• A Book Review of Strategies for Monetary Policy, John H. Cochrane and John B. Taylor, eds.1

Each year, the Hoover Institution hosts a conference on monetary policy at its Stanford University headquarters. The conferences bring together academics and Fed officials to discuss issues in monetary economics. The proceedings from the 2019 conference have now been published in a book entitled Strategies for Monetary Policy. The papers are mostly aimed at readers with knowledge of advanced undergraduate macroeconomics, although a few papers are fairly mathematical and aimed at those who have Ph.D.s in economics.

The first paper is by Federal Reserve Vice Chair Richard Clarida, and discusses the role of forecasts in setting monetary policy. Clarida distinguishes between three types of forecasts; those derived from macroeconomic models, from surveys of professional forecasters and/or the general public, and from asset prices (including yield spreads.) Each type of forecast has advantages and disadvantages.

The Fed traditionally relies on macroeconomic models of the economy when forecasting variables such as inflation. Unfortunately, these models have performed rather poorly in recent years. The models rely on assumptions about the “natural rate of interest” and the “natural rate of unemployment.” But it seems that these parameters are not stable, and excessive reliance on these “Phillips curve models” has caused the Fed to wrongly assume that inflation would increase as the economy recovered.

Forecasts of important macroeconomic variables can sometimes be inferred from asset prices, such as the spread between the yield on conventional bonds and the yield on inflation-indexed bonds (TIPS). That interest rate spread—termed the TIPS spread—would be exactly equal to the market’s inflation forecast if conventional and indexed bonds were perfect substitutes. In that case, the expected return on an indexed bond, which equals its guaranteed real interest rate plus the expected inflation rate, would be identical to the actual yield on a conventional bond of the same maturity.

Unfortunately, the TIPS spread can sometimes be distorted by shifts in risk or liquidity premia, which can make one type of bond more attractive than the other. In recent years, TIPS spreads have tended to slightly underestimate inflation. For this reason, Clarida indicates that he puts at least as much weight on the consensus forecast of inflation from various surveys. Unfortunately, these surveys have recently overestimated inflation—so there is no perfect forecasting technique.

Of course the Fed doesn’t just care about inflation; real economic growth and employment are also a part of their policy framework, and the stock market may anticipate some changes in the business cycle. Clarida is somewhat dismissive of the stock market as a guide to monetary policy, however, with some justification. Stock prices move around for many different reasons, some of which are difficult to explain.

Nonetheless, I suspect the Fed pays more attention to the stock market than they let on. Big drops in the stock market often lead to monetary policy shifts, especially if accompanied by other financial indicators suggesting an increase in risks to economic growth. One can see how the Fed would be a bit uneasy talking about this issue, as they would naturally wish to avoid the perception that they are trying to prop up stock prices, even if the justification were that doing so would help to stabilize the broader economy. Economists in academia are less constrained by public opinion, and Roger Farmer has explicitly argued for using monetary policy to stabilize a broad stock market index.2

In the second paper, Andrew Lilley and Kenneth Rogoff defend the use of negative interest rates as an expansionary policy tool. I found this paper to be very well argued, and perhaps the most important chapter in the book. The traditional view is that deeply negative interest rates are impossible, due to the alternative of holding zero interest currency. Japan and some European countries have been able to push interest rates slightly negative, as holding cash is costly, but this hasn’t really changed the basic view that there is an “effective lower bound” to interest rates. As a result, many Keynesian economists worry that monetary policy is “out of ammunition” once rates fall to zero, or slightly below.

Lilley and Rogoff suggest that with a few modest reforms it would be possible to push interest rates deeply negative, and thus provide almost unlimited monetary stimulus to the economy. This might include eliminating large denomination currency notes ($50 and $100 bills), and imposing increasingly steep fees on the withdrawal or deposit of large currency hoards. The basic idea is to allow a small amount of currency hoarding as well as zero interest bank accounts for average people, but to apply punitive fees to large currency hoarders. This would allow for sharply negative interest rates on large bank deposits held by wealthy individuals and financial institutions, and also allow interest rates on bank loans to go negative.

To see how this works, think about the traditional theory of money and inflation. In normal times when interest rates are positive, cash is a sort of “hot potato,” which people try to get rid of because alternative assets earn a higher interest rate. When the government prints money and injects it into the economy, people spend the excess cash balances, and this eventually boosts aggregate demand and inflation.

But when interest rates on alternative assets falls to zero, banks are likely to sit on any new cash injected into the economy. Putting a negative interest rate on bank reserves is one way of motivating banks to move newly created money out into the economy. But that won’t work if the bank reserves earning negative interest are merely converted into cash earning zero interest.

This is why economists like Lilley and Rogoff are so obsessed with currency. Unless something can be done to prevent massive currency hoarding, it is hard to see how the government can drive interest rates low enough to stimulate the economy in a major recession. Fortunately, it is not necessary to eliminate currency entirely, just make it inconvenient to hoard large quantities. After all, the government can inject literally trillions of dollars of new money into the economy. If a few hundred billion are hoarded as zero interest currency, that’s not enough to soak up anywhere near the increase in bank reserves from large scale asset purchases by the Fed. So I believe that Lilley and Rogoff are probably correct in assuming that their plan is technically feasible.

On the other hand, I’m not convinced the plan is politically feasible, nor am I convinced that it is the best way to stimulate the economy during a period of deflation or depression. The Fed has not even come close to exhausting the potential of conventional options that do not involve negative interest rates.

Lilley and Rogoff are skeptical of the effectiveness of policies such as “quantitative easing” (QE, which means printing lots of money) and forward guidance for policy. But the evidence they cite is not persuasive. It is true that governments that have done QE have not seen a robust recovery, but many studies suggest that QE does have an expansionary effect.3 And note that it is equally true that countries that have adopted negative interest rates have also not seen robust recoveries.

Lilley and Rogoff could argue that central banks in Europe and Japan haven’t driven rates far enough below zero, but one could just as well argue that central banks such as the Fed did not do enough QE, and did not adopt aggressive enough forward guidance. In particular, I’ve advocated a policy of targeting the level of nominal GDP, and adopting a “whatever it takes” approach to quantitative easing to hit the target.4 In my view, this would be more politically feasible in America than steeply negative interest rates, and just as reliable.

Lilley and Rogoff’s pessimism about the effectiveness of conventional monetary policy is part of a broader trend in modern macro, which confuses cause and effect. When low nominal interest rates and QE coincide with weak economic growth, it is often seen as an indication that monetary stimulus doesn’t work, whereas it is actually a sign that previous monetary policy was too tight. Historically, tight money drives nominal GDP growth rates to very low levels (as in the Great Recession), which reduces the equilibrium, or natural rate of interest. As the natural rate of interest falls close to zero, the demand for liquidity increases sharply and central banks respond (defensively) with QE programs.

If the Fed aggressively targeted NGDP growth at a higher rate—say 5%/year—then nominal interest rates would stay above zero and there would be no need for quantitative easing. This is how Australia avoided a recession during the Global Financial Crisis of 2008-09. Lilley/Rogoff reflect the consensus view of economists, which tends to see deflationary episodes as exogenous “shocks” that hit the economy, and not (as I believe) the failure of monetary policy to maintain expectations of adequate nominal GDP growth.

My favorite part of their paper discusses how the government could derive better market forecasts of inflation by creating new types of bonds, where the inflation compensation is capped at 3%/year. By comparing changes in the prices of these bonds with normal inflation-indexed bonds, policymakers could indirectly estimate the probability of inflation rising far above the target. This would provide the Fed with an early warning system for the “tail risk” of major changes in the rate of inflation.

In the comment section of the paper, Andrew Levin (p. 79) suggested that he shared their skepticism about the effectiveness of QE, citing the fact that economic growth did not increase sharply in 2013. But this overlooks the fact that growth did increase modestly in 2013, despite the fact that hundreds of Keynesian economists signed a letter warning that fiscal austerity threatened to push the economy into recession.5 At the beginning of 2013, large tax increases and spending cuts suddenly reduced the budget deficit from $1061 billion to $561 billion, and the modest increase in economic growth during 2013 was almost certainly due to the monetary stimulus adopted in late 2012. Market monetarists like myself predicted that QE would prevent fiscal austerity from slowing the economy, and we were right.

Levin also described a digital cash plan he developed with Michael Bordo, which envisioned 100% safe bank accounts where the deposits are invested in interest-bearing reserves at the Fed.6 Levin points out that these specific accounts would not need deposit insurance, but I’d go even further. If we are going to set up this sort of system, then we should use it as an opportunity to once and for all abolish all government deposit insurance, as the moral hazard created by FDIC has created an increasingly unstable financial system.

“Banks are currently encouraged to take socially excessive risks, knowing that a portion of any losses may be borne by taxpayers.”

Banks are currently encouraged to take socially excessive risks, knowing that a portion of any losses may be borne by taxpayers. With the safe banking alternative described by Levin and Bordo, depositors could choose between safe bank deposits paying lower rates of interest, and riskier deposits paying somewhat higher rates of interest. The funds in the riskier accounts would be used to make bank loans, and there would no longer be a moral hazard problem caused by taxpayer subsidized deposit insurance.

The paper by Thomas Mertens and John Williams looks at various interest rate rules, in the tradition of the famous Taylor Rule. In the 1990s, John Taylor described a policy rule where the Fed’s interest rate target was periodically adjusted according to a mathematical formula.7 Thus, when inflation rose about target, the Fed’s interest rate target would be increased by an even larger amount, to assure that the real interest rate also increased, thus slowing inflation. When output fell below trend, the Fed’s target interest rate would be cut. Much of the subsequent work by mainstream economists is in the tradition of the Taylor Rule, including various proposals in this paper.

For more on the Taylor Rule, see the EconTalk episodes John Taylor on Monetary Policy, Scott Sumner on Monetary Policy, Scott Sumner on Money and the Fed, and John Taylor on the State of the Economy. Consider also “What’s Wrong With the Taylor Rule?” by Jeffrey Rogers Hummel. Library of Economics and Liberty, Nov. 3, 2014.

While I believe that Taylor’s framing of the policy issue was extremely useful, I’m skeptical that any rule based on a mathematical formula would prove robust enough to handle a wide variety of economic shocks. These rules generally require policymakers to estimate both the natural rate of interest and the natural rate of unemployment. In recent years, however, the Fed has done a poor job estimating both variables, which have proved to be surprisingly unstable and unpredictable.

Interestingly, the comments on the paper focused less on the specific rules and more on broader issues, perhaps an indication that many economists don’t trust any specific mathematical policy rule. Several commenters advocated negative interest rates as a serious option, while others focused on the issue of what should the Fed be targeting—inflation, the price level, or nominal GDP growth. The Fed is currently looking at all of these issues in a major review of its policy strategy and tactics.

John Hamilton’s paper argues that the various QE programs were not very effective, citing the fact that bond yields often actually increased during the period when the Fed was purchasing large quantities of Treasury bonds. This might seem a bit surprising, as bond yields move inversely to bond prices. Thus bond prices actually fell during periods where the Fed purchased hundreds of billions of dollars worth of Treasury bonds.

On the other hand, a number of studies have reached a different conclusion, finding that QE was effective, despite falling bond prices. The paradox on falling bond prices coinciding with massive Fed bond purchases can be resolved if we recall that financial markets generally responded to QE announcements in a way that suggests investors believe the policy is effective in boosting economic growth, and faster economic growth tends to lead to higher nominal interest rates. Thus rising interest rates are not necessarily a sign that monetary stimulus has failed, indeed any truly effective stimulus measure is likely to boost growth and inflation enough to raise nominal interest rates over time.

Nonetheless, it is true that at least some QE programs have produced disappointing results. Paul Krugman and David Beckworth argued that temporary monetary injections tend to have very little stimulative effect.8 For instance, the Bank of Japan injected a great deal of new money into the economy in the early 2000s, and then sharply reduced the monetary base in 2006. The Japanese public knew that the Bank of Japan was committed to doing whatever it takes to keep inflation low, and mostly hoarded the temporary currency injections.

In the comment section to the Hamilton paper, Peter Ireland provided a very elegant explanation of the monetarist perspective on QE. Even when the Fed pays interest on bank reserves, the predictions of the quantity theory of money continue to hold for permanent injections of new money. A permanent doubling of the money supply will lead to a permanent doubling of the price level, in the long run.

John Cochrane responded to Ireland by suggesting that it was time to move past quantity of money approaches to policy and just target the interest rate. The public could determine how much money they wished to hold at a given interest rate. Ireland (p. 208) responded with a warning that central banks have occasionally gone far off course when they assumed they were merely responding to changes in the public’s money demand:

  • This is what you hear from central bankers during a hyperinflation. They’ll say, “But we have to keep printing money to keep up with the demand, because the price level is rising so fast.” I’m uneasy about an intellectual framework that appears to suggest, in exactly the same way, that an expansion in a nominal magnitude is just done exclusively to accommodate demand.

In the next paper, John Cochrane, John Taylor and Volker Wieland evaluate the role of policy rules in Federal Reserve monetary policy. They cite earlier studies showing that policy seemed especially rules-based during 1985-2003, which is a period when the economy was relatively stable. In contrast, they blame deviations from the basic Taylor Rule approach in 2003-05 for creating instability that may have contributed to the Great Recession.

I would not rule out the possibility that monetary policy had become a bit too expansionary by 2005, but I have trouble seeing how that could have played a major role in the Great Recession. We saw similar growth rates in aggregate demand (NGDP) during many other post-war expansions, without anything like the housing boom we experienced during 2003-05. Indeed during the 1960s and 1970s, monetary policy was far more expansionary than during 2003-05. Elsewhere I’ve argued that flaws in zoning and banking regulations were more to blame for the housing “bubble.” Kevin Erdman and I recently argued that excessively tight monetary policy was the primary cause of the Great Recession.9

Given that their paper seems generally supportive of the concept of interest rates rules, I wish there had been more discussion of the issue of whether interest rates are a reliable tool (or “instrument”) or monetary policy. On page 231 they present the optimal interest rate path under several versions of the Taylor Rule. These show the appropriate interest rate to be in the 2% to 3% range during 2011-15, a period when the actual rate was close to zero. And yet even that near-zero actual interest rate setting was far too contractionary to hit the Fed’s inflation and employment targets, so it’s not clear why a 2% or 3% rate would have done better.

John Taylor is generally regarded as a fairly mainstream macroeconomist, working in the New Keynesian tradition. According to this framework, an increase in interest rates is a contractionary monetary policy. John Cochrane has recently discussed an alternative and more heterodox model, termed ‘NeoFisherism’. The NeoFisherian model relies on the Fisher effect, which connects inflation and nominal interest rates. According to this view, higher inflation expectations are associated with higher nominal interest rates. Put simply, higher nominal interest rates are actually an expansionary monetary policy.

Because Cochrane and Taylor were two of the three authors of this paper, I wish they had addressed this issue. Given that these two prominent economists have very different views on how the economy moves when the Fed adjusts interest rate, it’s hard to have much confidence in any monetary policy rule that relies on this policy “steering wheel.” Do lower interest rates reflect low inflation expectations resulting from tight money, or are lower rates reflective of the liquidity effect of easy money? In a forthcoming Mercatus Center paper, I argue that interest rates are not a reliable tool of monetary policy, precisely because we don’t know which effect dominates in a given situation.

The book also contains two symposia that discuss various issues in monetary policy. The first examines the relationship between markets and monetary policy. George Shultz introduces this section with some important lessons learned during his experience in government, which goes all the way back to the 1960s. First, we should be very skeptical when people claim that monetary policy cannot control inflation. That was a widely held view during the 1970s, and turned out to be false. Today, we hear similar claims. Second, there are real benefits to resisting the temptation to bail out failing firms. Bailouts create moral hazard, making the financial system even less stable in the long run.

Laurie Hodrick looks at the relationship between monetary policy and stock prices. While there’s little doubt that monetary policy can affect the stock market, it is not easy to measure exactly how the two are related. This is partly because of the “identification problem,” the difficulty in figuring out which changes in interest rates reflect changes in monetary policy, and which reflect other economic forces.

Mickey Levy argues that the Fed has shifted too much in the direction of managing financial stability. I am not convinced by this claim, partly because after a major change in asset prices it is difficult to determine whether the Fed response reflects its desire to stabilize financial markets, or its belief that asset price changes are signaling actual economic shocks that need to be offset. For instance, Levy (p. 274) seems skeptical of the Fed’s response to a sharp break in stock prices in late 2018, but subsequent events showed that the Fed was correct in easing policy—inflation continued to run below target in 2019 and 2020.

One would think that if the Fed were taking its eye off the ball—putting less emphasis on stabilizing inflation and employment—then those variables would become less stable. In fact, during periods where the Fed is accused of propping up asset prices with low interest rates, the Fed usually falls short of its inflation target. So it is not obvious that they are responding to asset prices at a cost of greater inflation instability. In my view, the Fed isn’t actually more active than in the past, rather they seem more active due to the zero lower bound on interest rates, which forces them to adopt unconventional policy tools.

Nonetheless, Levy made a number of good recommendations. On page 276 he recommended that “every policy statement should begin with an assessment of monetary policy and whether it is consistent with achieving the Fed’s statutory mandate, rather than the Fed’s assessments of the economy and its subsectors.” That is, the Fed needs to make it clear that it not merely a bystander, rather it determines the path of nominal aggregates. On page 278, he pointed out that in recent years the interest rate futures markets have made more accurate predictions than the Fed, even about the Fed’s own future policy settings. And on page 280 he recommended that the Fed provide an explicit forecast of nominal GDP.

Scott Minerd argued that the success of inflation targeting is keeping interest rates relatively low and stable, which means there will be greater need for alternative tools such as QE and forward guidance. He recommends that the Fed allow somewhat greater variation in market interest rates, as this would convey useful information about the state of the economy.

St. Louis Fed President James Bullard led off the second symposium with a paper that points to some advantages of nominal GDP targeting. The traditional argument for NGDP targeting is based on New Keynesian models, and focuses on the way that it can help to stabilize the labor market while keeping inflation reasonably well anchored over the business cycle. Bullard points to an additional advantage in a world of nominal debt contracts; NGDP targeting can reduce financial risk, which would help to stabilize the financial system. The basic idea, building on earlier research by George Selgin, Evan Koenig, and Kevin Sheedy, is based on the idea that NGDP is aggregate income, and stabilizing NGDP growth is a way of insuring against income shocks to the economy.10

San Francisco Fed President Mary Daly argues that average inflation targeting is easier to communicate than either price level targeting or nominal GDP level targeting, although I would argue the exact opposite. When in a recession, it is much easier to explain to voters why the Fed is trying to increase national income than it is to explain why the Fed is trying to increase the cost of living for average Americans.

Dallas Fed President Robert Kaplan has a brief paper summarizing models that forecast inflation, an issue that is is still very poorly understood. The Fed tends to rely on Phillips Curve models that suggest that low unemployment should lead to high inflation. But inflation stayed low in 2019 even as unemployment fell to a 50-year low of 3.5 percent. Kaplan mentions technology as a possible factor holding down inflation, which is certainly true in some product categories. But it is hard to see how technology could be depressing the overall inflation rate, as productivity growth has been relatively slow since 2004.

In the final paper, Cleveland Fed President Loretta Mester has a number of good ideas about improving Fed communication with the public. She recommends that the Fed become more transparent about how policymakers are responding to specific economic data points. I applaud this proposal, although I worry that it will be difficult to implement under the current institutional framework. It seems clear to me that the Fed reacts to financial indicators such as the stock market, but equally clear that the Fed is a bit uncomfortable acknowledging that it responds to asset market prices.

I applaud the Fed for undertaking a major study of their policy regime, with an eye on improving policy to better meet the demands of a world where interest rates occasionally fall to zero. My fear is that the Fed will be too cautious—perhaps moving slightly in the right direction, but not far enough to overcome the zero bound problem. If that happens, we will likely continue relying on fiscal stimulus during recessions, a policy that is far more costly than monetary stimulus.


[1] John H. Cochrane and John B. Taylor, Strategies for Monetary Policy. Hoover Institution, 2020.

[2] Roger Farmer, “What Keynes should have said” VOXeu/CEPR, September 17, 2018.

[3] Joseph E. Gagnon, “Quantitative Easing: An Underappreciated Success.” PIIE Policy Brief, Peterson Institute for International Economics, April 2016.

[4] Scott Sumner, “Measurement, Accountability, and Guardrails: Nudging the Fed towards a Rules-Based Policy Regime.” Cato Journal 36, no. 2 (Summer 2016): 315-35.

[5] “350 Economists Warn Sequester Cuts Could Kill the Recovery,” Our Future website, February 26, 2013.

[6] Michael Bordo and Andrew Levin, “U.S. Digital Cash: Principles & Practical Steps.” Economic Working Paper 19101, Hoover Institution, January 2019.

[7] John B. Taylor, “Discretion versus Policy Rules in Practice.” Carnegie-Rochester Conference Series on Public Policy 39 (December 1993): 195-214.

[8] Paul Krugman, “It’s Baaack! Japan’s Slump and the Return of the Liquidity Trap.” Brookings Papers on Economic Activity 2 (1998): 137–87, and David Beckworth, “Permanent versus temporary monetary base Injections: Implications for past and future Fed Policy.” Journal of Macroeconomics. Volume 54, Part A, Pages 1-148 (December 2017)

[9] Scott Sumner and Kevin Erdmann, “Housing Policy, Monetary Policy, and the Great Recession.” Mercatus Center Working Paper, August 2020.

[10] George Selgin, Less Than Zero: The Case for a Falling Price Level in a Growing Economy, Cato Institute: 2018, Evan F. Koenig, “Like a Good Neighbor: Monetary Policy, Financial Stability, and the Distribution of Risk.” International Journal of Central Banking 9 No. 2, (June 2013): 57-82, and Kevin D. Sheedy, “Debt and Incomplete Financial Markets: A Case for Nominal GDP Targeting.” Brookings Papers on Economic Activity (Spring 2014): 301-61.

*Scott Sumner is Professor Emeritus in Economics at Bentley University in Waltham, Massachusetts, and Research Associate on monetary policy at the Mercatus Center. He earned his Ph.D. in economics at the University of Chicago in 1985. He blogs both at EconLog and also at his personal blog at The Money Illusion.