Economic Crisis Forecasting: Can It Predict the Next Crash?

The Forecast Afterward (0:00)

Hey, chibis! I'm Aiwee, and today we're talking about what economics gets wrong about predicting the next big crisis. If you enjoy stories like this, hit the like button and subscribe if you haven't already — let's go! Act One: The forecast that arrives after the storm In the autumn of two thousand eight, financial institutions were collapsing, credit was freezing, and governments were improvising emergency measures. Many of the most influential macroeconomic models had offered little warning.

Their equations could describe an economy in elegant motion, yet the financial machinery capable of bringing that system down was often represented weakly, or not at all. This is not a story about foolish people failing to guess one event. It is a question about what a discipline can honestly promise when it studies a moving target made of billions of decisions. If economics cannot reliably announce the next crisis, does that make it a pseudoscience? Not automatically.

Forecasting is only one test of knowledge, and complex systems are difficult to predict in fields far beyond economics.

Where Economics Breaks (1:15)

But economic advice can determine interest rates, taxes, employment, housing, and access to medical care. The stakes are too large for confidence to substitute for evidence. So the useful question is narrower and harder. Where does economics produce dependable knowledge, where does it break down, and what would intellectual honesty look like in between? Act Two: Why prediction is such a severe test A forecast is not the same thing as an explanation.

A theory may clarify how incentives work, or help design an institution, without naming the exact month a recession will begin. Still, public life rewards predictions because they are easy to quote. “Economists expect” sounds authoritative even when the underlying statement depends on assumptions about politics, technology, consumer behavior, and financial confidence. Philip Tetlock’s long-running research offers a useful warning about expert judgment.

Experts Versus Chance (2:15)

Across roughly two decades, his program collected more than eighty-two thousand predictions from two hundred eighty-four experts in geopolitics, economics, and related areas. On average, the experts performed only modestly better than chance. Some simple methods, such as extending recent trends forward, could compete surprisingly well. The lesson is not that expertise is worthless. It is that expertise needs measurement, feedback, and a record of being wrong.

Public reputation can make this problem worse. The qualities that attract attention are certainty, speed, and a memorable conclusion. Good forecasting often requires the opposite: explicit probabilities, changing one’s mind, and admitting that several outcomes remain plausible. Tetlock’s findings covered multiple fields rather than economics alone, so they cannot be treated as a verdict on economists specifically. But economics is unusually exposed to this incentive because its forecasts affect markets and elections, making dramatic claims profitable for institutions, commentators, and media platforms.

There is also a deeper technical difficulty. A country cannot be placed in a laboratory and assigned two different tax systems at the same time.

A Moving Target (3:30)

People learn from policy, react to expectations, change jobs, invent products, form institutions, and respond to one another. The object being studied changes while it is being measured. A policy that worked in one decade may produce a different result after technology, demographics, or global trade has changed. That does not excuse vague reasoning. It tells us what a responsible claim should sound like.

Instead of promising certainty, an economist might identify a mechanism, specify the conditions under which it should operate, compare it with evidence, and state what would change the conclusion. That style is less dramatic, but it is much easier to audit. Act Three: When elegance omitted the danger Before the financial crisis, dynamic stochastic general equilibrium models were a dominant framework in academic macroeconomics and central banking. These models were valuable tools for thinking systematically about households, firms, shocks, and policy. Their weakness was not that mathematics had no place in economics.

Models Omit Danger (4:45)

The weakness was that important features of the real economy could be simplified so aggressively that the model became least useful at precisely the moment of greatest danger. One major criticism after two thousand eight was that many influential versions did not adequately represent the financial sector or the possibility of cascading financial failure. A model can be internally consistent and still omit the mechanism that matters most. That is the difference between solving an equation and validating a description of the world. Paul Romer’s two thousand fifteen critique used the term “mathiness” for a related danger.

His argument, in substance, was that formal language can sometimes conceal a disputed assumption instead of clarifying it. Equations may create the appearance of precision while political or theoretical choices remain unexamined. Mathematics is not the culprit. Unmarked assumptions are. Paul Krugman’s two thousand nine essay argued that the profession had valued mathematical beauty over empirical realism.

After the Crisis (5:45)

Joseph Stiglitz made a similar criticism by describing models that were prepared for ordinary disturbances but poorly equipped for systemic illness. These critiques were influential, but they were not the end of the story. After the crisis, macroeconomic research expanded its attention to financial frictions, differences between households, and networks of interconnected institutions. Improvement does not erase the earlier failure, but it shows that criticism can produce a better research agenda. There is a limit here that no model can completely remove.

When policy changes behavior, historical relationships may stop holding. A forecast based on past rules can become unreliable once people anticipate the rule itself. Economists have formal ways to think about this problem, but no formula turns a changing society into a fixed object. The best model is still a map, and maps become dangerous when users forget that roads can move. Act Four: The parts that work If we judge economics only by dramatic macroeconomic forecasts, we miss some of its strongest achievements.

In narrower settings, researchers can define the problem, observe outcomes, test alternatives, and measure whether an intervention worked.

Trials That Test Claims (7:00)

The results may not predict an entire national economy, but they can improve a real institution serving real people. Randomized trials helped transform research on poverty reduction. In two thousand nineteen, the Nobel Memorial Prize in Economic Sciences recognized Abhijit Banerjee, Esther Duflo, and Michael Kremer for an experimental approach to alleviating global poverty. Their work examined practical questions in areas such as education and health by comparing interventions rather than relying only on broad theory. Trials still have limits.

Results from one place may not transfer perfectly to another, and measurement can miss long-term effects. But the method makes claims more testable. Market design offers another success story. Alvin Roth and Lloyd Shapley’s work on matching theory informed systems such as kidney exchanges and school-choice mechanisms. These systems do not simply assume that a free market will solve every coordination problem.

Designing Better Markets (8:00)

They define rules for matching people with resources when prices are incomplete, restricted, or ethically unsuitable. In two thousand twenty, Paul Milgrom and Robert Wilson were recognized for advances in auction theory and for designing auction formats used in settings including communications spectrum allocation. These auctions translated abstract ideas about information and incentives into public mechanisms capable of allocating valuable resources and raising substantial government revenue. Here, economics is not claiming to predict everything. It is designing a constrained process and checking whether the process performs.

This contrast is crucial. Economics often performs better when the question is bounded. How should scarce slots be assigned? Which intervention improves attendance? How can donors and recipients be matched?

It performs less reliably when asked to compress an entire society into a single forecast and then speak with certainty about the future. Act Five: The politics inside the numbers Economic analysis is never conducted outside society.

Incentives Shape Research (9:15)

Researchers choose what to measure, which outcomes matter, and which assumptions deserve attention. Funding can influence research agendas, though that does not mean every grant corrupts every result. Universities, public agencies, foundations, businesses, and political organizations all support work selectively. The danger is not one uniform ideology. It is the possibility that incentives reward conclusions that are useful to a sponsor, a faction, or a career.

Ideological disagreement is not automatically a flaw. Competing explanations can expose weaknesses and generate better tests. The problem begins when a school of thought treats criticism as disloyalty, or when a model is defended because it belongs to a tribe rather than because it survives evidence. Economics has sometimes been divided this way, especially when technical disputes carry direct consequences for taxation, regulation, wages, and public spending. Students in France captured part of this frustration in an open letter published in Le Monde in two thousand.

They objected to economics becoming too detached from real-world problems and called for greater pluralism. The movement later became associated with Real World Economics Review.

Disciplined Pluralism (10:30)

Whatever one thinks of its terminology or influence, its central challenge is reasonable: a discipline studying society should remain open to methods and evidence that do not fit one dominant template. That pluralism must also be disciplined. Adding every possible theory does not create knowledge. A useful approach is to make competing claims specific, test them where possible, compare predictive records, and separate evidence from values. Economics cannot decide the goals of society by data alone.

It can, however, clarify tradeoffs and show when a promised benefit depends on a cost being ignored. Act Six: What honest economics can promise The fairest conclusion is neither that economics is worthless nor that it is a mature predictive science. It is a developing social science with uneven performance. It has produced powerful tools for experiments, incentives, matching, and institutional design.

A Standard for Trust (11:30)

It has also built models that sometimes mistook simplicity for reality and forecasts that were delivered with more confidence than their evidence justified. That suggests a practical standard. Trust economic claims more when the mechanism is clear, the scope is narrow, the evidence is transparent, the result has been tested beyond one setting, and uncertainty is stated plainly. Be more cautious when a prediction covers an entire economy, depends on hidden assumptions, or arrives as a precise answer to a political question that the data cannot settle. Admitting “we do not know” is not surrender.

It is a safeguard against turning a model into an ideology. It also makes knowledge more useful. If economists distinguish between a tested mechanism, a plausible scenario, and a personal judgment, audiences can decide what deserves confidence and what deserves continued investigation. The strongest version of economics is therefore not a machine that predicts every boom and collapse. It is a toolkit for asking sharper questions about choices people make together, then learning from outcomes without hiding the misses.

That toolkit remains incomplete, but incompleteness is not fraud.

What Would Disprove It (12:45)

The real scientific failure would be refusing to measure the limits. So when you hear that economists predict the future, ask what kind of claim is being made, what evidence supports it, and what would prove it wrong. That habit is useful whether the forecast concerns inflation, housing, employment, or the next crisis. If this helped you think more clearly about economic evidence, you can explore another topic on the channel, or simply share the question with someone who treats certainty as proof.