INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4552

www.rsisinternational.org

The Effect of Tax Revenue on Unemployment and Income in

Developing Countries

Ohiomoje Iyemifokhae Abubakar

Department Of Economics, Faculty of Economics, Management and Social Sciences, University of

Ibadan, Ibadan, Oyo State,Nigeria,

DOI:

https://dx.doi.org/10.47772/IJRISS.2025.910000377

Received: 07 October 2025; Accepted: 13 October 2025; Published: 12 November 2025

ABSTRACTS

The need for developing countries to create the enabling environment for rapid economic development put them

in situation where they have to ensure that more households and firms are inclusive in their tax net. There have

been a growing debate among researchers on the effect of tax on unemployment and income in the emerging and

developed economies. While the findings from such studies showed positive, inverse relationships or their mix

among these variables in the developed economies, the experience of developing countries with regard to tax

revenue, unemployment and income is yet to be examined. This study aims to fill this vacuum in the literature.

The study made use of the fully modified ordinary least squares technique to evaluate this relationship. The

findings indicated that there was decreasing relationships between tax revenue and unemployment while, in the

case of income, this was indeterminable, given its insignificantly positive estimate in developing countries.

Keywords: Tax Revenue, Unemployment, Income, Developing Countries, Fully Modified Ordinary Least

Squares.

INTRODUCTION

Unemployment and low gross domestic product (GDP) per capita are some of the basic problems of developing

countries and the quest to escape from these challenges have been a daunting task to policy makers in these

economies. Compared to the developed economies, most developing countries are confronted with the shortage

of skilled labor and advanced technology which are essential for development in modern industrial processes.

The relative scarcity of these factors engenders low level of per capita income in developing countries and

hinders them from reaching their full potentials in terms of industrial development. This problem is orchestrated

by misguided economic policies, insecure property rights and political instability. The debates on the solutions

to overcome unemployment and low income in developing countries have been fierce among economists.

Among the arrays of economic policies prescribed by this group of social scientists is an effective and efficient

tax revenue drive. What are the probable effects of tax revenue on unemployment and income in developing

countries? A large number of empirical studies have shed light on this concern. Most of the evidence on the

impact of tax revenue on labour market have focused on understanding the increasing wage gap between skilled

and unskilled workers in the North (Freeman and Katz, 1991; Gaston and Trefler, 1997; Goldberg and Pavcnick,

2007). Infact, there were only few studies that focused on tax revenue and unemployment in developing countries

(Epifani, 2003; Hoekman and Winters, 2005; Saba Arbache, 2001). Given the diversified revenue base in

developing countries, a study of this nature that examines the influence of tax revenue on unemployment and

income in these countries is of great importance because of its potentially much larger impacts.

The need to expand social and economic infrastructures is a desideratum for developing countries to be on the

path of rapid economic development. However, the trend in most developing countries is worrisome as they are

unable to raise the targeted revenues from taxation to uplift their dilapidated state of infrastructures. It has been

observed that a large percentage of residents and firms in such economies evade paying the necessary taxes and

even those that do, underpay the required amounts to the government (Aizenman, J. and Jinjarak, Y., 2009).

Besides, given the statistics that, in developing countries, over 70% of the labour force work in the agricultural

sector and that they operate mostly on subsistence basis, the collection of direct taxes on their incomes is difficult

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4553

www.rsisinternational.org

because of the herculean task of engaging each of these subsistence farmers. According to Aizenman and Jinjarak

(2009), trade taxes are easier to collect and account for a major share of tax revenue in developing countries.

Tax is a compulsory levy imposed by the government on income, consumption, profits, capital gains and transfers

of citizens and firms that thrive in such a country. This can come in the form of direct and indirect taxes. The

former (direct tax) is a levy on the incomes of individuals and profits of firms. In the case of indirect taxation,

the imposition is done on goods and services. Examples of this are excise duties and the value added tax on

consumption items. These taxes are expected to generate income for the provision of economic and social

infrastructures which will create the enabling environment for the prevalence and growth of economic activities.

Most of the countries in Sub Saharan Africa (SSA) are considered to be underdeveloped. Some countries in Asia

and Latin America fall under this group. Developing countries differ widely among themselves and no single

list of features unilaterally describes them. In the 1960s, these countries were much more similar to one another

in their approaches to macroeconomic interventions. Soon, things began to change. East Asian countries

embraced export-oriented development strategy, abandoning import-substitution as a panacea to industrial

development. This strategy recorded greater successes. Also, some countries in Latin America adopted reduction

in trade barriers like tariffs and canvassed for more government roles in economic processes. The outcome of

this approach was mixed. Some developing countries like those of Sub Saharan Africa reformed their economic

policies in similar fashion with those of the successful industrial economies. While few of these economies (of

developing countries) were successful in such attempt, the overall assessment of most of their performances was

awful as their economies were still characterized by low gross domestic product per capita, high unemployment

rate and under-employment, dominance of agricultural produce in their exports, low production of industrial

output and high inflation rate.

The Background of the Study

Figure 1 depicts the aggregate revenue generated from taxation in some Sub-Saharan-African countries. The

taxes are represented in the vertical axis while the identifications of the samples and the series are shown on the

horizontal axis.

Countries

Total Taxes

Source: United Nations Government Revenue Dataset, 2021.

The maximum tax revenue was obtained by Burkina Faso (C_id 2) with a figure of $72,923,435 in 2017. This

was seconded by Cameroon (C_id 5) with a tax revenue of $66,444,688 in 2013. The revenue from tax in the

list of countries was at its lowest in Ghana (C_id 8) with a figure of $2276.11 in 2014.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4554

www.rsisinternational.org

Figure 2 represents the gross domestic product (GDP) per capita. This is used to represent income. The GDP per

capita is shown on the vertical axis and the horizontal axis reflects the countries that are examined in the study

with the sample period. The GDP per capita is the gross domestic product divided by the population of a country.

It is usually used to measure welfare across different countries. This metric was at its maximum in Equitorial

Guinea (C_id 9) with a score of $22,834 in 2008. The second country on the sequence was Botswana (C_id 3)

which recorded $7609.17 in 2012. Morocco (C_id 11) had the least GDP per capita of $311.54 in 2010.

Countries

Income (Gross Domestic Product (GDP) per capita)

Source: World Development Indicators Database, 2020.

The figure 3 shows the amount of the value added tax (VAT) that were generated by the samples in the period

under consideration. As can be seen from the graph, the revenue obtained through VAT was generally low and

non-existent in some of the samples.

Countries

Value Added Tax

Source: United Nations University (UNU-WIDER) Dataset, 2021.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4555

www.rsisinternational.org

The value added tax was highest in Congo with a revenue of $970,633.08 in 2013. Cameroon came next in the

ranking with a revenue of $958,153.08 in 2017. This was followed by Nigeria with a VAT of $953,663.78 in

2017.

One of the constraints confronting researchers in this area of study was the dearth of data on employed labour in

the sample of countries. Even when such data was available, it was randomly given. In other words, there was a

great vacuum of omitted data on employment. In figure 4, the Y-axis represents the annual growth in employment

and the X-axis depicts the sample period and countries.

Countries

Annual Employment Growth Rate

Source: International Monetary Fund, World Economic Outlook Database, World Bank Enterprise Survey Data,

2021.

Given the shortcomings in employment data, this study considers an indirect approach as a panacea to this

problem by examining unemployment data in the sample. This is represented in Figure 5 as shown below.

Annual Unemployment Growth Rate

Source: International Monetary Fund, 2020.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4556

www.rsisinternational.org

In the graph above, the growth rate of unemployment was at its peak in Botswana (C_iD 2) at 21.77% in 2004.

Congo (C_iD 5) had an unemployment rate of 19.94% in 2005. Other countries with relatively higher

unemployment rate were Egypt (C_iD 6) at 13.05% in 2015 and Mali (C_iD 11) at 11.71% in 2007. The growth

rates of unemployment were generally low in Cote d’voire (C_iD 3), Kenya (C_iD 9), Cameroon (C_iD 4) and

Nigeria (C_iD 12) during the period under consideration.

Tax Administration and Governance in Sub-Saharan Africa

Information on tax administration and governance in Sub-Saharan Africa (SSA) is important not just as an

academic exercise but also to provide fiscal authorities with robust compendium to proffer solutions to macro-

economic challenges. The capacity of the countries in SSA to generate the required taxes to counter such

problems reflects in the quality of governance. Quite often than not, there has been improvements in the amount

of tax that has been generated over-time, there is still gap between the quality of governance and the expectations

of the citizens.

Many countries in Sub-Saharan Africa establish quasi-independent tax authorities with the staff of these

institutions being remunerated more than their counterparts in other departments and agencies in the civil service.

Several reforms have been carried out in tax administration to re-evaluate tax assessments, collection and

outsourcing in Sub-Saharan Africa. However, fiscal implementations have been facing brick-wall in SSA’s

countries. Budgets are perceived to be better made than implemented. Some of the challenges thwarting efficient

implementation of fiscal policies include corruption, political instability, poor forecasting ability, macro-

economic shocks and poor quality data.

The quality of social and economic infrastructures that are provided in most of the countries in Sub-Saharan

Africa has leave less to be desired. This inadequacies calls for the need to come up with efficient tax drive that

will generate the requisite finance to put the economies of Sub-Saharan African countries on the road map to

increase employment and income.

LITERATURE REVIEW

Theoretical Review

In the word of Keynes (1936) in his General theory, employment can be increased by increasing the level of

consumption and investment. Fiscal measure like taxation is one of the vital instruments that can be used to

regulate consumption and investment. Consumption depends on income. For instance, when the fiscal authorities

want to regulate consumption, they adjust the tax system (i.e direct tax on income and profits). In the face of

decreasing consumption, the direct taxes may be reduced in order to increase the disposable income available to

economic agents. When income rises, consumption also increases but not as much as the former. Also,

investment can be jerked up by a fall in the rate of interest and a rise in the marginal efficiency of capital which

depends on the cost of capital and its yields. Given the constraints of increasing some of the components of these

variables (consumption and investment) to increase employment in the short run, the under-employment

equilibrium is what is attainable in such a system. The observed behavior of the consumption function cited

earlier widens the gap between income and consumption which basically cannot be filled due to insufficient

investment. This makes the quest of attaining full employment unrealistic. The need for the adoption of effective

tax system to counter this investment shortcoming to achieve full employment in developing countries is a

desideratum.

However, Professor Schlesinger considered Keynes treatment of the aggregate demand to be defective.

According to the critic, the relative prices in the different sectors of the economy partly determine supplies. This

view was shared by Professor Don Patinkin. In his criticism, the 45 percent line adopted by Keynes infers that

demand creates its own supply. This is a fallacy and neglect of the relevance of the supply side in determining

employment equilibrium. Other critics rejected Keynes proposition of the consumption function because it was

deficient of other vital components like wealth effect, technological change, expectations and attitude towards

assets. Also, Keynes proposition of the investment function met brick wall as it failed to show how this

transcended to technological progress. Furthermore, his theory on interest rate was short-lived as Keynes failed

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4557

www.rsisinternational.org

to incorporate the real forces that determined interest rate. Interest rate is not only a function of the demand and

supply of money but also other kind of assets. The General Theory offered solutions in countering cyclical

unemployment only while it neglected frictional and technological unemployment.

Capital formation is of strategic importance for developing countries to jump-start their economies and

experience rapid economic transformation. It is therefore necessary to achieve a higher ratio of savings to

national income. The state must rely on the instrument of fiscal policies to actualize this ratio. However, there is

an inherent dislike for direct regulation by the state by entrepreneurs viz what to be produced or price fixing.

Similarly, most consumers would not like to be physically ordered on their consumption pattern. The taxation

of articles whose consumption should be discouraged is therefore preferable. This strategy also applies to the

entrepreneurs in the form of tax incentives and increased tax on non-essential products. A coherent tax system

is a vital tool of raising the ratio of savings to the national income which is a sine-qua-non for developing

countries to attain a high level of economic emancipation.

A constant savings-income ratio was a fundamental assumption of the Harod-Domar (1956). The proponents of

the model like Kaldor and Pasinetti considered the savings-income ratio as a variable in the growth process. The

hypothesis was that there were two classes of workers – the wage earners and the profit earners whose savings

were functions of their income. However, the propensity to save of the profit earners was higher than the wage

earners. Thus, the aggregate savings of the community depended on the distribution of income. The classical

savings function was given as

S = sp.

𝜋

⁄𝛾

where sp = propensity to save;

𝜋

⁄𝛾 = the ratio of profit to national income.

If the hypothesis of the propensity to save between the wage earners and the profit earners was maintained (in

which the latter’s propensity was higher than the former), steady state of growth will be the outcome.

The propensity to save and the capital-output ratio of the Harod-Domar model were criticized because they were

assumed to be constant which is not true in real life. This altered the outcome of steady state in the model. Also,

the assumption that capital and workers were used in fixed proportion was a mirage. Furthermore, inflation

consideration was not taken into account in the model. Besides, the model ignored government interventions like

fiscal and monetary policies.

The Solow-Swan model was a neo-classical model that examined the long run growth of output based on the

growth of the population and the rate of technological progress. In the word of Romer (1990), this model

discounted the relevance of government in economic stabilization. The endogenous growth model was

formulated as a reaction of the shortcomings of the Solow-Swan model. The new growth model extended the

work of the latter by introducing endogenous technological progress emanating from the rate of investment, the

size of capital stock and the stock of human capital resulting from the creation of new ideas. According to this

model, knowledge and technological progress were more important than natural endowments. This may be a

major cause of underdevelopment of most developing countries. However, this model was criticized as it

neglected institutional reform programmes like taxation.

Taxation is an instrument of the fiscal policy measures to achieve economic stabilisation or growth. The

government may truncate undesirable levels in private consumption and entrepreneurial investments by adjusting

its tax structure. The end result of such policy is to achieve macroeconomic stability like full employment, growth

in output, high GDP per capita, amongst others. According to Jan Tinbergen (1952), the number of policy

instruments must equalised the number of objectives, otherwise the system would be under-determined if the

instruments fell short of the objectives or over-determined if vice-versa. This

Tinbergens’s proposition was needful for economic policies to be successful. Based on this, Robert Mundell

(1968) stated that in countries where employment and balance of payment policies were restricted to monetary

and fiscal instruments, monetary policy should be reserved for external balance viz balance of payment situation

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4558

www.rsisinternational.org

while the fiscal policy like taxation should be assigned for internal balance like employment. These two policies

would bring the economy to simultaneous equilibrium in the internal and external balances.

However, there was controversy whether monetary or fiscal authorities should carry out their policies based on

pre-determined rules that put into template how their policy directives will be determined in all future economic

situations or whether they should deploy the use of discretion in determining the policy variables as the situations

evolve. According to the Keynesian, the economy was subject to several vagaries caused by changes in

expectation, political events, oil crisis and war. These exogenous shocks led to cyclical fluctuations and this

resulted to uncertainties of varying intensity which may affect employment and income. This situation was

aggravated by the inherently unstable private sector in areas like employment generation, output and income.

Thus, Keynesians proposed that government should be aggressive in its fiscal policy drive to stabilize the

economy.

Economists like Lucas do not agree with the Keynesians. According to him in his rational expectation theory, it

is only unexpected changes in policies that led to changes in real GDP. He assumed that consumers and firms

had accurate information about future economic events like government actions and they adjusted their plans

appropriately. This made government policies to be ineffective. In this regard, in the face of active fiscal policy

actions by the government to counter unemployment, a tax cut or increased in government expenditure to solve

the unemployment situation would be effective if its short run effects on the economy were unanticipated by

economic agents. If this was on the contrary, when these agents expected price hike with such a policy, workers

would demand for more wages while firms raised the price of their goods to offset their cost, given the

expectation of increased wages. Consequently, the government good fiscal intentions would be made non-

effective. The Rational Expectation Hypothesis (REH) was criticised because the rational economic agents did

not have perfect knowledge of economic activities to counter fiscal policy actions of governments.

Empirical and Methodological Reviews

Empirics on this literature found that tax revenue was correlated with reductions in employment or rather increase

in unemployment, although the nature of this impacts differed in-between studies. In some studies, tax revenue

reduce unemployment while in others, the reverse was the case. In most studies, a greater number of statistically

significant coefficients suggested positive relationship between tax revenue and different strata of unemployment

(Marquez and Pages, 1997; Jaramillo and Tovar, 2006).

In models with increasing returns to scale, the impact of tax revenue on labour reallocation and income showed

that tax reduction increased agglomeration in specific geographic clusters and sectors (Melitz, 2003). It depended

on the size of the trade policy shock which included the size of the initial tariffs, the extent of tariffs reduction,

the structure of trade partners and other complementary macro-economic policies.

Ebrill et. al (1999) stated that the main effects of tax reductions on revenue collection in developing countries

was the price elasticity of import demand. If import demand was price inelastic, reduction in tariffs had smaller

effects on the quantities imported which tantamount to lower tax receipts. On the other hand, when the elasticity

of import was price elastic, reduction in tariffs translated to larger quantities of goods being imported. With

regards to unemployment, tax revenue could affect this concern but this was a function of the size of public

employment which was usually small in developing countries (Rama, 1999).

Muendler (2007) analysed the impact of tax revenue on labour reallocation in Brazil. He found that reduction in

tariffs led to reduction in unemployment in this country.

Cobham (2005) assumed that income generated in the shadow economy (in developing countries) enhanced the

tax gap at the national level in these economies (as this income was taxed) if effective tax auditing was possible.

This outcome was vehemently criticized by Auriol and Warlters (2005). They argued that many developing

countries deliberately created cost of entry into their formal economies in order to reduce competition. In this

direction, they created economic rent which was later taxed. According to these critics, this was because the size

of the shadow economies was endogenous. If this endogeneity was taken into consideration, the negative

correlation between the ratio of the tax revenue and the size of the shadow economies would disappear.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4559

www.rsisinternational.org

In terms of methodological review, most studies on this literature made use of the revenue/gross domestic product

(GDP) data provided by the International Monetary Fund (IMF).

Ahlerup and Bigsten (2015) made use of cross country panel data to examine the revenue impact of value added

tax (VAT) on Sub-Saharan African (SSA) countries.

Oliver et. al (2016) used pooled ordinary least squares, fixed effects and random effects with country specific

quadratic time trend (to allow for heterogeneity) to examine tax revenue performance in developing countries.

The standard errors were clustered by countries to control for serial correlation and heterogeneity within each

country.

Afuberoh and Okoye (2014) studied the impact of tax revenue on the gross domestic product in some sections

of Nigeria, namely Abuja, kogi State, Delta Sate, Niger State, Ebonyi State and Ondo State. The study made use

of primary data and estimated the regressors on the regressand using the SPSS 17.0.

Neway et. al (2018) examined the determinant of tax revenue in Ethiopia. The study made use of the multiple

regression analysis.

Velaji and Prendi (2014) studied the relationship between tax revenue and unemployment, inflation, the gross

domestic product and import in Ethiopia using the Pearson correlation and regression analysis.

METHODOLOGY

This study makes use of the panel data analysis and the fully modified ordinary least square technique in its

analysis. The former enables us to understand more complex behavioral patterns in the model while the latter

analysis is needed to explain the dynamic. To set the stage, the variables of the cross-sectional units in this study

are specified in the following model.

= α

+ β

+ µ

…………………………………………………………………. Eq (1)

Where T represents total taxes, α depicts the intercepts of the countries in the sample, β

is the vector of

coefficient estimates showing the relationship between the regressors and the regressand. UE is

unemployment variable, I is income,

i = 1, ……………, 13 where i is the number of countries in the sample, t

= 1, ……………., 13 where t is the time covered in the sample.

Equation (1) denotes the pooled effect regression in which all the cross sectional units or countries have

homogenous characteristics. Given the disparity in the socio-cultural environment of countries, the fixed effect

model is inculcated in the framework in which differential intercepts dummy variables are introduced like

= α

+ α

+,…………, + α

13i

+ β

+ µ

……………………………. Eq (2)

Where α

, ……………, α

13i

represents the different intercepts or heterogenous features across the various

countries in the sample.

Since only thirteen countries are examined in the sample, we only introduce twelve dummy variables in equation

(2) to avoid the dummy trap. Thus, the intercept of the first country is treated as the reference intercept category

while other intercept coefficients show the extent to which they differ from the former or base intercept category.

In equation (3), rather than treating α

as fixed, we hypothesize that it is a random variable with a mean value of

, having no subscript i. The intercept value for each country is expressed as

= α

+ 𝜀

…………………………………………………………………………………..Eq (3)

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4560

www.rsisinternational.org

where 𝜀

is a random variable with a mean value of zero and variance of 𝜎

𝜀

. The individual differences in the

intercept of each country is reflected in the random variable.

Substituting equation (3) into equation (1) gives

= α

+ β

+ µ

……………………………………………………………… Eq (4)

Equation (4) translates to

= α

+ β

+ w

……………………………………………………………. Eq (5)

Where w

= 𝜀

+ µ

………………………………………………………………………………. Eq (6)

consists of 𝜀

which is the individual-specific error component of the countries in the sample and µ

which is

the combination of time series and cross-sectional error components called error components model (ECM). The

assumption of the ECM is that the individual error components are not correlated with each other and are not

auto-correlated both in the cross-sections and time series units.

The Description of Variables and Sources of Data

The constituents of total taxes are taxes on income, profits and capital gains, taxes on goods and services, taxes

on international trade and transactions, the value added tax and other taxes. This information is obtained from

the United Nations University database. The gross domestic products per capita of the different countries in the

sample are used to define their respective income and this is measured in the US dollar. This data is obtained

from the world development indicator (WDI). The data on unemployment’s variable is sourced from the

international Monetary Fund. The sample period was between 2004 and 2017. The countries that were examined

in the study were Bourkina Faso, Botswana, Cote d’voire, Cameroon, Congo, Egypt, Ghana, Equitorial Guinea,

Kenya, Morocco, Mali, Nigeria and Senegal.

RESULT AND DISCUSSION

The panel data analysis is used to examine the relationship between tax revenue, unemployment and income in

developing countries. This is carried out with the pooled ordinary least square regression, fixed effect and the

random effect estimations. The pooled regression assumes that the cross sectional units have homogenous

characteristics. In other words, the sub-sectional units have a common intercept. The fixed effects regression

allows the cross sectional units to retain their unique features or varying intercepts but this should be time-

invariant. In the case of the random effect regression, the intercept of the cross subsections have a common mean

value, however the differences in the intercepts of the different units or sample countries are reflected in their

disturbance terms.

In appendix 1 of the pooled regression, a percentage increase in the annual growth rate of unemployment leads

to an increase in the tax revenue of 238811.0 in the cross-sectional units of the sample, all other things being

equal. This estimate is significant at less than 10% level of significance. This implies that there is a positive and

significant relationship between tax revenue and the growth rate of unemployment in the sample. Also, a percent

point rise in the gross domestic product per capita (which depicts income) leads to a negative growth of 95.47 in

the revenue from taxation in the sampled countries. However, this result is insignificant. Generally, the

assumption of pooled regression of homogenous characteristics of the cross sectional units of the sample is

unrealistic as the different countries vary in their economic, social and cultural attributes.

In appendix 2 of the fixed effect regression, all the countries (in the sample) have an intercept of 4,559,529 and

this estimate is insignificant at less than 10% probability level. A percentage increase in the annual growth rate

of unemployment in the sample results in 295,814.3 decline in the income generated from taxation by

governments, ceteris paribus. Thus, there is an inverse relationship between unemployment and tax revenue in

the group. Furthermore, if the gross domestic product per capita rises by a percentage, tax revenue increases by

94.50 percent. The results of the fixed effect regression vis-a-vis the coefficients of the unemployment growth

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4561

www.rsisinternational.org

rates and income (with respect to the total tax revenues) are insignificants in the cross sectional units.

In appendix 3 which considers the random effect regression, all the countries examined in this study have a mean

intercept of 5,730,008 and this estimate is significant at less than 10% probability level. If the annual growth rate

of unemployment rises by a unit, the revenue derived from the aggregate tax declines by

313,178.7. This reveals an inverse relationship between tax revenue and unemployment. Also, a percentage rise

in the GDP per capita (income) leads to a decline in tax revenue by 140.05 percent. Thus, there is negative

relationship between the gross domestic product per capita (which signifies income) and tax revenue. Like in the

case of the fixed effect regression, the estimates of the coefficients of unemployment’s growth rates and income

are insignificant, given their probability values.

To examine which of the two effects (fixed or random effect regression) to select in order to analyse the

relationship (if any) between tax revenue, unemployment and income, the Hausman test is conducted. The result

of this estimate in appendix 4 shows that the probability of Chi square statistics is 0.8547 which is insignificant.

This implies that the study rejects the null hypothesis that presupposes that the random effect is appropriate and

adopts the fixed effect estimates. In the fixed effect regression, there is positive correlation (94.50) between

income (GDP per capita) and tax revenue. When employees experience increased tax on their income as well as

consumption (indirect tax), they increase their number of man-hour in a bid to smoothen their consumption

pattern. Also, if the annual growth rate of unemployment soars by a unit, tax revenue reduces by 295,814.3 units.

It is important to note that all the estimates from the fixed regression are insignificants. This insinuates that there

are unobserved heterogeneity and measurement errors in the series, using the panel data analysis.

Given this shortcomings, the study analyses the relationship between tax revenue, employment and income in

developing countries by utilising the fully modified ordinary least squares. This technique solves the problem of

serial correlation and endogeneity. Table 5 reveals the Pedroni residual cointegration estimates for the individual

intercepts and the probability values of the eleven statistics (Panel V-Statistics, Panel Rho-statistics, Panel PP-

statistics, the Weighted Statistics of the listed statistics as well as their Group Statistics). Most of the values of

these listed statistics fall below the 5% probability threshold. Similar results are also obtained for the individual

intercepts and trend deterministic. Given these results, the study rejects the null hypothesis which states that

there is no co-integration among the variables. The Table 6 in the appendices corroborated this finding in the

Kao Residual Cointegration test. The probability value of the Augumented Dickey Fuller Test of the Residual

Variance and HAC Variance is below 5%. The Durbin Watson (DW) estimates of 1.93 reveals stationarity of

the series.

Since there is co-integration in the variables, the study examines the relationship between tax revenue,

unemployment and income by using the fully modified ordinary least squares technique. Table 7 (in the

appendices) shows that after correcting measurement and statistical errors in the analysis, a unit increase in

income generates 25,557.18 in tax revenues in the cross-sectional countries in the sample. The probability value

of this estimate is above the 10% threshold indicating its insignificance. However, a percentage increase in the

growth rates of unemployment in the sample leads to a decline of tax revenue by 3,789,372 in the cross-sectional

units. This estimate is significant as it falls below the 10% benchmark. Thus, there is negative relationship

between unemployment and tax revenue in developing countries. This indirectly suggests that there is the dearth

of virile private sector in developing countries. In other words, if there is reduction in taxation in developing

countries, the expectation is that this should translate to more investment, output and employment in developing

countries which results in fall in unemployment. However, this is not the case in the examined countries as the

estimates reveal inverse relationship between unemployment and tax revenues, suggesting that there is the lack

of enabling environment for the growth of the private sector and businesses in the sampled countries. Another

possibility is that the elasticity of demand of consumers with regard to taxable commodities and/or services is

highly elastic. In this case, such economic agents can evade the burden of taxation by drastically reducing the

quantities purchased or not consuming them at all. Such behavior and response adversely affect supplies,

investment, output, unemployment and income. Thus, there should be proper assessment of the elasticity of

demand of taxable commodities and services before such taxes are initiated in order to achieve desirable

macroeconomic outcomes like increases in income and reduction in unemployment level.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4562

www.rsisinternational.org

CONCLUSIONS AND RECOMMENDATIONS

Given the inverse relationship between tax revenue and unemployment in developing countries, these economies

should create the enabling environment for increased capacity utilization of the employers of labor (firms and

industries). Expansionary fiscal measures like reduction in taxation and increased spending in infrastructures

should be promoted. These policies will increase the employment of labor which is a desirable macroeconomic

objective in any economy. Additionally, fiscal authorities should endeavor to impose tax levies on goods and/or

services whose demand are inelastic. This reduces tax evasion from the intended tax payers. Such initiative

generates increased investment, reduction in unemployment, increased output and income. The empiric of the

study also showed that there is direct relationship between tax revenue and income in developing countries. This

is a progressive tax system in which more tax is imposed as income increases. Fiscal authorities should desist

from tax measures which are regressive in nature. The value added tax, for instance, is regressive in nature in

which both the poor and non-poor are subjected to the same rate. This could be a dis-incentive to the former

group of workforce (in terms of man-hour put in production) as it depletes their disposable income relative to

the latter group. Generally, most indirect taxes are regressive. Developing countries need to adopt interventionary

policies that will boost local production, increase disposable income and open their economies such that they

can leverage on the benefits of comparative advantage. Also, if policy makers can get it right in setting the right

tariffs on imports, promote exports through the initiation of appropriate incentives and institutionalise

progressive tax system in developing countries, these policies engender south-ward movement in the growth rate

of unemployment and causes the income level of individuals and profit of firms to soar in such economies.

Overtime, most firms (in developing countries) that do not have the stamina to compete with their foreign

counterparts will overcome such challenge, given these fiscal policy measures. Furthermore, the revenue

generated from tax should be channeled in economic developmental projects that are capable of enhancing

domestic production of goods and service, increase the employment of labour and income. The practice of

squandering tax revenue on activities of little or no economic value should be discouraged. There should be

fiscal discipline, transparency and accountability of tax resources.

Limitation of the Study

Initially, the study intended to examine the relationship between tax revenue, employment and income in

developing countries. This plan was limited by the number of observations of some of the variables of the model.

There was dearth of data on the employment variable across the sample. Moreover, it was observed that there

was a large presence of omitted data in the sample. This makes it difficult to conduct further statistical checks

on the estimates to counter some of this problem. Given the challenge of obtaining robust employment data in

the sample during the period under review, the study explored unemployment data in the cross section.

REFERENCES

1. Afuberoh Dennis and Okoye Emmanuel (2014). The Impact of Tax Revenue Generation in Nigeria: A

Study of Federal Capital Territory and Selected States. International Journal of Public Administration and

2. Management Reasearch, Vol. 2 (2): 22-47.

3. Aizenman, J. and Jinjarak, Y. (2009). Globalisation and Developing Countries- A Shrinking Tax Base?

Journal of Development Studies, Vol. 45 ( ): 653-671.

4. Auriol, Emmanuelle and Michel Warlters (2005). Taxation Base in Developing Countries. Journal of

Public Economies, Vol. 89 ( ): 625-646.

5. Cobham, Alex (2005). Tax Evasion, Tax Avoidance and Development Finance. Queen Elizabeth House

Working Paper No. 129.

6. Ebrill, L., Gropp, R., Stotsky J., (1999). Revenue Implications of Trade Liberalisation. IMF: Washington

DC.

7. Gupta, Abhijit (2007). Determinants of Tax Revenue Efforts in Developing Countries. IMF Working Paper

07/184.

8. Haldenwang, C. and Ivanyna, M. (2012). A Comparative view of Tax Performance of Developing

Countries: Regional Patterns, Non-Tax Revenue and Governance. The Open Access, Open-Assessment

EJournal, Vol. 6: 2012-2032.

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4563

www.rsisinternational.org

9. Jaramillo and Tovar (2006). Trade and Labour Market: Evidence from the Colombia Trade

Liberalisation Process. Bogota: Universidad de los Anales, CEDE.

10. Marques G. and Pages, C. (1997). Trade and Employment. Evidence from Latin America and the

Carribean. Washington DC: Inter-American Development Bank, Office of the Chief Economist.

11. Melitz, M. (2013). The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry

Productivity. Econometrica 71: 1696-1725.

12. Mkandawire, T. (2010). On Tax Efforts and Colonial Heritage in Africa. Journal of Development Studies,

46: 1647-1669.

13. Moore, M. (2008). Between Coercion and Contract: Competitive Narratives around Taxation and

Governance. In D. Brautigan, H. Fjeldstad and M., Moore (Eds.). Taxation and State Building in

14. Developing Countries: Capacity and Consent. Page 34-63, Cambridge University Press, Cambridge, M.

15. Muendler (2007). Trade and Workforce Change-over in Brazil. Discussion Paper 2001-01, Department of

Economics, University of California, San Diego.

16. Neway Gobachew, Kenenisa Lewis and Woldemicael Shibiru (2018). Determinants of Tax Revenue in

Ethiopia. Science Publishing Group.

17. Oliver Morrisey, Christian Haldenwang, Armin Von Schiller, Maksym Ivanyna and Ingo Bordon (2016).

18. Revenue Performance and Vulnerability in Developing Countries. Journal of Development Studies, Vol.

15 (2): 1689-1703.

19. Rama, M. (1999). Public Sector Downsizing: An Introduction. World Bank Economic Review, Vol. 13

(1); 1-22.

20. Robert Mundell (1968). The Appropriate Use of Monetary and Fiscal Policies for Internal and External

Stability. IMF Staff Papers, International Economics.

21. Romer, Paul (1990). Increasing Return and Long Run Growth. Journal of PE, Endogenous Technological

Change.

22. Tinbergen, J. (1952). On the Theory of Economic Policies.

23. Velaji, E. and Prendi, L. (2014). Tax Revenue- The Determinant Factor- the case of Albania. European

Scientific Journal, Volume 1 (ISSN: 1857-7881).

APPENDICES

Appendix 1

Dependent Variable: TOTAL_TAXES

Method: Panel Least Squares

Sample: 2004-2017

Periods included: 13

Cross-sections included: 12

Total panel (unbalanced) observations: 154

Variable

Coefficient

Std. Error

t-Statistic

Prob.

ANNUAL_UNEMPLOYMENT_GROWTH_RATE

238811.0

112423.6 2.124207

0.0353

INCOME__GROSS_DOMESTIC_PRODUCT

__GDP__PER_CAPITA_

-95.47168

176.1284 -0.542057

0.5886

R-squared

-0.048385

Mean dependent var

2626288.

Adjusted R-squared

-0.055282

S.D. dependent var

8910767.

S.E. of regression

9153759.

Akaike info criterion

34.91013

Sum squared resid

1.27E+16

Schwarz criterion

34.94957

Log likelihood

-2686.080

Hannan-Quinn criter.

34.92615

Durbin-Watson stat

1.595540

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4564

www.rsisinternational.org

Source: Author’s calculation.

APPENDIX 2

Dependent Variable: TOTAL_TAXES

Method: Panel Least Squares

Date: 02/19/22 Time: 02:44

Sample (adjusted): 2004 2017

Periods included: 14

Cross-sections included: 12

Total panel (unbalanced) observations: 154

Variable

Coefficient

Std. Error

t-Statistic

Prob.

4559529.

5026402.

0.90711

0.3659

ANNUAL_UNEMPLOYMENT_GROWTH_RA

-295814.3

631280.9

-0.468594

0.6401

INCOME__GROSS_DOMESTIC_PRODUCT

__GDP__PER_CAPITA_

94.50028

478.0809

0.197666

0.8436

Effects Specification

Cross-section fixed (dummy variables)

R-squared

0.149615

Mean dependent var

2626288.

Adjusted R-squared

0.070651

S.D. dependent var

8910767.

S.E. of regression

8590225.

Akaike info criterion

34.85666

Sum squared resid

1.03E+16

Schwarz criterion

35.13274

Log likelihood

-2669.963

Hannan-Quinn criter.

34.96880

F-statistic

1.894721

Durbin-Watson stat

1.971616

Prob(F-statistic)

0.035417

Source: Author’s computation.

Appendix 3

Dependent Variable: TOTAL_TAXES

Method: Panel EGLS (Cross-section random effects)

Sample (adjusted): 2004-2017

Periods included: 14

Cross-sections included: 12

Total panel (unbalanced) observations: 154

Swamy and Arora estimator of component variances

Variable

Coefficient

Std. Error

t-Statistic

Prob.

5730008.

2161798.

2.650574

0.0089

ANNUAL_UNEMPLOYMENT_GROWTH_RA

-313178.7

257999.3

-1.213874

0.2267

INCOME__GROSS_DOMESTIC_PRODUCT

-140.0501

230.6429

-0.607216

0.5446

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4565

www.rsisinternational.org

__GDP__PER_CAPITA_

Effects Specification

S.D.

Rho

Cross-section random

2642125.

0.0864

Idiosyncratic random

8590225.

0.9136

Weighted Statistics

R-squared

0.015781

Mean dependent

var

1830487.

Adjusted R-squared

0.002745

S.D. dependent var

8652508.

S.E. of regression

8619156.

Sum squared resid

1.12E+16

F-statistic

1.210605

Durbin-Watson stat

1.816350

Prob(F-statistic)

0.300892

Unweighted

Statistics

R-squared

0.026177

Mean dependent var

2626288.

Sum squared resid

1.18E+16

Durbin-Watson stat

1.722282

Source: Author’s Calculation.

Appendix 4

Correlated Random Effects - Hausman Test

Equation: Untitled

Test cross-section random effects

Test Summary

Chi -Sq.

Statistic Chi-Sq. d.f.

Prob.

Cross-section random

0.314070

0.8547

Cross-section random effects test comparisons:

Variable

Fixed

Random Var(Diff.)

Prob.

- -

Annual_Unemployment_Growth_Ra 295814.2897 313178.74017 33195192192 Te 26 0

4.31492 0.9760 Income__Gross_Domestic_Product 175365.24248

__Gdp__Per_Capita_ 94.500275 -140.050089 6 0.5754

Cross-section random effects test equation:

Dependent Variable: TOTAL_TAXES

Method: Panel Least Squares

Date: 02/19/22 Time: 02:50

Sample (adjusted): 2004 2017

Periods included: 14

Cross-sections included: 12

Total panel (unbalanced) observations: 154

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4566

www.rsisinternational.org

Variable

Coefficient

Std. Error

t-Statistic

Prob.

4559529.

5026402.

0.907116

0.3659

ANNUAL_UNEMPLOYMENT_GROWTH_RA

-295814.3

631280.9

-0.468594

0.6401

INCOME__GROSS_DOMESTIC_PRODUCT

__GDP__PER_CAPITA_

94.50028

478.0809

0.197666

0.8436

Effects Specifi

cation

Cross-section fixed (dummy variables)

R-squared

0.149615

Mean dependent var

2626288.

Adjusted R-squared

0.070651

S.D. dependent var

8910767.

S.E. of regression

8590225.

Akaike info criterion

34.85666

Sum squared resid

1.03E+16

Schwarz criterion

35.13274

Log likelihood

-2669.963

Hannan-Quinn criter.

34.96880

F-statistic

1.894721

Durbin-Watson stat

1.971616

Prob(F-statistic)

0.035417

Source: Author’s Calculation.

Appendix 5

Pedroni Residual Cointegration Test

Series: TOTAL_TAXES INCOME__GROSS_DOMESTIC_PRODUCT__GD

P__PER_CAPITA_ ANNUAL_UNEMPLOYMENT_GROWTH_RATE

Sample: 2000-2017

Included observations: 186

Cross-sections included: 11 (2 dropped)

Null Hypothesis: No cointegration

Trend assumption: No deterministic trend

User-specified lag length: 1

Newey-West automatic bandwidth selection and Bartlett kernel

Alternative hypothesis: common AR coefs. (within

-dimension)

Weighted

Statistic Prob. Statistic Prob.

Panel v-Statistic -0.210337 0.5833 -2.268016 0.9883

Panel rho-Statistic -1.716037 0.0431 -1.279651 0.1003

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4567

www.rsisinternational.org

Panel PP-Statistic -8.050991 0.0000 -7.578026 0.0000

Panel ADF-Statistic -3.213306 0.0007 -2.162200 0.0153

Alternative hypothesis: individual AR coefs. (between-dimension)

Statistic Prob.

Group rho-Statistic 0.373085 0.6455

Group PP-Statistic -8.899417 0.0000

Group ADF-Statistic -3.354920 0.0004

Cross section specific results

Phillips-Peron results (non

-parametric)

Cross ID AR(1) Variance HAC

Bandwidth

Obs

1 Dropped from Test

2 -0.439 4.66E+09 3.59E+09

3.00

3 -0.113 6.89E+13 5.94E+13

2.00

4 -0.242 2.44E+14 1.04E+14

4.00

5 Dro

6 -0.326

pped from Te

1.54E+11

2.03E+10

11.00

1.022

32151585

52776116

2.00

0.383

1.12E+11

9.52E+10

3.00

-0.230

6.39E+12

3.08E+12

4.00

-0.127

5.73E+11

0.00

-0.105

9.84E+13

8.68E+13

2.00

-0.085

3.27E+11

0.00

-0.167

4.13E+12

1.06E+12

5.00

Augmented Dickey-

Fuller resul

ts (parametric)

Cross ID

AR(1)

Variance Lag

Max lag

Obs

Dro

pped from Test

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4568

www.rsisinternational.org

-0.505

4.80E+08 1

-0.308

7.35E+13 1

-0.502

2.53E+14 1

Dro

pped from Test

-0.726

1.51E+11 1

0.843

26846592 1

0.234

1.12E+11 1

-0.660

6.37E+12 1

-0.244

6.20E+11 1

-0.255

1.05E+14 1

-0.128

3.12E+11 1

-0.440

4.22E+12 1

Source

: Author’s Calculation

Appendix 6

Kao Residual Cointegration Test

Series: TOTAL_TAXES INCOME__GROSS_DOMESTIC_PRODUCT__GDP__PER_CAPITA_

ANNUAL_UNEMPLOYMENT_GROWTH_RATE

Sample: 2000-2017 Included observations: 186 Null Hypothesis: No

cointegration

Trend assumption: No deterministic trend

User-specified lag length: 1

Newey-West automatic bandwidth selection and Bartlett kernel

ADF

t-Statistic

Prob.

-3.188628

0.0007

Residual variance

1.96E+14

HAC variance

1.12E+14

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(RESID)

Method: Least Squares

Date: 02/18/22 Time: 11:18

Sample (adjusted): 2006 2017

Included observations: 130 after adjustments

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue X October 2025

Page 4569

www.rsisinternational.org

Variable

Coefficient

Std. Error

t-Statistic

Prob.

RESID(

-1)

-1.282607

0.147416

-8.700581

0.0000

D(RESID(-1))

0.137742

0.104259

1.321159

0.1888

R-squared

0.48842

Mean dependent var

519350.8

Adjusted R-squared

0.484427

S.D. dependent var

12081236

S.E. of regression

8674742.

Akaike info criterion

34.80499

Sum squared resid

9.63E+15

Schwarz criterion

34.84911

Log likelihood

-2260.325

Hannan-Quinn criter.

34.82292

Durbin-Watson stat

1.926893

Source: Author’s computation.

Appendix 7

Dependent Variable: TOTAL_TAXES

Method: Panel Fully Modified Least Squares (FMOLS)

Sample (adjusted): 2005-2017

Periods included: 13

Cross-sections included: 12

Total panel (unbalanced) observations: 145

Panel method: Grouped estimation

Cointegrating equation deterministics: C

Long-run covariance estimates (Bartlett kernel, Newey-West fixed bandwidth)

Variable

Coefficient

Std. Error

t-Statistic

Prob.

INCOME__GROSS_DOMESTIC_PRODUCT__GDP__PER_C

APITA_

25557.18

17318.40

1.475724

0.1424

ANNUAL_UNEMPLOYMENT_GROWTH_RATE

-3789372.

2042929.

-1.854872

0.0659

R-squared

-192.137247

Mean dependent var

2748608.

Adjusted R-squared -211.303539 S.D.

dependent var 9169718. S.E. of regression 1.34E+08 Sum squared resid 2.34E+18 Long-run

variance 5.32E+13

Source: Author’s calculation.