determinants tax buoyancy in developing nations

92
UNIVERSITY OF ZIMBABWE FACULTY OF SOCIAL STUDIES DEPARTMENT OF ECONOMICS AN EMPIRICAL ANALYSIS OF THE DETERMINANTS OF TAX BOUYANCY IN DEVELOPING NATIONS: AN EXPERIENCE FROM THE SADC ECONOMIES USING PANEL DATA ANALYSIS BY BONGA WELLINGTON GARIKAI A DISSERTATION SUBMITTED TO THE DEPARTMENT OF ECONOMICS, UNIVERSITY OF ZIMBABWE, IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE MASTER OF SCIENCE DEGREE IN ECONOMICS. UNIVERSITY OF ZIMBABWE May 2009

Upload: jr-consultancy

Post on 16-Apr-2015

243 views

Category:

Documents


4 download

DESCRIPTION

AN EMPIRICAL ANALYSIS OF THE DETERMINANTS OF TAX BOUYANCY IN DEVELOPING NATIONS: AN EXPERIENCE FROM THE SADC ECONOMIES USING PANEL DATA ANALYSIS

TRANSCRIPT

Page 1: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

UNIVERSITY OF ZIMBABWE

FACULTY OF SOCIAL STUDIES

DEPARTMENT OF ECONOMICS

AN EMPIRICAL ANALYSIS OF THE DETERMINANTS OF TAX

BOUYANCY IN DEVELOPING NATIONS: AN EXPERIENCE FROM

THE SADC ECONOMIES USING PANEL DATA ANALYSIS

BY

BONGA WELLINGTON GARIKAI

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF ECONOMICS, UNIVERSITY OF ZIMBABWE, IN PARTIAL

FULFILMENT OF THE REQUIREMENTS FOR THE MASTER OF SCIENCE DEGREE IN ECONOMICS.

UNIVERSITY OF ZIMBABWE

May 2009

Page 2: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

i

DEDICATION To my beloved mother (Beauty), late father (Kassian), brothers (Kassian, Moses and Robert) and sisters (Virginia and Ellen).

“.......If it moves forward, tax it. If it stops moving, subsidise it. And if it

moves backwards, regulate it..............,” Reagan, President of USA

Page 3: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

ii

ACKNOWLEDGEMENTS First and foremost, I would like to express my sincere gratitude to my dear supervisors Dr

R.T Chifamba, Mr. R Makoto and Ms N Maphosa for their constructive suggestions and

guidance throughout the study. Without their supervision, I would have not gone this far.

Dr Nelson Wawire should also be credited for introducing the subject to me.

Special thanks should also be given to my family members, especially my brother

Kassian and mother, for their financial, spiritual and moral support throughout my

educational successes. Special recognition should also be given to my brother Mr DS

Gwande and family for their endless support and hospitality they gave to me during my

period of study.

I also wish to thank everyone in my MSc 2008/09 class for their encouragement and

academic help they provided during the course of the study. I will single out Chiguvu

Tadious, Musoro Audrey, Rugube Bridget and Mugayi Providence for their closest

friendship and continuous concern to the success of my research. To them I say, “Keep

up this spirit of industry.”

The African Economic Research Consortium (AERC) should also receive my great

thanks for their academic and financial support during my study.

With due respect, I also acknowledge, the guidance of the Almighty, God, for taking me

through this challenging time of my life. I owe everything to Him.

All views, errors, and omissions are my own and should not be attributed to any person or

organisation mentioned above.

Bonga Wellington Garikai

Page 4: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

iii

ABSTRACT

Southern African Development Community (SADC) countries, like many other

developing countries face difficulty in raising tax revenue for public purposes. This study

uses panel data for fourteen SADC countries during 1994-2005 to analyse empirically the

determinants of tax buoyancy in developing nations. Among the variables identified as

affecting annual tax buoyancy is monetization, with empirical results confirming its

importance. The results have shown that the way monetization is handled in the

developing nations affects annual tax buoyancy negatively. Other variables that have

been found to be affecting tax buoyancy in developing nations include the growth in the

agricultural and industrial sector contribution to national income, external aid growth,

growth of fiscal deficit and growth of total expenditure. The determinants of tax

buoyancy have been suggested following the tax handle theory advice. The study yielded

such results because quality dimension of tax performance have been considered, which

has been neglected by many previous authors1.

1 Previos studies include Tanzi(1987), Teera(2002), Chelliah, Baas and Kelly (1975), Harley 1965, Raja (1971), Lotz and Morss (1967) and Raja et al. (1975)

Page 5: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

iv

CONTENTS DEDICATION.................................................................................................................... i ACKNOWLEDGEMENTS ............................................................................................. ii ABSTRACT ...................................................................................................................... iii CONTENTS...................................................................................................................... iv LIST OF TABLES AND FIGURES ............................................................................... vi LIST OF TABLES ........................................................................................................... vi LIST OF FIGURES ......................................................................................................... vi ACRONYMS ................................................................................................................... vii CHAPTER ONE ............................................................................................................... 1 INTRODUCTION............................................................................................................. 1 1.1 RESEARCH PROBLEM .............................................................................................. 3 1.2 OBJECTIVES OF THE STUDY .................................................................................. 4 1.3 RESEARCH QUESTIONS .......................................................................................... 4 1.4 HYPOTHESIS .............................................................................................................. 5 1.5 SIGNIFICANCE OF THE STUDY.............................................................................. 5 1.6 SCOPE OF THE STUDY ............................................................................................. 7 1.7 OUTLINE OF THE STUDY ........................................................................................ 7 CHAPTER TWO .............................................................................................................. 8 BACKGROUND OF THE STUDY ................................................................................. 8 2.0 INTRODUCTION ........................................................................................................ 8 2.1 TAXATION AND THE GLOBAL WORLD ............................................................... 8 2.2 TAX DESIGN IN DEVELOPING NATIONS........................................................... 11 2.3 REVIEW OF TAX STRUCTURE IN DEVELOPING COUNTRIES ...................... 12 2.4 EVALUATION TAX SYSTEM IN SADC REGION................................................ 15 2.5 SEMI-AUTONOMOUS REVENUE AUTHORITIES AND TAX SYSTEM IN THE SADC REGION ................................................................................................................ 17 2.6 COMPOSITION OF TAXES IN THE SADC REGION ........................................... 19 2.7 ANALYSIS OF DETERMINANTS OF TAX BUOYANCY IN SADC ................... 21 2.8 CONCLUSION ........................................................................................................... 27 CHAPTER THREE ........................................................................................................ 28 LITERATURE REVIEW .............................................................................................. 28 3.0 INTRODUCTION ...................................................................................................... 28 3.1 THEORETICAL LITERATURE REVIEW ............................................................... 28 3.2 LITERATURE ON BUOYANCY AND ELASTICITY ESTIMATION .................. 30 3.3 EMPIRICAL LITERATURE REVIEW ..................................................................... 33 3.4 CONCLUSION ........................................................................................................... 38 CHAPTER FOUR ........................................................................................................... 40 METHODOLOGY ......................................................................................................... 40 4.0 INTRODUCTION ...................................................................................................... 40 4.1 MODEL SPECIFICATION ........................................................................................ 40 4.2 EMPIRICAL MODEL TO BE ESTIMATED ............................................................ 41 4.3 DISCUSSION AND JUSTIFICATION OF VARIABLES ........................................ 42 4.4 MODEL SPECIFICATION TESTS ........................................................................... 48 4.5 PARAMETER TESTS AND MISSPECIFICATION TESTS ................................... 51 4.6 DATA TYPES AND SOURCES................................................................................ 52

Page 6: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

v

4.7 CONCLUSION ........................................................................................................... 52 CHAPTER FIVE ............................................................................................................ 53 ESTIMATION AND INTERPRETATION OF RESULTS ........................................ 53 5.0 INTRODUCTION ...................................................................................................... 53 5.1 SUMMARY STATISTICS ......................................................................................... 53 5.2 CORRELATION MATRIX ....................................................................................... 54 5.3 TESTING FOR MODEL SPECIFICATION ............................................................. 54 5.4 PARAMETER TESTS ............................................................................................... 57 5.5 MODEL ESTIMATION ............................................................................................. 58 5.6 DISCUSSION OF RESULTS .................................................................................... 59 5.7 CONCLUSION ........................................................................................................... 63 CHAPTER SIX ............................................................................................................... 64 CONCLUSION AND POLICY RECOMMENDATION ........................................... 64 6.0 INTRODUCTION ...................................................................................................... 64 6.1 CONCLUSION ........................................................................................................... 64 6.2 POLICY RECOMMENDATION ............................................................................... 65 6.3 FUTURE AREAS OF RESEARCH ........................................................................... 66 REFERENCES ................................................................................................................ 67 APPENDICES ................................................................................................................. 77 APPENDIX I: STRUCTURE OF SADC CENTRAL GOVERNMENT TAXATION ... 77 APPENDIX II: SUMMARY OF REVIEWED LITERATURE ....................................... 78 APPENDIX III: SPECIFICATION TESTS ..................................................................... 79 APPENDIX IV: STATA OUTPUT FOR DATA ANALYSIS ........................................ 81

Page 7: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

vi

LIST OF TABLES AND FIGURES

LIST OF TABLES Table 1. 1: Fiscal Deficit for SADC region (percentage of GDP)………………………..3 Table 5. 1: Summary Statistics…………………………………………..………………53 Table 5. 2: Correlation Matrix .......................................................................................... 54 Table 5. 3: LM Test For Random Effects ......................................................................... 55 Table 5. 4: Hausman Test ................................................................................................. 56 Table 5. 5: General Pooled Ordinary Least Squares Regression Model ........................... 58 Table 5. 6: Specific Pooled OLS Model (Excluding DEBT And POP) ........................... 59

LIST OF FIGURES Figure 1: Average Annual Tax Buoyancy for SADC region ............................................ 15 Figure 2: Average Tax Buoyancy for the Period (1994-2005) ......................................... 16 Figure 3: Composition of Government Revenue in the SADC region (1994-2005) ........ 20 Figure 4: Tax Composition in the SADC Region ............................................................. 20 Figure 5: Money Supply Growth for SADC Region ........................................................ 21 Figure 6: Average Growth of Money Supply for the period 1994-2005 .......................... 23 Figure 7: Government Revenue and Expenditure trend for SADC region(1994-2005) ... 25

Page 8: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

vii

ACRONYMS

AERC - African Economic Research Consortium

AGR - Growth of Agricultural Sector Contribution

AID - External Aid Growth

ARA - Autonomous Revenue Authority

BUOY - Buoyancy

CLR - Classical Linear Regression

DEBT - External Debt Growth

DEF - Fiscal Deficit Growth

ECON - Level of Economic Development

EXP - Growth of Total Expenditure

GDP - Gross Domestic Product

GNP - Gross National Product

IMF - International Monetary Fund

IND - Growth of Industrial Sector Contribution

JSE - Johannesburg Stock Exchange

LM - Lagrange Multiplier

LRA - Lesotho Revenue Authority

M2 - Money plus Quasi Money

MERP - Millennium Economic Recovery Program

MRA - Malawi Revenue Authority

MSc - Master of Science

OECD - Organisation for Economic Cooperation and Development

OLS - Ordinary Least Squares

POP - Population Density

RED - Recent Economic Developments

RSS - Residual Sum of Squares

SADC - Southern African Development Community

SAP - Structural Adjustment Programmes

SSA - Sub Saharan Africa

Page 9: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

viii

TRA - Tanzania Revenue Authority

VAT - Value Added Tax

VIF - Variance Inflation Factor

WB - World Bank

XM - Trade Openness

ZIMPREST - Zimbabwe Programme for Economic and Social Transformation

ZIMRA - Zimbabwe Revenue Authority

ZRA - Zambia Revenue Authority

ZSE - Zimbabwe Stock Exchange

Page 10: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

1

CHAPTER ONE

INTRODUCTION

The traditional function of the tax system is to bring in sufficient revenue to meet the growing

public sector requirements. Common measures of the ability of the tax system to mobilize

revenues are buoyancy and elasticity (Asher 1989). A desirable property of a tax system is that

income elasticity and buoyancy should be equal or greater than unity. Such property ensures that

revenue growth keeps pace with that of Gross Domestic Product (GDP) without frequent

discretionary changes. More important, it imparts build-in stability to the tax system, hence

ensuring mitigation of cyclical variations in GDP over the course of the business cycle. A major

issue in public economics is the appropriate design of a tax system. Such a system is usually

viewed as balancing the various desirable attributes of taxation in that taxes must be raised

(revenue yield) in a way that treats individuals fairly (equity), that minimizes interference in

economic decisions (efficiency), and that does not impose undue costs on tax payers or tax

administrators (simplicity), (Joweria, 2000:17).

Concentration of the study will be on tax buoyancy, which indicates whether the tax “keeps up”

with growth in the economy. Tax buoyancy measures the total response of tax revenue to

changes in national income (Begum, 2007). Year to year buoyancy measures the volatility of the

tax and the ability of government to meet the demands of their constituents. As an economy

grows, income of taxpayers grows and the demand for public services tends to increase. If tax

revenues grow less quickly than the economy, then the public sector will not be able to meet

increased demand for better social amenities. Low tax buoyancy suggests that governments may

face increased public pressure for better and/or more services but with slower growing revenue

sources.

The Southern African Development Community (SADC), in line with the above indicator

(buoyancy), comprises of low tax performance countries with average regional buoyancy that is

less than unity (Matshediso, 2004) implying that the tax system is not responsive to the income

changes in the region. An effort to improve tax performance has been done over the years mainly

Page 11: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

2

noted by various reforms2 in the taxing system but no significant permanent solutions have been

reaped so far. On the other hand, policies relating the monetization to tax performance have not

yet received attention in the region.

The study attempts to examine the determinants of tax buoyancy, paying particular attention on

the effect of monetization on tax buoyancy. The tax performance analysis aims at finding out

whether there is a possibility of increasing tax revenue in developing nations through the

monetary policy. Taxation is an important instrument for attaining a proper pattern of resource

allocation, income distribution, and economic stability, in order that the benefits of economic

development are evenly distributed. Increased revenues are desired for many other purposes

including expanding socially desired government current expenditures, or even on pragmatic

grounds, as to impress foreign aid donors with evidence of the nation’s effort to develop on the

basis of domestic resources. The tax system can be used for non-tax objectives, for example to

correct market failures such as positive or negative externalities, (Bird and Zolt, 2003:34). Many

countries use a tax expenditure budget to account for the costs of tax provisions that are used to

promote non-tax objectives.

Tax systems should be adequately stable and buoyant3 in order to enable a country to meet its

increasing financial commitments as its gross domestic product (GDP) grows. If the tax revenue

of a country is stable and buoyant, there is a high probability that its public expenditure needs

will be adequately met over time. If GDP is growing more than tax revenues then it could be one

policy indicator that the tax structure needs reform. The study of tax buoyancy is of much

importance because it is both a quality and quantity measure of tax performance. Tax buoyancy

can also be used to summarize revenue growth over time, (Zolt, 2003:8). Finally, it shows the

strength of the tax system in the country when they are subjected to certain environments for

example when a certain sector is declining.

The manner in which different countries raise taxes differs as widely as do the amounts they

raise. The pattern of taxes found in any country depends upon many factors such as its economic

2 The reforms include the shift from Sales tax to Value added tax, changing various tax rates and the adoption of Semi Autonomous Revenue Authorities done in the respective countries. 3 The tax system should be responsive to changes in the tax base, usually Gross Domestic Product.

Page 12: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

3

structure, its history, and the tax structures found in neighboring countries (Bird and Zolt, 2003:

7). According to Zolt (2003:1), developing countries are no different: ideas, interests, and

institutions play a central role in shaping tax policy. Basing on this argument the study will be

focusing on SADC countries since they are close to each other and belong to a community.

Countries no longer have the luxury to design their tax systems in isolation.

1.1 RESEARCH PROBLEM

The deterioration of public services in the past years in developing nations should be of concern

given that revenue from taxation is mainly for that purpose (RED, 2004). The level of revenue

being raised from taxation is very low in these nations as compared to the tax base4 which shows

their tax potential. Rapid expansions in expenditure and declining or low revenue levels have

been the main cause of fiscal imbalances in SADC countries over the years (Ghura, 1998). The

trend of fiscal deficit is shown in Table 1.1 below.

Table 1. 1: Fiscal Deficit for SADC region (percentage of GDP) Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Fiscal Deficit -6.2 -4.5 -4.1 -4.3 -4.8 -5.0 -3.2 -3.8 -4.0 -3.9 -4.2 -4.8

Deficit (with

grants)

-4.7 -3.3 -2.9 -3.2 -3.6 -3.5 -1.9 -2.3 -2.7 -2.3 -2.8 -3.1

Source: African Development Indicators 2006

From Table 1.1 above it can be noticed that although there are variations in the magnitude of

deficit, the deficit is always recorded and adequate attention to reduce it is needed. The budget

deficits are regarded as a fiscal phenomenon which can be reduced by either limiting on

government expenditure or raising tax revenues. Budget deficits can produce negative effects on

economic activity, poor service provision, high interest rates, reduced real capital formation and

production.

Tax revenues appear to be highly volatile relative to GDP, the tax base (Ghura, 2004;

Greenaway, 2005). According to Ghura (2004), the changes include the effects of changes in tax

4 Gross Domestic Product is usually considered the tax base showing the level of national income generated.

Page 13: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

4

rates, deductions and compliance. Developing countries are characterized by high tax rates

(exorbitant tax rates, Matshediso 2004) as compared to developed nations, the obvious effect

being decreasing tax revenues collection due to the increased informal sector activities and hence

the governments are not able to meet public demand of public goods. The existing persistent

budget deficits in developing nations suggest that the tax system is not revenue productive, and

in such situations increasing revenue should be the main objective of tax policy.

On the other hand, money supply in the SADC economies has been growing at high levels but

has been named inflationary. There are high levels of tax erosion, due to high growth of money

supply (RED, 2006). Given the continuous reforms (leading to uncertainty and loss of

credibility) that have been happening in the taxing system of developing nations, it still remains a

wonder why tax performance is still low and even declining in some countries. The monetary

sector has not received attention as far as taxing policies are done in these nations and hence its

emphasis should be brought about, since some studies have proposed its importance.

Existing literature (for example Quazi, 1994 and Begum, 2007) of tax performance indicated that

monetization enhances improved tax performance. This means that as level of monetization

increases, documentation of financial transactions of the economy increases, and this increase in

documentation facilitates the collection of direct taxes. Monetization has been found to increase

the collection of indirect taxes such as excise duty, sales tax and value added tax

1.2 OBJECTIVES OF THE STUDY

The objective of the study is to establish the main determinants of tax buoyancy in developing

countries with special attention to the effect of monetization on the tax buoyancy.

1.3 RESEARCH QUESTIONS

The General and Specific research questions of the study can be stated consecutively as follows:

• What are the main determinants of tax buoyancy in developing nations?

Page 14: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

5

• How does monetization affect tax buoyancy?

1.4 HYPOTHESIS

The main hypotheses to be tested in this study are that:

• Monetization have a positive relationship with tax buoyancy.

• Growth of the industrial sector and agricultural sector, fiscal deficit, external debt, level

of economic development, total expenditure and trade openness increases tax buoyancy.

• Growth in external aid and the concentration of population reduces tax buoyancy.

These hypotheses are tested by determining the significance of the regression coefficients of

relevant regression equation that will be estimated.

1.5 SIGNIFICANCE OF THE STUDY

Countries no longer have the luxury to design their tax systems in isolation due to current wave

of globalization and regionalization (Bird and Zolt, 2003). With dramatic reduction in trade

barriers over the last two decades, taxes have become a more important factor in location

decisions. There is increased tax competition for portfolio investment, qualified labor, financial

services, business headquarters and foreign direct investment. This means that taxes do matter,

and any country with a tax system that differs substantially from other countries, particularly its

neighboring countries, may suffer. From this idea, analysis of determinants of tax buoyancy in

the SADC region can be undertaken, since the countries trade with each other, share labor

services and share national borders.

The previous studies (Harley(1965), Lotz and Morss(1967), Raja(1971), Raja et al.(1975) and

Roy(1979)) of tax performance dwelt much on quantitative measures of tax performance such as

the tax ratio. There is a need to incorporate both a quality and a quantity measure of tax

performance, in this case tax buoyancy. There are few studies (for example Teera, 2002 and,

Bird and Zolt, 2003) carried out in this area especially for African countries. It is a new area

which needs further investigation around the regions of the world. Also the study involves the

Page 15: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

6

determination of yearly buoyancy, of which several studies (Quazi (1994), Begum (2007) and

Teera (2002)) have been involved in the use of single averages over a period. The main base of

this study’s approach is that tax buoyancy changes over time even annually because of many

factors (discretionary changes) which may include the political environment among others.

The effect of monetization on tax buoyancy is a crucial issue to consider when making tax

performance decisions. This is because policy makers have to critically administer the optimal

level of money supply in the economy that will not have adverse effects on economic agents. If

money supply grows faster than the growth of the economy, inflation arises and the problem of

tax erosion occurs since there is a gap between the times taxes are to be paid and when they are

actually paid. Increased documentation of the economy can also arise as monetization increases

and hence this facilitates the collection of both direct and indirect taxes. From this idea the

impact of monetization on tax buoyancy has to be analysed. The results will be used to give

necessary policy advice on the link between monetization and tax performance. The

incorporation of money supply in taxing decisions of governments is also a contribution, the

variable have been left by many authors without any justification.

The study will contribute to existing literature on tax buoyancy for developing nations, this helps

in the continuous debate of the effects of various determinants. Analyzing the determinants

offers a guide to policy makers on which areas to put more emphasis. According to Teera (2002),

a poor tax performance, in terms of raising revenues can mean either deficiencies in tax structure

policy or an inadequate effort to collect, on the part of government, both of which are influenced

by various factors. Hence, the study concentrates on finding factors that affect tax performance.

Therefore there is need for more empirical input and guidance to carry out rational economic

decisions. To formulate strategies for achieving sustained increase in tax buoyancy relevant

information is necessary. Therefore examining determinants of tax buoyancy is an appropriate

way of finding where policies can rightly respond to those issues and as such we would gain

better understanding about the determinants. Knowledge of the determinants of tax buoyancy in

SADC will help preclude policy makers from (over) emphasizing only few variables to the

neglect of other important ones in promoting tax performance.

Page 16: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

7

1.6 SCOPE OF THE STUDY

A fully comprehensive coverage of all developing nations is practically impossible therefore in

the study, only fourteen countries5 in the SADC region are considered in the analysis of the

determinants of tax buoyancy. The information obtained in this analysis can be transferred across

to other nations for the purpose of policy formulation. A twelve year period is covered by the

study, from 1994 to 2005 based on data availability in its current form.

1.7 OUTLINE OF THE STUDY

In the first chapter an introduction is presented and the background of the study is well presented

in the next chapter. Theoretical and empirical literature review is done in the third chapter. The

focus of the fourth chapter is to outline the econometric methodology that is going to be used in

the study in examining the determinants of tax buoyancy. Definitions and justification of

variables used in the model and estimations to be carried out are also presented in the same

chapter. The fifth chapter presents a discussion and assessment of the estimation procedure and

the interpretation of results found. And finally, the sixth chapter is for conclusion of study and

policy recommendations.

5 SADC countries include Angola, Botswana, Democratic Republic of Congo, Malawi, Zambia, Lesotho, Swaziland, Mauritius, Seychelles, Zimbabwe, Tanzania, South Africa, Mozambique and Namibia.

Page 17: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

8

CHAPTER TWO

BACKGROUND OF THE STUDY

2.0 INTRODUCTION

The background chapter comprises discussion on taxation and the global world, tax design in

developing nations, composition of taxes in the region and a review of the tax system

performance after the introduction of Semi-Autonomous Revenue Authorities (ARAs). ARAs

were introduced particularly to improve revenue raising capacity and to avoid corruption. Trends

of both annual tax buoyancy and monetization in developing nations is presented and analysed in

this chapter. Finally, the chapter will give a brief summary of the selected determinants of tax

buoyancy in the region.

2.1 TAXATION AND THE GLOBAL WORLD

The requirements of every country can not be met by any single tax structure. The best system

should be determined taking into account economic structure, capacity to administer taxes,

public service needs, and other factors. What matters in tax policy is found by understanding

existing taxes around the world and this means the level and structure of taxes, and the way in

which taxing patterns have changed in recent years.

2.1.1 TAX LEVELS IN THE WORLD

The tax ratio6, on average, is higher in developed nations as compared to developing nations.

This is mainly because the developed nations utilise their tax base due to their well developed

infrastructure and administrative capacity. The fact is supported by Bird and Zolt (2003) who

found an average tax ratio of 18.8 percent for a sample to represent the world countries. Average

tax ratios of 40 percent and 10 percent have been found for developed and developing nations,

respectively. However, not all developed nations have high ratios and not all developing nations

6 Tax ratio can be defined as the ratio of tax revenue to GDP.

Page 18: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

9

have low tax ratios. Both opportunity and choice appear to affect tax levels. Countries with

access to rich natural resource revenues tend to have higher tax ratios than otherwise comparable

countries, though such revenues may also be highly volatile, reflecting commodity price

changes. Tax ratios in higher income countries appear to reflect more choice than chance

because of their stable economies.

However, in broad terms tax ratios do vary by income levels. According to Begum (2007) low

income group, middle income group and high income group recorded 17 percent, 22 percent and

27 percent, respectively. The variability is because demand for public services may rise faster

than income particularly in lower-income countries. For instance, urbanization tends to rise with

income, and the demand for public services is generally higher in urban areas. At the same time,

however, it is usually easier to collect taxes in urbanized areas. More generally, the capacity of

countries to collect taxes appears to rise as income levels increase.

2.1.2 TAX STRUCTURE IN THE WORLD

The pattern of taxes found in any country is a function of its economic structure, its history, and

the tax structures found in neighbouring countries. With dramatic reduction in trade barriers over

the last decade, taxes have become a more important factor in location decisions. There is

increased tax competition for portfolio investment, qualified labour, financial services, business

headquarters and foreign direct investment. This means that taxes do matter, and any country

with a tax system that differs substantially from other countries, particularly its neighbouring

countries, may suffer. Choice also plays a part, as different countries may also attach different

importance to such commonly accepted characteristics of a good tax system as fairness,

economic effects and collection costs. Of useful is also to consider briefly average patterns as

one approach to tax policy in any one country.

Consumption taxes and income taxes form the greatest part of all total taxes collected. Bird and

Zolt (2003) found consumption taxes to be 40 percent of total taxes collected and income taxes

to be equally important. Within consumption tax category, value-added taxes (VATs) contribute

more than other taxes and excises being almost equally important. The main reason for this is

Page 19: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

10

that consumption and income taxes are easy to collect and administer than other types of taxes

and this is supported by the tax handle theory. Personal income taxes are a bit more important

than corporate taxes within the income tax category, because of the political nature that

companies have. If company taxes are very high they may migrate to nearby countries or oppose

the political government. Most of the remaining tax revenues come from taxes on imports and

exports (Bird and Zolt, 2003).

Revenue structure appears to depend to some extent upon its location and economic structure. In

small island countries such as Barbados, for instance, international trade taxes may play an

unusually important role. Trade taxes tend on the whole to be more important in the lower-

income group compared to higher-income group. Begum (2007) found that trade taxes contribute

24 percent and 1 percent of total revenues for low income and high income countries

respectively. Trade taxes (mainly customs duties) appear to decline steadily as countries become

more developed. An interesting exception is the transitional countries which have traditionally

relied little on trade taxes (Martinez-Vazquez and McNab, 2000). In general, however, trade

taxes clearly decline in importance as income rises.

Low-income countries tend to raise more revenues at the border, where relatively few collection

points need to be controlled and hence they are more likely to rely more heavily on excise taxes

on tobacco, alcohol and so on. In contrast, direct taxes (and VAT) tend to require both a more

effective tax administration and taxpayers who are more sophisticated, conditions more likely to

exist in developed countries.

2.1.3 RECENT TRENDS OF TAXES IN THE WORLD

The relative importance of different taxes has changed in recent years. The most striking feature

is the increase in the share of revenues generated by consumption taxes. One reason being the

adoption of broad-based VATs as an improvement from commonly used sales tax (40 percent of

all consumption taxes from 34 percent in early 90s; Bird and Zolt, 2003). The use of VAT is

more efficient than sales tax given the growing underground economy and this allows the

Page 20: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

11

taxation of informal trade and production. Approximately 70 percent of the world’s population

lives in the 123 or more countries that now levy a VAT (Ebrill et al. 2001).

There has also been some increase in the share of revenues raised from direct taxes, especially

personal income taxes. Taxes on international trade have dropped dramatically, decreasing by

4.3 percent of total collections (a decline approximately offset by the 4.1 percent rise in

consumption taxes), (Bird and Zolt, 2003:9). The use of trade taxes has dropped even more over

the longer term. In 1981, for example, trade taxes accounted for 30.6 percent of developing

country revenues (Tanzi, 1987), compared to only 24.3 percent for the same countries in 1998.

On average tax revenues have grown more quickly than GDP, the overall average buoyancy was

1.04 (Bird and Zolt, 2003). However, buoyancies tend to be lower in Africa and especially Asia,

than elsewhere, Begum(2007). This is because developing countries are not able to utilize their

tax bases and account informal trading activities. The low levels of buoyancy in developing

nations shows poor tax performance and is the main worry of this study and hence have to be

examined.

2.2 TAX DESIGN IN DEVELOPING NATIONS

The best tax policy in the world is worth little if it cannot be implemented effectively. Tax policy

design must take into account the administrative dimension of taxation. What can be done may to

a considerable extent determine what is done in any country.

The design of a tax system is strongly influenced by economic structure. Many developing

countries have a large traditional agricultural sector that is not easily taxed. Many transitional

and developing countries have a significant informal (shadow) economy that also is largely

outside the formal tax structure. The potentially reachable tax base thus constitutes a smaller

portion of total economic activity than in developed countries. The size of the untaxed economy

is in part a function of tax policy. The resulting lower tax revenues often lead governments to

raise tax rates, further exacerbating incentives to evade taxes. Improving tax administration is

Page 21: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

12

thus central to the choice of tax structures and to improving taxation in developing and

transitional countries.

The medium to long run tax design may take into consideration the existing administrative

constraints but in the short run this may compromise both efficiency and effectiveness of the tax

system. As such strengthening tax administration is critical for medium to long term successful

reforms. The process of tax administration reform involves an overhaul of tax laws,

identification of tax payers, assessment of the tax, control and verification of assessments,

collection of the taxes, and enforcement in the event of non payment. At each of these stages

weaknesses in tax administration are likely to compromise efficiency and effectiveness. It is

necessary therefore that tax administration reform be aimed at improving the efficiency with

which taxes are assessed, processed and collected.

Experiences from Jamaica, Indonesia, Ghana and SADC as cited by the World Bank in 2001

have shown that improving tax revenue mobilization requires administrative reforms. Thus

computerization of tax systems, training of key personnel, modernisation of auditing procedures

and development of new management systems are essential in tax administrative reforms. Tax

administration reforms carried out in SADC yielded substantial gains in tax yield. However, the

implementation of these reforms have been undermined by high rates of corruption, low morale

of staff due to low salaries and benefits, inadequate link between rewards to amount of tax

collected, poor working environment, inability to fire corrupt and incompetent staff, and inability

to retain key professional staff such as accountants. The combination of these factors

necessitated the World Bank, IMF and other development partners to consider complete overhaul

of tax administration before any more tax policy and design reforms were pursued especially in

poor countries facing structural adjustment problems. The result has been the introduction of

semi-autonomous and autonomous revenue collection organisations.

2.3 REVIEW OF TAX STRUCTURE IN DEVELOPING COUNTRIES

Tax statutes in developing countries resemble those of industrial countries. If one compares the

actual tax structure of developing countries with ideal norms of a desirable tax structure, it would

Page 22: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

13

be difficult to find greater contrast. The prototypical tax structure of developing countries can be

criticized on stabilization, efficiency, and equity grounds.

Stabilization role of a tax system requires one or at most several predominant taxes with a rate

schedule that can be adjusted quickly and with a high degree of certainty to alter the purchasing

power available to the private sector. These predominant taxes can then be used to increase or

cut back private spending to achieve stabilization goals related to growth, prices, or the balance

of payments. The tax systems of developing countries are cluttered and tax bases are narrow,

adjustment of revenue for stabilization purposes must come about by piecemeal measures. From

a dynamic point of view, the continued reliance on these tax bases for revenue needs tends to be

self-defeating (higher rates on the narrow bases further distort the allocation of resources and

lead to evasion or to consumer resistance).

Efficiency role of a tax system requires least amount of distortion of relative prices, both in the

prices facing the consumer and the producer. A lump-sum poll tax would satisfy the criterion of

noninterference in all relative prices, but broadly based income or sales taxes with one or a few

rates are considered desirable in that they lead to less distortion in prices. Taxation of products

with low elasticity of demand would bring about less interference with consumer choices

(Hatwick theory). In developing countries many different import and excise tax rates on narrow

bases are combined with a variety of exemptions to produce almost random effects on consumer

choice and producer incentives. Efficiency goals are also compromised by selective

administration of the personal and corporate income taxes. Because the personal income tax is in

practice mainly a tax on wages, companies using more labour-intensive productive methods are

at a disadvantage. Large companies are forced to pay tax, whereas small companies and retailers

can avoid taxes.

The heavy reliance on foreign trade taxes can be singled out as the most undesirable aspect of tax

systems in developing countries from the point of view of both stabilization and resource

allocation (Greenaway, 1981). From the standpoint of stabilization, foreign trade taxes tend to tie

government revenue to unpredictable fluctuations of export commodity prices, and so to

aggravate fiscal stabilization problems. A vicious circle can also begin if a reduction in imports

Page 23: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

14

imposed for demand management purposes leads to a fall in import duties, thereby enlarging the

fiscal deficit. From the standpoint of resource allocation, trade taxes have no place in a "first-

best" tax structure. The discriminatory nature of trade taxes ensures that their use imposes both a

production distortion cost and a consumption distortion cost. Import duties, often relied upon for

revenue, lead to excessive effective protection and may encourage inefficient domestic industry.

Equity goals are also heavily compromised by the tax structure of developing countries. The tax

structure appear to be highly progressive because the rate structure of the personal income tax is

often highly progressive. On closer examination, however, the vertical equity of the system is

compromised by the fact that the personal income tax collects a relatively small share of total

revenue and applies mainly to wage earners, whose incomes may be lower than those of the self-

employed. High nominal rates on luxury consumer imports such as motor vehicles may equally

be offset by exemptions and evasion. Excises on such staples as cigarettes and beer are clearly

regressive when taken in the context of the monetized sector. Thus, an apparent vertical equity in

rate structure of personal income tax and import duties is negated by a combination of selective

administration and legal exemptions. With regard to horizontal equity, the tax structure is

perhaps more clearly inequitable. For the income tax, selective administration on wages favors

the self-employed, partnerships, and income from capital. Taxes on consumption also violate

horizontal equity because of their uneven and almost random application.

Selective administration of the statutory tax system in developing countries systematically

discriminates against the modern sector of the economy. The modern sector tends to fall into the

income tax net because it keeps better accounts, is more centralized, and thus is more revenue-

productive in the eyes of the tax collector. Although a heavier burden of taxation on the modern

sector may be appealing on equity grounds, one should bear in mind that the present tax system

tends to impede the growth of the dynamic sector that may be the key to more rapid long-term

growth in developing economies.

In view of the above criticisms, the tax administrator is left with a dilemma. Strict adherence to a

cost-benefit, revenue maximizing strategy would worsen efficiency distortions in the present tax

structure and would eventually undermine voluntary compliance. But political pressure for

Page 24: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

15

immediate revenue to reduce large fiscal deficits may be strong. In such circumstances, the tax

administrator should strive to balance immediate revenue objectives with the need to support

changes designed to produce a "better" tax structure in terms of the efficiency, equity, and

stabilization goals mentioned above. These latter efforts would seek to reduce dependence on

convenient tax handles and to tackle the more difficult problem of implementing more broadly

based income and consumption taxes.

2.4 EVALUATION OF THE TAX SYSTEM IN THE SADC REGION

The evaluation of the tax system of the SADC community is in terms of the analysis of the trend

of tax buoyancy over the study period. A good tax system, as noted earlier in the previous

chapter should be buoyant and stable. This means that the revenue from taxation should be well

responsive to the fluctuations in national income. The trend is well presented in Figure 1 below.

Figure 1: Average Annual Tax Buoyancy for SADC region

Average Annual Tax Buoyancy for SADC region

0

0.5

1

1.5

2

2.5

3

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

year

Ta

x B

uo

ya

nc

y

bouyancy

Source: World Development Indicators. Various issues

Figure 1 above shows clearly that buoyancy levels declined over the period from 1994 up to

2000 and increased slightly in 2001. However, it then decline until 2003 and increased a little

sharply in years 2004 and 2005, but it never reached its previous peak value of 1994. The general

Page 25: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

16

conclusion is that the tax system performance declined over the period, showing that the tax

revenue being registered in developing nations is not responsive to national income. The

authorities are failing to collect revenue matching their potential characterized by their tax base.

Poor tax administration and underdeveloped infrastructure are possible reasons. Corruption

among the tax authorities and evasion by tax payers contribute to such performance.

Hindriks(1999) supported that, because tax collection invites several forms of dishonesty and

malpractice: tax payers evade their tax liabilities, while tax inspectors solicit bribes in order to

connive at such evasions and hence they abuse the authority with which they are entrusted.

Average tax buoyancy for each country over the period can be analysed using Figure 2 below.

This helps to assess the level at which the countries differ during the period under study.

Figure 2: Average Tax Buoyancy for the Period (1994-2005)

AVERAGE TAX BUOYANCY FOR THE PERIOD(1994-2005)

ANGOLA B

OTS

DRC

LESOTH

O

MALA

WI M

AUR M

OZA

NAM

IBIA

SA

SEYC

SWAZ

TANZA

ZAM

BIA

ZIM

0

0.5

1

1.5

2

MEMBER COUNTRY

AV

ER

AG

E B

UO

YA

NC

Y

Source: World Development Indicators. Various issues

Figure 2 shows that many countries have tax buoyancy that is less than unity, implying that there

is poor tax performance as the tax system can be revealed that it is not responsive to changes in

respective national incomes. Botswana performed very well with an average buoyancy of 1.5 for

the whole period, which means that its tax system is very responsive to its tax base changes.

However, Botswana performed well above others because it managed to collect more revenue

only in the initial years of VAT adoption and also the introduction of its revenue authority during

the period of study. However, Namibia, Malawi, Zambia, South Africa and Lesotho have lowest

values of tax buoyancy around 0.35, while other nations averagely performed near the value

Page 26: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

17

equal to one. Although other countries are better as compared to their counterparts, they still

have to improve their tax systems because the magnitudes of buoyancy still indicate a poor

performance. The main reason for these differences comes from major sectors each country has

established and hence over taxing the sector.7

2.5 SEMI-AUTONOMOUS REVENUE AUTHORITIES AND TAX SYSTEM

IN THE SADC REGION

Historically, ministries of finance have existed to collect and manage government revenues.

They rarely have a complete monopoly over collection (Fjeldstad and Moore, 2008:2). The

creation of Semi Autonomous Revenue Authorities (ARAs) has improved relationships between

tax authorities and larger corporate taxpayers, and increased, at least marginally, the capacity of

governments to raise revenue. In the last two decades, there appears to have been a trend among

developing countries towards the creation of ARAs to replace their existing tax collection

agencies (Manasan, 2003:1). In many of these countries, the radical reform of the tax agency was

primarily intended to improve revenue performance in the face of deep-seated problems in tax

administration. Tax experts (e.g., Silvani and Baer 1997 and Jenkins 1994) have suggested the

imperative for radical changes in tax administration in countries where the tax gap is large (i.e.

40% or more). In the eyes of a few academics and external observers, the introduction of revenue

agencies has been seen as a step on the road to privatisation of the revenue collection process

(Kiser and Sacks 2007; Devas et al. 2001; Byrne 1995). Despite the rhetoric and debate about

‘autonomy’, there has been very little loosening of the political and bureaucratic grip of central

executive authorities over the revenue collectors. Presidents and Ministers of Finance are still

very much in control. President Museveni in 2000, described the Uganda Revenue Authority as a

‘den of thieves’ (Therkildsen, 2004:82).

In Zimbabwe, for instance the Zimbabwe Revenue Authority (ZIMRA) came into operation on 1

September 2001 to take over dealings in taxation transactions from the Ministry of Finance. Its

mission was to facilitate economic development, trade and travel, revenue generation and

7 For example Malawi is for fishing sector, Zimbawe agriculture, DRC mining, Zambia sugar cane, South Africa manufacturing.

Page 27: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

18

collection, to enforce compliance with revenue laws and enforce regulatory controls with

integrity, transparency and fairness (Saruchera, 2009:3). However the success of the ZIMRA in

improving tax performance did not come about due to several reasons. The organization was set

in a period of economic crisis therefore it did not receive full attention from relevant authorities

and also the underground economy was heavily growing hence hindering tax collection. There

were manpower shortages due to brain drain as revenue officers were moving out of the country

in search of greener pastures and this affected revenue collection as training of new staff is

expensive. Increased bribery and corruption was another problem such that due to economic

hardships tax payers were paying the officers privately so as not to face full tax charges

(Chiminya, 2008). Smuggling of goods was a contributing factor as officials connive with traders

(revenue leakage). Also there was tax erosion due to inflation, this was due to the gap between

tax payments dates scheduled and actual days it was paid, and there were delays in debt follow

up.

The South Africa Revenue Authority (SARA) was launched in 1997, leading to tax effort to rise

from 24% in 1996 to 26% in 1999 (with reduction in tax rates), (Manasan, 2003:8). Lesotho

Revenue Authority (LRA) was launched in January 2003 to deal in problems of tax

administration. Generally, the annual turnover of staff is low. In the Lesotho Revenue Authority,

for instance, turnover is around 1–2 % per year (FIAS/DFID 2006b).

Tanzania Revenue Authority (TRA) came into existence in 1996, with the main aim being to

improve revenue performance in the face of deep seated problems in tax administration. Tax

effort rose from 11% in 1995 to 12% in 1996 but declined since then reaching 10% in 1998

(Manasan, 2003:8). There is some evidence that the gains in revenue performance tend to be

eroded after some time. Following its launch not all previous tax collection staff were transferred

directly into ARAs. Initially disciplinary procedures tightened up within TRA, and more staff have been

penalized for corruption. However, the use of dismissals in the initial phases has not been sustained. In

Tanzania annual dismissals have in later years dropped to less than 2% of the staff total (Manasan, 2003).

All former staff members were dismissed and had to reapply for a position in the new revenue

authority. Almost 1200 staff members, equivalent to more than a third of the total former work

force, were not re-employed on evidence or suspicion of misconduct (Fjeldstad, 2003).

Page 28: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

19

Zambia Revenue Authority (ZRA) is one of the oldest in Africa after Ghana (1985) and was

launched in 1993. Tax effort dipped from 19% in 1992 to 15% in 1993 before increasing to 19%

in 1994 before settling at 17%-18% in 1995-1997 (Manasan, 2003:8). Thus, it seems that the

process of changing over to a new system may involve significant costs in terms of reduced

revenues in addition to the cost of retiring personnel of the old tax administration.

Similar trend have been noticed for all countries in the region and this refers to a commonly

noticed trend that revenue only increased in the years of the launch and thereafter it starts to

decline. The main reason for such revenue pattern is that the revenue authorities did not become

independent from the political authorities and hence corruption remains and also evasion by

private sector continued as there is mistrust of the revenue authorities. The heads of revenue

authorities were appointed by government officials and in some cases (Malawi) they are also

government officials. The conclusion to the establishment of ARAs in the region is that they did

not bring much change to the trend of revenue growth.

2.6 COMPOSITION OF TAXES IN THE SADC REGION

Government revenue that is always collected does not cover the expenses by the government and

hence SADC countries have persistent fiscal deficits. Although the national income is low in

these countries, revenue collected from taxes is below the potential level. However, the

government obtains revenue from various sources. Government revenue consists of tax revenue

and non tax revenue and taxes constitute a bulk of domestic revenue. Trend of both tax and non

tax revenue for the SADC region is shown in Figure 3 overleaf. Figure 3 shows that government

revenue constitutes two main sources that are not stable over the period; tax revenue is always

fluctuating and the main cause being continuous changes in tax policy over the period. This is

supported by the adoption of value added tax from sales tax by many countries during the period,

the development of autonomous revenue authorities as highlighted in section 2.5 and also the

changes in various tax rates.

Page 29: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

20

Figure 3: Composition of Government Revenue in the SADC region (1994-2005)

Composition of Government Revenue

0

5

10

15

20

25

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

%G

DP

Govt Revenue

Non TaxRevenueTax Renenue

Source: African Development Indicators 2006, World Bank

The main source of revenue include taxes on income and profits for both individuals and

corporates; taxes on goods and services in the form of value added tax, customs duty and excise

duty; revenue from Investments and properties, that is, interest and dividends; international aid

and grants, and miscellaneous taxes in the form of stamp duties, fees, estate duties and business

licenses. This is in detail in Figure 4 below:

Figure 4: Tax Composition in the SADC Region

DOMESTIC REVENUE CONTRIBUTION

38%

29%

14%

9%3% 7%

INDIVIDUAL TAXES

VAT

COMPANY TAXES

CUSTOM DUTY

EXCISE DUTY

0THER

Source: Regional Integration in Southern Africa, Vol.6, p.38.

Page 30: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

21

Individual taxes formed the greater part of tax revenue (37.9%), this is mainly because it is very

difficult to evade income taxation and it is well recorded. The other major contributor being

value added tax (29.1%) with excise duty (2.9%) contributing a smaller percentage of revenue

showing that the bulk of tax revenue is from individuals and companies. The probable reason is

that individual and company taxes are easier to collect than other forms of taxes and this is in

line with the tax handle theory to be discussed in literature. A detailed structure of the SADC tax

system is presented in Appendix I.

2.7 ANALYSIS OF DETERMINANTS OF TAX BUOYANCY IN SADC

MONETARY DEVELOPMENTS

The money supply in the region have recorded a significant magnitude during the period under

study. However, it has to be noted that different individual countries have been recording high

values for money supply due to crisis they have been facing for example the Zimbabwean nation.

The trend for money supply growth is presented in Figure 5 below.

Figure 5: Money Supply Growth for SADC Region

Money Supply Growth for SADC region

0102030405060

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

Year

Mo

ne

y S

up

ply

Gro

wth

ms

Source: World Development Indicators. Various issues

From Figure 5 above, initially in the period high rates of growth have been recorded (53.3, 50.8

for 1994 and 1995 respectively) then a sharp decline experienced in 1996 (21.5) and then on

Page 31: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

22

average constant rates have been experienced until 2001 when it starts to rise reaching a peak

(50.2) in 2003 and a decline again up to the end. The growth in money supply although

fluctuating in the period is very significant showing that it should have an impact in the

economy, for example the inflationary effects from the monetarist view.

An average expansion of 34.5 per cent in 2002 and 50.5 per cent in 2003 can be analysed. These

figures include Zimbabwe which registered high money supply growth rates of 164.8 per cent

and 413.5 per cent in 2002 and 2003, respectively on account of domestic credit expansion

caused by increased lending to both the private sector and the government. Zimbabwe registered

high levels of money supply growth in the period because it printed excess money to finance

external debt and its growing deficit. Excluding Zimbabwe, expansion of broad money supply in

the region fell from 23.6 per cent in 2002 to 20.2 per cent in 2003 (RED, 2005).

During 2003, nine SADC Member States namely Angola, Botswana, Lesotho, Mauritius,

Mozambique, Namibia, South Africa, Tanzania and Zambia experienced declines in the growth

rate of their money supply as compared to 2002. Six countries registered money supply

expansion of less than 15 per cent in 2003 and of these countries; Lesotho and Namibia had

growth rates of even less than 7 per cent. The increasing globalization in the region involving

summit arrangements led to similar behaviour of the SADC states concerning money supply

discipline. This is driven by bilateral agreements between countries and monetary unions.

The three other countries apart from Zimbabwe that witnessed an increase in the growth rate of

money supply in 2003 were DRC, Malawi and Swaziland with expansions of 31.7 per cent, 26.4

per cent and 14.1 per cent, respectively. In Malawi, the 26.4 per cent growth in money supply

was the result of increased borrowing by the government from the banking system that was

necessitated by non-timely disbursement of programmed or budgeted donor funds.

Except for Zimbabwe and Angola (which recorded an expansion of 66.3 per cent in 2003) money

supply growth in the remaining SADC Member States appears to be within controllable levels.

However, it is to be noted that although DRC’s growth rate in money supply was high, its

inflation rate was within controllable limits.

Page 32: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

23

Figure 6: Average Growth of Money Supply for the period 1994-2005

Average growth of Money Supply (1994-2005)

ANGOLA

BO

TSWANA

DRC

LESOTH

O MALA

WI

MAURIT

IUSM

OZA

MBIQ

UE

NAM

IBIA

SEYCHELS

SASW

AZIL

AND

TANZA

NIA

ZAM

BIA

ZIM

0

20

40

60

80

100

120

Country

Mo

ney S

up

ply

Gro

wth

Source: Regional Economic Development 2006

From Figure 6 above it can be clearly seen that Angola, DRC and Zimbabwe have the highest

levels of money supply growth in the period. This may be due to the crisis they have been facing

in the period. Botswana registered the lowest growth of money supply and this shows discipline

in its monetary sector and also signify stability. However, these averages can be misleading since

the more recent years have shown that money supply is growing fast, with Zimbabwe leading

significantly to the extent that average regional growth of money supply excluded it.

Monetization of an economy should be handled with care since there are various factors that may

arise related to the level of money supply growth. The growth of money supply should be linked

to the growth of the economy matching the reserve levels. If monetization levels are optimal then

the documentation of the economy is possible and hence this should be seen leading to

facilitation of tax collection of each tax thereby bringing improved tax performance. Utilization

of the growing money supply to improve tax collection is expected for every nation. It is

inconclusive to determine which level of growth brings in improved tax performance unless a

detailed study is carried out. The study seeks to examine the impact of monetization levels to tax

performance in the developing nations. The general idea revealed is that monetization is higher

in developing nations than developed nations and developing nations are low tax performers.

Page 33: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

24

EXTERNAL DEBT

Southern Africa is the least indebted region of the African continent (RED, 2004:13). However,

it is worth noting that there are wide disparities in the level of external indebtedness across

countries. In fact, Malawi, Mozambique and Zambia are all classified as heavily indebted poor

countries. By contrast, the ratio of external debt to GDP of Botswana, Mauritius, Namibia, South

Africa and Swaziland stood below the 30 per cent mark over the period 1999 through 2002.

However, most countries reported a rise in their foreign debt in 2003. Zambia is the only country

that reported a fall in its foreign debt position from US$7.1 billion at the end of 2002 to US$6.4

billion at the end of 2003 (WDI, 2005). Malawi’s external debt position was unchanged with

respect to the 2002 level at US$2.5 billion, which is quite high relative to the size of its economy

(RED, 2004). The same trend continued for both 2004 and 2005 for most of the nations.

With a large debt, the government needs to raise the necessary revenue to service it. When the

interest on the debt exceeds net borrowing plus the possible reduction in non-interest

expenditure, the level of taxation must go up unless the rate of growth of the economy is high

enough to neutralize the increase. Therefore, debt plays a role in determining the extent to which

countries may take advantage of their taxable capacity (Tanzi, 1987). Given that debt growth

levels are high in SADC does the tax system consider its magnitude?

FISCAL DEVELOPMENTS

Fiscal deficits have been arising because government spending is always greater than

government revenue. This shows that there is no fiscal discipline in developing nations, they

don’t balance their budget. One of the problems arising from this is policy inconsistency and also

loss of borrowing power from institution such as the World Bank, International Monetary Fund

and development partners. Tax performance should be affected in some way by fiscal deficit

growth since the deficits have to be covered by revenue from taxation or monetisation. The trend

of government revenue and expenditure over the period is presented in Figure 7 (next page)

showing that expenditure is greater than revenue generated.

Page 34: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

25

Figure 7: Government Revenue and Expenditure trend for SADC region(1994-2005)

Government Revenue and Expenditure

15

20

25

30

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

% G

DP

Govt Revenue

Govt Expenditure

Source: African Development Indicators 2006, World Bank issue

Fiscal deficits seems a tradition to the nations, however some nations like Botswana shows that

they are giving attention to its magnitude. Although all SADC Member States registered deficits

in 2003, the fiscal deficit to GDP ratio for the region which averaged 4.0 per cent in 2002

improved to 3.7 per cent in 2003, a right step towards attaining the SADC macroeconomic

convergence programme target (RED, 2004). Angola, Lesotho, Mozambique, Swaziland and

Zimbabwe recorded improvements in their fiscal deficit as compared to 2002 while the deficits

for Zambia and Mauritius were more or less unchanged at around 4 per cent and 6 per cent,

respectively.

Zimbabwe registered a fiscal deficit to GDP ratio lower than 1 per cent in 2003. The countries

that registered quite high fiscal deficit to GDP ratios of above 4 per cent include Angola (5.1 per

cent), DRC (4.3 per cent), Malawi (5.7 per cent), Mauritius (6.2 per cent), Mozambique (4.9 per

cent), Swaziland (4.6 per cent) and Zambia (4.1 per cent), (RED, 2004).

The increase in the fiscal deficit in Malawi was the result of some unplanned demands on the

budget to cover domestic arrears, an increase in wage bill on account of promotions and the

depreciation of the kwacha. In Mauritius, the deficit was financed essentially from domestic

sources and, more specifically, from commercial banks and the non-bank sector. Net financing of

the deficit by the central bank was negative for the fourth consecutive year. The unchanged

domestic budget deficit to GDP ratio of 4.1 per cent for Zambia was largely due to expenditure

Page 35: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

26

overruns in the recurrent budget. Fiscal policy in South Africa became moderately expansionary

in 2003, registering a deficit ratio of 2.6 per cent.

The International Monetary Fund, World Bank and other international lending agencies impose

different conditionalities to reduce the budget deficit. One of the conditionalities is to reduce the

budget deficit through increase in new taxes and improving existing taxes by removing different

allowances and exemptions. It is therefore expected that the increase in deficit will compel

governments in developing countries to increase collection through new as well as existing taxes.

LEVEL OF ECONOMIC DEVELOPMENT

In spite of the improved growth attained in 2005, economic conditions in the SADC region are

still fragile and many economic indicators are far below the targets set up in the macroeconomic

convergence programme (RED, 2OO6). SADC’s growth performance has been rising during the

period and was more or less unchanged in 2003. On an average basis, GDP growth hovered

around 3.5 per cent in 2003. This stable level of growth was mainly the result of the slowdown in

the growth rate, from 3.6 per cent in 2002 to 1.9 per cent in 2003, of South Africa, the largest

economy in the region, on account of declines in its manufacturing, agricultural and gold mining

outputs. In spite of the overall improvement recorded in SADC, only two countries, Angola

(20.6 percent) and Botswana (8.4 percent) achieved output growth rates above the target, the rest

performed below 7 percent. Zimbabwe’s output declined by 1.2 percent. Zimbabwe was the only

country that recorded a negative growth rate in 2003 to 2005. However, the negative growth of

9.0 per cent in Zimbabwe in 2003 was an improvement over the decline of 11.6 per cent

registered in the preceding year.

A higher per capita income reflecting a higher level of development is held to indicate a higher

capacity to pay taxes as well as a greater capacity to levy and collect them (Chelliah, 1971). Also

as income grows countries generally become more urbanized. Urbanisation per se brings about a

greater demand for public services while at the same time facilitating tax collection (Tanzi,

1987). Given the growth pattern of the SADC region discussed above, the effect to tax

performance have to be assessed.

Page 36: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

27

2.8 CONCLUSION

The chapter has managed to reveal that the buoyancy levels in the SADC region are very low in

the face of a significantly growing money supply. Results from the reforms in tax administration

have been illustrated using some selected countries examples. The trends of some selected

determinants of tax buoyancy have been described. This chapter automatically leads to a review

of the literature related to the study at hand in order to borrow some guidelines for this study.

Page 37: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

28

CHAPTER THREE

LITERATURE REVIEW

3.0 INTRODUCTION

The study seeks to investigate empirically the determinants of tax buoyancy in developing

nations. However, before the empirical estimations of the relationship could be carried out, it is

useful that we review and discuss the theoretical and empirical literature related to tax buoyancy

and its determinants. In light of the above, the current chapter will discuss and review the

theoretical literature as well as the empirical studies on tax buoyancy. Thus, the current chapter

is significant in as much as it discusses the foundation upon which the empirical model that we

shall adopt is based.

3.1 THEORETICAL LITERATURE REVIEW

Currently, there are no direct theories of tax buoyancy and its determinants. Previous theories

only concentrate on the calculation of tax buoyancy. However, we have theories related to tax

performance. These theories help in equipping us with the general idea of what determinants of

tax buoyancy may be. The argument being that tax buoyancy is a measure of tax performance

that incorporates quality aspect rather than quantity alone.

THEORIES OF TAX PERFORMANCE

The normative bent of the literature on tax policy deals with the questions of why a country

develops a particular tax structure and why this tax structure differs among countries and

changes during the process of economic growth. This strand of tax literature not only recognizes

the importance of administrative constraints on tax policy, but in contrast to the normative

literature places administrative factors at the forefront.

Page 38: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

29

The "tax handle" theory offers a sweeping historical explanation of tax structure change. It

argues that low-income economies are forced to collect revenue from easy-to-administer taxes

(or tax handles), but that this administrative constraint lessens as countries develop and become

able to choose "better" taxes as defined by the normative objectives discussed above. Measures

of tax handles typically include per capita income, trade taxes and the proportion of people living

in urban areas (Liebaman, 2003). Although economists differ on the recipe for a "better" tax

structure, there would be general agreement that broadly based income or consumption taxes are

preferable to a reliance on foreign trade taxes, and historical evidence suggests such a shift. In

any case, high-income economies have a certain freedom to manoeuvre in choosing tax systems,

whereas low-income economies are in large part constrained by administrative considerations.

In accordance with the tax handle theory, the effective tax structure of developing countries is

characterized by reliance on narrow tax bases, reflecting administrative convenience.

Considering import duties and export duties, the centralization of taxable goods in a port leads to

ease of administration. For excises, production of goods is concentrated in a few factories. For

the personal income tax, withholding of taxes on wages paid by large enterprises and

government is the meaning of administrative convenience. However, corporate income tax may

be inherently more difficult to administer because the base of net profits requires more

sophisticated accounting measures such as depreciation. In this case administrative convenience

may take the form of simply accepting, with only perfunctory audit that presents no serious

challenge, a "reasonable" amount of taxation as defined by the firm. If firms are unwilling to

contribute a reasonable sum, governments may then adopt simple rules of thumb, such as using a

percentage of gross sales to replace the concept of net profit. Musgrave (1984), argues that tax

structure is shaped by economic factors which account for the size of different tax bases and by

political and social factors which influence opinions on tax equity. Government centralization is

favourable for direct taxes on income and wealth because of its administrative advantages in

assessing these taxes.

The optimal tax theory, the reigning normative approach to taxation combines information on a

country’s economic structure, the set of available taxes to the government and the objectives of

tax policy to make recommendations on tax mix, structure and incidence (see Slemrod, 1990;

Page 39: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

30

Burgess and Stern, 1993). Optimal taxes are those that raise a desired amount of revenue with the

lowest marginal efficient cost, with few distortions and that promote the desired amount of

wealth. While optimal tax theory tackles the trade off of different taxes, it does not explain the

structure of government revenues. The standard optimal tax theory assumes that individuals and

firms voluntarily pay their tax liability. This assumption is inaccurate since individuals pursue

many illegal avenues to reduce their payments such as underreporting incomes, hence there is

evidence in many countries that tax evasion is extensive.

The Ricardian equivalence theory is based on the opinion that when the government borrows

instead of levying taxes to finance budget deficit, the current generation is under taxed. This

generation is rational and will realize that the loan will have to be repaid from income tax at

some time in the future; debt finance is therefore a postponement of the tax burden which will

fall on the future generation. The current generation will not want their heirs to be in a worse off

position on account of the under funded benefit which they are enjoying and will therefore

increase their bequest by an amount equal to the tax burden of the next generation. In short, the

effect of a budget deficit is that the burden will be borne in the end by the tax payers , despite

being financed from either borrowing or taxation. This can result in lags of budget deficits

having an impact on tax revenue as the revenue is not only influenced by the current deficits but

by deficits of previous years. The importance of this theory to tax performance is now

questionable given the continuous borrowing in developing nations and the associated budget

deficit in these economies. The theory suggests discipline in the monetary sector and effective

borrowing which does not affect generations to come.

3.2 LITERATURE ON BUOYANCY AND ELASTICITY ESTIMATION

The traditional model used to estimate tax buoyancy requires GDP to be a determinant of tax

revenue. Several studies (Wawire, 2000; Ariyo, 1997; Quazi, 1994; Osoro, 1993) of tax

buoyancy have used the following model:

εβα eYeT =

Page 40: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

31

Where; T – tax revenue, Y- income (GDP), α - constant term, β - buoyancy coefficient,

−e natural number.

The double-log version of the above equation is estimated using ordinary least squares (OLS).

The following estimating techniques have been suggested;

PROPORTIONAL ADJUSTMENT METHOD

This has been suggested by Sahota (1961) and Prest (1962) and was later described by Mansfield

(1972) and used by Omoruyi (1983), Osoro (1993) and Ariyo (1997). The method involves

isolating the data on discretionary revenue changes based on data provided by government. The

resulting data reflect only what the collections would have been if the base year structure had

been in force throughout the sample period (Osoro, 1993:13). The adjusted data is then used to

estimate the following equation;

pppp YT εβα ++= lnln

Where pβprovides an estimate of the income elasticity of the pth tax.

Ariyo (1997,20) cited shortcomings of this method. Data on revenue receipts directly and strictly

attributable to discretionary changes in tax policy are not available. In fact, it relies on budget

estimates for discretionary effects of tax revenue, which tends to differ substantially from the

actual tax revenue collected. The approach assumes that the discretionary changes are as

progressive as the underlying tax structure, hence it is contingent on the assumption that

discretionary changes are more or less progressive than the tax structure they modify (Leythold

and N’guessan, 1986 and Chipeta, 1998). Further the approach is highly aggregative compared

to other methods.

THE DUMMY VARIABLE METHOD

Developed by Singer (1968), this method introduces a dummy variable for each year in which

there was an exogenous tax policy change. The resulting estimating model becomes:

piipppp DYT εσβα +++= ∑lnln

Page 41: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

32

Where the dummy variable ,........)2,1( =iDi takes the value of zero (0) before the discretionary

change and one (1) after the change. The summation accounts for the possibility of multiple

changes in the period (Wilford and Wilford, 1978a:98 and Osoro,1993:14).

However, Ariyo (1997,25) suggested additional modifications of the two models above. The first

modification is to introduce a one-year lag in GDP. The argument is that new policy guidelines

contained in a budget speech may not be implemented until relevant circulars are issued. It

captures the potential effects on tax revenues due to implementation time lag. If there are

pronounced administrative lags or delayed remittances, the lagged value will be more

significantly associated with the dependent variable in each equation.

Another modification involves the introduction of a slope dummy. The argument is that over a

long period of time or under unusual circumstances, not only do the intercepts change but also

their slopes may change (Koutsoyannis, 1988:282-283). Ariyo (1997) argued that policy

proposals in the budget are based on the performance of each tax revenue source in the preceding

period. Those that perform above expectation are given more ambitious targets in the new fiscal

year can put under great surveillance. The modified models are given below.

.............lnlnln 121 ttttt YYT εββα +++= − first modification.

.............lnlnln 2413121 ttttt DDYYT εββββα +++++= − second modification.

Where; 1−tY- previous level of income (GDP), t - year, 1D -intercept dummy, 2D - slope dummy

( )tYDD *12 =.

Khan (1973) has applied this method in estimating the responsiveness of tax yields to increases

in national income and tax revenue forecasting. The shortcomings of this technique are that the

method becomes impossible to use where tax policy changes are too frequent and also creates a

potential multicollinearity problem from the inclusion of more than one dummy variable into the

tax function.

Page 42: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

33

CONSTANT RATE STRUCTURE

Used by Andersen (1973) for Denmark and Choundry (1975) for west Malaysia. The method

involves collecting statistics on actual tax receipts and data on monetary value of the legal tax

bases and corresponding revenues. The tax bracket of the base year is then multiplied by the

corresponding base values and the products summed up. The simulated tax revenue data is then

regressed on GDP. However this method can only be applied if the number of items is small, the

range of tax rates is narrow and the data can be compiled easily. Furthermore, this method

requires highly disaggregated data and detailed tax base series for all individual taxes and this

could be difficult to obtain, besides getting the same tax base over time.

DIVISIA INDEX

The method introduces a proxy for discretionary tax measures. The index measures technical

change, which is taken as the effects of discretionary changes in tax yields. The index is derived

from the estimated tax function analogous to the production function. The tax function must be

well defined, continuously differentiable and homogeneous of degree one. The method is best

suited where the revenue effects of discretionary measures are not available. It uses time trends

as proxies for discretionary changes. This is a major weakness because it introduces bias, leading

to either an overestimation or underestimation of the adjusted tax revenues (Choundry, 1979).

Choundry (1979) carried out a detailed comparison of these methods. The study concluded that

the Proportional adjustment method was more superior to the Dummy Variable approach, the

Divisia index and the Constant Rate Structure method. The study found that the Constant Rate

Structure method had the smallest elasticity estimates. This was probably due to the generation

of a simulated tax revenue series on the basis of the effective tax rate for a given reference year.

3.3 EMPIRICAL LITERATURE REVIEW

Empirical studies of determinants of tax buoyancy are not many; many studies of tax

performance have been interested in the analysis of the determinants of tax ratio (tax effort).

Page 43: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

34

Many studies (Choundry, 1979; Wawire, 2000; Ariyo, 1997 and Osoro, 1993) of tax buoyancy

have been interested in the calculation and interpretation of the buoyancy magnitude. This study

will mention some results from the studies of tax ratio as a way to guide the study answering its

objectives.

The principle determinants of tax performance from earlier work (Teera, 2001; Quazi, 1994;

Begum, 2007) supports the tax handle theory and includes the agricultural, industrial and

international trade sectors, because some sectors of the economy are more amenable to taxation

and they generate different taxable surpluses. The agriculture sector may be an important

determinant because small farmers are notoriously difficult to tax and subsistence agriculture

does not generate large taxable surpluses and many countries are unwilling to tax main foods that

are used for subsistence. Industrial sector can also be important as business owners typically

keep better books and records and can generate large taxable surpluses if production is efficient.

The international trade sector is typically the most monetized sector of the economy, the entrance

and exit to the country takes place in specified locations.

Quazi (1994) carried out a study of the determinants of tax buoyancy in developing nations using

35 countries for a period of ten years. The countries were chosen at random all over the world

but based on the level of national income. Zambia and Zimbabwe are the only SADC countries

that were selected. He used the ordinary least squares method in the regression of tax buoyancy

and its suggested explanatory variables. The model includes average growth of money supply

(monetization- M2), import sector output, industrial sector output, service sector output,

agricultural sector output, deficit, grant and Gross Domestic Product (GDP). He found

monetization to be positively related to tax buoyancy, he commented that an increase in

monetisation increases the documentation of the economy which increases tax collection. His

conclusion was that an increase in the level of monetization through an increase in

documentation facilitates the growth of tax revenue. Other variables found to affect buoyancy

include growth of industrial sector, growth of imports and growth of grants.

A study by Begum (2007) on the determinants of tax share and revenue performance (buoyancy)

is worth noting. It focused Bangladesh along with 10 other developing countries through a panel

Page 44: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

35

data analysis span for 15 years. The results obtained suggest international trade, broad money,

external debt and population growth to be significant determinants, with expected signs of the

estimated coefficients. The study identifies Bangladesh as the lowest tax effort country in the

sample, with an average tax effort index of 0.493. This has important policy implications that

Bangladesh and other countries having low tax effort (less than unity) are not utilizing their full

capacity of tax revenue, and therefore, have the potential for financing budgetary imbalance

through raising tax revenue. The tax effort index for both direct and indirect taxes is below 0.6,

implying that Bangladesh has the potential for raising revenue collection from both direct and

indirect taxes. In terms of tax buoyancy, Bangladesh ranks the second highest among the sample

countries, with a tax buoyancy ratio of 1.235, meaning that tax revenue is quite responsive to

GDP and effort has been made to increase tax revenue over the period. The study defined

monetization as broad money (money plus quasi money) divided by GDP, his argument was that

the degree of monetization affects the tax potential in the economy and a positive sign was found

in the study. This was in line with the expected sign, the reason being credited to the effect it

exerts on the overall economic activity of an economy.

Muriithi and Moyi (2000) applied the concepts of tax buoyancy and elasticity to determine

whether the tax reforms in Kenya achieved the objective of creating tax policies that made yield

of individual taxes responsive to changes in national income. They used the proportional

adjustment method to estimate the responsiveness of tax yields on income. The results showed

that tax reforms had a positive impact on the overall tax structure and on individual tax handles.

The study concluded that despite the positive impact, the reforms failed to make value added tax

(VAT) responsive to changes in income. However, VAT had been around for about eleven years

only and subjecting it alone in a regression model did not make statistical sense. There was need

to separate the effect of average monetary GDP and average total GDP on tax revenue and also

to use average figures instead of the annual ones (which this study used) because tax revenue

figures are on fiscal year basis that starts on 1st July while the GDP figures are on calendar year

that starts on 1st January.

A study by Osoro (1993) examined the revenue productivity implications of tax reforms in

Tanzania. In the study, the tax buoyancy was estimated using the double log form equation and

Page 45: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

36

tax revenue elasticity using the proportional adjustment method. The argument for using the

method was that a series of discretionary changes had taken place during the sample period

1979-89, making the use of dummy variable technique impossible to apply (Osoro, 1993:14). For

the study, the overall elasticity was 0.76 with a buoyancy of 1.06. The conclusion was that tax

reforms in Tanzania had failed to raise tax revenues. The reasons were that government granted

numerous tax exemptions in conjunction with poor tax administration.

Tax performance have been related to various combinations of explanatory variables by

Chelliah, Baas and Kelly (1975), using 47 countries averaged over the 1969-1971 period. They

obtained the best fit using the agricultural share and export share in GDP as explanatory

variables. They found that industry was positively related to tax performance, while agriculture

negatively related and the export ratio insignificant. An update of earlier work was done by Tait,

Gratz and Eichngeen (1979) using a sample of 47 developing nations with data averaged over the

1972-76 periods. They found stability in the results compared to the earlier studies. Overall, their

results suggested the Chelliah et al. (1975) specifications were appropriate, using their non-

export income per capita, the share of mining, and the share of non mineral exports in GNP as

explanatory variables or non export income per capita and the share of exports in GNP as

explanatory variables. They found that the share of agriculture was significant.

Panel data analysis has been used in a study by Tanzi (1987) for a sample of 86 developing

countries. He examined how the share of tax revenue in GDP was related to the logarithm of per

capita income. The study found a positive relationship between the two. In a subsequent study,

Tanzi (1992) extended this analysis to incorporate a specific time dimension by analyzing a

series of cross sections. For a sample of 83 developing countries over the period 1978-88, the

study found that the relationship between tax share and per capita income was weak. It was

hypothesized that other factors, such as macroeconomic stability, the need to service debt and the

changing structure of the economy had become more important determinants. It was found that

the share of agriculture in GDP was strongly inversely related to the tax share and its explanatory

power was greater than per capita income. Leuthold (1991) uses panel data on eight SSA

countries over the 1973-81 period to estimate a version of this model. Agriculture share was

inversely related and foreign trade was directly related to tax performance.

Page 46: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

37

Ariyo (1997) evaluated the productivity of the Nigerian tax system for the period 1970-90. The

aim was to devise a reasonable accurate estimation of Nigeria’s sustainable revenue profile. In

the study tax buoyancy and elasticity were estimated using two equations modified by Ariyo

respectively. The slope dummy equations were used for the oil boom and structural adjustment

programmes (SAPs). It was found that on overall, productivity was satisfactory. However, the

results indicate wide variations in the level of tax revenue by tax source. The variations were

attributed to the laxity in administration of non-oil tax sources during the oil boom periods.

Significant reduction in public expenditure and prudent management of financial resources were

suggested as solutions to the fiscal deficit. The study further asserted that there was need to

improve tax information system to enhance the evaluation of its performance and facilitate

adequate macro-economic planning and implementation, Ariyo (1997,33).

Tax buoyancy has been estimated using the proportional adjustment method in a study by

Chipeta (1998). He evaluated the effects of tax reforms on tax yields in Malawi for the period

1970-94. The results revealed a buoyancy of 0.95 and an elasticity of 0.6. The study concluded

that the tax bases had grown less rapidly than GDP. Kusi (1998) studied tax reform and revenue

productivity of Ghana from 1970 to 1993. Results showed a pre-reform buoyancy of 0.72 and

elasticity of 0.71 for the period 1970-82. The period after reform (1983-93), showed increased

buoyancy of 1.29 and elasticity of 1.22. The study concluded that reforms had contributed

significantly to tax performance.

The Divisia index methodology has been used by Milambo (2001) to study the revenue

productivity of the Zambian tax structure for the period 1981 to 1999. The results showed

elasticity of 1.15 and buoyancy of 2.0, which confirmed that the tax reforms had improved the

revenue productivity of the overall system. However, these results were not reliable because time

trends were used as proxies for discretionary changes.

Wawire (2000) used total GDP to estimate the tax buoyancy and income elasticity of Kenya’s

tax system. Tax revenues from various sources were regressed on their tax bases. The study

concluded that the tax system had failed to raise necessary revenues. However, the shortcomings

Page 47: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

38

of the study were, first, it never considered other important determinants of tax revenues, for

example, unusual circumstances that could have affected tax revenue productivity. Second, it

never disaggregated tax revenue data by source hence it was difficult to say which taxes and

bases contributed more to the exchequer. Third, it never took into account the time series

properties of the data.

Analysis of revenue buoyancy and its fiscal implications have been done in a working paper by

Isabelle and Christophe (2008). The paper provided an analysis of the factors behind recent

revenue buoyancy and examines past responses to unexpected revenue gains. It also discussed

whether improved information on fiscal positions and future fiscal challenges, combined with

relevant fiscal rules might help in avoiding a repetition of past errors in fiscal policy. They

mainly focused on the impact of political variables on revenue buoyancy.

A study carried out by Teera (2000) found that the results of the dynamic measure of tax

performance (tax buoyancy) indicate that the high-income OECD group has the least percentage

number of countries with a buoyancy ratio below unity, followed by the lower middle-income

group. This implies that the lower income groups have made less effort to increase tax revenues

over the period as compared to the higher income groups. The spearman’s rank correlation

coefficient shows that there is little correlation in the ranking order of countries according to tax

effort index and tax buoyancy ratios, implying that a low effort index does not necessarily

indicate that countries have not made efforts to increase tax revenues nor does a high effort index

indicate that countries have made efforts to increase tax revenues over the years. He mainly

hammered on the tax evasion variable. In the regression variables like total expenditure and time

trend were included.

3.4 CONCLUSION

In this chapter theories which help us to determine the determinants of tax buoyancy and their

expected effects have been discussed. Few studies have considered the variable monetization,

and this makes this study one of the pioneers to have it analysed. The tax handle theory is the

main theory associated with most variables found to be determinants of tax buoyancy (see

Page 48: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

39

Appendix I for more detail). The Ricardian equivalence theory can also explain the monetization

variable inclusion predicting a negative relationship to tax performance. According to the

literature reviewed the variables affecting tax buoyancy includes monetization, growth of

external aid, fiscal deficit growth, national income growth, external aid growth, growth of

economic sectors in the economy, total expenditure growth, demographic factors and external

growth. The signs of coefficients have been varying for many studies revealed in literature;

however explanations have been given for each result.

The study will adopt panel data methodology which have been used by Leuthold (1991), Begum

(2007), Tanzi (1987) and Teera (2001). Some methodologies have been used but panel data

analysis proved to be used for regional studies. This study will differ from other studies by the

fact that it uses annual data rather than periodic averages data as have been used before. The

study (due to its nature) adopts and utilises the basic definition of tax buoyancy rather than

several modifications developed by some researchers.

Page 49: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

40

CHAPTER FOUR

METHODOLOGY

4.0 INTRODUCTION

The main objective of this chapter is to discuss the econometric model to be applied in this study.

Specifically, it encompasses the model to be adopted, definition of variables, their proxies,

estimation techniques, data types and sources. It also justifies why panel data method was used

instead of the ordinary time series or cross section methods.

4.1 MODEL SPECIFICATION

THE GENERAL MODEL

The suggested model to be estimated for this study will be in a form that suits our data with both

cross sectional and time dimensions and the type of study which is a regional study. The panel

data model is presented below:

1.4..............................................ititiit XBOUY εβα +′+=

2.4..................................................................., itittitWhere νηλε ++=

Where i denotes countries and t denotes time. BOUYit is the dependent variable namely tax

buoyancy explained by Xit , which is a matrix of explanatory variables that affect buoyancy. α

and β are the usual constants. The error term itε, is a two way component which takes into

account country effects itη and time effects tλ

and the usual error term itν which satisfy the

classical linear regression model assumptions of a mean of zero and a constant variance.

Panel data methodology is suggested because the study is a regional study (across countries) and

has time dimension. There is need to consider these characteristics. Panel data admits that

countries are not homogeneous unless the homogeneity is tested. There is a possibility of having

Page 50: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

41

differences which are specific to each individual country and hence using panel data analysis will

cater for these. Panel data or longitudinal data can be defined as a data set that follows a given

sample of individuals over time and thus provides multiple observations on each individual in the

sample (Hsiao, 1996). Cameron and Trivedi (2005,697) defined panel data as repeated

observations on the same cross section, typically of individuals or firms in microeconomics

applications, observed for several time periods. Panel data provide information on individual

behaviour both across time and across individuals.

Panel data makes it possible to capture the behaviour of variables in both time series and cross-

section dimensions. According to Islam (1995) a panel data approach allows us to take into

account some potentially important factors that cannot be measured. In a pure cross-country

instrumental variable regression, any unobservable country specific effects becomes part of the

error term, which may bias the coefficient estimates.

4.2 EMPIRICAL MODEL TO BE ESTIMATED

The following model with support from various empirical literatures (for example Begum

(2007), Teera (2001) and Tanzi (1987)) is deemed appropriate. The model will help us in the

empirical examination of the determinants of tax buoyancy in developing nations.

3.4..............................1098

7654321

itititit

itititititititiit

EXPDFAID

XMPOPECONDEBTMSINDAGRBOUY

εβββ

βββββββα

++++

+++++++=

Where; α and 1β to 10β are constants to be estimated,

BOUY - Annual tax buoyancy.

AGR - is annual growth of the agricultural sector contribution to GDP

IND - is average growth of the industrial sector contribution to GDP.

−MS Annual growth of money supply (M2).

DEBT - Annual growth of external debt.

ECON - Level of economic development approximated by annual growth of GDP.

−POP Population density per 1000 ha.

Page 51: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

42

−XM Trade openness.

−AID Annual growth of external aid to GNP ratio.

−DF Annual growth of deficit and

EXP - Annual growth of total expenditure.

itittit νηλε ++=, where 14........3,2,1=i and 12............3,2,1=t

The model can be estimated using pooled ordinary least squares (OLS) model, random effects

model or fixed effects model. There are specification tests to be carried out to guide on the model

to be used. However, econometric researchers normally use fixed effects and random effects

models when dealing with panel data.

4.3 DISCUSSION AND JUSTIFICATION OF VARIABLES

4.3.1 DEPENDENT VARIABLE

Tax Buoyancy (BOUYit)

Tax buoyancy refers to the total response of the tax system due to both changes in national

income and the deliberate decision of the government to raise the tax rates, improve tax

administration, change in the tax code etc (Teera, 2002). The buoyancy of a tax system reflects

the total response of tax revenue to changes in national income as well as effects of discretionary

changes in tax policies over time (Jayasundera, 1991). According to Begum (2007, 3), the

buoyancy of tax can be estimated in two ways: (i) by calculating the ratio of the percentage

change in tax revenue to the percentage change in national income/GDP; or (ii) by regressing the

tax revenue on the tax base (e.g., national income/GDP) after taking the natural logarithm for

each of them.

The first definition is used in the study because tax buoyancy of each country each year is to be

used. The other definition can only be used when periodic buoyancy is needed. Studies by Quazi

Page 52: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

43

(2004) and Begum (2007) have used the same definition as of this study. Two consecutive years

are used in the calculation and the following formula is used;

GDPinchange

revenuestaxinchangeBouyancy

%

%=

Where buoyancy is to be calculated annually and is not a propensity. This should not be likened

to tax elasticity since it includes discretionary effects, we have not controlled for other variables

that affect GDP to change because we need to assess the total change due to all factors. This

means that buoyancy is both a quantity and quality measure of tax effort.

4.3.2 DETERMINANTS OF TAX BUOYANCY

Monetization ( MSit )

Monetization is the rate at which money supply (broad money known as M2) is growing in an

economy. M2 is chosen as a measure of monetization and not M3 or M1, because M2 measures

the amount of mobile money circulating in the economy while M3 include less mobile (active)

components of money supply. M1 will in the same way not preferred because it undermines

monetary activities in the economy that are related to taxation. In the study annual growth of

money supply has been divided by average growth in GDP to standardize the variable for

international comparison. This definition has been used also by Quazi (1994). A study by Begum

(2007) used the ratio of broad money to GDP as the proxy for monetization. Increases in

monetization will increase the documentation of the economy which will increase the collection

of each tax. The degree of monetization tends to affect the tax potential of the economy (Begum,

2007:14). Therefore the study expects a positive sign in the regression analysis.

The level of economic development (ECONit )

This is approximated by average annual growth of GDP. This lends support to the hypothesis

that as countries develop tax bases increase more than proportionately to the growth in income.

According to the tax handle theory the inclusion of this variable as a determinant of tax

Page 53: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

44

buoyancy is necessary. Musgrave (1969) argues that the lack of availability of ‘tax handles’

might limit revenue collection at low levels of income and these limitations should become less

severe as the economy develops. Economic development is assumed to bring about both an

increased demand for public expenditure (Tanzi, 1987) and a larger supply of taxing capacity to

meet such demands (Musgrave, 1969). A higher per capita income reflecting a higher level of

development is held to indicate a higher capacity to pay taxes as well as a greater capacity to

levy and collect them (Chelliah, 1971). There is also the consideration that, as income grows

countries generally become more urbanised. Urbanisation per se brings about a greater demand

for public services while at the same time facilitating tax collection (Tanzi, 1987). A positive

sign is expected for this variable.

Trade Openness (XMit)

The relative size of the overseas sector, which is a measure of openness, reflects the degree of

exposure of an economy to external economic influences. The same definition has been used by

Teera (2000). The variable has been captured by the annual growth of import and export sectors

contribution to national income divided by average growth of GDP. Algebraically it can be

represented as follows;

Where; is the annual growth of export sector contribution, is the annual growth of import

sector contribution and is average annual growth of national income. and represents country and time respectively.

Quazi (2004) used the export sector growth only for the same reasons (it is easy to collect tax

revenue from the international trade sector), and did not combine with the import sector growth.

His argument being that developing countries are net importers, however both exports and

imports are registered in developing nations and have to be accounted for. In the presence of

inward capital flows, the overall level of activity in the economy is artificially and or temporarily

increased through foreign borrowing and so is the aggregate tax base. As a consequence, tax

revenues become artificially buoyant (Seade, 1990) and volatile. Certain features of overseas

trade make it more amenable to taxation than domestic activities, and in developing countries,

Page 54: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

45

the overseas trade sector is typically the most monetized sector of the economy. The

administrative ease with which trade taxes can be collected makes them an attractive source of

government revenue when administrative capabilities are scarce (Linn and Weitzel, 1990). A

positive sign is expected for trade openness.

Average Growth of Agriculture Sector (AGRit)

Agriculture is considered to be a salient feature regarding the structure of the economy and as

Tanzi (1992) asserts, a country’s economic structure is one of the factors that could be expected

to influence the level of taxation. The influence of agricultural sector is from both the demand

and supply point of view (Tanzi, 1992). On the supply side, it is very difficult to tax the

agricultural sector “explicitly”, though it is often very heavily taxed in many implicit ways such

as; import quotas, tariffs, controlled prices for output, or overvalued exchange rates (Bird, 1974;

Ahmad and Stern, 1991; Tanzi, 1992). On the other hand, small farmers are notoriously difficult

to tax and a large share of agriculture is normally subsistence, which does not generate large

taxable surpluses, as many countries are unwilling to tax the main foods that are used for

subsistence (Stotsky & WoldeMariam, 1997). On the demand side, since many public sector

activities are largely city-oriented, it may be assumed that the more agricultural a country is, the

less it will have to spend for governmental activities and services. The expected sign is

ambiguous.

Average Growth of External Aid (AIDit)

External aid refers to the dollar value of an amount received in kind from abroad into a country.

The proxy for this variable has been taken as the average annual growth of external aid including

grants divided by average growth of GNP. Quazi (2004) used the same proxy for this variable.

Aid and grants have been a major source of development finance for the majority of developing

countries over the past few decades. Empirical literature has tended to evaluate the impact of aid

by including it as a variable in a regression for the determinants of some economic performance

indicator, emanating from the general. Servicing of the foreign debt requires a trade account

surplus, which in turn may require a reduction in imports. This affects revenue given the high

Page 55: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

46

dependence of the tax system on the external sector concern that it might have a negative impact

on some of such indicators. For instance, there is a general concern that aid may decrease

taxation revenue in recipient countries. In fact, the results in Franco-Rodriguez, Morrissey, and

McGillivray (1998)’s study on Pakistan were in agreement with this concern. A negative sign is

expected in the study.

Fiscal Deficits and Debt (DEFit and DEBTit)

Fiscal deficit variable is measured by the annual growth of the difference between revenues and

expenditures in the economy. External debt growth is approximated by the amount of money

which a given country owes to other countries. According to the tax handle theory the two

variables should be determinants of tax buoyancy. The growth of public spending has generated

large fiscal deficits in many countries, leading to increases in the share of public debt relative to

GDP. With a large debt, the government needs to raise the revenues necessary to service it.

When the interest on the debt exceeds net borrowing plus the possible reduction in non-interest

expenditure, the level of taxation must go up unless the rate of growth of the economy is high

enough to neutralize the increase. Therefore public debt plays a role in determining the extent to

which countries may take advantage of their taxable capacity (Tanzi, 1987). However, a high

debt burden can also create macroeconomic imbalances that may tend to reduce the tax level. In

general, however, on balance, a high debt burden would tend to raise the tax level, ceteris

paribus (Tanzi, 1992). On the other hand, however, countries faced with an increased trade

deficit may try to restrict imports as an alternative to exchange rate adjustment irrespective of the

source of the trade imbalance. This will reduce revenue from import duties. A positive sign is

expected for debt and a negative sign for deficit.

Population Density (POPit)

Population density refers to the total number of people living in area equal to a thousand

hactares. The variable is supported by the tax handle theory to be one of the determinants of tax

buoyancy. A study by Teera (2000) used the same definition as of this study. However, some

studies used population growth, logarithm of population but for the same reason as of this study.

Page 56: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

47

Population density is expected to have an adverse effect on the tax buoyancy, mainly because the

higher the density of population the higher will be the use of taxable sources (i.e. rising the tax

base), and the tax authorities could intensify their efforts to collect taxes at a relatively minimal

cost as compared to a sparsely populated country. Conversely, in a thinly populated area,

administrative costs are expected to be higher in terms of total yields and therefore, less

encouraging for collection of tax revenues. In such a situation, the degree of tax evasion and tax

avoidance may also be relatively higher than in the densely populated area (Ansari, 1982). The

expected sign for this variable is ambiguous.

Annual Growth of Industrial Sector (INDit)

The proxy for this variable is taken as the annual growth of industrial sector contribution divided

by annual growth of GDP. The same proxy has been used by Quazi (2004) and Teera (2000).

Manufacturing enterprises are easier to tax than agricultural enterprises since business owners

typically keep better books of accounts and records. Manufacturing can generate larger surpluses

if production is efficient. Therefore the variable is expected to be positively related to the tax

buoyancy.

Annual Growth of Total Expenditure (EXPit)

The variable captures the spending behaviour of the economy. It is approximated by the annual

growth of government spending. The tax handle theory allows it to be among the variables

affecting tax performance. Teera (2000) included this variable in his study and the same

definition in this study. As spending increases, this should force the revenue from tax to be

increased so as to finance the expenditure. In developing nations, tax revenue is increased mainly

by the increase in tax rates or by introducing new taxes to easy-to-tax points. However it can be

done by tax base broadening (for example, introduction of toll gates and new taxes), and this is

more efficient as it reduces the tax burden. Usually for many economies most of the revenue

comes from taxation, while other revenue comes from various sources including income

generating projects, profit from parastatals (many of them however rarely make profit) and donor

funds. A positive sign is expected in the study.

Page 57: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

48

4.4 MODEL SPECIFICATION TESTS

When estimating the panel data model one is faced with the choice of whether to use the pooled

model, the fixed effects model or the random effects model. The pooled model leads straight to a

Classical Linear Regression (CLR) formulation for which OLS will produce consistent and

efficient estimates for the parameters.

Specification tests shall be carried out to decide which model best fits the data. The following

tests are going to be performed; the fixed effects test, the Hausman test, the Breusch and Pagan

Lagrange multiplier test to determine which model best suits our data and the F-test to determine

the validity of the whole model.

4.4.1 The Fixed Effects test

The F-test is used to test whether the pooled or fixed effects model is appropriate. The null

hypothesis for the F-test is;

H 0 : the pooled (restricted) model is correct

H 1 : the data is panel (fixed effects model is correct)

The F-statistic is calculated as follows;

)/(

)1/()(),1(

KNNTRSS

NRSSRSSKNNTNF

U

UR

−−

−−=−−−

Where R= restricted model (pooled/OLS model)

U=fixed effects (unrestricted) model, N= number of countries, T=time period and K= number of

parameters, RSS = residual sum of squares.

If the null hypothesis is rejected then the study will either adopt the fixed effects or random

effects model. If the F statistic is significant it implies that there are significant individual (state

level) effects implying that pooled OLS would be inappropriate (Baum, 2006). The next step will

be to test between random effects and fixed effects model using the Hausman test. A

complimentary test at this level may be carried between pooled OLS and random effects model.

Page 58: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

49

4.4.2 The Hausman Test

In deciding whether to choose the fixed effects model or the random effects model if N, the

number of countries is large and T, which is the time frame is small and if the assumptions

underlying the random effects hold then the random effects model is more efficient and is then

used for estimation purposes. The fixed effects model is used if the errors and the observations

are correlated. The Hausman test (1978) is used to determine the suitability of the models. The

test is distributed Chi-Squared Asymptotic around the null hypothesis that the random effects

model is appropriate. The fixed effects model is consistent under both the null and the

alternative. The random effects model is consistent under the null and inconsistent under the

alternative. The hypothesis we wish to test is that

H o :Cor ( iti x,α 0= ) (Random effects model)

H)0(: ,1 ≠iti xCor α

(Fixed effects model)

Rejecting the null hypothesis implies that fixed effects model is appropriate. We can test the

suitability of the models using probability values. If the probability of the chi-square is less than

0.1 it implies that fixed effects model is more appropriate than the random effects model.

According to Baum (2006) it implies that individual effects appear to be correlated with the

regressors and hence the fixed effects since will give efficient estimates.

4.4.4 The Breusch and Pagan Lagrange multiplier test

As a complimentary test we can also decide whether to use the pooled model versus the random

effects model using the Lagrange multiplier (LM) method recommended by Breusch and Pagan

(1979). Under the null hypothesis,

H 0 : 2

µδ= 0 (The alternative hypothesis being 0: 2 ≠′

µδH ), Breusch and Pagan have shown that

.]1)([

)1(2

22

11

−−

= ∑∑==

it

T

t

N

it

NTg ε

Page 59: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

50

is asymptotically distributed as a chi-square distribution with one degree of freedom, where

the itεare the initial least squares (OLS) residuals from regressing itY

on itX.

Rejecting the null hypothesis implies that the random effects model is appropriate. If the LM

statistic is significant it implies that the random effects model is appropriate, hence pooling the

data will give biased results. The major drawbacks of the Breusch-Pagan statistic is that it is two

sided in spite of the fact we know that variances cannot be negative and that Lagrange multiplier

tests often have low power.

4.4.5 The Time effect test

The study shall employ the F test to test the significance of time effects. The null hypothesis is

that all time dummy parameters except one are zero. Rejecting the null hypothesis will imply that

time dummies are not relevant for the present study.

4.4.6 The Country effects Test

The study shall employ the F test to test the significance of country effects. Rejecting the null

hypothesis will imply that there is homogeneity in the data and hence pooled OLS model is

appropriate.

A joint test for both country and time effects may be carried out using an F-test. This enables us

to include both country and time effects in the final model. The test is important if country

effects are significant (Baumol, 2006). The reason being that at least one of the included

individuals (countries) is affected by shocks and hence can be analysed.

4.4.7 The F-test (Significance of the whole model)

The study shall perform the F-test to determine the significance of the whole model The F

statistic from the test will be compared against the critical and the decision criteria will be to

reject the null hypothesis that none of the variables explain tax buoyancy when it is greater than

the critical value.

Page 60: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

51

In deciding whether to choose a pooled model or fixed effects model one uses the F-test and

between a pooled and random model one uses the Lagrange Multiplier test. In choosing between

fixed effects model and random effects model the Hausman test is used.

4.5 PARAMETER TESTS AND MISSPECIFICATION TESTS

Parameter tests brings about parameter significance to the used variables. They are captured by

the statistical package to be used (Stata). This involves the dropping of highly insignificant

variables. This approach is usually termed the General to Specific Model Approach. If efficiency

is to be improved this approach will be followed.

Misspecification tests will also be carried out and this ensures that statistical assumptions are not

violated. These as well are automatically controlled by the Stata econometric package.

The Multicollinearity test will also be carried out. The problem of multicollinearity to

econometric research is that it will be difficult to separate the effects of one explanatory variable

on the dependent variable from the other. The effect can be detected by merely looking at the

correlation matrix for a relationship that exceeds 0.80 (Cameron and Trivedi, 2005). However a

test statistic known as Mean VIF (Bruderl 2000) can also be computed. A value greater than 4

will imply a strong multicollinearity among variables. The rule of thumb is to drop one of the

correlated variables in the regressions.

Heteroskedasticity tests will be carried out in order to yield consistent though not efficient

estimates. In the presence of heteroskedasticity standard errors will be biased and robust standard

errors (panel corrected standard errors) will be computed to correct possible presence of

heteroskedasticity. A way to check for heteroskedasticity is to compute the correlation between

the explanatory variables and the error term, if it is near zero then there is homoskedasticity and

hence no reason to use robust standard errors (Cameron and Trivedi, 2005).

Page 61: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

52

Autocorrelation test, if necessary will also be done in order to correct for autocorrelation. This

arises when error terms are not homoskedastic. A Durbin-Watson statistic will be computed, any

value not close to 2 will imply serious autocorrelation.

4.6 DATA TYPES AND SOURCES

The sample will span the period 1994 to 2005.8 The sample targets fourteen countries in the

SADC region. Data used in the study was collected from Government Finance Statistics, World

Tables, International Financial Statistics, World Development Indicators (various issues) and

SADC databases. The World Development Indicators being the main data source, with obvious

advantages that its statistics are sometimes not found in host countries statistics and also figures

from this source enables easy and meaningful international comparisons. A major practical

problem is data inadequacy which extends to inconsistency in published figures. However where

necessary, data have been supplemented with information from other mentioned sources.

Annual growth of each variable is computed and used as the exogenous variable in the regression

analysis. Each exogenous variable has been divided by growth of domestic product to make each

variable standardised except aid (AID) which has been divided by GNP, level of economic

development (ECON) and population density (POP). Whilst every care was taken to maintain

accuracy in manipulating the data, the possibility of errors and omissions cannot be completely

rejected.

4.7 CONCLUSION

This chapter has outlined the methodology and estimation procedures to be employed in the next

chapter. Besides justifying the use of panel data, the chapter gives a rich analysis of explanatory

variables, their proxies and data sources. The next chapter concentrates on estimation and

interpretation of results.

8 The period have been chosen according to data availability in most current periods. After 2005 up to date no or little statistics could be found especially for DRC, Namibia and Zimbabwe.

Page 62: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

53

CHAPTER FIVE

ESTIMATION AND INTERPRETATION OF RESULTS

5.0 INTRODUCTION

This chapter focuses on the estimation of the model in order to determine the coefficients, the

significance and reliability of the factors determining tax buoyancy in developing nations. The

chapter will present summary statistics, results of relevant test undertaken in the study and final

results to be used to conclude the study.

5.1 SUMMARY STATISTICS

Table 5. 1: Summary Statistics

Variable Obs Mean Std.

Dev.

Min Max

BUOY 168 .8865928 1.892238 -3.947929 13.70466

ECON 168 4.064729 29.9848 -92.09988 285.75

POP 168 837.8695 1463.119 19.56571 6124.399

AGR 168 .5674204 16.515 -19.2799 202.326

IND 168 2.33925 18.73489 -42.00838 148.744

MS 168 27.7022 75.8601 -60.31707 458.1371

AID 168 -.0100021 .9947225 -6.9842 7.643232

DEBT 168 .0280023 .1284755 -.4400696 .755631

DF 168 1.86742 18.05055 -1.578498 206.0018

XM 168 3.449623 26.2771 -19.09303 332.0241

EXP 168 .0261509 .3233261 -.8615385 2.928571

Table 5.1 above shows the commonly used descriptive statistics, the mean and standard

deviation. It also includes the minimum and maximum values for each variable; this helps us to

check for outliers in the data collected. The number of variables is 168 for each variable and

hence it is a balanced panel. There is much variability in the population density variable as

indicated by a standard deviation of 1463.119 (no alterations will be made)9 and this is because

9 The variable has to maintain as transforming it using logs to reduce variability changes the scale and hence disrupting panel tests. It is a major theoretical variable.

Page 63: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

54

of the different sizes of the economies, also the debt variable records the lowest standard

deviation of 0.1284755. The tax buoyancy variable has a mean of 0.8859137 indicating that on

average the tax system in SADC economies is not that buoyant, however it ranges from a

minimum of -3.947929 to a maximum of 13.70466 with a standard deviation of 1.897908. This

indicates that, although the tax systems are sometimes too buoyant the variability among the

countries is not much. The monetization variable has a mean of 27.8601 indicating that there is a

significant growth of money supply in developing nations, however there is also variability

among the nations of 75.8601, having a maximum growth of 458.1371%.

5.2 CORRELATION MATRIX

Table 5. 2: Correlation Matrix ECON POP AGR IND MS AID DEBT DF XM EXP

ECON 1.0000

POP 0.0016 1.0000

AGR -0.0703 -0.0443 1.0000

IND 0.2132 -0.1038 0.0391 1.0000

MS -0.1153 -0.1269 0.0436 0.0889 1.0000

AID -0.0942 0.1071 -0.0984 -0.2109 0.0615 1.0000

DEBT 0.0113 0.0285 -0.0477 0.0297 0.0548 0.1105 1.0000

DF -0.0053 0.0257 -0.0423 -0.0387 -0.0337 0.0017 -0.0150 1.0000

XM -0.0522 -0.0039 0.0385 0.0973 -0.0467 -0.0956 -0.0257 -0.0137 1.0000

EXP -0.1346 -0.0260 -0.0765 0.1720 0.1449 0.0430 0.0369 0.0570 0.0406 1.0000

Table 5.2 above presents the correlation between the variables used in the study. The relationship

between variables is far less than 0.8 implying the variables do not have a strong relationship and

hence this shows absence of multicollinearity, although this can be tested. The Mean Variance

Inflation Factor (VIF) is less than 4 as reported in Appendix III (A), meaning that there is no

strong multicollinearity among variables. Hence, this allows us to include all variables in the

regressions.

5.3 TESTING FOR MODEL SPECIFICATION

The model specification tests presents the necessary tests which will guide the appropriate model

which suits the data collected and used in the study, the model suggested will produce more

efficient and unbiased results to be used for policy conclusions. The tests includes tests between

Page 64: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

55

pooled OLS and fixed effects model, pooled OLS and random effects model and fixed effects

model against random effects model. Tests to include time and country effects in the model are

also presented.

5.3.1 FIXED EFFECTS TEST (POOLED OLS AND FIXED EFFECTS)

The F-test was used to test for the significance of country effects which determines whether to

include country effects as explanatory variables. Results in Appendix III (F) show an F-critical

value of 0.89 and a p-value of 0.5690. We therefore do not reject the null hypothesis and

conclude that the country effects are not significant and the data may be pooled. Pooling the data

will therefore give unbiased estimators as compared to the fixed effects model. We have detected

homogeneity in the data. This is also supported by a correlation value between the individual

specific effects factor and the explanatory variable which is not significant, that is

=);( iti XCor η 0.0262 (refer to Appendix III (F)). We conclude that there are no unobservable

specific country effects and therefore no need for panel data and hence the data has to be pooled.

5.3.2 LM TEST FOR RANDOM EFFECTS (POOLED OLS AND RANDOM EFFECTS)

The null hypothesis of the test is that there is no correlation between regressors and country

effects. The Breusch and Pagan Lagrange multiplier test reports a chi-square statistic of 0.36 and

a p value of 0.5511 thereby accepting the null hypothesis at all levels of significance. We can

therefore conclude that the random effects model is not appropriate for the present study. The

LM test suggest for pooled OLS over the random effects model. The test is complimentary to the

F-test and they are in agreement that data cannot be panel and hence should be pooled.

Table 5. 3: LM Test For Random Effects

buoy[country,t] = Xb + u[country] + e[country,t]

Estimated results: Var sd = sqrt(Var)

BUOY 3.580563 1.892238

E 2.327946 1.525761

U 0 0 Test: Var(u) = 0

chi2(1) = 0.36

Prob > chi2 = 0.5511

Page 65: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

56

5.3.3 HAUSMAN TEST (RANDOM EFFECTS AND FIXED EFFECTS)

The null hypothesis of the Hausman test is that differences between fixed and random

coefficients are not systematic. Results from the Hausman test are presented in Table 5.4 below;

Table 5. 4: Hausman Test

Coefficients

(b) Fixed

(B) Random

(b-B) Difference

sqrt(diag(V_b-V_B)) S.E.

ECON | .0084629 .0058325 .0026304 .001514

POP | -.0000342 .0000263 -.0000605 .0001884

AGR | .0174207 .0191475 -.0017268 .0023648

IND | .017672 .0243331 -.006661 .0031603

MS | -.0025617 -.0032016 .00064 .0007303

AID | -.7546385 -.7315459 -.0230927 .0565374

DEBT | -1.459248 -1.009444 -.449804 .5973097

DF | -.0182977 -.016493 -.0018047 .003622

XM | .0040773 .0054175 -.0013402 .0017591

EXP | .7685685 .7987246 -.030156 .0984933 chi2(10) = (b-B)'[(V_b-V_B)^(-1)](b-B)

= 7.51

Prob>chi2 = 0.6765

Table 5.4 above reports the results of the Hausman test obtained after running the fixed effects

and random effects model. The Hausman test reports a 2χ statistic of 7.51 with a p value of

0.6765. We therefore do not reject the null hypothesis of unsystematic differences of coefficients

at all levels of significance. This means that the random effects model is an efficient estimator of

the data as compared to the fixed effects model.

5.3.4 COUNTRY EFFECTS TEST

Country effects have also been tested using an F- test, the null hypothesis being that the effects

are significant and hence present. This will trace the differences among the countries. The test

reported an F value of 1.20 and a probability of 0.2829 (see Appendix III (C)). This shows that

the country effects are not significant and will not be included in the regressions in order to

achieve efficient results. The test shows that there are no individual specific factors for the

countries included. The test is in agreement with the fixed effects test, and pooled OLS is

suggested.

Page 66: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

57

5.3.5 TIME EFFECTS TEST

The study tested for the significance of time effects. The null hypothesis is that time effects are

not present while the alternative states that time effects are present. After running a fixed time

effects model in Stata we got an F-statistic of 4.59 and a p-value of 0.0000 (see Appendix III

(D)). Therefore we do not reject the null hypothesis at all levels of significance and conclude that

the time effects are present and should be included in the final model.

5.3.6 JOINT TEST FOR TIME AND COUNTRY EFFECTS

Joint test shows that country and time effects have a significant combined effect on the results.

The test reported an F-statistic of 2.72 and a probability value of 0.0002, showing the

significance of the joint test (see Appendix III (E)). Hence, we may include both of them if this

will lead to us getting efficient results. However the joint test is of importance if fixed effects

model have been chosen by other specification tests above, this enables the incorporation of time

effects when country effects are present. Therefore in this study we ignore the results of the joint

test since the pooled OLS model have been chosen.

From the specification tests carried above (see also Appendix III (B) for a summary table of the

tests) above we found that the pooled OLS model with time effects is appropriate for this study.

We failed to get any significant individual or country specific differences among the included

countries which also suggest that the tax system is the same in the developing nations. Cameron

and Trivedi(2005,699) indicated that if the panel data model is correctly specified and regressors

are uncorrelated with the error then it can be consistently estimated using pooled OLS.

5.4 PARAMETER TESTS

Test for autocorrelation has been carried using DW-statistic and it has recorded a value of

2.256308, this shows us that there is no serious autocorrelation. Therefore there is no need to

correct for autocorrelation.

Page 67: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

58

Heteroskedasticity test has been carried by computing the correlation between the individual

specific effects factor and the explanatory variables. A value of 0.0262 has been reported and the

conclusion therefore will be that there is homoskedasticity in the data since the value is close to

zero. From this information there is no need to use robust standard errors and corrected

coefficients, implying that the coefficients and standard errors reported are efficient.

5.5 MODEL ESTIMATION

Based on the specification test carried in Section 5.3 above especially the Fixed effects test,

Hausman test and the Breusch Pagan Langrage multiplier test, a pooled OLS methodology have

been the best model suggested. Hence in this section the General and Specific models will be

presented as guided by the methodology of the study.

Table 5. 5: General Pooled Ordinary Least Squares Regression Model

Dependent Variable BUOY

BUOY Coef. Std. Err. P>|t|

ECON .0055739 .0041014 0.176

POP .0000214 .0000811 0.792

AGR .0186351 .0071676 0.010***

IND .0224122 .0068147 0.001***

MS -.0041815 .0016613 0.013 **

AID -.745964 .1228295 0.000***

DEBT -.8554755 .9184389 0.353

DF -.0167008 .0064789 0.011**

XM .0055541 .0044908 0.218

EXP .8689488 .3789776 0.023**

Trend -.0719134 .035823 0.046**

_cons 144.6955 71.64356 0.045** R-squared = 0.4098 Adj R-squared = 0.3682

F( 11, 156) = 9.85 Prob > F = 0.0000*** * denotes statistical significance at 10%, ** at 5% and ***at the 1% level.

To improve results (efficiency) the external debt (DEBT) and population density variable (POP)

have been dropped because they were highly insignificant. The Specific pooled OLS model

regression results are presented in Table 5.6 next page.

Page 68: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

59

Table 5. 6: Specific Pooled OLS Model (Excluding DEBT And POP)

Dependent Variable BUOY

BUOY Coef. Std. Err. P>|t|

ECON .0055117 .0040867 0.179

AGR .0188142 .0071365 0.009***

IND .0219473 .0067639 0.001***

MS -.0043354 .001639 0.009***

AID -.7558855 .1211132 0.000***

DF -.0165895 .0064552 0.011**

XM .0056362 .0044746 0.210

EXP .8653563 .3776645 0.023**

Trend -.074938 .035562 0.037**

_cons 150.7422 71.12115 0.036** R-squared = 0.4063 Adj R-squared = 0.3725

F( 9, 158) = 12.01 Prob > F = 0.0000*** * denotes statistical significance at 10%, ** at 5% and ***at the 1% level

5.6 DISCUSSION OF RESULTS

The F statistic 12.01 (0.0000***) shows that the model is correctly specified and that the null

hypothesis of variable inclusion is rejected at the 1% level of significance and we therefore

conclude that at least one of the variables in the model explain the magnitude of annual tax

buoyancy in SADC economies.

The coefficient of monetization (MS) has a negative value and significant at the 1% level

indicating that growth of money supply (M2) seems to negatively affect the tax buoyancy of

SADC states. The results are not in line with the tax handle theory which poses for a positive

sign. This means that the growth of money supply does not facilitate the documentation of the

economy so as to improve tax administration. The reasons why monetization has a negative

influence on tax buoyancy in the SADC region might be due to the lack of capacity within the

tax administrators to take advantage of the growing supply of money to facilitate the tax

collection of each tax, over relying on printing money to finance government activities rather

than generating revenue elsewhere, the presence of distortions such as trade barriers, weak legal

and financial systems. Some SADC financial markets are not well developed (Johannesburg

Stock Exchange and Zimbabwe Stock Exchange are the most developed, but with no derivative

markets). The region has a narrow range of intermediaries and offers a limited number of

Page 69: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

60

financial instruments. This finding is not in line with the result obtained by Quazi (1994) who

found a positive and significant effect of monetization and tax buoyancy for a sample of 35

developing countries in a period of 10 years. Begum (2007) obtained a positive significant

coefficient for Bangladesh which is a developing nation, the reason being that there is utilization

of growing money supply to facilitate the documentation of the economy.

The coefficient of growth of the agricultural sector (AGR) is .0188142, with a p-value of 0.009

showing that the coefficient of domestic investment was positive and significant at 1% level.

Thus countries that are able to maintain and improve their agricultural sector will experience an

increase in their tax performance. This result is in line with the result obtained by Teera (2000)

who found a positive impact of growth of the agricultural sector and tax performance in a panel

of low income countries using pooled OLS, fixed effects model and random effects model.

However this opposes what Quazi (1994) obtained, he found a negative sign but insignificant.

Growth of the industrial sector (IND) has a positive and significant impact on tax buoyancy at all

levels of significance. This shows that it is one of the major variables that explain how the tax

system is performing in developing nations. This is in line with the tax handle theory which

predicts a positive impact. The possible reason for such results is that the sector is easy to tax,

companies keep records of transactions and many are located in urban areas which reduces costs

of tax collection. Industry includes mining companies which are very large and also few and

hence easy to monitor and audit for tax payment. The result obtained is similar to the result

obtained by Quazi (1994) in his study of the determinants of tax buoyancy. However the results

conflict with what Teera (2000) obtained, he found a negative relationship significant at all

levels for all three models of panel analysis. Begum (2007) obtained conflicting results as of this

study, the reason being that the sector is not well established. However he then separated direct

and indirect tax buoyancy, and direct tax buoyancy yielded a significant industrial sector growth

coefficient, the reason being that direct taxes are more common in industrial sector than indirect

taxes.

External aid (AID) has a negative and significant impact on tax buoyancy at all levels, indicating

a major variable. Economic theory (tax handle) predicts a negative impact, which is in line with

Page 70: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

61

the results. The findings in this study suggest that high levels of external aid growth in SADC

have influenced the tax performance of developing nations negatively. The reason may come

because an increase in foreign resources makes governments in the developing nations relaxed

and due to fear of any political unpopularity the governments rely less on domestic resource

mobilization. The results are in line with those of Quazi (1994), who found also that it was a

major variable significant at all levels with the appropriate sign.

Fiscal deficit (DF) variable is significant at 5% with a negative coefficient of -.0165895 and a p-

value of 0.011. The sign is not the expected, and this shows that as the deficit grows big it

demotivates the respective authorities from improving their taxing strategies to raise revenue

from taxes. This is not in line with Quazi (1994), who found a positive significant relationship of

the fiscal deficit and tax buoyancy at the same level of significance (5%).

The time trend coefficient (Trend) is significant at 5% with a negative sign. Implication is that

tax performance is changing over time due to the presence of shocks. The variable explains the

presence of time effects and hence economic shocks during the period under study. The possible

reasons for such results can be explained by the presence of droughts, civil wars and political

instability during the period10. The pooled OLS and random effects models by Teera (2000)

reported also a negative sign for low income countries but were insignificant. However for the

fixed effects model a positive sign was reported for the same study, but was also not significant.

For the SSA countries category the trend was negative and significant for both the fixed effects

and random models. The pooled OLS shows also a negative but insignificant for SSA

economies.

Total expenditure (EXP) is significant at 5% significant level and reports a correct positive sign.

This is in line with economic theory. The results for this variable are in line with what Teera

(2000) obtained. The results indicate that as the total spending increases, this causes the tax

system to be more buoyant, this is due to an extra effort to collect more revenue through taxation

to finance the increasing spending.

10 Countries like Zimbabwe and Mozambique have experienced droughts. Civil wars have been severe in DRC and also Kenya. Political instability have been seen mostly in Zimbabwe (1998 through 2005) and DRC, Kenya to some extend. For remaining nations, there is evidence that they have experienced the same shocks but at a lesser extend.

Page 71: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

62

However some variables have been found not explaining the dependent variable of tax buoyancy,

and they include trade openness (XM) and economic development (ECON). The variables have

also been used in several studies and have been yielding various impacts.

The economic development variable (ECON) has the correct positive sign but is insignificant.

This is in line with Teera (2000), in his random effects model estimation. However the pooled

and fixed effect model show significance, but fixed model showed a negative impact. The

implication of this is that the way the national income in developing nations grows does not

facilitate adequate tax collection. The rate at which income grows is less than the growth in tax

revenue.

Trade openness (XM) is also another variable which is not significant but has the correct sign, it

has a probability value of 0.210. A positive sign was also obtained by Teera (2000) but

significant at 5 % and 10% for low income group and SSA countries respectively. The possible

reasons for such results include the undervaluation of imported goods which applies to most

own-funds imports (Fjelstad, 1995). This is due to the fact that the importer has access to foreign

exchange without going through central bank records. Administrative constraints and corruption

at entry points increase the problem of undervaluation of imported goods (Basu and Morrissey,

1993: 22)

The overall model reported Adjusted R-squared of 0.3725. This tells us that approximately

37.25% variation in the annual tax buoyancy is explained by the explanatory variables included

in the model. The obtained Adjusted R-squared by Begum (2007) reports a value 0.51 for total

tax buoyancy regression and 0.46 for indirect tax buoyancy regression using pooled OLS, no

justification was given for such results. It automatically reports to us that there are some

variables that explain buoyancy that have been omitted. Some variables have been omitted such

as the shadow variable (tax evasion) due to the problem of measuring it. According to Teera

(2000), it is an indicator of the hidden economy and its sign is ambiguous, countries with large

hidden economies should be characterized by low tax revenues. If the shadow variable is

positive, people are holding more money than they would be expected to do given the

Page 72: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

63

characteristics of the economy. Pyle (1989) points out that one of the implications of existence of

the underground economy is that some income goes untaxed and also certain indirect taxes are

evaded. Naturally, this leads to the conclusion that tax revenues are lower than if everyone had

paid their taxes.

To add more, there are many factors often sited by many researchers11 that cause tax evasion in

developing nations. They include imperfect laws which leave loopholes for manipulations,

insufficient penalties leading to increase in the desire to avoid tax, inequity which also increase

the desire to evade taxes, low income levels creating a tendency of not reporting by low income

earners, probability of being detected is very low and also high tax rates.

5.7 CONCLUSION

In this chapter we have estimated and interpreted the regression results which have indicated that

monetization seem to have negatively affected the tax buoyancy of SADC countries and the main

possible reason being that the developing nations abuse the printing of money to finance

government activities despite the effects it exerts in the overall economy. Our major findings

seem to reject the hypothesis that monetization increases tax buoyancy. The hypothesis answered

by the study concern growth of industrial sector, growth of agricultural sector, total expenditure

growth and external aid growth. Apart from monetization variable, growth of fiscal deficit

(among the significant variables) has been found not to answer the hypothesis set for it. The

study managed to achieve its objectives of establishing the determinants of tax buoyancy in

developing nations and they include the growth of agricultural sector and industrial sector,

external aid, fiscal deficit and growth in expenditure. These findings are the basis for policy

prescriptions to be outlined in the next chapter.

11 Some researchers include Alm and Martinez (2007), Minorov (2006) and Eltony (2002) among others.

Page 73: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

64

CHAPTER SIX

CONCLUSION AND POLICY RECOMMENDATION

6.0 INTRODUCTION

This chapter contains a detailed conclusion to the study and also some policy lessons drawn from

the empirical results of the previous chapter. In addition, the chapter also gives possible areas of

future research.

6.1 CONCLUSION

The study has attempted to examine empirically the determinants of tax buoyancy in the

developing nations using SADC countries information for the period 1994-2005. The motivation

of the study sort to address the neglected quality dimension of tax performance leading to biased

policies, low tax revenue collection against potential revenue that can be raised from national

income, deterioration and shortage of public goods, continuous changes in tax rates that are also

high as compared to developed nations leading to the rapid expansion of the hidden economy,

persistent fiscal deficits and high growing rates of money supply.

Using a panel data, pooled OLS methodology we found out that monetization negatively affects

tax buoyancy. This result suggests that monetization seem not to increase tax performance,

instead it retards buoyancy levels. Possible reasons of this result may be due to the inability to

utilize the growth of money supply to increase the documentation of the economies to increase

the collection of each tax. The abuse of the printing of discretionary paper money to finance

government activities in developing nations is a possible reason for such results. This reduces

effort to raise adequate revenue from collecting taxes. Poor tax administration and also

underdeveloped infrastructure may be possible reasons.

Page 74: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

65

Apart from monetization, the study found that growth of agricultural sector, growth of industrial

sector, external aid growth, fiscal deficit growth and total expenditure growth to be the

determinants. The agricultural sector is still contributing positively to revenue generation

processes despite the industrialization going on in developing nations. The industrial sector is

worth to be noted, it contributes significantly to tax performance. Respective countries can

increase their tax buoyancy by actively encouraging growth in the agricultural and industrial

sectors; they should not undermine one sector since both have positive impacts. Fiscal deficit

growth and external aid growth have indicated a negative impact on tax performance.

Impact (negative) of economic shocks to tax performance has been found in the study. Economic

shocks have been shown by the presence of time effects and their significance in the regressions

undertaken. The main causes have been severe droughts due to inadequate rainfall, civil wars and

political instability.

6.2 POLICY RECOMMENDATION

Since our findings suggest a negative relationship between monetization and tax buoyancy,

maybe for SADC countries to reap benefits from monetization they need to have strong link

between fiscal and monetary sectors, improve tax administration through appropriate reforms,

invest in the improvement of infrastructure to facilitate the collection of taxes and stable

macroeconomic environment. A strong link between fiscal and monetary sectors facilitates the

utilization of the growth in money supply to improve tax performance through increased

documentation of the economy. Improvement in tax administration will produce efficient taxing

systems which discourages tax evasion and the growth of the hidden economy. Most of all there

should be central bank independence from political authorities, the government activities should

be financed not from discretionary paper money printing rather they should use other non-

inflationary ways such as tax collection until the potential level of the economy has been

reached.

Policies aimed at developing the domestic taxing systems are beneficial. The policies should be

aiming at taking special considerations of the findings in the study. Factors aimed at taking

Page 75: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

66

appropriate decisions on variables like fiscal deficits and external deficits since they have been

found to have a negative impact on tax buoyancy. In some SADC countries there are a few

numbers of financial intermediaries and stock markets are not well developed, an effort to

improve their significance will be an appropriate measure to be undertaken.

Development policies should not be biased towards the growth of one sector rather it has to be

across all sectors. Both the agricultural sector and the industrial sectors have proved to be of

significance in defining the level of tax performance, and hence they have to be treated without

disparity.

Developing nations should however consider maintaining their taxing systems or improve them

positively over time. The time trend variable have shown us that over time there is negative

effect on tax buoyancy, there is actually a decline and this is to be prevented. Policies aimed at

avoiding declining tax performance over time should be developed. Economic shocks like

drought should be avoided through forecasting and mitigation strategies developed.

6.3 FUTURE AREAS OF RESEARCH

Earlier studies on tax performance have been using the tax ratio as the indicator of tax

performance. The tax ratio is only a quantity measure of tax performance. This study attempted

to improve on this by using tax buoyancy which is both a quantitative and a qualitative measure

of tax performance. The study failed to separate the quality dimension to stand alone, this

involves a study of the determinants of tax elasticity. The study would have been more useful if

the study could find the determinants of the elasticity of these taxes, but due to non-availability

on the discretionary measures of each tax this was not feasible. So this is an ongoing area which

needs attention. The main shortfall of the study is data availability in current years; data could

not be collected to date.

Page 76: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

67

REFERENCES Ahmad, S. E. U. and N. Stern (1991), “Tax Reform and Shadow Prices for Pakistan.” The

Development Economics Research Program, London School of Economics.

Allen, D.S. and Ndikumana L.(1998). “Financial Intermediation and Economic Growth in

Southern Africa.” Working Paper Series 1998-004, the Federal Reserve Bank of ST. Louis.

Alm and Martinez (2007), “Tax Morale and Tax Evasion in Latin America,” International

Studies Program. Working Paper 07-04. March.

Andersen, S.P (1973), “Built-in Flexibility of Sensitivity of the Personal Income Tax in

Denmark,” Swedish Journal of Economics, vol. 75, no.3:34-98

Ansari, M. M (1982), “Determinants of Tax Ratio: A Cross-Country Analysis,” Economic and

Political weekly, June 19, pp. 1035-1042.

Ariyo Ademola (1997), “Productivity of the Nigerian Tax System: 1970–1990.” Department of

Economics University of Ibadan, Ibadan, NIGERIA. AERC Research Paper 67, African

Economic Research Consortium, Nairobi, November.

Basu, Priya and Oliver Morrissey. (1993), “The fiscal impact of adjustment in Tanzania in the

1980s.” (Draft). Nottingham: CREDIT/University of Nottingham.

Baum Christopher F (2006), “An introduction to Modern Econometrics using stata,” Old Square

School.Masachusetts.Stata Press

Begum Lutfunnahar (2007), “A Panel Study on Tax Effort and Tax Buoyancy with Special

Reference to Bangladesh. Policy Analysis Unit(PAU). Working Paper Series: WP 0715.

Research Department, Bangladesh Bank. Bangladesh.

Page 77: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

68

Bird, R.M. (1974), “Assessing Tax Performance in Developing Countries: A Critical Review of

the Literature,” in Taxation and Economic Development ed. By Toye, J.F.J., Frank Cass and

Company Ltd. London, pp. 33-57.

Bird and Zolt (2003), “Introduction to Tax Policy Design and Development. Practical Issues of

Tax Policy in Developing Countries,” World Bank, April 28- May 1.

Bruderl Joseph (2000), “Applied regression analysis using Stata” UCL press. London

Breusch, T. S., and A. R. Pagan (1979), “A Simple Test for Heteroskedasticity and Random

Coeficient Variation,” Econometrica 50, 987–1007.

Burgess, R and Stern, N (1993), “Taxation and Development,” Journal of Economic Literature,

31, pp762-830.

Byrne, P.D. (1995), “Privatization in the area of tax administration: an overview.” Development

Discussion Papers (DDP) 508: Taxation Research Series No. 24 (April). The Harvard Institute

for International Development. Harvard University. Cambridge, MA.

Cameron C. and Trivedi K. (2005), “Microeconomics. Methods and Applications.” Cambridge

University Press. Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São

Paulo.

Chelliah, R.J. (1971), “Trends in Taxation in Developing Countries,” IMF Staff Papers, Vol. 18,

No. 2, July, pp. 254-325.

Chelliah, R.J., Baas, H.J., and Kelly, M.R. (1975), “Tax Ratios and Tax Effort in Developing

Countries, 1969-71” IMF staff Papers, Vol. 22, No. 1, March, pp.187-205.

Chiminya A. (2008). “An Empirical Analysis of the Determinants of Tax Revenue Share in

Zimbabwe(1975-2005).” MSc Thesis. University of Zimbabwe.

Page 78: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

69

Chipeta C. (1998), “Tax reform and tax yield in Malawi.” Southern African Institute for

Economic Research Zomba, Malawi. AERC Research Paper 81. African Economic Research

Consortium, Nairobi March .

Choudhry, N. N. (1975), “Measuring the Elasticity of Tax Revenue: A Divisia Index Approach,”

IMF Staff Paper, vol. 36 pp 87-122.

Choudhry, N. (1979), “The Measurement of Tax Revenue Elasticity,” IMF Staff Papers, vol. 26,

March.pp 124-187.

Devas, N., Delay, S. and Hubbard, M. (2001), “Revenue Authorities: Are they the right vehicle

for improved tax administration?” Public Administration and Development, Vol. 21, pp. 211-

222.

Darlauf S.N. and Quah. (1998), “The new empirics of economic growth.” Discussion Paper

No.384. Center for Economic Performance.

Ebrill, Liam Stocky and Gropp R. (2001), “The Modern VAT,” (Washington: International

Monetary Fund).

Eltony M. Nagy (2002), “The Determinants of Tax Effort in Arab Countries,” API Publications,

Kuwait.

FIAS/DFID. (2006), “Sector study on the effective tax burden. Rwanda.” Washington D.C.:

Foreign Investment Advisory Service, International Finance Corporation/The World Bank.

FIAS/DFID. (2006b), “Sector study on the effective tax burden. Lesotho.” Washington D.C.:

Foreign Investment Advisory Service, International Finance Corporation/The World Bank.

Page 79: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

70

Fjelstad, O-H. (1995), “Taxation and Tax reforms in Tanzania: A Survey.” Working paper

1995:4. Chr Michelsen Institute, Bergen, November .

Fjeldstad, O.-H. (2003). “Fighting fiscal corruption. Lessons from the Tanzania Revenue

Authority.” Public Administration and Development, Vol. 23, No. 2, pp. 165-175 (May).

Fjeldstad, O.-H. and M. Moore (2008), “Tax reform and state building in a globalized world.”

Chapter 10 (pp. 235-260) in D. Bräutigam, O.-H. Fjeldstad and M. Moore (eds.) Taxation and

State Building Developing Countries. Cambridge: Cambridge University Press.

Ghura, D. (1998), “Tax revenue in Sub-Saharan Africa: Effects of economic policies and

corruption.” IMF Working Paper 98/135, Washington, DC: International Monetary Fund.

Franco-Rodriguez, S., Morrissey, O., And Mcgillivray, M. (1998), “Aid and The Public Sector in

Pakistan: evidence with Endogenous Aid,” World Development, Vol. 26, No. 7, July, pp. 1241-

1250.

Government Financial Statistics. International monetary Fund, 2006.

Greenaway David (1981), “Tax revenue instability, with particular reference to Sub-Saharan

Africa.” Journal of Development Studies, VoL. 31, No. 6, pp. 883-902.

Harley, H. Hinrichs (1965), “Determinants of Governments Revenue Shares among Less-

developed Countries.” The Economic Journal 75. Vol 15.pp 127-154.

Hindriks Jean, Casanegra J. and Fernando L. (1999), “Corruption, Extortion and Evasion.”

Journal of Public Economics. Vol. 74. No.5.pp 234-46

Hsiao, Cheng, (1986), “Analysis of Panel Data,” (Cambridge University Press).

International Monetary Fund, Government Financial Statistics, 2001.

Page 80: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

71

International Financial Statistics, International Monetary Fund, Government Financial Statistics,

2005.

International Monetary Fund, Government Financial Statistics, 2006.

Isabelle and Christophe (2008), “Tax Incidence Analysis of Developing Countries: An

Alternative View,” World Bank Economic Review, 5: 533-52.

Islam N. (1995). “Growth Empirics: A Panel Data Approach.” Quarterly Journal of Economics,

Vol.110 No.4 pp.1127-70.

Jayasundera. 1991. Buoyancy and Elasticity of Taxes. Report of the Taxation Commission,

1990, Sessional Paper No. 1, chapter 8. Colombo, Sri Lanka.

Jenkins (1994), “Modernization of Tax Administration in Low-Income Countries: the Case of

Nepal.” Consulting Assistance on Economic Reform II Discussion Paper 68, Harvard Institute

for International Development.

Joweria M. (2000), “Tax Performance: A Comparative Study.” University of Bath, Department

of Economics, BATH BA2 7AY.

Khan, M.Z. (1973), “Responsiveness of Tax Yields to Increase in National Income,” Pakistan

Development Review, vol. xii, no. 4.

Kiser, E. and A. Sacks. (2007), “Improving tax administration in contemporary African states.

Lessons from history,” Paper presented at the conference on “The Thunder of History: Taxation

in Comparative and Historical Perspective.” Northwestern University, May 4-5.

Koutsoyannis A. (1988), “Theory of Econometrics,” Harper & Row, New York.

Page 81: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

72

Kusi, N.K. (1998), “An Examination of Ghana’s Stabilisation and Structural Adjustment

Programme,” 1983-88. Unpublished PhD thesis, University of New England.

Leuthold, J.H. (1991), “Tax Shares in Developing Economies: A Panel Study,” Journal of

development Economics, Vol. 35, No.1, January, pp. 173-185.

Leuthold, J., and T. N’Guessan. (1986), “Tax buoyancy and elasticity in a developing economy,”

Bureau of Economic and Business Research Faculty Working Paper no. 1272. University of

Illinois, Urbana Champaign.

Liebaman, E (2003), “Race and Regionalism in the Politics of Taxation in South Africa and

Brazil,” Cambridge: New York, Cambridge University Press

Linn F.J. and Weitzel D. (1990), “Public Finance, Trade and Development: What Have We

Learned?” in Fiscal Policy in Open Developing Economies, ed. By Tanzi, pp. 9-28.

Lotz, Jorgen R and Morss, (1967), Theory of Tax Level Determinants,” Economic Development

and Cultural Change, Vol 18, No.3, pp478-479

Manasan Rosario G (2003), “Tax Administration Reform: (Semi-) Autonomous Revenue

Authority Anyone?” Philipine Institute for Development Studies, Discussion Paper Series No.

2003-05 June, (Revised).

Mansfield, C. Y. (1972), “Elasticity and Buoyancy of a Tax System: A method applied to

Paraguay,” IMF Staff Papers Vol. 19.

Martinez-Vazquez, Jorge and Robert M. McNab, (2000), “The Tax Reform Experiment in

Transitional Countries,” National Tax Journal, 53: 273-98.

Marcelo Piancastelli (2001), “Measuring the Tax Effort of Developed and Developing Countries:

Cross Country Panel Data Analysis - 1985/95.” Institute of Applied Economic Research (IPEA) -

Page 82: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

73

Directory of Macroeconomic Policy & Studies (DIMAC). September. IPEA Working Paper No.

818

Matshediso I.B. (2004), “A review of mining tataxion in selected SADC countries and Chile.”

Botswana Journal of Technology Vol.13 No.2 pp. 41-50.

Milambo (2001), “Tax Incidence Analysis of Developing Countries: An Alternative View,”

World Bank Economic Review, 5: 533-52.

Minorov Maxim (2006), “Economics of Spacemen: Tax Evasion and Firm Performance.

Evidence from Russian Banking Transaction data,” Graduate School of Business, University of

Chicago.

Muriiti M. K and Moyi (2000), “Tax Reforms and Revenue Mobilization in Kenya,” AERC

Research Paper no. 131.

Musgrave, R.A. and Musgrave, P.B. (1984), “Public Finance in Theory and Practice,” McGraw-

Hill, New York.

Musgrave R.A. (1969), “Fiscal Systems,” Yale University Press, London, UK.

Omoruyi, S.E. (1983), “Growth and Flexibility of Federal Government Tax Revenue: 1960-

1979”. Economic and Financial Review, vol. 21, no.1: 11–19.

Osoro N. E. (1993). “Tax reform in Tanzania.” Paper presented at the CREDIT-CSAE

Workshop on trade and fiscal reforms in sub-Saharan Africa, Oxford, 6-8 January.

Prest, Allan. R. (1962), “The Sensitivity of the Yield of Personal Income Tax in the United

Kingdom,” The Economic Journal, Vol.52. September.pp.576-96

Page 83: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

74

Quazi Masood Ahmed (1994), “The Determinants of Tax Buoyancy: An Experience from the

Developing Countries.” The Pakistan Development Review. 33:4 Part II (Winter 1994) pp. 1089-

1098.

Raja J., Chelliah, Hessel J. Baas, and Margaret R. Kelly, (1971), “Tax Ratios and Tax Effort in

Developing Countries,” Staff Papers, International Monetary Fund, Vol. 2, pp. 187-205.

RED, (2004), “Integrated Paper on Recent Economic Development,” Bank of Mauritious.

September.

RED, (2006), “Integrated Paper on Recent Economic Development,” Bank of Namibia.

December 1.

Roy, Bahl W. (1979), “A regression Approach to tax Effort and Tax Ratio Analysis.” IMF staff

Papers, Vol.18 no.3 pp. 570-612.

Sohato, G.S. (1961), “Indian Tax Structure and Economic Development,” New York:

Asian Publishing House.

Saruchera Allen (2009), “ZIMRA Computerisation,” (Presentation). ZIMRA Information

Technology. May 25.

Seade (1990), “Taxing Ourselves: A Citizen’s Guide to the Great Debate Over Tax Reform,”

Cambridge, Massachusetts: MIT Press.

Silvani, Carlos and Katherine Baer (1997), “Designing a Tax Administration Reform Strategy:

Experiences and Guidelines.” IMF Working Paper 97/30.

Singer, N. M. (1968), “The Use of Dummy Variables in Establishing the Income Elasticity of

State Income Tax Revenue,” National Tax Journal vol. 61 pp 83-90.

Page 84: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

75

Slemrod, Joel. (1990), “Optimal Taxation and Optimal Tax Systems,” Journal of Economic

Perspectives, vol 4, no.1 pp.157-78.

Stephen Knack (2004), “Measuring & Monitoring Governance in Developing Countries,” The

World Bank, 2nd International Roundtable, Marrakesh, February.

Stotsky, J.G. and WoldeMariam, A. (1997), “Tax Effort in Sub-Saharan Africa,” IMF Working

Paper, WP/97/107, September.

Tait, Alan and Wildfrid Gratz, (1979), “International Comparisons of Taxation for Selected

Developing Countries, 1972-76,” Staff papers, IMF, Vol. 26, pp. 123-56.

Tait, Alan, and Barry J. Eichengreen, (1978), “Two Alternative Approaches to the International

Comparison of Taxation,” IMF Department Memorandum 78/73 (Washington: International

Monetary Fund).

Tanzi V. (1987), “Quantitative Characteristics of the Tax Systems of Developing Countires”, in

The Theory of Taxation for developing Countries, ed. By Newbery and Stern, Oxford University

Press, New York, pp. 205-241.

Tanzi, V. and Zee Howell, H. (2000), “Tax Policy for Emerging markets: Developing

Countries,” National Tax Journal, Vol. 53, No. 2 pp. 299-322.

Tanzi, Vito, 1981, “A Statistical Evaluation of Taxation in Sub-Sharan Africa,” in Taxation in

Sub-Saharan Africa (Washington: International Monetary Fund), pp.45-50.

Tanzi V. (1992), “Structural Factors and Tax Revenue in Developing Countries: A Decade of

Evidence”, in Open Economies: Structural Adjustment and Agriculture, ed. by Goldin, I. and

Winters, A.L., Cambridge University Press, Cambridge. pp. 267-281.

Page 85: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

76

Teera J. M (2000), “Tax Performance: A Comparative Study.” University of Bath, Department

of Economics, BATH BA2 7AY. Bath

Teera, J.M. (2002), “Determinants of Tax Revenue Share in Uganda,” University of Bath, UK.

Taliercio, R. (2004), “Administrative Reform as Credible Commitment: The Impact of

Autonomy on Revenue Authority Performance in Latin America.” World Development, Vol. 32

No. 2, pp. 213-32.

Taliercio, R. (2002), “Designing performance: The semi-autonomous revenue authority model

in Africa and Latin-America.” Washington, DC: The World Bank.

Therkildsen, O. (2004), “Autonomous Tax Administration in Sub-Saharan Africa: The Case of

the Uganda Revenue Authority.” Forum for Development Studies, vol.31 no.1 pp. 59-88.

Wawire Nelson Were H. (2000), “The Determinants of Tax Revenue in Kenya.” Kenyatta

University, Nairobi.

Wawire, N.H.W. (2003), “Trends in Kenya's Tax Ratios and Tax Effort Indices, and their

implications for Future Tax Reform,” In Emilia V. Ilieva (eds) Egerton Journal Vol. IV, No. 243

Pp.256-279 (ISSN 1021- 1128)

Wilford and Wilford (1978), “Tax Effort in Sub-Saharan Africa,” IMF WP/78/107 73

(Washington: International Monetary Fund).

World Bank(2001), “Lessons From Tax Reform,” Washington.

World Bank. (2006). World Development Indicators. Washington

Zolt Eric M. (2003), “Intergovernmental Fiscal Transfers: Lessons from International

Experience,” World Development.

Page 86: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

77

APPENDICES

APPENDIX I: STRUCTURE OF SADC CENTRAL GOVERNMENT

TAXATION

Structure of SADC central government taxation, 1994- 2005 (percentage of total taxes) 1994 1995 1996 1997 1998 1999

Taxes on income and wealth 36.2 29.1 27.8 18.8 31.0 26.5

PAYE

10.3 9.0 4.3 3.2 3.5 3.1

Individuals 2.4 2.4 2.0 1.9 2.5 2.6

Companies 21.6 15.7 19.0 13.7 22 18.0

Other 1.9 2.0 2.5 0.0 3.0 2.8

Taxes on domestic goods and

services

41.8 54.5 44.0 40.9 31.5 31.9

Sales and excise tax-local 40.2 51.2 37.4 40.9 29.4 29.2

Other 1.6 3.3 6.6 0.0 2.1 2.7

Taxes on international

transactions

22.0 16.4 28.2 28.7 16.1 15.3

Import duties 11.9 8.8 14.3 14.8 10.8 9.4

Sales and excise tax – imports 3.0 7.4 13.7 13.9 5.2 5.6

Other 7.1 0.2 0.2 0.0 0.1 0.3

Other taxes* 0.0 0.0 0.0 11.6 21.4 26.3

2000 2001 2002 2003 2004 2005

Taxes on income and wealth 26.6 40.9 41.2 40.7 42.3 38.9

PAYE

3.0 5.6 5.4 5.1 6.4 5.4

Individuals 2.4 4.8 5.2 6.3 7.3 7.5

Companies 19.4 26.9 26.8 26.0 25.2 22.8

Other 1.8 3.6 3.8 3.3 3.4 3.2

Taxes on domestic goods and

services

30.4 32.1 30.9 31.6 30.2 32.4

Sales and excise tax-local 28.1 26.8 27.8 27.9 26.4 27.0

Other 2.3 5.3 3.1 3.7 3.8 5.4

Taxes on international

transactions

14.1 14.7 14.6 13.8 13.4 14.1

Import duties 9.2 10.2 10.0 9.7 9.3 9.0

Sales and excise tax – imports 4.0 3.8 3.7 2.9 1.7 2.7

Other 0.9 0.7 0.9 1.2 2.4 2.4

Other taxes* 28.9 12.3 13.3 13.6 16.1 14.6

*Other taxes consist of motor vehicles taxes, miscellaneous taxes and property taxes. Source: World Development Indicators 2006 (various years), IMF financial statistics (various years)

Page 87: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

78

APPENDIX II: SUMMARY OF REVIEWED LITERATURE

DETERMINANT SUPPORTING THEORY EMPIRICAL LITERATURE

Monetization Tax handle

Ricardian Equivalence*

Quazi(1994), Begum(2007)

Trade Openness Tax handle Teera(2001),Begum(2007)

Agricultural Sector Tax handle Quazi(1994), Tanzi(1992)

Chelliah et al(1975)

Leuthold(1991)

Industrial Sector Tax handle Quazi(1994),

Chelliah et al(1975)

External Aid Tax handle Quazi(1994)

External Debt Tax handle Quazi(1994), Begum(2007),

Tanzi(1992)

Population Density Tax handle Teera(2001), Begum(2007)

Fiscal Deficit Tax handle Quazi(1994),

Level of Economic

Development

Tax handle

Optimal tax theory

Quazi(1994), Tanzi(1987)

Tanzi(1992)

Total Expenditure Tax handle Teera(2001)

* The link is not direct but through transmission mechanism

Page 88: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

79

APPENDIX III: SPECIFICATION TESTS

(A) MULTICOLLINEARITY TEST

(B) Summary of specification test TEST CRITICAL

VALUE

P- VALUE DECISION IMPLICATION

F-test F(13,144)= 0.89 0.5690 Accept H0 Pooled OLS

Hausman Test Chi2(10) = 7.51 0.6765 Accept H0 Random effects

LM Test Chi2(1) = 0.36 0.5511 Accept H0 Pooled OLS

Country Effects Test F(13,133) = 1.20 0.2829 Accept H0 No Country effects

Time Effects Test F(11,133) = 4.59 0.0000 Reject H0 Time Effects

Joint Effects Test F(24,133) = 2.72 0.0002 Reject H0 Combined effects

present

(C) COUNTRY EFFECTS TEST

Page 89: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

80

(D) TIME EFFECTS TEST

(E) JOINT TEST FOR COUNTRY AND TIME EFFECTS

(F) FIXED EFFECTS TEST (F- test)

Page 90: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

81

APPENDIX IV: STATA OUTPUT FOR DATA ANALYSIS

Page 91: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

82

Page 92: DETERMINANTS TAX BUOYANCY IN DEVELOPING NATIONS

83