Is Deflation Actually Bad? It Depends Which Kind

Ask a working economist whether deflation is bad and the answer comes back fast: yes. Ask why and you'll usually get one of two examples: the Great Depression, or Japan's lost decades. Both happened. Both involved falling prices. Both were genuinely bad.

But here's the catch. The word deflation covers a second phenomenon that looks identical from twenty thousand feet — prices going down — and runs on a completely different mechanism underneath. In the late-19th-century United States, prices fell for roughly a quarter-century while the economy industrialized, real output expanded, and real wages rose. Same falling-price pattern. Opposite cause. Opposite consequences.

Conflating those two is the single most consequential category error in modern macroeconomics. It's the assumption underneath the Federal Reserve's 2% inflation target: when technology makes things cheaper, the response is to expand the money supply and keep prices drifting up rather than let them fall. And it's why falling prices set off alarm bells even when the economy is healthy.

This article walks through both kinds. The bad one deserves the fear. The good one doesn't. Telling them apart is the prerequisite for thinking clearly about money.

Start with the case for fearing deflation

Take the conventional view seriously first, because it isn't wrong — it's incomplete.

The economist Irving Fisher wrote his classic paper “The Debt-Deflation Theory of Great Depressions” in 1933, while the Great Depression was still unfolding around him. The mechanism he described is straightforward and grim.

Start with an economy carrying a lot of debt. A shock hits — a banking crisis, a stock market crash, a sudden tariff war, anything that drops demand. Prices start to fall. So far, so manageable.

But the debts don't fall with them. When the downturn cuts your hours and your paycheck shrinks, your house payment doesn't shrink with it — it's the same dollar amount it always was, now eating a bigger share of a smaller paycheck. Multiply that across millions of households and businesses, and people start selling whatever they can — dumping inventory, houses, stocks — to raise the dollars their loans still demand. Those forced sales push prices down further. Which shrinks incomes and makes the fixed debts heavier still. Which forces yet more selling. And so on.

That's the spiral. Falling prices → harder-to-service debts → forced asset sales → falling prices. The mechanism is real, the math is real, and it's exactly what played out from 1929 through the early 1930s. Real US output fell by roughly a quarter. One in four workers couldn't find a job.

Japan's “lost decades” after 1990 are the modern reference case. A real-estate and equity bubble unwound; debts taken on at peak asset values became crushing; consumer and corporate spending collapsed; deflation set in and stuck around for the better part of two decades.

Under these conditions, falling prices really do feed on themselves and drag the economy down. The fear is rational. Any honest treatment of deflation has to start here.

And then notice the case it doesn't fit

Now consider a different period.

From roughly 1873 to 1896, US wholesale prices fell at an average rate of about 1.5% to 2% per year. The fall lasted nearly a quarter-century. Prices in 1896 were materially lower than prices in 1873.

The economy was not collapsing. The opposite. Real GDP more than doubled over that window. The rail network was built out. Electricity arrived. Steel production scaled. Real wages rose.

Prices weren't falling because demand had collapsed. They were falling because the economy was getting better at producing things.

Same surface pattern — multi-decade falling prices — running on a completely different mechanism underneath. The cost of producing goods kept falling, because factories produced more per hour and new technology made old methods obsolete. Sticker prices fell with production costs. Workers got cheaper goods on top of higher wages.

This isn't an exotic claim. The economists Andrew Atkeson and Patrick Kehoe walked through 17 countries and 100 years of data in “Deflation and Depression: Is There an Empirical Link?” (2004) and found that 65 of 73 deflationary episodes weren't associated with depression at all. The Great Depression turns out to be the outlier, not the pattern.

The 19th-century US case is what productivity deflation looks like in the wild.

How to tell which kind you're looking at

Both kinds of deflation look the same from the outside — prices falling. What tells them apart isn't the falling prices. It's why they're falling.

Two signals will tell you which mechanism is running:

• Check real output. Is it growing or contracting? Productivity deflation comes with rising real output (the late-19th-century US grew while prices fell). Debt-deflation comes with falling real output (the early-1930s US contracted as prices fell).
• Check real wages. Are they rising or falling? Productivity deflation comes with rising real wages — the same paycheck buys more because the same labor produces more. Debt-deflation comes with falling real wages, often falling employment, and shrinking household incomes.

Apply both signals to the two reference cases:

• 1873–1896 US: real output rising, real wages rising → productivity deflation.
• 1929–1933 US: real output collapsing, real wages falling (alongside collapsing employment) → debt-deflation.

Same surface (prices falling). Different signal on every other variable. Different category of event.

The reason this matters: policy that's calibrated for one mechanism will misfire if it's the other.

The 2% target collapses the category

The Federal Reserve adopted an explicit 2% inflation target in 2012, formalizing a stance that had been operating informally since the 1990s. The target is positive specifically to keep a buffer above zero — to make sure the price level doesn't drift into deflation territory.

That logic is defensible only if productivity deflation is impossible or dangerous. It's neither. It happened — the late-19th-century US had it for more than two decades. And it isn't dangerous — Atkeson and Kehoe found most deflations came and went without a depression.

The consequence: when technology lowers the cost of producing a laptop, a TV, a phone call, or a streaming subscription, the Fed's 2% target keeps the aggregate price level drifting up instead of down. The productivity dividend gets absorbed into monetary expansion rather than paid back to dollar holders.

You can actually see it in a few places — the ones where the gains were too big to hide. Solar-panel cost per watt has fallen roughly 90% since 2010. Sequencing a human genome has dropped from millions of dollars to under a thousand.

But those are exceptions. Take groceries: we grow far more food per acre than we did decades ago, so it costs steadily less to produce — and your grocery bill went up anyway. The productivity gain is real; it just got buried under a rising price instead of showing up as a lower one. Real, but invisible. A companion piece lays out the data: the sectors where prices barely moved, the essentials that climbed instead, the dollar losing value faster than technology made things cheaper.

Whether that trade-off is defensible is a separate debate. The narrower point here is that the trade-off is built on a category collapse: it treats all deflation as if it were the 1933 kind, even though the 1873 kind is the more common one.

But what about Japan?

A careful reader will raise Japan, and the reader is right to.

Japan's deflationary stretch from the mid-1990s into the 2010s is the most-cited modern example of “bad deflation.” It was real and it lasted longer than most people expected. But it wasn't productivity-driven. The 1989–1990 real-estate and equity crash left borrowers holding assets worth 50–80% less against debts that didn't move. Aggressive central-bank response prevented the 1930s-style immediate contraction, but the underlying balance-sheet damage took two decades to work through.

The economists Michael Bordo and Andrew Filardo, in “Deflation in a Historical Perspective” (2005), reviewed the long historical record and proposed a three-part taxonomy: good deflations driven by productivity gains, bad deflations driven by recessions, and ugly deflations — the rare Fisher spiral, where falling prices and falling demand feed each other. Their empirical conclusion echoed Atkeson and Kehoe's: most historical episodes are not the ugly kind. Japan was bad bordering on ugly. The 19th-century US was good.

A bad case doesn't invalidate the good case any more than the Great Depression does.

What this changes about reading the news

Once the two kinds are separable, a lot of monetary commentary reads differently.

When you see “deflation is dangerous,” ask which kind. If the case rests on 1933 or Japan, fair — that mechanism is real. If the case rests on “any falling price is a warning,” it's overreaching — stretching the rare bad case to cover all the rest.

When the Fed defends the 2% target as insurance against deflation, ask which kind of deflation the insurance is buying. The buffer that prevents a Fisher spiral also prevents the productivity pass-through. Same policy, two different effects.

When prices fall in some corner of the economy — a streaming service cuts its price, a new gadget gets cheaper, shipping costs drop back to normal — notice what doesn't follow. No spiral. No crisis. Your money just goes a little further. Notice which kind of deflation that was.

A monetary system that can tell the two kinds apart can leave productivity-driven price declines alone and only act when the debt-deflation signals fire. A monetary system that can't has to absorb every price decline with offsetting expansion, on the assumption that the next one might be the dangerous kind.

Back to the clock

Bitcoin's fixed supply doesn't cause deflation. It just removes the offset. Productivity-driven price declines pass through to the holder of the unit rather than being absorbed into monetary expansion.

That's what each item card on Satoshi's Clock is tracking. As productivity gains continue to compress the cost of producing everyday goods, fewer satoshis are needed to buy them. The deflation the clock is denominated in is the 1873 kind, not the 1933 kind.

Tell the two apart, and a lot of arguments about “but deflation!” stop applying.