In 2001 security agencies in the United States proposed using prediction markets to assess the likelihood of a terrorist attack. That government program was cancelled in 2003 due to criticism from US Congress. Put forward a case for or against the use of markets to predict the likelihood of this type of event.

Prediction markets are a platform for participants to trade in futures contracts where the payoff is dependent on an event rather than a value. This essay explores how neoclassical and behavioural economics support the use of this tool as a better predictor than traditional methods of polls and surveys. Using these markets to predict terrorist attacks in theory would retrieve information more easily from disparate sources (Wolfers and Zitzewitz, 2004)¹: both within security agencies and from the general public. Different design and participants created the moral argument against these markets. I will explore how they can be controlled and beneficial in contributing to anticipating terrorist attacks, especially cyber-terrorism.

As technology and social media is now such a large part of western society there is another method of security threat through the huge amount of data available. Although they monitor for key words, prediction markets could be another means of picking out vital information from the public domain. A lot of what is available could however be opinion as opposed to informed knowledge (Tabarrok, 2002)², so a perfect market with traders looking to make profit would ‘de-bias’ this information (Berg and Gruca, 2007)³. The public information may be a good contributor for topics that security agencies do not have in depth knowledge of or it may just inform what the public is more interested in (Yeh, 2006)⁴.

A market needs uninformed participants to form trades. If the market was formed, for example, of security officers all from the same department then they would have common posterior beliefs, not disagree and not want to take alternate sides of a trade (Aumann, 1976)⁵. It is therefore necessary for the market to at least be across teams with different insider information. Yet these may still have the same beliefs, so including the public could be a way to ensure a market is diversified with those that are overconfident and speculating and those using it for entertainment purposes. In the latter case, terrorism is a topic that will attract a lot of interest from those looking to trade on new instruments (Wolfers and Zitzeweitz, 2006)⁶. However, the US proposal for prediction markets to test policy ideas and for geopolitical risks would not have attracted much interest (Wolfers and Zitzewitz, 2004)¹. Therefore, a subsidy may have been used to acquire participants. This generated public concern of people profiting in a market (Hanson and Oprea, 2007)⁷ which has negative moral implications.

Making a trade on a prediction market is with regards to a specific event that may occur, but with cyber-terrorism there are thousands more possible scenarios. Thus, it is a big challenge to speculate and form clear and simple contracts to trade. However, it can be argued that markets will do this more quickly and efficiently than the systems security units already have. Also, once the platform and contracts are established they can be replicated easily for many scenarios. A first use of prediction markets around cyber-terrorism could be looking at the likelihood of a defence agency being successfully penetrated (Yeh, 2006)⁴. This is a narrower and larger event than the thousands of other possibilities across institutions, locations and methods. There may be more knowledge on this already too because of the highly sophisticated systems which need to be implemented and controlled, as well as the interconnectivity of security and defence.

The Efficient Market⁸ and the Hayek hypotheses⁹ are the main theories behind prediction markets aggregating information in a sustainable manner. These tools allow the model to accurately price knowledge from different individuals, as well as on the environment. From these come the random walk theory. This is a caveat to the use of prediction markets, stating that: because information is reflected instantaneously it cannot predict the future movements, rather tomorrows price will simply reflect tomorrows unpredictable news (Yeh, 2006)⁴.  Despite this, prediction markets that are already in use have a high success rate. This is possibly because movements in prices reveal events occurring, (Yeh, 2006)⁴ even if they have no contribution as a predictor for future movements. Despite the confidence intervals from random walk projections increasing over a longer time horizon, they can be argued to be more accurate than the margins of error from a poll (Berg, Nelson, and Rietz, 2003)¹⁰. Therefore, prediction markets are still beneficial in measuring the degree of (un)certainty.

There aren’t many experts on computer programs and hacking so the prediction market created around cyber-terrorism would be thinner than those which back up the Efficient Capital Market hypothesis (Grady and Parisi, 2006)¹¹. In this case, there would be less speculation and more price volatility because the number of buyers for contracts and price they are looking for does not match the (number of) sellers (Hanson and Oprea, 2007)⁷. It is important for security agencies to understand if bubbles will occur in these prediction markets on terrorism contracts because it could damage their forecasts. As with asset markets, they should be equipped to decide whether it is better to burst the bubble or respond after to limit negative effects on the structure of the market (Yeh, 2006)⁴.

Any attempt to increase market participation to limit volatility as well as acquire more information should be done so with discretion due to the argument that potentially terrorists could enter, influence and even profit from the market changes. They could manipulate what the markets are formed from (coding in websites or software programs for example), thus changing prices reflected in the markets or simply by speculating in the market (Wolfers and Zitzeweitz, 2006)⁶. However, studies around this have shown that a trader whose target is uncertain and is trying to manipulate the market will not be successful in doing so because this extra noise (biases discussed earlier) gives more opportunity for profit making (Hanson and Oprea, 2007)⁷. This will bring prices to their correct equilibrium. In other words, trade reveals objectives of participants that are trading against or changing the market which in the case for terrorism can be supporting evidence. There is a trade-off between allowing the public to participate for added insight then recognising when a market manipulation is of any significance (i.e. how much the share price needs to deviate by before being an indicator of terrorist activity) and limiting participation which could just make the market more complicated with unknown benefits in prediction. It can be argued however that limiting participation is also not worthwhile because if it was terrorist’s intention to fund operations through the market they would manipulate a related market. This may have been seen pre-9/11 with unusual trading in airline stock but it is unclear whether adverse profits were made from this (Wolfers and Zitzewitz, 2004)¹.

Behavioural economics also introduces biases which could result in prices being reflected incorrectly in prediction markets. They are based on contracts on future events so, prices which arise are typically interpreted as probabilities an event will occur, which has been discussed by Manski (2006)¹²to be unsound reasoning because there would need to be more insight into why the trade has happened (Yeh, 2006)⁴. Nonetheless, according to Kahneman and Tversky’s Prospect theory (1979)¹³ these probabilities will be over-weighted when calculating rare vents such as terrorism (Wolfers and Zitzeweitz, 2006)⁶. Yet, cyber-terrorism is becoming much more frequent so this argument may reverse soon and studies show this has little influence on the reliability of these instruments anyway (Wolfers and Zitzewitz, 2005)¹⁴.

Similarly to how many of the public think it is unethical that bankers place large bets on financial markets, it has also been shown they view prediction markets as a novel idea (Hanson and Oprea, 2007)⁷. They may think it is a waste of taxpayer’s money (Yeh, 2006)⁴ to create similar models. Increased education for the public on how markets work will make the potential process less obscure. One view of prediction markets for terrorism is that they are putting values on people’s lives and a profit can be made, even by the attackers, which is immoral. Understanding of how to control the use and misuse should also reduce opposition to prediction markets. If a profit is made but results in detection of the threat (Wolfers and Zitzewitz, 2004)¹ and lives being saved this will be worthwhile. It should also be questioned whether it differs from the wide spread use of life insurance. In fact, the UK has a market for insurance against cyber-terrorism which is effectively pricing the likelihood, very similar to a prediction market.

The possibility and scale of potential cyber-attacks is a relatively new and obscure concept compared to the visible consequences of a physical terrorist attack. An awareness of the benefits of an extra resource for security in this new area could reduce opposition to prediction markets. Perhaps trialling these markets with a different name to see the participation and success rates could be a good start for introducing prediction markets but this is ethically dubious.

In conclusion, I support the use of prediction markets as another security measure. There needs to be a lot more persuasion of the value of results as they are relatively new tool generally. Then they may be accepted in the security industry and senior officers will aid their effectiveness (Yeh, 2006)⁴. With gaps in our abilities as humans to process data (Tabarrok, 2002)², prediction markets supplement solutions such as surveys and consultants. They could be extremely useful in aggregating and updating information over the long term (Yeh, 2006)⁴ where this work is cumbersome and time consuming. This will allow expert judgement to be put to better use on deeper analysis.

Bibliography

1.  Wolfers, J., and Zitzewitz, E. 2004. “Prediction Markets” Journal of Economic Perspectives. Volume 18, Number 2. Pages 107-126. https://www.aeaweb.org/articles?id=10.1257/0895330041371321
2. Tabarrok, Alexander. 2002. “Entrepreneurial Economics: Bright Ideas from the Dismal Science” Oxford University Press. https://books.google.co.uk/books?hl=en&lr=&id=5QFlWv5GSDUC&oi=fnd&pg=PA79&dq=Hanson,+Robin+decion+markets&ots=SFG9JvRjrS&sig=MvlWBbqovnUV74cGSLsQiJhIX88#v=onepage&q=Hanson%2C%20Robin%20decion%20markets&f=false
3. Berg, Joyce and Gruca, Thomas. 2007. “Public Information Bias and Prediction Market Accuracy” The Journal of Prediction Markets. 1 3. 219-231. http://www.bjll.org/index.php/jpm/article/viewFile/430/462
4. Yeh, P.F. 2006. “Using Prediction Markets to Enhance US Intelligence Capabilities” Studies in Intelligence. Volume 50, Number 4. https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/csi-studies/studies/vol50no4/using-prediction-markets-to-enhance-us-intelligence-capabilities.html
5. Aumann, Robert J. 1976. “Agreeing to Disagree”. The Annuals of Statistics. Vol 4, No 6. Pp 1236-1239 http://www.dklevine.com/archive/refs4512.pdf
6. Wolfers, Justin and Zitzeweitz, Eric. 2006. “Five Open Questions About Prediction Markets” NBER Working Paper 12060. http://www.nber.org/papers/w12060.pdf
7. Hanson, Robin and Oprea, Ryan. 2007. “A Manipulator Can Aid Prediction Market Accuracy” http://mason.gmu.edu/~rhanson/biashelp.pdf
8. Malkiel, Burton. 2003. “The Efficient Market Hypothesis and Its Critics” Journal of Economic Perspectives. Volume 17, Number 1. Pages 59-82 https://eml.berkeley.edu/~craine/EconH195/Fall_14/webpage/Malkiel_Efficient%20Mkts.pdf
9. Hayek, F. A. 1945. “The Use of Knowledge in Society” The American Economic Review. Vol 35, No 4, pp 519-530 http://www.kysq.org/docs/Hayek_45.pdf
10.  Berg, Joyce. Nelson, Forrest and Rietz, Thomas. 2003. “Accuracy and Forecast Standard Error of Prediction Markets” University of Iowa, Working Draft.
11. http://repository.binus.ac.id/2009-1/content/D0264/D026473379.pdf
12.  Grady, Mark, and Parisi, Francesco. 2006. “The Law and Economics of Cybersecurity” Cambridge University Press.
13. https://books.google.co.uk/books?id=_m_lFy1X-BYC&pg=PA301&lpg=PA301&dq=efficient+markets+hypothesis+cyber+terrorism&source=bl&ots=WeMniuWZt5&sig=tBkpUVPyjOJbQ5vhicZcWscmIZg&hl=en&sa=X&ved=0ahUKEwiv_pX6rb7ZAhWhF8AKHfv-DvQQ6AEILzAB#v=onepage&q&f=false
14. Manski, Charles. 2006. “Interpreting the predictions of prediction markets” Economic Letters. Volume 91 Issue 3. Pages 425-429. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.159.7574&rep=rep1&type=pdf
15.  Kahneman, Daniel and Tversky, Amos. 1979. “Prospect Theory: An Analysis of Decision under Risk”.
16.  Econometrica, 47(2), pp263-291 https://www.princeton.edu/~kahneman/docs/Publications/prospect_theory.pdf
17.  Wolfers, Justin and Zitzewitz, Eric. 2005. “Interpreting Prediction Market Prices as Probabilities” Discussion Paper Series IZA DP No. 2092. http://ftp.iza.org/dp2092.pdf
18.  Rhode, Paul and Strumpf, Koleman. 2004 “Historical Presidential Betting Markets” Journal of Economic Perspectives. Volume 18, Number 2. Pages 127-142. https://pubs.aeaweb.org/doi/pdf/10.1257/0895330041371277