Search interesting materials

Wednesday, October 12, 2016

How will your armed forces perform?

by Ajay Shah.


In the world of public policy, there are two dimensions to thinking about performance.

How hard is a problem? "High load problems" are those where there is more discretion, a large number of transactions, and the stakes are high.

Are there easy mechanisms for accountability? Performance is greater when measurement of outcomes is more feasible. As an example, it's relatively easy to get performance from a central bank which is tasked with delivering a 4% inflation target and nothing else, as success/failure are highly visible. Performance is measured every month. If the central bank performs poorly, within a few years, this becomes visible, and that would kick off remedial actions.

Feedback into superficial change or deeper change? When institutions are weak, a government agency is personified into a few individuals at the top. When failure is visible, the response is to seek staffing changes. As an example, the 1962 war led to the sacking of V. K. Krishna Menon; the scandal of 1991/1992 led to the sacking of S. Venkitaramanan. These were relatively superficial changes. For a display of better institutional capability, the Indian inflation crisis of 2006-2013 helped trigger off fundamental reform of RBI.

Rare events and feedback loops

The best feedback loops are those where measurement is frequent, and reports of low performance kick off small steps that address the problem. When measurement is infrequent, the feedback loop is much less effective.

A small number of actions, and thus a weak feedback loop, is something that we've seen in another domain: consumer finance.  In the field of consumer protection in finance, there has long been a sense about why consumers do badly when buying certain financial products like home loans or pensions: the transactions are so infrequent. When a consumer engages with toothpaste, there are many transactions through life, and a feedback loop is in place through which performance is improved. In contrast, many consumers buy only one or two home loans or pension products in their life. There isn't enough experience with small course corrections through which the feedback loops kick in. This leads to big mistakes by consumers. While this is an example about consumers and finance, the principle is general: infrequent actions hamper feedback loops.

The puzzle of building the armed forces

The armed forces pose a difficult challenge for public administration as it's hard to know whether they are working properly. Using vast public resources, an organisation is built for the purpose of waging war. In peace time, it's hard to know whether the organisation has the advertised capabilities. Wars often have a decisive win/lose outcome, but wars are infrequent, and there also, there are extraneous factors that limit accountability.

With the armed forces, we face the Principal-Agent problem where the Principal (the political and civilian authorities) want to get a target level of capability at the minimum cost, and the Agent wants to pursue self-interest. In this Principal-Agent problem:

  1. Building the armed forces in peace, or using them in war, is a high load problem as it involves a lot of discretion, high stakes. It takes a lot of capability in the political and policy system to pull this off.
  2. The checks-and-balances of democracy are weak given the opacity that envelops the armed forces.
  3. The Principal (the civilian political leadership) knows little about the performance of the Agent (the armed forces). In peace time, years and years go by with no feedback loop about performance. For example, Germany knows little about the capabilities of their armed forces as they have been tested so little in recent decades. By this yardstick, the US fares best as there are frequent wars which generate data about performance.
  4. For countries where nationalist feelings are prevalent, there is a greater risk of civilian authorities glorifying the armed forces and thus failing to employ public choice reasoning. By this yardstick, Germany fares better than the US.

In advanced countries, there is a considerable extent of the rule of law, and review of the armed forces, in times of peace and war. For example, see the 19 occurrences of the word `lawyer' in Ghost Wars by Steve Coll, and the long history of the political leadership sacking military leaders. But this is complex institutional machinery which is hard to build. A recurring feature of underdevelopment is that civilian authorities have limited power over the top leadership of the armed forces. When the generals claim excellence, it is hard for civilian authorities to be intrusive, impose management and staffing changes, etc. Weak oversight by the Principal is then the breeding ground for failure by the Agent. Claims of the greatness of the Agent increasingly diffuse through society, unchallenged.

Mistakes by the Principal

War is an expensive way to resolve a dispute; it seems like a shame that ordinary processes of bargaining are not able to avert more wars. Why do wars take place as much as they do?

Nobody would go to war if they did not expect to win. Yet, half the time, one party has gone to war with the mistaken notion that he was going to win. Why does the Principal make this mistake?

Some ingredients of this excessive willingness to mistakenly wage war may be as follows. In each country, the Agent claims I'm doing a great job of using your money, we're very capable, we're ready for a war, we will win it. Exaggerated respect for capabilities of the armed forces would arise when these claims are not challenged widely in society. There is a toxic interplay between places with high nationalism (which would tend to glorify the armed forces) and low institutional capability (which would tend to leave the Agent free to do as he likes). Under these conditions, the Principal may be prone to inefficiently continue politics by other means.


  1. The armed forces are one of the hardest problems in public administration: high discretion, high stakes, high opacity. When wars are infrequent, the civilian authorities (the Principal) know little about the capabilities of their Agent. When a country suffers from nationalism, this further damages the ability of the Principal to coldly analyse and reshape the Agent.
  2. Overconfidence bias would make the Principal over-estimate how well he is handling the Agent. Nationalism is likely to delude the Principal too. Hence, civilian leaderships may have a bias in favour of over-estimating the capabilities of their armed forces.
  3. Fewer countries would go to war if they had a more accurate assessment about the incompetence of their armed forces.

I thank Shekhar Hari Kumar, Ila Patnaik, Kaushik Krishnan, Renuka Sane and Susan Thomas for useful discussions on this.


  1. The war in practical sense are not so infrequent. To say that they are infrequent is to think of war in in its traditional definition where two sovereigns fight with each other openly. Atleast in India or some other countries, internal violence (may be extrenally sponsored) is nothing less than an event like war.

    Hence, if the issue is to asses the benefits of a very high investment (i.e. maintaing an army) with negligible performance events (i.e. traditional war), the solution is to redefine the perfomance event itself. This would be actually redefining the definition of the army. Hence the army is not only who fight with an another country but also take care of high impact internal violence.

    Obviously, this creates a sort of overlapping between army and "paramilitary forces". We have a big infrastructure of paramilitary forces. The common argument for creation of paramilitary forces is so called insensitivity or lack of expertise of army in dealing with issues related to civilians or which are political in nature.

    However, the above argument (i.e. insensitivity) is not a very a fundamental reason but seems more like an administrative or political issue. This issue can be avoided or made part of army training.

  2. A nice point!

    Wars are infrequent - think India/Pakistan 1965, 1971, 1999. Or India/China 1962. Or IPKF in Sri Lanka. These are rare events.

    But internal conflict in India is happening a lot. These are frequent events. This would imply that there *is* accountability and a feedback loop which is refining para-military organisations. Sadly, these skills are profoundly different from those required in fighting external wars, so having adequate positive feedback on one does not assure capabilities on the other.

  3. The pricipal agent construct with political/civilian as a principal appears to be problematic in my view.

    1) Political/civilian idoes not appear to be a perfect Principal

    In a standard setting, the principal is able to define his utility function and he has an honest interest in his utility function. Atleast, he has no other motive than to be happy or satisfied if his utilituy is maximized. For example, the shareholders knows that profit has to be maximized and they will be dissatisfied otherwise.

    The politician/civilian pricipal need not satify these characteristics. Defence investment by a political regime may not necessarily be based upon anticipated long term genuine defense objective. In some cases, it may be based upon social or cultural grouping of its traditional vote bases. For example, a democrat regime in US may not have necessarily initiated Iraq war in 2004. Hence, even if there is an objective, it may not necessarily be genuinely defence related.

    A politician may not be even genuinely interested in knowing or propogating whether the agent has really performed (i.e won the war). A propogated Illusion of a victory is all what they may need. For example, almost all of us believe that we Won the 1965 war. However, the same is believed by Pakistani people also.

    2) A politician is predominantly an agent but not a principal:

    The politician most probably appears to be an agent of its voters or traditional vote group. Hence, his primary objective may be to convince its voters that their army or war related notions or objectives are fulfilled. Hence, different voter groups may lead to different types of principals and thereafter different army related objectives. Hence, the primary objective of a political regime may be to ensure satisfaction of its traditional vote group.

    As an example, a political regime dominated by extreme rightist will increase defence investment not because it anticipates genuine defence benefit but this is what it identify itself with

    1. I think this is an interesting way of looking at this problem. This is not a "principle-agent" problem but a "principle - agent(intermediary) - agent(final)" problem. The principle here is the public. The government is agent(intermediary) and Army is the agent(final). The Principle-Agent-Agent problem faced here is similar to the one faced in consumer finance. For example in mutual funds, retail investor is principle, the financial advisor is the agent(intermediary) and the mutual fund itself is the agent(final). The complications in principle-agent-agent problems are much severe.


Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticising me is perfectly okay; uncivilised language is not. I delete any comment which is spam, has personal attacks against anyone, or uses foul language. I delete any comment which does not contribute to the intellectual discussion about the blog article in question.

Please note: LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.