Perverse Incentive: When the Reward Structure Consumes Its Own Objective

In nearly every capital-intensive industrial group and nearly every distribution network, order volume in the final two weeks before a quarter closes runs markedly above the average of the preceding ten weeks — a pattern that repeats with enough regularity to be treated as structural rather than incidental. The second half of the same pattern appears in the opening weeks of the following quarter, when return rates rise, payment terms lengthen and renegotiation requests from the customer side multiply. A third observation, rarely placed alongside the other two, sits in the financial statements: the correlation between the quarter-end volume spike and the deterioration in receivables turnover is tracked by two separate units, in two separate reports, entirely independently of one another. The sales function bears no responsibility for receivables turnover, and the finance function did not design the quarter-end closing behaviour.

A comparable pattern surfaces on the production side through a different aperture. Where line efficiency is measured by downtime, deferring planned maintenance improves the metric directly, while the failure arising from that deferral typically falls outside the measurement window — sometimes into a successor's tenure altogether. Where recruitment is measured by time-to-fill, loosening screening standards shortens the metric, and the cost of that loosening surfaces twelve months later in turnover figures, at a point when it no longer appears on the recruiting function's scorecard at all. In all three cases the behaviour is locally rational, for the simple reason that what is rewarded is precisely what is being done.

The name for this pattern is perverse incentive — a reward structure that systematically funds behaviour producing the inverse of the intended outcome. Its mechanism originates in a proxy gap: unable to measure directly the thing it actually wants, the organisation selects a measurable substitute. What is wanted is a profitable and repeating customer relationship; what is measured is revenue volume. What is wanted is asset availability across an economic life; what is measured is this month's downtime minutes. What is wanted is the right person staying for a long period; what is measured is the day the requisition closes. So long as the gap between proxy and objective stays narrow, the structure functions and genuinely lowers cost; as the gap widens, the shortest route to optimising the metric increasingly runs through consuming the objective.

This tendency is better understood as a shortcut than as an error. No organisation can measure every dimension simultaneously — measurement carries cost, managerial attention thins as the count of metrics rises, and a multi-criteria bonus scheme can collapse in practice into an ambiguity that directs behaviour toward no criterion at all. Single-metric simplicity is a powerful instrument for aligning behaviour rapidly during a growth phase, which is why most commission systems are entirely functional in the conditions that gave rise to them. The difficulty lies not in the selection of the metric but in the metric remaining fixed while the conditions that justified it change; once the product mix broadens, the customer profile institutionalises or supply lead times extend, a metric that was correct four years earlier is now rewarding an altogether different behaviour.

A second and less frequently noticed property of this mechanism is its timing asymmetry. The reward is paid at the close of the measurement period, whereas the cost lands in subsequent periods, often distributed across several of them. Recognising the reward in a single line item and in a single named individual's favour, the organisation carries the cost diffusely — through warranty expense, rework, return provisions, late-collection financing charges and customer attrition. Because this distribution severs the causal chain at the level of the accounts, the structure itself is rarely interrogated; what gets interrogated is usually the person exhibiting the behaviour.

The institutional cost accumulates most concretely in the working capital cycle. Term extensions granted under quarter-end volume pressure move days-sales-outstanding permanently upward, and once that upward shift settles into customer expectation, reversing it takes years — the commission is paid once, while the extended term becomes a standard carried forward into every contract renewal. On the inventory side the same mechanism operates through purchase discounts: a procurement function rewarded on volume rebates places orders that depress inventory turnover in order to capture the rebate, and that inventory line generates its problem through age rather than through carrying value. The maintenance-deferral pattern, meanwhile, pulls the capital expenditure calendar forward, since unplanned replacement substituting for planned maintenance raises both unit cost and financing requirement.

The table at which this cost is priced most expensively is not the company's own management table but an acquirer's diligence table. Examining revenue quality, the reviewing side looks past the revenue level to its intra-period distribution, return rate, discount intensity and collection behaviour; a revenue profile concentrated at quarter-ends reads, to an institutional buyer, as revenue with low repeatability. The typical consequence of such a profile is less a direct reduction of the valuation multiple than a migration of the issue into deal structure: a portion of the earnings is pushed into an earn-out, the working capital target is redefined against a normalised receivables day count, representations and warranties are broadened to cover revenue recognition practice, and the escrow percentage is raised. Sellers frequently read these adjustments as negotiating hardness, when in substance they are the priced form of behaviour the reward structure itself produced.

Neutralising this structure is a matter of measurement architecture rather than personal awareness, and that architecture has four separable components. The first is pairing every reward metric with a constraint metric: where volume is measured, collection days become a threshold; where downtime is measured, planned-maintenance compliance becomes one; where time-to-fill is measured, twelve-month retention becomes one — below the threshold, no bonus is paid. The second is bringing the timing of payment closer to the timing of cost, releasing a portion of the bonus only once collection occurs, the return window closes or the warranty period lapses. The third is defining the metric as time-bound rather than permanent, attaching to every reward structure a validity note specifying which change in conditions triggers recalibration. The fourth is separating the unit that sets the metric from the unit the metric rewards; any arrangement in which a function defines its own performance criterion drifts predictably toward proxy optimisation.

In the projects BEIREK manages, this architecture is embedded within the contracting and reporting layer rather than bolted on beside it. Contractor, supplier and internal team incentives are consolidated into a single incentive map, with each reward item set against an explicit statement of the dimension — schedule, quality, cash, scope — that optimising it could consume; that map forms a standard part of the annex presented to the investment committee before any progress-payment or bonus structure is accepted. In milestone definitions, payment is tied to verification of the work rather than to its delivery: the payment differential between mechanical completion and performance test acceptance, the calibration of the LD cap, and the retention percentage are the three junctions at which perverse incentive appears most densely, and all three are addressed at term sheet stage.

The second line of intervention is rhythm. Because behaviour generated by an incentive structure becomes visible not at the moment of definition but six to twelve months afterward, periodic project reviews record not only progress but which dimension that progress consumed — order acceleration reported alongside term extension, schedule compression alongside change order volume, cost saving alongside rework rate. A decision record is also maintained, capturing the assumption under which each incentive item was accepted, the condition that would render it void, and the party responsible for monitoring that condition. Absent such a record, the discussion inevitably migrates to individuals once the structure breaks down; with it, the discussion returns to the assumption itself, and correction takes weeks rather than months.

The relationship between perverse incentive and institutional maturity becomes legible at exactly this point. An immature organisation reads behaviour produced by the reward structure as a matter of character and attempts to solve it by replacing the individual; a mature organisation, observing the same behaviour repeat across three successive occupants of a role, looks past the person to the structure. This distinction is the difference between formal authority and earned legitimacy as it registers in a measurement system: the authority to change a metric sits with formal position, while the capacity to see what a metric is producing is acquired only through a discipline of recording what the structure does over time.

The most productive question that can be asked of a reward structure is not what it encourages but what it releases when it is satisfied completely. An organisation unable to put that answer in writing has not designed the behaviour its structure produces; it is merely waiting to observe it.