APRA is choosing a disastrous path on executive remuneration

Press/Media: Expert Comment

Description

New standards Rewarding managers for not behaving badly towards customers not only fails, it harms business in the long run. Elizabeth Sheedy One of themost important unresolved issues from the Hayne royal commission is thematter of executive remuneration. The first draft of the new prudential standards on remuneration (CPS511)was metwith howls of protest from the industry. APRAwent back to the drawing board, took a breather on account of COVID-19, and released its revised draft prudential standard on November 12.

There is much to like about the current draft.The focus on deferrals and adjustments for adverse risk and conduct outcomes is commendable and backed by logic and academic research. But APRA has persistedwith the requirement that non-financialmeasures be givenmaterialweight for determining eligibility for variable remuneration (paragraph 37). I am no apologist formisconduct in financial services, but this requirement is disastrous, inmy view, and one that is likely to lead toworse rather than better risk management outcomes. First, the requirement for non-financial measures to receive amaterialweighting is unnecessary from the perspective of accountability. If you are already adjusting the variable remuneration, potentially to zero, to account for adverse outcomes,what is the point of this additional requirement?

The issue of known bad behaviour has already been dealtwith. Aproponentofnon-financialmeasures could argue that theyprovide extra guidance to executivesregardingwhat theorganisation values.They enshrine theundoubted importanceofrisk andconductcriteria. But this should not be necessary if these criteria are baked into themalus and clawback provisions. In otherwords, if executives understand theywill lose their bonuses for bad risk/conduct outcomes, it should already be abundantly clear that these criteriamatter. The use of positive non-financial measures for determining bonus eligibility has no real benefit, yet it introduces significant risks. Charles Goodhart, aBritish economist, was one of the first to point out that linking measures to reward tended to reduce their usefulness.The harder something is to measure, themore Goodhart’s Law applies. Whenmeasures are linked to reward, people often findways tomanipulate or game themeasure. As a result, potentially usefulmanagement information is destroyed. Evenworse, if it becomes common knowledge that the risk/compliance measures tend to be gamed, this can have an adverse effect on risk culture.

If short-term performancemeasures can be gamed, it’s a short step to gaming other policies; the espoused risk policies are increasingly seen as ajoke. Even financial criteria have their measurement problems, despite the use of auditors to confirm their veracity.Butwhen it comes to themeasurement of nonfinancial criteria, andmeasures of risk/ compliance, it is impossible to find reliable short-termmeasures that are robust to gaming. We only know if performance has been truly good in the longer term.That’swhy deferral, ormalus (where a bonus has been awarded but not yet paid) and clawback of paid bonuses are the key to solving the remuneration problem because they rely on known outcomes rather than unreliable early signals.

Take the case of customer turnover data and complaints data. These are an excellent management tool for uncovering misconduct.But if youmake this data a criterion for rewards, it is likely to be manipulated by executives. For example, an executivemight find ways of placating an unhappy customer to prevent him or her frommaking a formal complaint.While the customerwalks away satisfied, the underlying business practice that led to the dissatisfaction continues. Other customerswho lack themotivation to express their dissatisfaction remain dissatisfied, and the complaints data shows nothing. A perfectly good piece of management information has been rendered useless, and frankly,we can’t afford to lose importantmanagement information. Take the case of risk/compliance measures. In the short term, thesemeasures always present a rosy picture relative to reality because non-compliance is hidden. Monitoring and controls are never perfect and policy violations come to light onlywith time.

Making the short-termmeasures into criteria for bonusesjust increases the odds that bad stuffwill be hidden. Goodhart’s Lawwins again. Staff engagement and culture surveys are another example of Goodhart’s Law.They are potentially useful formanagement purposes but are likely to lose their value once the scores are used for decisions about reward or promotion. Whensurveys areused inhigh-stakes contexts, the faking ofbehaviour increases and theresultsbecomemisleading.By linking staffsurveys to remunerationoutcomes, the risk is that a) remunerationoutcomeswillbe determined incorrectlysuch thatpoor conduct and gamingbehaviourwillbe rewarded, andb) the surveys lose their value as amanagement tool.Theusefulnessof management informationwillbedestroyed, fornoclearbenefit. We have awonderful chance to get executive pay right in the financial services industry.This is arguably themost important reform to reducemisconduct towards customers. Let’s notwaste it.

AFRGA1 A039 ProfessorElizabeth Sheedy is at theMacquarie Business School atMacquarie University. We only know if performance has been truly good in the longer term.

Period1 Dec 2020

Media contributions

1

Media contributions

  • TitleAPRA is choosing a disastrous path on executive remuneration
    Degree of recognitionNational
    Media name/outletAustralian Financial Review
    Media typeWeb
    CountryAustralia
    Date1/12/20
    DescriptionNew standards Rewarding managers for not behaving badly towards customers not only fails, it harms business in the long run. Elizabeth Sheedy One of the most important unresolved issues from the Hayne royal commission is the matter of executive remuneration. The first draft of the new prudential standards on remuneration (CPS511)was met with howls of protest from the industry. APRA went back to the drawing board, took a breather on account of COVID-19, and released its revised draft prudential standard on November 12.

    There is much to like about the current draft. The focus on deferrals and adjustments for adverse risk and conduct outcomes is commendable and backed by logic and academic research. But APRA has persisted with the requirement that non-financial measures be given material weight for determining eligibility for variable remuneration (paragraph 37). I am no apologist for misconduct in financial services, but this requirement is disastrous, in my view, and one that is likely to lead to worse rather than better risk management outcomes. First, the requirement for non-financial measures to receive a material weighting is unnecessary from the perspective of accountability. If you are already adjusting the variable remuneration, potentially to zero, to account for adverse outcomes, what is the point of this additional requirement?

    The issue of known bad behaviour has already been dealt with. A proponent of non-financial measures could argue that they provide extra guidance to executives regarding what the organisation values. They enshrine the undoubted importance of risk and conduct criteria. But this should not be necessary if these criteria are baked into the malus and claw back provisions. In other words, if executives understand they will lose their bonuses for bad risk/conduct outcomes, it should already be abundantly clear that these criteria matter. The use of positive non-financial measures for determining bonus eligibility has no real benefit, yet it introduces significant risks. Charles Goodhart, a British economist, was one of the first to point out that linking measures to reward tended to reduce their usefulness. The harder something is to measure, the more Goodhart’s Law applies. When measures are linked to reward, people often find ways to manipulate or game the measure. As a result, potentially useful management information is destroyed. Even worse, if it becomes common knowledge that the risk/compliance measures tend to be gamed, this can have an adverse effect on risk culture.

    If short-term performance measures can be gamed, it’s a short step to gaming other policies; the espoused risk policies are increasingly seen as a joke. Even financial criteria have their measurement problems, despite the use of auditors to confirm their veracity. But when it comes to the measurement of nonfinancial criteria, and measures of risk/ compliance, it is impossible to find reliable short-term measures that are robust to gaming. We only know if performance has been truly good in the longer term. That’s why deferral, or malus (where a bonus has been awarded but not yet paid) and claw back of paid bonuses are the key to solving the remuneration problem because they rely on known outcomes rather than unreliable early signals.

    Take the case of customer turnover data and complaints data. These are an excellent management tool for uncovering misconduct. But if you make this data a criterion for rewards, it is likely to be manipulated by executives. For example, an executive might find ways of placating an unhappy customer to prevent him or her from making a formal complaint. While the customer walks away satisfied, the underlying business practice that led to the dissatisfaction continues. Other customers who lack the motivation to express their dissatisfaction remain dissatisfied, and the complaints data shows nothing. A perfectly good piece of management information has been rendered useless, and frankly, we can’t afford to lose important management information. Take the case of risk/compliance measures. In the short term, these measures always present a rosy picture relative to reality because non-compliance is hidden. Monitoring and controls are never perfect and policy violations come to light only with time.

    Making the short-term measures into criteria for bonuses just increases the odds that bad stuff will be hidden. Goodhart’s Lawwins again. Staff engagement and culture surveys are another example of Goodhart’s Law. They are potentially useful for management purposes but are likely to lose their value once the scores are used for decisions about reward or promotion. When surveys are used in high-stakes contexts, the faking of behaviour increases and there sults become misleading. By linking staff surveys to remuneration outcomes, the risk is that a) remuneration out comes will be determined incorrectly such that poor conduct and gaming behaviour will be rewarded, and b) the surveys lose their value as a management tool. The usefulness of management information will be destroyed, for no clear benefit. We have a wonderful chance to get executive pay right in the financial services industry. This is arguably the most important reform to reduce misconduct towards customers. Let’s not waste it.

    AFRGA1 A039 Professor Elizabeth Sheedy is at the Macquarie Business School at Macquarie University. We only know if performance has been truly good in the longer term.
    Producer/AuthorProfessor Elizabeth Sheedy
    URLhttps://readnow.isentia.com/Temp/76455-38288611/1365957001.pdf
    PersonsElizabeth Sheedy