# Part of Europe's Decline Was Calculated, Not Measured Source: Realpha Blog (blog.getrealpha.com) Original article and charts: https://blog.getrealpha.com/en/blog/oddlots-2026-10-08-what-everyone-gets-wrong-about-the-economic-proble/ > Notes on Bloomberg Odd Lots (2026-10-08): economist Dominik Leusder takes apart the Draghi report's headline trend line — Europe's output per hour falling from 90% of the US to 70% — and shows how differences in how the US and Europe deflate tech prices compound into that slope, while Germany's loss of over a million manufacturing jobs since 2021 remains real. Educational notes only; no investment advice, no price targets, no stock recommendations. Published: 2026-10-08 Locale: en Tags: European economy, productivity, purchasing power parity, China trade shock, Odd Lots ![Two brass rulers with different scales lie end to end on a long statistics office table at dusk, pressing down a few line charts, with filing cabinets receding down a corridor](/covers/oddlots-2026-10-08-what-everyone-gets-wrong-about-the-economic-proble-cover.png) > The Hu people snap a bone, the Yue cut the arm, the people of the central states smear blood on the lips. The means differ; as pledges of trust they are one and the same. > > —— *Huainanzi*, "Equalizing Customs" (Western Han; translation mine) In the 2026-10-08 episode of Bloomberg's *Odd Lots* (recorded September 30), hosts Joe Weisenthal and Tracy Alloway talk with Dominik Leusder, an economist and policy analyst formerly at the LSE, about whether Europe is actually in decline. Leusder's claim: the headline slope in the November 2024 Draghi report — European output per hour falling from over 90% of the US level to around 70% since 1995 — owes a good chunk of its shape to the fact that the US statistical agency adjusts tech prices for quality more aggressively than European offices do, worth maybe 0.2 to 0.3 percentage points a year, compounding into that line. Switch to a single-year snapshot for 2024 and the US sits slightly above the Western European aggregate on output per hour, and that edge disappears once you drop the top 10% of the income distribution. What this argument kills is the *trend line*; it does not absolve Europe of its own problems, and Leusder is blunter than most about the biggest one: Germany has lost over a million manufacturing jobs since 2021. ## What I used to think Europe was Short hours, heavy regulation, no large tech champions of its own, therefore falling steadily behind the US. I could recite the numbers too — Jamie Dimon has said it repeatedly on his visits: European GDP was 90% of the US ten or fifteen years ago, now it's 70%. The Draghi report draws that ratio as a line and says it keeps falling unless Europe does X, Y and Z. The long French lunch and the August holiday I filed under cope: you lost on productivity, so you talk about quality of life. What I never asked was whether that line was measured or computed. ## The slope partly comes from how each side computes inflation Leusder goes straight at the denominator. To compare two economies you first have to strip out exchange-rate swings and price-level differences — that's purchasing power parity, a ratio of how many French euros buy a dollar's worth of consumption. The Draghi measure fixes prices at a benchmark year (2010 international dollars) and then extrapolates between benchmarks using the real growth rates that each country's national accounts produce. The trouble sits in how those real growth rates are built. The US Bureau of Economic Analysis applies quality adjustment to tech goods and services more aggressively than European offices do. Leusder uses the MacBook in front of him: same nominal price, better specs, recorded as a fall in the quality-adjusted price. The same nominal spending therefore shows up as more real growth in the US than in the EU, because measured inflation in that sector is lower. Some studies put the gap at 0.2 to 0.3 percentage points a year, and it compounds — over three decades that is the wedge. PPP series built from relative inflation rates instead of national-accounts real growth do not show the decline at all. ![Two lines start from the same point and separate year after year: the upper line is the United States and the lower line is Europe, with the wedge-shaped gap between them widening each year, annotated as a difference of 0.2 to 0.3 percentage points per year.](/figures/oddlots-1008-two-rulers-wedge-en.svg) His conclusion is deflating and honest: we do not have a measure that can compare levels over time. So the downward trend that gets passed around as a stylized fact has no solid data behind it. What we can do is compare snapshots. For 2024, the US sits a little above the Western European aggregate on output per hour, below Germany and the Nordic and low countries; strip out the top decile of income and the aggregate advantage goes; and the bottom half of the US distribution is worse off than any single European country except Portugal. ![Three rows of markers on one horizontal scale show the US position sliding left or right as the measure changes: slightly higher overall, overlapping Western Europe once the top 10% is removed, and falling to the far left for the bottom half of the population.](/figures/oddlots-1008-snapshot-2024-position-en.svg) Another metric sharpens it. Leusder prefers actual individual consumption, because it counts in-kind benefits like publicly funded healthcare, which makes it easier to compare than income. Adjust for hours worked and the US looks unremarkable. Healthcare is where he spends the most time: the US spends at least five percentage points more of GDP on health than most European countries, and his own model says it spends roughly 50% more per person than a country at that consumption level would be expected to — between $1.7 and $1.9 trillion a year in excess. Prices per unit of care are high, utilization is poor, outcomes (life expectancy, mortality for certain procedures) are worse, and hospitals, insurers and administrative middlemen take a cut along the way. So when PPP marks US output down, part of what it marks down is rent. ![Two horizontal bars of different lengths compare health spending as a share of GDP, with the extra stretch on the US bar marked as excess, and three cells below showing higher prices, lower utilisation and worse outcomes.](/figures/oddlots-1008-health-spend-rent-en.svg) Joe's reaction stuck with me: hearing all that, he said he came away more impressed with the rest of the US economy for performing while carrying that albatross. Tracy's line was better still — "the value of my leisure is above zero." Take that as a premise and the choice set changes shape: hours fall because productivity gains get converted into leisure, and European employment rates have held up, which is what you'd want to see if this were a choice rather than a labor-market failure. Leusder cites work showing the causation runs both ways, hours down raising productivity and productivity pushing hours down. ## The arithmetic was wrong; the illness is real Say a stock you own is down 40% and the news says the industry is in structural decline. How do you separate noise from structure? This episode gave me two layers to work with. The first is measurement: who computed that number, which base year it uses, whether both sides use the same units. The second is physical: are jobs disappearing, are lines moving, are customers buying from someone else. Leusder splits it exactly this way — he dismantles the trend line and then makes Germany's problem sound worse than the consensus does. Over a million manufacturing jobs gone since 2021, coinciding with the shift in Chinese industrial policy and the energy crisis. He adds a detail I wrote down: chemicals, the sector most exposed to both energy costs and Chinese competition, has shed the fewest jobs — because firms are hoarding labor they cannot replace, which makes the figures look better than the shock. ![Two stacked timelines: on the top layer, the measurement signal arrives early but is drawn as a dashed line because it can be recomputed and overturned, while on the bottom layer the physical signal arrives late yet is drawn solid and cannot be shifted.](/figures/oddlots-1008-two-layers-caliber-physical-en.svg) The physical layer moves slower than the measurement layer, and it cannot be deflated away. On China he points to Brad Setser's work: export price deflation after the housing bust, a thicker web of manufacturing subsidies, a third of the labor force in the gig economy, and more than half of GDP growth possibly resting on net exports. Europe's response is slow enough to be painful — the Industrial Accelerator Act, proposed this year, under review, perhaps in force early next year, with "buy Europe" provisions written so that in practice only Chinese imports fail the hurdles. Twenty-seven member states, unanimity or qualified majority voting, domestic coalitions underneath: Leusder calls it a joint decision trap that regresses to the lowest common denominator. He quotes his friend Adam Tooze on why it took so long — Europeans are "the Taliban of neoliberalism" on trade. For contrast: in late July the Commission added several Chinese entities to its sanctions list, and the next day China put a batch of major European industrial firms on its own. His worry is that Europe's willingness to escalate will outrun its ability to absorb retaliation. ![Two bars running in opposite directions: chemicals has the longest energy and China exposure but the shortest job losses, and the right side points to a box labelled labour hoarding.](/figures/oddlots-1008-chemicals-hoarding-en.svg) There was one coincidence during the recording worth repeating. Tracy was mid-sentence arguing that anyone bullish on AI-driven productivity should be bullish on Europe, since no region is more motivated to squeeze extra output out of each working hour, when a red headline crossed the terminal: BMW targets shedding a fifth of manager roles with the help of AI. Both sides of the coin in one line. Leusder is relaxed about Europe's position in AI — he thinks Europe will not build a frontier model soon and should not try, and that near-frontier models like Mistral cover most European corporate needs. The risk he names is sovereignty: Mistral ships models on Chinese weights, a Chinese chassis, because Europe lacks the compute to train its own. What I have not settled is Tracy's closing objection. The whole arrangement where leisure is valued above zero grew up inside an era of absolute trade gains; with security, critical inputs and defense industry back on the table, priorities shift. I find Leusder's arithmetic convincing, but his conclusion rests on the premise that the point of the comparison is welfare. Change the purpose to withstanding blockade and retaliation and absolute output becomes the answer again — and he concedes that by its own standards Europe has done poorly on macro policy for a decade and a half. ## Further reading - Bloomberg *Odd Lots*, 2026-10-08, Joe Weisenthal and Tracy Alloway with Dominik Leusder - The Draghi competitiveness report, commissioned by the European Commission, published November 2024 — the source of the contested trend line - Brad Setser's work on China's reliance on net exports and the effects in Europe - Adam Tooze on European attitudes toward trade - Uwe Reinhardt's research on US healthcare prices per unit of care, cited in the episode ## One Thing to Take With You **A multi-year trend line is computed, not measured. Before comparing how two things have moved over time, check that both sides use the same ruler; if you cannot check, compare single-year snapshots and skip the line.** Something I've tried: write down the last "A is worse than B" sentence you said out loud — "we spent more this month than last," "our department is less efficient than theirs," "my kid is weaker at math than his classmates." Then add one line each: who produced that number, in what units, and whether the two sides use the same units. If they don't line up, rewrite the sentence as a snapshot — this month was X, last month was Y — and drop the trend.