Book club: Scott Sumner’s ‘Alternative Approaches to Monetary Policy’
Book club: Scott Sumner’s ‘Alternative Approaches to Monetary Policy’
Chapter 3: The Princeton School and the Zero Lower Bound
In Chapter 3 of AAMP, Sumner examines the development of the New Keynesian understanding of the zero lower bound problem. The association with Princeton University is probably the least important thing these ideas have in common, but it provides a convenient label. The Princeton School is not a school in the sense of representing a unique theoretical or methodological approach. It is a school only in having a common research focus on the problem of monetary policy at the zero lower bound.
This chapter is an interesting retrospective for me. As a financial market economist and then graduate student with an interest in Japanese monetary policy in the late 1990s and early 2000s, I read much of this literature as it was first published. Sumner’s synthesis gives this literature a coherence I had not fully appreciated when I first read it. He draws a clear throughline from Paul Krugman’s 1998 Brookings paper on liquidity traps showing how the ideas associated with the Princeton School came to influence Fed policymaking and its adoption of flexible average inflation targeting in 2020. He also highlights key points of intersection and differences with market monetarism.
The zero lower bound for nominal interest rates is a problem in standard New Keynesian macro models in which monetary policy is conceived as operating through an official interest rate. Those models become dynamically unstable at the zero bound, unless the central bank can maintain traction over the real interest rate by managing expectations for future interest rates and inflation.
The zero bound was mainly a theoretical problem until Japan began to experience low interest rates in the 1990s. By the late 1990s, the zero bound loomed as a potential constraint on the Bank of Japan’s main operating instrument, the overnight call rate. In 1999, the BoJ adopted what became known as the zero interest rate policy or ZIRP, in which the call rate was held near zero while forward guidance was used to condition expectations for the duration of the policy. It was the first of many policy frameworks the Bank of Japan applied to the zero bound problem.
Paul Krugman revisits the liquidity trap
Japan’s experience drew the interest of Paul Krugman, whose 1998 Brookings paper on liquidity traps Sumner calls one of most important and influential papers in modern macroeconomics. As Sumner notes, Krugman’s paper is often misunderstood and the paper lends itself to at least two main interpretations. One (New Keynesian) interpretation is that monetary remains effective at the zero lower bound, so long as the central bank can credibly influence expectations for future inflation. In particular, Krugman argued that a permanent monetary expansion (for example, through the money base) should create expectations for higher future inflation, lowering the real interest rate and maintaining policymakers’ traction over the economy.
The other (old Keynesian, pre-Princeton School) interpretation, is that fiscal policy should step up to take over the demand management role of monetary policy at the zero bound. As Sumner notes, many of the discussants for his 1998 Brookings paper were actually critical of Krugman’s dismissal of fiscal policy in this context. The latter view is one that you still hear a lot, not least from policymakers and in popular discussion, but Chapter 3 shows that this perspective ignores key developments in New Keynesian macro since the late 1990s.
Krugman himself oscillated between these two interpretations, but clearly favoured the former view in the late 1990s (this was, perhaps not coincidentally, before Krugman’s partisan turn when the Bush Administration assumed office in 2001). In a 1999 paper, Krugman expressed disbelief that Japan could resort to what he characterised as ‘unsustainable deficit spending’ when a more orthodox monetary approach had not been fully employed. Revisiting his 1998 paper in 2018, Krugman changed tact and makes the case for fiscal stimulus instead.
Krugman’s 1998 paper was prescient in effectively predicting the main reason Japan’s approach to monetary policy in the late 1990s and early 2000s would fail. The BoJ not only failed to credibly signal a permanent monetary expansion, but validated those expectations by reversing course. The problem was not that monetary policy was ineffective at the zero lower bound, it was that the Bank of Japan was not doing it right. Krugman’s paper was well-known to Japanese policymakers. I recall discussing it with officials when I visited the BoJ in 2000 and they had no shortage of advice from monetarists around this time. But for the BoJ, the ‘wild inflation’ of the 1970s still loomed large and the idea that it should deliberately cultivate expectations for future inflation was not one they were willing to embrace.
Ben Bernanke on Japanese monetary policy
As a Princeton academic and then as Fed Governor from 2002, Ben Bernanke picked-up the criticism of Japanese monetary policy. A key element of Bernanke’s critique was recognition that a central bank can never ‘run out ammunition.’ If the price level were really independent of monetary expansion, then the central bank could in theory acquire infinite quantities of assets or goods and services. This is impossible in equilibrium, so there must be some amount of monetary expansion that will raise the price level. As an aside, MMT maintains that monetary-financed fiscal expansion can be used in this way, but with price controls and/or taxes managing the inflationary consequences. In reality, those controls would wreck a market economy.
Note that Bernanke’s perspective implies that monetary policy could be effective even if not fully credible. The monetary authority could brute force higher inflation though monetary expansion. But as Sumner notes, the more credible the commitment to monetary expansion, the less expansion the central bank actually has to do. Indeed, the ideal monetary policy is one in which the central bank does very little with its policy instruments, relying instead on open mouth operations. The RBNZ functioned without an official interest rate before 1999, relying only on an explicit threat to change the quantity of settlement cash. Sumner notes that the central banks with the largest balance sheets are those that have implemented less credible policies. The key to the successful conduct of monetary policy is to formulate credible commitments, for which the policy instruments are mainly a communications device. In this conception of monetary policy, the actual effects of those instruments are a function of the credibility of policy.
Bernanke rejected the argument that monetary expansion amounts to beggar-thy-neighbour currency depreciation. The claim that central banks are engaged in ‘currency wars’ through monetary expansion is one you hear a lot in popular discussion. But as Bernanke argues, monetary expansion creates trade as well as diverting it. The beggar-thy-neighbour argument has it origins in the Great Depression, when it was used to oppose the devaluations that were needed for recovery. One of Japan’s main bilateral trade partners, the United States, was consistently calling on Japan to do more with macroeconomic policy in the 1990s and 2000s precisely because it recognised that economic recovery in Japan would be good for the US. Bernanke also rejected the view that the central bank’s balance sheet should be a constraint on monetary policy, noting that any losses on the BoJ’s portfolio of government bonds would be offset by gains to the fiscal authority.
Bernanke’s critique of Japanese monetary policy during the late 1990s and early 2000s naturally raises the question as to why the Fed made similar mistakes when Bernanke was Fed chair during and after the financial crisis of 2008. Sumner argues, convincingly in my view, that Bernanke was institutionally constrained by the Federal Reserve system into adopting a policy stance that was too conservative, as shown in Sumner’s The Money Illusion. Sumner argues that given a free hand, Bernanke would have done much better.
Gauti Eggertsson and Michael Woodford on history dependent policy
The work of Eggertsson and Woodford shows in a more formal way how central banks can make credible commitments that maintain policy effectiveness at the zero bound. In particular, they argue for history dependent policy frameworks, rather than policies optimised for current conditions. History dependent policies create expectations for either the price level or nominal GDP that are themselves stabilising. These policies include price level or nominal GDP level targeting, in which the central bank seeks to correct previous undershoots or overshoots of the target path, thereby anchoring expectations. This is in contrast to the let bygones-be-bygones approach that characterises inflation targeting in Australia, for example. Eggertsson and Woodford propose nominal GDP level targeting in a New Keynesian framework, partly on the basis that this would be easier to explain to the public, at least relative to an output gap-adjusted price level target. As Eggertsson and Woodford note, this is just a modern version of Friedman’s k-percent money growth rule, but replacing Friedman’s intermediate target variable (M2 growth) with the goal variable.
The Fed’s flexible averaging inflation targeting regime from 2020, at least as it was originally conceived, owes a clear debt to this history dependent approach. It can be viewed as a compromise, which allows the Fed to continue to use the language of inflation targeting, while at least notionally introducing a history dependent level targeting regime. Of course, Fed policy has since departed from this notional framework in important ways.
Lars Svensson’s ‘foolproof way’ and forecast targeting
Lars Svensson’s foolproof way of escaping a liquidity trap is firmly in the Princeton School of New Keynesian approaches to the zero bound problem. We have previously discussed this in the context of Japan’s recent experience with yield curve control, so won’t elaborate on it here. But Svensson’s work is consistent with the Princeton School in noting that there is no in-principle limit on the central bank’s ability to promote inflation. As Sumner notes, ‘an assumption that a monetary policy achieves substantial currency depreciation is tantamount to an assumption that the policy is successful in creating higher inflation’ (p. 79).
Svensson’s other important contribution is to argue for forecast targeting over instrument rules. An instrument rule, like the Taylor rule, entails changing the policy instrument in response to variables like inflation and the output gap. Forecasting targeting, by contrast, argues that the monetary policy instrument should be set at the level required to realise the desired or optimal policy outcome. In other words, the central bank’s economic forecasts should be the same as the policy goals, with the policy instruments adjusted to ensure the forecasts are consistent with the goals. This arguably requires a published interest rate projection alongside economic forecasts consistent with that projection. Note that the RBA publishes economic forecasts conditional on market- and survey-implied expectations for its official cash rate, but not an explicit interest rate projection of its own. More seriously, the macro forecasts at the back end of the projection period are often inconsistent with the RBA’s stated goals, a fact acknowledged in the minutes of the April 2023 RBA Board meeting.
Market monetarism
Market monetarists adapt forecast targeting to advocate targeting an asset price linked to the policy goal on the basis that the market forecast for inflation or nominal GDP is the optimal forecast. In this way, market monetarists avoid conditioning monetary policy on a structural model of the economy that may be wrong or incomplete. This is what puts the ‘market’ in market monetarism. In this context, it is worth noting the WEAI monetary panel discussion we linked to previously, in which RBA Board member Ian Harper and Deputy Governor Michelle Bullock both agreed the RBA was led astray pre-pandemic by a structural variable, the NAIRU, although as Peter Tulip has argued, the relevant model output was ignored in this case.
Sumner addresses the circularity problem that might be thought to arise when markets are watching policymakers to determine asset prices at the same time that policymakers are watching asset prices for guidance on monetary policy. In a NGDP futures targeting regime of the type advocated by Sumner, the targeted asset price doesn’t change and to that extent doesn’t provide policy guidance. The market is implicitly predicting the instrument setting required to achieve the policy goal. As Sumner notes, this is no different from a fixed exchange rate regime in which the market effectively tells the central bank what monetary policy stance is required to keep the foreign exchange rate constant.
From a policymaker’s standpoint, the information that is valuable is not the market’s forecast of inflation or nominal GDP, it is the forecast of the policy instrument conditional on the nominal GDP target being met (or alternatively, what interest rate setting is most likely to achieve the target inflation rate in an inflation targeting regime).
Nominal GDP futures targeting is in many ways the logical evolution of forecast targeting and history dependent policy as mechanisms for reinforcing policy credibility, but without the reliance on a structural model to guide policy or even a view on the monetary policy transmission mechanism (as we saw in Chapter 1). The Fed’s 2020 long-term strategy review contains elements of history dependent policy, although subsequent monetary policy practice departs from it in important ways.
Summary
In at least one respect, the zero lower bound problem was a positive development because it forced mainstream macroeconomics to address its mistaken focus on interest rates as the only monetary policy instrument and measure of the stance of monetary policy. Macroeconomists had to think more fundamentally about how monetary policy works. It was beneficial to monetarism, because monetarists already viewed interest rates as at best an incomplete and at worst a seriously misleading indicator of the stance of monetary policy.
Chapter 3 shows the ways in which New Keynesian macro has evolved to address the issues raised in a low interest rate environment, issues the traditional Keynesian framework was ill-equipped to handle. In tackling this problem, those in the Princeton School have evolved monetary policy thinking and practice in a direction that has reaffirmed the effectiveness of monetary policy at the zero bound, consistent with long-standing monetarist views. The Princeton School landed on nominal income targeting from a New Keynesian rather than monetarist perspective.
The Princeton School and modern macro part company with market monetarism in viewing aggregate demand shocks as largely exogenous rather than due to monetary policy errors. The Princeton School also retains a higher level of conviction in fiscal policy for demand management, with policy practice, especially in the wake of the pandemic, still wedded to a mix of monetary and fiscal policy in responding to shocks. But the most significant point of departure remains market monetarism’s extension of forecast targeting to use asset prices as the conditioning variable for policy.
Chapter 3 is a very good, non-technical survey of the New Keynesian literature on the zero bound problem and how it has come to influence modern monetary policy theory and practice. For me, one of the main takeaways is the extent to which popular discussion has not caught up to developments in mainstream macroeconomics, much less market monetarism. Every time you hear claims that central banks are out of ammunition, that monetary easing is beggar-thy-neighbour policy (aka ‘currency wars’), that monetary policy is ineffective or that we need to rely on fiscal policy for demand management, you can be sure the person making those claims is out of touch with modern macro.
ICYMI
Come inside the RBA's media lock-ups to see how we report on interest rates and financial risks.
Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models.
How policymakers can improve the functioning of the US Treasury market.
More Flexible Zoning Helps Contain Rising Rents: New data from four jurisdictions that are allowing more housing shows sharply slowed rent growth.