How honest are civil servants & what does this mean for S'pore systems?

Are civil servants really more honest than the average person?

Amid an ongoing investigation about a U.S. study which claimed to have identified a statistical pattern of Singapore civil servants purchasing homes near MRT stations before their locations were publicly announced, there have been conversations about whether the existing safeguards are enough.

The paper questioned if "pro-integrity social norms can substitute for formal enforcement in deterring misconduct by public officials".

In response, the government said it is reviewing the data of property purchases made between 2007 and 2011 by officers who were involved in rail planning, and has even extended the scope to 2014 as a matter of prudence.

The government has since updated the parliament that eight public officers whose property purchases warrant further examination have been referred to the police, after reviewing property transactions made by 191 officers.

As we wait for the results, it's worth considering: exactly what is the basis for these "social norms", and what does it mean for Singapore's systems?

Professor Leong Ching, NUSS professor of Behavioural Public Policy at the National University of Singapore, writes about her experiment on the matter.

By Leong Ching

On Oct. 6, the coordinating Minister for Public Services Chan Chun Seng announced that eight public service officers had been referred to the police for further investigation.

This came on the back of a working paper from the US-based National Bureau of Economic Research (NBER) alleging that civil servants here “disproportionately” bought private homes near MRT stations before those stations were announced.

The minister added that this did not mean that wrongdoing had been established; rather the police would review the facts and that due process would be followed.

Whatever the results may be, it has brought new salience to a larger issue: the issue of honesty in the Singapore civil service.

Assessing people’s honesty

In policy studies, the motivations of civil servants have long been the subject of study and theorising.

But empirical experiments testing honesty are notoriously difficult. Like the Higgs-Boson particle, honest behaviour is almost impossible to observe in an experimental setting.

Whether people are inherently honest is an old question. Much of our public life is built on an assumption of good faith — that we act honestly with one another and in our relations with the government.

Yet, from tax evasion to false declarations, people often fall short; in terms of policy design, audits and penalties are costly ways to detect and punish such behaviour.

In search of a low-cost and effective intervention, my colleagues and I at the National University of Singapore conducted a field experiment with 34,543 adults in nine countries across the Americas, Asia, and Europe.

The experiment

We designed three experimental conditions.

The first, “private cash”, got participants to privately roll a die and report the outcome, with a cash reward proportional to the reported value.

The rules were simple: if you roll a 1, you do not get any money, and if you roll a 6, you get S$5.

The incentive, therefore, is to report as high a number as possible.

If all participants in a large sample were perfectly honest, we would expect an average of 3.5. Anything higher indicates a degree of dishonesty.

The second, “private lottery”, was similar, but we introduced an element of chance. We wanted to see if people were more likely to be dishonest if their chances of getting the cash reward were determined by a virtual lottery.

The higher the reported value of the die roll, the higher the chance of winning. Again, the incentive is to report a high number.

The third, “public cash”, gave participants a choice to roll the die either privately or on a platform visible to the researchers. In this group, they could choose whether to be observed or not.

Importantly, they would still get paid out whatever amount they reported whether honestly or not.

The results

We found that most people were indeed dishonest.

Rolling entirely in private — the first experimental condition — average reports ranged from 3.76 in China to 4.43 in Mexico.

But most people were not maximally dishonest. Rather than thinking “in for a penny, in for a pound” and declaring a 6 when they rolled a 1, most shaded the truth a little.

They were partially dishonest — or, as they might tell themselves, partially honest.

Adding an element of luck — in the second condition — made little difference.

The most surprising result came from the third condition. I had expected most people not to opt into the condition to roll the dice on the NUS server where “you may be observed by the researcher”.

Instead, 85 per cent chose this option — and as a result were completely or almost completely honest, with averages ranging from 3.46 in Germany to 3.61 in Indonesia.

In the formal model, we worked out a coefficient which showed that people derived utility from being “seen to be honest”.

An intriguing finding.

Are public servants different?

For public servants, empirical evidence on honesty has largely been drawn from proxy samples — students who want to join the public service, police applicants and counter staff.

The results are mixed, i.e. we don’t know whether they are indeed more honest.

But our earlier, unrelated experiment provided a basis for answering the age-old question.

To find out, I ran the same experiment with nearly 3,000 public sector workers in the United States.

Rolling in private, public servants reported an average of 3.87, against 4.02 for Americans in the general population study.

When no one was watching, public servants were measurably more honest than everyone else (even if not perfectly so).

Given the option to be observed, 91 per cent opted in. Their average fell to 3.69.

At this number, the gap with the general public, who had been offered the same choice, disappeared. Even the few who declined to be observed were about as honest as those who opted in.

The offer itself seems to have been enough.

In short, civil servants do seem to be able to call on their better angels when no one is watching. Why?

Being seen to be honest

In the NBER paper, the authors argue that their data shows that social norms have limited persistence and are not enough to maintain a low-corruption equilibrium without formal enforcement.

My study does not compare enforcement with social norms, although it makes two interesting points about civil servants’ internal beliefs and their outward behaviour.

First, in the study, the more honest people believed their peers to be, the more honest they themselves were. They also consistently rated their colleagues as more honest than themselves.

Second, a common assumption about civil servants is that their internal values (or public service motivation) is related to higher standards of moral behaviour.

In this study, I did not find a connection — that is, an individual’s attraction to public service, commitment to public values, compassion, and self-sacrifice did not predict honesty.

Yet, most participants want to be perceived as honest and would behave honestly if given the chance.

A high bar

What would this insight look like in practice?

Well, for a start, like Section 8 of the Prevention of Corruption Act (PCA) 1960 in Singapore —  "Presumption of corruption in certain cases".

Murali Pillai SC, Senior Minister of State for Law and Transport, explained in a 2024 speech on anti-corruption how this use of presumptions worked in Singapore law.

He said: “Where a person employed by the government or public body receives gratification from someone who had dealings with the government or public body, both the giver and receiver are deemed to have given or received the gratification corruptly as an inducement or reward, unless it is proven otherwise.”

This is an extraordinarily high bar.

As the Singapore High Court explained in 2014, once a public servant is shown to have received a gift from someone dealing with the government, "the evidential burden of proof shifts to the recipient", who must offer an innocent explanation to rebut the presumption of corruption.

In short, to be seen as honest.

Designed for scrutiny

This is exactly what we are seeing play out in the PSD review and the police investigation.

Whatever the results from the PSD review therefore, the broader lesson is the same: integrity cannot simply be assumed from who public servants are (i.e. their public sector motivation, which may not relate to specific behaviours) nor from social norms (e.g. a belief in the honesty of others, which does co-relate to honest behaviours).

Instead, institutions and rules must be designed for scrutiny and the opportunity to be seen to be honest.

Second, and I think more significantly, there is the role of formal enforcement.

Empirically, formal sanctions have been shown to be effective deterrents, but even in the highly improbable instance that social norms can be shown to achieve the same behavioural outcomes, such enforcement must not be replaced.

Punishment is not mere quotidian behavioural intervention, it is a moral requirement underwritten by theories of justice and a contractarian understanding of public policy.

An incorrupt civil service does not require that no one be ever tempted, but instead, that those who give in to temptation be predictably punished, and honest officers given every opportunity to be seen to be so.

An opportunity in practice

Our research suggests that we can engage the better angels of our nature, if we design policies that allow people not just to behave honestly, but to be seen doing so.

Providing individuals with the option to be observed, therefore, offers a way to balance effective governance with respect for individual autonomy.

Civil servants today already have to declare their property purchases. A simple improvement would be for the government to offer civil servants the option to to opt-in for an automatic notification to be given to the PSD, any time they make property purchases or other significant investments.

It would not be difficult to link the Singapore Land Authority with the PSD. Those who do not opt-in will still have to comply with the rule, with a self-declaration, while those who do will save themselves the trouble of doing so.

This balance between realism and trust is sometimes thought to be determined by whether we think people are good or deceitful.

If like Hobbes, one thinks that life in the state of nature is “solitary, poor, nasty, brutish, and short” then there would be more demands for strong formal enforcement; in contrast, if as Chinese philosopher Mencius says: “Human nature is good, just as water flows downward” then there would be a strong reliance on social norms.

In this current case, rather than Hobbes or Mencius, however, the voice that would be most useful is Confucius who believed that “by nature, men are nearly alike; in practice, they grow to be wide apart”.

In public policy, the key question is not whether we believe that individuals are by nature good or dishonest, but whether in practice, we design rules that — by nudges and formal enforcements — bends the arc towards honesty and the public good.