![]() |
Home
| Databases
| WorldLII
| Search
| Feedback
eJournal of Tax Research |
![]() |
Poh Eng Hin[∗]
Tax expenditures are a potentially expedient means through which politicians can implement spending programs that target benefits at a select few while ensuring that the cost and distributive effects of such programs remain largely imperceptible to the majority. This paper reports the results of an exploratory study to assess the extent and determinants of public awareness of pronatalist tax policy in Singapore, and of public perceptiveness of the cost and distributive outcomes of these pronatalist tax incentives. It is found that survey respondents largely are aware of the existence of the incentives. However, there is widespread ignorance (if not misperceptions) of the spending implications and hidden cost associated with these incentives and of their distributive biases along eugenic, income and ethnic dimensions. Non-beneficiaries of the tax expenditures not only are less likely to be aware of the existence of the incentives, they are also less likely to be perceptive of the distributive effects. Overall, the empirical findings are a testimony to why the tax expenditure route has proven to be such a politically expedient way for the Singapore Government to implicitly pursue its policy of selective pronatalism.
Voters’ ability to accurately perceive the burdens and benefits associated with tax and spending decisions underpins the effective working of any system of taxation by democratic government. The important role of fiscal consciousness in affecting fiscal choice has motivated extensive and wide-ranging empirical research. Surveys have been carried out to investigate public consciousness of various aspects of the fiscal system, including knowledge of taxes and spending programs[1], consciousness of tax burdens and the benefits from public expenditures[2], and cognisance of the fiscal connection between the benefits of public sector output and the tax costs of such output[3]. Econometric analyses of archival data[4] as well as experimental research techniques[5] have also been used to investigate the extent, sources and consequences of fiscal illusion. Fiscal illusion (which potentially results from the lack of fiscal consciousness) is described in the literature (e.g. Oates, 1988) as voters’ systematic misperception of important fiscal parameters, leading possibly to their inability to make informed decisions and hence the distortion of their fiscal choices. Various structural or institutional elements of the fiscal system have long since been identified as potentially contributing to fiscal illusion (see e.g. Puviani, 1903 and Buchanan, 1967). These include the following: fragmentation of the budgetary process into taxing and spending decisions; diversity in the government’s revenue base (including the relative importance of non-tax revenues in the form of debt and inter-government grants); and lack of visibility of taxes or tax increases owing to reasons such as the income elasticity of the tax system, reliance on indirect taxes that are hidden in prices or otherwise ambiguous in incidence, administering tax collections by requiring tax to be withheld at source or by spreading out tax payments over time, etc.
One aspect of fiscal consciousness where relatively little research appears to have been done concerns the potential illusory effects associated with the use of tax expenditures as a means of public spending. Since the seminal work of Surrey (1973), much has been written to conceptualise the idea of a tax expenditure as being equivalent to a direct spending program, as well as to underscore the need for, and to put into effect, processes for the budgetary oversight of tax expenditure spending. Hardly any research, though, has been done to study public perceptions and consciousness in connection with the use of these tools of government spending. In this regard, some specific issues that are rarely addressed empirically include the following:
1. To what extent is there public cognisance of the hidden (opportunity) cost associated with tax expenditure spending and, hence, of the equivalence between a tax expenditure and a direct expenditure in the sense that both are spending programs that entail effectively an allocation and re-distribution of public resources?
2. To what extent is there public consciousness of how benefits are distributed under particular tax expenditure programs?
3. Who are more likely to be cognisant or conscious of the matters referred to above? For instance, is there any difference in perceptiveness, say, between beneficiaries and non-beneficiaries of the spending program?
The issues outlined above are important from a public choice perspective. If voters, on the whole, are ignorant of the equivalence between a tax expenditure and a direct expenditure and, hence, are imperceptive to invisible spending through the tax system, politicians will be better able to conceal spending programs by structuring them as tax expenditures. Such concealment may be motivated by the desire to have the public sector appear small or to hide potentially unpopular spending programs such as those targeted at vested interest groups or those entailing a distribution of benefits that is likely to be contentious. Such covert spending strategies will meet with relatively less public resistance, particularly in the absence of tax expenditure reporting and where non-beneficiaries of the tax expenditure programs are not conscious of the costs of such spending and of how the benefits are distributed. In other words, if widespread misperception persists that a tax expenditure program is costless (or if the costs are underestimated) and if the distribution of the benefits is obscured, fiscal choice is potentially distorted since, in the absence of such misperception, voters may well have objected to the spending program.
The present paper makes an initial attempt at filling the gap highlighted above in the fiscal consciousness literature. It presents the results of an exploratory survey research on the extent and determinants of fiscal misperceptions arising from the use of tax expenditures to deliver pronatalist subsidies to married couples in Singapore. The remainder of the paper is organised as follows. Section 2 presents a policy tools approach in considering the political and administrative reasons behind the use of tax expenditures (vis-à-vis direct subsidies) as an instrument of fiscal spending. It reviews the literature on the political economy of tax expenditure spending and the few existing theories that speculate on the likely level of public consciousness of the costs and distributive effects of tax expenditures. Section 3 describes the pronatalist tax incentives in Singapore and the political, fiscal, social and institutional contexts within which the incentives operate. Sections 4 and 5 present respectively the methodology and the findings of the study. Conclusions and directions for future research are contained in Section 6.
This section presents a policy tools approach[6], advocated by writers such as Salamon (1981), Hood (1986), and Salamon and Lund (1989), in considering some of the political and administrative factors that favour the use of tax expenditures as an instrument of fiscal spending. Both a direct subsidy and a tax expenditure confer a fiscal benefit by lowering for the targeted beneficiary the relative cost of undertaking the favoured activity. However, while a direct subsidy entails a disbursement out of the government’s revenue, a tax expenditure operates through the government’s forbearance in the first instance to collect tax revenue otherwise due under the normative tax structure. There is therefore the absence of an audit trail in the traditional budgetary accounts linking the fiscal spending to an explicit expenditure of government revenue. This fusion of the government’s taxing and spending functions can be a source of fiscal illusion if the spending implications are not generally perceived.
Much of the traditional literature on tax policy design (e.g. Report of the Royal Commission on Taxation, 1966; Bittker, 1967; Pechman, 1977) argued for the removal of spending programs from the tax system in order to preserve a comprehensive tax base for tax neutrality, fairness and simplicity. Broadening the tax base was indeed one of the main features characterising tax reforms worldwide in the 1980s and early 1990s (US Department of the Treasury, 1984; Sandford, 1993). The idea of spending through the tax system was given prominence from the late 1960s by Surrey’s conceptualisation of the term ‘tax expenditure’. Since then, much has been said (e.g. Surrey and McDaniel, 1985) about the hidden cost and lack of public accountability associated with tax expenditure spending[7] as well as other supposed deficiencies of tax expenditures such as their ‘windfall’ and ‘upside-down subsidy’ effects.[8] More recent contributions, such as Toder (2000), and Weisbach and Nussim (2004), have acknowledged that many of the alleged problems with tax expenditure spending are actually criticisms of the content and design of the underlying spending programs rather than of the use of tax expenditures per se as the means to deliver those programs. These deficiencies therefore may be overcome by appropriate design of the tax expenditure provisions, and by subjecting tax expenditure proposals and legislation to formal budgetary control and periodic review. In short, there appears to be no reason why tax expenditures cannot be designed to replicate the effects of direct subsidies on resource allocation and income distribution. However, tax expenditures do differ from direct expenditures in that different government agencies or departments are vested with jurisdiction over the spending programs – a tax expenditure program inevitably requires the involvement of the tax administration whereas a direct expenditure program is administered by a separate spending agency. Weisbach and Nussim frame the question of whether a spending program should be implemented through the tax system or via a direct expenditure program as one of institutional design, wherein the concern should be with how best to implement overall government policy rather than focussing merely on tax policy design considerations in isolation. More specifically, a spending program should be implemented as a tax expenditure if the benefits of coordination resulting from integrating the program with the tax system exceed the benefits of specialisation resulting from the administration of the program by a separate agency with the requisite expertise to better evaluate applications and allocate resources. In line with this, Toder suggests that a tax expenditure is a particularly suitable means to implement a spending program if:
Apart from the administrative considerations above, there may also be perceptual differences associated with the different modes of fiscal spending. Voters may perceive differently the value, the distribution and the cost of the benefits delivered through the tax system as opposed to through a direct subsidy. The ability to perceive the cost and distributive implications may also differ between the few who benefit from the spending program and the vast majority of taxpayers in general (non-beneficiaries) who effectively fund that program. There has not been all that much literature that addresses the psychological aspects of tax expenditures. One strand of literature addresses the issue from the perspective of recipients of tax subsidies, and contends that subsidies delivered through tax expenditures are more likely to be better received because these subsidies are less likely to be seen as a form of government assistance compared to direct subsidies. For instance, Woodside (1983 and 1986) suggests that the indirectness and relative invisibility of tax expenditure spending permit such forms of government intervention to be perceived as less intrusive on the private sector and less easily associated with the government. Similarly, Surrey and McDaniel speculate that beneficiaries of tax expenditure programs psychologically may feel that they are not recipients of state subsidies because they perceive tax subsidies to be no more than allowing them to retain what is essentially their own money.
There has also been some attempt to explain the psychology behind tax expenditures based on what decision theorists refer to as the endowment effect, i.e. that people systematically value opportunity costs less than equivalent out-of-pocket costs. If voters are susceptible to the endowment effect, they will value the opportunity cost (in terms of foregone tax revenue) associated with a tax subsidy less than the explicit cost of funding an equivalent direct subsidy. This is on account of the fact that a tax subsidy, unlike a direct subsidy, does not entail any explicit outflow of money from the Treasury to recipients of the subsidy. The implication is that voters in general will perceive a tax expenditure as involving a smaller effective sacrifice or expenditure of public resources compared to an equivalent direct subsidy. Furthermore, Shaviro (1990) argues that not everyone is equally susceptible to the endowment effect – beneficiaries of tax subsidies are likely to be less susceptible to this bias than non-beneficiaries. The result is that beneficiaries generally are able to perceive the benefits that they enjoy from a tax expenditure whereas non-beneficiaries will be less perceptive of the spending implications and will either fail to perceive or, at the very least, underestimate the cost of those benefits in terms of the tax collections foregone. Hansen (1983) makes a similar assertion that the benefits under a tax expenditure program tend to be highly visible to beneficiaries targeted by the program but virtually invisible to non-beneficiaries. She further contends that the total costs of the overall program will be hidden from both beneficiaries and non-beneficiaries alike.
The psychological considerations discussed above have implications on program implementation from a political or public choice perspective. They underscore the political invisibility of tax expenditure spending in terms of its hidden costs[9] and in terms of non-beneficiaries’ inability to perceive the benefits enjoyed by beneficiaries. If these contentions are valid, then from a public choice perspective, tax expenditures offer a politically expedient means of concealing spending programs that voters may otherwise find objectionable if implemented in the full glare of a direct expenditure program. Various motivations may be behind such covert spending through the tax system. For instance, politicians may wish to reward political patronage by targeting spending programs that favour a minority or special-interest group without imposing any perceived cost on the majority.[10] They may desire to foster the illusion that the public sector is smaller than the true extent of the government’s allocative and distributive interventions since the use of tax expenditures, instead of direct subsidies, enables lower taxes and lower spending to be reported in traditional budgetary accounts.[11] Spending through the tax system may also be undertaken to pursue certain economic, social or even ideological goals that entail controversial or unpopular re-distributive effects. Implementing a spending program by way of an income tax deduction or non-refundable tax credit is an administratively convenient and politically expedient way to conceal regressively targeted benefits in view of the inherent ‘upside-down subsidy’ effect brought about by progressivity in the income tax rate structure. In recent literature, Sullivan (2000), and Collier and Luther (2003) attributed some of the above reasons as motivations behind tax expenditure spending by the Clinton Administration in the US and the Thatcher Administration in the UK respectively.[12] People’s inability to fully understand the distributive implications of tax deductions has been empirically shown by Baron and McCaffery (2004) in two recent experimental studies conducted on the World-Wide Web. They found that many participants who favoured tax deductions as a means of government provision of various goods and services had either ignored or failed to consider the fact that, given a progressive tax rate structure, tax deductions will benefit the rich relatively more than the poor. Participants’ tendency to favour tax deductions was contrary to their preferences in support of redistribution from the rich to the poor. When these participants were informed of the effects of tax deductions, they became less likely to favour the use of the deductions. However, the drop in support for the deductions was only moderate, indicating that many of these participants continued to misunderstand the effects of the deductions.
To the author’s best knowledge, there has not been any empirical validation of whether and, if so, to what extent the endowment effect or other cognitive biases contributes to any perceptual differences between a tax subsidy and an equivalent direct subsidy.[13] There has also not been any empirical work investigating the extent and determinants of public consciousness of and misperceptions associated with the costs and distributive outcomes of tax expenditure programs, an issue which is explored in this paper. Empirical testing of any hypothesis of fiscal misperception in this regard probably has to be undertaken on a case-by-case basis to evaluate the validity of the theory as it applies to specific tax expenditure programs in particular jurisdictions.[14] Conclusions drawn from any one such study may well have limited generalisability given the specific nature, objectives and structure of the particular tax expenditure program studied and in view of jurisdictional-specific variables that may have a bearing on the general level of fiscal consciousness. This paper presents one such specific inquiry in investigating the extent and determinants of fiscal misperceptions arising from the use of tax expenditures in the context of pronatalist policy in Singapore.
Fertility rates in Singapore declined dramatically between the 1960s and the mid 1970s as a result of social and cultural transformation brought about by economic development, the availability of labour market participation opportunities for women, and a comprehensive antinatalist policy on the part of the Government. However, persistent below-replacement fertility rates in the 1980s led to fears that a shrinking and ageing population would adversely affect the sustainability of economic growth and the adequacy of existing health-care and social support systems. Furthermore, a trend emerged whereby many highly educated women were either remaining single or marrying later and having significantly fewer children than their less-educated counterparts. This raised concerns that the higher-educated and more talented strata of the population were not adequately replacing themselves.[15] These concerns led eventually to a reversal in the national fertility policy in the 1980s from one of antinatalism to one of selective pronatalism, with various specific financial and non-financial incentives announced in 1984 and 1987.[16]
The 1984 changes consisted of eugenic measures aimed at improving the quality of the population. These measures were an attempt at correcting the observed lopsided fertility pattern mentioned above and were founded on the rather controversial premise that the intelligence of the child is primarily inherited and principally determined by the parents’ educational attainments. The 1987 changes were more extensive and involved relaxing some of the existing antinatalist measures whilst introducing certain new pronatalist initiatives in a bid to encourage third- and fourth-order births from couples who could afford the long-term financial commitment of having more children. The emphasis on affordability reflected the stand that parents remain primarily responsible for the upbringing of their children and that any pronatalist intervention on the part of the government should not therefore result in short-term incentives for low-income couples to have more children at the expense of longer-term social ills and a strain on public welfare resources.
By predicating the pronatalist policy on the basis of affordability, two other certainly more contentious social engineering goals could be implicitly pursued. Firstly, because higher-income couples generally also possess higher educational qualifications (a proxy measure for natural talent and intelligence), the incentives introduced in 1987 were consistent with the eugenic policy of encouraging higher fertility amongst the better-educated, and presumably more intelligent and talented, strata of the population. The 1984 changes, which explicitly addressed the eugenic concerns, had resulted in a vehement public debate, widespread resentment, and even protest votes cast against the governing political party at the 1984 General Elections. One particularly castigated measure was the Graduate Mother Priority Scheme, which gave priority in the primary school registration process to children whose mothers were university graduates with at least three children. This measure was widely seen as symbolic of the elitism associated with the policy changes in 1984. It attracted widespread adverse public opinion, including resentment from graduate mothers themselves, and had to be abolished just one year after its introduction. The subsequent fertility policy changes in 1987 dropped the explicit reference to eugenic considerations although this goal was still pursued, albeit implicitly, through tax expenditures that targeted the fertility incentives mainly at higher-income couples.
The second goal implicitly pursued by predicating the pronatalist policy on parents’ affordability was the correction of the fertility imbalance between the two main ethnic groups, i.e. the Chinese and the Malays. The Chinese, who make up the significant majority of the population, have the lowest fertility rate whereas the Malays, who form the largest minority ethnic community, have the highest.[17] The Malays generally also have lower incomes and lower educational attainments compared to the Chinese. By targeting benefits mainly at higher-income and better-educated couples, the inherent effect of the various policy changes was to encourage more births from the Chinese in an implicit attempt to address the fertility imbalance between the two ethnic groups.[18]
At the time of the survey reported in this paper (December 2001), the main pronatalist tax expenditures in Singapore (which essentially evolved from the 1984 and 1987 policy announcements) consisted of the Enhanced Child Relief (ECR) and, more importantly, the Special Tax Rebates (STR).
ECR was introduced to encourage educated married women to have children whilst remaining economically active. It took the form of an annual income tax deduction that could be deducted only against the tax-assessable income of the child’s mother. Furthermore, the mother would be eligible for ECR only if she possessed at least the prescribed minimum educational qualification.[19] The annual ECR deduction per child (up to the fourth child of the family) was a stipulated percentage of the mother’s annual earned income[20], and this percentage increased (from 5% to as high as 25%) with the birth-order of the child.
STR took the form of non-refundable tax credits that could be set off against the gross income tax liabilities of the eligible parent/s over a stipulated number of years. These tax credits were first introduced in 1987 in respect of the third child of the family born in/after that year. The incentive was subsequently extended to the fourth child of the family born in/after 1988, and to the second child born in/after 1990. The rebate for the second child comprised a one-time non-refundable tax credit, which could be shared between the child’s parents for set-off against their respective gross tax liabilities. The amount of the tax credit ranged from S$0 to S$20,000, depending on the mother’s age at the time of delivery of the child. The rebates for the third child and for the fourth child consisted of two components. Each component also took the form of a one-time non-refundable tax credit. The first component was a lump sum S$20,000 tax credit, which could be shared between the child’s parents. The second component amounted to 15% of the mother’s earned income in the year of birth of the child, and this tax credit could be set off only against the mother’s gross tax liabilities.
In most cases, the parent’s/parents’ gross tax liability/liabilities for the first year after the birth of the child would be insufficient to fully utilise the STR tax credit. In this regard, any balance of the tax credit remaining unutilised could be carried forward for set-off against the future tax liabilities of the eligible parent/s for up to the next eight years. In other words, the maximum period over which each eligible child’s STR could be set off was nine years.
The pronatalist tax expenditures were structured so as to provide disproportionately higher tax subsidies to higher-income couples.[21] The structural features that contributed to this income-regressive subsidy effect were:
Significant reforms to the personal income tax system in the 1990s and early 2000s contributed to a further skew of the pronatalist tax subsidies in favour of the rich. In particular, the major structural tax reform of 1994, which saw the introduction of the Goods and Services Tax in place of significant cuts in income tax, resulted in some 70% of resident individuals dropping out of the scope of income taxation (IRAS, 1995). This meant that all low-income and some middle-income couples effectively were excluded from enjoying any of the pronatalist tax subsidies from the mid 1990s. Although the significant income tax cuts in the 1990s and 2000s also reduced the tax subsidies enjoyed by high-income and upper middle-income couples, the effect was relatively minimal for high-income couples while the effect for upper middle-income couples was mitigated by an amendment in 1994 that extended the maximum set-off period for the non-refundable tax credits from seven years to nine years.[22]
In summary, the use of tax expenditures to deliver fertility incentives appears to be an administratively and politically expedient way for the Singapore Government to implicitly pursue its policy of selective pronatalism. Other factors that facilitate this strategy include the political dominance and perceived credibility of the governing political party, the culture of top-down policy decision-making, and the absence of any form of tax expenditure reporting that might have highlighted the costs of the incentives. The introduction of the pronatalist tax expenditures in the 1980s was not met with the level of adverse public reaction and widespread resentment that had greeted the equally controversial Graduate Mother Priority Scheme mentioned earlier. This paper investigates whether public acquiescence of these tax incentives may be potentially explained by fiscal illusion.
The findings reported in this paper are based on the responses to certain questions that were included in a survey of 350 married individuals in Singapore. The main purpose of the survey was not to investigate fiscal misperceptions, but rather to study the extent to which respondents take into account the pronatalist tax incentives in their decisions to have children – the main findings, in this respect, are reported in Poh (2004). However, a number of questions had been included in the survey questionnaire to capture respondents’ perceptions of the incentives and it is these responses that are of relevance to this paper.
In view of the main objective of the survey, only married couples where the wife was not older than 44 years of age (in the year 2000) were surveyed. It was decided that, between the two spouses, the person filling in the survey questionnaire should be the spouse who had completed the couple’s most recent annual income tax return. This requirement was imposed because it was often not possible to survey both spouses, and therefore the requirement ensured that the responses were supplied by that spouse who had at least some level of awareness of the couple’s income tax affairs and some level of knowledge of income tax matters in general. It also meant that couples were excluded from the survey in situations where only one of the spouses was involved in the completion of the couple’s tax return and he/she was not available to participate in the survey.
The survey was conducted in December 2001 through visits made to various randomly selected households living in public flats and private residential properties across the city-state. Respondents were asked to complete a five-page questionnaire available in either English or Mandarin. The survey administrators were on hand to render any clarifications/assistance required by the respondents.
The socio-economic profile of the sample of 350 respondents who participated in the survey is presented in Table 1, together with the profile of the relevant population of married individuals in the year 2000.
TABLE 1: SOCIO-ECONOMIC PROFILES OF THE SAMPLE AND THE POPULATION OF MARRIED INDIVIDUALS (WHERE THE WIFE’S AGE DOES NOT EXCEED 44 YEARS)
|
Sample |
Population
[1] |
||||
|
|
|
|
|
|
|
|
Number of respondents |
% |
|
|
% |
|
Gender -
Female -
Male |
127 223 |
36.3 63.7 |
|
|
50.0 50.0 |
|
|
N = 350 |
100.0 |
|
|
100.0 |
[2] |
|
|
|
|
|
|
|
Age -
under 30 -
30 to 39 -
40 and above |
67 199 75 |
19.6 58.4 22.0 |
|
|
14.4 49.2 36.5 |
|
|
N = 341 |
100.0 |
|
|
100.0 |
[2] |
|
|
|
|
|
|
|
Ethnicity -
Chinese -
Malay -
Indian -
Others |
288 42 16 4 |
82.3 12.0 4.6 1.1 |
|
|
77.2 14.7 7.7 0.4 |
|
|
N = 350 |
100.0 |
|
|
100.0 |
[3] |
|
|
|
|
|
|
|
Highest educational
qualification -
Primary -
Secondary -
Tertiary |
11 123 216 |
3.1 35.1 61.7 |
|
|
29.4 54.4 16.2 |
|
|
N = 350 |
100.0 |
|
|
100.0 |
[4] |
|
|
|
|
|
|
|
Annual income (of those
working) in the year 2000 -
S$24,000 and
below -
S$24,001 to
S$60,000 -
S$60,001 and
above |
46 201 82 |
14.0 61.1 24.9 |
|
|
37.3 45.3 17.4 |
|
|
N = 329 |
100.0 |
[5] |
|
100.0 |
[6] |
|
|
|
|
|
|
|
Occupational status (of
those working) -
Professional/managerial -
Technical -
Clerical/sales/service -
Others
(including non-classifiable occupations) |
97 77 80 61 |
30.8 24.4 25.4 19.4 |
|
|
27.1 31.8 25.4 15.6 |
|
|
N = 315 |
100.0 |
[7] |
|
100.0 |
[8] |
|
|
|
|
|
|
|
Number
of children -
0 -
1 -
2 -
3 -
4 -
5 or more |
85 83 125 45 10 2 |
24.3 23.7 35.7 12.9 2.9 0.6 |
|
|
18.9 22.9 35.4 17.7 4.1 1.0 |
|
|
N = 350 |
100.0 |
|
|
100.0 |
[9] |
|
|
|
|
|
|
|
[1]
Population characteristics are derived from census data relating to the year
2000. |
||||||
|
||||||
[2]
Data is based on COP (2001c) Table 60, relating to married couples. |
||||||
|
||||||
[3]
Data is based on COP (2001c) Table 60, relating to married couples.
Individuals in inter-ethnic marriages have been excluded.
This is because the
census data does not break down data on couples in inter-ethnic marriages by
the ethnicity of the individual
spouses. Since Malays are less likely than
other races to be in inter-ethnic marriages, the actual proportion of married
individuals
(with the wife not exceeding 44 years of age) who are Malay is
likely to be less than the reported 14.7%. |
||||||
|
||||||
[4]
Data is based on COP (2001a) Table 16, relating to married individuals of
ages not exceeding 44. |
||||||
|
||||||
[5]
Data excludes 15 respondents who are homemakers. Furthermore, the sample data
is based on annual income from all sources. |
||||||
|
||||||
[6]
Data is based on COP (2001b) Table 36, relating to income from work (i.e.
business and employment only) of married individuals
of all ages. Including
married individuals of all ages overstates the proportion in the lowest
income group because of the presence
of the elderly who are more likely to
possess only low-level skills (and therefore earn low incomes) or are more
likely to be
undertaking part-time work. |
||||||
|
||||||
[7]
Data excludes 15 respondents who are homemakers. |
||||||
|
||||||
[8]
Data is based on COP (2001b) Table 19, relating to married individuals of all
ages. |
||||||
|
||||||
[9]
Data is based on COP (2001a) Table 25, relating to ever-married females (i.e.
including divorcees and widows) of ages not
exceeding 44. |
The sample appears to over-represent males compared to females but this is essentially due to the fact that it is usually the husband (rather than the wife) who completes the couple’s income tax return and who is therefore selected to participate in the survey. Those with tertiary education are over-represented although, to some extent, this was intended by the researcher given the specialised and technical nature of the subject matter of the survey. Intuitively, those with low education (and incomes) are less likely to gain from the tax incentives, less aware of their existence and less knowledgeable of their effects. It therefore made sense to sample proportionately more of higher-educated (and higher-income) respondents since it is this group to whom the incentives are targeted and who will enjoy the largest proportion of the benefits. For this same reason, those in the lowest income group (annual incomes not exceeding S$24,000) are under-represented – in fact, since the major tax reform of 1994 (discussed earlier in Section 3), these individuals generally are not liable to pay any income tax and will almost certainly not benefit at all from the tax incentives. Finally, there is also a slight over-representation of married individuals who are younger, who have no children, and who are in white-collar occupations.
Survey Questions
The survey questionnaire incorporated a number of questions that sought to ascertain respondents’ awareness, knowledge and perceptions of the STR and ECR tax incentives. Five of the questions elicit respondents’ perceptions on various aspects pertaining to the cost and distributive effects of the incentives and these responses are of particular relevance to this paper. The five questions are reproduced below and are numbered Q1, Q2, Q3, Q4A and Q4B for ease of reference:
Q1: “Tax incentives, such as STR and ECR, reduce the taxes paid by those benefiting from the incentives. Unlike a direct cash subsidy, the Government does not directly pay out any money to those benefiting from the tax incentives. Which statement below do you agree with?
Q2: “Malay couples form about 15% of all married couples of child-bearing age. Which statement below do you agree with?
Q3: “Which statement below do you agree with?
Q4: “Assume that there are two married individuals, H and L. Both are allowed STR and/or ECR for the same number of children and are in exactly the same circumstances, except that H’s annual income (say, $60,000) is two times L’s annual income (say, $30,000).
[A] Which statement below do you agree with?
[B] Which statement below do you also agree with?
Responses to Q1 will reveal if there is misperception on the part of respondents in thinking that a tax expenditure is costless and, in that sense, not equivalent to a direct expenditure. Responses to Q2, Q3 and Q4 will reveal if respondents are able to perceive the distribution of the pronatalist tax subsidies as effectively biased against Malay couples but favouring higher-educated and higher-income couples. Q4A presents the distribution of the tax subsidies in absolute dollar terms whereas Q4B frames the distribution of the subsidies in terms of whether it is income-regressive or income-progressive.
A couple of limitations to this study ought to be noted. These stem from the fact that the original objective of the survey was not to investigate fiscal misperceptions but rather to gain an insight into the extent to which the pronatalist tax incentives are taken into account in married couples’ decisions to have children. The first limitation relates to the survey sample, which excludes, amongst others, all single persons even though findings relating to their awareness and perceptions of the tax expenditures are potentially of interest. Secondly, due to constraints imposed by the need to avoid an unduly long survey instrument, only a few questions could be devoted in the questionnaire to eliciting respondents’ perceptions of the incentives. Responses were sought on five aspects or dimensions of interest (see Section 4.3 and Table 3). However, only one question covered each dimension examined. Therefore, given the absence of multiple questions to elicit responses on the same dimension, it was not possible to assess quantitatively the reliability of the responses provided.
The analysis in the remainder of this paper proceeds as follows. Section 5.1 reports the extent to which respondents are aware of the existence of the pronatalist tax expenditures and examines whether such awareness is related to their socio-economic characteristics. Section 5.2 proceeds to examine whether, and to what extent, respondents are able to perceive each of the five aspects pertaining to the cost and distributive effects of the tax incentives. The analyses in Section 5.2 cover only those respondents who are aware of the existence of at least one of the two incentives. Respondents’ ability to perceive each of the five aspects is also correlated to their socio-economic characteristics.
In examining the correlation between respondents’ ability to perceive the cost/ distributive effects and their socio-economic characteristics, both bivariate and multivariate approaches are adopted. From a bivariate perspective, two measures of association, Cramer’s V and Somer’s d, are reported. Cramer’s V is a symmetric measure of the strength of the association between two nominal variables. On the other hand, Somer’s d provides a directional measure of the strength of the association between two ordinal variables, with respondents’ ability to perceive as the dependent variable in the analysis. From a multivariate perspective, a logistic regression is run to regress the log odds of respondents’ ability to perceive against various predictor variables that take into account respondents’ socio-economic characteristics. The regression equations are arrived at using the backward stepwise method based on the Likelihood Ratio Test and significance levels of 5% and 10% respectively for entry and removal of variables.
Eight socio-economic variables are used as independent variables in the exploratory research:[23]
Out of the 350 respondents surveyed, 275 (78.6%) claimed to be aware of the STR incentive and 264 (75.4%) of the ECR incentive. 318 (90.9%) knew of at least one of the two incentives, with the remaining 32 (9.1%) having not heard of either incentive. Table 2 presents the association between awareness of at least one of the two tax expenditures and each of the socio-economic variables.
TABLE 2: AWARENESS OF AT LEAST ONE OF THE TWO PRONATALIST TAX EXPENDITURES BY VARIOUS SOCIO-ECONOMIC VARIABLES
|
|
Cramer’s V
[1] |
Somer’s d [2] |
||
Awareness of tax incentives (No v Yes) |
N |
Value |
p-value |
Value |
p-value |
by the following socio-economic
variables: |
|
|
|
|
|
Gender
(Female v Male) |
350 |
.054 |
.314 |
-.032 |
.291 |
Age
(<30 v 30-39 v =40) |
341 |
.075 |
.386 |
.030 |
.314 |
Ethnicity
(Non-Malay v Malay) |
350 |
.096 |
[3] .085 |
-.085 |
.158 |
Education
(Non-Tertiary v Tertiary) |
350 |
.138 |
.010 |
.082 |
.017 |
Tax
training (Not trained v Trained) |
350 |
.110 |
.040 |
.087 |
.001 |
Income
(=$24K v >$24K-$60K v >$60K) |
344 |
.089 |
.259 |
.039 |
.186 |
Occupation
(Other v Professional) |
329 |
.066 |
.232 |
.104 |
.189 |
Occupation
(Other v Finance-related) |
329 |
.050 |
[3] .443 |
.040 |
.284 |
|
|
|
|
|
|
[1]
Symmetric measure of nominal-by-nominal association. |
|||||
[2]
Directional measure of ordinal-by-ordinal association, with awareness of tax
incentives as the dependent variable. |
|||||
[3]
Using Fisher’s Exact Test (rather than Chi-square Test) due to one cell
having an expected frequency of less than 5. |
Based on Table 2, tertiary-educated respondents and those who have prior training in personal income taxation are more likely to be aware of at least one incentive as compared to respondents without such education and training. There is some evidence that non-Malays are more likely than Malays to be aware of at least one incentive although this result is not statistically significant at the 5% level. Other associations, including the association with income, are not statistically significant.
For the remaining analyses in this paper, only the 318 respondents with awareness of at least one of the two pronatalist tax expenditures are included. The remaining 32 respondents without any knowledge of the existence of either incentive were excluded because they could not logically be expected to provide any useful responses of their perceptions of the distributive effects of the tax expenditures.
General
Table 3 reports the numbers and percentages of respondents who are, and who are not, able to perceive the five different aspects relating to the cost and distributive outcomes of the pronatalist tax expenditures.
TABLE 3: ABILITY TO PERCEIVE THE COST AND DISTRIBUTIVE OUTCOMES OF THE PRONATALIST TAX EXPENDITURES
|
|
Perceptive |
Not perceptive |
|
Aspect in respect of
which respondent’s ability to perceive is examined |
N |
[1] |
Deluded [2] |
Unsure [3] |
|
|
|
|
|
Spending implications and hidden cost of the tax
expenditures (Q1) |
317 |
87 (27.4%) |
115 (36.3%) |
115 (36.3%) |
|
|
|
230 (72.6%) |
|
|
|
|
|
|
Benefits enjoyed
disproportionately less by Malay than non-Malay couples (Q2) |
318 |
55 (17.3%) |
115 (36.2%) |
148 (46.5%) |
|
|
|
263 (82.7%) |
|
|
|
|
|
|
Benefits enjoyed more by
higher-educated than lower-educated couples (Q3) |
318 |
120 (37.7%) |
97 (30.5%) |
101 (31.8%) |
|
|
|
198 (62.3%) |
|
|
|
|
|
|
In absolute dollars, more
benefits accrue to higher-income than lower-income couples (Q4A) |
317 |
126 (39.7%) |
18 (5.7%) |
173 (54.6%) |
|
|
|
191 (60.3%) |
|
|
|
|
|
|
As a percentage of income,
more benefits accrue to higher-income than lower-income couples (Q4B) |
317 |
56 (17.7%) |
61 (19.2%) |
200 (63.1%) |
|
|
|
261 (82.3%) |
|
|
||||
[1]
‘Perceptive’ refers to a correct response to the relevant question in Section
4.3 (e.g. for Q1, agreeing with the statement
that tax incentives are
provided at a cost to taxpayers), thereby suggesting that the respondent was
able to perceive the aspect
examined (i.e. the spending implications and
hidden cost of the tax expenditures). |
||||
[2]
‘Deluded’ refers to an incorrect response (e.g. for Q1, agreeing with the
statement that tax incentives are provided without
any cost to taxpayers). |
||||
[3]
‘Unsure’ refers to a ‘don’t know’ response. |
There appears to be widespread misperceptions (or, at the very least, ignorance) of the cost and distributive outcomes associated with the pronatalist tax expenditures. Fewer than three in ten respondents could appreciate the spending implications of the tax expenditures and the hidden (opportunity) cost of such spending. The fact that higher-income couples enjoy more tax savings in absolute dollars than their lower-income counterparts with the same number of children and in the same circumstances (a seemingly obvious point) is evident to only four in ten respondents. When asked if the distribution of the subsidies is income-progressive or income-regressive, only 18% of the respondents knew that it is income-regressive. A higher percentage of the respondents (38%) were able to perceive the elitist slant of the tax expenditures favouring higher-educated couples. There are two plausible reasons for this relatively higher level of consciousness. Firstly, educational qualification was an explicit qualifying condition for the ECR incentive and the prescribed minimum qualification was clearly stated in Inland Revenue literature referred to by taxpayers when completing their annual income tax returns. Secondly, the very intense and, to some extent, acrimonious public debate that followed the announcement of the controversial pronatalist measures in 1984 had very much focussed public attention on the fertility imbalance between the higher and lower educated, and had highlighted the eugenic bias of the incentives. The distributional aspect least perceived is the implicit bias of the tax expenditures against Malay couples. Only 17% of respondents were perceptive of this ethnic bias, and more than double this number (i.e. 36%) were in fact deluded into thinking that Malay couples enjoy benefits commensurate with the number of children they have. The relatively low perceptibility rate is unsurprising given that the ability to perceive this aspect required respondents not only to be aware that the incentives do not favour the lower educated and the lower income, but also to be cognisant of the fact that the lower educated and lower income are disproportionately Malay.
It may also be observed that issues relating to the distribution of the pronatalist tax subsidies by income had the highest percentages of ‘don’t know’ responses. 63% of respondents stated that they did not know whether the benefits of the tax expenditures are distributed in an income-progressive or income-regressive manner while, somewhat surprisingly, as many as 55% of respondents stated that they did not know whether higher-income couples enjoy more or less dollar savings than lower-income couples in the same circumstances.
In the remaining analyses that follow, responses relating to each aspect will be classified into two categories – ‘perceptive’ and ‘not perceptive’, with the latter category incorporating the ‘deluded’ and ‘don’t know’ responses.
Table 4 reports the bivariate association between respondents’ ability to perceive the spending implications and hidden cost of the pronatalist tax expenditures and each of the socio-economic variables.
TABLE 4: ABILITY TO PERCEIVE THE SPENDING IMPLICATIONS AND HIDDEN COST OF THE PRONATALIST TAX EXPENDITURES BY VARIOUS SOCIO-ECONOMIC VARIABLES
|
|
Cramer’s V
[1] |
Somer’s d [2] |
||
Ability to perceive (No v Yes) |
N |
Value |
p-value |
Value |
p-value |
by the following socio-economic
variables: |
|
|
|
|
|
Gender
(Female v Male) |
317 |
.108 |
.055 |
.100 |
.046 |
Age
(<30 v 30-39 v =40) [3] |
309 |
.152 |
.029 |
.007 |
.857 |
Ethnicity
(Non-Malay v Malay) |
317 |
.104 |
.064 |
-.148 |
.030 |
Education
(Non-Tertiary v Tertiary) |
317 |
.273 |
<.0005 |
.253 |
<.0005 |
Tax
training (Not trained v Trained) |
317 |
.179 |
.001 |
.214 |
.005 |
Income
(=$24K v >$24K-$60K v >$60K) |
311 |
.262 |
<.0005 |
.197 |
<.0005 |
Beneficiary status (Non-ben v MI ben v HI ben) [4] |
312 |
.174 |
.009 |
.063 |
.239 |
Occupation
(Other v Professional) |
299 |
.233 |
<.0005 |
.228 |
<.0005 |
Occupation
(Other v Finance-related) |
299 |
.166 |
.004 |
.201 |
.011 |
|
|
|
|
|
|
[1]
Symmetric measure of nominal-by-nominal association. |
|||||
[2]
Directional measure of ordinal-by-ordinal association, with ability to
perceive as the dependent variable. |
|||||
[3]
There is no statistically significant difference in ability to perceive
between respondents in age groups <30 and =40.
However, respondents
in age group 30-39 are more perceptive than those in the other age groups.
The 2x2 classification (Other
v 30-39) yields the following statistics: N =
309; V = .150 (p = .008); d = .135 (p = .006). |
|||||
[4]
There is no statistically significant difference in ability to perceive
between non-beneficiaries and middle-income beneficiaries.
However,
high-income beneficiaries are more perceptive than non-beneficiaries or
middle-income beneficiaries. The 2x2 classification
(Other v HI ben) yields the following statistics: N = 313; V = .171
(p = .003); d = .270 (p = .013). |
The results suggest that education, tax training, income and occupational status have a bearing on whether the spending implications and hidden (opportunity) cost of the pronatalist tax expenditures are perceived. There is also evidence to suggest a higher level of perceptiveness in this respect amongst males (compared to females), amongst respondents between the ages of 30 and 39 (compared to respondents of other ages), and amongst non-Malays (compared to Malays). However, based on results of the multivariate analysis reported below, these variables are no longer statistically significant once education, tax training, income and occupational status are controlled.
With regards to beneficiary status, the bivariate analysis suggests that high-income beneficiaries are more likely to perceive the spending and cost implications of the tax expenditures than other respondents (Note [4] in Table 4). However, when compared to non-beneficiaries on the same incomes (i.e. controlling for income), high-income beneficiaries are not significantly more perceptive (Table 5). In other words, while a high-income respondent is certainly more perceptive than a lower-income respondent due probably to a greater incentive and the means to be fiscally knowledgeable and/or due to better financial acumen, the fact that the high-income respondent is also a beneficiary of a pronatalist tax incentive does not, in itself, make him/her any more cognisant of the spending and cost implications of the tax expenditures. One does not need to be a beneficiary of a specific tax expenditure program in order to appreciate the spending implications and opportunity cost to taxpayers that characterise tax expenditure spending in general.
TABLE 5: ABILITY TO PERCEIVE THE SPENDING IMPLICATIONS AND HIDDEN COST OF THE PRONATALIST TAX EXPENDITURES AMONGST HIGH-INCOME RESPONDENTS (INCOMES EXCEEDING S$60,000) BY BENEFICIARY STATUS
|
Non-beneficiary |
Beneficiary |
Total |
|
No. (%) of high-income respondents |
No. (%) of high-income respondents |
No. (%) of high-income respondents |
|
|
|
|
Not
perceptive |
26 (55.3%) |
13 (48.1%) |
39 (52.7%) |
Perceptive |
21 (44.7%) |
14 (51.9%) |
35 (47.3%) |
|
47 (100.0%) |
27 (100.0%) |
74 (100.0%) |
|
|
|
|
Minimum
expected cell frequency: 12.77 |
|||
Cramer’s
V = .069 (p = .552) |
|||
Somer’s
d (Ability to perceive as the dependent variable) = .072 (p = .552) |
To ascertain the relative importance of the various socio-economic variables in determining respondents’ ability to perceive the spending implications and hidden cost, a logistic regression was run, with the dependent variable being the log odds of being able to perceive, and the independent variables comprising the various socio-economic variables. The basis of arriving at the regression equation is as described in section 4.5 and the results are summarised in Table 6.[24]
TABLE 6: LOGISTIC REGRESSION – ABILITY TO PERCEIVE THE SPENDING IMPLICATIONS AND HIDDEN COST OF THE PRONATALIST TAX EXPENDITURES
N
= 288 |
Nagelkerke
R2 = .218 |
|
|||||||||
|
|||||||||||
Likelihood
Ratio Test for Overall Model: |
|||||||||||
-2LL
for final model = 141.395 |
Chi-square
= 46.605 |
df = 5 |
p-value
= <.0005 |
||||||||
|
|||||||||||
Hosmer and Lemeshow Test for Goodness-of-Fit: |
|||||||||||
Chi-square
= 9.369 |
df = 6 |
p-value
= .154 |
|||||||||
|
|||||||||||
Variables
in the Equation: |
|
|
|
|
|
|
|||||
|
Parameter estimate |
Std error |
Wald |
df |
p-value |
Odds ratio |
|||||
EDU
[1] |
1.016 |
.381 |
7.120 |
1 |
.008 |
2.762 |
|||||
TRAIN
[2] |
.892 |
.351 |
6.454 |
1 |
.011 |
2.440 |
|||||
OCC_PRO
[3] |
.711 |
.301 |
5.563 |
1 |
.018 |
2.036 |
|||||
Income |
|
|
9.514 |
2 |
.009 |
|
|||||
-
LOW_INC [4] |
-1.536 |
.573 |
7.175 |
1 |
.007 |
.215 |
|||||
-
MID_INC [5] |
-.788 |
.322 |
6.001 |
1 |
.014 |
.455 |
|||||
INTERCEPT |
-1.539 |
.434 |
12.576 |
1 |
<.0005 |
.215 |
|||||
|
|
|
|
|
|
|
|||||
Likelihood
Ratio Tests for Individual Variables: |
|||||||||||
|
-2LL of reduced model |
Chi-square |
df |
p-value |
|
||||||
INTERCEPT |
141.395 |
.000 |
0 |
|
|
||||||
EDU |
149.239 |
7.844 |
1 |
.005 |
|
||||||
TRAIN |
147.736 |
6.341 |
1 |
.012 |
|
||||||
OCC_PRO |
146.907 |
5.512 |
1 |
.019 |
|
||||||
Income
(LOW_INC and MID_INC) |
151.345 |
9.950 |
2 |
.007 |
|
||||||
|
|
|
|
|
|
||||||
[1] EDU (0 = Non-tertiary-educated,
1= Tertiary-educated) |
|||||||||||
[2] TRAIN (0 = Not tax trained, 1 =
Tax trained) |
|||||||||||
[3]
OCC_PRO (0 = Other occupation, 1 = Professional/managerial occupation) |
|||||||||||
[4] LOW_INC (0 = Other income levels,
1 = Income not exceeding S$24,000) |
|||||||||||
[5] MID_INC (0 = Other income levels,
1 = Income exceeding S$24,000 but not exceeding S$60,000) |
|||||||||||
The results indicate that income is one of the most important variables determining respondents’ ability to perceive the spending implications and hidden cost of the tax expenditures – in particular, the odds of a low-income respondent being able to perceive are a factor of only .215 of the odds for a high-income respondent, or put the other way, the high-income respondent’s odds are a factor of 4.646 of the low-income respondent’s odds. General education, special tax training and occupational status are the other significant determinants, whereas gender, age, ethnicity and beneficiary status are statistically insignificant when the four variables mentioned previously are controlled.
All bivariate associations between respondents’ ability to perceive the bias of the tax expenditures against Malay couples and the various socio-economic variables (except for beneficiary status) are weak and statistically insignificant. One conclusion therefore is that Malays are not significantly more, or less, likely than non-Malays to perceive that the incentives are effectively biased against them.[25]
The data (Table 7), however, provides some statistical support for the contention that beneficiaries are more perceptive than non-beneficiaries of the hidden ethnic bias of the tax incentives (although this association is a relatively weak one).
TABLE 7: ABILITY TO PERCEIVE THE ETHNIC BIAS OF THE PRONATALIST TAX EXPENDITURES BY BENEFICIARY STATUS
|
Non-beneficiary |
Beneficiary |
Total |
|
No. (%) of respondents |
No. (%) of respondents |
No. (%) of respondents |
|
|
|
|
Not
perceptive |
182 (85.4%) |
76 (76.0%) |
258 (82.4%) |
Perceptive |
31 (14.6%) |
24 (24.0%) |
55 (17.6%) |
|
213 (100.0%) |
100 (100.0%) |
313 (100.0%) |
|
|
|
|
Minimum
expected cell frequency: 17.57 |
|||
Cramer’s
V = .116 (p = .041) |
|||
Somer’s
d (Ability to perceive as the dependent variable) = .094 (p = .055) |
|||
|
|
|
|
Perceptibility of the eugenic bias
Table 8 presents the bivariate association between respondents’ ability to perceive the eugenic nature of the pronatalist tax expenditures and each of the socio-economic variables. An additional variable, ‘completion of married woman’s tax return’, has been included to take into account whether the respondent had any part in completing that section of the couple’s tax return that relates to the wife. Since only the wife could be eligible for the ECR incentive and the ECR incentive had an explicit eligibility condition that related to educational qualifications, completing the married woman’s tax return potentially exposes the respondent to this information and raises his/her consciousness of the eugenic bias of the incentive.
TABLE 8: ABILITY TO PERCEIVE THE EUGENIC BIAS OF THE PRONATALIST TAX EXPENDITURES BY VARIOUS SOCIO-ECONOMIC VARIABLES
|
|
Cramer’s V
[1] |
Somer’s d [2] |
||
Ability to perceive (No v Yes) |
N |
Value |
p-value |
Value |
p-value |
by the following socio-economic
variables: |
|
|
|
|
|
Gender
(Female v Male) |
318 |
.033 |
.554 |
-.033 |
.555 |
Age
(<30 v 30-39 v =40) [3] |
310 |
.187 |
.004 |
.136 |
.007 |
Ethnicity
(Non-Malay v Malay) |
318 |
.037 |
.508 |
.058 |
.517 |
Education
(Non-Tertiary v Tertiary) |
318 |
.008 |
.884 |
-.008 |
.885 |
Tax
training (Not trained v Trained) |
318 |
.218 |
<.0005 |
.282 |
<.0005 |
Income
(=$24K v >$24K-$60K v >$60K) |
312 |
.077 |
.400 |
-.057 |
.241 |
Beneficiary status (Non-ben v MI ben v HI ben) |
313 |
.069 |
.475 |
.007 |
.903 |
Occupation
(Other v Professional) |
299 |
.054 |
.350 |
.057 |
.354 |
Occupation
(Other v Finance-related) |
299 |
.158 |
.006 |
.207 |
.010 |
Completion
of married woman’s tax return (No
v Yes) |
316 |
.151 |
.007 |
.148 |
.006 |
|
|
|
|
|
|
[1]
Symmetric measure of nominal-by-nominal association. |
|||||
[2]
Directional measure of ordinal-by-ordinal association, with ability to
perceive as the dependent variable. |
|||||
[3]
There is no statistically significant difference in ability to perceive
between respondents in age groups <30 and 30-39.
However, respondents in
age group =40 are more perceptive than those in the other age groups.
The 2x2 classification (Other v
=40) yields the following statistics: N
= 310; V = .187 (p = .001); d = .219 (p = .002). |
Based on the bivariate analyses, the variables with statistical significance at the 5% level are age (Other v =40), tax training, occupation (other v finance-related) and completion of married woman’s tax return. Gender, ethnicity, education, income, beneficiary status and occupational status (other v professional) do not determine respondents’ ability to perceive the eugenic bias of the tax incentives. Age, tax training and completion of married woman’s tax return are also confirmed statistically significant from a multivariate perspective. Table 9 reports the results of the relevant regression.
Table 9: Logistic regression – Ability to
perceive the eugenic bias
N
= 291 |
Nagelkerke
R2 = .128 |
|
|||||||
|
|||||||||
Likelihood
Ratio Test for Overall Model: |
|||||||||
-2LL
for final model = 47.728 |
Chi-square
= 28.847 |
df = 4 |
p-value
= <.0005 |
||||||
|
|||||||||
Hosmer and Lemeshow Test for Goodness-of-Fit: |
|||||||||
Chi-square
= 1.743 |
df = 4 |
p-value
= .783 |
|||||||
|
|||||||||
Variables
in the Equation: |
|
|
|
|
|
|
|||
|
Parameter estimate |
Std error |
Wald |
df |
p-value |
Odds ratio |
|||
AGE_40
[1] |
.885 |
.301 |
8.652 |
1 |
.003 |
2.424 |
|||
TRAIN
[2] |
.926 |
.350 |
6.975 |
1 |
.008 |
2.523 |
|||
OCC_FIN
[3] |
.617 |
.362 |
2.904 |
1 |
.088 |
1.852 |
|||
WRET
[4] |
.529 |
.261 |
4.115 |
1 |
.042 |
1.697 |
|||
INTERCEPT |
-1.273 |
.228 |
31.178 |
1 |
<.0005 |
.280 |
|||
|
|
|
|
|
|
|
|||
Likelihood
Ratio Tests for Individual Variables: |
|||||||||
|
-2LL of reduced model |
Chi-square |
df |
p-value |
|
||||
INTERCEPT |
47.728 |
.000 |
0 |
|
|
||||
AGE_40 |
56.435 |
8.707 |
1 |
.003 |
|
||||
TRAIN |
54.785 |
7.057 |
1 |
.008 |
|
||||
OCC_FIN |
50.609 |
2.881 |
1 |
.090 |
|
||||
WRET |
51.899 |
4.171 |
1 |
.041 |
|
||||
|
|
|
|
|
|
||||
[1] AGE_40 (0 = Other age group, 1 =
Age group =40) |
|||||||||
[2] TRAIN (0 = Not tax trained, 1 =
Tax trained) |
|||||||||
[3] OCC_FIN (0 = Other occupation, 1
= Finance-related occupation) |
|||||||||
[4] WRET (0 = Did not complete
married woman’s tax return, 1 = Completed married woman’s tax return) |
|||||||||
The finding that respondents aged 40 years or older are more perceptive (than younger respondents) of the eugenic bias of the tax expenditures is robust even after controlling for the other variables. There may well be a very simple reason for this. It has to be remembered that these older respondents would have been in their mid twenties to mid thirties[26] at the time when the controversial tax and other incentives were first introduced in the 1980s. They would have experienced and, perhaps, even participated in the intense public debate that occurred at that time. This debate had evolved mainly around the issues of eugenics and elitism. Some had disputed outright the validity of the hypothesis that intelligence and talent are primarily inherited, while many others felt strongly that, regardless of any truth in the hypothesis, it was beyond the legitimacy of government action anyway to pursue a eugenic population policy and to institutionalise a system of incentives and disincentives that breeds elitism and promotes class-consciousness. The extensive media coverage and widespread publicity given to the heated exchanges in the 1980s appear, more than fifteen years on, to have left an imprint on older respondents of the present survey. Respondents old enough to recall the controversies and acrimony of the mid 1980s are found to be more likely to perceive the eugenic bias of the incentives. Younger respondents to the survey would have been too young back then to appreciate or remember the controversy.
Tables 10 and 11 report respectively the bivariate associations and the logistic regression relating to respondents’ ability to perceive that higher-income couples enjoy more dollar subsidies from the pronatalist tax expenditures than do lower-income couples in the same circumstances.
TABLE 10: ABILITY TO PERCEIVE WHETHER HIGHER-INCOME COUPLES ENJOY MORE TAX SUBSIDIES IN ABSOLUTE DOLLARS BY VARIOUS SOCIO-ECONOMIC VARIABLES
|
|
Cramer’s V
[1] |
Somer’s d [2] |
||
Ability to perceive (No v Yes) |
N |
Value |
p-value |
Value |
p-value |
by the following socio-economic
variables: |
|
|
|
|
|
Gender
(Female v Male) |
317 |
.012 |
.830 |
.012 |
.830 |
Age
(<30 v 30-39 v =40) |
309 |
.042 |
.761 |
.037 |
.459 |
Ethnicity
(Non-Malay v Malay) |
317 |
.105 |
.062 |
.163 |
.075 |
Education
(Non-Tertiary v Tertiary) |
317 |
.098 |
.080 |
.100 |
.075 |
Tax
training (Not trained v Trained) |
317 |
.258 |
<.0005 |
.338 |
<.0005 |
Income
(=$24K v >$24K-$60K v >$60K) |
311 |
.071 |
.455 |
-.012 |
.814 |
Beneficiary status (Non-ben v MI ben v HI ben) |
312 |
.122 |
.099 |
.113 |
.044 |
Occupation
(Other v Professional) |
298 |
.090 |
.121 |
.096 |
.125 |
Occupation
(Other v Finance-related) |
298 |
.223 |
<.0005 |
.296 |
<.0005 |
|
|
|
|
|
|
[1]
Symmetric measure of nominal-by-nominal association. |
|||||
[2]
Directional measure of ordinal-by-ordinal association, with ability to
perceive as the dependent variable. |
TABLE 11: LOGISTIC REGRESSION – ABILITY TO PERCEIVE THAT HIGHER-INCOME COUPLES ENJOY MORE TAX SUBSIDIES IN ABSOLUTE DOLLARS
N
= 293 |
Nagelkerke
R2 = .142 |
|
|||||||||
|
|||||||||||
Likelihood
Ratio Test for Overall Model: |
|||||||||||
-2LL
for final model = 75.292 |
Chi-square
= 32.640 |
df = 5 |
p-value
= <.0005 |
||||||||
|
|||||||||||
Hosmer and Lemeshow Test for Goodness-of-Fit: |
|||||||||||
Chi-square
= 2.563 |
df = 4 |
p-value
= .633 |
|||||||||
|
|||||||||||
Variables
in the Equation: |
|
|
|
|
|
|
|||||
|
Parameter estimate |
Std error |
Wald |
df |
p-value |
Odds ratio |
|||||
ETH
[1] |
.720 |
.391 |
3.386 |
1 |
.066 |
2.055 |
|||||
TRAIN
[2] |
.989 |
.358 |
7.644 |
1 |
.006 |
2.688 |
|||||
Beneficiary
status |
|
|
5.167 |
2 |
.075 |
|
|||||
-
MID_BEN [3] |
.453 |
.305 |
2.204 |
1 |
.138 |
1.573 |
|||||
-
HI_BEN [4] |
.860 |
.436 |
3.897 |
1 |
.048 |
2.364 |
|||||
OCC_FIN
[5] |
1.022 |
.365 |
7.856 |
1 |
.005 |
2.779 |
|||||
INTERCEPT |
-1.007 |
.182 |
30.698 |
1 |
<.0005 |
.365 |
|||||
|
|
|
|
|
|
|
|||||
Likelihood
Ratio Tests for Individual Variables: |
|||||||||||
|
-2LL of reduced model |
Chi-square |
df |
p-value |
|
||||||
INTERCEPT |
75.292 |
.000 |
0 |
|
|
||||||
ETH |
78.680 |
3.387 |
1 |
.066 |
|
||||||
TRAIN |
83.081 |
7.789 |
1 |
.005 |
|
||||||
Beneficiary
status (MID_BEN
and HI_BEN) |
80.491 |
5.198 |
2 |
.074 |
|
||||||
OCC_FIN |
83.255 |
7.963 |
1 |
.005 |
|
||||||
|
|
|
|
|
|
||||||
[1] ETH (0 = Non-Malay, 1 = Malay) |
|||||||||||
[2] TRAIN (0 = Not tax trained, 1 =
Tax trained) |
|||||||||||
[3] MID_BEN (0 = Other respondent, 1
= Beneficiary with income exceeding S$24,000 but not exceeding S$60,000) |
|||||||||||
[4] HI_BEN (0 = Other respondent, 1 =
Beneficiary with income exceeding S$60,000) |
|||||||||||
[5] OCC_FIN (0 = Other occupation, 1
= Finance-related occupation) |
|||||||||||
|
|||||||||||
Table 12: Ability to perceive whether higher-income
couples enjoy more tax subsidies as a percentage of income (i.e. the
income-regressive
benefit distribution) by various socio-economic variables
|
|
Cramer’s V
[1] |
Somer’s d [2] |
||
Ability to perceive (No v Yes) |
N |
Value |
p-value |
Value |
p-value |
by the following socio-economic
variables: |
|
|
|
|
|
Gender
(Female v Male) |
317 |
.003 |
.962 |
-.002 |
.962 |
Age
(<30 v 30-39 v =40) |
309 |
.034 |
.833 |
.003 |
.947 |
Ethnicity
(Non-Malay v Malay) |
317 |
.022 |
.701 |
.026 |
.713 |
Education
(Non-Tertiary v Tertiary) |
317 |
.020 |
.727 |
.016 |
.724 |
Tax
training (Not trained v Trained) |
317 |
.103 |
.067 |
.105 |
.109 |
Income
(=$24K v >$24K-$60K v >$60K) [3] |
311 |
.193 |
.003 |
-.052 |
.249 |
Beneficiary
status (Other v HI ben) [4] |
313 |
.127 |
[5] .034 |
.172 |
.077 |
Occupation
(Other v Professional) |
298 |
.028 |
.627 |
-.024 |
.619 |
Occupation
(Other v Finance-related) |
298 |
.050 |
.390 |
.052 |
.422 |
|
|
|
|
|
|
[1]
Symmetric measure of nominal-by-nominal association. |
|||||
[2]
Directional measure of ordinal-by-ordinal association, with ability to
perceive as the dependent variable. |
|||||
[3]
Respondents with incomes up to $24,000 are more perceptive than those in
higher income groups. The 2x2 classification (=$24K
v Other) yields the
following statistics: N = 311; V = .176 (p = .002); d = -.180 (p = .011). |
|||||
[4]
There is no statistically significant difference in ability to perceive
between non-beneficiaries and middle-income beneficiaries.
These two groups
are collapsed into one labelled ‘Other’ in order to
obtain a 2x2 classification to which Fisher’s Exact Test is applied. |
|||||
[5]
Using Fisher’s Exact Test (rather than Chi-square Test) due to one cell
having an expected frequency of less than 5. |
Tables 12 and 13 report respectively the bivariate associations and the logistic regression relating to respondents’ ability to perceive that higher-income couples enjoy more tax subsidies as a percentage of income than do lower-income couples in the same circumstances.
TABLE 13: LOGISTIC REGRESSION – ABILITY TO PERCEIVE THE INCOME-REGRESSIVE BENEFIT DISTRIBUTION
N
= 309 |
Nagelkerke
R2 = .083 |
|
||||||||
|
||||||||||
Likelihood
Ratio Test for Overall Model: |
||||||||||
-2LL
for final model = 23.494 |
Chi-square
= 15.880 |
df = 2 |
p-value
= <.0005 |
|||||||
|
||||||||||
Hosmer and Lemeshow Test for Goodness-of-Fit: |
||||||||||
Chi-square
= .015 |
df = 1 |
p-value
= .902 |
||||||||
|
||||||||||
Variables
in the Equation: |
|
|
|
|
|
|
||||
|
Parameter estimate |
Std error |
Wald |
df |
p-value |
Odds ratio |
||||
LOW_INC
[1] |
1.254 |
.358 |
12.278 |
1 |
<.0005 |
3.504 |
||||
HI_BEN
[2] |
1.283 |
.455 |
7.938 |
1 |
.005 |
3.607 |
||||
INTERCEPT |
-1.976 |
.202 |
96.025 |
1 |
<.0005 |
.139 |
||||
|
|
|
|
|
|
|
||||
Likelihood
Ratio Tests for Individual Variables: |
||||||||||
|
-2LL of reduced model |
Chi-square |
df |
p-value |
|
|||||
INTERCEPT |
23.494 |
.000 |
0 |
|
|
|||||
LOW_INC |
34.971 |
11.477 |
1 |
.001 |
|
|||||
HI_BEN |
30.571 |
7.076 |
1 |
.008 |
|
|||||
|
|
|
|
|
|
|||||
[1]
LOW_INC (0 = Other income levels, 1 = Income not exceeding S$24,000) |
||||||||||
[2]
HI_BEN (0 = Other respondent, 1 = Beneficiary with income exceeding S$60,000) |
||||||||||
The main findings from Tables 10 to 13 may be summarised as follows. Firstly, TRAIN and OCC_FIN are important determinants of whether a respondent can perceive the absolute dollar distribution of the tax subsidies (Tables 10 and 11). However, neither variable is statistically significant as a determinant of the ability to perceive the income-disproportionate distribution of the tax subsidies (Tables 12 and 13). It does appear that many people do not think in income-proportionate terms when evaluating the distribution of tax subsidies, and that this is the case even for those trained in personal income taxation and/or whose work involve dealing with finance-related matters (including taxation). Secondly, there is some evidence (at the 10% significance level) that Malays are more perceptive (than non-Malays) of the absolute dollar distribution of the tax subsidies, but they are not any more perceptive of the income-disproportionate distribution of those subsidies.
Thirdly, there is no statistically significant difference between lower-income respondents and higher-income respondents with regards to their abilities to perceive the absolute dollar distribution of the tax subsidies (Table 10). However, low-income respondents (incomes up to S$24,000) are more perceptive (than higher-income respondents in general) of the income-regressive distribution of the tax subsidies (Note [3] in Table 12; and Table 13). This may seem somewhat surprising but there is a credible reason for it. Respondents with incomes up to S$24,000 do not pay any income tax as a result of the major tax reforms in the 1990s (described in Section 3). Regardless of the number of children they have, these respondents cannot possibly enjoy any benefit from the pronatalist tax expenditures. Neither do they stand to gain from any other tax expenditure. By reflecting on their own obvious position (i.e. that they clearly receive no benefits through the tax system), some of these respondents are then probably able to reason that it is conceivable for the rich to enjoy multiples of the little or no tax savings that accrue to the poor. In other words, by being so obviously deprived outright of all benefits delivered through the tax system, some of these respondents probably have developed over the years a heightened consciousness of the disproportionate bias of tax incentives in general towards the rich. In contrast, higher-income respondents (incomes exceeding S$24,000) generally are liable to pay income tax and are not automatically excluded from enjoying tax savings from the pronatalist or other tax incentives. Unlike the low-income respondents, higher-income respondents lack an obvious personal reference point that could have helped them perceive the regressive benefit distribution.
While the argument referred to above of a heightened level of consciousness amongst low-income respondents may have some truth today, it is unlikely that this was the case amongst low-income taxpayers back in the 1980s when the pronatalist incentives were being introduced. When these incentives were first introduced in the 1980s (i.e. before the tax reforms of the 1990s), virtually all working individuals in Singapore were paying income taxes and therefore very few of those targeted by tax expenditures back then would have been denied outright of at least some tax savings due to low incomes. In this regard, it is doubtful that all that many low-income taxpayers were perceptive of the income-regressive bias at the time when it mattered most during the 1980s when the incentives were announced and being debated.
A fourth and final point is that amongst high-income respondents (incomes exceeding S$60,000), the evidence suggests that the ability to perceive the absolute dollar distribution (Table 11) and the income-regressive distribution (Tables 12 and 13) of the tax subsidies is statistically significantly greater for those who are beneficiaries of the incentives. To sum up this and the previous points, respondents who are more likely to perceive the income-regressive benefit distribution fall into one of two very distinct groups: (1) low-income respondents who, by the design of the tax incentives and the subsequent structural tax reforms, are now excluded from enjoying any benefit; and (2) high-income beneficiaries who stand to gain the most from the incentives (Table 14 summarises the position).
TABLE 14: ABILITY TO PERCEIVE THE INCOME-REGRESSIVE BENEFIT DISTRIBUTION BY INCOME AND BY BENEFICIARY STATUS (FOR RESPONDENTS WITH INCOMES EXCEEDING S$60,000)
Income |
=$24K |
>$24K to $60K |
>$60K |
Total |
|
|
|
|
Non-
beneficiary |
Beneficiary |
|
|
No. (%) of respondents |
No. (%) of respondents |
No. (%) of respondents |
No. (%) of respondents |
No. (%) of respondents |
|
|
|
|
|
|
|
|
|
42 (89.4%) |
18 (66.7%) |
|
Not
perceptive |
35 (68.6%) |
159 (87.4%) |
60 (81.1%) |
254 (82.7%) |
|
|
|
|
|
|
|
|
|
|
5 (10.6%) |
9 (33.3%) |
|
Perceptive |
16 (31.4%) |
23 (12.6%) |
14 (18.9%) |
53 (17.3%) |
|
|
|
|
|
|
|
|
|
|
47 (100.0%) |
27 (100.0%) |
|
|
51 (100.0%) |
182 (100.0%) |
74 (100.0%) |
307 (100.0%) |
|
|
|
|
|
|
|
This paper has presented the results of an exploratory study into the extent
and determinants of the awareness of and perceptions
pertaining to pronatalist
tax policy in Singapore. While there appears to be a high level of awareness of
the existence of the tax
expenditures, there is widespread ignorance, if not
misperceptions, of the cost and distributive outcomes of these incentives. This
is so despite the over-representation in the sample of respondents who are
tertiary-educated, on higher incomes and in professional/managerial
occupations.
Few respondents recognise the spending implications of the tax expenditures and
that granting these incentives imposes
a hidden cost on the majority of
taxpayers who do not benefit from them. Similarly, not many respondents can
perceive the predisposition
of the incentives in favour of the highly educated
and the rich, or the hidden bias against the Malays. This is especially true of
the bias against the Malays, which is the least perceived of the three
distributive biases and in respect of which the Malays themselves
do not seem to
be any more perceptive than respondents of other races. The overall findings of
extensive ignorance and misperceptions
may well be the key to explaining public
acquiescence of the tax incentives and the relative lack of protest against
their introduction
in the 1980s when compared to the acrimonious reception given
to the other, non-tax and more visibly elitist measures. This study has not investigated directly the extent to which beneficiaries of
tax expenditures are accurate in their assessments of
the benefits they receive,
or the extent to which non-beneficiaries misperceive the benefits received by
beneficiaries. This therefore
could be an opportunity for future research.
However, the evidence from the present study is that beneficiaries
(particularly, high-income
beneficiaries who stand to gain the most) are more
perceptive than non-beneficiaries of the overall distribution of the tax
subsidies
by income levels and, to some extent, of even the ethnic bias of the
incentives. Beneficiaries are more likely to understand the
distributive
outcomes of the incentives than non-beneficiaries; moreover, non-beneficiaries,
quite obviously, are less likely to
be even aware of the existence of the
incentives. These findings highlight the potential of tax expenditures as a
politically useful
covert spending instrument to target and deliver benefits to
a select few while ensuring that the underlying distributive effects
remain
relatively imperceptible to the majority. The findings in this study also appear to illustrate the importance of public
debates and media publicity in shaping public opinion
on fiscal matters. The
highly visible and widely reported controversies and criticisms in the 1980s of
the eugenic and elitist slant
of the overall pronatalist policy left a sustained
impression on many of those who had experienced the episode. While the debates
in the 1980s had raged on the eugenic aspects of the overall policy, with public
resentment directed particularly at the primary
school registration priority
scheme for children of graduate mothers, specific references and objections to
the income-regressive
bias of the tax expenditures (which were an integral part
of the overall package of policy measures introduced in the 1980s) were
relatively muted. Insofar as public opinion on the distribution of tax subsidies
is concerned, it is certainly the case that the
overwhelming majority of people
think in terms of absolute dollar savings rather than from the fiscal
economists’ perspective
of income progressivity or regressivity. Not
surprisingly then, the survey findings reveal more than double the proportion of
respondents
being perceptive of the absolute dollar distribution of the tax
subsidies than of the distribution of the tax subsidies relative
to income. To sum up, this paper has provided empirical evidence of extensive ignorance
and misperceptions associated with the cost and distributive
outcomes of the
pronatalist tax expenditures in Singapore. The findings are a testimony to why
the tax expenditure route has proven
to be a politically expedient way for the
Singapore Government to implicitly pursue its policy of selective pronatalism.
There is
potential for more elaborate research to be undertaken in the future to
investigate how such ignorance and misperceptions directly
affect attitudes,
preferences and support for these incentives. There is also scope for future
studies, possibly experimental-based,
to investigate directly whether and, if
so, how benefits delivered through a tax expenditure are perceived any
differently from the
same benefits delivered through an equivalent direct
subsidy. Baron J and McCaffery E J (2004), Masking Redistribution (or its Absence),
University of Pennsylvania Law School, Institute for Law & Economics,
Research Paper No. 04-09 Bittker B I (1967), A ‘Comprehensive Tax
Base’ as a Goal of Income Tax Reform, Harvard Law Review, 80,
925 Buchanan J M (1967), Public Finance in Democratic Process: Fiscal
Institutions and Individual Choice (Chapel Hill: The University of North
Carolina Press) COP(2001a), Singapore Census of Population 2000 – Statistical
Release 1: Demographic Characteristics (Singapore: Department of
Statistics) COP(2001b), Singapore Census of Population 2000 – Statistical
Release 3: Economic Characteristics (Singapore: Department of
Statistics) COP(2001c), Singapore Census of Population 2000 – Statistical
Release 5: Households and Housing (Singapore: Department of Statistics) Collier P and R Luther (2003), The Use of Tax Regulations as Covert Political
Policy Variables: A Case Study of Tax Expenditures during
the Thatcher Years,
Financial Accountability & Management, 19(2), 159-84 Cullis J G and A Lewis (1985), Some Hypotheses and Evidence on Tax Knowledge
and Preferences, Journal of Economic Psychology, 6, 271-87 Cullis J G and P R Jones (1987), Fiscal Illusion and “Excessive”
Budgets: Some Indirect Evidence, Public Finance Quarterly, 15(2),
219-28 Dollery B E and A C Worthington (1996), The Empirical Analysis of Fiscal
Illusion, Journal of Economic Surveys, 10(3), 261-97 Enrick N L (1963), A Pilot Study of Income Tax Consciousness, National Tax
Journal, 16(2), 169-73 Enrick N L (1964), A Further Study of Income Tax Consciousness, National
Tax Journal, 17(3), 319-21 Francken D A (1986), Consumer Perceptions and Preferences for Public
Expenditure, Journal of Economic Psychology, 7, 179-95 Fujii E T and C B Hawley (1988), On the Accuracy of Tax Perceptions, The
Review of Economics and Statistics, 70(2), 344-7 Furnham A and A Lewis (1986), The Economic Mind (Hemel Hemstead:
Harvester Wheatsheaf) Gemmell N, O Morrissey and A Pinar (2003), Tax
Perceptions and the Demand for Public Expenditure: Evidence from UK Micro-Data,
European Journal of Political Economy, 19, 793-816 Gemmell N, O Morrissey and A Pinar (2004), Tax Perceptions and Preferences
over Tax Structure in the United Kingdom, The Economic Journal, 114
(February), 117-38 Gensemer B L, J A Lean and W B Neenan (1965), Awareness of Marginal Income
Tax Rates Among High-Income Taxpayers, National Tax Journal, 18
(September), 258-67 Hansen S (1983), The Politics of Taxation: Revenue without
Representation (New York: Praeger) Hood C C (1986), The Tools of Government (Chatham, New Jersey: Chatham
House) IRAS (1995), Annual Report of the Inland Revenue Authority of
Singapore Lewis A (1978), Perceptions of Tax Rates, British Tax Review, 6,
358-66 Lewis A (1982), The Psychology of Taxation (Oxford: Martin Robertson
& Company Ltd) McCaffery E J and J Baron (2004), Framing and Taxation: Evaluation of Tax
Policies Involving Household Composition, Journal of Economic Psychology,
25, 679-705 OECD (1996), Tax Expenditures: Recent Experiences (Paris: Organisation
for Economic Co-operation and Development) Oates W E (1988), On the Nature and Measurement of Fiscal Illusion: A Survey,
in G Brennan, B S Grewal and P Groenewegen (eds.), Taxation and Fiscal
Federalism: Essays in Honour of Russell Mathews (Sydney: Australian National
University Press), 65-82 Page B (1983), Who Gets What from Government (Berkeley: University of
California Press) Pechman J A (ed.) (1977), Comprehensive Income Taxation (Washington
DC: Brookings Institution) Poh E H (2004), Procreation Tax Incentives and the Decision to Have a Child:
A Survey of Married Couples in Singapore, Electronic Tax and Accounting
Briefing, 5(July), 1-14 Poh E H (2005), Pronatalist Tax Expenditures in Singapore: Quantifying the
Subsidies and the Disparity between Actual and Expected
Subsidies,
International Tax Journal, 31(1), 43-65 Puviani A (1903), Teoria della Illusione Finangiaria (Palermo) Report of the Royal Commission on Taxation (1966), (Ottawa:
Queen’s Printer) Salamon L M (1981), Rethinking Public Management: Third-Party Government and
the Changing Forms of Government Action, Public Policy, 29(3), 255-75 Salamon L M and M S Lund (eds) (1989), Beyond Privatization: The Tools of
Government Action (Washington DC: The Urban Institute Press) Sandford C (1993), Successful Tax Reform (Bath: Fiscal
Publications) Sausgruber R and J-R Tyran (2004), Testing the Mill Hypothesis of Fiscal
Illusion, University of Copenhagen Economics Working Paper No. 04-18 Saw, S H (2005), Population Policies and Programmes in Singapore
(Singapore: Institute of Southeast Asian Studies) Schokkaert E (1988), Fiscal Preferences and Fiscal Knowledge at the Local
Level, Journal of Economic Psychology, 9, 29-46 Shaviro D N (1990), Beyond Public Choice and Public Interest: A Study of the
Legislative Process as Illustrated by Tax Legislation
in the 1980s,
University of Pennsylvania Law Review, 139(1) Shaviro D N (2004), Rethinking Tax Expenditures and Fiscal Language, Tax
Law Review, 57(2), 187 Steel B and N Lovrich (1993), Causes and Consequences of Public Knowledge
Concerning State and Local Taxes, Paper presented at the Western Political
Science Association meeting, Pasadena, California, United States Steurle G (1995), Some Thoughts on the Status of Tax Expenditures, Tax
Notes, 68(Jul), 485-6 Sullivan M (2000), Tax Expenditure Budget, Now More Than Ever, Tax
Notes, 86(Feb), 1187-8 Surrey S S (1973), Pathways to Tax Reform: The Concept of Tax
Expenditures (Cambridge, Massachusetts: Harvard University Press) Surrey S S and P R McDaniel (1985), Tax Expenditures (Cambridge,
Massachusetts: Harvard University Press) The Straits Times (15.12.1986), Disaster if Sparse Birth Patterns Remain
Unchanged The Straits Times (2.3.1987), Affordability Level Counts, not the Education
Level Toder E J (2000), Tax Cuts or Spending – Does It Make a Difference?,
National Tax Journal, 53(3), 361-71 Traub S and C Siedl (1999), Framing and Judgments of Fairness in
Taxation, available at http://citeseer.ist.psu.edu/211022.html US Department of the Treasury (1984), Tax Reform for Fairness, Equity and
Simplicity (Washington DC: US Government Printing Office) Van Wagstaff J (1965), Income Tax Consciousness under Withholding,
Southern Economic Journal, 32 (July), 73-80 Wanna J (2003), Invisible Hands? The Non-Budgeting of Tax Expenditures in
Australia and Canada, Paper presented at the Australasian Political Studies
Association Conference Weisbach D A and J Nussim (2004), The Integration of Tax and Spending
Programs, The Yale Law Journal, 113(5), 955-1028 Winer S L (1983), Some Evidence on the Effect of the Separation of Spending
and Taxing Decisions, Journal of Political Economy, 91(1), 126-40 Woodside K B (1983), The Political Economy of Policy Instruments: Tax
Expenditures and Subsidies in Canada, in M M Atkinson and M
A Chandler (eds.),
The Politics of Canadian Public Policy (Toronto: University of Toronto
Press), chapter 8 Woodside, K B (1986), Policy Instruments and the Study of Public Policy,
Canadian Journal of Political Science, 19(4), 775-93 Zelinsky, E A (1993), James Madison and Public Choice at Gucci Gulch: A
Procedural Defense of Tax Expenditures and Tax Institutions,
The Yale Law
Journal, 102(5), 1165 [∗] Assistant Professor at
the Nanyang Business School of Nanyang Technological University, Singapore.
Email: aehpoh@ntu.edu.sg [1] For instance, based on surveys
of public knowledge of the tax system in the US, Page (1983), Hansen (1983), and
Steel and Lovrich
(1993) concluded that people know very little and that there
are widespread misconceptions about federal, state and local taxation.
In the
UK, Cullis and Lewis (1985), and Cullis and Jones (1987) found widespread voter
ignorance of sources of government revenue,
with income tax being the relatively
more visible tax. Schokkaert (1988) surveyed voter knowledge of local government
financing in
a Belgian municipality while in the Netherlands, Francken (1986)
examined public knowledge of government contributions to the costs
of providing
semi-collective public services in the form of legal advice, public transport
and hospital services. [2] Early studies into taxpayer
consciousness of their burdens under the US federal personal income tax include
Enrick (1963 and 1964),
Van Wagstaff (1965) and Gensemer et al (1965). A more
recent study, Fujii and Hawley (1988), found no evidence of systematic
misperceptions
of personal marginal income tax rates. Marginal income tax rate
consciousness was also the subject of a study in the UK by Lewis
(1978), who
reported an underestimation in general of the tax rates. Gemmell et al (2003)
relied on micro survey data to investigate
public consciousness of the
additional income tax and value added tax (VAT) liabilities that households had
to pay under certain
alternative specified changes in tax rates. They found that
there was fairly extensive ignorance of personal income tax and VAT liabilities.
Contrary to the usual fiscal illusion hypothesis, there was overestimation
(rather than underestimation) of tax costs, particularly
for the relatively less
visible VAT. [3] Studies (see Lewis, 1982; and
Furnham and Lewis, 1986 for reviews) generally have found a failure on the part
of voters to translate
public service benefits into tax costs and that there is
the lack of a fiscal connection between choices made on taxes and those
on
public spending. Winer (1983) suggests that the separation of taxing and
spending decisions characteristic of most representative
democracies may
generate fiscal illusion and contribute to increased public spending. [4] These studies use field data,
typically at local government level, to establish the validity of various
hypotheses of fiscal illusion.
For instance, the revenue complexity hypothesis
postulates that the complexity and lack of transparency of sources of public
financing
cause voters to systematically underestimate the tax price of public
sector output, in turn, leading to excessive demand for such
output and an
increase in the size of the public sector. See Oates (1988), and Dollery and
Worthington (1996) for a review of the
theory and empirical work relating to
this and other fiscal illusion hypotheses. The findings from such studies have
been viewed
with some scepticism for a number of reasons. They do not observe
the magnitude of voter misperceptions directly but rather rely
on highly
aggregated proxies to represent fiscal complexity and, by implication, the
extent of fiscal illusion. Furthermore, they
fail to show unambiguously that the
alleged distortion of fiscal choice is a consequence of fiscal illusion. As a
response to these
criticisms, some recent studies have relied on surveys (e.g.
Gemmell et al, 2003 and 2004) and experimental designs (e.g. Sausgruber
and
Tyran, 2004) in order to measure tax misperceptions directly and to incorporate
these into formal models of the demand for public
expenditures or models seeking
to explain voters’ preferences amongst alternative changes to the existing
tax structure to
finance public expenditure increases. [5] Using an experimental design to
investigate the causes and consequences of tax burden misperceptions, Sausgruber
and Tyran (2004)
found that the tax burden associated with an indirect tax is
systematically underestimated but no such systematic misperception occurred
with
an equivalent direct tax. They also found that in the indirect tax scenario,
illusion-prone participants were more likely, compared
to under the direct tax,
to approve of an expenditure (redistribution) proposal that was not in their
material self-interest. [6] The policy tools approach
entails the study of the political economy of policy tools, and advocates
shifting the focus of policy analysis
from individual programs and policies to
the generic tools or instruments of government action (Salamon, 1981). [7] Tax expenditures represent
legal entitlements that, once enacted, tend to be relatively permanent features
of the tax system divorced
from regular parliamentary review and budgetary
oversight. Unlike direct expenditures, tax expenditures do not compete for
fiscal
resources, are not subjected to annual reviews and authorisations, and
are infrequently subjected to efficiency/effectiveness audits.
The hidden nature
of tax expenditure spending erodes government accountability for the underlying
use of public resources. Furthermore,
the cost, level of benefits and
distribution of benefits under a tax expenditure program (other than a
refundable tax credit) are
inherently affected by changes in the incidence of
the tax, in turn, resulting from changes in the fiscal characteristics of the
taxpayers and/or in the tax rate structure or other provisions of the tax code.
Tax expenditures are, in this sense, open-ended commitments
since their costs
(and effects) can change even without any explicit program decisions taken to
amend the relevant legislation regulating
the tax expenditures. [8] The windfall effect results
from a tax expenditure rewarding behaviour that would have been undertaken
anyway even in the absence
of the incentive. Like many of the other alleged
deficiencies of tax expenditures, this effect is not necessarily peculiar to tax
incentives but may also apply in the case of direct subsidies. The upside-down subsidy effect (i.e. benefits are distributed amongst
beneficiaries in an income-regressive manner) arises in the
case of a tax
expenditure structured as a reduction in the tax base (or as a non-refundable
tax credit) in a tax system characterised
by a progressive rate structure. This
deficiency is overcome by structuring the tax expenditure as a refundable tax
credit. [9] The identification and
estimation of tax expenditures in tax expenditure reports in many jurisdictions
have since led to their increased
political visibility. Even so, tax expenditure
reporting in many of these jurisdictions has not been fully integrated into the
annual
budget appropriations process and systematic legislative review of tax
expenditures through the formal budgetary process has not
been widely instituted
(OECD, 1996). For a more recent paper on the situation in Australia and Canada,
see Wanna (2003). Notwithstanding the institution of tax expenditure reporting, if voters at
large continue to be unable to appreciate the spending
implications of tax
expenditures, politicians will continue not to be pressured into reducing such
hidden spending. Indeed, the experience
in the US has been that despite tax
expenditure reporting, tax expenditures have proliferated and very few of them
have actually
been eliminated (Steurle, 1995). [10] Zelinsky (1993) offers a
different view. He argues, from a US perspective, that tax institutions are
exposed to more competitive and
visible political processes given the numerous
and diverse constituencies they serve. This is unlike the specialised and
limited
clientele served by institutions that design and administer direct
expenditure programs. Accordingly, tax institutions are less susceptible
to
interest group capture and better positioned to make decisions informed by
expertise. [11] Shaviro (2004) refers to
this as ‘spending illusion’, i.e. the confusion between the actual
size of government and the
nominal dollar flows between the government and the
private sector that are denominated as ‘taxes’ and
‘spending’. [12] Sullivan describes the
Clinton Administration’s pervasive use of the tax system to implement
spending programs as ‘tax
expend’ liberalism and attributes such a
strategy to political expedience. Collier and Luther describe how the Thatcher Administration used tax
expenditures to promote the supply side changes that were integral
to its
economic and ideological policies. The significant reliance on tax expenditures
occurred despite the well-known stand of the
Conservatives in advocating tax
neutrality, reduced government intervention, and free-market forces, and despite
the advent of tax
expenditure reporting in the UK. The authors also emphasised
that the various programs, many of which conferred the most benefits
to the
richest individuals or in respect of the most expensive properties, would not
have been politically acceptable had they been
implemented through payment of
direct monetary subsidies. [13] There are, however, studies
exploring the cognitive biases at play in different framing situations that
compare a tax subsidy with
an equivalent policy framed as a tax penalty –
see Traub and Siedl (1999), and McCaffery and Baron (2004). [14] Such an approach is
recommended by advocates of policy tools analysis in order “to relate the
characteristics of tools to actual
program operations and outcomes”
(Salamon, 1989, p 261). [15] This observation was first
given political prominence by the then Prime Minister, Lee Kuan Yew, in his 1983
National Day Rally speech.
He noted the strong negative correlation between a
married woman’s fertility and her educational attainment. For instance,
based on 1980 population census data, women with little or no formal education
had 3.5 children on the average compared to just 1.6
children for women who were
university graduates. [16] See Saw (2005) for a
comprehensive account of the measures introduced. [17] Based on population census
data, the Chinese made up 77.7% of the resident population in 1990, compared
with 14.1% and 7.1% respectively
for the Malays and the Indians. In that same
year, the total fertility rate of the Chinese was 1.71 compared to those of the
Malays
and Indians at 2.73 and 1.99 respectively. Furthermore, the Chinese
(especially Chinese female university graduates) also had the
lowest marital
rate and, amongst those who did marry, had the highest age at the time of
marriage. [18] Although this aim obviously
was not explicitly stated as an objective of the pronatalist policy, the
Singapore Government did express
its observation and concern that the decline in
overall fertility was largely attributed to the fact that the Chinese were not
replacing
themselves (The Straits Times, 15.12.1986 and 2.3.1987). [19] In 2001, the minimum
educational qualification was three passes at the General Certificate of
Education Ordinary Level (GCE ‘O’
Level) examinations, or equivalent
qualifications. [20] ‘Earned income’
refers to the sum of business income, employment income, and taxable pension
income, reduced by any allowable
business losses. Mothers who were not
economically active would not have any earned income and, accordingly, would not
have benefited
from ECR. [21] See Poh (2005) for the
valuation of these tax subsidies expressed in their direct subsidy
equivalents. [22] See Poh (2005) for a more
elaborate discussion of the impact of the personal income tax reforms on the
pronatalist tax subsidies enjoyed
by couples on different incomes. [23] Many of these variables are
similar to those included in prior studies on fiscal knowledge and tax burden
consciousness. For instance,
gender and social class (defined in terms of
various occupational categories) were among the background variables considered
by Cullis
and Lewis (1985) in evaluating the level of voter knowledge of the
various sources of UK central government financing. Similarly,
gender, age,
education, income and occupational status were included in Schokkaert’s
(1988) analyses of voter knowledge and
awareness of local government financing
in Belgium. Gender, age and education were among the potential determinants
considered by
Fujii and Hawley (1988) in their study on marginal income tax rate
consciousness in the US. [24] Collinearity is not a major
problem in the final equation. The strongest correlation between any pair of
predictor variables is that
between LOW_INC and MID_INC (R=.389). [25] The relevant statistics for
the bivariate association between ability to perceive the ethnic bias and
ethnicity are as follows: N
= 318; Cramer’s V = .052 (p = .356);
Somer’s d (ability to perceive as the dependent variable) = .062 (p =
.404). [26] Respondents aged 40 in 2001
would have been 23 years old in 1984. The oldest respondents in the sample of
318 were 51 years old in
2001, i.e. 34 years old in 1984.
AustLII:
Copyright Policy
|
Disclaimers
|
Privacy Policy
|
Feedback
Conclusions and
Directions for Future Research
References
URL: http://www.austlii.edu.au/au/journals/eJlTaxR/2007/1.html