Why central banks keep getting inflation forecasts wrong

The Federal Reserve’s 2021 forecast for year-end inflation was off by 6.5 percentage points, a statistical miss equivalent to a weather channel predicting a sunny day and instead receiving a Category 5 hurricane. Jerome Powell and his colleagues at the Eccles Building weren't just slightly optimistic; they were operating in a mathematical vacuum that failed to account for the most aggressive price surge in forty years.

This wasn't an isolated American failure. The European Central Bank, led by Christine Lagarde, and the Bank of England under Andrew Bailey, all sang from the same "transitory" hymnal while consumer prices tore through their 2% targets. Between 2021 and 2023, the collective forecasting errors of the G7 central banks represented the most significant breakdown in macroeconomic modeling since the stagflation of the 1970s.

📉 The math of a missed target

Central banks rely almost exclusively on Dynamic Stochastic General Equilibrium (DSGE) models. These are complex systems of equations designed to predict how the economy reacts to shocks, but they harbor a fatal flaw: they assume the economy always wants to return to a steady state of equilibrium. In the eyes of a DSGE model, a global pandemic followed by a land war in Europe is just a temporary "blip" that the system will naturally iron out without permanent structural shifts.

In early 2021, the Fed’s median "Dot Plot" forecast suggested that the Federal Funds Rate would remain near zero through 2023. By the time 2023 actually rolled around, the rate was sitting at 5.25%. This discrepancy exists because DSGE models are backward-looking; they use the last thirty years of low-inflation data to predict a future that no longer resembles the past. They are essentially driving a high-speed train by looking out the rear window.

Claudio Borio, head of the Monetary and Economic Department at the Bank for International Settlements (BIS), has argued that these models have become "intellectual straitjackets." They rely heavily on the concept of the "output gap"—the difference between what an economy is producing and what it could potentially produce. If the model says the output gap is wide, it assumes there is no room for inflation. But in 2021, the models were measuring "potential" based on 2019 labor participation, completely ignoring that millions of workers had retired, died, or simply vanished from the workforce.

📦 The supply-side blindness

Central bankers are trained to manage demand. By raising or lowering interest rates, they can make it more or less expensive for you to buy a car or for a business to build a factory. They are, however, utterly powerless when it comes to the supply side of the equation. No amount of interest rate hikes can force a semiconductor factory in Taiwan to work faster or clear a backlog of 100 container ships sitting off the coast of Long Beach.

In 2021, Philip Lane, the ECB’s Chief Economist, repeatedly pointed to "base effects" and supply chain disruptions as reasons to ignore the rising heat in the Eurozone. The logic was that since these were supply-side shocks, they would fade. What Lane and his peers missed was that supply shocks don't just sit there; they bleed into expectations. When a business owner sees the price of shipping a container rise from $2,000 to $20,000, they don't wait for the "transitory" phase to end—they raise prices immediately to survive.

This supply-side blindness was exacerbated by the "Just-in-Time" era's collapse. For decades, globalization acted as a massive deflationary force. China’s entry into the World Trade Organization in 2001 provided a seemingly endless supply of cheap labor and goods. Central banks got used to this "free" downward pressure on prices. When the gears of globalization ground to a halt during the lockdowns, the models had no "variable" for a world where goods became scarce instead of abundant.

🏗️ The broken Phillips Curve

For half a century, the Phillips Curve was the North Star of central banking. It posits a simple inverse relationship: when unemployment is low, inflation is high, and vice versa. It’s a clean, elegant trade-off that fits nicely into a spreadsheet. The problem is that the Phillips Curve has been effectively "flat" since the mid-1990s, and it completely shattered during the post-pandemic recovery.

In 2022, the U.S. unemployment rate hit a 50-year low of 3.4%. According to traditional Phillips Curve logic, this should have triggered a wage-price spiral that sent inflation into the double digits. While inflation did spike, the "spiral" didn't look like what the Fed expected. Real wages—adjusted for inflation—actually fell for many workers. The models couldn't explain why inflation was soaring even when labor didn't have the bargaining power they traditionally associated with such tight markets.

Former Fed Chair Ben Bernanke recently conducted a review of the Bank of England’s forecasting failures. His 115-page report was a polite but devastating indictment of institutional inertia. Bernanke noted that the BoE was using outdated software and failing to incorporate the reality of "shocks to energy and food prices" into their core narrative. They were essentially using a 1990s map to navigate a 2020s minefield.

🧠 The psychological trap of "Anchoring"

Central bankers are obsessed with the word "anchored." They believe that as long as the public believes inflation will eventually return to 2%, it will. This is a circular piece of logic that borders on the religious. The theory, championed by Janet Yellen during her time as Fed Chair, is that inflation expectations are self-fulfilling prophecies. If you think prices will go up, you demand a raise, and the business raises prices to pay you, thus creating the inflation you feared.

The failure of 2021-2022 proved that "anchoring" is a fragile illusion. When the price of eggs doubles and gas hits $5 a gallon, nobody cares what the Fed's 5-year breakeven inflation rate says. Consumers react to the prices they see at the checkout counter, not the academic papers published in Jackson Hole. The Fed stayed "lower for longer" because their surveys of professional forecasters—other people using the same broken models—told them expectations were still anchored.

This creates a feedback loop of institutional groupthink. The FOMC (Federal Open Market Committee) is comprised of brilliant PhDs from the same handful of universities, all reading the same research and attending the same conferences. When Larry Summers, a former Treasury Secretary, began shouting in early 2021 that the $1.9 trillion American Rescue Plan was too much stimulus for a supply-constrained economy, he was treated as an outlier. The "institutional anchor" was too heavy to move until it was far too late.

🛢️ The energy transition oversight

Central banks also failed to account for "Greenflation." As the world shifts away from fossil fuels, the cost of carbon-intensive energy rises, and the capital required for the transition—copper, lithium, nickel—becomes significantly more expensive. This isn't a cyclical change; it's a structural one. However, energy prices are often stripped out of "Core CPI" because they are considered too volatile for policy decisions.

By focusing on Core inflation (which excludes food and energy), the ECB ignored the massive signal coming from the energy markets in late 2021. Even before Russia’s invasion of Ukraine, natural gas prices in Europe were skyrocketing due to low storage levels and a pivot toward renewables that hadn't yet been fully backed by storage capacity. Lagarde’s team treated this as "noise." But energy is the fundamental input for every good and service in the economy. You cannot have a "core" price that isn't eventually impacted by the cost of the electricity used to produce it.

Isabel Schnabel, a member of the ECB’s Executive Board, was one of the few to eventually admit that the transition to a low-carbon economy poses a persistent upside risk to inflation that standard models simply don't capture. The models assume that "shocks" are random and temporary. They don't have a way to handle a deliberate, multi-decade policy shift that intentionally makes certain forms of energy more expensive.

⏳ The data lag and the "Last Mile"

Government data is notoriously slow. The Consumer Price Index (CPI) is a snapshot of the past, often lagging real-world price changes by weeks or months. During periods of slow economic change, this doesn't matter much. During a period of hyper-volatility, it's a death sentence for accurate forecasting. By the time the Fed saw the "official" numbers in 2022, the inflationary fire had already jumped the firebreak.

We are now seeing the reverse of this problem with the "last mile" of inflation. Central banks are struggling to get inflation from 3% down to their 2% target. Their models suggest that high interest rates should have cooled the economy by now, but the "Neutral Rate" (the interest rate that neither stimulates nor restricts growth) has likely moved. Because they don't know where the new Neutral Rate is, they are guessing. If the Neutral Rate has risen because of higher productivity or government deficits, then a 5% interest rate isn't as restrictive as the Fed thinks.

Commercial data providers like Truflation or the Billion Prices Project at MIT often show shifts in consumer behavior months before the Bureau of Labor Statistics. Yet, central banks remain tethered to official statistics to maintain their "credibility." This creates a paradox: to appear credible, they must use the most "official" data, even if that data is the least useful for predicting the immediate future.

⚖️ The ghost of 1970s credibility

The ultimate reason central banks keep getting it wrong is a deep-seated fear of repeating the mistakes of Arthur Burns, the Fed Chair in the 1970s who cut rates too early and allowed inflation to become entrenched. This fear has created a "reaction function" that is biased toward being late. They would rather wait for "definitive proof" that inflation is dead before cutting rates, even if their own forecasts suggest a recession is looming.

This bias is why we see "higher for longer" becoming the new mantra. They are overcompensating for their 2021 failure. If they were wrong on the way up, they are terrified of being wrong on the way down. This isn't data-driven policy; it's trauma-driven policy. The models are being adjusted manually to reflect a world that central bankers no longer feel they understand.

The era of the "Great Moderation"—that three-decade window of low volatility and predictable inflation—is over. We have entered a period of "The Great Volatility," characterized by geopolitical shifts, climate-related supply shocks, and massive fiscal deficits. Central banks are still trying to play 3D chess with a 2D board. Until they abandon the idea that the economy is a closed loop that always returns to "normal," their forecasts will continue to be more fiction than fact.

The real danger isn't that the models are wrong; it's that central bankers still believe they can be "right" if they just add one more variable to the equation. In a world of fragmenting supply chains and $34 trillion in U.S. national debt, the 2% inflation target isn't a scientific necessity—it's an arbitrary goalpost from a different century. The next decade won't be won by the central bank with the best model, but by the one with the most humility to admit that the old rules no longer apply.

🔧 Build Your AI Business: Guides, templates, and tools at SochiLab