This version: 10 September 1999
Competitive tendering and service quality
Waverley City Council
Senior Research Fellow
Australian Research Council
Fraser, L. and Quiggin, J. (1999), 'Competitive tendering and service quality', <<Just Policy>> 17, 53-7.
Competitive tendering and service quality
The policy of competitive tendering and contracting for the provision of public services has been widely adopted by Australian governments in recent years. There is strong evidence that competitive tendering and contracting leads to cost savings for governments (Industry Commission 1996). However, there is little agreement as to whether these cost savings are primarily realised through improvements in operating efficiency, or by shifting cost burdens from governments to employees and service users (Quiggin 1994).
In particular, there has been considerable debate over whether competitive tendering and contracting for services previously provided by government agencies is associated with reductions in service quality, and therefore, in effect, with a transfer from service users to governments. Some overseas studies surveyed by the Productivity Commission found that service quality had generally declined following contracting out (Ascher 1987), while others found no evidence of statistically significant change (Savas 1977). Australian studies have produced mixed results. Some studies found a general deterioration in serivces (Evatt Research Centre 1990; Rimmer 1993), while others found either no change or improvements (Domberger, Hall and Li 1995). Few of these studies have considered possible sources of bias in responses.
One difficulty in comparing the results of different studies is that there are few instances where the same contracting process has been studied by different groups. An exception is the case of school cleaning in New South Wales. Domberger, Hall and Li (1995) presented evidence on cleaning services in public schools, hospitals and offices in New South Wales. They concluded (p. 1469) that 'when controlling for contract and industry characteristics, competitive tendering generally lowered prices whilst at least maintaining ex post quality of service'. In contrast, evidence from surveys of cleaners (Fraser 1997) and principals (Egan, Montesin and Adena 1995) in New South Wales schools suggests that the quality of services provided by cleaning agencies has declined following the introduction of competitive tendering and contracting. These results are consistent with the reported results of a review of cleaning services undertaken following a change of government in New South Wales, which led to an upgrading of service specifications.
Because competitive tendering and contracting leads to income transfers between employees, contractors, service consumers, and the public purchasers of services, incentive effects must be taken into account in analysing evidence on the quality of service. The results reported by Domberger, Hall and Li (1995) may be biased as a result of attempts by contractors to mislead inspectors regarding their average service quality. Evidence from school cleaners may be biased because contracting out is associated with reductions in paid working hours and uncompensated increases in work intensity. The evidence from school principals may be subject to compliance bias associated with the fact that the survey in question was undertaken by the New South Wales Teachers Federation, which opposed competitive tendering and contracting.
Survey evidence on NSW school cleaning services
Until 1992, cleaning of NSW government schools was undertaken for the NSW Education Department by the Government Cleaning Service (GCS). During 1992 and 1993, cleaning was provided on the basis of a mixture of competitive tendering, with the GCS competing against private firms, and non-tendered services provided by the GCS. The majority of contracts awarded under competitive tendering went to three major contractors, Berkeley Challenge, Menzies International, and Tempo Services.
From the beginning of 1994, the services previously provided by the GCS were contracted out. The GCS was not permitted to tender. Contracts for 2221 schools were divided between the three major contractors. In most cases, GCS employees transferred to employment with the contractors.
Although many cases of competitive tendering and contracting have been studied, the case of NSW school cleaning is unusual in that evidence on service quality covering the same service has been collected from three different groups of economic actors, with very different results. A possible explanation of the differences in results is that biases in the assessment of service quality are correlated with income transfers associated with competitive tendering and contracting. In this section, the main results of the studies of Domberger, Hall and Li (1995), Fraser (1997) and Egan, Montesin and Adena (1995) are described and possible sources of bias are considered.
Domberger, Hall and Li
Domberger, Hall and Li (1995) base their analysis on inspections undertaken expressly for their study over the period from October 1992 to June 1993. They compare non-tendered services provided by the GCS with tendered services provided by private contractors. The research was supported by the New South Wales Treasury and was undertaken with the co-operation of cleaning contractors. Although Domberger, Hall and Li analysed the results of inspections of public schools, hospitals and offices, this article is concerned solely with the results relating to public schools.
Domberger, Hall and Li conclude (p. 1468) that:
In special schools, the effect of introducing tendering and private contractors is to lower prices by some 50 per cent and to raise ex post quality by 35 per cent. Both were statistically significant. In other schools, the downward price effect of 49 per cent was statistically significant, but the more modest quality effect was not.
In assessing the reliability of the results reported by Domberger, Hall and Li it is important to consider the possibility of strategic manipulation by contractors arising from the fact that inspections were not undertaken on a random basis, but with advance notice provided to contractors. Domberger, Hall and Li (1995, p. 1463) do not directly address the possibility of manipulation by contractors. However, they state:
Precautions were taken to avoid the cleaners' efforts being influenced by the onset of inspections. Although access arrangements were often made in advance, individual cleaners were not notified as to which areas were to be inspected.
While these precautions raise the cost of strategic responses from individual cleaners that are designed to manipulate the monitoring process, they do not preclude such responses from contractors. To generate misleadingly high estimates of service quality, contractors must ensure that the entire school, rather than merely the areas selected for inspection, is cleaned unusually well prior to the inspection visit.
Evidence suggesting that the results of inspections may be biased is presented by Fraser (1997, p. 41) who quotes one of her respondents as saying:
We worked a long weekend but were only paid normal rates. The contractor wanted it cleaned because the monitoring team was coming. We were told that if we did not do the extra work the contractor will lose the contract. We got letters saying thanks for our help but we weren't paid the proper rates.
This evidence suggests that at least one of the three contractors was responding to the contract monitoring process by providing a higher standard of cleaning immediately prior to the monitoring visit than at other times. The fact that staff received letters of thanks implies a centrally organised response rather than an initiative by an individual supervisor. This is a rational response to the incentives associated with the contract.
The monitoring team referred to in this case was employed directly by the purchasing agency, and was not the inspection team used by Domberger, Hall and Li (1995). However, contractors would have similar incentives in either case. The fact that full contracting out of school cleaning services was under consideration was known, as was the fact that the policy advice provided by Domberger and his colleagues would play an important role in determining policy outcomes. In view of the oligopolistic structure of the contract cleaning industry and the small size of the sample studied by Domberger, Hall and Li, contractors had both a collective and an individual interest in providing misleading information. No comparable incentives or opportunities for manipulation arise for individual cleaners employed by government agencies potentially subject to competitive tendering and contracting, although it is possible that the managers of the Government Cleaning Service may have had an interest in maintaining the contract in public hands.
A survey of 45 school cleaners, concerned primarily with the impacts of contracting out on cleaners from non-English speaking backgrounds, was commissioned by the Commonwealth Bureau of Immigration, Multicultural and Population Research in 1995. Results were reported in Fraser (1997). Of 40 respondents to a question about changes in the quality of service, 29 (72.5 per cent) reported that quality had worsened, and 10 (25 per cent) reported no change or did not know. One cleaner reported an improvement in service quality.
Cleaners also reported that competitive tendering and contracting was associated with uncompensated increases in work intensity. The cost savings noted by Domberger, Hall and Li were achieved primarily through a reduction in the number of paid hours of cleaning provided to schools. Because of the labour-intensive nature of the service and because rates of pay were governed by awards determined through the arbitration system, it would have been difficult to reduce cleaning costs except by reducing the hours of paid work allocated to cleaning a given area.
Of the cleaners interviewed by Fraser (1997), 32 (80 per cent of respondents) reported a decrease in the number of cleaners employed following contracting out, while one reported an increase. The effects on hours worked by each employee were slightly more mixed, with nineteen cleaners (48 per cent of respondents) reporting a reduction in hours, and five reporting an increase. The net impact on total pay per employee was similarly mixed, and was further complicated by an exogenous increase in rates of hourly pay arising from an industry-wide change in award rates determined through the arbitration system. Sixteen respondents reported an increase in total wages and nine a reduction. However, nearly all respondents reported a reduction in the amount paid for a given work effort. All cleaners reported an increase in workload. The proportion of cleaners reporting unpaid extra work rose from 30 per cent before contracting out to 60 per cent after contracting out.
A reduction in hours of cleaning work might be achieved without loss of quality as a result of technological change. About 50 per cent of cleaners reported a change in the cleaning agent in use, and some regarded this as an improvement. Some changes in equipment, and replacement of old equipment took place immediately prior to contracting out. However, 80 per cent of respondents reported that there had been no change in equipment after contracting out. There is no general evidence that technological change is a major factor explaining reductions in the cost of cleaning services.
It is important here to consider possible sources of bias in responses elicited from cleaners. Since individuals cleaners' responses to a survey were unlikely to have any effect on their own conditions of employment, rational responses to the incentives associated with the survey would not imply strategic misrepresentation of changes in service quality. However, since competitive tendering and contracting has been associated with reductions in welfare for cleaners as a group, respondents who act in the interests of cleaners as a group would have an incentive to report a decline in the quality of service, and thereby raise the probability of a return to the provision of cleaning services by government agencies.
Alternatively, the phenomenon of cognitive consistency may produce a predisposition to report a decline in the quality of service. For workers, cognitive consistency normally entails a belief that an adequate standard of skill and effort is being supplied, and this belief implies that a reduction in the number of hours allocated to a service task must entail a reduction in the quality of service provided. Hence, even if effort levels prior to competitive tendering and contracting were low enough to permit a reduction in paid hours with no loss of service quality, cleaners might be predisposed to report a reduction in service quality.
Egan, Montesin and Adena
In 1994, following members' complaints that the quality of cleaning services had declined, the NSW Teachers Federation undertook a mail survey of all school principals in New South Wales, the results of which were analysed and reported by Egan, Montesin and Adena (1995). Responses were received from 1182 principals, a response rate of 56 per cent. Like cleaners, principals reported a decline in the quality of service. The proportion of school principals dissatisfied or very dissatisfied with the overall quality of service rose from 41 per cent prior to contracting out to 61 per cent after contracting out while the proportion satisfied or very satisfied fell from 32 per cent to 19 per cent.
Responses from school principals also supported the proposition that reductions in cost were achieved primarily through a reduction in the number of paid hours allocated to cleaning a given area. Seventy-seven per cent of the school principals reported a reduction in the number of hours of cleaning services provided, while 2 per cent reported an increase. Similarly, 45 per cent of principals reported that contracting out was associated with extra unpaid work. Most principals also reported that, after contracting out, unpaid cleaning work was undertaken by teachers, ancillary staff and students, with 20 per cent reporting unpaid work being undertaken by parents. It is not clear, however, to what extent such unpaid work has increased since contracting out.
Prior to contracting out, cleaners provided services that were not included in the contract specification. Some of these services were performed during paid hours and others were provided without pay. For example, extra unpaid hours were worked when special functions were held at the school or after a break-in. These extra services were not provided after contracting out. Unpaid hours worked after contracting out were required for the completion of contractually-specified work.
The loss of services not specified in contractual terms is partly reflected in the response of principals to the question 'Does the contractor recognize the cleaner's role in the school community'. Fifty-four per cent of principals answered 'No'. Similar concerns are evident in cleaners' comments reported by Fraser (1997).
Responses from school principals should display less bias than those from other sources. Since cleaning services are provided at no direct cost to schools, principals and other teachers will benefit from changes if and only if they result in an improvement in service quality. It is possible that compliance bias may arise from the fact that the survey in question was undertaken by the NSW Teachers Federation, which was known to be opposed to, or at least concerned about, competitive tendering and contracting. Respondents to a survey may exhibit compliance bias by making the response they believe is desired by the person or agency undertaking the survey.
However, this argument simply pushes the problem one stage further back. If competitive tendering and contracting of school cleaning services leads to improved service quality, teachers as a group will benefit from tendering. Hence, it is necessary to explain the opposition of the Teachers Federation and the complaints which prompted the survey on grounds such as inaccurate perception of service quality or ideological opposition to competitive tendering and contracting.
Competitive tendering and contracting raises the potential for large transfers of income. The results of officially-sanctioned studies such as that of Domberger, Hall and Li (1995) may have a significant impact on the policy process. Economic actors therefore have an interest in taking actions that may bias the results of such studies.
The results reported in this article support the common finding (Ganley and Grahl 1988; Milne and McGee 1992) that reductions in the cost to governments of providing services following competitive tendering and contracting arise at least in part because of reductions in effective wages per unit of effort. The transfers associated with reductions in effective wages will flow to employers and the purchasers of public services, in this case, governments.
The divergence between the results of Domberger, Hall and Li (1995) and those of Fraser (1997) may be explained by the observation that respondents and others in a position to influence the results of the study were beneficiaries of competitive tendering and contracting in the first case and losers from competitive tendering and contracting in the second. The respondents to the survey of principals analysed by Egan, Montesin and Adena (1995) had no incentive to give inaccurate responses. Although the results of the survey of principals may be affected by compliance bias, it is difficult to believe that compliance bias could account for a 35 per cent improvement in quality, as estimated by Domberger, Hall and Li, being translated into a significant decline in reported consumer satisfaction.
The debate over the impact of competitive tendering and contracting on service quality remains unresolved. At a minimum, the results reported in this paper show that improvements in official measures of performance may not be matched by an increase in satisfaction for service users.
Ascher, K. (1987), The politics of privatisation : contracting out public services, Macmillan Education, Basingstoke.
Domberger, S., Hall, C. and Li, E. A. L. (1995), 'The determinants of price and quality in competitively tendered contracts', Economic Journal 105(433), 1454-70.
Egan, J. T., Montesin, H. J., and Adena, M. A. (1995), NSW Teachers Federation School Cleaning Services Survey 1994, INSTAT Australia Pty Ltd.
Evatt Research Centre (1990), Breach of Contract: Privatisation and the Management of Australian Local Government, Pluto Press, Sydney.
Fraser, L. (1997), Impact of contracting out on female NESB workers: Case study of the NSW government cleaning service, Ethnic Communities' Council of NSW Inc.
Ganley, J. and Grahl, J. (1988), 'Competitive tendering and efficiency in refuse collection: a critical comment', Fiscal Studies 9(1), 80-5.
Industry Commission (1996), Competitive Tendering and Contracting by Public Sector Agencies, AGPS, Melbourne.
Milne, R. and McGee, M. (1992), 'Compulsory competitive tendering in the NHS: a new look at some old estimates', Fiscal Studies 13(3), 96-111.
Quiggin, J. (1994), 'The fiscal gains from contracting out: Transfers or efficiency improvements?', Australian Economic Review September, 97-102.
Rimmer, S. (1993), Aspects of competitive tendering and contracting in local government administration (part of PhD thesis), University of New England.
Savas, E. (1977), 'An empirical study of competition in municipal service delivery', Public Administration Review Nov/Dec, 717-24.