The Cash Nexus: Money and Politics in Modern History, 1700-2000

by Niall Ferguson

  Do voters view the economy differently according to which class they belong to?

  Do voters act asymmetrically, punishing failure more than they reward success?

  Are voters backward-looking or forward-looking?

  Are voters myopic or rational?

  Is ‘feeling good’ – meaning the perception of prosperity – more important to voters than actually being better off?

  The following paragraphs summarize the main answers political scientists have come up with.

  1. Economic Indicators. It is clear from numerous individual country studies that there is considerable variety in the economic indicators that matter to voters. For instance, one study of US congressional elections between 1896 and 1964 found a significant correlation between changes in real income and election results, a slightly weaker negative correlation between prices and voting, but no real relationship with unemployment.64 Looking at twentieth-century presidential elections, Ray Fair found changes in both real GNP and the rate of unemployment in the pre-election year to have been significant: specifically, a 1 per cent rise in real GNP gave the incumbent an additional 1.2 per cent of the vote, while an equivalent rise in the rate of unemployment cost as much as 2.3 per cent of the vote.65 Growth-based models were a source of encouragement to Al Gore’s supporters during the 2000 campaign, when their candidate was lagging behind in the polls.66

  In Britain more attention has been paid to the trade-off between inflation and unemployment. A pioneering study of British elections published in 1970 found that both indicators had influenced British government popularity in the post-war period (after allowing for a six-month lag in the effect of unemployment changes, ‘euphoria’ after the election of a new government, and ‘backswing’ to a government just before an election).67 Comparable results have since been produced for the UK under the Thatcher government.68 However, other analysts have argued that inflation was the more important indicator for Britain in this period.69 There is in fact some evidence of shift of attention in the historical record. It was a sign of the changing economic climate by September 1977 that Jim Callaghan no longer agreed with Tony Benn that ‘the key’ to electoral success was ‘full employment’:

  ‘Well, honestly,’ said Jim, ‘I don’t know what to do. Does it matter very much? I remember the 1935 election when there were 3 million unemployed and Labour still did badly.’

  [Benn:] ‘In 1945 it was the heart of the whole election.’

  ‘The awful truth is,’ [Callaghan] said, ‘that inflation affects everybody and unemployment comparatively few.’70

  As inflation had risen, Callaghan had come to doubt the efficacy of pre-election ‘reflation’. The then Chancellor Denis Healey agreed. On 14 September 1978 Healey told his Cabinet colleagues that ‘the only reason unemployment had come to the top of the list [of public concerns expressed in polls] was that prices were stable. If they began rising, inflation would go back to first place.’71

  Growing public concern about inflation may help explain why Britain produced a more radical conservative reaction in the 1980s than other European countries. For West Germany, by contrast, there is evidence that unemployment mattered more than inflation between 1971 and 1986 in persuading voters to change parties.72 In France between 1978 and 1987 both indicators correlated closely with public approval for the party in power.73 But then inflation in those countries never reached British levels in the 1970s. The maximum inflation rate in Britain was 27 per cent in August 1975, at a time when French inflation was just 11 per cent and German less than 6 per cent. On the other hand, the British Conservatives had to work hard to defend an anti-inflationary policy which had as its principal side-effect a doubling of unemployment. ‘Inflation had fallen from 20 to 4 per cent – its lowest level for 13 years,’ Thatcher later recalled of her first term in office. ‘Success against inflation was the single achievement to which we drew most attention as we approached the [1983] election…. The black spot in the record was of course unemployment, which was still well over three million. It [was] vital in the campaign to explain why this was so and what we were doing about it.’74 Nigel Lawson also saw the 1983 election as overturning ‘the conventional wisdom … that rising unemployment … would mean certain disaster for the party in office’. But by this time, he suggested, unemployment had ceased to be ‘a good indicator of the general sense of economic well-being’: ‘Unemployment caused by the end of decades of overmanning … at a time when the economy was manifestly improving and the living standards of those in work rising was a different matter altogether.’75

  Although most studies of the ‘vote function’ focus on real incomes, inflation and unemployment, there are of course other possibilities. Tax policy probably deserves more investigation, given the importance in recent years of promises like George Bush’s infamous ‘Read my lips: no new taxes’ pledge in August 1988.76 Trade policy too can play an important part in elections (as it did in the nineteenth century): witness the extraordinary swings in party allegiance on the eve of the 1988 Canadian general election as a result of the Liberals’ decision to stall the passage of the North American Free Trade Agreement.77 There is also some evidence that privatization did attract new voters to the British Conservatives in the 1980s.78 One strong possibility is that the importance of specific indicators changes over time (and indeed from place to place).79 In the United States, for example, the relationship between family income and government popularity was strong in the 1950s, weak in the 1960s and then strong again in the early 1970s.80 This is precisely the volatility that made the feelgood factor model so unreliable in Britain after the 1960s.

  2. Self-interest or the Common Good? Some research on the United States has suggested that voters are more likely to be influenced by general economic conditions than by their own individual economic circumstances in their voting decisions, implying that voters are ‘sociotropic’ or possibly altruistic in their behaviour; or, alternatively, that ‘self-reliant’ Americans tend not to attribute their own individual fortunes to the government.81 In 1984, for example, only around 5 per cent of American voters attributed changes in their own personal economic circumstances to government tax policy.82 However, it is not at all easy to distinguish personal and general economic influences on voting decisions.83 And a number of forecasters still regard the question ‘Are you better off or worse off than you were a year ago?’ as a good indicator to voting intentions.84

  3. Partisan Asymmetry. A further complication is that voters may have different economic expectations of different parties: they may expect lower inflation but higher unemployment from right-wing governments, and the converse from left-wing governments.85 If nothing else, this seemed a helpful way to rationalize the failure of voters to turn away from the Thatcher and Reagan governments when unemployment rose during the early 1980s.86 However, closer inspection has required some substantial qualifications to this story: for example, it emerges that in 1979 British voters regarded Labour as better than the Conservatives at dealing with inflation, which polls still showed to be the most important economic issue of that election. Yet Labour lost. Moreover, when unemployment rose in the subsequent three years, voters turned not to Labour but to the new Social Democratic Party.87

  4. Class Voting. Thus far we have tended to imagine voters as a homogeneous group. However, political scientists have long sought to demonstrate that voters’ economic preferences are in some measure a function of their social class. It is often argued by British political commentators that the link between social class and party allegiance has fallen since its zenith in the 1950s.88 Early proponents of ‘feelgood’ politics such as Anthony Crosland were deeply influenced by this, arguing that as voters abandoned old class allegiances they would simply follow whichever party delivered the lowest prices and the highest employment. More recently it has been argued that class has been superseded by other social attributes: now it is ethnicity or church attendance that matter more (though some of the ‘new’ determinants of voting
behaviour such as trade union membership and home ownership can hardly be distinguished from class).89 This raises the possibility that the stronger correlations between economics and politics in the 1950s and 1960s were in fact related to the persistence of class allegiance; whereas after around 1970 the dissolution of class identities reduced the significance of macroeconomic indicators. Nevertheless, there are those who maintain that class still matters. True, it appears to have declined if one uses a simple measure like the ‘Alford index’ which measures the difference between the percentage of manual workers and non-manual workers who vote for Labour. But more sophisticated methods (for instance, controlling for the decline of manual workers as a percentage of the population) seem to show the persistence of class-political affinities. Studies are legion showing that, even in more recent times, a rise in unemployment tends to increase support for the Left among the working class more than among other social groups.90

  5. There’s No Gratitude … Some political scientists argue that voters act ‘asymmetrically’, punishing politicians for bad economic developments more than they reward them when things go well.91 Among British voters polled in 1963 and 1964, for example, those who believed their economic situation had worsened over the previous year were twice as likely to ‘swing’ against the government than those who believed their situation had improved.92 Yet the rapid growth of the economy since the Tories had come to power in 1951 should have ensured that the losers were in a clear minority. Contemplating the government’s plummeting popularity since 1960, Harold Macmillan mused that perhaps ‘after ten years of unparalleled prosperity, the people are bored’.93 In other words, the majority who had never had it so good were less ready to reward the government than the minority of economic losers were to punish it.

  6. Peasants or Bankers? However, this assumes that voters judge politicians retrospectively on their past performance, as opposed to prospectively on their expected future performance. To pose the question in academic jargon: are voters backward-looking ‘peasants’ or forward-looking ‘bankers’? There is some evidence of retrospective voting in the United States,94 though the effect appears to be stronger in presidential than congressional elections, while retrospective judgements are clearly mingled with other factors, such as long-term political allegiance (‘We’ve always voted Republican in this family/street/town’).95 It would, of course, be wiser for voters to act on the basis of what they think a future government will do rather than what a past government has done. And some American research does indeed suggest that forecasts rather than memories are what motivate voters.96 In Britain too the 1992 election can be seen as an example of prospective voting: rather than punishing them for the past recession, voters re-elected the Conservatives on a wave of rising financial expectations.97 According to one argument, Tory voters had indeed voted prospectively throughout the 1980s – though their judgement of Labour remained firmly retrospective.98 For this reason, answers to the question ‘Who would make the best prime minister?’ may provide a more reliable guide to voting intentions than approval ratings, supposedly based on past performance.99

  Of course, this is almost certainly a false dichotomy, since expectations must in some measure be based on past experience.100 Yet even a model based on both the past and the future can be wrong. The Conservatives’ ‘great hope’ before the 1997 election was that ‘rises in real personal disposable income in the previous twelve months would be followed by rises in householders’ net expectations about their financial position in the next twelve months, which would be followed by rises in voting intention for the Conservative Party’.101 It did not happen.

  7. Fooling the People. The peasants-or-bankers question raises a more fundamental theoretical division between those who assume – with the original proponents of the political business cycle theory – that voters are ‘myopic’, and do not foresee the inflationary costs of pre-election ‘bribes’.102 However, neo-classical economists who think about these things prefer to assume that voters act rationally and with foresight, and therefore ‘cannot be routinely fooled by the government’. This, the argument runs, is why American voters tend to switch allegiance in the middle of a presidential term.103

  8. Feeling versus Being. A final and related possibility is that ‘feeling good’ may be something different from actually being better off: in the words of Maurice Saatchi, it is ‘economic perceptions’ not ‘economic facts’ that count.104 If that is so, fooling people may be more important than filling their pockets.

  When voting behaviour is open to so many different interpretations – nearly all of which rest on some empirical evidence – it is perhaps no wonder that the history of economics and politics has an apparently ‘chaotic’ character, in the sense that electoral behaviour appears stochastic and is unpredictable. The political equations are in fact non-linear.105 Another way of putting this is to say that economic relationships, even if they do have some element of regularity, are frequently ‘drowned out’ by political events. Three notable examples in recent British history are the Falklands War, victory in which undoubtedly helped to avert a Conservative election defeat in 1983;106 the entirely avoidable crisis precipitated by the introduction of the poll tax in 1990, a measure which defied political as well as economic rationality; and the departure of sterling from the Exchange Rate Mechanism in 1992, which was at root a consequence of the rise in European interest rates caused by German reunification. The only way to rescue some vestige of linear correlation between economics and politics is somehow to incorporate such events into the model. Thus Conservative electoral performance since 1979 has been explained – and on at least two occasions forecast with considerable accuracy – using a model incorporating political dummies to allow for the impact of precisely these events.107

  Does such an exercise salvage the theory of the feelgood factor? It seems doubtful. Of the eight components of David Sanders’s equation designed to explain the Conservative vote between 1979 and 1997, only two are economic variables – the change in the tax index and the balance of positive over negative household financial expectations – compared with four political dummy variables (in addition to the three listed above, the advent of Tony Blair as a credible Labour leader has been included).108 There is of course no reason why more such dummies should not be added at the author’s discretion whenever the need arises to realign the model’s predictions with an inconveniently diverging reality.

  POLITIKVERDROSSENHEIT

  If the advertising agents and ‘spin doctors’ who have come to dominate British and American politics are right, however, then the breakdown of the relationship between real economic indicators and government popularity may not matter. If it is only perceptions that count, then politicians need only concentrate on buying the most effective possible election campaign, the theme of which should ideally be: ‘You have felt good under our party for the past few years; you will feel even better if you re-elect us for another few.’

  There are three reasons why this strategy is unlikely to be effective for long. The first is simply that marginal increases in human satisfaction from economic growth appear to be subject to diminishing returns. Richer people may be happier than poorer people, according to some surveys; but if all the people surveyed are asked again after five years of rapid economic growth, their reported happiness will not have increased as much as their incomes. In Japan, to give an example, real incomes have risen by a factor of five since the 1950s, but according to the surveys there has been no change in happiness.109 Nor are the richest countries in the world home to the happiest people: according to one attempt to gauge global happiness, the United States is only the thirteenth-happiest country in the world. Iceland is the happiest. Yet in terms of GDP per capita, the United States is ranked seventh to Iceland’s sixteenth.

  One interpretation is that an increase in economic growth may provide a boost to the credibility of the government order, but then the effect fades as prosperity becomes the norm, or levels off. In other wor
ds, it may have been the novelty of growth that caused its electoral impact in the 1950s and 1960s: the essence of the experience of the 1950s and 1960s was not so much feeling good as feeling better-off.110 A more radical possibility is that beyond a certain point rising incomes do not necessarily increase ‘well-being’: in the psychologist Donald Campbell’s striking phrase, the rich find themselves trapped on a ‘hedonic treadmill’ of unfulfilling consumption.111 The more frenetic pace of life in conditions of rapid growth may also heighten feelings of insecurity.112 Alternatively, it may simply be that the conventional indicators we use to measure prosperity do not capture ‘disamenities’ – negative side-effects – of growth, such as pollution and traffic congestion.113

  The second reason why voters may be less and less inclined to believe political pledges to make them feel better is the phenomenon of disillusionment with politics in general. The Germans call it Politikverdrossenheit, which is perhaps best translated as ‘politics fatigue’. One obvious symptom of this is the phenomenon of falling turnout at elections. There are those who maintain that turnout is mainly governed by the perceived ‘salience’ of elections and that, provided an election appears to be important and that voting in it seems likely to have some bearing on the result, voters will not stay at home. But this overlooks the point that elections in general may seem less ‘salient’ than they did thirty years ago.114 Turnout in British general elections has fallen from 85 per cent in 1959 to 70 per cent in 1997.115 It has declined even more markedly in local and European elections: in the latter, from 57 per cent in 1994 to just 49 per cent in 1999.116

  At the same time, there has been a marked increase in electoral volatility: the fickleness of voters. The proportion of British voters with ‘very strong’ party affinity was 47 per cent in 1964; in 1987 it was just 16 per cent.117 In Ireland in 1981, 24 per cent of voters said they did not feel ‘close’ to any party; eight years later the proportion had more than doubled to 58 per cent.118 On the continent too volatility has risen in the 1990s. The Pederson index of volatility for thirteen European systems (which adds together the change in each party’s share of the vote between elections and divides by two) shows an increase from around 8 per cent for the period 1948–89 to 14 per cent for 1990–1994.119 Using an alternative measure of volatility based on the net aggregate shift in votes from one election to the next (equivalent to the total gains of all winning parties), the Italian elections of 1994 saw volatility of 37 per cent, one of highest figures for any European election between 1885 and 1989.120 (For Germany in the post-revolutionary year of 1919 the figure was 48 per cent, for France after the end of the Nazi Occupation, 36.)