Over the past 15 years, the Federal Reserve has made two major policy errors. In 2008, it adopted an excessively contractionary monetary policy, worsening the Great Recession. In 2021 and 2022, it adopted an excessively expansionary monetary policy, adding to the recent inflation. Adopting three key policy reforms would greatly reduce the risk of repeating these mistakes.

During some periods, a discretionary approach to monetary policy can work tolerably well, as it did during Alan Greenspan’s 19-year tenure at the Fed. When the economy is hit by an unforeseen shock, however, reliance on the discretion of policy-makers has often resulted in significant mistakes. We need a policy regime based on credible and transparent objectives and a reliable method for implementing those policies.

Toward that end, the Fed should first replace its target for the average inflation rate over time (“inflation-targeting”) with a target for the total-dollar size of the U.S. economy — a figure also known as “nominal spending” or “nominal gross domestic product” (NGDP). First, the Fed might aim for 4 percent annual growth in total spending. Second, the Fed should commit to level-targeting, which means returning the economy to the target after periods when it has undershot or overshot the growth-rate target. And finally, policy should be guided by market forecasts, not the hunches of policy-makers.

Congress has mandated that the Fed seek three things: stable prices, high employment, and moderate long-term interest rates. Because these goals sometimes conflict, the Fed has adopted a flexible average-inflation target of 2 percent, with some variation allowed as needed to achieve its employment goals. Thus the Fed may allow brief periods of above 2 percent inflation during supply shocks such as the Ukraine war.

Over the past 18 months, though, the U.S. has experienced much higher inflation than any plausible interpretation of the flexible inflation target could justify. The Fed is not allowing this high inflation in order to address an unemployment problem; indeed, in 2022, many companies were short of workers. Clearly, something has gone wrong.

A look at nominal spending clarifies where the Fed has gone off course. Nominal spending includes both growth in real output and inflation. Because real economic growth tends to average 1.5 to 2 percent per year in the long run, nominal spending should grow by 3.5 to 4 percent per year on average for inflation to remain close to 2 percent.

Instead, the Fed allowed nominal spending to grow more than 9 percent over the past year, which largely explains why inflation has stayed so high. The media often focus on prices of particular goods and commodities such as oil and natural gas, but these prices are volatile and explain only a modest portion of the inflation overshoot.

If spending growth is steady, a negative supply shock will lead to a temporary rise in inflation and a reduction in economic growth. Because monetary policy was too expansionary in 2021–22, however, the rise in spending has increased the prices of many goods and services even though supply shortages have not reduced their availability.

The Fed does not currently target nominal spending. But as a practical matter, the Fed cannot achieve its goal of low inflation and high employment without stable nominal-spending growth. Whenever spending growth is far below 4 percent, unemployment will rise sharply. Whenever it is far above 4 percent, inflation will rise. Because spending growth is the sum of inflation and real GDP growth, it is the best single indicator of the Fed’s performance.

A policy target of 4 percent spending growth is much more credible and transparent than a vague policy aimed at 2 percent inflation and high employment. How should high employment be defined? The Fed doesn’t tell us. How does it balance the goals of avoiding high inflation and encouraging high employment? The Fed doesn’t tell us. What does it do if it misses a policy goal? Again, the Fed doesn’t tell us.

For a brief time in 2020, it looked as if the Fed was answering some of those questions. After it announced a 2 percent “flexible average-inflation target,” many economists (including me) assumed this meant that if inflation overshot 2 percent for a few years, the Fed would try to undershoot the target for the next few years, to hit the average in the long run.

But when an inflation overshoot actually occurred in late 2021, Fed chairman Jay Powell indicated that the Fed had no intention of undershooting in order to return to the average. As a result, inflation will not average 2 percent in the long run: “Flexible” defeated “average.” A nominal-spending target would be clearer and more transparent than even a well-designed average-inflation target, not least because it would not require as much flexibility. Total spending does not need to adjust to temporary supply shocks.

Two questions may arise at this point. First: Why should we target a positive rate of change, instead of zero, for either prices or nominal spending? Here, it’s worth thinking about the Fed’s third mandate: “moderate long-term interest rates.” In countries where nominal-spending growth from inflation averages close to zero, such as Japan, nominal interest rates often turn negative. While this sort of monetary policy is feasible, a variety of practical factors make it unlikely to be adopted in the U.S.

Second: Why not let bygones be bygones? If nominal spending grows too slowly or too quickly in one period, why not just aim for 4 percent growth going forward? It turns out that level-targeting has two surprising benefits, which can be illustrated by looking at the major policy errors of 2008 and 2021.

Let’s start with the more recent case, the Fed’s excessively expansionary policy of 2021–22. By late 2021, it was clear that spending had fully recovered from the Covid slump and was strongly overshooting the pre-2020 trend line. Unfortunately, despite this overheating, the Fed kept its interest-rate target at close to zero, where it stayed well into 2022.

Now assume that this policy misjudgment happened while the Fed was trying to keep spending on a stable long-term path. Market participants would have known that any overshoot would lead to sharply higher interest rates in 2022, intended to bring spending growth back to the 4 percent target. Those market expectations would have immediately pushed up longer-term interest rates even if the Fed had (foolishly) delayed raising its short-term rate target until early 2022. In a sense, markets would have been correcting an oversight by policy-makers.

Economic commentators often speak of “long and variable lags” in monetary policy. But in some respects, monetary policy works with long and variable leads. Anticipated policy changes can affect the economy long before they are implemented; what matters is when investors begin anticipating the policy change. The expectation that the Fed will sharply raise rates in the future can cause long-term interest rates to rise immediately, which will tend to slow important sectors such as housing.

The 2008–09 recession illustrates another major advantage of level-targeting. By December 2008, nominal interest rates had fallen to close to zero. These low rates did little to help the economy, since the public also expected very weak growth in spending. What matters for monetary policy is the level of interest rates relative to equilibrium interest rates. And with rates already low, the Fed had little room to cut them.

If the Fed had committed to return spending to the previous trend line after the sharp decline of 2008–09, however, then the same near-zero interest rates would have constituted a much more expansionary policy. A commitment to eventually raise nominal spending back to the old trend line would boost the equilibrium interest rate, which would make any given policy rate more expansionary. When interest rates are stuck at zero, monetary policy is much more effective with level-targeting of NGDP than with a bygones-be-bygones approach to growth-rate targeting.

Instead, nominal spending remained weak in 2009 because investors and the broader public did not expect a compensatory policy in the future. We shifted to a new and lower trend line for nominal spending. Ironically, the lessons of this policy mistake were so well learned that the Fed erred in the opposite direction in 2021. It tried hard to bring the economy back to the previous trend line and in doing so overshot its target.

This is why it’s so important that targeting be symmetric, making up for both undershoots and overshoots. If the Fed tries to offset some mistakes and not others, then investors will be flying blind. No one knows where spending will be three or five years from today, making it much harder for businesses to plan for borrowing, investing, and negotiating new wage contracts.

The Fed should adopt a better target. But it should also adopt new methods of hitting it.

Control of short-term interest rates and adjustments in the quantity of money are powerful tools that the Fed could use. The problem is how best to use those tools. Where should the Fed set interest rates and the money supply? I have argued that the ideal policy would set interest rates and the money supply at a level that a consensus of market participants expects to lead to hitting the spending target. But how do we know how much spending growth the market expects?

Financial markets provide some hints. Financial assets such as inflation swaps and indexed bonds provide indications of market expectations about inflation. A crude estimate of the other component of nominal-spending growth — the growth of real output — can be generated by looking at financial-asset prices that are highly cyclical. These include stock prices, commodity prices, high-risk bond spreads, and the slope of the yield curve. While these market indicators are far from perfect, they do give some indication of the expected path of NGDP; indeed the Fed already uses them to some extent.

But we can and should do better. I have proposed a sort of financial-market guardrail for monetary policy, which would be triggered whenever policy had drifted too far off course. The Fed could, for instance, create futures contracts linked to nominal-spending growth. If the target were 4 percent, then the Fed could announce that it would take a long position on anyone selling futures contracts at 3 percent growth and a short position on anyone selling at 5 percent.

Under this sort of policy regime, the Fed would profit from buying 3 percent NGDP contracts whenever actual NGDP growth exceeded 3 percent, and it would profit from shorting 5 percent NGDP contracts whenever NGDP growth fell short of 5 percent. If investors saw that monetary policy was drifting far off course and expected NGDP growth to fall short of 3 percent or exceed 5 percent, they could profit by trading these contracts with the Fed. If almost all investors were taking short positions, it would be a signal to the Fed that policy was too tight, and vice versa. These market guardrails would be analogous to the beeping sound a modern car makes when it’s about to back into a wall.

The purpose of these improvements to monetary policy is to establish a necessary condition for economic success. The economy never does well with highly unstable spending levels, and it generally does fairly well when spending grows at a steady rate (as during 2017–19, when it hewed pretty closely to 4 percent a year).

Staying on that growth path would be easier if the Fed took it as a goal, which must include committing to return to it when it veers off course. It would be easier still if it harnessed market forces to help chart its course.