Rethinking Public Sector Productivity in the Age of AI

In an era where artificial intelligence (AI) promises to revolutionize public service delivery, how we measure government productivity is due for a radical rethink. For decades, policymakers have relied on traditional productivity metrics designed for factory floors and farming fields. But today, as public services increasingly rely on knowledge work, digital tools, and relational tasks, it has become clear: the old metrics no longer measure what truly matters.

Recent research from the Productivity Institute, the Bennett Institute for Public Policy, the Asian Productivity Organization, and the World Bank converge on a critical insight—our current productivity measures often undercut, rather than advance, the goals of effective, equitable governance.

Productivity in government is often equated with cost-cutting and throughput: how many forms processed, how much budget saved, how few employees used. But this model misses key realities. Outputs are often intangible. Preventing a fire or improving mental health doesn’t show up neatly in an output ledger. Outcomes matter more than volume. More services delivered doesn’t always mean better outcomes. Public value is hard to price. Unlike in the private sector, public goods often lack market signals.

This legacy logic leads to perverse incentives—rewarding speed over accuracy, efficiency over equity, and surface-level improvements over systemic ones.

AI introduces powerful tools that can automate routine tasks, mine data for insights, and reshape citizen engagement. But without rethinking how we measure success, we risk using these tools to chase the wrong targets. Automating bad metrics can mean an AI system optimizes for "cases closed" without addressing the complexity—or humanity—of those cases. Cutting costs without improving services might result in chatbots that reduce staffing budgets but fail to resolve citizen needs. Bias amplification becomes a real danger when AI inherits and scales systemic biases, yet traditional metrics may fail to detect the resulting harms.

In short, if we continue to evaluate AI deployments using outdated metrics, we may entrench inefficiencies and inequities under the guise of innovation.

A new framework for measuring public sector productivity needs to move beyond these legacy assumptions. The traditional private-sector approach—focused narrowly on inputs and outputs, or on maximizing efficiency relative to cost—is not fit for purpose in a public service context where goals are multidimensional, long-term, and often relational. Public agencies are not factories, and their work cannot be reduced to unit cost per transaction.

As economist Mariana Mazzucato has argued, we need to rethink who creates value in the economy—and how that value is measured. Public sector institutions are too often treated as passive administrators or cost centers, when in fact they play a vital role in shaping markets, driving innovation, and delivering public value. Applying private-sector productivity models obscures this role. It mistakes value extraction—cutting costs or outsourcing functions—for value creation, which depends on long-term investment, mission-driven leadership, and the co-production of outcomes with citizens. A modern productivity framework should reflect this reality by recognizing the state as an active value-creator, not merely a manager of inputs.

Instead, a modern productivity framework should be built on three pillars: effectiveness, experience, and equity. Effectiveness asks whether the intended public outcomes are achieved. This could mean reductions in poverty, improvements in education attainment, or better public health. Experience considers the quality and accessibility of public services from the perspective of the user—whether services are timely, responsive, and trusted. Equity evaluates whether services are reaching those most in need, and whether benefits are distributed fairly across populations.

Supporting this framework requires richer data and more sophisticated measurement tools. This means developing systems that track both quantitative outputs and qualitative outcomes in real time. It means adopting a value chain model that considers the relationship between inputs, activities, outputs, and long-term impact. And it means embedding continuous feedback loops into performance management, so AI-enabled services can adapt and improve as they scale.

The role of AI in this new framework should not be to replace public workers or shrink budgets for their own sake. Instead, AI should augment human capabilities, streamline administrative burdens, and help tailor services to community needs. Success will be defined not by cost per interaction, but by improvements in public wellbeing. This shift is essential if AI is to fulfill its promise of making government not just more efficient, but more effective, inclusive, and humane.

We’re entering a new era of public service—one where the tools of the twenty-first century demand measures that reflect twenty-first-century values. AI can enable smarter, fairer, faster government—but only if we change what we mean by productivity. That starts with asking not just how much we do, but how well we serve.

References

Asian Productivity Organization. (2019). Measuring public-sector productivity: A practical guide. Tokyo: APO. Retrieved from https://www.scribd.com/document/789804982/Measuring-Public-sector-Productivity

Bennett Institute for Public Policy. (2024). Redefining public sector productivity. University of Cambridge. Retrieved from https://www.bennettinstitute.cam.ac.uk/blog/redefining-public-sector-productivity/

Mazzucato, M. (2013). The entrepreneurial state: Debunking public vs. private sector myths. London: Anthem Press.

Mazzucato, M. (2018). The value of everything: Making and taking in the global economy. London: Penguin Allen Lane.

Productivity Institute. (2022). Making public sector productivity practical (with Capita). Retrieved from https://www.productivity.ac.uk/research/making-public-sector-productivity-practical/

Productivity Institute. (2024). Public sector productivity review: Fifteen questions. Retrieved from https://www.productivity.ac.uk/research/public-sector-productivity-review-fifteen-questions/

World Bank. (2021). Public sector productivity part 1: Why is it important and how can we measure it? Retrieved from https://documents1.worldbank.org/curated/en/913321612847439794/pdf/Public-Sector-Productivity-Part-One-Why-Is-It-Important-and-How-Can-We-Measure-It.pdf