The central importance of relationships among economy, polity and society to the wealth of nations and the well-being of people has attracted much interest among academic researchers and non-academic observers for many years (see, for example, Smith 1776; Keynes 1936; Acemoglu and Robinson 2012). Since at least the early 1970s, this interest has been heightened by the varying fortunes of governing parties and their opponents, of government policies and of economic conditions (see, for example, Vig and Schier 1985: Chapter 1; Clarke et al. 1992: Chapter 1; Clarke et al. 2016). One area of interest involves reaction functions and outcomes functions, whereas another is popularity-vote (V-P) functions (see, for example, Whiteley 1986; Clarke et al. 1992: Chapters 1 and 8; Lewis-Beck and Stegmaier 2013; Stegmaier, Lewis-Beck and Park 2017). Reaction functions refer to government effects on economic policies, as well as policy effects on economic outcomes, notably whether and how government policies can stimulate employment, growth and investment and reduce inflation and poverty. V-P functions pertain to the effects of economic conditions (the objective economy) and economic evaluations (the subjective economy) on people’s party support, including their economic voting.
Economic-voting research has become a well-developed but a still-progressing program (for reviews, see Norpoth, Lewis-Beck and Lafay 1991; Lewis-Beck and Stegmaier 2000, 2007; van der Brug, van der Eijk and Franklin 2007; Duch and Stevenson 2008; Lewis-Beck and Costa Lobo 2017). This chapter reviews the contributions of this program, and offers several recommendations for its progress in terms of models, measures and methodologies.
A major model of economic voting is the reward-punishment model (Downs 1957; Key 1968; for reviews, see Clarke et al. 1992 and Lewis-Beck and Stegmaier 2007). According to this model, people who evaluate economic conditions as getting better vote for, thereby “rewarding,” a governing party or leader. But people who judge that these conditions have worsened vote for an opponent, thereby “punishing” an incumbent party or leader. This model involves at least three assumptions (see, for example, Clarke et al. 1992: Chapter 1). First, people apply a rationalchoice, “optimizing” strategy by “rewarding” a governing party that pushes economic conditions toward, while “punishing” one that pulls such conditions away from, their preferred performance. Second, people use retrospective judgments by focusing on past or recent performance of the country’s economy and/or their own economic circumstances during the time that a party is in government (see Kiewiet 1983; see also Healy and Malhotra 2013). This focus is sensible since economic performance that is currently “known” is presumably more reliable than that which requires future guessing. Third, people attribute responsibility to governing parties for the performance of the country’s and/or their own economic circumstances. When doing so, they credit the government for better circumstances or blame it for worsening ones (see, for example, Powell and Whitten 1993; Anderson 1995).
Although the reward–punishment model is straightforward and has attracted much attention among academic researchers and non-academic observers, its coronation as the superior explanation of economic voting remains premature. There are two major reasons for this. One set of reasons involves the reward–punishment model itself (see, for example, Monroe 1979; Clarke et al. 1992; Lewis-Beck and Stegmaier 2007). The model is fundamentally misnamed because voters may not have reward/credit or punitive/blame motivations but, rather, a best-performance incentive of wanting the best economic conditions for their country and for themselves. Moreover, several assumptions that inform the model are not necessarily consistent with a rationalchoice view (see, for example, Monroe 1991). According to this view, people should not be retrospective but, rather, prospective thereby using current or future-focused information to update their evaluations, with little or no discounting of this information by their past evaluations of economic performance. Indeed, other research shows that people can be more prospective than retrospective when making their vote decisions (see, for example, Lewis-Beck 1988; MacKuen, Erickson and Stimson 1992; Clarke and Stewart 1994; van der Brug, van der Eijk and Franklin 2007).
Responsibility attributions raise other considerations. Contexts that permit the formation of clear responsibility attributions tend to display stronger economic effects on vote choices (for discussion, see Lewis-Beck and Stegmaier 2007). But, responsibility attributions can be asymmetric with governing parties “blamed/punished” for bad performance but not necessarily “credited”/“rewarded” for good performance (see, for example, Bloom and Douglas Price 1975; Clarke et al. 1992; van Brug, van der Eijk and Franklin 2007). And people who live in institutionally complex settings, such as those with multiple parties in coalition governments or (con)federal systems with two or more government levels having overlapping economicmanagement jurisdictions, might be excused for lacking valid information about who is responsible for what (Powell and Whitten 1993; Anderson 1995; Schwindt-Bayer and Tavits 2016). In such settings, people’s information and knowledge can be confused rather than clarified, because responsibility for economic performance is confused rather than clarified. A vote-seeking party in a governing coalition and/or a (con)federal system strategically shirks when it tries to offload responsibility for economic performance to another party or another level (Clarke et al. 1992; van Brug, van der Eijk and Franklin 2007).
Cognitive difficulties may be further imposed on people’s economic evaluations (the subjective economy) when they try to obtain accurate information about economic conditions (the objective economy) from multiple, streaming sources. Thus, the links between evaluations and conditions might be best described as “through a glass darkly,” with distorted evaluations based on incomplete information about economic conditions (see, for example, Stimson 1989; Clarke et al. 1992: Chapter 3). The linkage becomes blurrier when the conditions being considered are country-level (that is, sociotropic; see, for example, Kinder and Kiewiet 1981), rather than personal (that is, egocentric), since voters might be expected to know relatively less about the former than about the latter.
A second set of reasons why the coronation of the reward-punishment model is premature is that rival models and alternative frameworks lay claim to the crown of superior explanation of economic voting. According to the issue-priority model, public evaluations of political parties as “owning” particular issues influence voting choices (see Budge and Farlie 1983; Clarke et al. 1992: Chapter 1). For example, people would vote for, not against, a governing conservative/ right-of-center party when they think that rising taxes are an important issue and that this party’s proven record will guide it to lower taxes in comparison with that of a left-of-center/liberal party. In turn, people would support a governing left-of-center/liberal party when they think that rising poverty is a real problem and that this party’s known reputation will guide it to reduce poverty much more than would that of a conservative/right-of-center alternative.
The spatial and valence models distinguish between economic issue types (see Downs 1957; Stokes 1963, 1992; for discussions, see Clarke et al. 1992: Chapter 1, 2004: 21–25, 2009, 2016; Whiteley et al. 2013). In a spatial (or positional or issue-proximity) model, parties express different issue preferences, people form specific issue preferences, and they estimate which party’s preferences approximate their own when choosing how to vote. Income (re)distribution is an example of spatial issues. In a valence (or performance) model, parties and people agree on a preferred issue outcome, and people then assess which governing or opposition party is most able to produce this outcome when deciding how to vote. Two examples of valence issues are a good economy and a safe society.
Overall, the four models described above share several similarities (for additional discussion, see Clarke et al. 2004: 21–25). They acknowledge the role of economic performance evaluations in vote choices and they do not meet all of the strict requirements of rational-choice theory (see, for example, Monroe 1991). But, unlike the reward–punishment model, the issuepriority, spatial and valence models recognize that vote choices are influenced, albeit in sometimes counter-intuitive ways, by retrospective and/or prospective as well as by egocentric and/ or sociotropic economic evaluations. These other models also allow that most voters’ evaluations of economic performance are not based on full and perfect information. Rather, experimental economists and cognitive psychologists have conducted numerous experiments that have contributed important insights into how people make decisions, including their use of cognitive “short-cuts” or cues, in political-economic settings (see, for example, Sniderman, Brody and Tetlock 1991; Gigerenzer 2008; Gigerenzer, Hertwig and Pachur 2011; Kahneman 2011; Thaler 2016). These cues include, but are not limited to, their party (non)identifications and/or their attitudes about parties and candidates/leaders.
As is well known, party identification and candidate/leader attitudes are important factors in two major and alternative theoretical frameworks of voting behavior. According to the socialpsychology framework, the “funnel of causality” begins with socialization learning and social groups leading to the formation of party identification as “the unmoved mover,” which, in turn, affects candidate attitudes, issue attitudes and voting behavior (Campbell et al. 1960; Butler and Stokes 1969). This framework has received much attention, including challenges by empirical analyses of partisan change and by the individual choice framework. In this framework, the “running tally” operates when people update their candidate/leader and issue attitudes and, in turn, their party (non)identifications and electoral choices (Downs 1957; Key 1968; Fiorina 1981).
The above review leads to three principal recommendations for further progress in economicvoting research. One recommendation is that it develop and apply an “anthropology” of voters. This anthropology would recognize that people do think and they do form evaluations, but they are not omniscient and they do use cognitive short-cuts or cues about performance when forming their vote decisions. Second, multiple models – not one model – of economic voting exist and vie for analytical attention. This recommendation is not a plea for the proliferation of models and the rejection of none. Rather, the models and frameworks discussed above require ongoing collections of reliable and valid data and analyses having strong diagnostic tests. These enable assessment of which model is inferior, which model is superior since it can do what other models do and do not, or whether several models contribute coherently to the explanation of economic voting. A third and related recommendation recognizes that models of economic voting are fundamentally nested in models of electoral behavior. Accordingly, the former need to include, rather than ignore, economic attitudes as well as candidate/leader, other issue and party (non)identification attitudes to avoid the risk of misspecification error and its threat to causal inference.
In economic-voting research, the principal dependent variables are vote intentions and vote choices. The former is typically based on survey questions that ask people about whether they intend to vote and, if so, then for which candidate or party in a forthcoming election. Vote choices are usually based on survey questions that ask people whether they did vote and, if so, then for which candidate or party in a just-held election.
The principal explanatory or independent variables tend to be of two types. As discussed above, one type is the objective economy, which involves employment, inflation, investment or growth indicators based on government or other statistical data. These objective indicators have been argued to be the key drivers (see, for example, Kramer 1983; Hibbs 1987; Fair 1978), or to play an important role (see, for example, van der Brug, van der Eijk and Franklin 2007; van der Eijk et al. 2007), in people’s voting and other party support decisions. However, and as also noted above, objective economic indicators say little about how voters think as well as form and act on attitudes about economic performance. Accordingly, another type pertains to the subjective economy. Since approximately the mid-1950s, various election studies and other major surveys have included one or more of several questions enabling multiple measures of economic evaluations. Some questions ask people about the economy in general or about inflation or unemployment in particular, whereas other questions ask them about their own personal economic circumstances. These questions inquire whether people think that the country’s economy has gotten better, gotten worse or stayed the same over the past year (sociotropic retrospection); whether it will get better, get worse or stay the same over the upcoming year (sociotropic prospection); whether their personal economic circumstances have gotten better, gotten worse or stayed the same over the past year (egocentric retrospection); and whether these circumstances will get better, get worse or stay the same over the upcoming year (egocentric prospection) (see Lewis-Beck 1988; Lewis-Beck and Stegmaier 2007).
These four measures have provided valuable service to economic-voting research. However, further research progress depends in part on at least two important recommendations. First, survey instruments need to avoid periodically asking just one, two or three questions, or asking the two egocentric questions combined and the two sociotropic questions combined. Instead, all four evaluation questions need to be asked consistently for proper measurement construction and model testing. A second recommendation is related to the anthropology of the voter. It recognizes that people’s attitudes are evaluative as well as affective and cognitive, and all three have experiential bases. In this regard, more attention needs to be paid to voters’ emotional reactions, such as anger or happiness, and fear or confidence, regarding their own or the country’s economic circumstances (see, for example, Conover and Feldman 1986). And more thought needs to be given to how people acquire economic information indirectly by accessing news provided by various media outlets (see, for example, Mutz and Kim 2017) and/or directly through economic experiences. These experiences include, but are not limited to, employment and income situations or prospects, purchases of food, clothing, shelter and other needs and wants, wealth assets and liabilities, as well as conversations with others that exchange information based on both news and experiences (see, for example, Conover, Feldman and Knight 1987; Feldman and Conley 1991; Lewis-Beck, Nadeau and Foucault 2013).
To date, economic-voting research has relied heavily on observational methods. Much of this research has used aggregate- or individual-level data and multivariate statistical models. These models typically have been single-equation specifications whose parameters are estimated using ordinary least squares regression or maximum likelihood techniques.
Some researchers have long advocated the use of aggregate-level data (see, for example, Kramer 1971, 1983). Indeed, path-breaking studies conducted during the revival of political-economy research in the 1970s and 1980s tended to be simple aggregate-level models that regressed party support in successive national elections (see, for example, Fair 1978) or vote intentions as measured in monthly public opinion polls (see, for example, Mueller 1970), on objective economic indicators involving employment, inflation and/or economic growth rates. The relative simplicity of these models and their promise of providing a tool for forecasting the future dynamics of party support conditional on economic outcomes stimulated considerable excitement in the research community.
However, initial enthusiasm waned when it became evident that these first-generation, aggregate-level models of party support suffered from several intertwined theoretical/modeling and methodological problems. As discussed above, a fundamental theoretical problem is that regressions with macroeconomic aggregates, such as unemployment, inflation or growth rates, as explanatory variables told researchers little about the psychology underpinning voters’ party support decisions. Are voters primarily retrospectively or prospectively oriented? Are they egocentric or sociotropic actors (see, for example, Kinder and Kiewiet 1981)? Are they a homogeneous group whose members all react the same way, or is there significant heterogeneity reflecting varying levels of cognitive capacity and political knowledge (Converse 1964; see also Luskin 1987; Lupia 2016)? Perhaps most basically, what is the anthropology of voters? Are they rational actors closely akin to the “homo economicus” who populates introductory economics texts, or “homo heuristicus” who is found in the experimental research of behavioral economists and cognitive psychologists? Answers to these questions have major implications for the specification and interpretation of models of economic voting and political choice (see van der Brug, van der Eijk and Franklin 2007; Clarke et al. 2009; Clarke, Elliott and Stewart 2015).
Related theoretical/modeling and methodological difficulties that occurred in early aggregatelevel studies and continue to bedevil much contemporary scholarship deserve further mention. As discussed in the “Major models” section above, one difficulty involves “issue-priority” or “issue-ownership” considerations. It has long been evident that party reputations for competence in managing economic affairs, and for being concerned about and responsive to particular economic problems, such as unemployment or inflation (see, for example, Budge and Farlie 1983; Clarke et al. 1992: Chapter 1), can vary over time. Such variations occurred in Great Britain when the currency crisis in September 1992 and the Great Recession starting in 2008 both caused serious erosion in the economic competence reputations of governing parties of the day (see, for example, Whiteley et al. 2013; Clarke et al. 2016). Accordingly, the impact of economic conditions on party support can differ depending on these reputational variations. This can produce parameter instability in dynamic models for analyses conducted in single countries, as well as differences in parameter values in cross-national comparative studies.
A related difficulty is that economic issues are perennially important, but their salience on the political agenda changes over time, co-varying with other domestic and (less frequently) foreign policy issues. However, variables measuring the impact of these noneconomic issues on party support were not included in the early aggregate-level studies and attention to them remains quite uneven. A further difficulty is that individual-level data analyses have long shown that attitudes about party leaders are important explanatory variables in some political systems (Campbell et al. 1960; Clarke et al. 2004; Clarke, Kornberg and Scotto 2009). These variables also were noticeably absent from the early aggregate-level political economy models and, with few exceptions, their omission continues.
The basic point here is that extremely parsimonious specifications of aggregate-level economic-voting models are likely to be a serious liability for assessing the impact of the economy on the dynamics of party support within and across political systems. However technically sound their econometrics, regressions of party vote shares in successive elections or monthly/quarterly polling data on macroeconomic statistics are unlikely to tell us what we need to know about how variations in the economy affects party support within and across countries. Adequate specification of aggregate-level vote and popularity functions requires attention to voters’ reactions to noneconomic events and conditions, but time series data on these events and conditions may be unavailable.
Aggregate-level economic-voting models also have encountered a variety of estimation problems. For example, famous simulations (see, for example, Granger and Newbold 1974) and formal analyses (see, for example, Phillips 1986; Sims, Stock and Watson 1990) convincingly show that regressions involving nonstationary (trending) measures of party support and non-stationary macroeconomic variables encounter “spurious regression” risks that pose serious threats to inference. As conjectured nearly a century ago (Yule 1926), type 1 errors are rife in such regressions. Although early studies typically ignored these risks, most researchers now employ a set of standard unit-root tests to determine whether their data are nonstationary and they difference variables as required before estimating economic-voting models.
Other problems exist. For example, when doing post-estimation diagnostics, researchers who find that residuals are serially correlated often try to “fix” this problem by inserting a lagged endogenous (dependent) variable as an additional predictor variable. If the data are stationary, this practice may be technically sound (if there is no simultaneity bias induced by the presence of the lagged endogenous variable), but it has a significant theoretical implication that may be overlooked. The revised model with the lagged endogenous variable implies that all of the explanatory variables have dynamic effects on the dependent variable, with the size of these effects eroding at the same rate at successively longer lags. Skeptics are right to require additional testing against rival specifications that permit dynamic effects to vary across predictors or to be absent entirely for some of them. Similar considerations regarding theory and interpretations of estimation results apply to other models, such as those estimated using feasible generalized least squares techniques (see, for example, Greene 2011), with implications that are hidden from naive analysts and their readers.
Economic-voting models that use time series data pooled across multiple groups (typically countries) also can be problematic. Like analyses of data for single groups, these panel studies require stationary data, and the use of lagged endogenous variables may be problematic for the reasons just stated. However, there are additional difficulties with panel designs. The most wellknown is unit-induced heteroskedasticity, and researchers have long debated the advantages and disadvantages of using fixed- and random-effects estimators to address this difficulty (see, for example, Bell and Jones 2015).
A basic difficulty involves the pooling operation itself. This operation assumes that all predictor variables have exactly the same effects for all units, that is, there is parameter invariance across space and time. But this assumption is highly implausible since it is very unlikely that, for example, economic-voting effects are the same in different countries over time. Panel unit-root tests can help researchers to assess whether the dynamics of economic conditions or party support vary significantly across units. Moreover, rather than pooling data a priori, analysts often would be well advised to pool only if tests for parameter invariance provide a “green light” or, at a minimum, after “jack-knifing” to investigate whether particular units are driving theoretically interesting findings by exercising undue leverage on parameter estimates.
A final problem with aggregate-level economic-voting models concerns the aforementioned conjecture that endogeneity is causing simultaneity bias. Although critics seldom conjure the specter of endogeneity when party support variables are regressed on objective economic conditions, such as unemployment and inflation, they often do so when covariates are measures of voters’ cognitive or emotional reactions to those conditions. This is especially the case if the model under consideration is an error correction specification that involves a long-run (cointegrating) relationship between economic attitudes and party support. Critics often recommend that the analyst should conduct Granger causality tests (Granger 1969) to check whether endogeneity is present and simultaneity bias poses a threat to inference. The check for endogeneity is prudent. But, Granger causality tests are relevant only for conditional forecasting exercises where the t + i feedback from Y to X needs to be incorporated into a multi-equation specification for forecasts extending beyond the time horizon of the feedback loop.
Given the above, weak exogeneity tests have been recommended to determine whether Y is likely to affect X contemporaneously (see, for example, Charemza and Deadman 1997). But, such tests tend to have weak statistical power and are subject to misspecification because of the absence of strong theoretical rationale for the model of X. Accordingly, the specifications of vector autoregressive (VAR) and, if co-integration is present, vector error correction (VECM) models have been recently recommended (Whiteley et al. 2016; see also Sims 1980; Johansen 1996). The VECM approach has proven to be very useful for modeling the inter-related dynamics among economic evaluations and other key variables in a valence politics model of party support in Britain (Whiteley et al. 2016; see also Clarke et al. 2009). The approach is an attractive way of handling mutually endogenous relationships, but it requires relatively high frequency (e.g., monthly) time series data on all variables under consideration (e.g., economic evaluations, attitudes about other issues and about party leaders). Such data are available for Britain over the 1992–2015 period, but this often is not the case for other places and other times.
Still other economic-voting researchers have used individual-level data that typically rely on survey data, especially those gathered in national election studies. Since its inception nearly 70 years ago, the American National Election Study (ANES) has served as a template for these surveys in a wide range of established democracies as well as emerging ones, such as the Taiwan Election and Democratization Study (TEDS) (see Ho et al. 2013). The usefulness of these national election studies has been significantly enhanced by other survey data collections for conducting cross-national research on economic voting and other topics in comparative perspective. These include the Comparative Study of Electoral Systems (CSES), which places common modules of questions on successive national election studies in multiple countries; the Eurobarometers (see, for example, Lewis-Beck 1988); the European Parliament Election Studies involving 42 cross-sectional surveys conducted in all EC/EU countries after the 1989, 1994 and 1999 elections (van der Brug, van der Eijk and Franklin 2007); the European Social Surveys (ESS); and other projects (see, for example, Duch and Stevenson 2008; for a review, see Lewis-Beck and Stegmaier 2007).
Although many of these national surveys are either “one-shot” post-election cross-sections or two-wave pre- and post-election panels, some are multi-wave panels conducted over several months or entire election cycles. These panel designs were introduced several decades ago (see, for example, Campbell et al. 1960; Butler and Stokes 1969), and they help researchers to assess stability and change in party support and public attitudes about economic conditions and other theoretically relevant variables. This task is not as simple as it might appear because observed change in responses to survey questions at two or more points in time may reflect random measurement error rather than true change (Green, Palmquist and Schickler 2002). But, the task is not insuperable. Four or more waves of panel data permit estimation of parameters for mixed Markov latent class (MMLC) models that assess the magnitude of true (latent-level) change controlling for random measurement error (Clarke et al. 2004, 2009; Clarke and McCutcheon 2009). Estimates of the proportions of “movers” and “stayers” for theoretically important variables like party identification (see, for example, Fiorina 1981; Green, Palmquist and Schickler 2002; Clarke et al. 2004) can help researchers resolve longstanding controversies in the voting behavior literature that are relevant for understanding how economic conditions affect party support.
Using multi-wave panel surveys to estimate MMLC models of economic voting is an example of the more general point that panel surveys can provide significant leverage for addressing statistical issues that can threaten inference in research designs that rely on cross-sectional data. Although their statistical power makes multi-wave panel surveys attractive to economicvoting and other researchers, they are in short supply because they are very expensive. Multiwave surveys also can suffer from sampling problems due to panel attrition and/or measurement problems due to panel-conditioning effects (see, for example, Weisberg 2005).
Persistent problems with cross-sectional designs also arise when a covariate in an estimating equation is correlated with the random error term in that equation. As is well known, this correlation will produce a biased and inconsistent estimate of the effect of that covariate (see, for example, Greene 2011). In economic-voting models, these biases may occur because of endogenous relationships between the dependent variable (a measure of party support) and a covariate (e.g., evaluations of national or personal economic conditions over retrospective or prospective time horizons). The “endogeneity conjecture” is that party support (partisan attachments, vote intention or the vote itself) influences voters’ attitudes about economic conditions, and this reciprocal causal influence creates a simultaneity bias in models that seek to estimate the effects of those attitudes on party support (see, for example, Wlezien, Franklin and Twiggs 1997; Evans and Andersen 2006; van der Eijk et al. 2007).
Some economic-voting students have attempted to minimize or to eliminate endogeneity problems in several ways. One way uses pre-post election panel data, with all covariates measured in the pre-election wave panel and voting reports measured in the post-election wave. Assuming voters cast their ballots on election day (or any time after the first panel wave), it is argued that the act of voting could not have caused pre-election responses to queries about economic evaluations (see, for example, Lewis-Beck 1988). However, critics contend that the endogeneity problem has been inadequately controlled for in many studies that rely on crosssectional or two-wave panel survey data (see, for example, Evans and Chzhen 2016). They further contend that statistical models using data drawn from three or more panel waves show that the dominant flow of influence is from party support to economic evaluations rather than vice versa. Another way uses objective measures rather than subjective assessments of economic conditions (van der Brug, van der Eijk and Franklin 2007). Still other ways of dealing with the problems exist (e.g., Duch and Stevenson 2008; but see Franklin 2009).
In response, researchers who argue that economic attitudes have sizable and consequential influences on party support also argue that the statistical models employed by proponents of the endogeneity hypothesis are seriously flawed for several reasons. These include model misspecification, mismeasurement of key variables and “timing problems” in the conduct of multi-wave surveys such that data are not gathered at points in time coordinated with the flow of economic information in the electorate (Whiteley et al. 2016). The latter problem also may cause these models to have their own unappreciated simultaneity biases.
Attempts to deal with endogeneity problems in individual-level economic-voting models via standard econometric instrumental variables procedures incur difficulties. These include the absence of suitable instruments (but see Nadeau, Lewis-Beck and Belanger 2013), the lack of statistical power of standard endogeneity tests and the absence of strong theory to mandate the specification of models used for endogeneity testing. The latter two problems can create a further problem if researchers join the “frequent regressors club” by conducting specification searches that continue until a preferred (exogeneity or endogeneity) test result is obtained.
Since the revival of political-economy research in the early 1970s, the economic-voting field has witnessed substantial progress as well as enduring controversies. Some scholars initially attracted to the field by the “promise of parsimony” discovered the field to be more difficult than originally thought. Similar to other areas of social-science inquiry, economic-voting studies experience theoretical, modeling and methodological challenges involving specification and analysis of multivariate models of observational data. The usefulness of these analyses depends on the quality of the theories guiding model specification and the availability of data that permit analysts to estimate parameters with confidence. In addition, and motivated by a desire to make broad generalizations across countries and over time, economic-voting students have encountered significant “contextual challenges” posed by varying combinations of economic, political and institutional differences between political systems and significant change within systems over time.
As discussed in this chapter, economic-voting researchers have reacted to these several challenges with industry and ingenuity and many valuable studies have appeared. More remains to be done. Particularly important will be the acquisition of high-frequency time series data that can facilitate dynamic analyses which address model specification and parameter estimation problems discussed above. Coordinated cross-national data collections of this type would be especially valuable for systematic analyses of the influence of contextual factors on economic voting.
Equally important, researchers need to address basic questions about the anthropology of the voter and their implications for theories and models of political choice. Doing so requires close attention to the results of research by behavioral economists and cognitive psychologists, as well as those by political psychologists that document significant heterogeneity of people’s levels of political sophistication. This recommendation may be seen as a variant of the challenge that Herbert Simon (1956) posed over half a century ago when he scandalized fellow economists by daring to conjecture that the world is populated by real people who satisfice, rather than by abstract agents who optimize. A similar challenge is relevant today for students of economic voting who wish to advance their field of inquiry in theoretically interesting and realityoriented ways.
Acemoglu, D. and Robinson, J. A. (2012) Why Nations Fail: The Origins of Power, Prosperity and Poverty, New York: Crown Publishers/Random House.
Anderson, C. J. (1995) Blaming the Government: Citizens and the Economy in Five European Democracies, Armonk: M. D. Sharpe.
Bell, A. and Jones, K. (2015) “Explaining Fixed Effects: Random Effects Modeling of Time-Series Crosssectional and Panel Data,” Political Science Research and Methods, vol. 3, no. 1, January, 133–153.
Bloom, H. and Douglas Price, H. (1975) “Voter Response to Short-Run Economic Conditions: The Asymmetric Effects of Prosperity and Recession,” American Political Science Review, vol. 69, no. 4, December, 1240–1254.
Brug, W. van der, Eijk, C. van der and Franklin, M. N. (2007) The Economy and the Vote: Economic Conditions and Elections in Fifteen Countries, New York: Cambridge University Press.
Budge, I. and Farlie, D. J. (1983) Explaining and Predicting Elections: Issue Effects and Party Strategies in Twenty-Three Democracies, London: George Allen & Unwin.
Butler, D. and Stokes, D. E. (1969) Political Change in Britain: Forces Shaping Electoral Choice, London: Macmillan.
Campbell, A., Converse, P. E., Miller, W. E. and Stokes, D. E. (1960) The American Voter, New York: John Wiley & Sons.
Charemza, W. W. and Deadman, D. F. (1997) New Directions in Econometric Practice, 2nd Edition, Aldershot: Edward Elgar Publishing.
Clarke, H. D. and McCutcheon, A. (2009) “The Dynamics of Party Identification Reconsidered,” Public Opinion Quarterly, vol. 73, no. 4, Winter, 704–728.
Clarke, H. D. and Stewart, M. C. (1994) “Prospections, Retrospections, and Rationality: The ‘Bankers’ Model of Presidential Approval Reconsidered,” American Journal of Political Science, vol. 38, no. 4, November, 1104–1123.
Clarke, H. D., Elliott, E. W., Mishler, W., Stewart, M. C., Whiteley, P. F. and Zuk, G. (1992) Controversies in Political Economy: Canada, Great Britain, the United States, Boulder: Westview Press.
Clarke, H. D., Elliott, E. W. and Stewart, M. C. (2015) “Heuristics, Heterogeneity and Green Choices: Voting on California’s Proposition 23,” Political Science Research and Methods, available online: https://doi.org/10.1017/psrm.2015.39.
Clarke, H. D., Kellner, P., Stewart, M. C., Twyman, J. and Whiteley, P. (2016) Austerity and Political Choice in Britain, Basingstoke: Palgrave Pivot/Macmillan.
Clarke, H. D., Kornberg, A. and Scotto, T. J. (2009) Making Political Choices: Canada and the United States, Toronto: University of Toronto Press.
Clarke, H. D., Sanders, D., Stewart, M. C. and Whiteley, P. (2004) Political Choice in Britain, Oxford: Oxford University Press.
Clarke, H. D., Sanders, D., Stewart, M. C. and Whiteley, P. (2009) Performance Politics and the British Voter, Cambridge: Cambridge University Press.
Conover, P. J. and Feldman, S. (1986) “Emotional Reactions to the Economy: I’m Mad as Hell and I’m Not Going to Take It Anymore,” American Journal of Political Science, vol. 30, no. 1, February, 50–78.
Conover, P. J., Feldman, S. and Knight, K. (1987) “The Personal and Political Underpinnings of Economic Forecasts,” American Journal of Political Science, vol. 31, no. 3, August, 559–583.
Converse, P. E. (1964) “The Nature of Belief Systems in Mass Publics,” in Apter, D. E. (ed.) Ideology and Discontent, Glencoe: The Free Press: 206–261.
Downs, A. (1957) An Economic Theory of Democracy, New York: Harper and Row.
Duch, R. M. and Stevenson, R. T. (2008) The Economic Vote: How Political and Economic Institutions Condition Election Results, New York: Cambridge University Press.
Eijk, C. van der, Franklin, M. N., Demant, F. and van der Brug, W. (2007) “The Endogenous Economy: ‘Real’ Economic Conditions, Subjective Economic Evaluations and Government Support,” Acta Politica, vol. 42, no 1, April, 1–22.
Evans, G. and Andersen, R. (2006) “The Political Conditioning of Economic Perceptions,” Journal of Politics, vol. 68, no. 1, February, 194–207.
Evans, G. and Chzhen, K. (2016) “Re-Evaluating the Valence Politics Model of Political Choice,” Political Science Research and Methods, vol. 4, no. 1, January, 199–220.
Fair, R. C. (1978) “The Effects of Economic Events on Votes for President,” The Review of Economics and Statistics, vol. 60, no. 2, April, 159–173.
Feldman, S. and Conley, P. (1991) “Explaining Explanations of Changing Economic Conditions,” in Norpoth, H., Lewis-Beck, M. S. and Lafay, J. D. (eds.) (1991) Economics and Politics: The Calculus of Support, Ann Arbor: University of Michigan Press: 185–206.
Fiorina, M. P. (1981) Retrospective Voting in American National Elections, New Haven: Yale University Press.
Franklin, M. N. (2009) “Book Review – The Economic Vote: How Political and Economic Institutions Condition Election Results by Raymond M. Duch and Randolph T. Stevenson,” Perspectives on Politics, vol. 7, no. 1, March, 202–203.
Gigerenzer, G. (2008) Rationality for Mortals: How People Cope with Uncertainty, Oxford: Oxford University Press.
Gigerenzer, G., Hertwig, R. and Pachur, T. (eds.) (2011) Heuristics: The Foundations of Adaptive Behavior, Oxford: Oxford University Press.
Granger, C. (1969) “Understanding Causal Relations by Econometric Models and Cross-Spectral Methods,” Econometrica, vol. 37, no. 3, August, 424–438.
Granger, C. and Newbold, P. (1974) “Spurious Regression in Econometrics,” Journal of Econometrics, vol. 2, no. 2, July, 111–120.
Green, D., Palmquist, B. and Schickler, E. (2002) Partisan Hearts and Minds, New Haven: Yale University Press.
Greene, W. (2011) Econometric Analysis, 7th Edition, Upper Saddle River: Pearson Education Ltd.
Healy, A. and Malhotra, N. (2013) “Retrospective Voting Reconsidered,” Annual Review of Political Science, vol. 16, May, 285–306.
Hibbs, D. A. (1987) The American Political Economy: Macroeconomics and Electoral Politics, Cambridge, MA: Harvard University Press.
Ho, K., Clarke, H. D., Chen, L. K. and Weng, D. L. C. (2013) “Valence Politics and Electoral Choice in a New Democracy: The Case of Taiwan,” Electoral Studies, vol. 32, no. 3, September, 476–481.
Johansen, S. (1996) Likelihood-Based Inference in Cointegrated Vector Autoregressive Models, Oxford: Oxford University Press.
Kahneman, D. (2011) Thinking, Fast and Slow, New York: Farrar, Strauss and Giroux.
Key, V. O. (1968) The Responsible Electorate: Rationality in Presidential Voting, 1936–1960, New York: Vintage Books.
Keynes, J. M. (1936) The General Theory of Employment, Interest and Money, London: Macmillan.
Kiewiet, D. R. (1983) Macroeconomics and Micropolitics: The Electoral Effects of Economic Issues, Chicago: University of Chicago Press.
Kinder, D. R. and Kiewiet, D. R. (1981) “Sociotropic Politics: The American Case,” British Journal of Political Science, vol. 11, no. 2, April, 129–161.
Kramer, G. H. (1971) “Short-Term Fluctuations in U.S. Voting Behavior, 1896–1964,” American Political Science Review, vol. 65, no. 1, March, 131–143.
Kramer, G. H. (1983) “The Ecological Fallacy Revisited: Aggregateversus Individual-Level Findings on Economics and Elections and Sociotropic Voting,” American Political Science Review, vol. 77, no. 1, March, 92–111.
Lewis-Beck, M. S. (1988) Economics and Elections: The Major Western Democracies, Ann Arbor: University of Michigan Press.
Lewis-Beck, M. S. and Costa Lobo, M. (2017) “The Economic Vote: Ordinary vs. Extraordinary Times,” in Arzheimer, K., Evans, J. and Lewis-Beck, M. S. (eds.) The Sage Handbook of Electoral Behaviour, London: Sage Publications: 584–607.
Lewis-Beck, M. S. and Stegmaier, M. (2000) “Economic Determinants of Electoral Outcomes,” Annual Review of Political Science, vol. 3, June, 183–219.
Lewis-Beck, M. S. and Stegmaier, M. (2007) “Economic Models of Voting,” in Dalton, R. J. and Klingemann, H. D. (eds.) The Oxford Handbook of Political Behavior, New York: Oxford University Press: 518–537.
Lewis-Beck, M. S. and Stegmaier, M. (2013) “The V-P Function Revisited: A Survey of the Literature on Vote and Popularity Functions after over 40 Years,” Public Choice, vol. 157, no. 3, December, 367–385.
Lewis-Beck, M. S., Nadeau, R. and Foucault, M. (2013) “The Compleat Economic Voter: New Theory and Evidence,” British Journal of Political Science, vol. 43, no. 2, April, 241–261.
Lupia, A. (2016) Uninformed: Why People Seem to Know So Little About Politics and What We Can Do About It, New York: Oxford University Press.
Luskin, R. (1987) “Measuring Political Sophistication,” American Journal of Political Science, vol. 31, no. 4, November, 856–899.
MacKuen, M. B., Erikson, R. S. and Stimson, J. A. (1992) “Peasants or Bankers: The American Electorate and the U.S. Economy,” American Political Science Review, vol. 86, no. 3, September, 597–611.
Monroe, K. R. (1979) “God of Vengeance and Reward?: The Economy and Presidential Popularity,” Political Behavior, vol. 1, no. 4, December, 301–329.
Monroe, K. R. (ed.) (1991) The Economic Approach to Politics: A Critical Reassessment of the Theory of Rational Action, New York: Harper Collins.
Mueller, J. E. (1970) “Presidential Popularity from Truman to Johnson,” American Political Science Review, vol. 64, no. 1, March, 18–34.
Mutz, D. C. and Kim, E. (2017) “Economic Voting in a New Media Environment: Preliminary Evidence and Implications,” in Arzheimer, K., Evans, J. and Lewis-Beck, M. S. (eds.) The Sage Handbook of Electoral Behaviour, London: Sage Publications: 733–758.
Nadeau, R., Lewis-Beck, M. S. and Belanger, E. (2013) “Economics and Elections Revisited,” Comparative Political Studies, vol. 46, no. 5, May, 551–573.
Norpoth, H., Lewis-Beck, M. S. and Lafay, J. D. (eds.) (1991) Economics and Politics: The Calculus of Support, Ann Arbor: University of Michigan Press.
Phillips, P. (1986) “Understanding Spurious Regressions in Econometrics,” Journal of Econometrics, vol. 33, no. 3, December, 311–340.
Powell, G. B. and Whitten, G. D. (1993) “A Cross-National Analysis of Economic Voting: Taking Account of the Political Context,” American Journal of Political Science, vol. 37, no. 2, May, 391–414.
Schwindt-Bayer, L. and Tavits, M. (2016) Clarity of Responsibility, Accountability and Corruption, New York: Cambridge University Press.
Simon, H. A. (1956) “Rational Choice and the Structure of the Environment,” Psychological Review, vol. 63, no. 2, May, 129–138.
Sims, C. A. (1980) “Macroeconomics and Reality,” Econometrica, vol. 48, no. 1, January, 1–48.
Sims, C. A., Stock, J. H. and Watson, M. W. (1990) “Inference in Linear Time Series Models with Some Unit Roots,” Econometrica, vol. 58, no. 1, January, 113–144.
Smith, A. (1776) An Inquiry into the Nature and Causes of the Wealth of Nations, Volume 1, London: W. Strahan & T. Cadell.
Sniderman, P. M., Brody, R. A. and Tetlock, P. R. (eds.) (1991) Reasoning and Choice: Explorations in Political Psychology, Cambridge: Cambridge University Press.
Stegmaier, M., Lewis-Beck, M. S. and Park, B. (2017) “The V-P Function: A Review,” in Arzheimer, K., Evans, J. and Lewis-Beck, M. S. (eds.) The Sage Handbook of Electoral Behaviour, London: Sage Publications: 608–630.
Stimson, J. A. (1989) “Perceptions of Politics and Economic Policy: A Macroanalysis,” in Clarke, H. D., Stewart, M. C. and Zuk, G. (eds.) Economic Decline and Political Change: Canada, Great Britain, the United States, Pittsburgh: University of Pittsburgh Press: 195–221.
Stokes, D. E. (1963) “Spatial Models of Party Competition,” American Political Science Review, vol. 57, no. 2, June, 368–377.
Stokes, D. E. (1992) “Valence Politics,” in Kavanagh, D. (ed.) Electoral Politics, Oxford: Clarendon Press: 141–164.
Thaler, R. H. (2016) Misbehaving: The Making of Behavioral Economics, New York: W. W. Norton & Company.
Vig, N. J. and Schier, S. E. (eds.) (1985) Political Economy in Western Democracies, New York: Holmes and Meier.
Weisberg, H. (2005) The Total Survey Error Approach: A Guide to the New Science of Survey Research, Chicago: University of Chicago Press.
Whiteley, P. (1986) Political Control of the Macro-Economy, London: Sage Publications.
Whiteley, P., Clarke, H. D., Sanders, D. and Stewart, M. C. (2013) Affluence, Austerity and Electoral Change in Britain, Cambridge: Cambridge University Press.
Whiteley, P., Clarke, H. D., Sanders, D. and Stewart, M. C. (2016) “Hunting the Snark: A Reply to ‘Re-Evaluating Valence Politics Models of Political Choice,’ ” Political Science Research and Methods, vol. 4, no. 1, January, 221–240.
Wlezien, C., Franklin, M. N. and Twiggs, D. (1997) “Economic Perceptions and Vote Choice: Disentangling the Endogeneity,” Political Behavior, vol. 19, no. 1, March, 7–17.
Yule, G. U. (1926) “Why Do We Sometimes Get Nonsense-Correlations Between Time-Series – A Study in Sampling and the Nature of Time-Series,” Journal of the Royal Statistical Society, vol. 89, no. 1, January, 1–63.