Political forecasting

---

Note: This article is intended to replace the current "political forecasting" page, the latter of which I intend to relocate to a new page called "election forecasting." The current political forecasting page is narrowly focused on election forecasting, which is a subfield of political forecasting. Political forecasting as a field has greatly expanded over the last ~twenty years both in academia and in the private sector, with ample credible sources in both that detail new methodologies in forecasting broad ranges of developments including economic developments, political regime evolution, forecasts of diplomatic agreements, forecasts of trade and economic shifts, and the outcomes of social movements to name a few. The use of index forecasting is also now common place. I propose this new article because the current election forecasting article is too well-developed to not be its own article at this point, especially if it narrowly describes a subset of the field.

For election forecasting, see Political forecasting (disambiguation).

Political forecasting is the process of predicting the outcomes of events involving politicians, political institutions, and matters of policy. The field of political forecasting is broadly concerned with predicting the outcome of various political events, especially diplomatic agreements, legislative and policy outcomes, as well as electoral outcomes and the outcomes of events as the consequence of leader survival strategies and exogenous factors. Political forecasting is a part of political science.

The methodology involved in predicting the outcome of political events is diverse and oftentimes multiple methodologies are employed simultaneously. Quantitative political forecasting includes the use of statistical models, expected utility models, game theory, and economic forecasting; notably, quantitative methodology may rely on input data derived from qualitative sources. Qualitative political forecasting involves the use of expert opinion, survey data, punditry by experts or personalities, and testimony by current or former political operatives. Solely qualitive forecasting is typically considered to be less reliable than quantitative forecasting, especially for complex events, though remains effective in its own right. Additionally, mainstream and heterodox perspectives exist that use established profiles of quantitative and qualitative methodology, and these perspectives have realized varying levels of success.

The applications of political forecasts are vast. Political forecasting has been used in American intelligence agencies to provide decision makers with the information needed to implement policies that achieve objectives faster and with less waste than without forecasts. Political forecasting can be used to update voter's beliefs and help inform decisions amongst voters based on the perceived consequences of various electoral outcomes. Political forecasting has also been used to forecast the outcome of elections, which is related to branch of political science concerning the study of elections known as psephology.

Political forecasting is difficult to disaggregate from political science more generally, where findings by political scientists and economists are often used to inform policy makers about the parameters of future policy events. In many ways the implications from the broader field of theoretical political science can be interpreted as predictive in nature, in the same way that experiments in physics can bear implications for the study of systems. Regardless, the invocation of both general political scientific knowledge and explicit predictive models is common place amongst forecasters and decision makers.

History of Political Forecasting

Premodern Forecasting

The act of political forecasting using any methodology dates back at least to ancient human civilization. The Pythia was the name of the high priestess who served as the oracle at the Temple of Apollo at Delphi in Ancient Greece where she was consulted by numerous Greek military leaders from as early as 1400 BC to the late 4th century AD. Other ancient forecasters include Thucydides and Chanakya. Michel de Nostredame, popularly known as Nostradamus, is well known in popular culture for his predictions, yet the current consensus amongst scholars is that his works are so vague as to remain unfalsifiable. The methodologies used by pre-19th century forecasters range from using primitive structure models, an example is Thucydides' conceptualization of what is now called Thucydides' Trap, to practicing spiritual or otherwise religious rituals.

History of Predictive Models in Political Science

The history of models in political science concerns the history of the introduction of predictive models to the study of political events. In modern times, especially since the 1940s and 50s, predictive mathematical models have played an increasingly important role in the evolution of political forecasting. However, the history of the use of these models in political science is difficult to isolate from other coterminous developments in the social sciences. This is due to the fact that many of the elementary models used throughout political science, especially in rational choice theory and decision theory, as well as in game theory, were originally developed for application in the economic sciences, and economic and political policy analysis often times involve similar audiences and employ similar methodologies. Because of this, the history of models in political science is best described as the history of models in the economic and political sciences more generally with later developments in scientific theory, technology, and data gathering expanding the applicability of models to primarily political processes, such as in the analysis of decisions by individuals in government or by public institutions.

The use of mathematical models in the social sciences began long before the development of political science as a highly empirical field. Some of the earliest general models relating to choice were developed in the 18th century, such as the discovery of the St. Petersburg paradox by Nicolas Bernoulli. The use of probability theory and decision theory remained integral, through not dominant, in the study of economic sciences for the next century and a half, until the expansion of economics into a rigorously mathematical, scientific framework by economist in the late 19th century, especially by Italian economist, sociologist, and philosopher Vilfredo Pareto and English economist Alfred Marshall.

More narrowly amongst political scientists, the use of models in political science especially accelerated beginning with the introduction of various mathematical concepts and the invention and development of formal modelling techniques in the beginning of the 20th century. The development of the mathematical framework of game theory in the early 20th century, most notably during the 1940s and 50s, was especially beneficial for political forecasting. There have been subsequent major developments in political forecasting in every decade since. Perhaps the most consequential and unique concept introduced in the mid-20th century was the formalization of the median voter theorem by economist Duncan Black in 1948. Black's median voter theorem states that for voters in a unidimensional policy space (meaning that tradeoffs only exist upon one dimension, as opposed to multidimensional policy spaces where changes in one aspect of a policy might affect another) the voter with the median preference casts the deciding vote in a majoritarian vote between two alternatives. The median voter theorem is now typically interpreted in either a strong or weak form, where the strong form means the median voter decisively determines a policy outcome while the weak form means the median voter is the deciding factor in the election of a decision maker, who may or may not implement policies with varying levels of efficacy.

In present times, political science continues to benefit heavily from the development of models in the economic sciences. While many of these models are not explicitly designed for forecasting purposes, their use is difficult to disaggregate from forecasting methods because so many forecasting methods predicate their logic on theories developed in the field of economics and politics at large.

The rise of modern political economy

Political economy is a subfield of political science that seeks to analyze interactions between economics and political institutions. Modern political economy should not be confused with the field of the same name that predated modern economics but was dissolved in favor of economics at the end of the 19th century. Modern political economy analyzers interaction of market fundamentals, decisions made by firms, and decisions made by leaders of governments. The study of political economy has proven valuable to political forecasting, where modern political economic theories have informed forecasters about the expected utility of different decisions.

Quantitative Forecasting Models

Quantitative methodology in political science has seen tremendous growth in the previous decades; however, alleged erroneous use of various methods by political scientists has received criticism from within the field. Nonetheless, substantial progress has been made in the application and efficacy of quantitative models to political science as a whole, especially those employed in forecasting efforts.

Data Used in Quantitative Models

Data used in quantitative political forecasting models may be either purely objective, often economic data or data that encodes qualitative judgments. Quantitative models may make use of exclusively primary data or estimated data, or may contain both. Examples of primary quantitative data would be data that is externally verifiable, such as economic data or voter rolls. Examples of estimated quantitative data would include expert opinion translated into quantitative data. An example of such translated data could include a consultant estimating another party's risk aversion on a normalized binary interval. Such estimated data could be used in any number of models.

Rational Choice and Expected Utility Models

Rational choice theory is the study of decision making within a narrow band of axiomatic assumptions about actors making decisions. Expected utility models use rational choice axioms to evaluate the outcome of broad ranges of events. Rational choice modeling is popular amongst political forecasters. The most successful expected utility models incorporate known empirical phenomenon in political science in their forecasting of political events.

Expected utility models exist within the study of expected utility theory. The theory of expected utility was developed by John von Neumann and Oskar Morgenstern, who are also the founders of game theory. The four rationality axioms proposed by von Neumann and Morgenstern that define rational actors are the following on a system U consisting of entities u, v, w, ...:

1. Completeness and transitivity: for any two u, v either u > v, u < v, or u = v and where u > v and v > w imply u > v > w.
2. Monotonicity: EU(u) > EU(v) if and only if u > v where EU(x) denotes the expected utility of x.
3. Continuity: for v > w > u, an α implies αu + (1 – α)v < w.
4. Independence of substitutability: EU(αu + (1 – α)v) = EU(L + (1 – α)v) where L = αu.:

Expected Utility in Multiple Dimensions

A substantial literature exists that is dedicated to understanding policy outcomes in n-dimensional policy spaces. The results of this literature have demonstrated the limits that are a consequence of expected utility. Economist Charles R. Plott discovered that some supermajority of actors will always prefer a different outcome to any defined point in a multidimensional bargaining space. Game theorist Kenjiro Nakamura in 1979 introduced the Nakamura number, which finds that complex rules on in a multidimensional bargaining space are not sufficient to prevent the existence of a supermajority of voters preferring a different status quo at any time. Nakamura's number implies that any bargaining space without a player with veto power is incapable of delivering an outcome that can't be beaten, meaning there is no certainty with which multi-dimensional issues can be forecasted. It is of note that Nakamura's finding was preceded by mathematician and meteorologist Edward Lorenz who developed the Lorenz system in describing the unpredictability of particles in a hydrodynamic medium using ordinary nonlinear differential equations. Lorenz' work highlights the instability of object position in a multi-dimensional, interactive space, which is highly analogous to the study of voters with preferences of varying magnitudes in a multidimensional policy space.

Bruce Bueno de Mesquita

Perhaps the most well-known forecaster to use rational choice assumptions is political scientist Bruce Bueno de Mesquita. Bueno de Mesquita claims a declassified Central Intelligence Agency report on his median voter theorem based Policon model system attests that his model is 90% accurate. This model has been superseded in his own work by a Bayesian model that probabilistically forecasts policy outcomes in a multidimensional setting using Schofield's mean voter theorem. The accuracy of this latter model has not been publicly disclosed.

Publicly available information on Bueno de Mesquita's forecasting models details that he assigns at least four values to each actor described in his forecasting model, namely measures of their position, their salience, their total capabilities, and the total bargaining range they are expected to tolerate. Data are collected from consultation with experts. Bueno de Mesquita notes that, although the experts he consults with often project wildly different future outcomes of policy events from one another, there is little variance in the data he collects from them. Bueno de Mesquita has published many predictions in academic journals that detail his results. He has made thousands of predictions, most of which are performed privately for clients.

Despite Bueno de Mesquita's models' fame, none of this models have ever been made full public. Despite this, Bueno de Mesquita has claimed that his publications on the model's components (published in academic journals, separately from his predictions) contain enough information on the model that "anyone (may) replicate something close to [my] work." The Policon model has been essentially recreated by Jason B. Scholz, Gregory J. Calbert, and Glen A. Smith of the C31 Division, Defence Science and Technology Organisation in South Australia. At least one full rendering of Scholz, Calbert and Smith's recreation is publicly available, though several discretions between that implementation and Scholz, Calbert, and Smith's writings exist. There is no currently available version of his Bayesian model.

Criticism of Bueno de Mesquita's 90% accuracy figure

The statement by Bueno de Mesquita that his predictions are 90% accurate has been called into question by Major Jonathon S. Seal of the United States Air Force. Maj. Seal writes that Bueno de Mesquita had lifted the figure from a study authored by an analyst who used to work for the CIA, and claims that the analyst merely commented that Bueno de Mesquita's predictions are highly detailed 90% of the time and aren't necessarily accurate. However, these notes are possibly misinterpretations, as the author of the study did have access to forecasts performed by Bueno de Mesquita for the CIA, and the highly detailed figure is in reference to the quality of specifically accurate forecasts. Furthermore, Bueno de Mesquita's forecasting methods were criticized by Maj. Seal for being highly accurate only in combination with other traditional methods. While the 90% figure is derived from a combination of traditional forecasting methods and Bueno de Mesquita's methods, Bueno de Mesquita has never claimed to not use traditional methods in his forecasting. Thus, the claim that Bueno de Mesquita's predictions are 90% accurate and are more specific than using purely traditional methods is still substantiated by one external audit.

Forecasts Using Game Theory

Game theoretic methodology is often used in quantitative political forecasting. Game theory a branch of mathematics. Game theory was originally designed to predict the outcome of economic events yet has been increasingly applied to political science. The literature concerning the application of game theory is vast and is not limited to forecasting.

Simple game theoretic forecasts

Game theoretic predictions are possible using very primitive models. In some sense, all game theory models are potentially forecasting models if given enough prior information.

Imagine that two players, Abdul and Sara, player a simultaneous game with no Nash equilibrium. John Forbes Nash Jr. proved that for every finite game there is at least one Nash equilibrium with mixed strategy. If Abdul and Sara both have strategy priors that are known to an outside observer, the outside observer can use those priors to solve for the Nash equilibrium out their interaction ex-ante.

Spatial Models

Spatial models are models that take advantage of the dimensionality of various policy spaces; in particular, policy spaces that can be ranked or ordered on a set. The use of spatial modelling in political science is not new but has increased over the last fifteen years. Spatial components are widespread in quantitative political forecasting.

Qualitative Forecasting Models

Expert Opinion

Expert opinion can be used to forecast the outcome of an event. Policy experts are required to have an informed understanding of all of the actors involved in a political event in order to make an informed prediction of the outcome of an event. There is evidence that expert opinion is less reliable than some poll aggregation methods in election forecasting. Furthermore, the use of expert opinion as the sole means for predicting the outcomes of policy events has been described by at least one prominent political scientist and forecaster as being inadequate; however, the same political scientist makes regular use of expert opinion in soliciting data used in formal models.

Survey Data

The use of survey data can be used to forecast future events by interviewing or collecting data on large numbers of people. The use of rigorous models on survey data is the most common form of election forecasting.

Punditry

Further information: pundit

The consultation of people for their opinion in a mass media environment is called punditry. Pundits can be experts in a specific field or simply political ideologues. The difference between an expert pundit and an ideological pundit is that expert pundits may make falsifiable predictions whereas ideological pundits offer their opinion which may or may not be a function of loyalty to a political movement. However, punditry is a broad field and many people who are called upon for their expert opinions might also offer ideological reasons for their predictions.

Perspectives in Political Forecasting

There are several different perspectives that define unique applications of quantitative and qualitative forecasting methods. Perspectives in political forecasting may also be rigidly philosophical. The following paradigms are commonly employed perspectives within political forecasting and political science as a whole. The following perspectives variously make use of quantitative methodologies, qualitative methodologies, or different philosophies and are commonly used by analysts today. The unit of analysis in each perspective ranges from focusing dogmatically on nation states to focusing on individuals. The degree of success of each of these various perspectives is wide-ranging, with several of the following perspectives receiving substantial criticism from within academic publications, while others have received praised. There is currently no single preeminent perspective in political forecasting.

International Relations Theory

International relations theory is the study of international relations from a theoretical perspective. International relations theory is not to be confused with the broader field of international relations: international relations theory has come to define, for various reasons, a number of specific theories in political science, most famously realism, liberalism, and constructivism, though many others exist, such as Marxist international relations theory. International relations theory is widely believed to be its own subdiscipline in political science. Many of the theories contained within international relations theory are intended to be predictive in nature and many analysts ascribe to standard theories in international relations theory; however, the subdiscipline as a whole and various theories within it have been criticized by other political scientists.

Constructivism

Constructivism is a perspective in international relations theory. Constructivists claim that norms in international relations are historically constructed. Constructivists assume that actors act according to their identity and that it is impossible to predict with their identity becomes visible or not.

Liberalism

Liberalism is a theory in international relations theory that provides that world events are shaped by actors seeking greater amounts of liberty that provides the rationale for the proliferation of trade and commerce in a world dominated by an ascendant market economy based world. It predicts that market driven forces pacify world politics and leads to increased living standards.

Liberalism's main hypotheses have been falsified. Several political economists have pointed out the role of domestic politics in shaping international trade, especially economists studying global value chains and the efficacy of free trade disputes. There is little evidence to suggest that free trade norms have been Pareto optimal or that increased trade leads monotonically to reductions in violence. Furthermore, the study of the efficacy of protectionism has provided substantial evidence to suggest that free trade does not universally increase wages, demonstrating that capabilities are not always the primary concern for forging more liberal trade policies.

Neorealism and Realism

Neorealism is a structural approach in international relations theory that forecasts that states will prioritize their own security needs above everything else.

The main central hypotheses of neorealist analysis were identified by Emerson Niou and colleagues as follows, and were restated in a political science textbook by political scientist Bruce Bueno de Mesquita

1. Essential states never become inessential.
2. Essential states are never eliminated from the international system.
3. Inessential states never become essential states.
4. Inessential states are always eliminated from the international system.

Bueno de Mesquita has written that these hypotheses have failed to demonstrate predictive power in forecasting political events.

Balance of Power

The balance of power perspective is a component of neorealism and realism that analyzes the ratio in capabilities between states as a causal mechanism for studying patterns in political events. It is a structural model. The balance of power perspective asserts that states seek to form coalitions that prevent the emergence of a hegemonic power that has the unilaterally advantaged position that threatens the security of other states.

The balance of power perspective has received substantial criticism by political scientists. Political scientists Emerson M. S. Niou and Peter C. Ordeshook have found that stability can exist in anarchic international systems under unique circumstances.

Positive Approaches

Main article: Positive political theory

The application of positive theory to political science is the process of applying purely quantitative methods to political science. It is a broad term that describes large, occasionally disparate fields of political science. The term is essentially synonymous with the broad application of quantitative methodology to political science.

Power Transition Theory

Power transition theory was first posited by political scientist A. F. K. Organski in his 1958 book World Politics. The power transition theory hypothesizes that states exist in a hierarchical structure with a hegemonic power satiating a coalition of great powers that provide the hegemon additional support, and that approval of the hegemonic state declines as states become weaker. It is a structural model. Organski posited that the majority of the world's population lives in middle power states or weak states that are too disorganized to rebuke the hegemon. The hegemon comes under existential threat whenever formerly weak states become more powerful, and the hegemon declares war on rising powers to secure its position future advantaged position. Organski therefore hypothesizes that world wars (and other regional wars) are a consequence of hegemons and rising powers coming into conflict over periods of transition.

Empirical evidence in support of the power transition theory is tentative, as noted by Organski himself, alongside fellow political scientist Jacek Kugler. Their work finds inconclusive evidence that power transitions are a necessary and sufficient condition for war, though there is a suggestion that equal power amongst a hegemon and a rising power leads to war more often than other arrangements. However, Kim and Morrow find intervals in which states are willing to engage in conflict need not depend on equality or even near equality of capabilities. Political scientist Robert Powell has also found substantial empirical evidence that weak states will sometimes engage in war even the weaker state fully expects to lose, anticipating that the intermediate payoff from war provides a higher utility to the weaker state than a peacetime when the weaker state is thoroughly dominated by a stronger state.

The Strategic Perspective

The strategic perspective is a political science perspective with significant predictive qualities developed by political scientists Bruce Bueno de Mesquita, Alastair Smith, James D. Morrow, and Randolph M. Siverson. These political scientists created the strategic perspective as an outgrowth of their Selectorate Theory, which is a comprehensive theory of government that defines institutions on continuous variables that allows for the theoretically infinite reclassification of regimes on a non-categorical spectrum. The selectorate theory is a highly cited concept in political science. The perspective has inspired numerous implementations of advanced, quantitative forecasting models that continue to see increasing success and adoption in the field of political science. The strategic perspective borrows heavily from the understanding of government derived from Bueno de Mesquita's forecasting model, the expected utility model.

Success

The most successful means of political forecasting remains the use of quantitative models guided by expert opinion, though such methodology is still not preeminent in the field of political forecasting. As the field of empirical social sciences becomes more technologically advanced, advanced quantitative methods have in general become more common and more precise. This advancement is due to a confluence of factors such as increased levels of data and increased access to data, the increased capabilities of computers, the number of researchers in political science, coterminous developments in the field of the economic sciences (especially in econometrics), and the advent of more and more advanced methods of causal inference.

Election Forecasting

Main article: election forecasting.

Election forecasting is the practice of predicting the outcome of an election. Election forecasting is the application of forecasting methodologies to the narrow band of political events concerning leadership succession in places with regularly scheduled, meaningful, and binding elections (read: democracies), especially where publicly available polling data is common. Election forecasting is a broad field in its own right. Election forecasting is often conducted for mass media markets, though private forecasts are used by decision makers within governments to make foreign policy decisions. Election forecasting has seen a rise in popularity on the internet, yet suffers from deficiencies such a lack of data, which is a result of very few elections that have featured high quality polling data. Additionally, methods to predict the outcomes of elections remain heterodox and are often ad hoc in nature.

Criticisms of Political Forecasting

Political forecasting shares many of the same criticisms as many choice theories, with many arguing that fields such as rational choice theory are best thought of as prescriptive or normative studies. Additionally, the definition of utility continues to be elusive, with the concept essentially being unverifiable. As many quantitative political forecasting methods make frequent use of utility functions to assign weight to strategy profiles in forecasting models, these models suffer from the same criticism that is generally directed at the use of utility in the economic sciences. British post-Keynesian economist Joan Robinson is a notable objector to the logical basis of utility.

Apart from criticism at the level of fundamentals, large numbers of political scientists remain skeptical of the efficacy of political forecasting. Many prominent voices in the political scientific community have expressed skepticism that current political forecasting approaches are credible, and express concerns that political forecasts might be over emphasizing quantitative methodology used to derive otherwise easily predictable or banal results.

Additionally, several prominent figures in the history of economic and political studies have expressed skepticism that their insights have appropriate predictive power. Notably, Oskar Morgenstern, one of the founders of game theory, can be quoted as saying "Economic prognosis is...impossible for objective reasons," though his statement significantly precedes his publication of game theory's most foundational text, Theory of Games and Economic Behavior.

Data Gathering

Data gathering remains difficult in forecasting models. Some political forecasting models require insights of experts or individuals with insider information, which may represent highly sensitive data that, for security reasons, cannot be disclosed to the general public. Thus, some forecasting methods remain opaque. This problem may constitute information bias in some cases that provides barriers to entry to would-be forecasters who lack proprietary information or have relationships with well-informed parties. This also prevents the auditing of political forecasting results by third parties in a way that is public and transparent.

Representation

The field of political science, and thus the field of political forecasting, is overwhelmingly composed of Caucasian males. Groups such as the American Political Science Association have made efforts to make the field more inclusive by issuing reports on the potential missed opportunities stemming from a lack of diversity.

Notable Contributions to Political Forecasting

The practice of forecasting events in political science remains a disjoint field with numerous paradigms existing conterminously that make consensus on the best forecasting methods difficult to achieve. Nonetheless, several academics have been praised for their contributions to the study of different outcomes in conflict that have advanced methods in forecasting.

Dr. James D. Fearon is the most highly cited political scientist in international relations. His work in developing models that explain war as the outcome of commitment failures and his work on the impact of commitment problems on the longevity of civil wars has contributed to the field of political forecasting.

Dr. Robert Powell's work on inefficiency criteria contends that negotiations amongst actors may break down when the capabilities of actors have high intertemporal variance. Powell hypothesizes that temporary fluctuations in capabilities provide temporarily advantaged actors the ability to secure concessions that are unachievable when their rivals are strong. His work has had profound impact in the study of revolutions, political instability, and has become integral in projections of future events following exogenous shocks.

Related Studies

Choice modelling

Further information: choice modelling

Choice modelling is the study of choices made by actors from a discrete set of choices.

Rational Choice Theory

Further information: Rational choice theory

Rational choice theory is the study of outcomes of interactions of players that are assumed to be rational. Elementary rational choice model the expected utility of

Notable Political Forecasters

• Bruce Bueno de Mesquita, New York University
• James Fearon, Stanford University
• Philip A. Schrodt, Parus Analytical Systems

See also

• Bruce Bueno de Mesquita
• Choice modelling
• List of political forecasters (I will also author this)
• List of political scientists
• Political Instability Task Force
• Political Science
• Punditry
• Rational choice theory
• War's inefficiency puzzle

External links

• www.example.com

This article "Political forecasting" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Political forecasting. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.

This page exists already on Wikipedia.

---