Government performance occupies a central place in economic debates. Yet a fundamental difficulty remains: how can we measure the productivity of institutions whose services are generally not sold in markets? An OECD study published on October 23, 2026, highlights the limitations of the statistical tools used to assess public-sector output. As digital transformation and artificial intelligence reshape public administrations, this question extends far beyond accounting methodology. It challenges how we understand value creation, institutional efficiency, and the economic role of the state.

A Paradox at the Heart of Public Economics

A company that produces more with the same resources achieves, all else being equal, a productivity gain.

In public administration, establishing this relationship is considerably more difficult.

A court may reduce the time required to process cases. A hospital may improve the quality of care. A tax authority may simplify its procedures. A school may provide better support to its students.

These improvements can generate substantial benefits for citizens and the economy.

Yet they do not necessarily translate into measurable increases in public-sector productivity.

The difficulty stems from a fundamental characteristic of government activity: many public services have no market price through which their economic value can be directly assessed.

In market sectors, prices, production volumes, and costs provide the basis for constructing productivity indicators, despite their own methodological limitations.

For non-market public services, this information is often unavailable.

Statisticians must therefore rely on conventions to estimate government output.

Some of these conventions create a paradox: public-sector production is measured using the resources employed to produce it.

In other words, statistical instruments may sometimes struggle to distinguish an administration that has become more efficient from one that simply continues to mobilize the same resources.

This is precisely the challenge examined by Orsetta Causa, Maxime Nguyen, Paul Schreyer, and Tomomi Tanaka in a study published by the Organisation for Economic Co-operation and Development on October 23, 2026.

Entitled Measuring Public Sector Productivity: Concepts, Challenges and International Practices, the study raises an essential question: when we conclude that government productivity is barely improving, are we measuring actual performance or the limitations of our statistical instruments?

When Output Is Calculated from Inputs

Productivity rests on a relatively straightforward relationship: the amount of output generated per unit of resources employed.

In an industrial company, it is possible to measure the number of vehicles produced, hours worked, equipment used, and intermediate inputs consumed.

In public administration, defining the output itself becomes more complicated.

What exactly does a ministry produce?

Administrative decisions? Regulations? Authorizations? Services delivered to citizens? The institutional capacity to organize society?

And how should these different outputs be compared?

A judicial ruling, a medical consultation, and a food safety inspection differ fundamentally in nature and economic consequences.

Historically, national accounting systems have addressed this difficulty by valuing many non-market public services according to their production costs.

These costs include employee compensation, intermediate consumption, and consumption of fixed capital.

This approach makes it possible to incorporate public administration into national accounts.

However, it introduces a major limitation when measuring productivity.

If the volume of output is estimated from the volume of resources employed, efficiency gains become difficult to identify.

An administration delivering more services with the same workforce may therefore fail to see this improvement fully reflected in official statistics.

Conversely, an increase in resources may raise the accounting value of production without demonstrating an equivalent improvement in services delivered.

An important distinction must nevertheless be made between measuring output in monetary terms and measuring its volume.

National accounting methodologies are not uniform, and several countries already use direct output measures for certain public services.

The problem is therefore not that all government production is systematically equated with expenditure.

Rather, existing methodologies remain heterogeneous, imperfect, and sometimes insufficient to capture the actual transformation of public administration.

Three Ways of Measuring Government, Three Interpretations of Performance

The OECD study examines three broad approaches to measuring public-sector output.

The first is based on inputs.

It estimates production using the resources employed, including labor, capital, and other factors required to deliver public services.

Its principal advantage is that it relies on data that are generally available.

It also makes it possible to construct relatively consistent statistical series when information on actual services delivered is insufficient.

However, this method substantially limits the ability to observe productivity improvements.

The second approach measures the volume of services delivered directly.

In education, this may involve the number of students receiving instruction.

In healthcare, it may include consultations, procedures, or treatments.

In the justice system, it may involve the number of cases processed.

This approach brings the measurement of public-sector output closer to the methods used in market activities.

Nevertheless, it introduces another major difficulty: producing more services does not necessarily mean producing better services.

A hospital that increases the number of consultations without improving patient outcomes is not necessarily more effective.

A court that accelerates proceedings at the expense of judicial quality does not necessarily generate a collective benefit.

The third approach therefore seeks to adjust output volumes for quality.

Depending on the sector, this may involve educational outcomes, healthcare quality, or other measures of service performance.

Conceptually, this approach comes closer to capturing the actual purpose of government activity.

But it introduces additional challenges.

How should service quality be defined?

How can improvements be attributed specifically to public administration?

How should outcomes be compared across societies with different demographic, economic, and social characteristics?

Measurement consequently becomes not only a statistical exercise but also a conceptual one.

Measuring the quality of public services requires defining what constitutes good performance.

Government Does More Than Deliver Services: It Shapes Economic Conditions

The challenge becomes even more significant when considering the institutional functions of the state.

Public administration does not merely produce individual services.

It also helps establish the conditions under which economic activity can develop.

A predictable judicial system facilitates contract enforcement.

Reliable land administration reduces uncertainty surrounding property rights.

Effective regulation can improve market functioning.

Efficient customs administration can reduce international trade costs.

These institutional functions may generate substantial economic benefits without corresponding to outputs that are directly sold or easily quantified.

Consider an administrative reform that reduces the time required to obtain an investment authorization from several weeks to a few days.

The number of civil servants may remain unchanged.

The budget may not decrease.

Even the number of authorizations issued may remain relatively stable.

Yet the reform may reduce transaction costs, accelerate investment, and improve resource allocation.

Part of the value created will appear in business performance rather than in the administration's own accounts.

This does not mean that all these benefits should be recorded as government output.

Doing so could lead to double counting and attribute to public institutions outcomes influenced by numerous other factors.

Nevertheless, the distinction reveals a fundamental limitation in debates about administrative efficiency.

The economic contribution of an institution does not necessarily coincide with the statistical value of the services it produces.

Government can exert a decisive influence on productivity throughout the economy without that contribution appearing directly in public-sector productivity indicators.

The Problem with International Comparisons

This difficulty becomes particularly important when comparing government performance across countries.

Economic debates frequently rely on indicators such as public expenditure as a share of GDP, government employment, or the cost of public services.

These indicators are useful.

They help assess the structure of public finances and the resources mobilized by governments.

But they are insufficient to establish whether one state is more productive than another.

Two countries may devote comparable shares of national income to education while achieving very different outcomes.

Conversely, two administrations may report different unit costs because they serve populations with fundamentally different needs.

Geographic dispersion, demographic structures, income levels, regulatory obligations, and service characteristics all influence production costs.

An administration operating across a sparsely populated territory may need more resources to guarantee comparable access to public services.

A healthcare system facing population aging may experience rising expenditure without any deterioration in efficiency.

International comparisons must therefore distinguish differences in productivity from differences in missions, constraints, and service quality.

The OECD study also highlights the uneven development of statistical practices.

Some countries possess administrative data that support relatively detailed direct measures of output.

Others remain heavily dependent on input-based methodologies.

Comparing their performance without accounting for these methodological differences can produce misleading conclusions.

This limitation is particularly relevant for emerging economies, where the availability, coverage, and quality of administrative data may be weaker.

Artificial Intelligence Could Make the Measurement Problem Much More Visible

The digital transformation of government introduces a new dimension to this debate.

For decades, much of administrative production relied on labor-intensive processes.

Processing applications, verifying documents, retrieving information, and preparing decisions required substantial human resources.

Automation is gradually changing this organizational model.

Procedures can be digitized.

Information can be verified automatically.

Documents can be analyzed more rapidly.

Requests can be directed to the appropriate departments without systematic human intervention.

Artificial intelligence expands these possibilities further.

It can assist public employees with document retrieval, information synthesis, case analysis, and anomaly detection.

But how should the resulting economic gains be measured?

Imagine an administration capable of processing twice as many applications with the same workforce because of automation.

If output is measured directly through the number of applications processed, a productivity gain can become visible.

If output is estimated primarily from the resources employed, that gain may be much less apparent.

The problem becomes even more complicated when technology improves quality rather than quantity.

An administration may reduce errors, strengthen the traceability of decisions, or make public services more accessible.

These improvements can be substantial without producing a proportional increase in the number of services delivered.

Conversely, automation may also accelerate incorrect decisions, reinforce certain biases, or transfer additional costs to users.

A faster administration is therefore not automatically a more productive administration in economic and social terms.

The challenge is to measure output volumes, resources, processing times, service quality, and outcomes simultaneously.

Artificial intelligence does not merely make administrative transformation possible.

It also requires reconsidering the instruments used to evaluate that transformation.

An Administrative Revolution That Could Remain Statistically Invisible

The potential consequences extend well beyond national accounting.

If productivity gains associated with digital transformation are inadequately measured, public decisions may rely on an incomplete assessment of their results.

An effective reform may appear disappointing because its benefits are not captured by existing indicators.

Conversely, a reform that reduces expenditure may be considered successful even when it undermines access to services or their quality.

This confusion between cost reduction and productivity improvement is particularly significant.

A reduction in administrative expenditure does not necessarily constitute a productivity gain.

It may reflect better organization.

But it may also result from fewer services being delivered, deteriorating quality, or costs being transferred to citizens.

Similarly, higher expenditure does not automatically indicate declining efficiency.

It may finance improvements in service coverage, quality, or resilience.

Evaluating administrative transformation therefore requires distinguishing several questions.

How much does government spend?

How many services does it produce?

What resources does it use?

What level of quality does it deliver?

And what effects do those services have on society and the economy?

These dimensions are related, but they are not interchangeable.

Productivity concerns the relationship between output and the resources employed.

Effectiveness concerns the achievement of objectives.

Quality refers to the characteristics of services delivered.

Public value encompasses the broader collective benefits that government activity seeks to generate.

Confusing these concepts means expecting a single indicator to answer fundamentally different questions.

The Risks of Governing Through the Wrong Indicators

The way an administration is evaluated directly influences how it operates.

When an organization is assessed primarily on the number of cases processed, it may be encouraged to prioritize simpler cases.

When evaluated on cost reductions, it may be tempted to limit certain services.

When judged on average processing times, it may improve its statistics without necessarily resolving the most complex situations.

This reflects a familiar problem in public policy analysis: once an indicator becomes a target, it may gradually lose its ability to represent the reality it was intended to measure.

The challenge is particularly pronounced in public services, where missions are multiple and sometimes conflicting.

A judicial system must be efficient but also impartial.

A healthcare system must be cost-effective but also accessible.

A tax administration must collect revenue while respecting taxpayers' rights.

A regulatory authority must protect the public interest without imposing disproportionate burdens.

Reducing these missions to a single indicator can produce trade-offs that undermine their underlying objectives.

The solution is not to abandon measurement.

It is to develop evaluation frameworks capable of combining several dimensions.

Quantitative indicators should be assessed alongside quality data, costs, processing times, and outcomes.

Where possible, the effects of reforms should be evaluated using methods that distinguish their specific contribution from independent economic or social developments.

Statistical results must also be interpreted in light of the actual responsibilities assigned to institutions.

More measurement does not automatically produce better governance.

What matters is measuring the right things.

A Decisive Challenge for Emerging Economies

For emerging economies, this question carries particular importance.

Administrative modernization is often central to development strategies.

Improving the business environment, simplifying procedures, digitizing services, and reforming institutions are regularly presented as essential conditions for investment and competitiveness.

Yet their actual economic contributions can be difficult to establish.

An administration that improves business registration, secures land transactions, or reduces customs clearance times may help transform an economy.

These effects may become visible through investment, business formalization, or lower transaction costs.

They do not necessarily appear in administrative productivity statistics.

The challenge is therefore twofold.

Governments must modernize public services while developing information systems capable of measuring their performance.

This issue is directly relevant to African countries pursuing digital transformation across their administrations.

Morocco, in particular, has undertaken reforms aimed at digitizing public services, simplifying administrative procedures, and modernizing government operations.

Evaluating these transformations cannot be limited to counting the number of digital platforms created, procedures digitized, or services made available online.

Such indicators describe technological deployment.

They do not, by themselves, demonstrate productivity gains.

A more comprehensive evaluation would examine actual reductions in processing times, costs borne by administrations and users, error rates, service accessibility, and the extent to which public needs are met.

It should also account for differences between territories and population groups.

A fully digital procedure may reduce costs for some users while creating new difficulties for those with limited access to technology.

Digital transformation should therefore not be evaluated solely through its ability to automate existing processes.

It should also be assessed according to its contribution to genuinely better public services.

Government Productivity Cannot Be Separated from Economic Productivity

One of the most important implications concerns the relationship between administrative performance and economic growth.

Productivity analysis generally distinguishes economic sectors according to their output and the resources they employ.

This distinction is essential for constructing coherent statistics.

But it should not obscure the interdependencies between public institutions and private economic activity.

A company may improve productivity through new equipment, better organization, or technological innovation.

It may also benefit from improved infrastructure, education, legal certainty, or administrative efficiency.

These factors influence the conditions under which businesses employ their own resources.

Institutional reform can therefore generate effects that spread throughout the economic system.

Its collective return may exceed the savings directly achieved by the administration implementing it.

This does not mean that all public expenditure is productive.

Some spending may be inefficient, poorly targeted, or inadequately evaluated.

Other expenditures may generate benefits that fall short of their costs.

The central challenge is precisely to develop instruments capable of distinguishing value-creating expenditure from spending that produces insufficient results.

Yet this distinction becomes difficult when measures of public-sector output remain incomplete.

The quality of measurement therefore influences the quality of budgetary decisions.

A fiscal consolidation strategy that ignores outcomes may weaken essential functions.

Conversely, an expansionary spending strategy that neglects costs and productivity may perpetuate structural inefficiencies.

The question is not whether government should systematically become larger or smaller.

It is what public resources actually enable institutions to produce.

Toward a New Framework for Measuring Public Performance?

The methodological developments examined by the OECD open several possibilities.

The growing availability of administrative data makes it possible to observe services delivered with greater precision.

Digitization facilitates the monitoring of procedures and processing times.

Information systems can improve the measurement of unit costs.

Statistical methods can help distinguish output volumes from their qualitative characteristics.

However, these advances do not eliminate the fundamental challenges.

The quality of an administrative decision remains harder to measure than the number of decisions issued.

Educational outcomes also depend on family and social circumstances.

Healthcare performance is influenced by individual behavior, environmental conditions, and demographic structures.

The economic effects of institutional reform may emerge several years after implementation.

It would therefore be unrealistic to attempt to reduce all public value to a single indicator.

A more robust evaluation architecture should distinguish measures of output used in national accounts from performance indicators used to manage public institutions.

The former must preserve statistical consistency and avoid double counting.

The latter can incorporate broader dimensions, including quality, accessibility, equity, speed, and outcomes.

This distinction would make it possible to understand government performance more accurately without artificially attributing all observed economic and social benefits to the state.

It would also provide a stronger basis for evaluating digital investments and artificial intelligence projects.

One of the risks of the coming years is that governments will multiply administrative modernization programs without developing the instruments required to determine which initiatives actually create value.

Rethinking Performance Before Rethinking the Size of Government

Debates over public finances frequently revolve around two competing perspectives.

The first emphasizes the need to reduce expenditure, public employment, and the overall size of government.

The second stresses the importance of public services, collective investment, and institutional capacity.

Both perspectives address legitimate concerns.

Yet they sometimes share a common weakness: they focus more on the resources mobilized than on the results actually achieved.

The OECD study does not settle this debate.

Instead, it highlights the conceptual and statistical limitations of the instruments used to assess public-sector productivity.

This observation should influence how administrative reforms are designed and evaluated.

The question cannot be limited to how much government costs.

It must also address what government produces, the quality of its services, the resources required to deliver them, and their contribution to the wider economy.

The distinction becomes particularly important as digital technologies transform administrative processes without necessarily producing immediate changes in budgetary structures.

A government can become more effective without becoming smaller.

It can also become less expensive without becoming more efficient.

And it can mobilize additional resources without proportionately improving the services it delivers.

None of these situations can be properly interpreted without sufficiently accurate measures of output and quality.

The question of public-sector productivity is therefore not simply about government efficiency. It is also about our ability to recognize, measure, and compare what public institutions actually produce.

As artificial intelligence promises to transform public administration, this distinction could become decisive.

A revolution in government productivity that statistical systems fail to identify would be difficult to evaluate, manage, and replicate.

And an economic policy framework incapable of accurately measuring the value created by public institutions risks continuing to confuse control over resources with control over results.