the standard errors are also employed such as the Bootstrapping and Huber-White estimation. We found the regression outcomes are robust to the techniques.
Additional variables in the second stage: We also consider variables presumably static and reflecting structural factors affecting tax collections (e.g., land-locked country, inequality, informality of the economy, and control of corruption). ²⁹ The operational strength of tax administrations remains significant with the regressions using these additional control variables in the second stage.

VI. Conclusions

31. The design and practices of tax administration agencies matter significantly for tax performance. As far as we know, ours is a novel research empirically examining the importance of the design features and actual practices of tax administration agencies on tax collection. The operational strength of the agency is closely associated with tax collections—among EMs and LIDCs, countries at the top 25 percent (in terms of operational strength index) collect larger tax revenues (by 3.25 percent of GDP) than countries at the lowest 25 percent, assuming other conditions are equal.³⁰

32. Our empirical analyses emphasize the importance of CRM and the use of third-party data. Our analyses find that adopting key tax administration practices and characteristics—such as stronger CRM, active use of third-party data, higher level of digitalization, enhancing service orientation, public accountability, and autonomy—are associated with higher tax collections, with CRM and the use of third-party data seemingly delivering stronger tax collection. To enhance tax collection, tax administration reform efforts could prioritize: (i) strengthening CRM by adopting automated risk profiling and electronic audits; and (ii) utilizing third party data by adopting computer systems for processing the data and prefilling returns. These efforts would likely also facilitate effective functioning of LTOs.

33. Furthermore, properly staffing tax administration agencies is important, but excessive staffing may not necessarily bring intended effects. Our analyses find that larger staffing of a tax administration agency improves revenue performance but its marginal benefit decreases as staff increase and becomes counter-productive. To this end, it is worth noting that, among all countries, the critical threshold point seems to be 0.25 percent of labor force, beyond which the marginal benefit of additional staffing may turn negative. Interestingly, our findings suggest that EMs and LICDs have not passed this threshold and additional staffing is likely to be associated with higher tax collection.

34. This paper also empirically confirms the importance of broadening a taxpayer base. The relative size of the number of active taxpayers is found to be relevant for tax collection. Fuller analyses by tax item and type of taxpayers (e.g., corporate and individual) would be needed to draw more robust conclusion. For countries where the relative size of active taxpayers is low (e.g., LIDCs), it is relevant to identify if this is owing to any issues associated with tax administration or tax policy (e.g., tax exemptions), with a view to exploring an appropriate strategy to address it.

35. In light of challenges faced by tax administration during the recent global pandemic situation, these findings have important policy implications. Countries with weaker tax administrations could be more susceptible to relatively large tax revenue losses during the pandemic. They should aim to prioritize and address specific risks to revenue arising out of pandemic-related measures implemented to provide relief to taxpayers (IMF, 2020). In all countries, striving to preserve taxpayer compliance using data from third parties through digital technology would be useful in mitigating the impact of pandemics on budget revenues and maintaining continuity in tax administration. Moreover, where the tax administration is already understaffed, preserving existing staff is likely to pay off well in terms of safeguarding revenue collections.

References

Akitoby, Bernardin, Anja Baum, Clay Hackney, Olamide Harrison, Keyra Primus, and Veronique Salins. “Tax Revenue Mobilization Episodes in Emerging Markets and Low-Income Countries: Lessons from a New Dataset,” IMF Working Paper 118/234, International Monetary Fund, Washington, D. C., 2018.
- Search Google Scholar
- Export Citation
Alesina, Alberto, Ricardo Hausmann, Rudolf Hommes, and Ernesto Stein. “Budget Institutions and Fiscal Performance in Latin America.” Journal of Development Economics 59.2 (1999): 253–273.
- Search Google Scholar
- Export Citation
Almunia, Miguel, and David Lopez-Rodriquez. “Under the Radar: The Effects of Monitoring Firms on Tax Compliance,” American Economic Journal: Economic Policy, 10 (1), 2018.
- Search Google Scholar
- Export Citation
Baum, Anja, Sanjeev Gupta, Elijah Kimani, and Sampawende Jules Tapsoba. “Corruption, Taxes and Compliance,” IMF Working Paper 17/255. International Monetary Fund, Washington, D. C. 2017.
- Search Google Scholar
- Export Citation
Baum, Anja, Clay Hackney, Paulo Medas, and Mouhamadou Sy. “Governance and State-Owned Enterprises: How Costly Is Corruption?” IMF Working Paper 19–253, International Monetary Fund, Washington, D. C., 2019.
- Search Google Scholar
- Export Citation
Besley, Timothy, and Torsten Persson. “The origins of state capacity: Property rights, taxation, and politics.” American economic Review 99.4 (2009): 1218–44.
- Search Google Scholar
- Export Citation
Boukbech, Rachid, Ahmed Bousselhami and Elhadj Ezzahid. “Determinants of Tax Revenues: Evidence From a Sample of Lower Middle Income Countries,” MPRA Paper No. 90268, München, 2018.
- Search Google Scholar
- Export Citation
CASE, “Study and Reports on the VAT Gap in the EU-28 Member States: 2018 Final Report,” CASE Report No. 496, Warsaw, 2018.
- Search Google Scholar
- Export Citation
Crivelli, Ernesto. “A Basic Tool to Assess Tax Administration Strength In Emerging Europe.” Economics of Transition, 2018.
- Search Google Scholar
- Export Citation
Dabla-Norris, Era, Florian Misch, Duncan Cleary, and Munawer Khwaja. “Tax Administration and Firm Performance: New Data and Evidence for Emerging Market and Developing Economies.” International Monetary Fund, Washington D. C., 2017.
- Search Google Scholar
- Export Citation
Dom, Roel. Semi-Autonomous Revenue Authorities in Sub-Saharan Africa: Silver Bullet or White Elephant, CREDIT Research Paper, No. 17/01, The University of Nottingham, Centre for Research in Economic Development and International Trade (CREDIT), Nottingham, 2017.
- Search Google Scholar
- Export Citation
Ebeke, Christian, Mario Mansour, and Grégoire Rota Graziosi. “The Power to Tax in Sub-Saharan Africa: LTUs, VATs, and SARAs,” FERDI Working Paper 154, Clermont-Ferrand, 2016.
- Search Google Scholar
- Export Citation
Fenochietto, Ricardo, and Carola Pessino. “Understanding Countries’ Tax Effort,” IMF Working Paper 13/244, International Monetary Fund, Washington D. C., 2013.
- Search Google Scholar
- Export Citation
International Monetary Fund, 2015. “Current Challenges in Revenue Mobilization: Improving Taxpayer Compliance,” Staff Report, International Monetary Fund, Washington D. C., 2015.
- Search Google Scholar
- Export Citation
International Monetary Fund, 2018. “Domestic Revenue Mobilization and Private Investment,” Regional Economic Outlook: Sub-Saharan Africa, International Monetary Fund, Washington D. C., 2018.
- Search Google Scholar
- Export Citation
International Monetary Fund, 2019. “Curbing Corruption,” Fiscal Monitor, Washington D. C., April 2019.
- Search Google Scholar
- Export Citation
International Monetary Fund, 2020. “Revenue Administration: Reinvigorating Operations to Safeguard Collection and Compliance,” Special Series on Fiscal Policies to Respond to COVID-19, IMF Fiscal Affairs Department, International Monetary Fund.
- Search Google Scholar
- Export Citation
Gaspar, Vitor, Laura Jaramillo, and Philippe Wingender. “Tax Capacity and Growth: Is There a Tipping Point?” IMF Working Paper 16/234, International Monetary Fund, Washington D. C., 2016.
- Search Google Scholar
- Export Citation
Greene, William. “Fixed Effects Vector Decomposition: A Magical Solution to the Problem of Time-Invariant Variables in Fixed Effects Models?” Political Analysis, 19:135–46, 2011.
- Search Google Scholar
- Export Citation
Gupta, Abhijit Sen. “Determinants of Tax Revenue Efforts in Developing Countries,” IMF Working Paper 07/184, Washington D. C., 2007.
- Search Google Scholar
- Export Citation
Ghura, Dhaneshwar. Tax revenue in Sub-Saharan Africa: Effects of economic policies and corruption. No. 98–135. International Monetary Fund, 1998.
- Search Google Scholar
- Export Citation
Hsiao, Cheng. “Analysis of Panel Data,” Cambridge University Press, United Kingdom, 2003.
- Search Google Scholar
- Export Citation
Kidd, Maureen, and William Joseph Crandall. “Revenue Authorities: Issues and Problems in Evaluating their Success.” Journal Issue 2006: 240.
- Search Google Scholar
- Export Citation
Leuthold, Jane H. “Tax shares in developing economies a panel study.” Journal of development Economics 35.1 (1991): 173–185.
- Search Google Scholar
- Export Citation
Morrissey, Oliver, et al. “Tax Revenue Performance and Vulnerability in Developing Countries,” Journal of Development Studies, 52:12, 2016.
- Search Google Scholar
- Export Citation
Ngoma, Marina Mavungu, and Nikolina Krsic. “Assessing the impact of tax administration reforms in Sub-Saharan Africa,” MPRA Paper No. 89275, München, 2017.
- Search Google Scholar
- Export Citation
Plümper, Thomas, and Vera E. Troeger. “Efficient Estimation of Time-Invariant and Rarely Changing Variables in Finite Sample Panel Analyses with Unit Fixed Effects.” Political Analysis 15.2 (2007): 124–139.
- Search Google Scholar
- Export Citation
Pomeranz, Dina. “No Taxation without Information: Deterrence and Self-Enforcement in the Value Added Tax,” American Economic Review, 105 (8), 2015.
- Search Google Scholar
- Export Citation
Russell, Barrie. Revenue Administration: Developing a Taxpayer Compliance Program. No. 2010/17. International Monetary Fund, 2010.
- Search Google Scholar
- Export Citation
Schlotterbeck, Stephane. “Tax Administration Reforms in the Caribbean ‘Challenges, Achievements, and Next Steps’”, IMF Working Paper 17/88, Washington D. C., 2017.
- Search Google Scholar
- Export Citation
World Bank. “Innovations in Tax Compliance Conceptual Framework”, Policy Research Working Paper 9032, World Bank Group, Washington D. C., 2019.
- Search Google Scholar
- Export Citation
Yohou, Hermann Djedje and Michaël Goujon. “Reassessing Tax Effort in Developing Countries: a Proposal of a Vulnerability-Adjusted Tax Effort Index (VATEI),” Working Paper 186. Fondation pour les Études et Recherches sur le Développement International, 2017.
- Search Google Scholar
- Export Citation

Appendix I. Sub-Indices Compiled to Reflect Facets of Revenue Administration³¹

Sub-Indices	Components	Description	Weight
Compliance risk management	3	Formal approach for identifying; Assessing and prioritizing key compliance risks; Automated risk profiling; electronic audit	1/3 1/3 1/3
Use of third-party data	12	Computer based information systems for processing various forms of third-party data: Financial institutions, International exchange, Online trading, Wage and Salary, Insurance company, Property sale, Other government agencies, Asset leasing, Prescribed contractors with report of payment, VAT invoices, and Others Use of third-party data in prefilling returns	1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/2
Degree of digitalization	10	E-filing mandatory for some/all taxpayers E-payment mandatory for some/all taxpayers Web-based information and communication services; Tools and calculators on the webpage, Online application for taxpayer, Capture data from third parties, Digital mailbox for communication with taxpayers, Information on the webpage, Integrated taxpayer account, Electronic invoicing system, and Others	1/3 1/3 1/24 1/24 1/24 1/24 1/24 1/24 1/24 1/24
Service orientation	14	Measures to facilitate taxpayer compliance and improve services to taxpayers: Have a formal service strategy, Have a formal set of service standards, Conduct taxpayer satisfaction surveys, Registration possible through other agencies, Provision of rulings to taxpayers, Availability of online application for taxpayers, Make special provisions for taxpayers with disabilities, Provision of services in unofficial language(s), Use of information on compliance burden, End-user testing of new services, End-user involvement in design of new services, Simultaneous Registration for multiple tax types, Formal document covering taxpayer rights, and Mechanism for managing taxpayer complaints	1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14
Public accountability	12	Measures enhancing tax agency’s accountability to the public: Publish its strategic plan, Publish its annual business/operation plan Make public a formal set of service delivery standards Publish its achievements vis-a-vis the standards Publish its annual reports Use of an external auditor Make key compliance risk public Make reports of outcomes in addressing compliance risk public Publish the results of taxpayer satisfaction surveys Document that formally set out taxpayer rights Have a specific mechanism for managing taxpayer complaints, and Publish periodic estimates of the tax gap	1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12
Autonomy	14	Institutional form, degree of autonomy in managing expenditure and human resources: Autonomous vs. operating within Ministry, Discretion over designing internal structure, Discretion over operational budget, Discretion over capital budget, Authority to set performance standards, Determination of work requirements, Appointment of new staff, Promotion of existing staff, Decide on qualifications for appointment, Decision whether work is carried out by permanent or contractual staff, Placement of staff in salary band, Termination of employment, Responsible for debt collection and enforcement, and Provision of tax policy advice	1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14
Large Taxpayer Office (LTO)	1	An office or a program dedicated to large taxpayers is in operation	1
Operational Strength	7 sub-indices(65 individual components)	Compound index determined as average of seven indices

Sub-Indices	Components	Description	Weight
Compliance risk management	3	Formal approach for identifying; Assessing and prioritizing key compliance risks; Automated risk profiling; electronic audit	1/3 1/3 1/3
Use of third-party data	12	Computer based information systems for processing various forms of third-party data: Financial institutions, International exchange, Online trading, Wage and Salary, Insurance company, Property sale, Other government agencies, Asset leasing, Prescribed contractors with report of payment, VAT invoices, and Others Use of third-party data in prefilling returns	1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/22 1/2
Degree of digitalization	10	E-filing mandatory for some/all taxpayers E-payment mandatory for some/all taxpayers Web-based information and communication services; Tools and calculators on the webpage, Online application for taxpayer, Capture data from third parties, Digital mailbox for communication with taxpayers, Information on the webpage, Integrated taxpayer account, Electronic invoicing system, and Others	1/3 1/3 1/24 1/24 1/24 1/24 1/24 1/24 1/24 1/24
Service orientation	14	Measures to facilitate taxpayer compliance and improve services to taxpayers: Have a formal service strategy, Have a formal set of service standards, Conduct taxpayer satisfaction surveys, Registration possible through other agencies, Provision of rulings to taxpayers, Availability of online application for taxpayers, Make special provisions for taxpayers with disabilities, Provision of services in unofficial language(s), Use of information on compliance burden, End-user testing of new services, End-user involvement in design of new services, Simultaneous Registration for multiple tax types, Formal document covering taxpayer rights, and Mechanism for managing taxpayer complaints	1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14
Public accountability	12	Measures enhancing tax agency’s accountability to the public: Publish its strategic plan, Publish its annual business/operation plan Make public a formal set of service delivery standards Publish its achievements vis-a-vis the standards Publish its annual reports Use of an external auditor Make key compliance risk public Make reports of outcomes in addressing compliance risk public Publish the results of taxpayer satisfaction surveys Document that formally set out taxpayer rights Have a specific mechanism for managing taxpayer complaints, and Publish periodic estimates of the tax gap	1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12 1/12
Autonomy	14	Institutional form, degree of autonomy in managing expenditure and human resources: Autonomous vs. operating within Ministry, Discretion over designing internal structure, Discretion over operational budget, Discretion over capital budget, Authority to set performance standards, Determination of work requirements, Appointment of new staff, Promotion of existing staff, Decide on qualifications for appointment, Decision whether work is carried out by permanent or contractual staff, Placement of staff in salary band, Termination of employment, Responsible for debt collection and enforcement, and Provision of tax policy advice	1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14 1/14
Large Taxpayer Office (LTO)	1	An office or a program dedicated to large taxpayers is in operation	1
Operational Strength	7 sub-indices(65 individual components)	Compound index determined as average of seven indices

Appendix II. Correlation Between Tax Administration Sub-Indices in 2015 and 2017

Variables	Correlation
Compliance risk management	0.8677
Use of third party data	0.8439
Degree of digitalization	0.7290
Service orientation	0.8559
Public Accountability	0.8825
Autonomy	0.8319
Est. of LTO	Status the same for 93 percent of tax administrations
Tax admin staff out of labor force	0.9561
Active taxpayer out of labor force	0.9267

Variables	Correlation
Compliance risk management	0.8677
Use of third party data	0.8439
Degree of digitalization	0.7290
Service orientation	0.8559
Public Accountability	0.8825
Autonomy	0.8319
Est. of LTO	Status the same for 93 percent of tax administrations
Tax admin staff out of labor force	0.9561
Active taxpayer out of labor force	0.9267

Appendix III. Cross Section Regression Results³²

The results suggest the significance of tax administration in explaining tax performance but this approach has critical shortcomings, requiring an alternative approach. The effect of tax administration measured by the operational strength is positively associated with tax collection, indeed the single most relevant factor in the regressions. Unlike the existing literature, many macro-economic variables turn out to be statistically insignificant. However, this approach significantly limits time variation and reduces the effectiveness of the analysis. While tax performance is often affected by cyclical factors, such factors would not be taken into account, as a period average is used for the analysis. The two-stage regression exploits both within country changes over time and cross-country differences to address the problem.

Appendix Table 1.

Cross Section Regression Outcomes

Note: Estimation of equation by OLS regressions with robust standard errors. Standard Errors in Parentheses. *, **, *** denote statistical significance at the 1, 5, and 10 percent levels, respectively. Bold italics means statistically significant at least 10 percent level.

Appendix Table 1.

Cross Section Regression Outcomes

	(1)	(2)	(3)	(4)
*PerCapitaGDP_ PPP*	0.0518 (0.0673)	0.1151 (0.0938)	0.0270*** (0.0052)	0.0281*** (0.0064)
Sq(PerCapitaGDPPPP)	-0.0019 (0.0042)	-0.0051 (0.0059)
Trade openness (out of GDP)	0.0124 (0.0134)	0.0134 (0.0158)	0.0138 (0.0143)	0.0142 (0.0170)
External debt (out of GDP)	0.0031 (0.0080)	0.0106 (0.0107)	0.0033 (0.0077)	0.0095 (0.0099)
CPI	0.1725 (0.1808)	0.0860 (0.1825)
*Terms of Trade*	-0.0254* (0.0148)	-0.0029 (0.0133)
*Oil export (as a % of GDP)*	0.1269 (0.1371)	-0.1966 (0.1295)	0.0647 (0.1353)	-0.2606*** (0.0978)
Log (Agri, %GDP)	-0.0054 (0.0136)	0.0054 (0.0121)
Control Corruption	0.0097 (0.0087)	0.0144 (0.0118)
Budgeted(t)-Actual Rev(t-1)	-0.6087 (0.4181)	-0.3909 (0.4712)	-0.3904 (0.3799)	-0.3886 (0.4167)
*Operational Strength*	0.1129*** (0.0358)		0.0973*** (0.0327)
Constant	-0.1881 (0.2635)	-0.4832 (0.3588)	-0.1530*** (0.0384)	-0.1063** (0.0512)
Observations	96	122	96	122
R-squared	0.5651	0.4061	0.5373	0.3877

Appendix Table 1.

Cross Section Regression Outcomes

	(1)	(2)	(3)	(4)
*PerCapitaGDP_ PPP*	0.0518 (0.0673)	0.1151 (0.0938)	0.0270*** (0.0052)	0.0281*** (0.0064)
Sq(PerCapitaGDPPPP)	-0.0019 (0.0042)	-0.0051 (0.0059)
Trade openness (out of GDP)	0.0124 (0.0134)	0.0134 (0.0158)	0.0138 (0.0143)	0.0142 (0.0170)
External debt (out of GDP)	0.0031 (0.0080)	0.0106 (0.0107)	0.0033 (0.0077)	0.0095 (0.0099)
CPI	0.1725 (0.1808)	0.0860 (0.1825)
*Terms of Trade*	-0.0254* (0.0148)	-0.0029 (0.0133)
*Oil export (as a % of GDP)*	0.1269 (0.1371)	-0.1966 (0.1295)	0.0647 (0.1353)	-0.2606*** (0.0978)
Log (Agri, %GDP)	-0.0054 (0.0136)	0.0054 (0.0121)
Control Corruption	0.0097 (0.0087)	0.0144 (0.0118)
Budgeted(t)-Actual Rev(t-1)	-0.6087 (0.4181)	-0.3909 (0.4712)	-0.3904 (0.3799)	-0.3886 (0.4167)
*Operational Strength*	0.1129*** (0.0358)		0.0973*** (0.0327)
Constant	-0.1881 (0.2635)	-0.4832 (0.3588)	-0.1530*** (0.0384)	-0.1063** (0.0512)
Observations	96	122	96	122
R-squared	0.5651	0.4061	0.5373	0.3877

Appendix IV. The Impact of Staffing and Taxpayers on Tax Collection

Unlike most ISORA data based on binary questions, there are some quantitative questions. We tested how two quantitative variables measuring the size of tax administration in terms of staffing and taxpayer base affect tax performances.³³

Staff in the tax agency and tax performance

Staff level is positively associated with tax collections, but with diminishing returns. The staff level variables are statistically significant, while the coefficient of the quadratic term is negative and significant. This suggests the decreasing marginal tax revenue as the staff level incrementally increases. This declining trend of efficiency leads to a threshold, above which additional staffing does not increase revenues. Based on the second regression of ^{Table 3}, Appendix Table 2 illustrates a marginal increase in tax collection with an increased in the staff level by 0.01 percent of labor force. When the initial level of staff is low, the increase would bring larger tax collection on average, while the marginal increase in tax collection would turn negative when the initial level of staff is high (at 0.25 percent of labor force).

^{Appendix Table 2.}

Tax Gains From Increasing Staff (by 0.01 percent of labor force)

EMs and LIDCs do not reach this threshold but a wide variation in tax performance at staff level implies room for efficiency gains. Appendix Figure 1 shows the staff level in EMs and LIDCs is below the threshold while some AEs pass that point. Especially, in LICs, tax administrations appear understaffed, suggesting potential for ramping up revenue by hiring staff, desirably combined with other productivity enhancing reform measures. On the other hand, some countries suffer from low tax collection despite relatively high level of staffing, which means that they have room for efficiency gains.

Taxpayer base and tax performance

The share of active taxpayer out of labor force is positively and significantly associated with tax collection.³⁴ The size of active taxpayers is measured by the sum of active taxpayers in Personal Income Tax (PIT), Corporate Income Tax (CIT), tax withheld by employers from employees (e.g., PAYE), and VAT. There is a wide variation in the scale of taxpayer base amongst advanced, emerging and low-income economies. In general, advanced economies have larger taxpayer base than those of emerging economies, followed by low income economies.

The share is significantly low in many LIDCs, suggesting a scope for expanding their taxpayer bases. Almost all LIDCs show far lower share of active taxpayers than others. The low rate is not only accredited to the administrative efforts to ensure compliance to tax laws, but also reflects the size of the informal economy, the underlying tax policy and provisions and the tax paying culture of countries, etc.³⁵ To enhance tax collection, it may be worth exploring the factors for the low taxpayer base and, where possible, expanding the taxpayer base.

Appendix V. Second Stage Regression Outcomes Based on Different First Stage Regressions – Operational Strength Index

(1) All Countries

* Estimation of equation by panel between effect with least Square Estimation. Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 (2) Emerging and Low-income Countries

(1) All Countries

	(1) Tax Rate	(2) Bdgt-Actal	(3) Bdgt-Bdgt	(4) Tax Rate	(5) Bdgt-Actal	(6) Bdgt-Bdgt	(7) Tax Rate	(8) Bdgt-Actal	(9) Bdgt-Bdgt
*Operational Strength*	0.1902*** (0.0404)	0.2226*** (0.0349)	0.2317*** (0.0373)	0.1196*** (0.0401)	0.1767*** (0.0358)	0.1864*** (0.0372)	0.0901* (0.0476)	0.1228*** (0.0417)	0.1297*** (0.0420)
*#Tax. Staff/LaborForce*				0.1059*** (0.0242)	0.0802*** (0.0235)	0.0875*** (0.0244)	0.0739** (0.0278)	0.0535** (0.0250)	0.0616** (0.0254)
*Sq(#Tax. Staff/LaborForce)*				-0.0239*** (0.0083)	-0.0157* (0.0083)	-0.0170* (0.0087)	-0.0179* (0.0090)	*-0.0130* (0.0087)	-0.0152* (0.0089)
*#ActiveTaxpayer/LaborForce*							0.0329*** (0.0116)	0.0443*** (0.0122)	0.0472*** (0.0128)
Constant	-0.1159*** (0.0276)	-0.1187*** (0.0222)	-0.1239*** (0.0237)	-0.1402*** (0.0285)	-0.1444*** (0.0235)	-0.1537*** (0.0246)	-0.1212*** (0.0341)	-0.1122*** (0.0273)	-0.1209*** (0.0278)
R-squared	0.2049	0.2911	0.2841	0.4721	0.4937	0.5089	0.5432	0.5753	0.5934
Number of countries	88	101	99	74	85	84	62	70	70

* Estimation of equation by panel between effect with least Square Estimation. Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 (2) Emerging and Low-income Countries

(1) All Countries

	(1) Tax Rate	(2) Bdgt-Actal	(3) Bdgt-Bdgt	(4) Tax Rate	(5) Bdgt-Actal	(6) Bdgt-Bdgt	(7) Tax Rate	(8) Bdgt-Actal	(9) Bdgt-Bdgt
*Operational Strength*	0.1902*** (0.0404)	0.2226*** (0.0349)	0.2317*** (0.0373)	0.1196*** (0.0401)	0.1767*** (0.0358)	0.1864*** (0.0372)	0.0901* (0.0476)	0.1228*** (0.0417)	0.1297*** (0.0420)
*#Tax. Staff/LaborForce*				0.1059*** (0.0242)	0.0802*** (0.0235)	0.0875*** (0.0244)	0.0739** (0.0278)	0.0535** (0.0250)	0.0616** (0.0254)
*Sq(#Tax. Staff/LaborForce)*				-0.0239*** (0.0083)	-0.0157* (0.0083)	-0.0170* (0.0087)	-0.0179* (0.0090)	*-0.0130* (0.0087)	-0.0152* (0.0089)
*#ActiveTaxpayer/LaborForce*							0.0329*** (0.0116)	0.0443*** (0.0122)	0.0472*** (0.0128)
Constant	-0.1159*** (0.0276)	-0.1187*** (0.0222)	-0.1239*** (0.0237)	-0.1402*** (0.0285)	-0.1444*** (0.0235)	-0.1537*** (0.0246)	-0.1212*** (0.0341)	-0.1122*** (0.0273)	-0.1209*** (0.0278)
R-squared	0.2049	0.2911	0.2841	0.4721	0.4937	0.5089	0.5432	0.5753	0.5934
Number of countries	88	101	99	74	85	84	62	70	70

* Estimation of equation by panel between effect with least Square Estimation. Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 (2) Emerging and Low-income Countries

(2) Emerging and Low-income Countries