AI, Productivity, and Why Market Monetarists Think Nominal GDP Targeting Fits the Moment

It’s early 2026, and the conversation around artificial intelligence is as loud as ever. Leaders from companies like OpenAI, Anthropic, and Nvidia talk about AI agents entering the workforce, intelligence becoming dramatically cheaper, and potentially big shifts in white-collar jobs over the next few years. Interestingly, bond markets have often reacted to major AI announcements since 2023 by pushing yields lower—as if investors see a productivity surge on the horizon that might not bring the usual inflationary pressures.

Yet when you look at the broader economic numbers through 2025, the picture is more modest. Overall U.S. labor productivity grew around 2.1% for the year, with quarterly swings but no clear economy-wide breakout yet. AI shows up strongly in labs, pilot projects, and specific tasks—think 14-55% gains in areas like coding, customer service, or consulting—but it remains largely invisible in the national accounts, much like personal computers did in the 1980s.

As someone who follows market monetarist ideas, I’m not interested in either hyping AI to the moon or joining the chorus of skeptics. The more useful question is how monetary policy can best support an economy undergoing a potential positive supply shock from smarter tools. Market monetarism, which emphasizes stabilizing the growth path of nominal GDP (NGDP), offers a practical framework for handling exactly this kind of change. Here’s why it seems especially relevant right now.

A Quick Primer on Market Monetarism

Market monetarism, associated with economists like Scott Sumner and David Beckworth, argues that central banks should aim for steady growth in nominal spending across the economy—essentially targeting a predictable path for NGDP (the total dollar value of goods and services produced, before adjusting for inflation).

The appeal is that financial markets process information quickly. If we had liquid NGDP futures or similar prediction markets, they could give policymakers a timely read on whether money is too tight or too loose, without relying solely on lagging indicators like unemployment or inflation.

Traditional inflation targeting worked reasonably well during the relatively stable period from the 1990s through the 2010s. But it can struggle with large supply shocks. A big productivity boost from AI should show up as faster real growth and, all else equal, softer price pressures. Under a strict 2% inflation target, however, a central bank might feel compelled to tighten policy to prevent inflation from drifting too low—potentially choking off demand just as the economy’s productive capacity is expanding. NGDP targeting avoids that trap: it lets real output rise, allows prices to adjust downward if productivity warrants it, and keeps overall nominal spending on a steady trajectory. In short, it doesn’t mistake “good” disinflation for a problem.

AI as a General-Purpose Technology

AI stands out because it reduces the cost of cognition—one of the most fundamental inputs in modern economies. It speeds up code writing, legal research, medical image analysis, creative brainstorming, and supply-chain optimization. Early studies and company reports show meaningful task-level gains, sometimes in the 30% range for specific uses, though scaling those across entire organizations and the broader economy is proving slower and messier than some expected.

If these gains compound and spread, sustained real GDP growth of 3–4% (or higher in optimistic scenarios) becomes more plausible than the 1.5–2% trend many had grown accustomed to. From a market-monetarist viewpoint, this is clearly positive. As Sumner has pointed out, under inflation targeting, stronger real growth from AI would tend to push NGDP growth higher if the central bank holds the inflation line—say, 5% real growth plus 2% inflation equals 7% NGDP. The economy wouldn’t suddenly lack “demand”; nominal spending would naturally expand to match the new level of abundance, and velocity of money would likely pick up too.

A classic market-monetarist (or productivity-norm) response to a positive supply shock is gradual, benign disinflation: prices ease while demand remains stable. Consumers benefit from higher real purchasing power, and workers’ sticky nominal wages stretch further. The idea, drawing from thinkers like George Selgin, is to let productivity improvements flow through to lower prices rather than forcing the central bank to manufacture scarcity to hit an inflation target. If AI really does make certain kinds of intelligence much cheaper, letting the price level reflect that reality makes sense.

The Risk of Policy Missteps

Now consider what could go wrong under the current inflation-targeting approach. As AI drives marginal costs down in more sectors, CPI might drift toward 1% or lower. Some policymakers or commentators could interpret that as a “deflationary spiral” and respond by tightening—raising rates or scaling back accommodation. The result: NGDP growth falls short, real wages rise too sharply relative to productivity in the short run, and we layer avoidable cyclical unemployment on top of the structural shifts AI will bring.

History offers parallels. The 19th century saw strong real growth alongside periods of gentle deflation, fueled by technologies like railroads, electricity, and steel. It wasn’t economic disaster; it was broad progress. Today’s central banks, fixated on a 2% floor for inflation, might inadvertently restrain that kind of expansion.

NGDP-level targeting handles this more gracefully. Markets would signal the productivity surge quickly through asset prices and forward-looking indicators. The central bank’s job simplifies to keeping nominal spending on its pre-announced path—no dramatic over- or under-reactions. Labor markets can reallocate more smoothly because overall demand doesn’t collapse. Structural change from technology still happens (that’s creative destruction), but without the added pain of demand-shortfall recessions.

Jobs, Displacement, and Distribution

AI will displace some tasks and roles—faster in some cases than new opportunities emerge. That’s familiar from past technological shifts. Yet there’s also a Jevons-like effect: cheaper intelligence can increase overall demand for complementary human skills, such as judgment, creativity, empathy, and handling genuine uncertainty. Current large language models are impressive pattern-matchers, but they’re not yet reliable oracles for causal reasoning in novel situations. Many experts see them augmenting professionals (radiologists, lawyers, programmers) more than fully replacing them in the near term.

The market-monetarist perspective doesn’t default to universal basic income proposals or heavy-handed regulation as the first response. Instead, it stresses maintaining stable NGDP growth so that rising real wages and any gentle price declines help diffuse the benefits widely. If technological change shifts labor’s share of income, that’s primarily a question for tax and transfer policy—not something monetary policy should try to override by restricting accommodation of higher potential output.

What Markets Are Pricing

The bond market’s reaction to AI developments has been noteworthy: yields have sometimes fallen following breakthroughs, even amid growth optimism. Investors appear to be betting on abundance and possibly lower neutral rates over time, rather than immediate scarcity-driven inflation. Market monetarism puts a lot of weight on these forward-looking signals, preferring them to rigid rules based on backward-looking relationships like the Phillips curve.

Silicon Valley forecasts may still be too aggressive on timelines—aggregate productivity data through 2025 show only modest AI fingerprints so far—but the underlying direction looks promising. The bigger near-term risk may not be mass technological unemployment, but central banks overreacting to the very disinflation that abundance would naturally produce.

A Framework for the Intelligence Era

AI won’t eliminate scarcity anytime soon. Constraints around energy, compute hardware, data quality, regulation, and integration will likely create bumps—including occasional inflationary pressures from supply bottlenecks. A good monetary framework needs to handle both positive and negative shocks. NGDP targeting along a clear, level path does that by letting markets guide expectations while the supply side works its magic.

In the end, an intelligence explosion that expands the economy’s potential isn’t a headache for monetary policy—it’s an opportunity. The challenge is ensuring central banks don’t stand in the way by fighting the symptoms of progress.

What are your thoughts? Do you expect NGDP targeting to gain traction if and when AI-driven productivity gains become unmistakable in the official stats? Or are there practical hurdles I’m underestimating? I’d love to hear perspectives in the comments—the future of abundance depends on getting the policy basics right.