Marked to Model

The Fed's instrument panel has no gauge for fiscal risk.

The FOMC is meeting right now (July 28–29) in DC to decide whether or not to hike rates. Before every FOMC meeting in DC, the Fed staffers run different policy choices through a whole stack of models that translate new data into forecasts and different policy options into outcomes.

Central bankers fly on their instruments. When Fed officials stated that quantitative easing carried no risk for inflation and that the balance sheet posed no fiscal risks, they were largely relying on this instrument panel. These instrument readings are highly noisy, and Fed staffers are fully aware of this. Macro-economic forecasting is not for the faint of heart. The problem wasn’t simply that the instruments gave noisy readings. Some of the instruments needed for the post-2020 economy were not visible to the flight crew.

Inside virtually all of the models used by central banks in advanced economies, fiscal policy is assumed to be passive. What does that mean? The fiscal rule quietly adjusts tax rates in the background whenever the debt-to-GDP ratio drifts from its target. Basically, every fiscal stimulus is fully funded and paid for by future tax increases or spending cuts. The government runs deficits today. The model starts legislating surpluses tomorrow.

The model is incapable of asking what happens if these projected government surpluses in the future do not arrive. Anyone who has ever looked at the CBO’s projections knows this assumption is suspect.

When you open up the box to look inside these models, you realize that the government’s cost of funding does not respond to fiscal policy at all. Government debt is completely safe. The debt is never marked to market in response to fiscal news. There is no fiscal risk premium. To be fair, all of this is internally consistent. Remember: every government spending increase is already fully paid by taxpayers. Bond investors can sleep on the job.

In these models, there is no downward-sloping demand curve for Treasurys either. Governments face a perfectly elastic demand curve for bonds at the risk-free rate. There are no convenience yields — the premium on Treasurys above AAA-rated corporates, compensation for their safety and liquidity. Whether government debt is at 60% of GDP or 150% of GDP doesn’t really matter in these models. They’d be rolling over the debt at pretty much identical yields. Instead, the government’s cost of funding is fully determined by the future path of interest rates set by the central bank.

It’s easy to understand the appeal of this class of models for Fed staffers: it absolves them of the need to model the fiscal decisions of Congress — the same Congress that also exercises oversight over the Fed. And obviously, given the Fed’s narrow price stability mandate, they probably want to stay as far away as they possibly can from modeling (or even thinking about) active fiscal policy.

This class of models was stress-tested in March 2020. Over the course of the next year, the U.S. federal government spent an additional 20% of GDP, north of $5 trillion, with no offsetting revenue coming down the pipeline. Ask these Fed models what an unfunded expansion of that size does, and the models will start legislating future surpluses to fully offset the effect of the COVID spending spree.

To be fair, this is not really a built-in forecast of what the Fed thinks fiscal policymakers will actually do. It is rather a technical requirement that needs to be met in order for an equilibrium to exist in this class of models where the Fed is actually in the driver’s seat, i.e., the Fed is fully in control of inflation. Economists call this regime monetary dominance. But the model nevertheless values government liabilities as though the required fiscal backing will arrive.

In the model, when COVID hits, the debt-to-output ratio initially increases but then gradually goes back down as the government starts to produce surpluses. The government’s funding costs are unchanged, and inflation doesn’t budge.

Compare that scenario to what actually happened. The yield increases and surprise inflation in the aftermath of the COVID-19 pandemic imposed huge haircuts on outstanding Treasury debt. This was arguably the largest revaluation of US government debt since the 1940s. In the real world, bondholders are being forced to bear a big part of the burden, not taxpayers.

Now, most central bankers privately agree that COVID was an unfunded spending shock. I mean, how could they not? There’s just no evidence that the federal government has any plans to increase surpluses in the near future to offset COVID deficits. When I suggested at a conference of central bankers that maybe inflation and yield increases were actually the mechanism through which government debt was marked to market in light of the unfunded fiscal expansion, I was told “that’s just not the way central bankers think about this.” I replied by asking: what’s your model of government debt valuation? How does the debt get marked to market when there is an unfunded increase in government spending, like COVID? No answer was forthcoming. In this class of macro models, government debt is safe by fiat. The debt is marked to model.

It’s almost as if central bankers assume that bond market investors are not clever enough to notice that those surpluses are not forthcoming — and so never change their assessment of what the debt is really worth. That seems like a bizarre view. Bond investors did notice. That is what the COVID-era announcement-day yield jumps were.

The Fed is not alone in flying with gauges missing from the panel. You can cross the Atlantic and take a look at the class of models used by the ECB in Frankfurt. Each of the euro area’s twenty-one fiscal authorities is assumed to dutifully stabilize its own debt-to-output ratio through a fiscal rule. In the case of the ECB, this approach requires a certain doublethink, because the ECB is one of the world’s leading authorities on the riskiness of sovereign debt, having presided over the European sovereign debt crisis in the early 2010s. And the ECB has repeatedly rolled out instruments and measures to counter sovereign risk, such as the Outright Monetary Transactions in 2012 and the Transmission Protection Instrument in 2022, both designed to cap sovereign bond yields.

What would a more informative instrument panel look like? It should, at the very least, include models in which inflation occasionally does the work that surpluses don’t, and models in which bond risk premia respond to fiscal shocks. Regime-switching frameworks exist in which fiscal policy is sometimes active and monetary policy sometimes passive, and then inflation does the adjusting that future government surpluses will not. Bianchi and Melosi presented exactly such a model at Jackson Hole in 2022 — at the Fed’s own conference — arguing that a fiscal component of the post-pandemic inflation could not be cured by rate hikes alone. Once you have active fiscal policy, the government’s cost of funding will start to respond to fiscal choices.

In August 2016, on the eve of the Fed’s annual Jackson Hole conference, Kevin Warsh published an opinion piece in the Wall Street Journal under the headline “The Federal Reserve Needs New Thinking.” Warsh had served as a Fed governor from 2006 to 2011, so this was not an outsider lobbing grenades. His diagnosis: the conduct of monetary policy had been deeply flawed, and the deeper problem was intellectual. He identified groupthink within what he called the academic economics guild as a key culprit. And rather than confronting its forecasting record, “the guild tightens its grip when it should open its mind” — to new data, new analytics, and especially new economic models.

And let’s face it: there is enough model uncertainty in macroeconomics to warrant humbly entertaining more than one model when evaluating policy choices and forecasts. As it happens, the author of that opinion piece is chairing the FOMC meeting this week. We are about to find out whether the guild’s grip has loosened.