Category Archives: heuristics

India’s behavioral science policy unit – Challenges and wayforward

Two recent stories that appeared in Indian media suggest that the Indian Central (Federal) Government is looking to set up a behavioral sciences policy unit under the Niti Ayog, a Government of India policy think-tank established by the Narendra Modi government.

This news item that appeared in The Economic Times  suggests that the government has tied up with The Bill & Melinda Gates Foundation to set up the unit.

There have been several examples of Behavioral Insights units, starting with the one in the UK cabinet office. The Behavioral Insights Team is now partly owned by the cabinet office and calls itself a social purpose company. 

Niranjan Rajyadhyaksha of Mint had written this compelling piece on why the Modi government needs a Nudge unit. The Indian Prime Minister himself on occasions alluded to the behavioral nature of some of the problems, particularly sanitation.

Needless to say there are several advantages of such a unit. This well written editorial in Mint takes a more balanced view to such a unit. Incidentally, 4 of the problems outlined in the opening paragraph of the piece are problem areas Final Mile has experience using Behavioral Sciences.

The piece also points out some potential limitations of such a unit. There are areas where a nudge simply is not good enough, behavioral scientists themselves are not immune to bias and the fact that India is complex. We though believe that the complexity argument is over stretched. There is diversity in every country. Successive governments have been making policies  accounting for the complexity. Our experience in general has been that there are more similarities than there are differences. Dilip Soman, a well know behavioral economist suggests that “Complexity makes it more likely that soft interventions will work better than other options”.  A good next step  would be to recommend such a policy unit at the state level as well.

As pioneers in the field of applying behavioral science to solving real world problems, this is highly encouraging news. There are some challenges such a unit needs to navigate and, based on our experience, these are some of those. We understand that most of these if not all, would have been taken in to consideration by the decision makers.

  1. There is an inherent danger in assuming that a particular behavioral science principle is universally valid. There have been cases where using a principle blindly have backfired. There was a recent experiment where a company used social norms with a view to increase savings by it’s employees. It proved counter productive. In context testing is therefore key.
  2. One of the big crisis that hit the world of behavioral sciences and psychology is where many ‘successful’ experiments could not be replicated. This was particularly true of social priming. This paper co-authored by one of the senior employees at Final Mile has more detail  . There is a need to generate strong evidence before a policy or an intervention can be deployed. Rigorous testing is vital. As Richard Thaler, widely considered the father of behavioral economics says “You can’t have evidence based policy without evidence”
  3. Complex and wicked problems need a multi-disciplinary approach. A nudge unit team needs to bring in diverse skills. One that if filled with Behavioral Scientists may not be the best approach. In our experience, integrating design thinking with behavioral sciences can lead to powerful results. Equally important are measurement and evaluation experts
  4. Navigating through the government system and particularly the famed Indian bureaucracy is going to call for incredible amount of patience and tact.  
  5. Establishing value of such a unit is obviously critical. At a conceptual level, all this makes sense but government officials and ministers are keen on quick results. There are realities of electoral politics. A good approach would be take one or two areas and demonstrate value rather than trying to spread thin across ministries. Peter Kalil, Deputy Director for Technology and Innovation, Office of Science and Technology policy in the white house made some observations on this subject at the recently held Behavioral Science summit. It is far easier to take life sciences in to application. It’s tangible and we have experience and set systems. Taking behavioral science to people is not the same.  Framing results and writing for policy makers is quite different from writing an academic paper. And that working with existing programs is a much better way to overcome Status Quo bias. Launching new programs may not be the best way to go. 
  6. Libertarian Paternalism is a phrase that Prof. Richard Thaler and Case Sunstein coined. It might sound like an oxymoron, but it isn’t. In their own words “The idea of libertarian paternalism might seem to be an oxymoron, but it is both possible and legitimate for private and public institutions to affect behavior while also respecting freedom of choice. Often people’s preferences are ill-formed, and their choices will inevitably be influenced by default rules, framing effects, and starting points. In these circumstances, a form of paternalism cannot be avoided. Equipped with an understanding of behavioral findings of bounded rationality and bounded self-control, libertarian paternalists should attempt to steer people’s choices in welfare-promoting directions without eliminating freedom of choice.”  However, such a unit is likely to come under criticism from both the right and left of the political spectrum. The left would argue that you cannot call poverty a behavioral problem, the right might term this a “nanny state” initiative. These are extreme arguments but ones that have been made several times. Considering the possibility of sensationalism by the Indian media, such a unit needs to be prepared to effectively deal with criticism.

Ultimately, the success of this unit depends on government support and patience. The mandate needs to come from the highest level, like the White House Social and behavioral science team where President Obama issued an Executive order “that directs all Federal agencies to use insights from the behavioral sciences to make government programs easier to access, more user-friendly, and more effective” 

Obama also notes that “Adopting the insights of behavioral science will  help bring our government into the 21st century in a wide range of ways – from delivering services more efficiently and effectively; to accelerating transition to a clean energy economy; to helping workers find better jobs, gain access to educational opportunity, and lead longer, healthier lives”

The Indian unit could do with a similar endorsement from Prime Minister Modi.

To B, or not to B, that is the question


The ‘B’ in the title stands for Brexit- the portmanteau monicker given to the possibility of Britain’s departure from the European Union. This June, many British citizens will face their second referendum on secession in three years. Their bone of contention is the burdensome legislation from the European Council- seen by many Brits as an infringement of their gleefully miserable approach towards language, food and life in general. Since the seeds of European integration were sown, the island nation has experienced bouts of collective outrage- on Brussels deciding how bendy British bananas should be, or how sausages should be renamed as emulsified high-fat offal tubes. Some of these stories have turned out to be practical jokes planted by journalists or even pure satire- the sausage story was in fact an episode of Yes Minister. But the resulting outrage in every case has been real and present. It is with this long-term cynicism of continental appropriation, as well as coercion from extreme right-wing groups such as the UK Independence Party, that Prime Minister David Cameron decided to announce the referendum last month.

Pertinent to the referendum discussion is the syntax and construction of the question that will be posed before voters. Some may consider this a minor detail, but the Electoral Commission of the UK has conducted eleven detailed assessments on it since 2001 alone. Learnings from behavioural science justify the earnestness shown by the election watchdog, and evidence from the following historical referendums further corroborates this claim:

Scotland, 2014

The framing of the question for the 2014 referendum- where Scottish citizens decided whether or not to remain a part of the United Kingdom- became a contentious issue. The Scottish National Party, who were in support of their independence, had originally proposed this question:

Do you agree that Scotland should be an independent country?

  • Yes
  • No

The Electoral commission pointed out the leading nature of the question. Behavioural science calls this  technique of influencing responses as ‘framing’: the manner in which the question was constructed would encourage neutral voters to vote ‘yes’ rather than ‘no’. This also follows from the claim that people are more likely not to disagree with a notion that is presented in a somewhat coercive manner.

Earlier during the movement, those in support of the ‘let’s stick together’ campaign preferred this question instead:

Should Scotland leave the United Kingdom?

  • Yes
  • No

Here, the negative connotation of the word ‘leaving’ would make voters experience loss aversion- a state where the prospect of negative outcomes looms larger than that of positive outcomes. At the pivotal moment of a voter making their choice, loss aversion would likely bring to their minds, all the risks and tribulations were Scotland to exit the UK, thereby encouraging them to tick the ‘no’ box. Additionally, considering this nation-of-many-countries has been at peace for seven decades, secession exists more conspicuously in a negative frame than independence does in a positive one.

The Electoral Commission, which pointed out that the final question should be presented in clear and neutral terms, eventually had both sides agree on the following format:

Should Scotland be an independent country?

  • Yes
  • No

Although the majority voted against independence, the marginally positive framing of the final question may have impacted the overall numbers supporting independence- which at 45%- amounted to a far closer race than was expected.

Greece, 2015

The Greek Bailout Referendum was put before its citizens ten days after its announcement by Prime Minister Alexis Tsipras in June 2015. The poll asked Greeks if the government should accept the bailout conditions set by the European Commission and International Monetary Fund. A ‘yes’ would mean years of spartan hardship for citizens and businesses, but a ‘no’ would lead to a possible exit from the Eurozone.


On voting day, the question featured in the ballot was as convoluted as the Euro crisis itself: it included sixteen page appendices, and the ‘no’ option was placed above the ‘yes’. It may not come as a surprise that 61% of the voters rejected the IMF’s proposition, and the government received a handsome political payoff- domestically and internationally- for its snap poll.

Nazi Germany, 1938

Many would describe such industrial-strength nudging by governments as unethical and distasteful. They are likely to find the arrangement of ‘Yes’ and ‘No’ options of the 1938 German referendum (asking Germans if they wanted to reunite with Austria and ‘vote for the list of the Fuhrer Adolf Hitler’) as astonishing as the revelation that the Third Reich actually bothered themselves with such an extensive democratic exercise. The large circle isn’t merely a nudge, but also a hint.


Learning from history and from convoluted referendum forms in its own past, the Canadian parliament passed the ‘Clarity Act’ in 2000 that restricted questions on secession from being overtly complex. Since then, length of referendum questions has reduced from 36 words- in the case of Quebec- to the six words for Scotland. But constraints of brevity may bring with them nudges of their own- which leads us back to the referendum facing Britons this June. The Electoral Commission has already recommended a somewhat long-winded question. But it manages to nullify all the behavioural biases that would plague the standard ‘should the UK leave the EU’ or ‘should the UK stay with the EU’ formats:

Should the United Kingdom remain a member of the European Union, or leave the European Union?

  • Remain a member of the European Union
  • Leave the European Union

Whatever the results turn out to be, it will be heartening to note that sub-conscious behavioural science will not play a role in deciding Britain’s vote, and by extension the fate of the European Union. It is important for nudge units and behavioural sciences firms across the world to question the ethics behind nudges that tilt public opinion to any one side in a democratic debate- particularly where neither side has engaged in violent, illegal or commonly-accepted immoral behaviour. Renaming Chicken Tikka as ‘high-flavour skewered British roast’ would, obviously, be an example of all three.


Images Sources: Merkel-Cameron: Business Insider/Greek Bailout: Wikimedia/Germany Ballot: BBC

Articles Cited

Grafting Loss Aversion

It would be fair to call corruption the single most important determinant of political fortunes in recent India. The party that rode to power by defining corruption a matter of absolute morality has now begun to appreciate it as a nuanced economic problem. The decision of the Delhi government to increase the salaries of their state legislators by 400% in order to ensure their unwavering scrupulousness, has been under intense scrutiny. It begs the question: do increases in public functionaries’ salaries truly inhibit corruption?

A new study under the LSE’s International Growth Centre, mentioned in The Economist, attempts to prove that it does not. Two American economists examined the effects of such a change in compensation in Ghana, where under a new directive salaries of the police force were doubled. They studied the frequency of bribes offered by lorry drivers- a common source of illegal income for the police- over long-distance journeys to neighbouring countries. The results of econometric analysis showed that after the new law, the level of corruption among the highway police increased by several measures: more roadblocks were erected, more lorry drivers were stopped with increased frequency, and more money was extracted.

This claim goes against the conventional viewpoint supported by different economic models. A discussion on them is too nuanced to be started here, but empirical data (see graph) of economic indices (per capita income of countries against their corruption index score) shows the strong relationship between low income and corruption.

Screen Shot 2016-02-16 at 18.52.21

Concepts from behavioural science may explain the strong correlation. Increased pay creates a sense of fairness in the minds of employees, and they are hence more likely to be virtuous simply out of reciprocity. Moreover, with increased salaries, employees less susceptible to moral hazard are likely to enter the workforce. As this new cohort reaches critical mass, social proof dictates a change in behavior in even the most unscrupulous among the old guard.

Most importantly, the concept of loss aversion may demonstrate why incentive-led measures are more often than not an effective anti-corruption tool. Every act of corruption has a built-in risk component: the payee charges the payer for the risk of facing legal penalty or even sudden unemployment. The alternative for the payer is to follow the official but usually circuitous route. Abrupt increases in salaries cause employees to suddenly feel loss averse- they now face the chance of losing a better-paying government service.

A closer reading of the LSE study may strengthen this argument. Of all the lorry drivers stopped by policemen, 19% were allowed to pass through without paying a bribe, an increase from 10% before the increase in salary. This reduction may arise from an increased risk averse attitude, as the frequency of asking for bribes is directly proportional to the probability of legal action against the policeman. The real risk of action also increases because of more perfect information: well-informed drivers may not only bargain harder to escape paying an illegal fine, but also threaten the police with a complaint to superior officers.

RiskQuadrantThis shift in risk perception is well represented by Daniel Kahneman’s risk quadrants (see image). Prospect Theory shows us that in situations with lower probabilities of gain (such as in the lower quadrant) the agent is more likely to take greater risk. By increasing the legal component of the employees’ ‘total’ incomes, the government shifts the overall probability of them reaching a certain income target into the upper quadrant, with a high probability of gains. The agent thus becomes averse to minor losses- even in weak institutions where the risks of punishment are insignificant.

Economists- as is their wont- will attempt to box corruption into a predictive model. Robert Klitgaard had famously concluded C=M+D-A (corruption equals monopoly plus discretion minus accountability). It is nevertheless an issue with many causal components- several theories from policy, economics and ethics may prove why Ghana policemen defied the conventional view. It may be important to note that law enforcement in Ghana has historically been the least well-paid of all government departments- and the compensatory catching-up under the new directive may not have rid the police force of its reference bias as well as a historical feeling of unfairness. Additionally, in light of an enhancement in social and financial stature, pressures from superior non-field police officers for their ‘cut’, or from demands by family and friends, may also cause redistributive pressure and an increased graft appetite on part of the highway police.

Finally, in the absence of a concurrent advent of effective anti-corruption laws, their strict implementation and a substantial increase in the number of convictions- which was the case in Ghana- increased incentives will not have the desired effects in abetting corrupt behaviour. As is the case with several policies dealing with human behaviour, liberal, incentivising legislation must be complemented by unflinching tough love by the executive.


Images Source: Own work. Fair use.

Paris attack and the problem of categorization

Blog - Paris Attack

The Paris attacks have been a major shock and we are understandably upset. Most of us have been reminded of some of the earlier acts of terror such as the 26/11 attack in Mumbai or even 09/11 in New York. It also reminded us of the mixed emotions (fear, anger and despair) that we felt during the attacks.

 Since the earlier attacks were also targeted at civilians, we consider this act in the same vein. However, this Atlantic Article – What ISIS really wants suggests that, maybe, this attack is different. The article refers to the origin of the Islamic State from al-Qaeda, the formation of its leader al-Baghdadi, its strong belief of being a key agent of the apocalypse and many other qualities. But the most important point that it tries to make is that ISIS is not like al-Qaeda or any other organization that we tend to categorize as terrorists. The author suggests that we seem to have misunderstood jihadism as monolithic or of a single kind which it is not.

We can probably call this a problem of categorization or essentially mis-categorization.

Continue reading Paris attack and the problem of categorization