Most forecasters failed to correctly read the tea leaves or coffee grounds for 2023. A global recession did not occur and interest rates did not fall. The interesting question is why they keep getting it wrong.
First, the economic tools are weak. Emanuel Derman, a physicist turned financier, lamented that most economists had never really seen a successful model and could not differentiate between good and bad ones. The basic theories of growth, inflation, fiscal or monetary policy, currency values or debt—all essential to predictive modelling—are vague and in constant flux. US Federal Reserve Chairman Jerome Powell graciously admitted in 2022: “We understand better how little we understand.”
Reductionist in nature, economic models focus on a few measurable variables because of limited data availability. Statistics and mathematics are deployed to give the appearance of rigour and science. Casual correlation masquerades as causality in the absence of tractable hypotheses. Assumptions such as rational, utility-maximising actors are frequently not satisfied in the real world. Economics is infatuated with the ideal of unattainable equilibriums.
Unlike hard physical measurements, economic and financial data, if available, is suspect, sometimes arbitrary choices between competing methodologies. Timeliness is problematic as are subsequent data revisions as more detailed information becomes available. Former Chinese premier Li Keqiang once referred to official Chinese statistical releases as “man-made” and “for reference only”. Unfortunately, the Middle Kingdom is not the only place where this holds true.
The intersection with politics means government, central bank or regulatory decisions affect outcomes. Unfortunately, the ability of economists to anticipate major geopolitical events is poor. Few foresaw the Ukraine and Gaza conflicts and or their trajectory, even after they started.
Economists seem oblivious to basic data dynamics. In authoritarian systems, underlings must ensure that statistics meet forecasts to avoid punishment or ensure career progress, even if this means falsifying data. Bad news must always be the fault of external forces or actually good news. Even irrefutable measures of poverty levels can be manipulated by changed thresholds.
Complex feedback loops are ignored. Economic models and forecasts are integral to the formation of expectations and actual real world behaviour. Predictions may shape policy which, in turn, influences results. If the estimated outcomes were wrong in the first place, then incorrect action taken pursuant to these affects results. For example, improved modelling can encourage banks and investors to assume greater exposures, changing systemic risk.
American economist Frank Knight defined uncertainty as the absence of quantifiable knowledge about an event. It denotes fundamental ignorance or the boundary of knowledge making certain outcomes entirely unpredictable. Risk, a subset of uncertainty, is quantifiable or measurable. Analysts mistake uncertainty for risk and vice versa.
Second, forecasts betray behavioural biases identified by Daniel Kahneman and Amos Tversky. There is ‘induction’—formulation of general rules without sufficient information. ‘Availability’ means conclusions rely on the information obtainable, rather than that needed. Forecasts are subject to ‘contamination’ that is based on irrelevant but superficially connected data. Economists are prone to overconfidence in ‘calibration’, underestimating the statistical confidence levels and error terms around each data input.
There is ‘confirmation’—using models or seeking out data that confirms a preconceived hypothesis, while ignoring evidence to the contrary—and ‘affect heuristic’, allowing preconceived value judgements to impede our conclusions. ‘Hindsight’ encourages unhelpful reliance on history. Another problem is the Dunning-Kruger effect, a cognitive predisposition in which people with limited competence overestimate their abilities.
Third, the prevalent incentive structures are influential. Economists provide forecasts not because they know, but because they are asked to by clients or employers. The payoff structure is unusual. Forecasting’s key performance indicator is relative rather than absolute. This means an economist’s forecasts tend to cluster around a point with few if any outliers. As John Kenneth Galbraith held: “It is far, far safer to be wrong with the majority than to be right alone.”
Fourth, forecasting reflects its target audience. Recipients are interested in using the information for financial gain, primarily through investments and business or policy decisions which benefit from the predictions.
But the real (macro- and micro-economic factors) and financial (asset prices) economy no longer exhibit any strong relationship. Mundane indicia, like earnings, dividends or growth, do not correlate to values. Forecasts, even if accurate, cannot be used necessarily to make money. Returns necessitate taking real risk based on actual uncertainty, and without uncertainty there is no risk.
There is an essential forecasting paradox. If someone knows the course of future events, then would they provide you with this knowledge for a modest outlay or for free. Besides, for every expert opinion there is an equal and opposite one that can be found.
These shortcomings mean that economists rarely meet the professional standards that John Maynard Keynes specified as “humble, competent people, on the level with dentists”. Their forecasts too are of dubious utility. This poses a different question—why do we read or listen to predictions which we know are deeply flawed?
There are some possible explanations. Financial institutions employ economists to cloak their activities in a modicum of seriousness. They are a form of marketing. Many economists who spend time on air as media talking heads are advertising their employers—a kind of product placement.
Investors, businesses and governments employ economists as they do management consultants—as scapegoats for when things go awry. They cannot be held responsible if they acted on the best advice. Of course, occasionally the forecasts might prove correct in which case the person can claim the credit. Alternatively, it might be for the entertainment or similar value.
The enthusiasm for forecasts speaks to the modern search for certainty and predictably. But physicist Wolfgang Pauli was probably correct in his view that the best we can hope to achieve is simply to misunderstand at a deeper level.
Satyajit Das, Former banker and author; his latest book is Fortune’s Fool: Australia’s Choices