The End of Proxy Economics?

For much of its modern history, development economics has been constrained by a remarkably simple problem.

We cannot improve what we cannot observe.

Every development intervention, whether in agriculture, education, trade, health, energy, governance or private sector development, ultimately seeks the same objective: to improve people's lives. Yet measuring whether this has actually happened has always been one of the discipline's greatest challenges.

Development economists have therefore become masters of inference.

We infer improved livelihoods from economic growth. We infer better education from school enrolment. We infer agricultural success from increased yields. We infer institutional effectiveness from policy implementation. We infer poverty reduction from household expenditure. These measures have served the profession well because, for decades, they represented the best evidence available.

There was little alternative.

Collecting information was expensive. Surveys were infrequent. Administrative systems were incomplete. Entire sectors of the economy remained invisible to policymakers. Measuring every household, every firm or every economic interaction was simply beyond the reach of governments and researchers alike.

Consequently, development economics evolved around carefully selected indicators, statistical sampling and well-established proxies. They became the language through which we understood development.

But what if the constraints that shaped those methods are beginning to disappear?

Today, economies generate unprecedented volumes of digital information. Mobile money transactions, electronic customs systems, digital payment platforms, satellite imagery, remote sensing, social protection databases, Internet-connected devices, business management systems and electronic government services all produce information that scarcely existed a generation ago.

Artificial intelligence adds another dimension.

Its greatest contribution may not be replacing economists, but helping them interpret evidence at scales and speeds previously unimaginable. Large and diverse datasets can now be organised, linked, analysed and interrogated in ways that would have required years of manual effort only a decade ago.

The significance of this transformation is often described as a "data revolution". I believe it is something more fundamental.

It is an observation revolution.

Development economics has historically depended upon observing fragments of reality and constructing statistically rigorous explanations around those fragments. Increasingly, technology allows us to observe larger parts of that reality directly.

This distinction matters.

GDP remains indispensable for measuring aggregate economic activity. Poverty rates remain essential for monitoring living standards. Household surveys remain irreplaceable for understanding dimensions of wellbeing that cannot be captured digitally. None of these measures become obsolete because better technology exists.

Rather, they become part of a much richer evidence ecosystem.

The question is no longer whether GDP is useful. It unquestionably is.

The more interesting question is whether GDP alone can continue to shoulder expectations it was never designed to meet.

Can aggregate production adequately explain changes in household resilience?

Can national averages reveal how technology transforms local markets?

Can annual indicators capture economic systems that increasingly evolve in real time?

Can development interventions still be evaluated primarily through periodic surveys when digital evidence is generated continuously?

These are not criticisms of traditional economics.

They are questions about its future.

Technology is changing not only the economies we study, but also the methods available to study them.

This shift has profound implications for development policy.

Imagine agricultural programmes informed not only by annual production statistics but by continuous streams of weather data, market prices, satellite imagery and farm-level digital transactions.

Imagine trade facilitation measured through the movement of goods, electronic documentation, customs processing times and logistics networks almost as they occur.

Imagine education policy supported by learning data that extends beyond enrolment registers and examination results.

Imagine monitoring systems that observe institutional performance continuously rather than periodically.

Such possibilities are no longer confined to theory.

Many already exist in various forms around the world.

The challenge is that our conceptual frameworks have not evolved as rapidly as our technological capabilities.

We continue to ask twentieth-century questions using twenty-first-century data.

Development economics therefore finds itself at an unusual moment.

For generations, the discipline focused on overcoming data scarcity.

The emerging challenge is learning how to make sense of data abundance.

This requires more than better software.

It requires new ways of thinking about evidence itself.

From Data Scarcity to Evidence Abundance

The profession is entering a period in which the limiting factor is no longer access to data, but the ability to transform data into credible evidence.

This raises a new generation of questions.

What should count as evidence?

How should different forms of evidence be integrated?

How do we distinguish observation from attribution?

How do we prevent more data from creating greater confidence than the evidence actually supports?

These are methodological questions, but they are also philosophical ones.

As technology expands what can be observed, economists must become equally rigorous in defining what can legitimately be concluded.

Artificial intelligence does not eliminate the need for careful reasoning.

If anything, it increases it.

The next generation of development economics will not be distinguished simply by bigger datasets or faster algorithms. It will be distinguished by the quality of its judgement.

Towards Better Questions

The future of development economics does not lie in replacing economic theory with technology, nor in abandoning established indicators. It lies in building richer, more integrated systems of evidence capable of bringing measurement closer to the realities that development policies seek to change.

This series explores that journey.

Future Insights will examine how technology is reshaping development economics across agriculture, trade, business development, evaluation, institutions and public policy. Along the way, they will explore why many existing measurement frameworks deserve to be revisited and how emerging approaches may help us understand development with greater clarity, precision and humility.

The objective is not to provide definitive answers.

It is to ask better questions.

Because every significant advance in development economics has begun not with a new dataset, but with a new way of seeing.

Quadripoint Reflection

If technology is changing what development economists can observe, it must also change how development is measured, evaluated and understood. The future of the discipline will depend not only on better data, but on asking better questions.