Late one night, a drunken adult is stooped low on the side of a street. Their eyes scan a small area of pavement illuminated by a nearby street light. “Is everything okay?” asks a passing police officer. “I’ve lost my wallet”, replies the drunk. The police officer looks at the ground. “Are you sure this is where you dropped it?”, the officer asks. “No”, replies the drunk, “it’s more likely I dropped it across the road, but this is where the light is”.

The ‘streetlight effect’ illustrates the problem of observational bias, where researchers search for their ‘answers’ in the places that are easiest to look. They follow the availability of data, funding or theoretical frameworks. And it pervades even the most prestigious areas of natural science, whether in physics or medical research.

The problem is particularly prevalent in economics. And the Institute for Fiscal Studies (IFS) – the tax and spending watchdog treated with reverence by much of the UK’s mainstream media – is no exception. Indeed, the ‘IFS view’ of the economy is narrowed by the spotlight problem in three important ways.

First, the IFS’s work is based primarily on microeconomic analysis (the study of individuals and firms) using the data that is most easily obtainable. Yet we know that macroeconomic dynamics (relationships between aggregate, economy-wide factors like interest rates) can move the goalposts. Indirect effects from government decisions can snowball, such as through known ‘multipliers’ on public spending (for example as estimated by the Office for Budget Responsibility and International Monetary Fund), and through the possibility of either crowding-in or crowding-out private investment and innovation, as discussed by economists such as Marianna Mazzucato.

Paying too little attention to such dynamics means at least half the story can be lost. Microeconomics can tell us how much it costs to employ a new teacher, but it is far less well equipped to tell us what the value of a transformational education programme might do for overall wages, taxes and wider public good. The more significant the proposed changes are, the more powerful these indirect effects can be – and therefore the worse the observational bias of discounting them can become.

Second, the IFS’s view on ‘fiscal risk’ is disproportionately focused on risks to the public balance sheet. But this ignores a crucial point. Public balance sheets are one of the most efficient ways to pool risk across society. Private debt, which is currently high and rising, is potentially far more risky to the economy. Arguably, governments should be taking on more risk, not less. The costs of an aging population are far better borne by government, than if left to families to pay privately, and is probably far better for government to run even a prolonged deficit, than to risk global warming. The right question isn’t always how much will public debt rise, often it is what will happen if it doesn’t, and which outcome is likely to be worse?

Finally, like many economists, the IFS relies on neoclassical economic theory to guide its interpretation of the evidence. One consequence of this is that it assumes markets can deliver ‘optimal’ outcomes. Consequently, actions that might disrupt market ‘efficiency’ (such as tax or subsidy) are seen as inherently more risky than those that don’t. The consequence for policy is that disrupting markets requires a high burden of evidence, whereas leaving them alone is deemed preferable by default.

Is this the right framework? In part, the answer depends on how well markets are actually working. In the UK today, inequalities in earnings are higher (and still rising) than at almost any other time in the post-war era. The past 12 years have failed to yield a real terms increase in average pay for the first time in 200 years. And now, at least 20 years after scientific consensus over climate change was first reached, there is precious little time left for advanced economies to transform their energy and resource consumption, without risking catastrophic social cost. There is a strong case that the burden of evidence for not intervening in today’s markets, should be far higher.

This is not to single out the IFS. Limitations in policy appraisal are ubiquitous across all think tanks and university departments. And within the fields of public accountancy and microeconomic fiscal study, the IFS consistently produce among the most careful, clear and transparent work around.

The problem, rather, is in the relationship between the IFS and the UK media: especially broadcasters, and especially during elections. Broadcasters are, too often, too quick to assume they have found an infallible voice of authority, and seemingly unwilling, or unaware of the need to better distribute their airtime among economists with different perspectives.

As a young Albert Einstein is alleged to have said, “if we knew what we were doing, it wouldn’t be called research”. But the answer is neither to stop asking the questions, nor to stick with one voice and hope for the best. With the IFS set to report on all the main parties’ manifestos today, it is vital that the public are given enough information to decide who is looking under the right streetlights.