C ha p t e r

P ub lic C hoice
10 I n P ra c t i c e

Politicians are the same the whole world over. They promise to build a bridge even when there is no river.

Nikita Krushchev

Politicians and Public Policies ♦ Public Servants and the Supply of Public Services ♦ The Role of Special
Interest Groups ♦ General Causes of Government Failures ♦ Predictions of Public Choice Analysis ♦
Policy Implications

I
n this chapter we discuss how public policies are actually made and whether these
processes are likely to produce outcomes that enhance collective welfare. The analysis
focuses on the roles of three groups (politicians , public servants and special interest
groups) and the environment in which they work. The actions of members of these groups are
influenced by social norms, individual ethics and private incentives. On the other hand, the
public sector often offers opportunities for acquisition of private benefits or economic rent s.
And, certainly, some politicians and public servants as well as members of special interest
groups may pursue these benefits depending on the opportunities available and the constraints
imposed by the political environment. This environment includes the voting processes, the
requirement that politicians must get elected, the monopoly n ature of much bureaucratic
supply and the asymmetric nature of information in the public service.
In the first half of this chapter we examine how politicians, public servants and interest
groups may influence public policy. The second half of the chapter discusses general causes
of government failures, describes empirical results of public choice analysis and discusses
policy implications.

Politicians and Public Policies

In democracies, elected representatives make most important collective decisions. Th is
reduces the cost of making these decisions. But it raises the principal–agent problem. In
general, the issue is how can the principal A motivate their agent B to work in the interest of A
rather than follow their self-interest, when and where the objectives of the principal and the
agent diverge? In a democracy, how can citizen voters (the principals) ensure that their
elected representatives (their agents) represent the preferences of voters rather than their own
agendas? To answer this question, we need to identify the objectives of politicians, the
constraints on their actions and the likely outcomes of the representative process.
The public choice literature usually assumes that an elected representative seeks to
maximise their own utility subject to the constraint that he or she must be elected. A
168 Part 4 Public Choice

politician’s utility is assumed to depend on their personal position, income and power and, in
some cases, on achievement of social objectives. However, because these elements of utility
are difficult to measure, the literature usually assumes that a politician’s main objective is to
get elected. As a recent Australian Prime Minister, John Howard, remarked: ‘I am in the
business of winning elections’1 (not, it may be noted, in the business of maximising social
welfare)! The public choice analysis then examines how election processes affect collective
decisions and how freedom from election constraints may allow a politician to pursue their
personal agenda rather than represent the preferences of the voters.
There are two difficulties with this approach. One is the pejorative starting assumption that
politicians are largely out to further their own interests and, by implication, put little or no
value on social ideals. The second difficulty is that outcomes depend on the constraints on
politicians posed by the election process. Because election processes and voting rules vary
significantly across countries , outcomes may also vary. We consider below some implications
of two-party and multi-party models of legislatures. A key issue is whether all votes for
politicians are equally important or whether some votes are more important than others.

Two main parties and critical votes

Traditionally, many legislatures have featured two main parties. Examples include the House
of Representatives in Australia (Liberal and Labor), the House of Representatives in the
United States (Republicans and Democrats) and the House of Commons in the UK
(Conservatives and Labour). Notwithstanding the emergence of minority parties in balance of
power positions in Australia and the UK in recent years, dominant two-party systems may
emerge for several reasons. One is the winner-takes-all nature of elections. To be elected to
represent a constituency in the House of Representatives in Australia, a candidate must obtain
50 per cent plus one vote. If this does not happen in the first round of voting, the candidate
with least votes is eliminated. This process continues until one candidate achieves the
required vote. For the UK House of Commons, the winning representative is the candidate
who achieves most votes in the first round of voting. There are no prizes in either case for
coming second. Second, national parties require substantial organisation, marketing and
resources. Third, the party with most seats usually forms the government. A candidate who
seeks power must join a party that can win such a majority. 2
A key finding of public choice theory is that in a two -party state and under certain
conditions, the median voter has the decisive vote. This applies both to the election of
representatives and to public policy making. The median voter is the voter whose preferences
are in the middle of the distribution of preferences. In voting for representatives, the
conditions are that the winning candidate must attract 50 per cent plus one vote, voting is
compulsory and voter preferences are single-peaked about a single major issue (such as the
amount of government expenditure). Under these conditions, a candidate maximises his or her
vote by winning the support of the median voter. In Figure 10.1 a candidate to the right or left
of centre will not obtain the required majority. If this holds, this implies that both parties must
offer policies that reflect the preferences of the median voter.
There is much casual evidence for the median voter hypothesis. As Persson and Tabellini
(2002) point out, many welfare programs are oriented to assist the middle classes. Also ,
governments tend to favour labour market policies that protect the wages and working
conditions of the employed at the expense of unemployed workers who are not represented by
the median voter.

1
As reported in The Sydney Morning Herald, 1 March 2000.
2
T hese conditions do not always result in two main parties. Italy changed from proportional representation to the UK’s
first-past-the-post system, but still has a multi-party outcome.
 Chapter 10 Public Choice in Practice 169

Turning to econometric studies, as Mueller (2003) shows, many studies have attempted to
test the median voter hypothesis by estimating a cross -section demand equation of the
following form:
ln G = a +  ln tm +  ln Ym +  ln Z +  (10.1)
where G is government expenditure (e.g. on schools) in various jurisdictions, tm and Ym are
the tax price and income of the median voter and Z is a vector of preference parameters, such
as number of children in a household. Although many studies find that the coefficients α and
β have the expected sign and are statistically significant, similar results are obtained in
equations employing mean tax price and income variables. This suggest that the average
income voter is significant, but it is hard to distinguish between median and mean income.
In a different kind of study, Bruner and Ross (2010) analysed the voting patterns in two
popular referendums on expenditure in public schools in California where both high and low-
income households preferred low public expenditure and middle-income households favoured
high expenditure. In both cases , a majority of voters favoured low public expenditure.
However, in these cases, the voter with the median income was not decisive.
There are indeed several conditions under which the p references of a median voter may not
be decisive. First, people may vote for a representative based on party positions on several
issues, not just a single issue. In this case, a candidate may win by attracting votes from
swinging voters on several issues. This may include supporting some radical positions rather
than one central position. Second, with optional voting, many people not vote, especially
when there is little to choose between the parties and voting incurs time and travel costs.3
Thus, candidates may move away from the centre to attract people to vote. Third, and most
importantly, in ‘safe-seat’ electorates where one party is certain to obtain the required
majority, the elected candidate will be the person who wins party selection. In these common
conditions the successful party candidate must obtain the median vote of the party selectors.
Thus, the views of many elected representatives may be quite different from those of the
median citizen voter (as we saw in the same sex plebiscite in Australia in early 2018)! Fourth,
and critically, to win the most seats in the country each major party will aim to win the most
marginal seats. Therefore, each party targets the votes of swinging voters in marginal seats.
This may encompass only a small number of voters in a few seats. Accordingly, the political
No. of voters

Winning
candidate


 Losing
candidate

Social Median voter Conservative Position of

democratic candidates

Figure 10.1 Median voter and candidate positions

3
Voting is generally compulsory in Australia but optional in most other countries.
170 Part 4 Public Choice

parties may design policies with a view to winning these votes, which may not reflect median
voter preferences across the country.
A potentially more important perversion of the democratic process is the gerrymandering of
seats. Gerrymandering is the setting up constituency boundaries for political gain. This
typically involves establishing a few seats with large majorities of opposition p arty voters so
that they cannot vote in other areas. By doing this , a party can win a majority of seats without
a majority of votes. Where governments can determine political boundaries, as in the United
States, this can lead to undemocratic outcomes (as has happened). In Australia, where
political boundaries are determined by independent electoral bodies this is less likely.

More than two parties

As we noted above, even in traditional two-party legislatures, two dominant parties are not
inevitable. In other cases, multi-party legislatures may be common, especially with a
proportional representation voting system.4 Under this method, elected places are allotted in
proportion to the total votes cast. Proportional representation increases the number of parties
in a legislature, causes candidates to spread out over the issues and results in representation of
a greater number of views. However, even in this case some views may not be represented.
Even with multiple parties, outcomes may tend towards the policies preferred by the elected
representative who represents a central position. To win government, and to win votes for
major resolutions, parliamentary coalitions are required. Successful coalitions tend to be
dominated by central parties because they can attract and hold majority support more easily
than extreme parties can.
But on other occasions a party may win power or major votes with the support of a small
minority group. In Australia, minority parties have often held and exerted the balance of
power in the Upper House (the Senate). Following the 2010 election for the Lower House,
three independent members with a wide range of disparate policies held the balance of power.
In New Zealand, the small New Zealand First Party has held the balance of power on several
occasions since 1998, when a form of proportional representation was introduced into the
elections. In such cases a minority group, and sometimes only one or two individuals, has an
influence over public decisions that is wholly disproportionate to their electoral support.
Another common consequence of coalitions is instability. Because of the constant pressure
to negotiate the perks of power, coalitions are liable to break up and reform. This is akin to
the unstable cycling over gains that occurs in votes over distributional issues. Mueller (2003)
cites international studies that show that the duration of governments is negatively correlated
with the number of parties in the parliament and that the probability of a government falling is
positively related to the number of parties in the coalition that forms the government.

A politician’s freedom of action

An elected representative often has an opportunity to pursue their own personal interests.
Their freedom to do so depends on the power of th e government to make discretionary
decisions. This power depends in turn on the size of the government majority, which makes a
government safe from losing confidence votes and facing a new election, and the length of
time between elections. Also important is the ability of the voters to monitor government
actions, the scrutiny of government actions and the freedom of the media.
The length of time between elections varies. National legislature elections are held within
three years in Australia, four years in the United States and five years in the UK. In France, a
president holds office for seven years. But even three years allows some freedom between
elections. Political parties generally offer the public a package of policies. Voters have limited

4
Members of the Australian Senate are elected by proportional representation.
 Chapter 10 Public Choice in Practice 171

choice and mandates are not clear. Government may decide which parts of its policy package
it will implement and not infrequently it breaks election promises. After the 1996 election, the
Australian Prime Minister, John Howard, distinguished between ‘core’ election promises that
could not be broken and ‘non-core’ promises that were not real commitments! In contrast,
directors of private companies can be sued for failing to provide accurate information and are
accountable to shareholders formally once a year. More importantly, they are accountable
daily in the capital market. If a company’s share price falls, the company and the management
are vulnerable to takeover and change. By comparison, the constraints on politicians are
weak. Governments may even declare war without the consent of the governed.
This freedom from constraints invites corrupt practices. As Mill (1859) remarked,
There is little doubt, that if power is granted to a body of men, called representatives, they, like
any other men, will use their power, not for the advantage of the community, but for their own
advantage if they can.
Recent major books by Rose-Ackerman and Palifka (2015) and Cockroft and Wegener (2017)
document extensive corruption in government including in Western democracies.
Corruption arises partly because of the opportunities for politicians to enrich themselves at
the public expense. While corruption can be constrained by frequent elections, a free media
and rules and procedures for dealing with corruption (including imprisonment ), opportunities
for private gain in the exercise of public office occur in most countries. Moreover, the need to
finance elections creates incentives for delivery of large favours to campaign contributors in
any democratic system. When Time magazine asked Narasimha Rao (Prime Minister of India
1991–96), ‘How significant was the corruption issue in the election?’ Rao replied ‘Corruption
as an issue is there in all countries in all elections’. 5

Conclusions
In representative democracy, citizens have only indirect influence on public decisions. Often
there are only two or three main parties. When political issues can be represented by a one-
dimensional right-to-left political spectrum, a party may secure power by winning the support
of the centre and there is some evidence that parties focus on voters with preferences in the
middle of the distribution of preferences (the median voter). Moreover, to win elections,
parties often need to attract only a small proportion of swinging voters in marginal seats from
other parties so voters in marginal seats tend to have disproportionate influence over policies.
In addition, special interest groups (as discussed below) and politicians holding the balance o f
power also tend to have disproportionate influence over public p olicies. Finally, elected
representatives may have scope to pursue their own interests unless there are strong legal or
administrative constraints on them. Thus , overall there are many reasons why the preferences
of voters may be poorly represented in the political process.

Public Servants and the Supply of Public Services

To introduce our discussion of the behaviour of public servants and the impact of their
behaviour on the supply of public services, it is useful to distinguish three potential and actual
roles of public servants. They are expected to:

1. serve the public interest and that is what they do;

2. serve the elected government and that is what they do;
3. serve the elected government and/or the public interest, but in practice they also pursue
their own interests.

5
Time, 19 August 1996.
172 Part 4 Public Choice

Traditionally public servants in Australia, and other countries such as France and the UK,
were expected to represent the public interest and to provide objective and independent advice
to government. Indeed, the Public Service Act 1999 declared that the Australian Public
Service is ‘apolitical’ and the Act was designed in part to give public servants an obligation to
the public as well as to the government.
An alternative view and probably the dominant one today is that public servants should
implement the policies of elected representatives. Emy and Hughes (1991) express this view
in rather extreme form: ‘the public service is fundamentally a political instrument. There is no
public interest above and beyond the government of the day’. Politicians are accountable to
the electors, whereas public servants are not. In practice, in Australia public services are
increasingly politicised. Ministers rather than public servants generally decide senior public
service appointments and appoint people to carry out their wishes. Senior public servants no
longer have tenure of office and the independence of view and action that this allows.
Ministers also rely increasingly on political advisers rather than on expert advice. 6 Under this
view of government, public servants are first and foremost government agents and only
partially if at all independent servants of the public interest.
There remain some public agencies in which public servants have some autonomy.
Although governments often appoint judges, once appointed the judges are usually
independent. In Australia, many statutory authorities, such as universities, public trading
enterprises and independent commissions such as the Australian Competition and Consumer
Commission and the Productivity Commission, are established under legislation that provides
for only limited directions from ministers. However, these are mainly executive agencies
rather than policy-making bodies.
Be that as it may, how do public servants behave? The safest assumption is that the actions
of public servants are influenced by a combination of their contractual obligations to
government, their views of the public interest and, in part, by private interest considerations.
It is certainly possible to point to examples of corrupt behaviour by public servants in many
countries. As discussed below, some may pursue economic rents in the public sector or their
own advancement and this may affect how public services are supplied. But I would have to
say that from quite extensive experience, in state and local government especially, most
public servants are conscientious and have high integrity, though they are often conflicted
between carrying out political demands and their personal views of the public interest,

A public choice model of public supply

To model the impact of public servants on the supply of public services , assumptions are
required about the objective of public servants and the constraints under which they act. In a
classic article, Niskanen (1971) assumed that public servants have substantive private
interests and are motivated by the ‘3 Ps’—pay, power and prestige. Further, in order to
achieve these 3 Ps the public servant must acquire bureaucratic power. In particular, the 3 Ps
are generally correlated with the size of the budget controlled and the number of employees
managed. Therefore, a public servant has a strong incentive to maximise his or her budget and
output. This increases all the key elements in the public servant ’s utility function, except
perhaps a desire for a quiet life. Many public finance explanations of the supply of public
services draw in some way on these assumptions.
Of course, these assumptions, like those in any model, are a simplification. As we have
noted, many public servants hold altruistic social objectives. However, at the heart of the
model is the insight from principal–agent theory that principals and agents may have different
objectives. In this case, the principal is the elected government and the agent is the public

6
For a detailed analysis of similar trends in the United Kingdom see Foster (2005).
 Chapter 10 Public Choice in Practice 173

servant. Elected representatives contract public servants to carry out certain functions.
However, public servants may have different objectives, different information and some
freedom of action and may not always act as the government wishes. It follows that there are
many potential causes of inefficient output.
What are the constraints on public servants? Because they spend other people’s money,
they have no personal incentive to economise public expenditures unless constrained to do so.
However, public sector budget constraints are soft ones. A hard constraint exists where a
business must cease operating when it has run out of funds. A soft budget constraint is a
constraint that may be breached without serious consequences. In the public sector, funds can
be supplemented from consolidated revenue. Thus , the constraint on expenditure must be
some form of government control. To understand how government controls public
expenditure, we need to understand the environment in which government operates.
This environment has three main characteristics. First, most government departments are
monopolies. Ministers must obtain most, if not all, services from their department. There is
usually little direct competition within government or from outside. Government departments
often fight fiercely for funds and territory but rarely compete directly in provision of services.
This avoids wasteful duplication. But, without competition, there is little incentive to achieve
efficiency and only weak measures of it. Second, there are rarely ready measures of the value
of non-marketed public services. The scale of government activities, such as police or fire
services, may be known but their value hard to measure. In this world, more inputs often
count as more outputs. Third, information is often asymmetric, especially in large
departments. Senior public servants generally know what the government wants. But the
government has little idea of the real costs (or benefits) of the services. Ministers obtain most
of their information from public servants. But public servants have no incentive to reveal the
real costs to ministers, while they often have an incentive to exaggerate the benefits.
Economic models of government output based on these characteristics are illustrated in
Figure 10.2. In this figure, the quantity of services is shown on the horizontal axis and their
total value and cost on the vertical axis. Total benefit rises with output, but at a declining
marginal rate. Cost is assumed here to rise linearly with output, implying a constant marginal
cost of supply. The efficient level of output, where social net benefit is maximised, is QE . At
this point marginal benefit equals marginal cost (the slopes of the total benefit and cost curves
are equal). Higher output is inefficient because marginal cost exceeds marginal benefit.

$ total cost
Value / cost

Total benefits

QE QF QG QH Quantity of service

Figure 10.2 Non-marketed output in a public service model

174 Part 4 Public Choice

Actual output depends on the behaviour of public servants and ministers. A common
prediction is that output will rise to QG , which is the point where the total benefit of output
equals total cost. Public servants are assumed to seek the highest level of output for which
they can obtain funds. On the other hand, ministers are expected to be able to understand
approximately the total benefit and cost of a program and to be unwilling to fund expansion
beyond QG . Only public servants can measure the marginal cost and benefit of a program and
they conceal this information from elected representatives. However, by controlling the flow
of information to ministers, public servants can expand output from QE to QG .
Other behavioural assumptions and levels of output are of course possible. Ministers may
not know even the total benefit and cost of a program. Public servants may exaggerate the
benefits of programs; for example, teachers have an incentive to emphasise the benefits of
smaller class sizes. If ministers believe that the benefit curve is higher than shown in Figure
10.2, the equilibrium quantity would move to the right from QG towards QH . On the other
hand, public servants may favour a higher cost structure than that shown in Figure 10.2. This
could reflect inefficient production methods, rent seeking (salaries in excess of opportunity
costs) or a simple preference for a quiet life (the traditional perk of a monopoly). In any of
these cases, the cost curve would shift upward and the equilibrium output, where total benefit
equalled total cost, would fall below QG .
In this model, the actual supply of public services depends on the outcome of bilateral
monopoly negotiations: the government is sole purchaser of a department’s services and the
department is sole supplier of the services. The outcome depends on negotiations between th e
two. Also, other departments (notably including the central Treasury agenc y) may support or
oppose the department supplying the service. As in most bargaining models, various
predictions of output are possible.
In summary, this model of public service output (often described as the Niskanen model)
suggests that government is likely to oversupply public services, to supply the kinds of goo d s
favoured by public servants and to adopt relatively high -cost methods of production. The
classic TV series Yes Minister and numerous anecdotes suggest that the model contains much
reality. Khursheed and Borcherding (1998) and Mueller (2003) provide useful summaries of
empirical tests of these predicted outcomes. As shown later in Chapters 16 and 18, there is a
fair amount of evidence that government supplies services at a higher cost than does the
private sector. In the absence of user charges , it is harder to test whether public services are
oversupplied or whether the wrong kinds of services are supplied. Empirical studies have
produced a variety of results. For example, Staaf (1977) found th at salaries in the US public
school system were linked to education budgets and the size of school districts (supporting
the Niskanen model). And it may be remarked that there appears to be a s trong correlation
between the growth of student numbers in Australian universities and the size of vice-
chancellor and other managerial salaries. On the other hand, Johnson and Libecap (1989)
found no evidence of a general relationship between agency growth and bureaucratic salaries.

The Role of Special Interest Groups

Special interest groups play an important role in most economies. A special interest group is
any organised group of people which works to further the common interests of its members.
Interest groups exist in many forms. Employer and employee groups were traditionally
powerful interest groups. Examples include trade associations (e.g. for the coal, motor vehicle
and pharmaceutical industries), farming associations, professional associations (such as for
doctors and lawyers) and trade unions. Sometimes a company is sufficiently powerful to
constitute a significant interest group on its own. In recent years, other influential interest
groups have emerged, including consumer and environmental groups, and welfare
associations to promote child care and the interests of the elderly.
 Chapter 10 Public Choice in Practice 175

Most interest groups aim to influence public policy. In some cases, the aims are idealistic
such as protection of wildlife or provision of benefits to disadvantaged groups in the
community. However, many interest groups seek to increase the income of their members,
often at the expense of other sections of the community. They may do so by seeking
government regulation of the market or by direct transfer of income to their group.
Lobbying the government to obtain higher than normal returns through regulations is
known as rent seeking. Regulations may control entry into the industry, output, prices or
advertising, competition from imports and so on. They may also require firms to employ only
members of unions or workers with particular qualifications, for example locally obtained
qualifications. Interest groups (employers and employees) aim to obtain such regulations to
gain economic rent by reducing competition and increasing the market power of their
members. Rent seeking typically has two economic costs: it leads to restrictions on output and
it uses resources without providing any output. 7 Rent seeking may also result in inequitable
distributions.
Special interest groups often influence public decisions in a way that is disproport ionate to
the size of their membership because of their ability to mobilise financial support for political
parties. Political parties require very large funds for both their ongoing organisation and
elections. Twenty years ago, The Economist reported that annual spending in political
elections in the United States exceeded $3 billion. Election to a seat in the Upper House (the
Senate) cost an average of $5 million.8 Under the Citizens United (Supreme Court) decision
in 2012 in the United States, political spending is protected speech under the First
Amendment of the Constitution and government cannot restrict contributions to support
candidates at elections.
Interest groups also derive influence from their ability to deliver votes. Some interest
groups such as unions, environmental and seniors ’ groups can deliver large number of votes.
The American Association of Retired Persons has 40 million members. When voting is
optional, members of interest groups that are offered preferential policies have more ince ntive
to vote than other citizens. A related factor is the ability of some interest groups, such as large
firms and ethnic groups, to deliver a concentration of votes that may influence the outcomes
of particular seats. In these cases , the votes of protected workers are likely to have more
influence on policies than votes of consumers and taxpayers who are more widely dispersed.
Evidently special interest groups can affect both the quantity and price of goods. Industry
protection for primary producers, manufacturers and professional services is common in
many countries. As Mueller (2003) shows, protection tends to be greatest in concentrated
industries and in labour-intensive industries which can establish effective and powerful
interest groups. Another interest group that has been effective in many countries is the elderly
who have obtained significantly increased pensions and health care benefits in recent years
(Persson and Tabellini, 2002). Both traditionally, since it was established in 1871, and
recently, the National Rifle Association of America has had extraordinary, and literally
deathly, influence in promoting and retaining the gun industry in the United States.

General Causes of Government Failures

We have described some of the resource misallocations that are likely to arise from the
political process, the public supply of goods and interest groups. However, this discussion,
which is typical of the public choice literature, has not provided a systematic analysis of why
government may fail to allocate resources efficiently similar to the analysis of market failure
in Chapter 4. In this section we discuss some systemic features of the supply of non-market
goods that are likely to result in a misallocation of resources —due essentially to the lack of a

7
For further discussion of rent seeking costs, see Chapter 14.
8
The Economist, 8–14 February 1997.
176 Part 4 Public Choice

market. Some points have been mentioned above and other points are taken up in later
chapters, especially Chapter 16, which discusses the supply of non -market goods.9
A key feature of non-market goods and a major cause of resource misallocations is the
absence of property rights. In the absence of property rights over public revenue or
expenditure, the whole government budget becomes effectively a common property resource.
It presents an opportunity for rent seeking by those who can exert the greatest po litical
strength. Most citizens want a share of the rent. But, because the rent is a common property
resource, individuals have more incentive to exploit it than to conserve it. A major related
problem with non-market goods is the absence of prices. Prices convey critical information
about the value of goods. Government has to estimate the value of non -market goods without
this information. When outputs are measured by inputs (as government output is measured in
the national accounts), more inputs may be regarded as more output regardless of whether this
is the case.
In the absence of prices, the demand for public expenditure is ‘decoupled’ from the cost.
In markets, the benefits of consumption are coupled to expenditure. In the public sector,
individuals can obtain benefits at no personal cost. This creates a ‘free riding’ environment in
which individuals can demand that services be supplied without paying for them. In effect
they are gaining rents. When a service is free, there is almost always excess demand for it and
pressure for more supply. Similarly, with regulations, individuals or groups can lobby for
regulations that raise their incomes without paying the costs associated with the regulations.
On the other hand, the absence of a pricing mechanism creates a need to ration supply, which
provides public servants with opportunities for corrupt behaviour. The decoupling of benefit
and cost is most extreme with income redistribution programs, where one group of people can
lobby for programs to be paid for by other people. Moreover, because of the high discount
rate of politicians facing elections, politicians may satisfy the demand for services or other
benefits now regardless of future costs that tend to be discounted.
On the supply side there are also major failures due to the lack of markets. First, single-
source production is often inefficient because it is not subject to competition. In the absence
of competition, evaluation of the cost and quality of the output is difficult. Second, when the
revenues that fund a service are decoupled from the costs of supplying it, there is no clear
benchmark for efficiency and the scope for misallocation of resources is greatly increased.
Third, in the absence of direct financial constraints on the supply of a servic e, there is no
termination mechanism.
A fundamental challenge is the wide variety of views of fairness and justice in society.
Fairness may be viewed as equal total income or as equal wages per hour of work, as equality
of opportunity or as fair rewards for work, as assistance for low-income earners or more
generally as assistance for disadvantaged groups . Equity may mean horizontal equity (treating
like individuals in the same way) or vertical equity (individuals should pay tax based on their
capacity to pay tax) and so on. Citizens (voters) may well have differing views about fairness
for these are ethical views on which individuals may differ. However, the variety of views
about equity creates considerable potential for inconsistent government actions.

Predictions of Public Choice Analysis

Our analysis has indicated several general causes of government failures and inefficient
resource allocation. Does public choice analysis provide specific predictions about
government behaviour and economic outcomes? In this section we examine some effect of
politics and bureaucracy on short-run macroeconomic policy, the size of government, the
allocation and distribution of resources , and methods of production.

9
T his section draws on Wolf (1988), especially Chapters 3 and 4.
 Chapter 10 Public Choice in Practice 177

Elections, macroeconomic policy and the business cycle

A common public choice hypothesis is that elections influence the business cycle. The
presumption is that, to win elections, government presents popular (high expenditure/low tax)
budgets before elections. Fiscal and monetary policies are designed to maximis e economic
growth and minimise unemployment before elections. If there is spare productive capacity,
increases in prices lag increases in output and any large inflationary effects are experienced
after the election. Governments therefore increase spending, including transfer payments, and
government deficits before elections, and increase taxes after elections. These policies create
a business cycle.
This political business cycle hypothesis is based on one or more strong assumptions: that
voters do not recall the overall performance of the government in office; that voters are not
concerned about short-term budget deficits; or that voters are not concerned about possible
inflationary consequences of short-run increases in aggregate demand.
Mueller (2003) reviewed empirical studies of the political business cycle hypothesis. Many
studies cited support the political business cycle hypothesis, with increasing expenditures,
lower taxes and increased deficits just before elections found in the United States, Can ada and
Japan, and more generally in OECD countries. See, for example, Bhattacharya and Wassmer
(1995) and Reid (1998). Crosby et al. (1997) also found that government expenditure rose
before elections in Australia. A casual review suggests that Australian government
expenditure also increased above trend before each of the elections in 2001, 2004, 2007 and
2010.
Another prediction from public choice theory is that macroeconomic policies generally
reflect the ideological preferences of parties. Right-of-centre parties are expected to favour
price stability over reducing unemployment. Left-leaning parties to have reverse priorities.
This prediction assumes that parties can follow their own preferences and promote the
interests of their main constituencies rather than pursue the support of the median voter (or
the marginal swinging voter), in which case their policies would converge. Support for this
prediction is strong in the United States. Between 1952 and 1988 Republican administrations
reduced the inflation rates and increased the unemployment rates that existed when they took
office, whereas Democratic administrations increased inflation rates and reduced
unemployment rates. Similar outcomes have been found in Europe (Mueller, 2003). However,
Crosby et al. (1997) found no evidence of such partisan influences over the macroeconomy in
Australia. And in recent times with close elections there has been considerable policy
convergence especially in the 2010 election.

Total government expenditure

As we saw in Chapter 2, total public expenditure in Australia rose substantially in relation to
GDP from 1960 to 1985 but has been broadly constant around 35% of GDP since then. We
also discussed there how economic factors influenced these outcomes. We now consider the
role of political factors. These factors include (1) the demands of voters and special interest
groups for more free or subsidised public services, such as health services, and transfer
payments and (2) the role of representatives and public servants in sup plying more goods or
income.
There are several demand-side political explanations of the growth of public expenditure.
One is based on the political importance of the median voter. When income distribution is
skewed with top income earners receiving very high incomes, the median income is less than
the mean and the median voter prefers an increase in tax rates because their tax contribution
will be less than the value of public services he or she receives. Providing that income
redistribution does not create too many disincentives, politicians can win the support of the
majority by promising to redistribute income to less well-off households. Also, in many
countries low-income and other disadvantaged groups have become more powerful politically
178 Part 4 Public Choice

as the vote was extended to more groups and eventually to all adult citizens. Increasing
inequality in market incomes in many countries (see Chapters 20 and 34) has also increased
demands for income transfers.
Another possible explanation for rising public expenditure is that special interest groups
tend to seek increased public expenditure on goods and transfers (see Olsen, 1982). These
groups include (1) producer groups of all kinds (farmers, miners, importers, artists and so on)
who seek subsidies and (2) consumer groups who seek more public funds for education,
health, child care and aged care and so on. These interest groups often have more effective
power than the mass of taxpayers who may favour lower taxation , although we note below
some exceptions in the use of democratic power in Switzerland and the United States
However, the demand for more services and benefit transfers is not limited to interest groups.
It is a widespread phenomenon and a natural corollary of the free-rider characteristic of many
public services, such as local hospital, educational or transport services, that people lobby for
services funded out of consolidated revenue and provided to them at no cost.
On the supply side, politicians gain by providing projects to their constituents funded from
consolidated revenue. And, as we have seen, public servants may gain higher incomes from
managing larger budgets. Chapters 16 and 18 provide some support for the view that in -
house production of services increases project costs and hence government expe nditure.
Another popular supply-side explanation for the growth of public expenditure is the ratchet
or displacement effect generally attributed to Peacock and Wiseman (1961). Peacock and
Wiseman argued that governments like to spend public money but are constrained by their
perception of what the public will bear. In crises such as war the public will bear higher taxes.
The expansion of government-funded services increases the tolerance of the public to higher
levels of taxation. After the crisis, public expenditure continues to displace private
expenditure, as it is racheted up to a permanently higher level. There is international and
Australian evidence for this hypothesis , though the evidence is now mostly historic.

Fiscal illusion is another possible explanation for the rise in government expenditure. Fiscal
illusion occurs if voters do not understand the impacts of government fiscal actions. For
example, there may be fiscal illusion about the results of pump -priming expenditure to
promote economic growth before elections. These extra goods and transfers may be paid for
after the election by inflation or by lower government spending. More generally, whatever the
cause(s) of inflation, unless tax rates are indexed to allow for price changes, inflation
increases taxes as taxpayers move into higher tax brackets and in effect experience real tax
increases. Australia, like most other countries, does not index tax rates.

Evidence. Drawing on Mueller (2003), various studies support the view that political fa ctors
influence the size of government spending. First there is evidence that public expenditure falls
when voters have more control over the budget. In a study of public expenditure in 110
cantons/municipalities in Switzerland, Pommerehne and Schneider (1982) concluded that
public expenditure was 28 per cent higher in the 62 cantons that were governed by elected
representatives and had no direct voting on any major issues than in the 48 cantons which
operated under direct democracy. Funk and Gathmann (2011) found likewise that direct
democracy reduced canton spending, though by less than had been previously estimated.
Public expenditure is also much lower in New Hampshire, where there are constitutional
constraints on public spending, than in neighbouring s tates in the United States. Perhaps mo s t
famously, the popular plebiscite vote in California, Proposition 13, under which property
taxes were capped and halved represents an example of the power of the mass of people
rebelling against higher taxes and pres umably, by implication, higher public expenditure.
Second, several studies have shown that interest groups affect the size of government (e.g.
Mueller and Murrell, 1985 and 1986). Based on a cross-sectional study of 12 OECD
 Chapter 10 Public Choice in Practice 179

countries, Lybeck (1986) reported that government size was significantly influenced by the
degree of unionisation, the number of public employees and by unemployment. He also found
that the size of government in relation to GDP in Sweden varied over time with the fraction of
employees who were members of interest groups. Plotnick (1986) reported that income
support in the states of the United States depended on the size of the pro -welfare interest
groups.
Third, public expenditure is generally a smaller proportion of GDP in federal state s than in
unitary states. The inference is that voters have more power in a federated country where
there is also more competition between states. Blankart (2000) found that public expenditure
in Germany rose as it became more centralised.
In Australia, Hackl et al. (1993) reported that, in addition to conventional economic
variables such as income and population, several political factors explained changes in
government expenditure. The political factors included interest group effects, the
bureaucracy, the political complexion of government, the size of the budget deficit, tax share
and price inflation. Moreover, the ratchet effect of war has had a major and sustained impact
on the level of government expenditure, notably after the Second World War during which
the Commonwealth obtained income tax powers.

The allocation of public expenditure and regulations

Much public choice literature suggests that both policy and the allocation of public
expenditure can be predicted from models of the political process . It predicts, for example,
that governments will deliver regulations and goods that are attractive to interest groups
(especially to narrowly concentrated groups), to voters in marginal seats and to swinging
voters especially, who can provide party finance or critical votes. Because of the strategic
voting position of the median voter, government will provide programs of special benefit to
the perceived median voter. More generally, government will redistribute income towards
middle-income households.
In Australia, governments have regularly provided programs for special interest groups.
Examples are tariff protection for the textile and clothing and motor vehicle industries,
accelerated depreciation allowances for primary producers which reduces their tax liability,
and licensing requirements for professional groups such as doctors, pharmacists and lawyers.
Before elections, Commonwealth and state governments invariably announce major
expenditure projects in marginal seats. At all times, the Commonwealth g overnment provides
substantial support for middle-income earners in health care, education, home ownership,
child care subsidies and so on. Although overall government tax and expenditure programs
reduce income inequality (see Chapter 20), there is considerable support for middle-income
groups and marginal seats.
Worldwide, probably the most generous industry support goes to agricultural interests. Half
the whole European Union budget goes to agricultural producers. This is striking testimony to
their political power. Lopez and Pagoulatos (1994) show that industry rents from tariff
protection in the United States are a significant positive function of the size of political
donations. Goldberg and Maggi (1999) show that non -tariff protection is also a positive
function of an industry’s campaign contributions.

Methods of production
The public choice literature predicts that both politicians and bureaucrats will prefer public
production of government output. Government ministers may favour public ownership of the
means of production because financial transfers can be made to politically favoured groups in
hidden ways, for example by cross -subsidies. Bureaucrats may prefer public production
because it expands their area of responsibility, which in turn enhances the 3 Ps (pay, power
180 Part 4 Public Choice

and prestige). The fact that public production is often preferred to private despite higher costs
(see Chapters 16 and 18) provides some support for these predictions in the literature.

Corruption in government
The public choice analysis of government also predicts that some corruption is likely. The
monopoly power that politicians and bureaucrats exercise over the provision of many services
enables them, unless accountable, to extract a monopoly rent from supplying the service.
Corruption is an extreme form of rent seeking and can result in an extreme misallocation of
resources. In a survey of the literature, Gruber (2016, pp. 266-269)) shows that corruption is
associated positively with electoral systems such as proportional represen tation where voters
elect slates of representatives rather than individuals (thus reducing accountability) and the
degree of red tape for business operations (which increases the benefits of bribery to
business), and negatively with the wages of public servants (who have less to lose from being
punished for corrupt practices). Rose-Ackerman (1999), Rose-Ackerman and Palifka (2016)
and Cockroft and Wegener (2017) document large amounts of corruption in many countries.10
Australian politicians are not immune from corruption. Over the last 20 years several
Australian politicians have been jailed for corrupt practices. 11

Policy Implications
Evidently, for various reasons government may misallocate resources and fail to provide the
services that citizens want. We discuss below how government decisions can be made more
responsive to citizen preferences. In Chapter 16 we discuss the related issue of how public
services can be supplied efficiently.

Direct democratic measures. The most direct ways to increase citizen control over public
policies and services are by constitutional controls or by various direct voting processes.
Some writers, for example Milton Friedman, have argued for constitutional rules that would
allow citizens to trade across national borders without government restrictions or that would
enshrine limits on government expenditure or budget deficits. However, moves along these
lines to date have been limited and had mixed success. The US Congress has made several
attempts to reduce the US budget deficit by legislation that would bind both the President and
Congress. However, Rosen and Gayer (2014) observe that Congress has been skilful at
circumventing the budget caps. Recently Congress classified expenditure over US$100 billion
spent on the wars in Afghanistan and Iraq as emergencies and thus not to count under the cap .
Nearly all states in the United States have also introduced rules in their constitutions that
forbid operating deficits. This requires clear distinctions between operating and capital
revenues and expenditures, which is not always easy. There is also the issue of enforcement.
How does a legislature deal with a situation where an operating budget surplus is planned but
not delivered? It is hard to test whether the statutory caps have been effective because the
capped outcomes may reflect more fiscally conservative legislators rather than the statutes
themselves. However, Auerbach (2008) found that the fiscal rules did have some impact on
deficit control.
Direct voting procedures include voting for major representative positions, recall elections,
referendums and voter initiatives.12 A recall election is a special election initiated by citizens

10
See also the regular reports of Transparency International, a non-profit organisation established to promote clean
government.
11
Jailed state government politicians included Burke and Parker (an ex-premier and ex-deputy premier respectively
in Western Australia), ‘Buckets’ Jackson, who had been Minister for Corrective Services in New South Wales,
“Lunch-a-lot” Macdonald and mate Minister Eddie Obeid (also NSW) and Nuttall in Queensland (for 12 years).
12
T he US information in this paragraph is drawn from Gruber (2016).
 Chapter 10 Public Choice in Practice 181

with the aim of replacing a sitting elected representative. In the United States, 18 states allow
the recall of state officials and 36 states allow the recall of local officials. A referendum
allows citizens to vote on state laws. All US states allow legislatures to invite a popular ballot
and 24 states allow citizens to collect enough signatures to require the legislature to take a
popular ballot. Twenty-four states also allow voter initiatives that allow citizens to place their
own legislation on the ballot for voters to accept. Many cantons in Switzerland also have a
tradition of referendums. There is no such tradition in Australia. To be effective, a plebiscite
must be binding on government, not simply advisory. There is no doubt that these various
means of direct democracy have had a significant political impact in the United States, where
among other results it led to the election of Arnold Schwarzenegger as Governor of California
in 2003, and in Switzerland.
However, whether direct voting improves public resource allocation is an open question.
According to The Economist (19 March 2011) California is ‘now widely studied as an
example of what to avoid’. Seventy-five per cent of the state’s budget is outside of the
government's control, the current budget deficit is US$25 billion and ‘the roads and colleges
are crumbling’. These are challenging observations. A fuller and wider review of the impacts
of direct voting might provide a more balanced conclusion about the merits or otherwise of
direct voting and more especially when it is beneficial and when not.

Indirect methods of increasing citizen control. There are many indirect ways to increase
voter control over policies. One strategy would be to adopt political processes that encourage
more political parties to emerge. This would allow more representation of voter preferences
and introduce more checks and balances into the political process, as occurs in the Senate in
Australia. On the other hand, an increase in the number of parties may reduce the stability of
government.
Probably a more effective way to improve voter representation is by dece ntralising
government and encouraging competition between governments. Several studies (for example
Zax, 1989; Oates, 1989) have shown that public spending falls with greater fragmentation of
political jurisdictions. A federal state encourages competition between the constituent states
in taxation and the provision of services. Citizens can choose their preferred public package
by moving to another jurisdiction (voting ‘with their feet’). Competition provides an incentive
to state legislators and bureaucrats to discipline their spending. Providing public services at
the most local level of government compatible with efficient production helps to ensure that
voter preferences are recognised in the provision of services.
As we have seen, the cost of elections is another major issue with implications for the
power of special interests over resource allocation. The standard policy responses to this
include increased public financing of election campaigns, greater disclosure of contributions,
cost controls on elections and more enforcement of electoral rules. However, there are
apparently no such limits in the United States.
Another electoral issue is the transparency of information provided by government to
voters. Under the Australian Charter of Budget Honesty, government must provide forward
spending and revenue estimates with each budget and a mid -year outlook. After an election is
called, the Secretaries of Treasury and Finance must sign off on revised budget estimates and
issue an economic and fiscal outlook. These requirements make the existing government more
transparent and accountable. New governments no longer have the excuse that they ‘did not
know how bad things really were’. The National Commission of Audit (1996) proposed that
the Secretaries of Treasury and Finance also provide independent economic reports at budget
time and half yearly, but given the nature of the government/public servant relationship this is
impractical and has not been implemented.
182 Part 4 Public Choice

Tax hypothecation, the process whereby tax revenu es are raised for specific projects, is
another way to increase the transparency of public decisions. However, treasuries tend to
argue that all tax revenues should go into consolidated revenue and then be allocated
efficiently to priority projects (see Chapter 28).
Finally, two other strategies should be mentioned. One involves greater use of statutory
agencies and officials. These are agencies and officials that are charged by statute to carry out
various functions with given objectives, including acting in the public interest, and who once
appointed are subject to limited ministerial control. This would reduce day -to-day political
decision making in these designated areas.
A more general strategy for increasing voter control over public programs is to p rovide
consumers with choices by subsidising use of services rather than supply. For example,
government can provide parents of school children with educational vouchers that can be used
to purchase school services, including services from private suppliers. People requiring health
care can be given the choice of health care provider. This gives citizens more power over the
use of resources. However, politicians and public servants often resist this strategy because
they perceive that it dilutes their control.
In summary, there are many government failures and consequential resource misallocations
—and no single solution. However, there are many possible strategies that, taken together,
can increase voter control and mitigate the extent of the resource misallocation.

Summary
• In representative democracy, citizens have limited • When the price of a service is zero, demand exceeds supply
influence on collective decisions. Often there are only and there is pressure for an inefficient increase in supply.
two or three main parties. Elected representatives rarely • Government failures in supply also include the difficulty of
represent a broad cross-section of the voters. measuring output, single-source production, weak measures
• Median voters and swinging voters in marginal seats and of efficiency and soft budget constraints on the supply of a
special interest groups tend to have disproportionate service.
influence over policies. Also, elected representatives • Empirical studies suggest that election politics influence
often have significant scope to pursue their own interests. macroeconomic policy and possibly the business cycle. Also,
• Public servants also have an interest in the supply of political factors strongly influence the size and composition
government goods. In the absence of competition, this is of government expenditure. Public choice theory also
likely to lead to an oversupply of government goods, a predicts that corruption in government is likely.
supply of goods preferred by public servants and public • Government decisions can be brought under greater citizen
means of production. control in various ways. Direct methods involve more
• Special interest groups often have a disproportionate citizen control over legislation. Indirect methods include
influence on policies because of their ability to deliver decentralisation of public services, more limits to funding
funds and votes to politicians. of elections, greater transparency of government actions,
• Underlying these political issues is the systemic nature of increasing the number and responsibility of statutory
non-market goods. Public income can be viewed as a agencies and offices and providing more choice of services
common property resource. Individuals can obtain to users of government services.
benefits at no personal cost.
 Chapter 10 Public Choice in Practice 183

Questions
1. Should public servants serve elected politicians or the 8. Draw on public choice theory to explain why most
public interest? Or is this an immaterial question governments fail to index tax rates.
because they will serve their own interests in any 9. Assume that a government department consistently
case? provides its minister with an underestimate of the
2. How do public servants influence the supply of public total costs of a program. Using the Niskanen model
services? of public output, show how this is likely to lead to a
3. What are the main systemic causes of government level of output in excess of the efficient level.
failure? 10. Assume that government supplies a publicly financed
4. Does the financing of elections give special interest good with a total cost function C = 40 + 10Q. The
groups too much influence? If so, what might be done total community benefit is given by B = 20Q – 0.5Q2 .
to curb this influence? What is
5. How can the hypothesis that government’s i. the maximu m level of output that can be supplied
macroeconomic policies are influenced by political without a net loss to the community?
(re-election) considerations be tested? What is the ii. the efficient level of output?
evidence, if any, that electorally induced
macroeconomic policies influence the business cycle? 11. European Union (EU) countries frequently violate
the EU rule that deficits be kept below 3 per cent of
6. What is the evidence, if any, that governments GDP despite the formal sanctions of large fines
oversupply public services in aggregate or that they (which are not actually imposed). Is it feasible and
supply the wrong kinds of public services? desirable to have legislative rules that govern the size
7. What evidence is there that direct voting affects of government deficits?
government revenue or expenditure? What 12. Why are economic rents a feature of government?
conclusions, if any can be draw about the effect of And why do economic rents encourage corruption?
direct voting on social welfare?

Further Reading
Funk, P. and Gathmann, C. (2011) ‘Does direct democracy M ueller, D. (2003) Public Choice III, Chapters 10 to 22,
reduce the size of government? New evidence from Cambridge University Press, Cambridge.
historical data, 1890-2000’, Economic Journal, 121, Persson, T. and Tabellini, G. (2002) ‘Political Economy and
1252-1280. Public Finance’, pp. 1551–1658 in Handbook of Public
Gruber, J. (2016) Public Finance and Public Policy, Chapter 9, Economics, Vol. 3, A. Auerbach and M . Feldstein (eds),
M acM illan Education. Elsevier, Amsterdam.
Khursheed, A.F. and Borcherding, T.E. (1998) ‘Organising Rose-Ackerman, S. and Palifka, B.J. (2016) Corruption and
government supply: the role of bureaucracy’, Chapter 2 in Government: Causes, Consequences and Reform,
Handbook of Public Finance, F. Thompson and M .T. Cambridge University Press.
Green (eds), M arcel Dekker, New York. Wolf, C. (1988) Markets or Government, Chapters 3 and 4,
M IT Press, Cambridge, M A.
184