A paradox of modern democracy is that, whereas politicians are generally prohibited from buying votes with their own money, they feel free to do so with other people’s.Tom Bostock.
The ability to ‘buy votes’ is widely presumed to be one of the most effective tools available to a politician. Spending promises tend to be centrepieces of election campaigns. The long-term increase in Commonwealth government spending per capita suggests that governments have considerable faith that higher government spending is what people want and is the way to procure their support—from 1962-63 to 2002-03, after accounting for inflation, Commonwealth payments per person tripled.[1] By contrast, GDP per person (inflation adjusted) only went up two-and-a-third times over the same period.[2] Commonwealth expenditure has grown faster than both the population and the economy.
The wealthier we get, the more help we apparently need or desire from government. Indeed, the apparent need or desire increases faster than our rise in income. Government becomes a ‘luxury good’ reflecting, at least in part, the increasing demand for services such as health and education. Such growth may ‘feed on itself’ as rising tax burden on families creates a ‘tax trap’ where paying for private provision—in a sense, paying twice—becomes too prohibitive an option.[3]
Yet accusations of profligate spending seem to have some bite. Politicians don’t promise to spend on anything, they attempt to justify them according to some higher policy purpose. There is at least some constraint on their spending patterns—hence, Commonwealth payments as a percentage of GDP peaked at 28% in 1984-85.
(For simplicity, unless specified otherwise, I continue the budget fiction that the GST is both a state/territory expenditure and tax. I will take it as read that expenditure policies are generally aimed at increasing, or at least retaining, voter support. Use of regulatory policies to the same end are not examined, except in the general context of GDP growth.)
And what does this presumed ability to ‘buy votes’ actually rest on? It rests on the notion that the costs of raising revenue, predominantly from taxation—including the taxes themselves, compliance costs imposed in paying taxes and lost economic activity from the displacement effects of taxes—plus administrative costs, inefficiency and wastage (plus any envy effects from non or lesser recipients) are nevertheless exceeded by the gratitude of the voting public for the services received, which said public has compulsorily paid for. Commonwealth revenues per person, adjusted for inflation, went up two-and-three-quarter times from 1962-63 to 2002-03, as can be seen in Figure 1.[4]
This presumes a remarkable level of government efficiency. It also presumes a remarkable specificity of gratitude—given it is a reasonable expectation, that if one side of politics establishes a spending programme, the other side won’t abolish it (remember that tripling of payments per person). It further presumes that such gratitude is greater than is any alienation of other voters. Yet, I can remember far more hostile comments about government spending on other people than I can ever remember hearing of expressing gratitude for spending received.
Commonwealth money is citizens’ money. It is hardly surprising that people have a different view about ‘receiving back’ their money than ‘their money’ going elsewhere. To spend citizens’ money on citizens is to give them their due. To spend citizens’ money on one set of citizens and not another is to deny the latter their due—even insult them as ‘less worthy’. Votebuying expects people to be both selfish—wanting money spent on them—and altruistic—indifferent or positive to collective funds spent on others.
Do voters really believe that governments are better spenders of their own money than they are? Presumably not very much—hence the appeal of tax cuts. They may, quite rationally, believe governments are better spenders on things that they could not buy themselves (for example, a defence force). They may be quite in favour of the idea that the government spend other people’s money on them. But, we can surely wonder how much that game is a net electoral winner for governments, particularly as the level of lifetime redistribution is quite small. Ann Harding has estimated that the bottom income decile receives 21% of its lifetime income in government cash transfers—but pays out 12% of said income in income tax. (Other tax payments were not estimated.) Services are more mixed—the bottom income decile are strong net beneficiaries from government health expenditure, mainly due to the willingness of top deciles to pay for private health care.[5]
Services other than health are less redistributive, since upper income groups often receive more government services. The ratio of highest to lowest decile use of education services is 45:39 and would be higher if the subsidy to private school places was not lower than those to government school places. Middle-income groups tend to come out about even. Moreover, the electorate covers people of all stages in their life-cycle, tending to ‘cancel out’ effects. Particularly as some of the biggest net recipients—the under 18s—don’t vote (though their parents do).
Poll Preference for Education and Health Spending
Recent polling data has suggested popular preferences are for increased education and health spending rather than tax cuts. The proportion of voters preferring more spending on social services has gone up from 15% in 1987 to 30% in 2001, with those preferring reducing taxes falling from 65% to 42%. More recently, when the question is changed to education and health expenditure specifically, the figure jumps to 48% for increased spending compared to 28% for reducing taxes.[6]
Health and education expenditure is more clearly expenditure on most voters than ‘social services’ which, as Katherine Betts points out, includes expenditure on marginal groups, increasing the direct appeal of such expenditure. If tax cuts are seen as applying more to upper-income groups, that would decrease their appeal. As Andrew Norton has also suggested, the recent history of Commonwealth budget surpluses may also suggest that increased spending need not imply increased taxes.[7]
How much such polling results express a preference to be seen to be compassionate is hard to say. The ‘revealed preference’ of voting behaviour seems fairly clearly against higher taxes. Both the 1993 and 1998 general election results can be very plausibly read that way. This article concentrates on the revealed preference of voting behaviour rather than the expressed preference of polling behaviour.
So, when we examine the matter, there seems to be grounds for doubt about the efficacy of politicians’ ability to ‘buy votes’ (with the voters’ own money).
Well, what’s the evidence? Does increased spending increase electoral support? At the level of Commonwealth politics the evidence does not support the proposition that increased Commonwealth spending makes governments effective purchasers of votes. How so? Because incumbency—at least in the period when the ALP has been genuinely competitive with the Coalition (the ALP has won 7 of the 14 House of Representative elections from 1969 on, compared to 5 of the previous 26)—is clearly a wasting electoral asset. If increasing government spending was a net electoral winner, one would presume that incumbency (being able to actually deliver on spending, rather than merely promise it) would be a distinct electoral advantage. Yet, in 12 of the 14 elections from 1969 onwards, there has been a net two-party preferred swing against the incumbents (Figure 2).
The only two exceptions were 1993, when the Opposition was promising a big new tax, the GST, and 2001 when, along with other considerations, people had already punished the incumbents for the GST in the previous election which had, perhaps, proved more palatable than feared. They were also the only two occasions when the incumbents were sitting on a minority of the two-party-preferred vote—in all but one occasion (1969) when the incumbents had been elected on a minority of the two-party-preferred vote since 1949, there was a swing back to the incumbents the next election. (Since 1949, only incumbents have won federally on a minority of the two-Party-preferred vote.)
When one matches shifts in two-party-preferred votes with increases in Commonwealth payments per person, there is a moderate negative correlation (-0.4). Looking at that hardy perennial—the ‘election budget’—only marginally reduces the negative correlation (-0.3). Thus, the most ‘generous’ government received one of the largest swings against an incumbent (1975), while the Hawke government managed to achieve the two lowest swings against an incumbent government while cutting payments per person (1987 and 1990). Such ‘hostile indifference’ to increased expenditure does not imply much success in vote-buying.
The rate of growth of government spending has been lower in the economic reform period (1984 to 2001) than previously. In the earlier period, increased spending over the term of Parliament shows a clear negative correlation. In the reform period, voters appear to have become indifferent to changes in spending.
There is, however, a clear difference in correlations between incumbent Coalition and ALP governments. The Coalition gets no persistent benefit either way from changes in expenditure. The ALP gets punished for increasing expenditure and rewarded for cutting it. Various factors may be relevant here; such as voter antipathy to fiscal indulgence, quality of services delivered being more of a problem for the party which sets higher expectations for government delivery (and whose welfarist credentials provide ‘cover’ for cuts) and that voters may simply not much notice Coalition expenditure one way or another. That, for example, Commonwealth health and education expenditure has increased more rapidly under Howard than Keating, as it has, is hardly the common perception.
The behaviours of Oppositions may also be pertinent here: Coalition Oppositions may be keener in drawing attention to instances of government waste and inefficiency than Labor Oppositions, which would help punish Labor incumbents for increasing spending while tending to deny political salience to spending by Coalition incumbents.
Another factor may be cultural distance. As Katharine Betts says
The most marked division within Australian politics is not between different groups of voters (working versus middle class) but between a majority of voters, including the traditional working class, and candidates for the Labor, Green and Democrat parties.[8]
If increased spending is directed, or simply delivered, by knowledge class ‘cosmopolitans’ in ways which reflect different priorities and values from those of voters, that would tend to decrease spending’s appeal as well. (Indeed, patterns of polled political preferences suggest strongly that the real comparative advantage of the Howard-led Coalition has been in cultural politics.)
Incumbents generally get re-elected. There was a change of government in only 4 of the 14 elections. But one can reasonably doubt that increased spending accounts for that, particularly given the weak negative correlation between spending increases and shifts in two-party-preferred vote. Nor does it provide any reason to believe that somehow support for incumbents would have fallen even more strongly if spending hadn’t increased—especially as swings against incumbents have been lower in the period when growth of Commonwealth spending was also lower.
Indeed, the long-term behaviour of the Commonwealth—wearing the odium for collecting taxes which it then hands over to the states and territories to get the kudos for spending—is deeply irrational if there is any strong belief that voters tend to vote on the basis of some fiscal gain-loss calculus.
But if there is one thing voters clearly don’t like, it is paying for increased government (Table 2). The negative correlation between increased Commonwealth receipts per head and votes is very strong, particularly for the Coalition. Very clearly, voters expect the Coalition to restrain growth in tax payments—not good news for the Coalition incumbents when tax as a share of GDP is the highest in our history. (An election winning theme for Mark Latham?) By contrast, the ALP gets some tolerance in election years (presumably lower expectations on tax, better expectations on its use), but not over the life of a Parliament.
The rate of growth of Commonwealth receipts is lower in the economic reform period than before, though average growth is higher than previously in election years. Apart from being somewhat more unimpressed by increased taxes in election years, antipathy to increased taxation does not seem to have changed much.
Even more depressingly for any fond belief that there are sure-fire levers to gain electoral support, the story for incumbents is no better if we examine changes in GDP per head.
Voters tend to be negatively indifferent to economic growth, with increase in per head GDP over parliamentary term moves mildly in the opposite direction to shifts in two-party-preferred support for incumbents. Economic growth in election years is less contrary—there is simply no correlation at all.
On the other hand, there is again a clear difference between Coalition and the ALP. Voters seem to be ‘hostilely indifferent’ to economic growth under Coalition incumbents, perhaps again a failure of expectations to match perceived benefits. Conversely, the ALP gets some reward for good economic growth in an election year, perhaps because it is easier for them to achieve above expectations as economic managers.
What marks a bigger difference is the period of economic reform beginning with the floating of the $A in 1983. The growth of per head GDP has been markedly greater, and the average swing against incumbents in the economic reform period is less than half it was beforehand, with only one change of government in seven elections, rather than three. On the other hand, the economic growth correlation results seem more perverse, with strong negative correlations. So electoral movements are more forgiving of incumbents but more strongly counter-cyclical. This may flow from wealthier voters being more aware of the difficulties governments have had to deal with, so being more forgiving when things go poorly, and more interested in other things when things go well. Perhaps also incumbents are more sharply focused on retaining support in poor economic conditions and more complacent and more easily ‘wrong-footed’ in good economic conditions. (A similar dynamic of Government focus, and Opposition complacency, may underlie the pattern of swings to incumbents sitting on minorities of the two-party-preferred vote.)
To make sense of these results, which go against conventional wisdom, we need to consider what it means to have two evenly matched contenders for the vote (the Coalition averaging 50.2% of the two-party-preferred vote, the ALP 49.8% over the 14 House of Representatives elections from 1969 on).
Voters are weighing up two contenders; the performance in office of the incumbents, and future expectations derived from that experience, against the alternatives. Incumbents are, of course, the more ‘known’ quantity, so one would expect some advantage from that—and it is the normal pattern for government opinion poll support to improve as the election date gets closer. Incumbents have been re-elected in 10 of those 14 elections, though twice (1990 and 1998) with a minority of the two-party-preferred vote (so, one can claim that, in a manner of speaking, incumbents actually ‘lost’ 6 out of the 14 elections).
But, governments actually do, oppositions only promise. Since actions have a somewhat greater capacity to alienate than mere promises, particularly as politicians’ promises are probably fairly heavily discounted anyway, incumbency in a position of genuine competitiveness tends to result in a net expenditure of ‘political capital’. Hence the strong pattern (12 out of 14 elections) of swings against incumbents, and the tendency of opinion polls in periods when an election is not imminent to be adverse for incumbents.
Regarding such political capital: increased spending seems to be, on balance, generally a form of expenditure of political capital. Nor can we really be surprised that a majority of voters actually prefer to have funds in a form they can control directly. After all, it would be fairly odd if people genuinely preferred to have to have funds shifted from a form they can control (private income) to a form they cannot (government expenditure). It would suggest a deeply implausible level of confidence in, and deference to, politicians and public servants. The continuing shift to private schooling, for example, is surely, in significant part, about control and evidence against such deferences. (On the other hand, the aforementioned ‘tax trap’ can foreclose options.)
Hence the need for politicians to argue for higher purposes—that expenditure would be for things people want but would not otherwise get or imply that someone else would be paying for it. Of course, analysis of actual distributional effects shows that there is remarkably little of the latter, particularly for middle-income groups. If people broadly have some grasp of that truth, it might help explain why higher spending does not seem to be useful for incumbents.
Even if we can see ourselves as net-gainers from a particular area of government spending, the sheer scale of government action must dilute the effect, by increasing the areas where we are et losers. More is, in fact, less. Which, again, the expenditure and electoral evidence would support—particularly as new spending has a greater natural political salience than established spending.
For those interested in psephological prognostication: the two-party-preferred result which, based on Commonwealth Budget forecasts for 2004-05, best preserves the existing correlations for Coalition incumbents on all three measures (changes in Commonwealth payments, receipts and GDP per head), is a 2.6 percentage points swing against the Coalition, which would give the ALP an 8-seat majority. There is no particular reason to suppose those correlations will be preserved—particularly if cultural politics become salient, or the Opposition engages in behaviour more hostile to the expressed preferences of voters. It does, however, point to the underlying electoral tendency for Coalition incumbents, based on past voting behaviour, of the current Government’s tax-and-spend approach.
Sugar Seats
The classic example of Australian local vote-buying expenditure is the sugar industry and the ‘sugar seats’ of Hinkler, Kennedy, Dawson, Leichhardt, Herbert and Wide Bay.[9] What is striking about those seats is that, from 1987 to 2001, they actually have a higher average swing against incumbents than the rest of the country (Hinkler was first created in the 1984 expansion of Parliament). The national average two-party-preferred swing against incumbent governments was -1.3 percentage points. In the sugar seats it was -2.2 percentage points, the same as for the rest of Queensland. Given the cost of sugar support, that is expensive ingratitude. Moreover, swings in the sugar seats march along with the general trend in the Queensland vote.
That, in the 2001 election, the biggest sugar seat swing was enjoyed by Bob Katter in Leichhardt (8.5 percentage points) after becoming an Independent, and thus eschewed belonging to any possible governing Party, particularly suggests the limited value of government vote-buying. It does, however, suggest the power of expressive politics.
The notion that Commonwealth governments are effective aggregate purchasers of votes (with the voters’ money) rests on an implausible presumption of government efficiency and voter gratitude that is not supported by the evidence.
Michael Warby is a Melbourne writer, educator and analyst. mwarby@mira.net.
A more complete analysis of the historical pattern up to and after 1966 is provided in ‘Incumbency as Wasting Asset: Structural shifts in federal politics’, Australasian Parliamentary Review 17(1) Autumn 2002, 17-25.
