Average Inflation Targeting
“To achieve buy-in by households and businesses, a comprehensible, credible, and actionable makeup strategy will need to be followed by years of central bank policy consistent with that strategy.” FRB Chair Jerome H. Powell, Opening Remarks at the Conference on Monetary Policy Strategy, Tools and Communications Practices,” June 4, 2019.
The Federal Open Committee’s first-ever comprehensive monetary policy review looks to be coming to an end. Since the announcement on November 15, 2018, the Fed has focused on strategies, tools, and communications practices, and engaged the public through numerous Fed Listens events, including a conference at which invited experts proposed new approaches (see our earlier post). It also has held extensive internal discussions.
Following COVID-19 delays, policymakers are now signaling that they are nearing a conclusion. At its July meeting, the FOMC discussed potential changes to its Statement on Longer-Run Goals and Monetary Policy Strategy—the “foundation for the Committee’s policy actions”—with the aim of finalizing those changes soon. And, Chairman Powell is scheduled to speak this week about the “Monetary Policy Framework Review” at the annual Jackson Hole Economic Policy Symposium.
Perhaps the most important issue on the review agenda is the FOMC’s inflation-targeting strategy. Since 2012, the FOMC has explicitly targeted an inflation rate of 2% (measured by the price index of personal consumption expenditures, PCE). A key objective of FOMC strategy is to anchor long-term inflation expectations, contributing not only to price stability, but also to “enhancing the Committee’s ability to promote maximum employment in the face of significant economic disturbances.” Yet, since the start of 2012, PCE inflation has averaged only 1.3%, prompting many policymakers to worry that persistent shortfalls drive down expected inflation (see, for example, Williams). And, with the Fed’s policy rate now back down near zero, falling inflation expectations raise the expected real interest rate, tightening financial conditions and undermining policymakers’ efforts to drive up growth and inflation.
While the framing of the review implied little desire to alter the 2% level of the inflation target, it left open the possibility of other changes. In this note, we discuss one alternative to the current approach that has gained wide attention: namely, average inflation targeting. The idea behind average inflation targeting is that, when inflation falls short of the target, it creates the expectation of higher inflation. And, should inflation exceed its target, then it would reduce inflation expectations. Even when the policy rate hits zero, the result is a countercyclical movement in real interest rates that enhances the effectiveness of conventional policy.
A central bank implementing an average inflation targeting framework chooses a period of time—call it the averaging window—over which it measures inflation with the goal of meeting a numerical target (say 2%) on average. By including some number of past years, this system differs sharply from a conventional inflation target. In the latter case, bygones are bygones: the conventional strategy ignores past misses, aiming for a constant numerical target over the next few years (or the medium term). Even if inflation falls short of the target, the target remains unchanged going forward. In contrast, if policymakers’ strategy is one of average inflation targeting, then they will try to make up for some part of past misses.
To see the difference, consider the U.S. experience since 2012. As we noted above, over the past eight years, inflation has averaged 0.7 percentage points below the Fed’s 2% goal. Despite this shortfall, the Fed’s forward-looking medium-term inflation objective has remained unchanged. If the FOMC’s target had been average inflation, their forward-looking target would have risen to offset the previous shortfall and achieve the average over time.
To understand the mechanics of average inflation targeting, it is useful to consider the case of price level targeting. This strategy is the polar opposite from pure inflation targeting, as bygones are never bygones. Put differently, price level targeting is average inflation targeting with an infinite time window for measuring inflation. That is, the central bank aims at keeping aggregate prices close to a long-run path that increases at the target rate, seeking to compensate for any past misses regardless of when they happened. Many economists have advocated price-level targeting precisely because—when credible—it can lower the expected real interest rate at the lower bound for nominal interest rates, providing more scope for expansionary policy without having to resort to forward guidance or balance sheet tools (see, for example, Woodford).
To illustrate the difference between the two polar approaches, consider two central banks with an identical goal of a 2% annual growth in the price level. The first follows a pure inflation targeting strategy where bygones are bygones, while the second follows a pure price level targeting strategy, seeking to make up any misses. In the absence of any shocks that drive inflation from the target, the two strategies are indistinguishable: the price level rises along a constant 2% growth path. (We borrow this example from George Kahn.)
Suppose, however, that there is a disinflationary shock that lowers inflation in year 1 by one percentage point from 2% to 1% (for the case of an inflationary shock, see Chart 1 in Kahn). Suppose further that both central banks successfully aim to return to their target two years hence, in year 3 (we call this the restoration period). The chart below shows the paths of the price level under these two strategies. Under inflation targeting, the central bank aims to raise inflation from 1% to 2% gradually over the restoration period (see solid blue line). Importantly, while the price level returns to a path that is parallel to the original price path, it is permanently lower: the blue line is below the black dotted line. It is in this sense that bygones are bygones.
The dashed red line shows the price level targeting strategy. Note that this line gradually rises back to the black dotted line. Importantly, during the period from years 1 to 3, the dashed red line is steeper than the black dotted line—that is, inflation exceeds the long-term 2% target during the two-year restoration period. This is what it means for a central bank to make up for past misses: a period of above-target inflation follows a period of below-target inflation. Put differently, under price level targeting, policy is history-dependent. Under inflation targeting, it is not.
Price-level paths under inflation and price-level targeting when inflation falls 1% below target in year 1
Source: Authors' calculations based on Kahn.
From this we can see that average inflation targeting is a hybrid between inflation targeting and price level targeting (for a technical derivation, see Cecchetti and Krause). The longer the averaging window, the more it resembles the latter. Furthermore, the two polar strategies have different implications for the long-run variability of the price level. By minimizing deviations from the target path, a pure price level target makes the level of prices predictable in the long run, but adds to the short-run volatility of inflation. Under inflation targeting, deviations from the original price-level path grow over time, diminishing the predictability of prices in the long run.
In the United States, the practical differences between inflation and price-level targeting have been limited. Arguably, the Fed has been a de facto inflation-targeter since the mid-1980s. Nevertheless, over the past 30 years, the price level (excluding food and energy) has increased broadly in line with that of a pure price-level targeting regime, resulting in an average annual inflation rate of 1.86% (see chart). At the same time, over shorter intervals, deviations have tended to persist. In the early years, inflation typically exceeded the target. As a result, by the second quarter of 1995, the actual price level had surpassed the long-run path by 4.6%. More recently, the pattern has reversed. As of the second quarter of 2020, the price level shortfall is a sizable 3.5%.
Price index of personal consumption expenditures (excluding food and energy): Actual vs. 2% annual rate of increase (Jan 1990=100), 1990-June 2020
Source: FRED and authors’ calculations.
The implication is that a shift by the Fed to an average inflation targeting regime would be a moderate, rather than a radical, step (see also Reifschneider and Wilcox). To see this, we can do a simple computation. Start by assuming (as in the simulation above) that the central bank targets average inflation over five years, looking back three years and seeking to restore the price level its implied 2% path over the next two years. Given the limits on the effectiveness of monetary policy, we suspect that the actual restoration period would have to be longer, so this assumption results in a more aggressive response to shocks and adds to the volatility of the Fed’s annual inflation target. (On the other hand, as the restoration period lengthens, it becomes increasingly difficult for the FOMC to make a credible commitment to meet its average inflation target.)
In the chart below, taking the averaging window to be the previous three years, we plot the implied average annual inflation target over the two-year restoration period. (Note that this a static assumption: it does not take into account how the inflation process changes as the Fed adopts this five-year average inflation targeting approach.) Under these assumptions, the implied target since 2011 would have ranged between 2.20% and 3.04%, with an average of 2.65%. That is, with a three-year historical averaging window and a two-year restoration period, the Fed’s inflation target would have been 0.65 percentage points higher for the past decade. If this central bank behavior was both credible and anticipated, it would have lowered the average real short-term interest rate by 65 basis points; a welcome (if moderate) boost at the zero lower bound for nominal interest rates. (As an aside, we note that shifting from the price excluding food and energy to the headline measure boosts the volatility of the two-year ahead inflation target, widening its range over the 30-year sample by about one percentage point.)
Implied annual average inflation target over the next 2 years based on a past averaging period of 3 years, 1990-2Q 2020
Note: inflation is measured using the price index for personal consumption expenditures excluding food and energy). Source: Authors’ calculations.
That is surely good news, but there is a downside to such a framework. It is visible in the figure. In the first four years of the 1990s (including the 1990-91 recession), a five-year average inflation-targeting regime that includes a two-year restoration period likely would have resulted in substantially tighter policy than observed. Indeed, the central bank would have aimed at a deflation ranging between -1.21% and -0.17%!
The point is that average inflation targeting—like price level targeting—can result in episodes when the central bank wishes to lower the price level, not just lower inflation. This may be why no central bank has explicitly adopted such a framework—at least, not yet. We also note that an implied target of deflation is more likely the longer the averaging window, and the shorter the restoration period. To offset this, the central bank could make both of these time frames flexible, but that probably would diminish the credibility of the policy commitment and its impact on inflation expectations. Another alternative, proposed by Bernanke, is to follow an asymmetric strategy: target inflation in upturns and switch to price-level targeting only in downturns. While this makes logical sense, we suspect that it would be difficult to communicate such an approach in a way that automatically lifts inflation expectations at just the right time, as a recession begins.
Perhaps the Fed will simply further improve how it communicates what it considers to be a reasonable inflation target over the next two or three years—deviating modestly from 2% when there have been persistent shortfalls or excesses—while maintaining some flexibility in defining the averaging window and restoration periods. This appears to be what Fed Governor Brainard has in mind when she said: “Flexible inflation averaging would bring some of the benefits of a formal average-inflation-targeting rule, but it could be more robust and simpler to communicate and implement. Following several years when inflation has remained in the range of 1.5 to 2 percent, the Committee could target inflation outcomes in a range of 2 to 2.5 percent for a period to achieve inflation outcomes of 2 percent, on average, overall.”
So, what are we likely to hear from Chairman Powell later this week, and what is the FOMC likely to conclude at its mid-September meeting?
Our best guess is that the Committee will shift in the direction of average inflation targeting, allowing for some makeup of past misses, and signaling a corresponding change in the Statement on Longer-Run Goals. Policymakers’ approach to their strategic review makes us confident that they will be as clear as possible, even if the strategy falls short of a formal average-inflation-targeting regime. Indeed, FOMC members fully understand that achieving their goal of influencing inflation expectations depends both on transparency and commitment. If, as we suspect, Fed policymakers choose to shift to a version of average inflation targeting, it will still take considerable time and experience for the public, including financial markets, to learn precisely what the change implies for the paths of inflation and interest rates.