Previously, when comparing with the private sector, the government always used the big companies (with 100 employees and above) as a benchmark, rather than small firms, even though 80 per cent of the working population is employed by firms with less than 50 staff.

This year, the pay-trend survey has, for the first time, included companies with 50 to 99 employees, whose data has been given a 25-per-cent weighting in the consolidated results. This has gone some way towards a more balanced representation, though the ratio can be reviewed in the future.

A pay system based on market surveys obviates the need for tedious and even acrimonious negotiations with the staff side. Whether any alternative systems would result in fewer conflicts and uncertainties is questionable.

The condition, of course, is that pay-level and pay-trend surveys are conducted objectively and regularly. A mechanism to adjust pay up and down, in line with market movements, should also be instituted to avoid disputes.

The government has always been among the better-paying employers in relation to the lowest-paid workers. Hence, the practice of bringing up pay-rise rates for lower-band civil servants to those for the middle band.

This, in effect, widens the pay gap with the private sector, which would backfire on junior staff during a recession, when a pay cut was necessary. The practice may well need to be re-examined later.

Stability and predictability come at a price. Broad pay comparability does not guarantee similar levels of performance or market demands for different job types. The present unified salary structure, where different grades are linked to one another in a complex network, is unduly rigid. Thus, adjustments to some grades would, by convention or pressure, result in similar changes to the linked grades, sometimes without strong economic reason. In future, the centralised pay scales should be gradually broken up, to facilitate more flexible and devolved reward management, based on departmental and grade features.

This would also make it easier for the pay levels of some civil service jobs to be adjusted in line with changing market trends, without triggering intra-service adjustments simply for the sake of parity. Strictly speaking, civil servants are currently paid for their job positions. To reassure the public of value for money, greater attention should be paid to performance evaluation and cost efficiency. In 2002, a pay review taskforce recommended that performance-related pay and flexible pay ranges (in lieu of fixed salary scales) be introduced in the medium term, initially at the directorate level. Pay movements within the pay range could then be used as a tool to encourage better performance and to discourage underperformance.

A performance-related pay system only works if the 'performance' of individual civil servants can be reliably measured by objective indicators acceptable to both staff and management.

Otherwise, performance evaluation would become biased towards those elements of work easier to measure, overlooking the non-quantifiable aspects. It may also be vulnerable to manipulation by unscrupulous supervisors.

At the end of the day, we should not assume that the pay system can provide all the incentives for good performance or all the answers to the problems of staff management. To achieve an efficient, productive and streamlined civil service, non-monetary incentives, like career development and job enrichment - and non-pay-related measures like reorganisation, de-layering, and the modernisation of work procedures and job reviews - should also be considered.

Anthony Cheung Bing-leung is an executive councillor and founder of SynergyNet, a policy think-tank