Analysing and evaluating the taxpayer's demand for merit goods: the case of public sector education and health in Mauritius

Abstract

This paper analyses the average taxpayer's demand for merit goods, namely, education and health, in Mauritius, a small developing island state. To study these development-related goods, a fresh methodology is employed that captures the characteristics of the Mauritian economy. The empirical evidence relates to the post-independence period of more than three decades, 1973 to 1999 and beyond. The respective demand elasticities for each merit good are analysed and evaluated. The findings indicate that, contrary to the conventional theory of demand, beneficiaries tend to demand more of such goods when their prices are increased, and less when they are decreased. The rationale for this behaviour is that people, in general, and taxpayers, in particular, being quality conscious, increase their demand when they are asked to pay more, expecting in return an improvement in the quality of services. However, there is evidence that over these three decades users' preferences have shifted, particularly in the case of education, towards private provision, indicating less reliance on public sector provision, and that during the years of structural adjustments the demand for these services was severely suppressed.

¹This paper emanates from a research project undertaken by the author and which was fully sponsored by the Research and Consultancy Centre of the University of Mauritius in the year 2000.

1. Introduction

The main justification for the government's provision of a merit good, which in simple terms means an essential service, such as education, health or public transport, in an economy has to do with two economic doctrines: efficiency and equity. Such commitment reinforces the rationale for state intervention, namely, allocation of resources and distribution of income (Musgrave, 1959; Brown & Jackson, 1991). While efficiency gains are achieved through the generation of the socially optimum level of output by internalising any external benefit, equity gains are derived whenever merit goods are supplied at highly subsidised or concessionary rates to the poor and needy citizen. If it were left to the private sector to supply these goods such gains would far from materialise, owing to market imperfections, and there would be little or no welfare improvement.

On the whole, as these publicly provided private goods rely on the effective theory of demand, it is expected that there are specific factors that determine the users' ability and willingness to pay for them. In fact, it follows from the fundamental theory of collective decision making that publicly provided goods in general do not have a direct price or cost, but consumers do perceive their average tax contribution as the price they have paid or would pay to benefit from the consumption of the goods in question (Niskanen, 1978; Cornes & Sandler, 1996). Hence the demand for a merit good may also follow the conventional utility maximisation exercise conditional upon a given budget constraint.

It is of interest to study empirically the demand functions for merit goods, so as to identify whether such goods are considered to be luxuries or necessities. This identification is crucial to gauge economic decision making. If education or health has an elastic demand, it means that individuals in a given population have not yet reached the standard of recognising the importance of such goods in deriving socio-economic benefits. This occurs because the level of economic development might be too low. In such a case, the state has to further subsidise the provision of merit goods until they reach a desirable weight in the consumption basket of each citizen. Moreover, as it is publicly provided private goods that are in question here, it is essential to analyse any dynamic changes occurring in their demand. Economic development and the evolution of users' preferences can alter these demand functions and have specific bearing on public sector budgets. Yet another factor that prompts this investigation is the congestion hypothesis. Publicly provided goods, in general, and merit goods in particular, are subject to lumpiness in supply. Consequently, with an increasing number of beneficiaries because of population growth, congestion effects set in. In other words, while accessibility to benefit from provided services becomes constrained, the quality of public services eventually deteriorates. Congestion effects in fact increase the effective demand further and put more pressure on fiscal imbalances of developing economies. (Further such demand aspects are discussed in section 2 of this paper, related to their theoretical foundations.)

Empirical studies that track the determinants of the demand for public services in general have been rather sparse and vary significantly according to sample periods, country variations and methodologies applied. Early studies, for example by Borcherding & Deacon (1972) and Bergstrom & Goodman (1973), usually constitute the basis for modelling the demand function of publicly provided private goods. The study by Borcherding & Deacon (1972) reveals income elasticities for public services varying from + 0.2 to + 1.0, basically indicating that such services were necessities. On the whole, the computed income elasticities in this study were higher for public education than for health services, although price elasticities did suggest an inverse relationship between quantity demanded and the perceived tax price of each public service. In a more general context, Rubinfeld (1987) reports that income elasticities for publicly provided goods and services are, by and large, inelastic. Such goods are thus essential and, as might be hypothesised, could be regarded as necessities in the average voter-taxpayer's basket of consumption.

Other analyses pertaining to developing countries by Provopoulos (1982) and Khan (1988) also find a positive relationship between total real public expenditure and income, and substantiate a negative relationship between price and quantity of public services consumed. All in all, these findings tend to support the characteristics of a normal good. The evidence by Gertler & Hammer (1997), who emphasise the demand for health care, establishes that higher prices reduce the demand for medical services. The authors point out that there is a host of socio-economic factors, such as income, age, and gender, which dictate this relationship, particularly the size of the price elasticity coefficient. Jack (1999) provides a broader perspective on health-care demand in developing economies. Several demand elasticities were computed and compared in an attempt to generate policy recommendations. He found that his results were sensitive to the level of economic development, inter-regional disparities and peculiarities in household data. Sobhee (2003) studied the overall demand for public goods in Mauritius and found that the sensitivity of demand for such goods with respect to price changes depends on time periods. Both magnitude and direction are dictated by temporal elasticities.

There are two major limitations with the current empiricism on the modelling of demand for merit goods or publicly provided private goods: first, small developing economies have scarcely been addressed in the literature and, second, country specificities – that can provide new insights – have not been taken into account in the modelling of such demand functions. Hence, this paper extends this literature by exclusively modelling the demand for education and for health as merit goods in the context of Mauritius. Education and health expenditure lumped altogether in this small island economy constitute around 25 per cent of overall demand for public goods (pure public and merit goods altogether). Hence, taxpayer's money is to a large extent absorbed in financing such goods. The case of Mauritius is interesting because of its remarkable macro-economic performance among sub-Saharan economies in recent years. As such, it has grown at an average rate of 5 per cent over the last decade but, in contrast, has often incurred fiscal deficits exceeding 5 per cent of GDP. The growing demand for education and health puts considerable pressure on the fiscal imbalances. Uncontrolled fiscal deficits necessitated structural adjustments in the 1980s, primarily to improve macro-economic performance and restore confidence among private investors, through prudent fiscal retrenchment. To date, no assessment has been made to ascertain the impacts of such policies on the demand for merit goods. As an aftermath of the structural changes, there has been increasing competition through higher investment in the private sector in extending its education and health care services both qualitatively and quantitatively. Several hospitals and schools have been established, providing greater scope for competition and having a direct bearing on the demand for public education and health.

Essentially, this paper analyses and evaluates the evolution of the demand for each service and tries to capture specific factors that could provide more insights into the possible factors that influence such demand. The rest of this paper is organised as follows: section 2 develops a theoretical beneficiary – taxpayer model and articulates the theoretical underpinnings of demand elasticities emanating from the reduced form, section 3 deals with the empirical findings, and section 4 concludes.

2. An average Taxpayer's model

This section is split into two parts. Section 2.1 provides an introduction to the methodology used. (The technical aspects of the entire model are discussed and resolved in Appendix 1.) Section 2.2 deals with the theoretical underpinnings that govern the possible values or range of values that elasticity coefficients may assume in the reduced form of this model.

2.1 Methodological issues

It is assumed that the average taxpayer derives utilities from private goods X (or a vector of privately consumed goods) and two sets of publicly provided goods corresponding to pure public goods, on the one hand, and merit goods on the other. In particular, individual j's utility function is written in general form as:

X represents the private goods and G _i s represents publicly provided goods. A specific utility function is provided in Appendix 1 and is maximised subject to a given budget constraint. The whole optimisation exercise is performed and described in Appendix 1. The reduced form equation pertaining to the demand function to be estimated empirically can be expressed as:

Equation (8) can be written in log linear form as a time series equation (with ϵ_t as a stochastic term):

Indeed, the reduced form function suggests that G _i depends on P _i , Y, t, N and Z. In other words, the extended demand function for an impure public good depends on the perceived price of that good or service (P _i ), the average taxpayer's income (Y), preferences of the agent (t), number of beneficiaries (N) and government policy (Z). The theoretical underpinnings pertaining to each variable are discussed in the following subsection.

2.2 Theoretical postulates of demand elasticities for merit goods

The response of the average voter or taxpayer to changes in the factors influencing the demand for merit goods is a complex issue as such goods combine the characteristics of both public and private goods. What follows is a review of the theoretical underpinnings and channels that can influence the magnitude and direction of the coefficient of each variable emanating from the reduced form equation.

2.2.1 Perceived price of the good or service

The price of a merit good is represented by the average tax price perceived by users or taxpayers and is basically inversely related to the amount consumed as postulated by the conventional law of demand. However, this does not rule out the possibility of exceptions, for instance, quantity consumed being directly related to price in the event that users judge quality by the average amount they contribute. Consumers perceive a higher price as being compatible with a potentially better quality of the product supplied by the public sector. Yet another possibility could be the insensitivity of consumers to price changes; in which case, the coefficient of price elasticity would turn out to be insignificant or zero. Such an extreme possibility may occur whenever taxpayers are myopic and do not view their tax contribution as their willingness and ability to pay for public goods.

2.2.2 The income of consumers

By and large, the per capita income variable is expected to contribute positively to the amount consumed of a merit good as long as the latter is viewed as a normal good. However, it is erroneous to believe that income and quantity demanded for either education or health always have to be positively related. The nature of the good, largely dependent on household's income, will be indicated by the relevant sign. In Mauritius, private provision of education is often viewed as being superior to its public counterpart. Hence, as his or her income rises a consumer has a strong tendency to shift to a private provision. This belief is often substantiated by the fact that the public provision of merit goods involves congestion effects (Brown & Jackson, 1991) and is characterised by red tape and uniformity in the services. Thus, following the conventional Engel's Law, such behaviour is not irrational. When income is low, the demand for either education or health is positively related to income, and a high proportion of it is spent on merit goods.

In extremely poor countries, university education is still regarded as a luxury. [See Brown & Jackson (1991), who emphasise the relevance of economic development in the consumption of specific public goods.] Eventually, as a household's income rises, this share decreases proportionately and it is said that this commodity has become a necessity, with inelasticity of income. As per capita income rises, still further, the good then becomes inferior as the user has become relatively better off and can afford much better quality services supplied by private prestigious institutions. Therefore, generally speaking, whenever this stage is reached, the public provision turns out to be an inferior service.

2.2.3 Population

Regarding population, both the size and the composition play a major role in influencing this demand and thus these two elements are treated separately.

The size of population matters automatically as it directly determines the number of consumers for all goods in general, including merit goods. As the population increases, it is expected that the demand for publicly provided merit goods, in particular, education and health, will also rise. Because public goods are subject to lumpiness, as population increases congestion effects become unavoidable, thereby making services difficult to access. This prompts consumers or users to demand more of such services or to shift to private substitutes whenever they can afford them. In fact, what matters more is the compositional change of an expanding population as the amount consumers demand of each of these goods may greatly diverge in this context.

The composition matters because, for instance, an increase in the proportion of young people (age 0–15) can inevitably cause an increase in public demand for at least primary and secondary education but not necessarily for health. In contrast, a rise in the ageing population (age + 65) may accompany a sharp increase in the demand for public health but not necessarily in the demand for education. Finally, if the proportion of the working population (age 16–64) increases to total population, one may not foresee significant evolution in the demand for public health and education.

2.2.4 Preferences

Preferences may be highly related to habit persistence, the quality of the product and the demonstration effect. Habit persistence tends to dictate a positive correlation over time such that, by and large, families stick to a public provision of merit goods. Hence, habit persistence could be classified as a historical factor and could explain why consumers do not shift their preferences over time. Some families are either status- or quality-conscious and reveal their preferences in favour of an alternative provision at the expense of public sector services. In this case eventually, over time, one can anticipate for such families an inverse relationship between quantity consumed and time as a variable factor. The demonstration effect to a large extent tends to influence the consumption patterns of people vis-à-vis their neighbours or foreigners. Low-income consumers emulate the consumption behaviour of higher income classes, and these in turn apply the same principle by emulating the consumption patterns of the still higher income class to which they aspire.

2.2.4 Rationing policies

Government policies may very often bring about significant changes in the market conditions for publicly provided goods. Prolonged structural adjustment programmes, for instance, whereby public finances are strictly revisited and monitored, render the budget constraint tighter. A reduction in public services with unchanged tax policies (or higher taxes) may frustrate the demand for merit goods. Thus, it is expected that the quantity consumed may fall if the state rations the consumption of such goods. In Figure 1, a theoretical scenario is shown in which prolonged state policies may affect the demand for merit goods.

Figure 1: Demand for a merit good under rationing

Figure 1 shows the effects of a policy such as structural adjustment in the quantity demanded of a ‘welfare’ good. Adjustments originally tend to reduce quantity demanded from its initial state with no change in the actual demand function. The equilibrium point then moves from E₀ to an intermediate position E₁. However, the relatively higher tax price and the congestion effect described previously would tend in the long term to discourage demand whenever the policy is not short-lived. The equilibrium would then be established at E*, in which case the amount demanded would fall from OQ₂ to OQ*.

3. Empirical findings

The data used in this exercise were obtained from combined national and international sources. The national data relate to the Annual Digest of Statistics (1972–98) publications by the Central Statistical Office, while the international ones consist of annual issues of Government Finance Statistics (1997) and International Financial Statistics (1997). The data set pertaining to the Mauritian Economy spans the period 1973–99. The author has estimated only the demand for merit goods, in particular education and health, and in each equation real per capita public expenditure on education and health is used to represent the quantity demanded of each good.

This approach is generally used to reflect the average voter's amount consumed but may well suffer from some problems such aggregation and high volatility, especially during business cycles. In fact, this is a plausible way to capture the empirical demand function for public goods and has been applied in works by Khan (1988), Ashworth (1995) and Sobhee (2003). In the case of Mauritius, the insularity against volatility of public spending is maintained, given that tax revenue used to finance public spending comes from indirect taxes (about 75 per cent) and no unemployment benefits system exists. Further, in the absence of disaggregated data, the author advises following Sobhee (2003) for a similar exercise. The perceived tax price is represented by the average tax rate, computed as the ratio of real total tax revenue to real total public expenditure. Hence, this perceived tax price is identical for both merit goods, essentially because of the absence of tax earmarking and the fact that the beneficiary can consume both these goods from whatever he or she pays.

Regarding the evolution of preferences, as mentioned earlier, a time trend variable is applied. Regarding the policy variable, the author captures the structural adjustment period in the economy of Mauritius in the late 1970s and early 1980s. The basic aim of these programmes was to restore macro-economic stability through serious public finance reforms. To curb the high fiscal deficits, there were correspondingly major cuts in overall public expenditure. In fact, the author uses a dummy variable, defined as follows:

The empirical findings from time series regressions estimates are shown in more detail in Table 1.

Table 1: Empirical demand elasticities for education and health

Dependent variable	Constant	Explanatory variables					Reported test statistics
		P t	Y t	t	N	D		DW	F
	−9.3	0.42	0.49	−0.05	15.7	−0.15	0.92	2.04	24.7
	(−6.0)	(2.54)	(1.1)*	(−2.6)	(6.6)	(−3.3)
	−17.8	0.25	0.6	−0.02	2.96	−0.05	0.96	2.01	61.7
	(−1.7)	(2.2)	(2.0)	(−0.2)*	(1.8)	(−1.6)
Source: author's computation.
Notes: (i) Hausman's (1989) specification test was applied to detect the presence of any possible simultaneity bias, which was absent in this case; (ii) t-ratios are shown in parentheses; (iii) *indicates insignificance of the variable; (iv) represents the amount demanded for education; (v) represents the amount demanded for health; (vi) the author has not used unit roots tests here as described by Dickey & Fuller (1979 1981) as ‘t’ appears as an explanatory variable in his formulation.

From these findings, it can be inferred that per capita income has no significant impact on public education, suggesting that there is weak evidence that the taxpayer's income is a major determining factor of the amount he or she consumes of public education. Other variables appear to play a more effective role in influencing this demand. However, the average tax-price is a significant variable and appears to contradict the normal law of demand here. Instead of more being consumed at a lower price, the reverse seems to occur, implying that beneficiaries of publicly provided education in Mauritius have a tendency to judge quality by the price at which the good or service is provided. As the average tax price rises, consumers interpret this increase as a sign of potentially more reliable public output and are willing to demand more of it.

Similarly, regarding population, a positive impact is found, as postulated by economic theory. As population rises by 1 per cent, public expenditure on education rises by 15.7 per cent. This clearly shows that in recent years the quantity of this good that has been demanded has responded more than proportionately to population growth. In addition, such a finding supports a belief in the congestion or crowding hypothesis, where users demand more of such a service as their access to benefits from it is, to a large extent, constrained by population growth.

Next to be considered is the evolution of consumers' preferences. The negative relationship indicates that public education has become less popular over the years. It is but a truism that many users in Mauritius have been shifting to a private provision because they perceive it to be of superior quality. Lower student-to-teacher ratio, better infrastructure and higher rates of pay for teachers (an incentive for being more productive) in the private sector are likely to be the factors which have encouraged this shift to a private provision in the context of Mauritius. This is a prominent trend among rich families. The result is also consistent with the sign of the coefficient on the price variable. That is, users are quality-conscious; those who are not happy with publicly provided services go to the private sector.

Regarding rationing, restriction on the supply of merit goods by the state has consequently led to a greater congestion effect and an expansion of the demand for privately supplied substitutes in the 1980s and 1990s. In search of better quality, taxpayers have shifted to private schools, or the more demanding entry qualifications owing to limited places in public schools have led to a very high dropout rate especially among the lower income groups.

As a comparison, public demand for health was estimated for the same sample period, as shown in Table 1. There is little contrasting evidence obtained; with the exception of Y, all other impact elasticities are lower than in case of education. In the case of health, it is found that the commodity is viewed as a necessity, as indicated by the inelasticity of per capita income spent on the good. The latter variable is comparatively a more determining factor here. However, over time, users are found to have shifted their preferences to a private substitute, as in the case of education, but the effect is still weak. This would also indicate that for both historical reasons and the inability to afford private health care facilities, people have not been really successful in adapting to a private substitute. Private medical facilities are highly expensive and medical schemes are only very recent developments. Moreover, in recent years, public health care has improved greatly through the acquisition of sophisticated health equipment to ensure better delivery.

4. Conclusion

This paper has modelled the average taxpayer's demand for merit goods, namely education and health in Mauritius. Using an optimising exercise, this paper extends the literature by showing how the demand for merit goods could be dynamic. Empirical results on impact elasticities emanating from the Mauritian economy for education and health sectors differ in terms of their magnitude and explanatory power. It is found that Mauritians are willing to pay higher prices to obtain a better quality of public service. Further, the average voter and taxpaying citizen perceives publicly provided goods as basic necessities. There is evidence of the crowding or congestion hypothesis; that is, as the number of beneficiaries increases, there is a greater tendency to demand more of the goods in question. This occurs because the quality of services tends to decline and people reveal their desire through an extension of the current services provided. There is evidence that users have been shifting their preferences to a private provision for such goods in search of better quality–most probably, an outcome of the congestion problem. However, this is more pronounced in the case of education than health.

Lastly, the dummy variables used in both regressions clearly reveal that rationing impeded the demand for both these goods, but more significantly in the case of education. These findings have strong implications for policy: a more market-friendly approach may in the long term reduce the public demand for merit goods in Mauritius. The characteristics of this approach therefore have to be worked out carefully. Such an endeavour would clearly warrant an in-depth investigation so that the result will not be a classic case of improved public sector efficiency at the expense of equity.

Notes

¹This paper emanates from a research project undertaken by the author and which was fully sponsored by the Research and Consultancy Centre of the University of Mauritius in the year 2000.

Appendix 1

Optimisation exercise

It is assumed that the taxpayer's preferences are represented by a nested Cobb–Douglas utility function as specified below:

U _j is the utility derived by individual j, X represents the private good, represents an nth order vector of publicly provided goods (both pure and merit public goods), θ,θ _i and β are the usual impact elasticities. Here β registers the impact of all publicly provided goods and services on the beneficiary's utility function. Moreover, it is assumed that β is a dynamic impact elasticity coefficient, which is allowed in the long term to vary, owing to several factors that indirectly affect the utility of the optimising agent. In particular, the dynamism of β can be explained by the following functional relationship:

where t is a trend variable, N is number of beneficiaries and Z is a policy variable. The essence of the temporal variable t is to track the evolution of the individual's preferences in the presence of private substitutes. The lumpiness of public sector output often dictates that the population level (N) also matters in an environment where population density tends to be on the rise, as is the case in Mauritius. More specifically, this variable captures the extent of crowding or congestion effect whereby the quality of services can be affected as the number of beneficiaries rises (see Niskanen, 1978; Khan, 1991; Brown & Jackson, 1991). Lastly, a policy variable is deemed necessary to register significant changes in government policy that can affect the availability of the G_is, and hence the voter's utility. All in all, the policy variable is essential for developing countries that are often subjected to structural reforms, and Mauritius is not an exception.

On the other hand, the budget constraint of the taxpayer is given by:

where P _x represents the price of the private good X, P _i represents the perceived (tax) price of the beneficiary for each publicly provided good and Y is the average taxpayer's income.

Hence, the beneficiary is expected to maximise U(X, G _i s) subject to the budget constraint denoted by equation (3). Setting the problem in the usual Lagrangean multiplier framework yields:

First-order conditions yield:

Equating (5) and (6) yields:

Substituting (7) in the budget constraint yields:

Because β = f (t, N, Z) from (2) then, broadly defined as:

Equation (8) can be written in log linear form as a time series equation (with ϵ _t as a stochastic term):