Private & public sector management

Does Centralizing Public Purchasing Really Save Money?

By Abderrazzak Jirari

· 8 min read

The clock on the central tower of the Parliament Building in Québec City
Photo: Cephas

On September 1, 2020, Québec handed its group purchasing to a single body. The Centre d'acquisitions gouvernementales (CAG) took over the procurement activities of five entities: the Centre de services partagés du Québec and four non-profit organizations from the health and education networks (Vérificateur général du Québec [VGQ], 2024). The bet was clear. According to the government, the previous model did not allow the benefits to be maximized across government (Gouvernement du Québec, 2019). Buying together, in larger volumes, was supposed to cost less.

Six years later, the question is no longer whether the idea was sound. It is whether anyone can show that it works. In November 2024, the Auditor General of Québec concluded that the CAG was failing to carry out a significant share of the group purchases it had planned, and that it was not adequately assessing the savings it reported (VGQ, 2024).

My argument here is that these difficulties are not primarily a matter of execution. They stem from three gaps built into the very design of the organization: between the plan and the capacity, between the structure and the needs, and between control and information.

A plan the organization cannot deliver

The statute placed planning at the heart of the system. The CAG must prepare a government acquisitions plan that reflects the needs of public bodies, then send it to the Conseil du trésor, Québec's treasury board (Act respecting the Centre d'acquisitions gouvernementales [ACAG], s. 8). It must also consult the bodies concerned (s. 6). This is what the literature calls a deliberate strategy: a direction set in advance, formalized, then translated into operations (Whittington et al., 2020).

The problem appears when the plan is compared with what is actually delivered. Measured against the files scheduled at the start of the year, the plan's completion rate was only 52% in 2022-2023, then 70% in 2023-2024. The CAG itself published 69% for 2022-2023, because it measured its result against files that had been rescheduled during the year (VGQ, 2024). In one year, the value of group acquisitions fell by nearly 26%, from $5.8 billion to $4.3 billion.

These delays have concrete consequences. In 2022-2023, 56 files were awarded by mutual agreement, without a call for tenders, 23 of them because of a delay. In other cases, departments and institutions had to buy for themselves what the group purchase was supposed to supply (VGQ, 2024). The body created to pool purchases thus ends up producing what it was meant to eliminate: individual purchases, concluded under time pressure.

This is an emergent strategy, but one that is endured rather than chosen. The tension between the planned and the emergent is documented in the public sector (Axelsson & Höglund, 2023). It becomes a problem when the plan serves more to announce than to steer.

A machine for professionals' needs

Creating the CAG also meant choosing an organizational form. Mintzberg (2023) uses the term "programmed machine" for the configuration that coordinates work through standardization: rules, procedures, a clear hierarchy. This form is efficient for producing simple goods in volume. It is far less so when needs call for a diagnosis and a tailored response, which is typical of professional organizations such as hospitals.

The CAG displays these traits: the standardized call for tenders as its main coordination mechanism, centralized planning, and mandatory use that four ministers can impose by order (ACAG, s. 9). The Auditor General's figures show how far this standardization goes. Of the 159 group purchases concluded in 2022-2023, 94 were awarded to the lowest bidder and 56 by mutual agreement. Only 7 used the best quality-price ratio, and none took the total cost of acquisition into account (VGQ, 2024).

The issue is not new. In 2019, while the bill was being studied, the Chair of the Conseil du trésor had promised to review the lowest-bidder rule at a later stage (Desrosiers, 2019). The same finding applies to buying from Québec suppliers. The share of the value of supply contracts concluded with Québec suppliers went from 60% in 2022-2023 to 54% the following year, then to 50% in 2024-2025, while the target was rising to 64% (CAG, 2025).

Yet Québec research offered grounds for caution. Studying assistance to homeless people in Montréal, Dupuis and Farinas (2009) described a system of complex services coordinated through mutual adjustment among many actors, in an organic rather than hierarchical mode. In their view, such a system copes with complexity better than an integrated whole controlled by rules. Brunet and Savoie (2003), for their part, point out that an informal organization, made up of networks of expertise and initiative, attaches itself to every formal structure and compensates for its shortcomings.

Merging five entities with different practices, processes and systems (VGQ, 2024) risks losing that tacit knowledge without replacing it. The statute does provide counterweights: the CAG must consult public bodies and may set up user advisory committees (ACAG, s. 6). But these lateral links remain grafted onto a structure that standardizes.

Results-based control without verifiable results

The CAG is not a government department. It is a legal person and a mandatary of the State, headed by a president and chief executive officer appointed by the government (ACAG, ss. 1 and 19). This agency model rests on an exchange: managerial autonomy in return for measured results. Emery (2005) described the distance between that idea and its implementation in public organizations.

On paper, the control architecture is complete. On the executive side, the CAG manages its acquisitions in accordance with the directions of the Chair of the Conseil du trésor. A governance committee chaired by the Secretary of the Conseil du trésor, made up mostly of senior government officials alongside two independent members, ensures compliance with government directions (ACAG, ss. 5, 27 and 28). On the legislative side, the financial statements and the activity report are tabled in the National Assembly, and the Auditor General audits the organization's books every year (ss. 42 and 43). These are the two channels of control distinguished by Proulx (2010): the central agencies and Parliament.

The whole edifice depends on one condition: reliable information. That is precisely what is missing. The CAG reported $407.9 million in savings for 2022-2023, but that amount covers only 20% of the 1,275 agreements in force, and the calculation method varies from one file to the next. The Auditor General writes that it "was not possible to calculate the actual value of the savings" (VGQ, 2024, my translation).

Management data are incomplete as well. For a single file, 124 complaints were sent to the CAG and only 12 appear in its register (VGQ, 2024). Results-based control then loses its grip. Neither the Conseil du trésor nor members of the National Assembly can verify whether the organization is delivering the savings promised in exchange for its autonomy. Control exists formally. It does not provide the assurance it announces.

What has changed since the audit

It would be unfair to stop at the 2024 picture. The CAG accepted the Auditor General's four recommendations and produced an action plan (VGQ, 2024; CAG, 2025). Its 2024-2025 annual management report records 101 group purchases completed out of 142 planned or added during the year, and $591.9 million in reported savings, above target. A new acquisitions management system is being prepared (CAG, 2025).

Three reservations are nonetheless in order. First, the rate of contracts completed within the timelines agreed with clients stands at 71%, below the 75% target (CAG, 2025). Second, the savings are still reported by the organization itself: the review carried out by its internal audit unit does not constitute an audit opinion (CAG, 2025). Third, the Committee on Public Administration, which heard the president and chief executive officer on December 2, 2025, expressed concern about inconsistencies in the data of the acquisitions plan. It asked for a status report on the new system and a summary of the measures taken to ensure data quality (Commission de l'administration publique, 2025).

The legislature, for its part, chose another path. The Act assented to on April 2, 2026 makes the CAG's contracts subject to the conditions applicable to government enterprises, and the resulting policy came into force on October 2, 2026 (Act to reduce bureaucracy, 2026; CAG, 2026). The CAG sees this as a gain in agility. The response is consistent with the second gap, that of rigidity. On its own, it does not close the third: more flexibility in awarding contracts does not make results more verifiable.

What the case reveals beyond procurement

The CAG is an almost complete application of new public management: an autonomous agency, quantified targets, accountability for results. Osborne (2006) criticizes this paradigm for looking inside the organization, when public services are increasingly produced between organizations. The case proves him right. The savings from a group purchase are decided in the relationship between the CAG and its clients: how the need is defined, whether deadlines are met, how much of each contract is actually used. That is where information is lacking.

I have described elsewhere, writing about the Institut national d'excellence en éducation, a State that borrows the vocabulary of collaboration in order to recentralize. The CAG shows the other side of the same problem: a State that centralizes in the name of performance without giving itself the means to measure it.

I draw three lessons for anyone who designs or leads a public organization. A plan is worth only the capacity behind it: better to plan what can be delivered than to publish what one hopes for. Form must follow the nature of the needs: specialized medical equipment is not bought the way paper is. And an agency earns its autonomy through the quality of its data, since data are the only currency of results-based management.

The idea of pooling the State's purchases remains defensible. What is less defensible is announcing the gains before being able to demonstrate them.

(The author)

Portrait of Abderrazzak Jirari

Abderrazzak Jirari

Strategy, Digital and AI Consultant

Trained in political science (from an honours bachelor's to a master's at Université de Montréal), I'm now completing a graduate diploma (DESS) in private- and public-sector management at HEC Montréal.