Abstract: The dominant indicators used to analyze battlefield progress in counterinsurgency warfare – enemy-initiated attacks, civilian casualties, public perceptions – fail to capture the core dynamic of the fight: the contest for the support of a decisive majority of the population. This disconnect has led to erroneous assessments and formed the basis for ill-informed decisions. This article builds on recent work to apply rational-expectation theory to close this gap. It argues for the need to shift the focus of assessments away from the stated perceptions of the populace towards their actual behavior to better capture the complex developments on the ground. The population’s behavior reflects their actual perception of the relative strength and staying power of the host state and the insurgency. The main contribution of this article is a “horizon framework” which links behavioral indicators with varying expectation horizons and leverages this to gain a greater awareness of the true perception of the population. The framework can be used by analysts to untangle the complex reality of a population that can have positive expectations to the near future and at the same time harbor negative expectations to a more distant future (and vice versa). The framework provides a basis for selecting indicators, assessing developments, and strengthening the basis for timely action in counterinsurgency warfare.
The indicators employed to assess developments in Afghanistan have been detached from a core rationale of counterinsurgency strategy: the contest for the support of the population. How the public perceives the fight matters to its eventual outcome. But the indicators used analyze counterinsurgency progress have largely failed to capture this dynamic. The error of the Vietnam War of recreating the chaos of the battlefield in the numbers has not been duplicated in Afghanistan, but the indicators have been unable to fully capture whether the desired goal of the counterinsurgency effort was being advanced. This article addresses this disconnect by showing how different behavior-based indicators can be selected, structured and analyzed to allow for a more nuanced understanding of the true perceptions of the population. We argue that behavioral indicators should be used to complement existing indicators used to assess counterinsurgency campaigns.
The underlying rationale of a counterinsurgency campaign is to ensure that a decisive majority of the population will come to support the state and break with the insurgency or at least not actively support it. This requires, as the theory goes, that the state is perceived as the most reliable provider of basic public goods such as security, essential services and possibly some level of political inclusion. But too often the indicators employed to assess counterinsurgency developments were detached from this core rationale. This disconnect has grave implications as it leaves analysts and practioners with little – or even worse: false – information about the core of the counterinsurgency conflict and how the campaign affects the outcome.
The indicators applied to assess the counterinsurgency effort in Afghanistan have exhibited a strong focus on security incidents, enemy-initiated attacks, enemy losses, civilian casualties, and territorial control. In this light, indeed, the indicators legacy of Vietnam lives on in the Hindu Kush. Such indicators provide insight into some parts of the conflict, such as the operational proficiency and tactics of the insurgency. This information is important in itself. But it reveals little about the political core of the counterinsurgency battle: the actions and perceptions of the population. To this aim, coalition countries have employed indicators such as trends in public perceptions of security, public perceptions of state actors, public perceptions of threats and key problems, and so forth. These indicators bring civilian and military officials closer to gauging the prevailing winds of the political contest between insurgents and counterinsurgents for the support and control of the population. But the indicators are limited by three important shortcomings.
First, there is a gap between attitudes and behavior. This is old news to economic and political behaviorist theory, but it is a prevalent fallacy in much counterinsurgency theory and practice. A significant trend of positive perceptions of the state does not necessarily translate into strong, active support. Second, it is difficult to assess whether, or under what circumstances, a shift or hardening of attitudes will lead to a shift in actual behavior. Perception-centered indicators provide little or no insight into the underlying mechanisms that bind attitudes to behavior. Third, if perception indicators show a positive trend of increasing active support for the state, how can officials discern the robustness or strength of this development? It is difficult to confidently identify and delineate short-term spikes from fragile trends and decisive waves.
Ethan Kapstein argues that rational expectation theory can be used to resolve the disconnect between the core of the fight and the indicators used to assess it. He stresses the need to develop indicators to observe the behavior of the population as reflected in their decision-making. Measured actions – rather than stated beliefs – reveal real attitudes towards current and future developments. Kapstein proposes a focus on “the economic choices that people make during periods of intense conflict” to improve the basis for linking attitudes to behavior. If the public is optimistic about the future, then this should manifest itself in their behavior.
For instance, one can examine trends in the number of registered small businesses as a behavioral indicator of public rational-expectations of the future. Starting a business entails sunk costs and investments with break-even points that may come several years down the road. A rise in the number of business should, therefore, reflect that the rational expectations of the population are increasingly optimistic about the future as they foresee a development which is increasingly conducive to economic growth. Conversely, capital flight and a decline in registered business indicate declining levels of confidence in future developments.
Resolving the Disconnect: Bringing in the Temporal Dimension
A rational expectations perspective holds great potential for identifying valuable indicators, as does the emphasis on behavior rather than attitudes or perceptions as indicators of outcome. However, to gain a deeper understanding of the progression of the counterinsurgency campaign, we argue that analysts must distinguish between the time perspectives that are addressed by the indicators. One must appreciate that positive expectations of the near future do not necessarily mirror positive expectations of a more distant future (and vice versa). Otherwise analysts run the risk of judging the long-term progression on the basis of populations’ expectations to the near future. In other words, a temporal dimension is necessary.
We propose a horizon framework for measuring counterinsurgency progress based on behavioral indicators with varying expectation-horizons. It is structured to better capture the robustness of the behavioral shift that is necessary for a counterinsurgency success: a decisive majority of the population supports the state thereby draining the support base of the insurgency and marginalizing the remaining active supporters of the insurgency. The robustness of the support has an important temporal element and this is what the horizon framework is modeled to capture. The temporal dimension strengthens basis for selecting and analyzing behavioral indicators and enables assessments of the robustness of the observed patterns and trends.
The following table provides an example for its application:
The Horizon Framework
The first column states the behavioral indicator. Examples of each indicator are included for illustrative purposes. The next column categorizes the time horizon related to the rational expectations that are captured by the indicator. For instance, sending your children to school reflects an expectation that the school will not be attacked today or in the near future. It does not capture the rational expectations of the specific population for the next year or onwards. Recruitment for public services is listed as a medium-term indicator, as it is less sensitive to day-to-day events but reflects rational expectations beyond the immediate future. The decision to seek a job as a civil servant or enlist in the military is, however, more consequential and “costly” than going to the bazaar or sending ones children to school. It reflects the active decision to choose sides in a conflict where the state is a party. This entails considerable risk and hence reflects the rational expectations toward at least the next six months. However, enlisting as an army cadet does not, unlike registering a business, entail long-term investments or sunk costs that require a long period to recover. Accordingly, for reasons mentioned above, the number of registered business is included as an example of a long-term indicator.
The third column notes whether the indicator primarily indicates a positive or negative outcome or can indicate trends in either direction. Some indicators are strong indicators of counterinsurgency regression (-), while others are stronger indicators of progress (+). For instance, as mentioned above, while an increase in registered businesses can probably be assumed to indicate a rise in the population’s positive expectations for the future – and hence counterinsurgency progress – a constant level or decline in registered business may be due to several other obstacles such as drought, a shift in world prices, or general economic decline. The number of registered business is therefore primarily an indicator of counterinsurgency progress. Inducing counterinsurgency regression from a drop or constant level of registered business should be done with great caution and corroborating evidence. In contrast, school attendance is listed as a binary indicator as there is expected to be a direct, binary link between near-term trends in counterinsurgency outcome and school attendance. Parents would only send their children to school if they do not fear for their safety; merchants would not travel to the local bazaar if they have reason to fear suicide attacks and so on.
The forth column marks whether the specific indicator can be applied to special groups to gain further insight into the counterinsurgency outcome. For instance, since the Afghan insurgency is overwhelmingly Pashtun-based, a rise in the number of Pashtuns who are recruited to public civilian and military service in a given area would suggest that the Pashtuns expect the state to prevail in the counterinsurgency (while an increase in non-Pashtun recruits can just as likely reflect aggressive hedging against a Pashtunized insurgency).
Each shade of grey signifies the ascending levels of robustness of the outcome-tendency. The levels of “robustness” correspond to the ascending stages of time horizons. This allows the framework to capture the complexity of co-existing tendencies in the development of a counterinsurgency conflict: positive and negative developments are not mutually exclusive, but can co-exist and manifest themselves as optimistic and pessimistic rational expectations toward different future horizons. For instance, a new and fragile near-term optimism can co-exist alongside a rooted long-term pessimism, reflecting a public reaction to a short-term positive spike or change, which has yet to consolidate into a more robust trend or a decisive shift.
The framework allows analysts to capture this gradual spill-over-effect as positive trends are consolidated. The initial signs of fragile counterinsurgency progress are reflected in positive NT-scores. Then, if the tendency hardens and becomes gradually more reliable, public confidence grows accordingly, leading to a positive shift in MT-scores. Finally, if the trend continues to advance in strength and momentum so as to mark a robust trend or decisive shift, this would also manifest itself in positive LT-indicators. Conversely, analysts may also reach the conclusions that, while indicators do suggest some counterinsurgency progress, manifest in weak, but positive NT-scores, the overall trend is still so mixed and fragile that the medium- and long-term behavioral indicators portray a population still unconvinced that it is safe to bet on the state as the guarantor of order in the future. No indicator can stand alone as a single indicator of near-, medium or long-term rational expectations. Given the in-theater complexities and the often poor quality of battlefield data, the singular explanatory power of each indicator is too weak, too exposed to spurious relationships.
The Merit of the Horizon Framework
The horizon framework makes four contributions to theorists, analysts and practitioners of counterinsurgency warfare. The principal contribution lies in the framework’s ability to assist in untangling the wicked problem of measuring the intricate and oblique relationships and dynamics of trends in counterinsurgency warfare. The framework is designed to capture the behavioral shift vital to counterinsurgency success: that the population actively supports the government because it is viewed as increasingly effective and legitimate. A context-specific and systematic operationalization of the framework to identify, apply, and analyze indicators will lead to much more valid assessments of counterinsurgency progress or decline.
Second, the framework provides a theoretically informed basis for selecting indicators, and collecting and prioritizing data. The sheer wealth of data on counterinsurgency conflicts, and the innumerous indicators applied, necessitates a systematic method of filtering the information to enhance accuracy. The framework provides this by privileging behavioral indicators, emphasizing the need to differentiate between time horizons, and enforcing a focus on capturing the behavioral shift among the population necessary for counterinsurgency success as the overriding principle of choosing indicators and collecting data.
Third, the framework provides a lens to interpret developments and assess the robustness of a trend. It conceptualizes the link between public behavior and the significance or robustness of a trend. Consequently, it is better placed to capture the spill-over effects between rational expectations towards different time horizons that occurs as the population change their behavioral dispositions in response the gradual hardening of a fragile trend into a clear or even robust trend.
Fourth and finally, the horizon framework provides a basis for assessing and calibrating courses of action, based on a more accurate comprehension of public attitudes within a given area of operation. This would provide critical input into the planning and execution of counterinsurgency, as its effectiveness is acutely dependent upon situational awareness and understanding the dynamics of the human terrain.
 Christian Bayer Tygesen and Kristian Knus Larsen are both PhD Fellows at the Department of Political Science, Copenhagen University.
 Kapstein, Ethan B. (2012): “Measuring Progress in Modern Warfare”, Survival, Vol. 54, No. 1, pp. 144-148; Upshur, William P., Jonathan W. Roginski & David J. Kilcullen (2012): “Recognizing Systems in Afghanistan”, PRISM, Vol. 3, No. 3, pp. 87; Gos, Eric, Jonathan Schroden, Ryan McAlinden, Marcus Gaul & Bret Van Poppel (2012): ”Challenges of measuring progress in Afghanistan using violence trends: the effects of aggregation, military operations, seasonality, weather, and other causal factors”, Defence and Security Analysis, Vol. 28, No. 2, p. 110; Campbell, Jason, Michael E. O’Hanlon & Jeremy Shapiro (2009): “How to Measure the War”, Policy Review, No. 157; Weitz, Richard (2009): “Afghanistan: War by Metrics?”, Central Asia-Caucasus Institute, http://www.cacianalyst.org/?q=node/5206, accessed: 7-4-2013; Kilcullen, David J. (2009): “Measuring Progress in Afghanistan”, The United States Army http://hts.army.mil/Documents/Measuring%20Progress%20Afghanistan%20(2).pdf, accessed: 7-4-2013;
 Connable, Ben (2012): Embracing the Fog of War, RAND Corporation, Santa Monica p. 149; Daddis, Gregory A. (2011): No Sure Victory, Oxford University Press, New York, p. 223
 Daddis, 2011
 Kapstein (2012) and Upshur et al (2012) have, among others, also made the case for behavioral indicators chosen on the basis of an explicit theory. See also: Kilcullen, David J. (2006): “Twenty-Eight Articles: Fundamentals of Company-Level Counterinsurgency”, Military Review, May-June, pp. 103-109
 Kapstein, 2012
 Kapstein 2012: 148-152
 Various experts with field experience in counterinsurgency list these indicators as indicators of near term change in counterinsurgency outcome, but these indicators have yet to be included in a structured and theory-based framework delineating different expectation horizons and ascending levels of trend strength. For instance, see Kilcullen 2006 and Upshur 2012
 Rittel, Horst W. & Melvin M. Webber (1973): “Dilemmas in a General Theory of Planning”, Policy Sciences, Vol. 4, pp. 155-169