Yongjian Ke

Social infrastructure PPPs: when the asset is a service

· 7 min

A motorway either carries traffic to a standard or it does not, and payment can be tied to that. A hospital or an aged care home is harder. What is being bought is care given to a person, and most of what makes that care good or bad is difficult to write into a schedule. Four papers, three with Kun Wang and Shankar Sankaran and one led by Liang Ma at Shanghai University, look at how public-private partnerships behave when they are applied to that kind of asset (Wang, Ke, & Sankaran, 2019; Wang, Ke, Liu, & Sankaran, 2022; Ma, Hu, Zhu, & Ke, 2023; Wang, Ke, & Sankaran, 2025).

How far the contract reaches into the service

Hospital PPPs are usually sorted by how far private participation extends. The first model covers the building and its maintenance, and sometimes non-clinical services such as catering and cleaning. The second adds clinical services. The third goes beyond the hospital to the primary care centres around it, so the operator also influences where patients are treated.

The 2019 paper takes the Chinese Ministry of Finance project database as at April 2018, isolates the 94 non-profit hospital PPPs with complete records, and finds the distribution weighted heavily towards the narrowest model: 74 projects in the first model, 19 in the second and one in the third. Build-operate-transfer accounted for 66 of the 94, and a hybrid of user charges and government payment for 52. We then compared one project from each model on their published documents: the value for money report, the fiscal affordability report, the feasibility report, the procurement documents and the PPP agreement.

Two results are worth putting together. The first is about competition. Capacity requirements rise sharply with scope. In the first-model case, no hospital experience was required, the first open tender was abandoned because fewer than three qualified applicants came forward, and the project was eventually won by a construction firm whose main business was ground and foundation work and steel structures. In the second-model case, where bidders had to show prior hospital investment and operation, four passed prequalification, and the implementation plan recorded that the market did not have the conditions for full competition. Three passed in the third-model case.

The second is about measurement. In the second-model case, the operating performance indicators covered auxiliary services such as cleaning, security and washing, and no clinical service indicators appeared in any published document. In the third-model case, no performance assessment was described at all, and the agreement said the appraisal would be settled by the parties at some later point.

The further a contract reaches into the service, the fewer firms can bid for it, and the less of what they deliver is measured.

Sydney: what a thin specification costs

The 2022 paper looks at what that gap produces in operation. Northern Beaches Hospital opened on 30 October 2018, replacing Manly and Mona Vale hospitals, with Healthscope responsible for design, construction and operation under a 20 year concession for public services and a further 20 years for private patients. Within months the NSW Legislative Council opened an inquiry. It received 236 submissions and nine supplementary submissions, held three public hearings and one in camera hearing, and reported in February 2020.

We coded that report against a social sustainability framework adapted from Labuschagne et al. (2005), using five categories: employees, patients, local community, society, and stakeholder participation. Patients were added as a group, since the original framework was built for the process industry and has no equivalent.

Employees and patients absorbed most of the problems. Wages were delayed in the first months of operation. Staff transferring from the two closed hospitals lost entitlements they had been promised on migration. Junior medical officers were asked to help manage private patients although the contract tied those positions to public patients, and their hours were heavily overloaded. Orientation for permanent staff was announced six days before the training week, one month out from opening. Supplies were reordered as they ran out rather than held in reserve. The electronic medical record system could not exchange information with the system used across the state's public hospitals. Access was its own problem: no direct bus route from the northern beaches, a shortest public transport journey of about an hour and a half, and a catchment reaching 30 kilometres to Palm Beach along a road that closes in floods.

The two advantages we identified, co-location of public and private services and the scope for innovation, were raised by the operator and by NSW Health. The disadvantages were recognised by nearly everyone else. The committee recommended that no future public hospital in the state be delivered as a PPP, and in August 2020 the government confirmed that none were being pursued.

None of this proves that the model caused the failures. A newly opened hospital of any ownership can be short of staff and supplies. What the case does show is that the things that went wrong were largely the things the contract did not specify or measure: rostering, orientation, supply buffers, and system compatibility. Site access was a different kind of problem, and one the government rather than the operator was placed to solve.

Shenzhen: what a thick specification looks like

The 2025 paper is the same question from the other end. Project X is an elderly care PPP in a Shenzhen district, based on a welfare centre that had been built and operated publicly since 2015. It was tendered in June 2019, contracted in July, and has 791 beds, a 15 year concession and a user-pays mechanism. Four hundred beds are subsidised for local residents, including people with no children, no income and no capacity to work, recipients of the minimum living allowance, veterans and the very elderly. The remaining 391 are sold at market rates.

We set the social sustainability indicators as the goal and worked backwards to the decisions that produced them, identifying 33 practices across the project lifecycle. The instructive ones are early and concrete. The number of subsidised beds offered served as the bid price and carried 10 per cent of the evaluation, with the technical and commercial proposals taking 45 per cent each, weighted towards the operating and maintenance plan and towards elderly care experience. The output specification set a minimum food ingredient budget of 800 RMB per bed per month. Care and meals were defined as core services the operator may not subcontract. Prices had to be fixed before residents moved in, so that quality could not be cut or additional services pushed on residents later. Care services were then given 60 per cent of the operating performance score, and payments moved with the assessment.

None of that is novel. It is ordinary procurement practice applied to the social content of the service rather than to the building. It is also all settled before signature, which is the point. There is no comparable lever afterwards.

What the wider literature says

The fourth paper steps back to 234 journal articles published between 1998 and 2021. Two features of that body of work matter here. The evidence is geographically narrow: of the papers that named a country, 72 studied the United Kingdom, followed by Australia with 21, China and Italy with 13 each, and the United States with 11. And the topic weighting is financial: financing and economic viability accounted for 44 papers, ahead of governance and regulation with 25, performance management and contract and relationship management with 23 each, risk management with 22, and facilitating and hindering factors with 16.

The share of national PPP activity that is social also varies widely. Education and healthcare together account for roughly 55 per cent of UK PPP projects. In China the figures are about 5 per cent for education, 2 per cent for healthcare and 1 per cent for sport. In India, social and commercial infrastructure is under 10 per cent. The countries with the deepest social infrastructure PPP experience are also the countries producing most of the research, which is worth remembering before treating any of it as general.

The review's answer to its own question is qualified. Risk allocation is where the model most clearly does something that traditional procurement does not. Value for money and performance remain unresolved, in part because financial information on these projects is rarely public and concessions outlast the studies that examine them.

What follows for practice

Set scope by what can be measured, not by what can be transferred. If the buyer cannot describe and assess care or clinical quality, that part of the service does not belong in the contract yet. In the Chinese cases, scope had been extended past the point where the buyer had indicators to attach to it.

Put the social content into the priced documents. Four instruments carry it: the output specification, the payment mechanism, the evaluation weights, and the definition of what may not be subcontracted. Intentions recorded in a business case do not survive into operations. Those four do, because they are what the operator is paid against.

Expect competition to thin as scope grows, and plan for that. A prequalification bar set high enough to guarantee capability can be high enough to remove the competitive pressure that was the reason for tendering in the first place. Where only two or three bidders are realistic, the discipline has to come from the contract, since it will not come from the market.

Treat the transition as the exposed period. Staff transfer terms, orientation, supply buffers and records systems all fall due in the same few weeks, and Sydney is the case of what happens when none of them is ready.

What the papers do not settle

None of the studies has a counterfactual. The Sydney hospital is not compared with a public hospital opening at similar scale, and the Shenzhen project is not compared with the same facility under its earlier public operation, although four years of that operation preceded the PPP and would have made the strongest available test.

Both case studies were selected on outcome. Northern Beaches was chosen because it had a public inquiry, so the evidence is a record of complaint assembled from parties with reasons to submit. Project X was chosen because practitioners in the sector called it successful and it had no negative press coverage. Neither describes the middle of the distribution.

The Shenzhen evidence rests on three senior managers from the consultant and the project company, interviewed in July 2021 with two follow-up calls in October. No employee, resident, family member or community member was interviewed, in a study about their well-being. Seven of the 21 indicators, all of them concerning the local community and society, could not be traced at all, because the practices meant to deliver them were stopped by the pandemic.

The 2019 comparison is of documents rather than outcomes. All three projects were under construction or newly operating, so the paper says nothing about how the three models perform once running, which is the question the models were built to answer. The review is confined to English language journal articles indexed in Scopus, and in this field a good deal of the financial evidence sits in audit office reports rather than journals.

The data also age quickly. The Chinese database snapshot is from 2018, and PPP arrangements there have since been reorganised, with new projects steered towards user-pays franchise arrangements. The distribution described in the 2019 paper belongs to a period that has closed.

Journal papers discussed on this page