Scholarly article on topic 'Yes, Minister: The Impact of Decision-making Rules on Geographically Targeted Particularistic Spending'

Yes, Minister: The Impact of Decision-making Rules on Geographically Targeted Particularistic Spending Academic research paper on "Political Science"

Share paper
Academic journal
Parliam Aff
OECD Field of science

Academic research paper on topic "Yes, Minister: The Impact of Decision-making Rules on Geographically Targeted Particularistic Spending"

Parliamentary Affairs (2014) 67, 935-954 Advance Access Publication 10 April 2013


Yes, Minister: The Impact of Decision-making Rules on Geographically Targeted Particularistic Spending

Jane Suiter1 and Eoin O'Malley2*

1School of Communications, Dublin City University, Dublin, Ireland;

2School of Law & Government, Dublin City University, Dublin, Ireland


This article tests a theory to explain particularistic political spending not normally used on parliamentary systems. Using constituency-level data, we evaluate the merits of theories predicting whether parties reward their core voters or target floating or swing voters to maximise the party's electoral return. In order to bring new insights into the process of parliamentary pork in a system which incen-tivises garnering a personal vote, we introduce the decision-making rule (ministerial autonomy) as a variable and argue that the level of ministerial discretion in allocating funds coupled with the electoral system's incentives leads, in Ireland, a

to a form of distributive authoritarianism where the interests of ministers trump a

those of their party. u

1. Introduction

The distribution of public benefits is the essence of politics, and the electoral connection with distributive spending is regarded as axiomatic (Lasswell, 1936). This often means that parties will target distributive spending on groups of voters. While some spending is targeted at social groups, such as the middle classes, spending is also geographically targeted, a phenomenon known as pork-barrel politics. Pork-barrel spending is generally thought of as important for the re-election of certain candidates, but it can also help a party achieve or maintain a majority in the parliament. Furthermore, it could be used to maintain party unity (Lyne, 2008; Carroll and Kim, 2010) or to advance certain public policies (Bertelli and Grose, 2009). Thus, pork projects are visible, expensive and particularistic: that is, targeted to a politician's preferred locale or constituency (Diaz-Cayeros et al., 2006).

# The Author 2013. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For permissions, please e-mail:

This article seeks to explain why pork-barrel spending is targeted at certain geographic areas and not others. We contend that the decision rule, i.e. how the decision as to where money is spent is made, will explain much of the variation in the geographic distribution of public resources. Furthermore, we look specifically at the importance of the executive in parliamentary democracies and argue that the decisive actor in most established parliamentary democracies is not based on median legislator or the committee chair, as the US-dominated literature does, but the controlling minister (Laver and Shepsle, 1996). By studying non-presidential systems we question the generalisability of existing theories and add a useful new data set in this area. Ireland is a useful case to test theories of 0

particularistic spending because there is a strong incentive for garnering the per- n

sonal vote. It is also a case where the effective decision-maker is easily identifiable: d

the responsible cabinet minister. In turn because ministers in Ireland are also ff

senior-elected politicians who must defend their own seats they have clear and |

predictable incentives in targeting spending. Using data on two spending pro- P

grammes that are clearly geographically based but where the level of ministerial p

autonomy varies, allows us to see if variation in this decision rule explains vari- x

ation in where the spending programmes target. This can be used to evaluate d

existing theories of pork-barrel spending, which we discuss next. We then set §

out the justification for our case selection and provide information about it. s The fourth section describes the data, its collection and the method of analysis. Section 5 provides the estimation and results, which provide support for our hypothesised relationship.

2. The determinants of pork

Much of the original work on pork-barrel spending sought to describe its presence rather than explain variation in pork (Key, 1949; Mayhew, 1974; Fenno, 1978). Later explanations for pork-barrelling behaviour use the different incentives among electoral systems to cultivate a personal vote (Cain etal., 1990; Carey and Shugart, 1995). According to Carey and Shugart where the ballot structure allows voters choose between party's candidates, those candidates have an incentive to cultivate a personal vote. Thus, the USA with open primaries and candidate-based voting will strongly incentivise personal vote cultivation. However, evidence of geographically targeted spending (hereafter GTS) aimed at electoral gain has been found around the world, almost regardless of the electoral system. While there is even some evidence of geographically based representation in the hyper-proportional cases of Israel and the Netherlands (Latner and McGann, 2005), fiscally based geographical representation tends to be more common in presidential systems in the Americas. However, Australia (Denemark, 2000), Brazil (Ames, 1995, 2001), Canada (Milligan and Smart, 2005), France

(Cadot et al., 2006), Germany (Stratmann and Baur, 2002), Japan (Thies, 1998; Horiuchi and Saito, 2003) and Scandinavia (Dahlberg and Johansson, 2002; Tavits, 2009) among others, all show GTS. The absence of empirically supported explanations of cross-country variation (Milesi-Ferretti et al., 2002 is a possible exception) is understandable because identifying and measuring what is Pork and what it is not, is rarely clear.

If the electoral system was the first explanatory variable used by scholars to explain the existence and extent of GTS, then a related literature attempts to explain where GTS goes. Cox and McCubbins (1986) ask whether politicians

direct resources to their core supporters, to the opposition or to swing voters 0

and find that politicians will favour those voter blocs that promise higher rates n

of return on their investment. These higher rates, they say, are invariably d

associated with core supporters, easily identified from previous voting behaviour. ff

A different perspective, based on persuasion and conversion rather than mobil- |

isation of core voters, was laid out by Lindbeck and Weibull (1987), who offer 5

a stylised theoretical account arguing for the logic of targeting swing voters. p

They assert that the expected electoral return of any given transfer is maximised x

when outlays are directed at swing voters. Transfers to loyal supporters or to com- d

mitted opposition voters cannot be expected to affect voting choices, as these §

voters' choices are generally between abstention and voting for their party. s

Both theories' competing expectations have been corroborated in different research and generally results are mixed. Some studies of distributive benefits in the USA find that parties target pork projects to swing districts (Bickers and Stein, 1994, 1996), while other studies find a bias toward core supporters (Levitt and Snyder, 1995; Ansolabehere et al., 2007).

McGillivray (2004) resolves these seemingly conflicting hypotheses by specifying under which condition each is likely to occur. The outcome will be due to interaction of the electoral system with the party type, conceived as the variable ability of the party leaders to impose direction on legislators (McGillivray 2004, p. 45). Where weak parties exist there is an expectation that bills receive support from both sides of the House (Weingast et al., 1981), that is leaders cannot direct legislators, and parties are not cohesive. McGillivray hypothesises that in a system with weak parties, and under a majoritarian electoral system, distributive funds will be targeted at safe districts which elect the more senior legislators. She asserts that in countries with electoral systems with single-member districts, where political parties are internally cohesive, legislators will target local public goods to marginal or swing electoral districts (see Table 1).

Much of this literature tends to be silent on the mechanism through which GTS is distributed, assuming the party leadership or median legislators will matter most. But the institutional rules that allow pork to be distributed are of great importance. In the USA it was found that legislators in more influential

Table 1. Electoral districts likely to receive most resources under alternative electoral and party arrangements

Governing parties

Strong Weak

Electoral system Candidate based Marginal districts Safe districts

Party based Party strongholds Party leader bailiwicks

Adapted from Golden and Picci (2008, p. 273) with permission from John Wiley and Sons.

committee positions are often more successful in targeting their home districts with additional funds, through their agenda setting powers (Shepsle and Weingast, 1987). And more generally we see fiscal transfers are linked to institutional rules (Baron and Ferejohn, 1989). For instance Hallerberg and Marier (2004) found that the decision rule has an impact on budgetary spending in Latin America. But, except in presidential systems, the mechanisms through which GTS is delivered are ignored.

In most parliamentary democracies, the executive or cabinet is generally thought to be much more powerful than legislatures (King, 1975). If governments make decisions on particularistic spending, how are these made and what incentives do the relevant actors who make the decisions have? There are various theories about where power lies within government (Elgie, 1997) and while Laver and Shesple (1994) argued for ministerial autonomy, there was limited empirical support for this assertion in their own volume. However, nor was there much support for the classical notion that cabinet is a collective decision-maker. If we assume ministers have some autonomy, where will they target particularistic spending?

This will depend on the minster's incentive structure. While all ministers have some electoral goals, in some places they are clearer and stronger than others. In some countries ministers are part of the parliament and will have seats of their own to defend—some of which may be far from secure. However some ministers, for instance in the UK, will have security of re-election by virtue of having already secured their party's nomination in a 'safe seat'. Other ministers, who are not parliamentarians, may have no personal electoral needs, but may look to the greater needs of the party. The outcomes set out in Table 1 show how the electoral system and party system are contingent on each other. Table 2 sets out that in candidate-based electoral systems (we do not think that variation in ministerial autonomy will have a variable impact on outcomes according to party strength and cohesion), where ministers have autonomy, they will look after their own needs, which will often be to target spending at their own constituencies and those of

Table 2. Electoral districts likely to receive most resources under alternative electoral systems and decision rules

Ministerial autonomy

Strong Weak

Electoral system Candidate based Minister's bailiwicks Marginal or swing districts Party based Party strongholds Party strongholds

The cells set out a probabilistic tendency rather than a deterministic law.

their supporters. Where ministers have less autonomy, the party leadership will expect or demand that the party's specific interests in re-election or gaining a majority will trump the personal needs of the minister (who of course will also want his or her party to maintain or improve its strength in parliament) and so should focus on marginal or swing districts. Next we outline the distinctive aspects of the Irish case and set out the particular incentives and expectations for GTS.

3. The Irish cases

Ireland provides diverse cases where the case selection allows us to hold constant all but the key independent variables we are testing (Gerring, 2007, p. 98). In particular, the spending programmes we study vary on the level of ministerial autonomy giving us some leverage over this variable. It is a useful case to test theories of particularistic spending in which we expect to observe the hypothesised behaviour for a number of reasons. First, the electoral system, the Single Transferable Vote strongly incentivises intra-party competition and personal vote gathering (Marsh, 2007). This is further facilitated by the small size of constituencies. Each Teachta Dala (TD or MP) represents just over 20,000 people, and in 2007 between 7000 and 8000 first preference votes were usually enough to see a candidate elected (Gallagher, 2008, p. 79). During campaigns, candidates stress how much they have done for the constituency, while opponents concentrate on how particular incumbents have failed to deliver. Local service is top of many TDs' priorities. When asked how strongly they emphasise service to their local constituency, 64 per cent of 2007 candidates said that they emphasise this service strongly (Marsh et al., 2008).

Unusually for a political system with strong incentives to garner a personal vote, the party in the parliament is strong in the sense that TDs will almost invariably vote with their party whip (Gallagher, 2010). The electoral organisation tends to be local, at the individual level, but government TDs normally

only rebel if their constituency is being subject to specific measures which will harm it.1

Secondly, the effective decision-maker in Ireland, the minister, has clear and predictable incentives in targeting spending. As in all parliamentary democracies cabinet is formally charged with making government decisions, which if necessary, are then usually approved by the parliament. However, cabinet is made up of individual ministers responsible for policy areas. There is then some variation in the degree to which ministers have autonomy—or can be policy dictators in their domain. Ireland is comparatively unusual in that all Irish ministers are elected members of the lower house.2 And because ministers nearly always 0

defend their seats, ministers will compete with incumbent TDs from their own n and other parties for re-election.3 Additionally, voters' expectations of ministers

for the delivery of public goods may be higher than for backbench TDs. Crucially, f

there is a clear link between the party's performance in a constituency and the |

ability of the senior TD in that constituency's likelihood of appointment to min- P

isterial office (O'Malley, 2006). Nor can ministers easily switch constituency as p

most are local notables with a long attachment to an area. So while ministers x

have a clear incentive to secure a majority for their party and to distribute d

spending widely, there is a stronger incentive to deliver to their own bailiwick. §

Ministers must temper this latter incentive, as being seen to be too locally fo- s

cussed can cause reputational damage. An example of this in 2012 was where the Minister for Health was found to have added two projects in his own constituency to an independently devised list of projects. This led to the resignation of a junior minister in the department.

Having the incentive to promote one's constituency does not mean one has the ability. But, the Irish case is also appropriate because the nature of the governmental system means that ministers are better able to actually deliver government spending to constituencies. The Irish executive is among the most dominant in Europe and within that there is a hierarchy, with the Minister of Finance and the Taoiseach (prime minister) clearly at the top. In part because of the overlap of party leadership (in parliament and party) with leadership in the executive, the Irish executive dominates the Dail (lower House) to a greater extent than in most other European countries (O'Malley and Martin, 2010). There are also many formal rules which guarantee executive dominance. Irish legislators cannot create temporary coalitions for specific spending because the government

'The current coalition is an exception to this, where most resignations have been as a result of national issues.

2There is provision for two (unelected) senators to be made minister, but this is almost never used.

3Ministers usually resign from ministerial office when they reveal they will not seek re-election, something which almost caused a constitutional crisis in 2011.

retains a constitutionally binding sole power to initiate spending or allocate the state's revenue (Article 17.2). The government also controls Dail time and its agenda through the House's Standing Orders. This means that opposition parties or backbenchers are limited to proposing motions, which may have no impact except to send signals to voters. For instance, an opposition party might call for the government to guarantee that a certain hospital in an area is not closed.4 Such motions are not binding on the government, but are designed to force opponents to vote in such a way that might be electorally embarrassing.

Within government, Irish ministers are considered to be given a good deal of autonomy (Farrell, 1994). The rules of government decision-making as set out in 0

the Cabinet Handbook (Department of the Taoiseach, 2006, p. 19) necessitate that n

'[p]roposals requiring a Government decision should be the subject of a memo- a

randum from the responsible Minister'. Only the responsible departmental min- ff

ister can make proposals in her area. Other ministers whose domains may overlap |

or have a departmental interest should be consulted about the decision, but there P

is evidence that many decisions are not thoroughly examined, with the cabinet p

acting as a decision affirming institution (O'Malley, 2012). x

So the major restriction on ministerial autonomy is not the cabinet, but the d

Taoiseach (prime minister), the Minister for Finance and other coalition party §

leaders. The control the Taoiseach has over the careers of ministers ensures 1

that ministers will generally be responsive to his direction, though he has fewer formal policy-making powers than one might expect (O'Malley and Martin, 2010). The political realities of coalition politics also means that the opinions and needs of government party leaders, who are invariably also members of the cabinet, must also be met. The Minister for Finance must sanction all spending, and does so through the estimates procedure (Department of Finance, 2008, p. A4, 9-16). Finance has some powers to ensure that spending is within the 'subheads' agreed by the Oireachtas (parliament). Within a 'sub-head', once a scheme or programme is established, ministers usually have discretion as to where money goes, subject to compliance with procedural requirements.

The schemes used in this study are subheads of the estimates and so ministers have some discretion over spending. Crucially, there is variation in the decision

4An example of such a motion was voted on in Dail Eireann on 6 October, 2010. The motion read; "That Dail Eireann, noting with concern:

— that there are more than 46,000 adults and children on hospital waiting lists across the country, which is 5,400 patients more than last year;...

calls on the Government to:...

— suspend the loss of frontline health services and capacity at hospitals, for example at Clonmel, Merlin Park, Nenagh, Roscommon, Navan, Sligo, Letterkenny, Portiuncula, Wexford, Monaghan, Ennis and Louth County; and

— immediately open the €16 million community hospital facility in Dingle, Co. Kerry."

rule and in the key independent variable, ministerial autonomy. In some areas the 'vote' authorised by parliament is quite specific and sets down guidelines for how it should be administered. The responsible minister has much more discretion in sports allocation grants than in education grants because education grants are subject to specific rules and because the objective need for investment in education varies greatly. Relying solely on the McGillivray's analysis we would expect that GTS would go to swing voters or constituencies. If we add the nature of decision-making we get a different, or at least a refined sense of the direction of GTS, one where powerful ministers will first look after their own home districts. This leads us to the following hypothesis:

Where ministerial discretion is high responsible ministers will direct higher levels of expenditures to their home constituencies, and those of other influential ministers. Where ministerial discretion is lower, relatively more money will be targeted at marginal constituencies.

4. Data and methods

In general, scholars of particularistic spending tended to focus on infrastructure spending or social welfare policy, or indeed geographically targeted discretionary grants such as sports grants (Denemark, 2000). The 2007 Irish Candidate Study reveals the areas or which candidates themselves believe they can claim credit. When asked what sort of local benefits TDs are most likely to claim credit for, many of the candidates mentioned more than one category and these guided our case selection decision. Almost 33 per cent mentioned schools and, in particular, school buildings, amounting to some 22 per cent of all mentions. Just under that (29 per cent) mentioned sports facilities and lottery grants, making up about 19 per cent of total mentions. Thus, sports clubs and schools are the top two areas that legislators themselves believe it is worth claiming credit for. This accords with elected TDs' behaviour in an analysis of parliamentary questions raised in the Dail (Martin, 2011). These programmes also varied in the level of ministerial autonomy, as discussed below.

We use two separate detailed data sets to test these hypotheses, one on sports capital grants and the other on primary school building programme in Ireland over a sequence of six calendar years from 2002 to 2007. The Sports Capital Grants scheme was set up in 1988 funded exclusively from National Lottery income. It was initially a multi-annual scheme, but since 1998 it has been allocated annually. For these monies the National Lottery Act (1986) gives the Minister for Finance discretion to decide the amount that will go to for each 'purpose', i.e. sports, arts, etc., but these funds are then left to relevant departments for dis-bursal. Since 2005 the scheme is also funded through the normal 'Vote' allocated

by the Dail for each Department's expenditure. Though there are criteria for the disbursal of funds, these are vague, and the minister has the ultimate decision. The fund was soon subject to controversy and there was an early (failed) attempt to ensure it was not used for what one TD described it as 'a re-election slush fund' (Dail Debates, 8 November 1988, vol. 383, col. 2120).

The second spending programme ensures variation in the decision rule. The School Building Programme, administered by the Minister for Education, is designed to facilitate the building and refurbishment of school buildings. It is administered by the Planning and Building Unit of the Department and has responsibility for planning education accommodation provision and managing the 0

capital funding allocated by the government each year for upgrading, replacing n

and expanding school and third-level college buildings and infrastructure. This d

includes the purchase of sites, provision of new buildings (including furnishing f

and equipping) and extending and refurbishing existing buildings. The overall |

amount is agreed with the Department of Finance each year, and the Minister P

for Education allocates the funds, but there are guidelines setting out priorities p

for the programme, and changing demographic needs will trump most other x

factors. Interviews with the protagonists involved in these programmes revealed d

that while this was an important programme for TDs, the ability to 'deliver' Pork §

was more imagined than real. Because of this we expect that relevant ministers' s

impact in this area will be lower than that of the sports grants where the minister is much less constrained. Both sets of grants are demand driven rather than formula-based aid programmes. So recipients must apply for grants. As we are §

interested in the funds delivered to a constituency we see no reason why this §

should bias our results. 3

The data used encompass two general elections (2002 and 2007), with up to three changes of minister, allowing for some variation in each data set. The data for each scheme are at the level of the individual grant and are allocated to the relevant constituency. The data were obtained under Ireland's Freedom of Information Act (1996, amended 2003) where details of every individual grant in the appropriate years was applied for. The resulting Sports Capital Grants data set contains entries for some 4444 different grants, amounting to €403 million over the period. The data contain the name of the sports club, its address and the amount it won. The Department issued data that break down the grants by county. However, in general, constituency and county boundaries are not coterminous. Indeed, some constituencies contain two counties, while Dublin has 12 constituencies within it, electing 47 TDs. Using a constituency map and address information, we allocated each individual grant application to the constituency in which it is located. Considine et al. (2008) provide some evidence that, from 1999 to 2007, the counties of the Minister for Arts, Sport and Tourism received the highest per capita allocation of Sports Capital

Grants in all 26 counties, even after changes in office holder. However, this bivari-ate analysis was done on the basis of county rather than constituency, and lacked the necessary controls.

The School Building Programme data set incorporates all capital spending on the primary education sector in the Republic of Ireland from 2001 to 2007. The original data set contains the name of each school, its address, the amount of each individual grant and its purpose. It also records the type of school (mixed sex or single sex, type of patron, e.g. church) as well as the number of children enrolled. Each individual school is then allocated to its constituency—these changed in

years where the boundaries changed. q

Because both data sets included large numbers of items, from minor repairs to n

small schools building to funding for stadia, and because the research question relates to constituency-level spending, the data were then collapsed to the per ff

annum constituency level amounting to some 255 observations for each pro- |

gramme [three years (2002-2004) with 42 constituencies and three years P

(2005-2007) with 43 constituencies]. So each observation represents the p

amount of money spent on the programme in each constituency per year. Collaps- x

ing the data in this way also removes any problems with measurement at different d

levels as the independent and control variables are all at the constituency level.5 §

The data thus vary greatly with observations for any one constituency in a es

given year varying, in the case of sports facilities from €56,000 to €4.58 million and for schools capital spending, varying from €283,684 to €2.26 million. As the spending variable is not normally distributed we use the log of the grant. Every constituency in every year was awarded funds and thus there is no censoring of data. This is a fixed effects panel data set; however, due to issues of co-linearity where some of the population level variables do not vary much over time, we have implemented random effects panel estimation. This does mean that we cannot control for time-constant unobserved heterogeneity. Panel data will often correct (inflate) standard errors. This is because each additional year of data is not independent of previous years. Thus, we may now have serially correlated error terms uit, and standard errors are biased. However, in this data set there is low serial correlation: Cor[yt, yt- 5] = 0.13. We are utilising use cluster-robust standard errors as this allows for heteroskedas-ticity and general correlation over time for given i. The two-way effects model allows intercept to vary over i and t:

Yit = bXu + a + Uit + Sit

5These two datasets are thus entirely new sources for scholars and will be made available by the authors.

Because Cov(xit, Vi) = 0 or close to, the bias of the random effects estimator will be low and this is no longer a significant problem.

Each data set has six periods (years) and is unbalanced as it contains 38 constituencies which did not change and have four constituencies which were abolished and five constituencies which were introduced mid-way through the period. However, random effects estimation techniques can deal with unbalanced data. We also argue it is appropriate to utilise dummy variables for the time effects.

The central empirical strategy is to regress per capita spending for each policy area by each electoral district and year on measures of political factors, together

with economic and demographic controls that may influence funding decisions. 0

Stein and Bickers (1994) argue that the number of grants rather than the level of n

grant are a more appropriate dependent variable. However, given the structure of d

the data where any one project can be the recipient of a large number of grants, as f

parts of the project are allocated monies in successive years, we argue that the |

overall amount is the key variable of interest. The variables used are set out in P

Table 3. Our major expectation is that the variables that capture the political p

actors who control the decision-making process will be significant. We also x

expect that there will be some evidence of a more general electoral calculus d

with core and/or swing voters likely to be targeted to some extent. The decision- §

rule variables are dummies which capture whether the constituency was repre- s§ sented by the relevant minister, the finance minister, the Taoiseach or any other cabinet minister. The variables designed to test for the party core and

swing voter models measure the percentage support for the main government §

party, which controlled all the relevant ministries, Fianna Fail; and marginality— §

how close Fianna Fail was to winning or losing a seat.7 3

In addition, we control for the various socio-demographic features of the §

constituencies in which the investments are made. These include the population O

in the constituency, as it is likely that additional funds and greater resources M

could be expected to be targeted at constituencies with larger numbers of residents. §

We also include the proportion of the population under 18 as both schools and 2

sports facilities are likely to be in greater demand in areas with proportionally 5 greater number of young people and voter turnout. Unemployment, expressed in terms of the proportion of adults registered as unemployed, and the proportion of adults educated to a minimal level are included, in order to control for areas of disadvantage which are generally the specified target for additional Exchequer

6We excluded small coalition parties because they are captured by the cabinet minister variable. When included in the model they were not shown to be significant.

7Other variables, the Fianna Fail seat proportion and whether Fianna Fail had lost or gained a seat in the constituency were also used, but these were not significant and were removed because of multicollinearity.

Table 3. Variables in the analysis

Variable name Description

Unit of Expectation

measurement (direction/ significance)

Ln Grant Logged grant allocation for each €Ln per n/a

constituency by year constituency

Core support

FF vote Vote for Fianna Fail (FF governing Raw percentage + /ns


Marginal seats

Swing An absolute value of difference Interval data + /ns

between number of seats and between 0 and 1

number of quotas. The higher the

figure the more unsecure the

seat/or closer to a target seat

Decision rule

Education Is this the education minister's 1 /0 dummy + /s

minister constituency?

Sports minister Is this the sport minister's 1 /0 dummy + /s


Taoiseach Is this the Taoiseach's (PM's) 1 /0 dummy + /ns


Finance minister Is this the finance minister's 1 /0 dummy + /s


Cabinet minister Is this any other cabinet minister's 1 /0 dummy ns



Urban Does the constituency contain a city? 1 /0 dummy —/ns

Turnout Turnout at previous elections Percentage + /ns

Education Percentage educated only to age 15 Percentage —/ns

Unemployment Numbers unemployed Raw numbers ns

Under18 Numbers under 18? Raw numbers + /s

Population Population of constituency Raw numbers —/s

resources. There is also an expectation that rural TDs are more likely to lobby for additional resources and engage in clientelistic practice (O'Leary, 2011) and thus we also include a dummy variable denoting whether a constituency is urban or rural.8 This is to capture whether rural ministers are likely to respond differently to demands for particularistic goods than urban ministers. Monroe and Rose (2002) argue that urban interests tend to be worse at translating political support into effective representation, a position that is supported by anecdotal evidence in Ireland where some interviewees suggest that rural TDs are better at delivering grants.

8All Dublin constituencies and those with large urban areas in Cork, Limerick and Galway are all coded as urban.

5. Estimation and results

We present results of two sets of models: one with spending on sports as the dependent variable and the other with money spent on schools buildings as the dependent variable.9 The main independent variables of theoretical interest are those measuring the political influence of legislators in constituencies, these are the dummies for the Minister for Finance and Minister for Education or Sports as well as the dummy for other cabinet ministers. Also of interest is a variable which measures the strength of the governing party's vote in each constituency thus testing the core voter hypothesis. A variable which captures the swing voter hypothesis is included (see description in Table 3). °

Our expectation is that the signs on the variables measuring the decision- O

making cabinet ministers will be positive (see Table 3). The hypotheses are e

tested cumulatively and so GTS directed at core voters, or party strongholds o

is tested first for its significance on its own; then the marginal voter hypothesis h

is tested, following by the decision-rule hypothesis. All three competing hypoth- /

eses are then tested against each other. Results appear as Columns 1, 2, 3 and 4, .

respectively, in Tables 4 and 5. xfo

Examining first sports grants in Table 4, the variable measuring the percentage o

constituency vote for the governing party is not statistically significant. This does g

not mean that no additional grants were targeted at core voters or credit claimed O for them, but rather that the targeting was not systematic or large enough to show

up in this model. Another variable testing the core voter model proposed by Cox ^

and McCubbins, the governing party's proportion of seats in the constituency, was |

also tested (not reported) but was not significant. This is perhaps not surprising as U

Ireland has reasonably strong, unified parties and so McGillivray would expect v

GTS would be targeted at marginal constituencies. Of the control variables, the t.

population is highly significant and positive, which indicates that larger constitu- on

encies attract more pork, presumably because they apply for more grants. |

Model 2 tests the marginal voter hypothesis proposed by Linbeck and Weibull. 6

The variable measuring marginality is also insignificant. So neither of the standard 0

explanations appear to be important in accounting for the variation in the amount 5

of GTS delivered to individual constituencies in particular years. Again we would stress that this is not due to a complete absence of such targeting and credit claiming but that it was neither large nor systematic. Our theoretical expectations are that the decision rule is most important, and in this case the line minister has greater autonomy. Model 3 tests for this with dummy variables for the relevant

9Visual inspection implied that the model suffers from non-constant variance of the residuals. Thus we report robust (Huber/White) standard errors, as these are less likely to mislead about the significance of the independent variables. We utilise the log of the grant variable as the data is skewed by outliers.

Table 4. Partisan spending effects and sports grants : cross-sectional analysis 2002-2007

Variables Core Swing Decision All

FF vote 0.000555 -0.00113

(-0.00572) (0.00800)

Swing 0.0468 -0.192

(0.250) (0.222)

Sports minister 0.856*** 0.948***

(0.125) (0.185)

Finance minister 0.682*** 0.698***

(0.222) (0.239)

Prime minister 0.160 0.114

(0.180) (0.200)

Cabinet minister -0.136 -0.134

(0.0865) (0.0856)

Under 18 Pop (pc) 1.393 1.294 1.541 1.448

(-1.469) (1.512) (1.670) (1.668)

Education until 16 0.00418 0.00504 0.00387 0.00383

(-0.00752) (0.00908) (0.00867) (0.00813)

Population 1.00e - 05*** 9.97e - 06*** 1.10e - 05*** 1.09e - 05*

(1.59e - 06) (2.08e - 06) (2.00e - 06) (1.97e - 06)

Unemployment (pc) 4.511 4.945 6.658 6.379

(-4.53) (5.694) (5.237) (4.917)

Turnout - 0.00042 0.00161 0.000695 0.000365

(-0.00937) (0.0103) (0.00951) (0.0101)

Urban rural 0.237* 0.223 0.180 0.191

(-0.115) (0.174) (0.167) (0.165)

Constant 12.49*** 12.36*** 12.27*** 12.43***

(0.607) (0.719) (0.699) (0.785)

Observations 255 255 255 255

R-squared between 0.289 0.3497 0.4631 0.4708

F-test 124.17*** 121.96*** 285.73*** 285.74***

All models are estimated by OLS. The numbers in parentheses are cluster-robust heteroscedasticity-consistent standard errors. Dependent variable is defined in logarithms. ***p < 0.001, **p < 0.01, *p < 0.5.

ministers. The results are both positive and highly significant and the model shows that constituencies of the decision-making ministers do significantly better than other constituencies. Recall that these data include changes in personnel in both Finance and the line departments, and the models control for demographic variables. So the explanation is unlikely to be due to an idiosyncratic minister or that that their constituencies were most needy. The sign relating to their other colleagues in cabinet is not significant, which is not surprising as ministers may equally want to respond to demands from backbenchers; a pattern that persists in the full model. This provides evidence against any idea of cabinet log-rolling. Thus, there is strong corroborating evidence for the hypothesis that the interests of the individually powerful ministers will prevail over those of the party as a whole.

Table 5. Partisan spending effects and education grants: Cross-sectional analysis 2002-2007

Variables Core Swing Decision All

FF vote -0.00108 (0.00926) -0.00169 (0.00935)

Swing 0.453* (0.267) 0.516** (0.254)

Education minister 0.443** (0.202) 0.453** (0.211)

Finance minister 0.462** (0.199) 0.448*** (0.160)

Prime Minister 0.397* (0.208) 0.562** (0.241)

Cabinet minister -0.0962 (0.145) -0.117 (0.145)

Under 18 pop. 5.367* (2.793) 5.645** (2.624) 5.230* (2.714) 5.799** (2.574)

Education until 16 0.000743 (0.0176) 0.000808 (0.0162) 0.000601 (0.0193) 0.00215 (0.0179)

Population 1.18e-05*** (2.81e-06) 1.21e-05*** (2.70e-06) 1.10e-05*** (2.77e-06) 1.15e-05*** (2.69e-06)

Unemployment 4.001 (8.911) 5.409 (8.507) 5.041 (10.20) 5.374 (9.497)

Turnout -0.00802 (0.0121) -0.00835 (0.0110) -0.000805 (0.0131) -0.00109 (0.0118)

Urban/Rural -0.0351 (0.155) -0.0615 (0.149) -0.0829 (0.147) -0.116 (0.149)

Constant 12.62*** (0.927) 12.32*** (0.849) 12.20*** (0.917) 11.88*** (0.898)

Observations 255 255 255 255

R-squared Between 0.3539 0.3954 0.4018 0.4517

F-test 103.85*** 107.33*** 116.97*** 137 95***

All models are estimated by OLS. The numbers in parentheses are cluster-robust heteroscedasticity-consistent standard errors. Dependent variable is defined in logarithms. ***p < 0.001, **p < 0.01, *p < 0.05.

SI0Z '9! ^^Vi 110 AiJSJSAiun MBSJByW jb /§jos[BiunofpjqjxoBd//:d:ui[ iuojj pspBojuMOQ

Most control variables, such as urban, unemployment rate, under-18 population or turnout are not significant.10 One interesting, if unsurprising, aspect is that the dummies controlling for each year of the programme show that significantly less money is spent in the years following an election and mid-term years and more money is spent in elections years or the year before an election.

Table 5 reports the same models for primary schools capital spending. Our expectation set out in Table 2 is that the minister's bailiwick will become less important, and that marginal constituencies should be targeted. Again there is no evidence for the core voter model on its own or in the full model. The swing model presents a more conventional picture, with those constituencies in 0

which Fianna Fail was closest to either winning or losing a seat based on the pre- n

vious election results receiving more monies. Once again the evidence for the d

decision-rule hypothesis is there, with the key decision-making ministers' con- ff

stituencies receiving additional funds, the finance minister and the education |

minister both significant at the 0.05 level. But the substantive effect of these P

two decision-making actors is much weaker than it was for sports grants. This p

is in line with our theoretical expectations and provides support for our hypoth- x

esis that the decision-rule matters. As the decision rule in this case is based on §

objectively measurable and observable criteria such as demographic projections §

we expect and see that the under-18 population is most important. We can s

also compare the standardised beta coefficients across models; the sports minister accounts for a quarter of a standard deviation change in sports grant allocations, whereas the education minister only accounts for about 8 per cent.

In terms of the raw amount of funding delivered (using unlogged grant allocations) we can see in the full sports model that the constituency of the Finance Minister can expect some E1.25 million in additional funding each year while the constituency of the Minister for Arts, Sport & Tourism can expect almost €600,000. In contrast, the constituency of other cabinet ministers will generally receive over €140,000 less in sports capital funding. In primary school capital funding the constituency of the Minister for Education will receive about € 1.45 million in additional annual funding, while the constituency of the Minister for Finance will receive about €2.8 million additional funds.11

10We also ran the models without these controls and a similar pattern of significance was found albeit with slightly different coefficients.

11 These figures derive from an analysis of the raw amount of funds as the dependent variable as in the sports model. We report these figures as it is difficult to interpret the log derived coefficients in the table above. Both models produced similar results with only small changes in significance levels.

6. Discussion

Pork-barrel politics is thought of as the product of opportunity structures and incentives in the electoral and party systems within which parties and individual politicians seek to win and maintain power. Many works tend to disregard the decision-making rules or have simplifying assumptions with regard to them. Ireland is a good case to test theories of partisan particularistic spending. There are reasons to expect GTS is important and that, everything else being equal, this would be targeted at marginal constituencies. But we selected and compared cases of geographically targeted programme spending in which the

decision-rule varied. The result is that, GTS is not primarily used as rewards or °

carrots for either potential or committed party voters nationally. With two O

types of programmes, which vary in the discretion given ministers we can e

further test if the level of ministerial autonomy is important. Where there is f§o

more autonomy (as with sports grants) the level directed at the decision-makers h

constituency is much greater than where ministerial discretion is somewhat /

diluted. Thus we can say, at least in this area, the interests of individual ministers .

often trump the interests of the ruling party. Other legislators are only able to O

signal to voters that they had influence through the use of parliamentary ques- o

tions. This of course leaves open the question of whether delivery of such §n

funds is associated with garnering more votes at subsequent elections than O rival candidates from the same party who do not have the ability to distribute

pork locally. Another puzzle is why the parties do not act in a more concerted ^

effort to maximise seat share. These are obvious next steps in further research. |

Decision rules and incentive structures vary between and within in countries. U

The data on Ireland supported the general hypothesis that the decision rule v

matters. Having studied the geographic distribution of resources to Irish t.

constituencies from 2002 to 2007 we find that districts which elect powerful n

decision-making cabinet ministers win additional capital investments. |

However, the governing party is not always successful at winning additional 6

funds for their core voters or for the districts which elect non-decision-making 0

ministers. Thus, the decision-rule for the allocation of resources is the first 5 factor that scholars should look at when seeking to explain where GTS goes. Only when these have been taken into account can we assess the validity of studies testing whether core or swing voters are the primary target of partisan spending.


Ames, B. (1995) 'Electoral Rules, Constituency Pressures, and Pork Barrel: Bases of Voting in the Brazilian Congress', The Journal of Politics, 57, 324-343.

Ames, B. (2001) The Deadlock of Democracy in Brazil: Interests, Identities and Institutions in

Comparative Politics, Ann Arbor, University of Michigan Press. Ansolabehere, S., Hansen, J. M., Hirano, S. and Snyder, J. M., Jr. (2007) 'The Incumbency Advantage in U.S. Primary Elections', Electoral Studies, 26, 660-668.

Baron, D. P. and Ferejohn, J. A. (1989) 'Bargaining in Legislatures', American Political

Science Review, 83, 1181-1206. Bertelli, A. M. and Grose, C. R. (2009) 'Secretaries of Pork? A New Theory of Distributive Public Policy', Journal of Politics, 71, 926-945.

Bickers, K. N. and Stein, R. M. (1994) 'Response to Barry Weingast's Reflections', Political

Research Quarterly, 47, 329-333. O

Bickers, K. N. and Stein, R. M. (1996) 'The Electoral Dynamics of the Federal Pork Barrel', l

American Journal of Political Science, 40, 1300-1326. e

Cadot, O., Roller, L.-H. and Stephan, A. (2006) 'Contribution to Productivity or Pork to

Barrel? The Two Faces of Infrastructure Investment', Journal of Public Economics, 90, 3

1133-1153. /

Cain, B., Ferejohn, J. and Fiorina, M. (1990) The Personal Vote: Constituency Service and .

Electoral Independence, Cambridge, MA, Harvard University Press. xf

Carey, J. M. and Shugart, M. S. (1995) 'Incentives to Cultivate a Personal Vote: A Rank O

Ordering of Electoral Formulas', Electoral Studies, 14, 417-439. n

Carroll, R. and Kim, H. A. (2010) 'Party Government and the "Cohesive Power of Public °

Plunder" ', American Journal of Political Science, 54, 34-44. a

Considine, J., Crowley, F., Foley, S. and O'Connor, M. (2008) 'Irish National Lottery Sports a

Capital Grant Allocations, 1999-2007: Natural Experiments on Political Influence', |

Economic Affairs, 28,38-44. U

Cox, G. W. and McCubbins, M. D. (1986) 'Electoral Politics as a Redistributive Game', e

The Journal of Politics, 48, 370-389. Jy

Dahlberg, M. and Johansson, E. (2002) 'On the Vote-Purchasing Behavior of Incumbent §

Governments', American Political Science Review, 96, 27-40. a

Denemark, D. (2000) 'Partisan Pork Barrel in Parliamentary Systems: Australian §§

Constituency-Level Grants', The Journal of Politics, 62, 896-915. 1

Department of Finance (2008) Public Financial Procedures: An Outline. Dublin, The Stationary Office.

Department of the Taoiseach (2006) Cabinet Handbook, Dublin, The Stationary Office.

Diaz-Cayeros, A., Estevez, F. and Magaloni, B. (2006) Vote-buying, Poverty, and Democracy: The Politics of Social Programs in Mexico, 1989-2006, Palo Alto, CA, Department of Political Science, University of Stanford. Elgie, R. (1997) 'Models of Executive Politics: A Framework for the Study of Executive Power Relations in Parliamentary and Semi-Presidential Regimes', Political Studies, 45, 217-231.

Farrell, B. (1994) 'The Political Role of Cabinet Ministers in Ireland'. In Laver, M. and Shepsle, K. A. (eds) Cabinet Ministers and Parliamentary Government, New York, Cambridge University Press.

Fenno, R. F. (1978) Home Style: House Members and Their Districts, Boston, Little, Brown.

Gallagher, M. (2008) 'The Earthquake That Never Happened: Analysis of the Results'. In Gallagher, M. and Marsh, M. (eds) How Ireland Voted 2007, Basingstoke, Palgrave.

Gallagher, M. (2010) 'Parliamentaries Parties and Party Whips'. In MacCarthaigh, M. and Manning, M. (eds) The Houses of the Oireachtas: Parliament in Ireland, Dublin, IPA.

Gerring, J. (2007) Case Study Research: Principles and Practices, New York, Cambridge University Press.

Golden, M. A. and Picci, L. (2008) 'Pork-Barrel Politics in Postwar Italy, 1953-94', American Journal of Political Science, 52, 268-289.

Hallerberg, M. and Marier, P. (2004) 'Executive Authority, the Personal Vote, and Budget Discipline in Latin American and Caribbean Countries', American Journal of Political Science, 48, 571-587.

Horiuchi, Y. and Saito, J. (2003) 'Reapportionment and Redistribution: Consequences of Electoral Reform in Japan', American Journal of Political Science, 47, 669-682.

Key, V. O. (1949) Southern Politics in State and Nation, Knoxville, Alfred A. Knopf.

King, A. (1975) 'Executives'. In Polsby, N. W. and Greenstein, F. I. (eds) Handbook of Political Science, Reading, Addison-Wesley.

Lasswell, H. (1936) Politics: Who Gets What? When? How? New York, McGraw-Hill.

Latner, M. and McGann, A. (2005) 'Geographical Representation Under Proportional Representation: The Cases of Israel and the Netherlands', Electoral Studies, 24, 709-734.

Laver, M. and Shepsle, K. A. (1994) 'Cabinet Government in Theoretical Perspective'. In i

Laver, M. and Shepsle, K. A. (eds) Cabinet Ministers and Parliamentary Government, New York, Cambridge University Press.

Laver, M. and Shepsle, K. A. (1996) Making and Breaking Governments, New York, Cambridge University Press.

Levitt, S. D. and Snyder, J. M., Jr. (1995) 'Political Parties and the Distribution of Federal Outlays', American Journal of Political Science, 39, 958-980.

Lindbeck, A. and Weibull, J. W. (1987) 'Balanced-Budget Redistribution as the Outcome of Political Competition', Public Choice, 52, 273 -297.

Lyne, M. M. (2008) 'Proffering Pork: How Party Leaders Build Party Reputations in Brazil', American Journal Of Political Science, 52, 290-303.

Marsh, M. (2007) 'Candidates or Parties? Objects of Electoral Choice in Ireland', Party Politics, 13, 500-527.

Marsh, M., Suiter, J. and Muinechain, S. O. (2008) Candidate Study 2007, Dublin, Trinity College, Dublin.

Martin, S. (2011) 'Using Parliamentary Questions to Measure Constituency Focus: An Application to the Irish Case', Political Studies, 59, 472-488.

Mayhew, D. R. (1974) Congress: The Electoral Connection, New Haven, Yale University Press.

McGillivray, F. (2004) Privileging Industry: The Comparative Politics of Trade and Industrial Policy, Princeton, Princeton University Press.

Milesi-Ferretti, G. M., Perotti, R. and Rostagno, M. (2002) 'Electoral Systems and Public Spending', The Quarterly Journal of Economics, 117, 609-657.

Milligan, K. and Smart, M. (2005) Regional Grants as Pork Barrel Politics. CESifo Working Paper Series No. 1453, accessed at SSRN: on 4 October 2012.

Monroe, B. L. and Rose, A. G. (2002) 'Electoral Systems and Unimagined Consequences: Partisan Effects of Districted Proportional Representation', American Journal of Political Science, 46, 67-89.

O'Leary, E. (2011) 'The Constituency Orientation of Modern TDs', Irish Political Studies, 26, 329-343.

O'Malley, E. (2006) 'Ministerial Selection in Ireland: Limited Choice in a Political Village', f

Irish Political Studies, 21, 319-336. j

O'Malley, E. (2012) 'The Apex of Government: Cabinet and Taoiseach in Operation'. In g

O'Malley, E. and Mac Carthaigh, M. (eds) Governing Ireland: From Cabinet Government 0

to Delegated Governance, Dublin, IPA. /

O'Malley, E. and Martin, S. (2010) 'The Government and the Taoiseach'. In Coakley, J. ^

and Gallagher, M. (eds) Politics in the Republic of Ireland, 5th edn, Abingdon, Routledge. |

Shepsle, K. A. and Weingast, B. (1987) 'The Institutional Foundations of Committee ^

Power', American Political Science Review, 81, 85-104. v

Stein, R. M. and Bickers, K. N. (1994) 'Congressional Elections and the Pork Barrel', The Journal of Politics, 56, 377-399. §

Stratmann, T. and Baur, M. (2002) 'Plurality Rule, Proportional Representation, and the y

German Bundestag: How Incentives to Pork-Barrel Differ across Electoral Systems', ,

American Journal of Political Science, 46, 506-514. 0

Tavits, M. (2009) 'Geographically Targeted Spending: Exploring the Electoral Strategies of Incumbent Governments', European Political Science Review, 1, 103 -123.

Thies, M. F. (1998) 'When Will Pork Leave the Farm? Institutional Bias in Japan and the

United States', Legislative Studies Quarterly, 23, 467-492. Weingast, B. R., Shepsle, K. A. and Johnsen, C. (1981) 'The Political Economy of Benefits and Costs: A Neoclassical Approach to Distributive Politics', The Journal of Political Economy, 89, 642-664.