Why Experts Keep Getting the Economy So Wrong

Key takeaways

• The U.S. economy behaves like a "hundred-leg stool" instead of a three-legged one: because services, tech, and diverse sectors now dominate over manufacturing, knocking out any single leg (oil, tariffs, one industry) barely tips the whole structure, unlike in the 1970s and 80s when the economy was far more concentrated and interest-rate sensitive.
• Economic models are usually right; commentators are usually wrong. Furman notes that serious models (Yale Budget Lab, CBO, investment banks) predicted the Liberation Day tariffs would shave roughly half a percentage point off growth—real money, but nowhere near the "recession is inevitable" rhetoric that dominated TV and op-eds.
• Monetary policy credibility has fundamentally improved since the 1970s and even since 2011. Inflation expectations stayed anchored during the 2021-22 spike partly because markets trusted the Fed to act, unlike the early 1980s when it took a brutal recession to restore that trust.
• People and institutions adapt in real time, which forecasters routinely underestimate. When the Strait of Hormuz closed, China cut oil imports by roughly 5 million barrels a day, capping global prices well below the $200-a-barrel doomsday scenarios; fuel efficiency gains since 1979 and 2011 also made the U.S. economy far less oil-sensitive than during prior shocks.
• Furman distinguishes "events" (tariffs, wars, Fed decisions) from "trends" (productivity growth, technology adoption): events can only subtract from GDP by disrupting existing capacity, while trends build new capacity gradually; meaning most one-off events skew negative but most long-run trends, including 100 years of steady 2% per-capita GDP growth, skew positive.
• On AI, Furman argues the truly optimistic scenario is modest: roughly half a point of added annual GDP growth, not the 5-10% some in Silicon Valley float. Growth that explosive would require impossible amounts of physical construction (data centers, housing) and would trigger Baumol's cost disease-style inflation in services; wrenching social change, not utopia.

Notable moments

• Thompson's opening riff: economists have gone from predicting "nine of the last five recessions" (Samuelson, 1966) to something like "100 out of the last two."
• Furman recalls Greg Mankiw and Paul Krugman both warning the Fed to stop hiking rates in December 2022; right before six more months of hikes with no recession.
• On fertilizer and the Iran war: sounding smart by citing an obscure supply chain (Strait of Hormuz fertilizer inputs) is "galaxy brain"; the real economist move is quantifying each step until the impact shrinks to near zero.
• Furman on oil prices during the Iran war: "If you had told me this set of inputs, I would have thought more like $150 a barrel than the $100-ish a barrel we are right now."
• Comparing 2011 to today: oil hit an inflation-adjusted $150/barrel after Obama's Libya strikes and stayed there for two years; yet nobody remembers it as a crisis.
• Furman on runaway growth predictions: "Do you realize how much change that implies?"; pushing back on Silicon Valley figures who'd shrug at 15% or 100% productivity growth.
• Closing exchange: both agree the reason AI-driven change might not be catastrophic is the same reason past doomsday predictions failed; flexible pricing, worker reallocation, and the economy's hundred-leg-stool structure.

Time saved: 55 minutes.