DETERMINANTS OF ECONOMIC GROWTH

A PANEL DATA OF 60 COUNTRIES

Thomas Mwebaze

March 2002

Introduction

This paper is based on the study by William Easterly and Sergio Rebelo; Fiscal policy and Economic growth, an empirical investigation. The study is an extension of the work by Robert Barro. We use a panel of annual data for 60 countries from 1960 to 1996 instead of a cross section data.

Easterly and Rebelo state that fiscal policy is an important growth determinant.

In their study, using cross section data of 100 countries, Easterly and Rebelo came up

with the following findings;

- Government consumption is robustly correlated with growth

- Government revenue is also consistently correlated with growth and private investment.

According to Easterly and Rebelo, high levels of inequality in income distribution observed prior to 1970 were associated with higher levels of publicly provided education in the period from 1970 –1988. They also come to a conclusion that high public spending on infrastructure investment raises growth.

The results of Easterly and Rebolo are presented in table 1 and 2 below.

The t-statistics are given in parentheses.

Table 2.

Partial Correlation between fiscal aggregates and Growth

The coefficients reported in table 2 above are all significant.

Literature Review

Empirical findings of Barro Robert J in August 1990 strongly support the general notion of conditional convergence. That for given starting level of real per capita GDP, the growth rate is enhanced by higher initial schooling and life expectancy, lower fertility, lower government consumption, better maintenance of rule of law, low inflation and improvement in the terms of trade. For given values of these and other variables, growth is negatively related to the initial level of real per capita GDP. Political freedom has only a weak effect on growth but there is some indication of a non-linear relation.

In the initial phase, growth in developing countries is hindered primarily owing to the lack of the basic economic foundations in the shape of the growth of environmental facilities and services. Their growth demands considerable investments in social and economic overheads. This is the initial phase of the process of development where the government of an under developed country is required to create the basic economic foundation for development by direct investment in social and economic overheads

According to Barro, higher initial per capita GDP is substantially negatively related to subsequent per capita growth. Per capita growth is positively related to the proxies for the initial human capital, holding fixed GDP60 and the other variables.

The idea that government policy is important in shaping the growth process has been a recurrent theme of the development literature for the past three decades (see for instance Krueger (1978) and Chenery (1986)). Growth theorists have also shared the interest in the role of public policy and the neoclassical growth model of Solow (1956) has been used as a testing ground to study the dynamic effects of numerous policies.

Under the revolutionary impact of Keynes’ book published in 1936, professional economists shifted their interest to an aggregate analysis of the economy. Keynes’ theory of national income and employment emphasized the fact that government’s fiscal operations could be administered so as to influence the level of income and employment. In addition to the aim of securing collective consumption, taxes could be imposed to curtail effect ive demand, and expenditure could be incurred in order to increase investment, income, and employment. Compensatory financing was placed at the core of the new macroeconomics. The budgetary weapons –taxes, borrowing and expenditure, -were analysed in the light of their effectiveness in promoting full employment with stable prices. In this way, the modern concept of fiscal policy as an instrument of economic growth, emerged.

The Keynesian exposition was a short-run analysis of the capital formation process. It took account of the income –increasing effects of investment. Later developments saw a shift from the Keynesian short-run emphasis to the post –Keynesian growth analysis of the Harrod Domar type. These developments considered both the short-term incomes creating effects and the long –run capacity –creating aspects of investment. In the final analysis, capital formation was assigned a decisive role in development process.

By the early 1950’s the Keynesian prescription of government intervention in the nation’s economic activities in order to promote growth received wide acceptance. Keynesian analysis with its later refinements was not applicable to an underdeveloped country because it had been cast in terms of the conditions existing in advanced countrie s. The policy measures, which flowed from the analysis, would yield different results when applied to a developing country. Suitable changes in economic priorities and in fiscal policy measures are required for countries at the stage of development.

Simon Kuznets (1952) observes that there is enormous variety in the different growth processes in the world.

For which reason, where there’s formulating or evaluating theoretical assumptions as to economic growth, any attempt to go beyond mere lists of factors would have to be made, bearing in mind the characteristics of countries and of the periods relating to the growth process it is intended to explain.

The determinants of growth methodology initiated by Moses Abramovitiz (1956) and Robert M Solow (1956) re lies heavily on the classical conception of an aggregate production function.. Solow’s (1956) growth theory, in its stripped –down model, asks the question: What are the determinants of sustainable growth of real income per person in an isolated economy producing a single product? This simple model assumes that the state of technology does not change over time, and that the labour does not grow, and capital stock does not depreciate.

Ricardo demonstrated that the ultimate outcome of the growth process rested up three basic propositions: (i) the Malthusian law of population, which held that population unless checked by disease, famine or war tends to expand at an exponential rate. (ii) The fundamental economic principle of diminishing return especially as applied to the scarce resource of agricultural land .

(iii) Theory of capital accumulation in which profit is a key variable.

The great importance of fiscal policy in many developing countries arises from the fact that the state, under democratic auspices, is called upon to play an active and important role in promoting economic growth. For various reasons, the governments of all developing countries virtually have been forced to play this role. In order to do this effectively, they have to interfere in the economic life of the country, control and regulate economic activities, and compel or induce people to behave in different ways at the cost of worsening its disincentive effects.

The only feature that is most common to most developing countries is the shortage of revenue, which makes it impossible for them to provide essential public services on the required scale. The inadequacy of public revenue has two important consequences. It forces undue economies precisely in those fields of public expenditure (like health and education) which are more easily sacrificed in the short run but are the most important from the point of view of long-run development. It also yields persistent budgetary deficits, which force the monetary authorities to follow highly restrictive credit policies (to protect the balance of payments and to limit the pace of inflation), which in turn has highly undesirable effects on the pace of economic growth.

Those who believe that insufficient growth and investment is mainly a consequence of a lack of resources are chiefly concerned with increasing the resources available for investment through additional taxation, even at the cost of worsening its disincentive effects. The only feature that is most common to most developing countries is the shortage of revenue which makes it impossible for them to provide essential public services on the required scale. The inadequacy of public revenue has two important consequences. It forces undue economies precisely in those fields of public expenditure (like health and education) which are more easily sacrificed in the short run but are the most important from the point of view of long-run development. It also yields persistent budgetary deficits, which force the monetary authorities to follow highly restrictive credit policies (to protect the balance of payments and to limit the pace of inflation), which in turn has highly undesirable effects on the pace of economic growth.

The resurgence of interest in growth theory may be linked to the twin contributions of Romer (1986) and Lucas (1988), who redirected the traditional focus on the accumulation of physical capital as a transitory source of growth to the accumulation of skills through education and training. Building on the foundations that Solow had established, these theorists incorporated technological change and imperfect competition to explain economic growth. One of the predictions of the neoclassical growth theory is that the growth rate of real income per capita of countries with identical preferences, technologies and population growth will converge over time due to higher rates of physical investment in the initially poorer countries, as capital moves in search of the higher rates of return.

A key assumption, following Lucas (1988), is that technology in each country is virtually identical; differences in income and productivity are thus entirely attributable to differences in relative factor endowments, human capital and physical capital stocks per worker.

Dhura Dhaheshwar and Micheal T., in their study on a large number of African countries, investigates empirically the determinants of economic growth for the period 1981-92. Their results indicate that an increase in private investments has a relatively large positive impact on per capita growth. They state that growth is stimulated by public policies that lower the budget deficit in relation to GDP. That convergence of per capita income occurs after controlling for capital development and public policies.

Hsing, Y. and Hsieh, W. examine the determinants of the growth rate of real output for China with an emphasis on institutional, social and political changes and developments. They found out that growth rate of real out put is positively correlated with employment growth or the change in employment to real output ratio, investment /output ratio, human capital, but negatively associated with the great leap forward, and the cultural revolution. The coefficients of deficit financing the openness of the economy , and dummy variables for economic and agricultural reforms are found to be insignificant.

Data and Variables

This work follows Easterly and Rebelo’s regression using data published by World Bank and from OECD National Accounts: Volume II (1983,1990,1996). A sample of 60 developing countries have been analyzed and the results compared with the arguments advanced by Easterly and Rebelo . We scrutinise the observations and drop the outliers. The countries dropped are Lebanon and Rwanda, otherwise if they were to be included, they would affect our results badly.

In preliminary work, we found that pushing the sample period back before 1970 sharply reduced the number of countries for which the data were available.

Summary statistics

Table 3. All the 60 countries

Variable Tax revenue Primary School Secondary school Total Revenue Government Debt Gov't consumption

As reflected in the table above the observation for schooling are few. This is mainly in the 60's and the 70's where data collection and storage was very expensive in developing countries.

Hypotheses

The following hypotheses are tested

1- Government Revenue /GDP rises with per capita income

2- Government consumption correla ted with growth.

3- Human capital is consistently correlated with growth

4- High government debt has a strong negative impact on growth.

5- Domestic taxes /GDP falls with increase per capita income

Model

gdp= ? + ^? 1 ^{tax +?} 2

prm +^? 3 ^{sec +?} 4 ^{rev +?} 5 ^dod

+ ^? 6 ^{con +?}

where ; gdp – Growth rate per capita GDP(1960-1996) tax - Tax Revenue (% of GDP) prm - primary school enrollment(% gross) sec - secondary school enrollment(% gross) rev - Total Revenue(% of GDP) dod - Government debt (% of GDP) con - Government consumption (% of GDP)

School enrollment variables are included as proxies for the level of human capital

Empirical Analysis

Below we present simple correlation of fiscal variables with per capita growth rate Data range 1960 -1996 Correlation coefficient Tax revenue % of GDP -0.0875

Total Revenue % of GDP 0.0915 Government debt % of GDP -0.1036 Government consumption % of GDP -0.0915

We now look at the regression, starting with simple tax

regression as presented below.

Gdp= ? + ? tax +?

Below(table3), we have the coefficients and the t-statistics for tax revenue. As Easterly and Rebelo, our model predicts a detrimental effect on growth. Taxes affect the rate of growth by reducing the private returns to accumulation.

Table 3

Comparing African countries with other countries, we get a positive coefficient for tax

revenue in the case of African countries and a negative for others. This is explained by the fact that most African countries depend on taxes as a source of revenue to finance investment projects which lead to growth. Most investments in African countries are public investments thus the impact is greater than the distortionary effects.

When we expand our analysis by including the variable for primary and secondary education, we get a positive coefficient for primary, which is highly significant, but a negative coefficient for secondary. This explains that human capital is important for growth but it also indicates something to do with convergence. This will be explored later when we look at different groups of countries.

Table 2-7 shows results for the 57 countries by including one variable at time

Tax Revenue(% of GDP) Primary school enronment Secondary School enronment Total Revenue Government Debt Government consumption

Table 5

------------------------------------------------------------------------------

------------------------------------------------------------------------------

As indicated in tables above,we get a positive coefficient for government revenue as Easterly and Rebelo confirming the argument that the two are positively correlated. As shown in table 5 and 6 above, the effects of government debt is more complex. Government debt should reduce the rate of growth but we find the coefficient to be insignificant. There is a possibility of government debt to be correlated with government consumption. Regression results without debt variable gives a negatively significant coefficient for Government consumption.

Table 7.

------------------------------------------------------------------------------

------------------------------------------------------------------------------

We also find out that GDP growth is negatively correlated with the government consumption levels as in Barro(1991), and Easterly and Rebelo.

When we expand our re gression by including other explanatory variables, one at a time, we find that the coefficients tend to be insignificant. This means that as we add on more variables, we lose some efficiency.

We continue to investigate the differences between African countries and others by adding more explanatory variables and the results are presented in tables 9 to 16. When we add the coefficient for primary and secondary for African countries, the coefficient of tax revenue is effected in fact it becomes insignificant hence a possibility that the three are correlated. With a panel set, containing a number of observations for each country, there may be country –specific effects, which are not captured by the explanatory variables, in which case the error term will be autocorrelated within country observations. Greene (1990) explains how it is possible to test for such effects and obtain unbiased estimates if they are present.

Table 9: African Countries

Gap | Coef. Std. Err.

t P>|t| [95% Con. Interval]

---------
+
--------------------------------------------------------------------

------------------------------------------------------------------------------

Table 10: Others

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 11: African Countries

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 13: African Countries

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 14: others

------------------------------------------------------------------------------

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 16: Others

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table17-24 below, shows the results for the group wise regressions depending on the level of development. We have four groups arranged according to the level of development. Group 1 represents less developed countries.

We find a positive relationship between gdp and tax for less developed countries. As Easterly and Rebelo, we get a negative relationship for more developed countries.

Table 17: group 1

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 18: group 2

------------------------------------------------------------------------------

gdp | Coef.

Std. Err. t

P>|t| [95% Conf. Interval]

---------
+
--------------------------------------------------------------------
tax | -.4808518 .1135395 -4.235 0.000 -.7075408 -.2541628

_cons | 9.97274

1.63171 6.112

0.000 6.714926

13.23055

------------------------------------------------------------------------------

Table 19: group 3

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 20: group 4

--
----------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 22: group 2

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Table 24: group 4

------------------------------------------------------------------------------

------------------------------------------------------------------------------

For group one, which represents less developed countries, the coefficient for tax revenue is insignificant. This is an indication that tax revenue in those countries does not affect GDP growth rates. From our results for the level of development above, we get a negative coefficient for secondary school enrollment for 3 groups. There is an indication of convergence hence in line with barro’s and Dhura’s findings about conditional convergence.

In this paper we also investigate the presence of endogeneity between GDP growth rate, tax revenue and government consumption, implying that the regressions that we reported are contaminated by simultaneous equation bias. We use the lagged variables as our instruments and the 2SLS test does not indicate presence of endogeneity

Summary and Conclusion

We have found out that government consumption and tax revenue are correlated with growth. That tax increases have detrimental effects on growth. This is because most countries depend on income tax as a source of tax revenue. The sharp increase in public expenditure relative to GDP and increase in tax burdens and fiscal deficits are seen as major cause of the difficulties many countries are facing in terms of sluggish economic growth.

The group wise results show that the impact of government consumption levels on GDP growth is indeed as predicted by Easterly and Robero.

Due to the shortage of voluntary savings, developing countries, especially African countries, are compelled to resort to high taxation as a device for mobilising resources needed for their development plans. We get positive coefficient for tax revenue for African countries hence a difference with Easterly and Rebelo’s results.

The empirical results in this study tend to suggest that the stages of economic development do indeed matter in identifying the variables that determine economic growth.

We also note the importance of human capital in the process of growth. The coefficient for secondary school is mainly negative.

The results thus predict conditional convergence.

The most important factors determining the process of economic growth appear to be tax revenue, government consumption and human capital.

Reference

Abramavitz, M 1956.” Resources and Output Trends in the United States since 1870” . American Economic Review .46 5-23

Barro, R.J.(1990) Economic Growth in a Cross-Sction of Countries. Quarterly Journal of Economics 56. 407-443.

Chenery H., S. Robinson and Syrquin (1986) “Industrialisation and Growth: A comparative Study “ Oxford: Oxf ord University Press

Dhura Dhaneshwar , Micheal T., “Growth in Sub-Saharan Africa”. International Monetary Fund , Working Paper: 95/136. P21

Evsey D. Domar, “ Capital Expansion, Rate of growth and employment,” Econometrica XLIV no.2

Greene W. H. (1990), Econometrics Analysis. New York, Macmillan.

Harrod, R.F.,” An Essay in Dynamics Theory”, Economic Journal Vol. XLX, no.193 March 1939 pp 14-33.

Hsing, Y ; Hsieh, W. , “The Determinants of Economic Growth and Policy Implication”: The case of China. Journal of Economics. Vol.22 (1). p 15-21 1996

Keynes J.M., The general theory of employment, Interest and Money; London; Macmillan and co.ltd 1936.

Krueger, A. (1978) “ Foreign Trade Regimes and /Economic Development: Liberalisation Attempts and consequences “. Cambridge MA: NBER.

Kuznets, Simon. “Economic Growth and Income Inequality”. Development Economics Volume 2. Lal, Deepak, ed

Mankiw NG, “The Growth of nations Bookings papers on economic activity” Sept. 1995 The Bookings Institute

Ricardo Campbell Rita “US social Security Under Low Fertilty”. Population and Development Review Volume 12 p296-312 supp 1986.

Romer, P. (1990) “Endogenous Technological Change”. Journal of political Economy 98 S71-S102.

Robert E. Lucas, “ On the Mechanism of Economic Development”, Journal of Monetary Economics, vol. 22 (1) (July 1988).

Solow, R. M. (1956), “ A contribution to the theory of economic growth”. Quarterly Journal of Economics 70 :65-94

OECD: National Accounts Volume II: Datailed Tables Paris OECD, 1983,1990,1996.