Why Central Banks Are Struggling to Read Their Own Models

For decades, central bank forecasting models were built on a relatively stable set of assumptions: predictable supply chains, gradually shifting labor markets, and inflation that responded in fairly linear ways to interest-rate changes. Those assumptions are increasingly being tested.

The past several years have exposed the limits of models calibrated on historical data that no longer describes the present. Supply shocks, geopolitical fragmentation of trade routes, and rapid shifts in energy pricing have introduced volatility that traditional models struggle to capture, let alone predict with the precision policymakers would like.

This has left central bankers in an uncomfortable position: forced to make consequential decisions on interest rates while acknowledging, sometimes explicitly, that their own forecasting tools carry wider error bands than officials are used to admitting publicly. Several monetary authorities have quietly begun supplementing traditional models with alternative, higher-frequency data sources — from shipping data to real-time card-spending figures — precisely because the old models are proving too slow to adapt.

The practical consequence for markets and businesses is a policy environment that is genuinely harder to predict than in the pre-pandemic decade. Forward guidance, once a reliable signal of a central bank's intentions, has become more conditional and more frequently revised, reflecting the underlying uncertainty in the models themselves rather than indecision on the part of policymakers.

None of this means the models are useless — they remain the best tools available for organizing a coherent view of a complex economy. But both markets and the public would benefit from central banks being more transparent about just how much uncertainty surrounds even their most confident-sounding projections.