Facts, Fears, and Functionality of NGDP Level Targeting: a Guide to a PopuLar Framework for Monetary Policy David Beckworth

Facts, Fears, and Functionality of NGDP Level Targeting: A Guide to a Popular Framework for Monetary Policy David Beckworth

SPECIAL STUDY David Beckworth. “Facts, Fears, and Functionality of NGDP Level Targeting: A Guide to a Popular Framework for Monetary Policy.” Mercatus Special Study, Mercatus Center at George Mason University, Arlington, VA, September 2019.

ABSTRACT Nominal GDP level targeting (NGDPLT) has become an increasingly popular monetary policy framework over the past decade. This rising popularity has led to increased interest in, as well as some confusion over, how this framework actually works. This paper attempts to address this interest by summarizing basic facts of NGDPLT and addressing some of the fears surrounding it. The paper also demonstrates how NGDPLT might work in practice.

JEL codes: E50, E52, E58

Keywords: nominal GDP targeting, NGDP, nominal income targeting, monetary policy frameworks

This paper can be accessed at https://www.mercatus.org/publications/monetary -policy/facts-fears-and-functionality-ngdp-level-targeting-guide-popular

The views expressed in Mercatus Special Studies are the authors’ and do not rep- resent official positions of the Mercatus Center or George Mason University.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 2 CONTENTS

Introduction 5

I. Facts about Nominal GDP Level Targeting 7

Fact 1: NGDPLT Is a Dollar-Denominated Target 7 Fact 2: NGDPLT Is a Growth-Path Target 7 Fact 3: NGDPLT Is a Velocity-Adjusted Money Supply Target 9 Fact 4: NGDPLT Is a Work-Around to the Supply Shock Problem 11 Fact 5: NGDPLT Is a Work-Around to Incomplete Financial Markets 14 Fact 6: NGDPLT Is an Anti–Zero Lower Bound Tool 16 Fact 7: NGDPLT Is a Way to Do Rules-Based Monetary Policy 17

II. Fears about Nominal GDP Level Targeting 19

Fear 1: Changes in Potential Real GDP Will Create Problems for NGDPLT 19 Fear 2: Data Revisions Make NGDPLT an Impractical Framework 20 Fear 3: The Public Will Not Understand NGDPLT 22 Fear 4: NGDPLT Does Not Satisfy the “Price Stability” Part of the Dual Mandate 23

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 3 Fear 5: NGDPLT Does Not Address Financial Stability Concerns 26 Fear 6: Overshoots of an NGDP Target Would Be Politically Tough to Correct 26 Fear 7: NGDPLT Is Too Radical. Why Not Settle for Price Level Targeting? 27

III. Functionality: How to Implement Nominal GDP Level Targeting 28

IV. Conclusion 31

Appendix. Supply Shocks in a Monetary Policy–Phillips Curve Model 32

I. The MP-PC Model 32

II. A Temporary Supply Shock 33

III. A Permanent Supply Shock 35

Notes 37

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 4 INTRODUCTION

oes the Federal Reserve need a new mone- One proposal for revamping the monetary tary policy framework? A growing number policy framework is for the Federal Reserve to Dof observers believe the answer is yes. Some adopt a nominal GDP level target. This approach see the Great Recession, the slow recovery from it, would have the Fed target a stable growth path and the persistent shortfall of inflation relative to for the total amount of spending in the economy. its target as evidence that the current approach to While not a new idea, nominal GDP level target- monetary policy is inadequate.1 Others worry that ing (NGDPLT) became popular in the aftermath the secular decline in interest rates is permanent of the Great Recession.3 and leaves insufficient room for the Fed to cut This rise in popularity can be seen in panel A interest rates in future recessions.2 These concerns of figure 1, which shows the results of Google and have caused many to call for an overhaul of the way Google Scholar searches for NGDP targeting. the Fed does monetary policy. Panel B of figure 1 indicates that not only does

FIGURE 1. GOOGLE SEARCH RESULTS FOR ARTICLES ON NOMINAL GDP TARGETING

Panel A. NGDP Targeting Search Panel B. Monetary Policy Frameworks Searches 180 5,000 5,000 160 140 4,000 4,000 120 3,000 100 3,000 80 All Articles All Articles 2,000 2,000 60 Scholarly Articles 40 1,000 1,000 20 0 0 0 1983 1988 1993 1998 2003 2008 2013 2018 2000 2003 2006 2009 2012 2015 2018 All Articles: Google Search Price Level Targeting Scholarly Articles: Google Scholar Higher Inﬂation Target Nominal GDP Targeting

Note: The search criteria used for nominal GDP targeting are as follows: “nominal GDP targeting” or “nominal GDP target” or “NGDP targeting” or “NGDP target” or “nominal income targeting” or “nominal income target.” For price level targeting, the search criteria used are “price level targeting” or “price level target.” For the higher inflation target, the search criterion is “higher inflation target.” The total number of search results for each year are used and were pulled from Google on April 9, 2019.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 5 NGDPLT remain popular, but it appears to be placed. This paper illustrates why by clarify- generating more interest than some other mon- ing what exactly NGDPLT is and by addressing etary policy framework proposals, such as price the misconceptions about it. Ultimately, this level targeting or a higher inflation target.4 paper shows that this monetary policy framework Like many popular phenomena, however, can be implemented in a stabilizing manner. NGDPLT has been misunderstood by some To that end, this paper provides a guide to observers, and this has led to confusion over what the facts, fears, and functionality of NGDPLT. It it would mean for monetary policy. Some believe, does so in an accessible and executive-summary- for example, that it would fail to anchor inflation styled format so that interested parties can easily expectations or that it would increase economic find specific issues surrounding NGDPLT while volatility.5 Others worry that NGDPLT is an still being able to get a complete picture of this impractical framework for monetary policy approach to monetary policy. The paper first looks because of changes in potential real GDP, data at seven facts and seven fears regarding NGDPLT. revisions, public confusion, and the mechanics of It then shows how NGDPLT might be imple- implementing it. mented in practice. The paper concludes with While all of these concerns about NGDPLT some practical suggestions for making the tran- are understandable, they are ultimately mis- sition to NGDPLT.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 6 I. FACTS ABOUT NOMINAL GDP LEVEL TARGETING

FACT 1: NGDPLT IS A DOLLAR- where P is the price level and Y is real income. DENOMINATED TARGET Real income is inflation-adjusted income and is NGDPLT anchors the dollar size of the US econ- the real gains from working in the economy. omy. This mooring is accomplished by having Since NGDP = NGDI, (1) and (2) can be combined the central bank target the level of total dollar to get the following: spending in the economy. The total amount of MV = PY. (3) money spent, in turn, generates an equal amount of money earned in the economy. The former is Equation (3) is the famous equation of officially called nominal gross domestic product exchange, an accounting identity that relates (NGDP), while the latter is called nominal gross money transactions to money incomes. It implies domestic income (NGDI).6 Because these are that an NGDP target that stabilizes MV is equiva- equal, NGDPLT can be viewed as a target for lent to a nominal income target that stabilizes PY. both total dollar spending and total dollar In either case, the monetary policy goal is a income earned. As a consequence, some com- dollar-denominated target. mentators call this approach nominal income All of this means, as mentioned earlier, that targeting. NGDPLT anchors the dollar size of the economy. To better understand this connection, note Put differently, NGDPLT is a nominal anchor that that total dollar spending on the economy is keeps the growth of prices, wages, and other equal to the stock of money multiplied by how dollar-denominated activity moored so that often it is used. This decomposition of NGDP can they do not expand too rapidly. be summarized as follows:

NGDP = MV, (1) FACT 2: NGDPLT IS A GROWTH-PATH TARGET where M is the total amount of money and V is NGDPLT requires the Fed to stabilize the level or the velocity, the number of times money gets growth path of NGDP. This means the Fed would used. Total dollar income can similarly be decom- have to make up for past misses from its target so posed into two parts: the price level times real that the targeted dollar level of NGDP is always income. This can be summarized as follows: maintained. This is illustrated in figure 2. It NGDI = PY, (2) shows a scenario where the Fed is targeting

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 7 FIGURE 2. A NOMINAL GDP LEVEL TARGET

Panel A. Correcting for Below-Target Growth Panel B. Correcting for Above-Target Growth Nominal GDP Level Target Nominal GDP Level Target

h Target Target Growth Path Growth Pat Slow-down Growth

Catch-Up Growth

h NGDP (Dollars ) NGDP (Dollars )

Target Growth Pat Target Growth Path

Y1 Y2 Y3 Y4 Year (Yt)Y1 Y2 Y3 Y4 Year (Yt) some growth rate—the slope of the line—for NGDP set of rows shows what happens when NGDP and makes up for periods of below- and above- grows below 4 percent in 2019 (Y1). The missed target growth so that the trend growth path is growth is made up in the next two years by hav- maintained. ing NGDP grow faster than 4 percent. The third Panel A shows a case where NGDP falls in set of rows shows the scenario where NGDP year one (Y1). The Fed would make up for this grows faster than 4 percent in 2019 (Y1). This miss in the next two years (Y2, Y3) by growing miss is corrected for in the next two years by hav- total dollar spending faster than the target—the ing NGDP grow lower than 4 percent. In all cases, steeper slope—until it has caught up to its tar- the dollar size of NGDP returns to its targeted get dollar level. Panel B shows that a similar value by the end of 2021 (Y3). NGDP, then, is also response would follow a spending boom that able to return to its normal target growth rate of pushed money spending above the targeted path. 4 percent in 2022 (Y4). The growth rate of NGDP would temporarily If this target were understood by the public slow down until the targeted growth path was and were credible, it would create expectations reached. of stable money spending growth that would Table 1 further illustrates this idea of mon- become self-fulfilling. That is,house holds and etary policy correcting for past misses by put- firms would have less incentive to rapidly spend ting numbers to the scenarios in figure 2. The or hoard money in the first place if they believed actual NGDP dollar size in 2018 is used as a start- the Fed would always correct past misses in its ing point, while the table assumes the NGDP targeted growth path.7 A credible NGDP level growth path is targeted at 4 percent trend target, in other words, would lead to the public growth. doing most of the spending adjustments needed The first few rows oftable 1 outline the base- to keep NGDP on its target growth path. This sta- line case where the NGDP level target is main- bilizing feature is why most proponents of this tained every year at 4 percent growth. The second framework favor a growth path target rather

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 8 TABLE 1. NOMINAL GDP LEVEL TARGETING SCENARIOS

On-target NGDP growth 2018 (Y0) 2019 (Y1) 2020 (Y2) 2021 (Y3) 2022 (Y4) Dollar size (trillions) $20.87 $21.70 $22.57 $23.47 $24.41 Growth rate 4.00% 4.00% 4.00% 4.00% 4.00%

Below-target NGDP growth Dollar size (trillions) $20.87 $20.66 $22.02 $23.47 $24.41

Growth rate 4.00% −1.00% 6.59% 6.59% 4.00% Above-target NGDP growth Dollar size (trillions) $20.87 $22.53 $23.00 $23.47 $24.41 Growth rate 4.00% 8.00% 2.05% 2.05% 4.00%

than a growth rate target: only the former makes These offsetting actions do not require spe- up for past misses. cial effort by the Fed since they happen automatically when the Fed stabilizes the growth path of total dollar spending. These naturally occurring FACT 3: NGDPLT IS offsets that characterize NGDPLT can be viewed A VELOCITY-ADJUSTED as a “monetary seesaw” as seen in figure 3. MONEY SUPPLY TARGET This monetary seesaw view of NGDPLT can Total dollar spending, as noted earlier, is equal to be illustrated by comparing two periods in US the money supply multiplied by its use. Since NGDP history. The first is theGreat Unanchoring NGDPLT stabilizes total dollar spending, it is period of 1960–1979, when total dollar spending therefore effectively stabilizing the interactions growth was not anchored and rapidly accelerated. between the stock of money and its velocity. The second period is the Great Reanchoring of Some observers, consequently, call NGDPLT a roughly the period 1985–2007, when the Fed did velocity-adjusted money supply target. effectively stabilize the growth of total spend- This concept can be better understood by ing.9 These periods can be seen in figure 4, which noting that changes in the supply of money are shows the year-on-year growth rate for NGDP automatically offset by changes in the demand and its trends. The trend growth rate during the for money under NGDPLT. For example, mone- Great Unanchoring was increasing 0.33 per- tary conditions would automatically loosen centage points a year, while it was flat during when people were more inclined to hold money the Great Reanchoring. balances—when, for example, they were afraid of The seesaw view of NGDPLT implies that economic trouble and wanted liquid assets—and during periods of unstable NGDP growth, such as monetary conditions would automatically the Great Unanchoring, changes in the money tighten when people were rapidly spending supply and velocity are not offsetting each other, money. The Fed, therefore, would be indirectly while during periods of stable NGDP growth, constraining money supply growth when money such as the Great Reanchoring, they are offsetting was circulating quickly and encouraging it when each other. The scatterplots in figure 5 show this turnover was low.8 to be the case, using two measures of the money

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 9 FIGURE 3. MONETARY SEESAW VIEW OF NGDPLT

Automatically Osets Increases (Decreases) Automatically Osets Decreases (Increases) in Supply (Velocity) of Money in Supply (Velocity) of Money

Money Money Supply Velocity Growth Growth

Money Money Velocity Supply Growth Growth

Stable NGDP Stable NGDP Growth Path Growth Path

FIGURE 4. TWO NGDP PERIODS

The Great 12 Unanchoring

10 The Great 8 Reanchoring

4 Percent Change from a Year Ago

0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Nominal GDP Trend

Source: FRED database and author’s calculations supply: Divisia M1 and Divisia M4. The former is row, however, shows a much stronger negative a narrow measure of retail money assets, while the correlation with slopes that are close to −1. This latter is a broader measure and includes retail and means the Fed was proportionally offsetting institutional money assets.10 swings in money supply and money demand such The scatterplots in the first row of figure 5 that there was a relatively stable growth path for reveal a very weak negative correlation between total dollar spending during the period 1985– changes in the money supply and changes in 2007. This is the velocity-adjusted money supply velocity during the period 1960–1979. The second targeting view of NGDPLT.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 10 FIGURE 5. A TALE OF TWO NGDP SEESAWS

Broken Seesaw: 1960–1979 Broken Seesaw: 1960–1979 3.5% 5.0%

3.0% 4.0% 2.5% 3.0% 2.0% 1.5% 2.0% 1.0% 1.0%

0.5% 0.0% Change in Velocity 0.0% Change in Velocity –1.0% –0.5% R = 15.42% –1.0% R = 3.53% –2.0% Slope = –0.24 Slope = –0.49 –1.5% –3.0% –1.0% 0.0% 1.0% 2.0% 3.0% 4.0% –1.0% 0.0% 1.0% 2.0% 3.0% 4.0% Change in Divisia M1 Money Supply Change in Divisia M4 Money Supply

Functioning Seesaw: 1985–2007 Functioning Seesaw: 1985–2007 4.0% 3.0% 3.0% 2.0% 2.0% 1.0% 1.0% 0.0% –1.0% 0.0%

–2.0% –1.0% –3.0% Change in Velocit y Change in Velocit y –4.0% –2.0% –5.0% R = 85.99% –3.0% R = 67.22% –6.0% Slope = –1.06 Slope = –1.10 –7.0% –4.0% –2.0% 0.0% 2.0% 4.0% 6.0% 8.0% –1.0% 0.0% 1.0% 2.0% 3.0% 4.0% Change in Divisia M1 Money Supply Change in Divisia M4 Money Supply

Notes: Changes are calculated as the a percent change from the previous quarter. Divisia M1 is a narrow measure of retail money assets, and Divisia M4 is a broader measure that includes retail and institutional money assets. Source: FRED database and author’s calculations.

FACT 4: NGDPLT IS A policy. In this case, however, tightening would WORK-AROUND TO THE further choke an economy already weakened by SUPPLY SHOCK PROBLEM the reduction in its productive capacity. When One of the tougher challenges Fed officials face is movements in the price level reflect changes to how to deal with supply shocks. These are unex- the productive capacity of the economy, it is best pected changes to the productive capacity of an for the Fed to ignore them. economy that push economic activity and infla- On the other hand, the Fed should tackle in tion in opposite directions. A sudden reduction flation arising from demand shocks. These shocks in the labor force, oil supply, or technology, for push economic activity and the price level in the example, would increase production costs and same direction and are therefore easier to handle. temporarily raise inflation. This development The Fed’s influence on the economy, moreover, might tempt a central bank to tighten monetary comes from altering demand conditions that

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 11 FIGURE 6. UNDERLYING CAUSES OF INFLATION Central Bank also misread supply-side-caused Supply Shock Demand Shock inflation in 2008 and 2011 and, as a result, tight- ened monetary policy. These actions by the central bank helped create and deepen the eurozone Inﬂation crisis.12 NGDPLT provides a simple work-around to this supply-shock problem: focus directly on Tighten or Loosen Monetary Policy? demand and ignore inflation in the short run. NGDPLT automatically does this by stabilizing the growth path of total dollar spending. underlie such activity. For example, if there were Table 2 illustrates this feature. It assumes an an unsustainable surge in spending that raised NGDP target of 4 percent, potential real GDP inflation too high, the Fed’s tightening of mone- growth of 2 percent, and a resulting trend inflation tary policy would simul ta neously fix inflation and rate of 2 percent. These values are depicted in the rein in the excessive spending growth. first row of thetable. The second row assumes a The problem is that Fed officials are unlikely positive supply shock that temporarily raises real to know in real time what kind of shock is causing GDP growth to 3 percent. Because NGDP is tar- inflation, as seen in figure 6. Nonetheless, know- geted at 4 percent growth, inflation temporarily ing the difference is crucial because responding to falls to 1 percent. The third row shows a negative supply-shock-driven movements in inflation is supply shock that temporarily lowers real GDP generally destabilizing to the economy. growth to 1 percent. Now the 4 percent NGDP tar- Recent examples include the period 2002– get temporarily raises inflation to 3 percent. 2004, when a productivity boom (a positive sup- Table 2, in short, shows how NGDPLT stabilizes ply shock) created a disinflationary environment. total dollar spending growth while allowing its The declining inflation caused Fed officials to composition of inflation and real economic growth worry about deflation and keep interest rates low to temporarily vary. for an extended period, even though the credit One country that illustrates what this boom was emerging. This response intensified the approach might look like in practice is Israel. business cycle.11 In 2008, Fed officialswere con- Though officially targeting an inflation range of 1 cerned about rising inflation coming from surging to 3 percent, the Bank of Israel has effectively sta- commodity prices (a negative supply shock). As bilized the growth path of NGDP over the past a result, the Fed decided against further easing decade. The bank, therefore, has prevented between April and October 2008 despite the eco- demand shocks from destabilizing overall demand nomic slowdown. This too intensified the busi- growth in Israel, as seen in panel A of figure 7. ness cycle. Across the Atlantic, the Euro pean

TABLE 2. NOMINAL GDP TARGETING AND SUPPLY SHOCKS

NGDP Target Type of Shock %ΔPY ≈ %ΔP + %ΔY 4% No shock %ΔPY ≈ 2% + 2% 4% Positive supply shock %ΔPY ≈ 1% + 3% 4% Negative supply shock %ΔPY ≈ 3% + 1%

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 12 FIGURE 7. WHAT NGDP TARGETING WOULD LOOK LIKE IN PRACTICE

Panel A. Stable NGDP Growth Path in Israel 350₪

330₪

310₪

290₪

270₪

250₪

Billions of Shekels 230₪

210₪

190₪

170₪

2011 2008 2009 2010 2012 2013 2014 2015 2016 2017 2018 Nominal GDP Trend

Panel B. Countercyclical Inﬂation in Israel 7

1 Percent Change from a Year Ago

–1

2011 2008 2009 2010 2012 2013 2014 2015 2016 2017 2018 Target Inﬂation Range Real GDP GDP Deﬂator

Source: FRED database and author’s calculations.

On the other hand, the Bank of Israel has tionship is strong enough that the inflation rate has allowed supply shocks to manifest themselves in been allowed to temporarily move outside the countercyclical inflation. This can be seen in panel inflation target range when there have been large B of figure 7, where real GDP growth has been supply shocks. For example, in 2009 during the matched by almost mirror opposite movements in global financial crisis, the inflation rate just topped the GDP deflator growth rate. This inverse rela- 5 percent. Despite this inflation flexibility, inflation

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 13 over the entire period has averaged near the cen- To see how, note first that the countercyclical ter of its targeted range at 1.9 percent. tendency of inflationunder NGDPLT means that An explicit NGDP target for the Fed would a sudden decline in real GDP will lead to an unex- similarly result in short-run inflation flexibility pectedly higher price level and, as a result, a while anchoring long-run inflation. In such a lower real debt burden for the debtor. The cred- framework, the Fed would cease worrying about itor, consequently, receives a lower real debt pay- the inflation rate in the short run—while still ment than expected and shares in the loss. It is anchoring it in the long run—and therefore avoid not all borne by the debtor. The risk of a real making destabilizing mistakes like those in the income loss is shared more evenly between the period 2002–2004 and in 2008, when it got con- debtor and the creditor under NGDPLT. fused by supply-side-driven changes in inflation. Now note that a sudden increase in real GDP This feature of NGDPLT is further illustrated in will lead to an unexpectedly lower price level and section two of the appendix using a Monetary an unanticipated higher real debt payment from Policy–Phillips Curve model. the debtor to the creditor. This feature can be seen as providing insurance to creditors against having their funds locked up in a fixed-price FACT 5: NGDPLT IS A WORK- dollar-denominated loan while real earnings in AROUND TO INCOMPLETE the rest of the economy rise. This way, the creditor FINANCIAL MARKETS gets to share in some of the unexpected “windfall One implication of NGDPLT highlighted earlier is gains” in the economy. that it will tend to create countercyclical inflation. NGDPLT should be, then, a tool for enhanc- A spate of recent papers shows that this feature of ing financial stability given the global growth of NGDPLT leads to better risk sharing between debt over the past few decades. One implication debtors and creditors.13 The basic idea is that the of this understanding is that those countries countercyclical inflationwill cause real debt bur- whose NGDP stayed closest to its expected pre- dens to change in a procyclical manner. As a result, crisis growth path during the financial crisis debtors will benefit during recessions and credi- should have experienced the least financial insta- tors will benefit during booms. Fixed nominal- bility. I tested this implication for 21 advanced price loans will act more like equity than debt and economies using a number of empirical tests and therefore promote financial stability. found it was borne out.14 Another way to view this feature is that in a Figure 8 provides a glimpse of this analysis. world of incomplete financial markets where one It plots the NGDP gap—the percent difference cannot insure against all future risks, an NGDP between where NGDP was expected to be and level target provides a work-around solution to this where it actually ended up—against a number of market deficiency. NGDPLT effectively provides financial measures for the period 2008–2013. A insurance against future risks that could affect negative NGDP gap means NGDP ended up debtors’ ability to repay their debts and also pro- being less than expected, and vice versa. In gen- vides insurance against potential returns creditors eral, the scatterplots indicate there is a system- might miss out on because their funds are locked atic relationship between realizations of stable up in fixed-price dollar-denominated loans. NGDP growth and financial stability.15

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 14 FIGURE 8. NGDP GAPS AND FINANCIAL STABILITY

NGDP Gap and Private Credit Growth NGDP Gap and M3 Money Supply Growth 3% 1% JPN R = 47.86% DEU CHE KOR SWE BEL 1% R = 58.41% CAN AUS –2% NLD FRA ISR KOR FIN AUT 1% CAN NZL ISR ITA GBR 1% GBR SWE –7% DNK AUS CHE FIN DNK USA NZL DEU 0% NLD AUT ITA FRA –12% BEL 0% USA JPN GRC PRT 2008–2013 Average 0% ESP –17% Average M3 Growth Rate PRT ESP GRC

∆ Avg. Private Credit Growth Rate 0% 2008–2013 Avg. minus 2003–2007 –22% 0% –20% –15% –10% –5% 0% 5% –20% –15% –10% –5%0%5% Average NGDP Forecast Error: 2008–2013 Average NGDP Forecast Error: 2008–2013

NGDP Gap and Stock Price Growth NGDP Gap and Home Price Growth 10% 15%

R = 53.57% R = 64.14% 5% KOR 10% ISR DNK ISR AUT SWE USA CAN DEU CHE CAN CHE 0% GBR AUS 5% SWE FRA NZL KOR AUS NLD BEL FIN BEL PRT JPN FRA NZL DEU –5% 0% JPN AUT ITA ESP FIN GBR USA DNK ITA NLD –10% –5% PRT GRC 2008–2013 Average 2008–2013 Average ESP

–15% GRC –10% Average Home Price Growth Rate Average Stock Price Growth Rate

–20% –15% –20% –15% –10% –5% 0% 5% –20% –15% –10% –5% 0% 5% Average NGDP Forecast Error: 2008–2013 Average NGDP Forecast Error: 2008–2013

NGDP Gap and Nonperforming Loans (NPL) NGDP Gap and Equity Risk Premium 16% 14%

14% GRC 13%

12% 12% GRC ITA 11% 10% 10% 8% PRT 9% 6% ESP GBR PRT FRA 8% 4% DNK BEL ITA JPN USA 7% ESP KOR ISR 2008–2013 Average

DEU 2008-2013 Average NLD JPN AUT ISR Equity Risk Premium FRA BEL NPL as % of Gross Loans NLD GBR 2% SWE CAN NZL R = 61.03% AUS 6% SWE AUS KOR R = 55.98% USA DEU 0% FIN CHE 5% NZL FIN CHE DNK CAN AUT –2% 4% –20% –15% –10% –5% 0% 5% –20% –15% –10% –5% 0% 5% Average NGDP Forecast Error: 2008–2013 Average NGDP Forecast Error: 2008–2013

Source: David Beckworth, “The Financial Stability Case for a Nominal GDP Target,” Cato Journal 39, no. 2 (2019): 419–47.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 15 I went on to more carefully test these rela- This presents a serious problem for conventional tionships and found support for causality running monetary policy. from NGDP stability to financial stability. The evi- NGDPLT is an effective way to deal with such dence, then, also points to NGDPLT being a work- ZLB experiences for two reasons. First, as noted around to incomplete financial markets.16 earlier, NGDPLT makes up for past misses and allows for more inflation flexibility over the business cycle. Specifically, NGDPLT makes inflation FACT 6: NGDPLT IS AN ANTI–ZERO countercyclical so that it would temporarily rise LOWER BOUND TOOL in a ZLB environment. This temporary surge in A big monetary policy concern is that the decline inflation serves to ease real debt burdens, as in the natural real interest rate over the past noted above, and to lower real interest rates to decade will push the Fed to the zero lower bound their market-clearing levels. NGDPLT, in short, (ZLB) more regularly. A ZLB situation arises generates the temporary rise in inflation needed when a sharp negative aggregate demand shock to escape a ZLB, something that is difficult to do contracts the economy and, as a result, lowers the with the Fed’s current inflation target. short-term natural real interest rate below zero. Figure 9 shows a counterfactual exercise These forces also pull down nominal short-term that demonstrates what NGDPLT might have interest rates until they get stuck near zero meant for inflation following the Great Reces- percent. As a result, real short-term interest rates sion. It shows, starting in mid-2009, a series of fail to reach their market-clearing levels and the counterfactual inflation forecasts conditional recession is prolonged.17 Kiley and Roberts esti- on NGDP returning to its trend path. It is based mate the US economy will hit such ZLB traps 30 on an estimated reduced-form vector autore- to 40 percent of the time going forward given the gression containing core PCE inflation, NGDP, sustained fall in the natural real interest rate.18 and the output gap.19

FIGURE 9. RETURNING TO TREND NGDP GROWTH PATH

Panel A. Actual NGDP versus Counterfactual Returns to Trend Panel B. Actual Core PCE Inﬂation versus Counterfactual Forecasts 4.0 $19.5 3.5

$17.5 3.0

2.5 $15.5 2.0

Trillions $13.5 1.5

1.0 $11.5 0.5 Percent Change from a Year Ago $9.5 0.0

2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 2-Year Recovery 3-Year Recovery 2-Year Recovery 3-Year Recovery 4-Year Recovery Actual 4-Year Recovery Actual

Note: PCE = personal consumption expenditure. Source: Reproduced from David Beckworth, “Permanent versus Temporary Monetary Base Injections: Implications for Past and Future Fed Policy,” Journal of Macroeconomics 54 (2017).

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 16 Panel A of figure 9 shows three paths of which modern macroeconomics is clear however— NGDP returning to its precrisis trend: a two-year and on which there is substantial consensus—it is path, a three-year path, and a four-year path. that policy rules have major advantages over dis- These three counterfactual paths for NGDP are cretion in improving economic performance.”20 plugged into the estimated VAR to create three For many observers, then, decision-making guided counterfactual forecasts for core PCE inflation. by monetary policy rules is a desired feature for The inflation forecasts are shown in panel B central banking. of figure 9. Although temporary, inflation is nota- This understanding emerged over the past bly higher than both the actual inflation rate that half century and is based on several influential occurred and the 2 percent target rate under each arguments. First, monetary authorities have a of the counterfactual NGDP return paths. The hard time committing to time-consistent behavior inflation rate would get as high as 3.8 percent for in the absence of rules. That is, with full discre- the two-year path and 3.2 percent for the four- tion, it is easier for central bankers to make prom- year path. Overall, counterfactual inflation paths ises than to keep them as circumstances change.21 would average about 2.5 percent during the Second, monetary authorities face “long and vari- catch-up periods. This compares to an actual able” lags in the conduct of monetary policy and average inflation rate of 1.5 percent. Given that therefore can be inadvertently destabilizing to the the ZLB was binding during this time, the short- economy, if not constrained by simple rules.22 term real interest rate would have fallen further Third, central bankers are subject to the same cog- in these three scenarios than it actually did with nitive biases that afflict any human decision- the Fed’s inflation target. The recovery would making process. Monetary policy rules guard have been stronger. against such cognitive biases.23 Finally, historical The second reason NGDPLT is an effective periods when the US monetary policy acted in a tool is that it creates expectations that should pre- non-rule- like fashion also happened to be periods vent the ZLB from arising at the outset. That is, if when there was less macroeconomic stability.24 the public understands NGDPLT and finds it cred- Taylor showed how monetary policy rules ible, then people will have less incentive to cut that worked and were robust to different situa- back on spending in the first place. Put differently, tions could be characterized by a simple reaction the public’s expectation of stable total dollar function.25 Specifically, monetary policy that sys- spending growth will become self-fulfilling. This tematically responded to deviations of inflation feature makes NGDPLT an “ounce of prevention from target and deviations of real GDP from is worth a pound of cure” solution for the ZLB. potential real GDP appeared to work well at stabilizing the economy. This feedback rule became prominently known as the Taylor Rule, and most FACT 7: NGDPLT IS A WAY TO DO monetary policy rules today are some version of it. RULES-BASED MONETARY POLICY A Taylor Rule, however, can be modified so A widely held view among monetary economists that monetary policy responds to an NGDP is that central bankers perform best when their target. This is demonstrated later in the paper behavior is constrained by monetary policy rules. and builds on the work of Bennett McCallum, As John Taylor noted, “If there is anything about who shows that NGDP targeting can provide a

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 17 rules-based approach to monetary policy.26 An growth seen in figure 4 was a consequence of the NGDP target, moreover, is arguably closer in Fed’s reaction function changing so that it began spirit to Milton Friedman’s call for simple rules to systematically respond to changes in fore- since it responds to only a single, nominal tar- casted nominal income growth.29 His findings get.27 Recent work corroborates this by showing suggest that the Fed was doing a rules-based in a standard New Keynesian model how a simple approach to monetary policy during the period monetary policy rule that targets NGDP often 1985–2007 implicitly based on something like an improves economic outcomes relative to other NGDP level target. monetary policy rules.28 Both theoretically and empirically, then, Empirically, Joshua Hendrickson provides NGDPLT has been shown to provide a rule-like evidence that the Great Reanchoring of NGDP approach for the conduct of monetary policy.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 18 II. FEARS ABOUT NOMINAL GDP LEVEL TARGETING

FEAR 1: CHANGES IN POTENTIAL potential real GDP.30 This would give the Fed REAL GDP WILL CREATE PROBLEMS enough time to be certain of trend changes and, as FOR NGDPLT noted earlier, should not cause problems in the As noted above, one desirable feature of NGDPLT short run because of the increased inflation is that it allows more inflation flexibility over the flexibility. short run. Over the medium-to- longer run, how- A second approach, championed by George ever, the trend inflation rate will be tied down to Selgin, is simply to allow changes in potential real the difference between the targeted growth rate GDP to translate into changes in the trend infla- 31 for NGDP and the trend real GDP growth. In tion rate. Doing so would still provide a nominal anchor since total dollar income growth is other words, since %ΔPY ≈ %ΔP + %ΔY, the trend inflation rate that emerges from an NGDP target stabilized. More importantly, though, this is as follows: approach avoids the potentially destabilizing effects on the economy that could arise if the Fed Trend Target Trend %ΔP ≈ %ΔPY − %ΔY . (4) were to correct such changes in the trend inflation rate. The trend real GDP growth rate, in turn, is It is hard to know in real time if a supply determined by “potential” real GDP growth, the shock is temporary or permanent since both fastest sustainable growth of real GDP given kinds push, on impact, inflation and output in technology and resource constraints. Potential opposite directions. Only a permanent supply real GDP growth is therefore a big determinant of shock can change potential real GDP, but it can the trend inflation rate under NGDPLT. take several years to know the temporal status of One fear about NGDPLT is that once it is up a supply shock. The concern is that by the time and running, potential real GDP may change and the Fed figures out a supply shock is permanent, cause the trend inflation rate to be different than the change in the trend inflation rate it has cre- what Fed officials expected it to be when they ated is likely to have changed the inflation expec- established the NGDP target. tations that are priced into wage, debt, and other There are several ways to address this con- financial contracts. If so, changing the trend cern. Some observers, such as Jeffrey Frankel, inflation rate after the fact could exacerbate the would have the Fed recalibrate its NGDP target business cycle because of sticky prices in these every three to five years to account for changes in contracts. This idea is further illustrated in section

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 19 FIGURE 10. COUNTERFACTUAL 5 PERCENT NGDP TARGET

Counterfactual 5% NGDP Target Implied Trend Inﬂation Since 1960 8%

6% Potential Real 5% GDP Growth

inﬂation rate 3% Average Inﬂation 2%

1% Implied Trend Inﬂation

0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

Source: FRED database and author's calculations. three of the appendix using a Monetary Policy– FEAR 2: DATA REVISIONS Phillips Curve model. MAKE NGDPLT AN Figure 10 provides an example of what this IMPRACTICAL FRAMEWORK second approach might look like. It assumes a NGDP is a quarterly measure of nominal eco- 5 percent NGDP target and subtracts from it the nomic activity that is released by the Bureau of year-on-year growth rate of the Congressional Economic Analysis in three stages: the advance, Budget Office’s estimates of potential real GDP. preliminary, and final estimates. The estimates The result is the black line and represents a coun- are sequentially released three months after the terfactual implied trend inflation rate that quarter ends. Typically, each estimate is a revi- changes over time. This inflation series averages sion of the previous one. The bureau further 1.96 percent over the full sample and gets as low revises its estimate of NGDP in the years that as 0.25 percent in 1967 and as high as 4.00 percent follow. in 2010. There is no runaway inflation or sharp These revisions cause some observers to deflation, but instead increased inflation flexibil- question whether NGDPLT is a practical frame- ity that is anchored near 2 percent over the long work for monetary policy. They question how run. To the extent the Congressional Budget the Fed can guide monetary policy according to Office’s estimate of potential real GDP is not an indicator that is revised so often. Will not this completely exogenous to the business cycle, as uncertainty over NGDP create more macroeco- some recent studies suggest, a credible NGDP nomic instability? target that stablizes total dollar spending and This fear can be addressed in three ways. lessens business cycles might actually lead to First, given the data revision problem, one can even less inflation volatility than depicted in show that NGDPLT is actually easier to imple- 32 figure 10. ment than the common approach of using a Taylor

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 20 Rule to guide monetary policy. The standard Tay- the NGDI and the CI-BI to calculate both a year- lor Rule takes the following form: on-year growth rate and an annualized quarter- on-quarter growth rate. These real-time esti- i = i* + φ π! + φ y! , (5) t t π t y t mates are fairly close to the actual, final growth rate of NGDP. For example, these real-time mea where the target policy interest rate, it, is set * sures would have indicated in real time to the equal to a baseline neutral interest rate, it , and responds to deviations of inflation from its target, Federal Reserve in early to mid-2008 that total dollar spending growth decline was accelerating. π!t , and the output gap, y!t . The output gap mea sures the amount of slack in the economy and is NGDPLT could have used these metrics to better defined as thepercent difference betweenactual inform the stance of monetary policy during this real GDP and potential real GDP. time. The Taylor Rule requires Fed officials to Another approach to real-time forecasts of know both real GDP and potential real GDP in real NGDP is to use the higher-frequency “nowcasting” time. An NGDP target requires Fed officials to techniques created by the Atlanta and New York know only NGDP in real time. The information Federal Reserve Banks. These measures provide requirements are therefore greater for the stan- multiple real-time updates of real GDP in the cur- dard Taylor Rule than for an NGDP target. This is rent quarter. The nowcasting models could be eas- an important distinction. Athanasios Orphanides ily adjusted to also provide “nowcasts” for NGDP shows how real time uncertainty over the output in the current quarter to help guide a Fed that used gap contributed to the high inflation of the 1970s.33 NGDPLT. Beckworth and Hendrickson show that output Relatedly, there are various “big data” endeav- gap uncertainty continues to be a problem for Fed ors to measure real-time transactions that could policy and that an NGDP target provides a work- inform monthly estimates of NGDP. For example, around solution to this knowledge problem.34 the JPMorgan Chase & Co. Institute has a monthly A second way to address this fear is to note aggregate measure of debit and credit card spend- that there are ways to get better real-time esti- ing for over 64 million anonymized Chase custom- mates of NGDP. Boragan Aruoba and his coau- ers across 14 metro areas. Researchers at the Fed thors show that GDI often is subject to less revi- have also begun constructing their own measure sion than GDP and use it to create a measure of real-time transactions using similar payment called GDP-Plus, a more reliable real-time esti- methods. The Fed could use such data to get 36 mate of GDP.35 Along the same lines, figure 11 monthly estimates of NGDP. provides two real-time estimates of NGDP using A final way to address this fear about data real-time estimates of NGDI and another real- revisions is to use NGDP forecasts. Fed officials time proxy for the NGDP. This second measure is could target the NGDP forecast as suggested by 37 the summation of the growth rate of the Philadel- Scott Sumner. He suggests creating an NGDP phia Fed’s coincident indicator for the US econ- futures market for the Fed to target, but other omy and the expected “break-even” inflation rate more modest forecasting approaches could also implied from a five-year inflation-indexed Trea be adopted. The Fed could use, at the monthly sury bond, hereafter called the CI-BI. Figure 11 frequency, the year-ahead Blue Chip NGDP takes the average of the real-time estimates of forecast or the year-ahead household nominal

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 21 FIGURE 11. REAL-TIME ESTIMATES VERSUS ACTUAL GDP

Percent Change from a Year Ago Annualized Quarterly Growth Rate 8 10

8 6 Actual NGDP 6 Actual NGDP 4 4

2 2 Real-Time Estimate 0

0 –2 Real-Time Estimate –4 –2 –6

–4 –8 2005 2006 2007 2008 2009 2010 2011 2012 2005 2006 2007 2008 2009 2010 2011 2012

Source: FRED database and author’s calculations. income forecast from the University of Michi- tive and be explained to the public as a policy that gan consumer sentiment survey. Alternatively, aims to stabilize their dollar income growth. This at the quarterly frequency, the Fed could use is something the public can understand and pre- the year-ahead NGDP forecast from the Phila- sumably appreciate. delphia Federal Reserve Bank’s Survey of Pro- Some narrower versions of an NGDP target fessional Forecasters (SPF). The Fed could use explicitly call for the Fed to stabilize labor any one of these measures in a Taylor-like Rule income growth.38 In this case, the Fed could to help guide the setting of the Fed’s target describe its framework as a policy that aims to interest rate. The final section of the paper stabilize wage and salary growth. In that vein, the illustrates one such rule that uses the SPF NGDP Fed could point to charts like figure 12, which forecast. shows the Atlanta Fed’s nominal wage tracker and the year-ahead household nominal income forecast from the University of Michigan con- FEAR 3: THE PUBLIC WILL NOT sumer sentiment survey as indicators of wage UNDERSTAND NGDPLT inflation that would guide monetary policy. Pre- Most people do not know what NGDP is, let alone sumably, over time, indicators like these would what it means to devise a target based on it. If the become as common to public discourse on mon- Fed announced NGDPLT as the new monetary etary policy as is now the case with the Consumer policy framework, could it be effective given the Price Index. In general, the idea would be to shift public’s unfamiliarity with it? Why abandon a the public’s focus from changes in the cost of liv- framework the public does understand—inflation ing to changes in dollar income growth. targeting—for one that is not understood? To be clear, this framing would not be with- This fear of NGDPLT can be addressed by out challenges. Just as the public often confuses recalling that NGDPLT stabilizes the growth path relative price changes for price level changes in of total dollar income. That is, this target can be their assessment of inflationunder the current framed from a nominal income targeting perspec- framework, it is likely that similar confusion could

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 22 FIGURE 12. LABOR INCOME GROWTH INDICATORS

Labor Income Growth Indicators 7.0%

6.0% Atlanta Fed Wage Tracker 5.0%

4.0%

3.0% Median Increase

2.0% Expected Household Income Growth Over Next Year 1.0%

0.0%

2011 1990 1993 1996 1999 2002 2005 2008 2014 2017 Note: Data for the Atlanta Fed Wage Tracker were unavailable from June 1995 to October 1996. Source: Author’s calculations based on Federal Reserve Bank of Atlanta, “Wage Growth Tracker”; University of Michigan, “Surveys of Consumers.” arise between relative and aggregate income from Congress. Likewise, the broader public changes under an NGDPLT framework. This chal- would also probably be fairly receptive to the idea lenge, though, is common to any monetary policy of the Fed allowing temporarily faster income framework and underscores the importance of growth in order to hit an NGDP level target. good communication by the central bank. One big advantage of NGDPLT is that it should be relatively easy to marshal support for it FEAR 4: NGDPLT DOES NOT in a severe recession. This is not the case for a SATISFY THE “PRICE STABILITY” price stability target. Former Fed chair Ben Ber- PART OF THE DUAL MANDATE nanke, for example, had a hard time explaining to Some observers worry that NGDPLT fails to sat- Congress why the Fed was trying to generate isfy the price stability part of the Fed’s dual man- additional inflation with its unconventional date legislated by Congress. While there is monetary policy programs. There are good rea- increased inflation flexibility under NGDPLT, one sons for some temporarily higher inflation dur- can still view this framework as satisfying the ing a severe downturn, including the explana- price stability mandate in two ways. First, an tions outlined earlier. But they are not intuitive NGDP level target creates a nominal anchor that and often seem unfair to a public already bur- determines the dollar size of the economy, a fea- dened with a contracting economy. Had Bernanke ture consistent with the spirit of the price stability instead pointed to a chart like figure 12 and advo- portion of the mandate. In partic u lar, nominal cated for temporarily faster growth in salaries and income growth is stabilized so that wage and sal- wages, he might have received a warmer reception ary growth remain well anchored. Second, over

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 23 longer periods, the trend inflation ratewill also be responds over time to the typical TFP shock. anchored by the NGDP target. As seen in figure 10, The solid line is the point estimate, and the the average inflation rate over the period 1960– dashed lines show confidence intervals. 2018 in a counterfactual US economy where there Figure 13 shows that positive TFP shocks had been a 5 percent NGDP target would be close typically created disinflationary pressures, as to 2 percent. There is no “Great Inflation” of the seen in panels A and B. This is a clear example of 1970s in this counterfactual world. Trend infla- a supply shock pushing real economic activity tion, in other words, is bound over the long run and inflation in opposite directions. while being more flexible over the short run. If The Fed’s typical response to such positive adopted, an NGDP level target would therefore TFP shocks and the disinflation they created was avoid any explosive inflationary or deflationary to lower the federal funds rate, as seen in panel C experiences. of figure 13. As Selgin, Beckworth, and Bahadir It is worth recalling that there is often a trad- note, however, the temporary acceleration of the eoff between price stability and full employment TFP growth rate implied by the one-time increase in the dual mandate. Supply shocks, in partic ular, in the TFP level should lead, all else equal, to a create a challenge because they push output and similar temporary rise in the neutral real federal inflation in opposite directions.39 Negative oil funds rate.43 This is the level of the inflation- shocks, for example, can temporarily raise the adjusted federal funds rate that does not cause inflation rate while lowering economic activity. If tightening or loosening of monetary policy. the Fed responded to the higher inflation by tight- Panel D of figure 13 reports the implied ening, it would further weaken the economy. Ben change in this neutral real federal rate given the Bernanke, Mark Gertler, and Mark Watson show positive TFP shock against the implied change in that the Fed has effectively done just that in its the actual real federal funds interest rate. The systematic response to oil shocks since the Fed’s response to this supply shock pushes these 1970s.40 This has worsened the business cycle. two measures of interest rates in opposite direc- Positive supply shocks create a similar tions. This easing of monetary policy gives rise to dilemma for the Fed. They raise economic activity a boom-bust cycle in real economic activity, as and temporarily lower inflation. If the Fedwere to seen in panels E and F. These panels report a tem- offset the lower inflation, it might turn a sustain- porary decline in the unemployment rate and a able expansion based on the positive supply shock temporary realization of a positive output gap. into an unsustainable boom. Figure 13 illustrates In short, figure 13 shows that the Fed’s sys- this point by replicating an empirical exercise tematic attempts to offset inflation movements found in Selgin, Beckworth, and Bahadir.41 created by productivity shocks intensify the busi- This figure shows, using an estimated vec- ness cycle. This example demonstrates the trad- tor autoregression, how the Fed systematically eoffs the Fedfaces in managing its dual mandate. responded to total factor productivity (TFP) A key takeaway, then, is that the dual man- shocks between 1954 and 2008 and the effects date actually requires inflation to fluctuate at this response had on the economy.42 The figure times if it is to be implemented in a stabilizing does this by reporting impulse response func- manner. NGDPLT provides this inflation flexibil- tions, which reveal how each economic variable ity exactly when it is needed during the business

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 24 FIGURE 13. FED RESPONSE TO A TYPICAL PRODUCTIVITY SHOCK (1954 Q1–2008 Q3)

Panel A. Total Factor Productivity (TFP) Panel B. PCE Inﬂation Rate 1.8 0.10

1.6 0.08

1.4 0.06 0.04 1.2 0.02 1.0 0.00 0.8 –0.02 0.6 –0.04 Percentage Points 0.4 –0.06 Percent Change in TFP Level 0.2 –0.08 0.0 –0.10 1 357911 13 15 17 19 21 23 1357911131517192123 Quarters after TFP Shock Quarters after TFP Shock

Panel C. Federal Funds Rate Panel D. Actual versus Neutral Real Federal Funds Rate 0.60 0.40

0.30 0.40 0.20 Implied Neutral Real Federal Funds Rate 0.20 0.10

0.00 0.00

–0.20 –0.10 –0.20 Percentage Points –0.40 Percentage Points Implied Real Federal Funds Rate –0.30

–0.60 –0.40

–0.80 –0.50 135 7 911131517192123 13579111315 17 19 21 23 Quarters after TFP Shock Quarters after TFP Shock

Panel E. Unemployment Rate Panel F. Output Gap 0.15 0.6

0.10 0.5 0.05 0.4 0.00 0.3 –0.05 0.2 –0.10 0.1 –0.15 0.0 Percentage Points

Percentage Points –0.20 –0.1 –0.25

–0.20 –0.2 –0.35 –0.3 1357911131517192123 1357911131517192123 Quarters after TFP Shock Quarters after TFP Shock

Source: Figure 13 is based on figure 6 from George Selgin, David Beckworth, and Berrak Bahadir, “The Productivity Gap: Monetary Policy, the Subprime Boom, and the Post-2001 Productivity Surge,” Journal of Policy Modeling 37, no. 2 (2015): 196.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 25 cycle while still providing a secure nominal growth to be manifested in higher real growth anchor over the medium-to-long run. and lower inflation. An inflation-targeting central bank, on the other hand, would lower interest rates just as the natural interest rate was ris- FEAR 5: NGDPLT DOES NOT ing.46 NGDPLT, in short, would empower the Fed ADDRESS FINANCIAL to keep its target interest rate closer to the natural STABILITY CONCERNS interest rate and thereby better avoid the buildup The Great Recession created a new appreciation of financial imbalances in the first place. This is for the role financial fragility plays in the busi- NGDPLT’s preventive feature. ness cycle. Some observers fear NGDPLT is just Should a financial cycle emerge anyway, another way of doing monetary policy that fails NGDPLT also has a curative feature that mini- to address these new financial stability concerns. mizes financial distress. This feature is the This framework, however, actually has both pre- improved risk sharing between creditor and debtor ventive and curative features that better align outlined in fact 5. To recap that section, the idea is monetary policy with financial stability. that the countercyclical inflation created by an The preventive features of NGDPLT begin NGDP level target will cause real debt burdens to with the observation that financial boom-bust change in a procyclical manner. As a result, debtors cycles in advanced economies often start with will benefit during recessions and creditors will above-average economic growth and below- benefit during booms. Fixed nominal-price loans average inflation. That is, many financial cycles will consequently act more like equity than debt, start with improved fundamentals, including and this will promote financial stability. improved productivity growth, before turning In summary, NGDPLT provides both preven- into a boom period of unsustainable growth in tive and curative features that are not currently asset prices and credit.44 One explanation for found in the Fed’s inflation targeting framework this pattern is that central banks respond to the and that would cause monetary policy to operate in low inflation in the early stages of the financial a manner that better supports financial stability. boom-bust cycle by easing monetary policy. This easing helps turn a sustainable expansion into an unsustainable boom. Lawrence Christiano and FEAR 6: OVERSHOOTS OF AN NGDP his coauthors formally demonstrate this process, TARGET WOULD BE POLITICALLY while Selgin, Beckworth, and Bahadir show that TOUGH TO CORRECT it helps explain the housing and credit boom of Another fear some folks have with NGDPLT is that the early to middle portion of the first decade of it will be politi cally tough to correct an overshoot the 21st century.45 of the target. That is, a level target requires a course A key part of this financial boom-bust story correction back to the targeted growth path if is that low inflation caused by real economic NGDP rises above it. Can the Fed actually engineer gains, particularly productivity growth, is accom- a tightening without politi cal blowback? modated by the monetary policy. NGDPLT would This concern is more academic than practi- avoid this temptation since it does not worry cal because the big challenge since the 1980s has about inflation over the short run. It would allow usually been NGDP falling below its trend, not a surge in current and expected productivity above it. Nonetheless, there are several ways to

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 26 address this unease. First, as mentioned earlier, if more modest, like price level targeting or average the public understands NGDPLT and finds it cred- inflation targeting? While these frameworks ible, then it will have less incentive to engage in would incorporate the properties that are useful excessive spending that pushes NGDP above tar- at the ZLB, they still suffer from all the chal- get in the first place since the Fedwill be expected lenges outlined earlier with price stability tar- to offset it. The public’s expectation of stable gets. Specifically, they would still be subject to total dollar spending growth, then, will become the supply shock confusion problem and not con- self-fulfilling. This lowers the likelihood of the sistently deliver the real-time countercyclical Fed having to correct an overshoot of the target. inflation of NGDPLT that causes it to enhance Second, even if there were an overshoot, the Fed financial stability. need not engineer an outright contraction of Imagine, for example, if the Fed had been fol- NGDP. As seen in panel B of figure 2, the Fed could lowing a price level target going into 2008. The simply slow down the rate of NGDP growth until surge in inflation that year probably would have its level returned to the targeted growth path. caused the Fed to do several rate hikes, given its Finally, the existing monetary policy framework concerns about inflation at that time. NGDPLT faces its own version of this fear. Yet policymakers would have seen through the temporary spike in have found ways to tighten monetary policy inflation and instead focused on total dollar when needed. It should be no different under spending.47 Alternatively, imagine a future NGDPLT. where current and expected productivity growth is rapidly accelerating and pushing down inflation. A price level target would force the Fed to FEAR 7: NGDPLT IS TOO RADICAL. ease even though the natural interest rate would WHY NOT SETTLE FOR PRICE be rising. It would be similar in spirit to the con- LEVEL TARGETING? fusion the productivity boom of the period 2002– A final concern is that NGDPLT is too radical a 2004 created for the Fed.48 For these reasons, change for the Fed. Why not settle on something NGDPLT is a more robust framework for the Fed.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 27 III. FUNCTIONALITY: HOW TO IMPLEMENT NOMINAL GDP LEVEL TARGETING

o how would the Fed actually implement Philadelphia Federal Reserve Bank’s SPF. The NGDPLT? Many of the studies that for- baseline monetary policy rule is as follows: mally examine NGDP targeting do so using S i i* NGDP Forecast Gap, (6) = + λ a modified Taylor Rule that replaces the inflation t t 1 t, t+h gap and output gap terms with some measure where the target policy interest rate, it, is set of NGDP deviating from its targeted value. To * equal to a baseline neutral rate, it , and responds calculate this NGDP gap, one could use a real- to the NGDP forecast gap, the percent difference time NGDP measure listed earlier—such as GDP between the SPF NGDP forecast at horizon t + h Plus, nowcasting, big data, or labor income—and and its targeted value. Since the SPF is at a quar- subtract it from the target NGDP growth rate. terly frequency, a year-ahead forecast sets h = 5. This would be a straightforward extension of the The framework being considered is a level Taylor Rule. target, so the baseline rule also needs to account A slightly diff erent tactic is implemented here. for past misses. Equation (6) is therefore This paper follows the “targeting the forecast” amended so that it also has a makeup term: approach of Lars Svensson, in which the stance of * Forecast Gap Level Gap monetary policy is adjusted so that the forecast of it = it + λ1NGDPt, t+h + λ2NGDPt , (7) 49 NGDP converges to its targeted value. This idea is Level Gap where NGDPt is the percent difference not new and can be dated back to at least James between the actual and targeted level value of Tobin, who made the case for targeting the forecast NGDP at time t. This term, inde pen dent of the of NGDP: “I think it would be preferable for the forecast gap, forces the stance of monetary pol- Federal Reserve to announce target ranges for MV icy to change if NGDP drifts off its targeted growth a year ahead, indeed several years ahead.”50 growth path. Sumner more recently has promoted target- Figure 14 illustrates what this rule might have ing the NGDP forecast using an NGDP futures implied for the Fed’s target interest rate for the market that the Fed would set up and run.51 A period 1985–2018 given several assumptions about more modest forecasting approach is adopted monetary policy. here, where existing NGDP forecasts are used in First, the two-year treasury yield is used for a modified Taylor Rule. Specifically, this paper Target the baseline neutral rate. Second, Δ%NGDP is uses the year-ahead NGDP forecast from the set to 5.5 percent for the period 1985–2008 and

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 28 FIGURE 14. AN NGDP LEVEL TARGET RULE

15%

10%

0% Target interest rate λ1 = 0.5, λ2 = 0.5

–5% λ1 = 0.0, λ2 = 1.0

λ1 = 0.5, λ2 = 1.0 Federal Funds Rate –10% 1985 1989 1993 1997 2001 2005 2009 2013 2017

Source: FRED database and author’s calculations.

FIGURE 15. STICKY FORECAST LEVEL

$22

$20

$18 Sticky Forecast

$16 Trillions $14 Nominal GDP

$12

$10

$8 1997 2000 2003 2006 2009 2012 2015 2018

Source: FRED database and author’s calculations.

4 percent for the period 2009–2018 to reflect the The sticky forecast measure is the average actual trend NGDP growth rates allowed by the level of NGDP the public expected during the five Fed during those periods. Third, the NGDP level years leading up to a certain point in time. It can gap is calculated as the difference between the be viewed as the expected NGDP level the Fed actual and sticky forecast level measure of implicitly created through its monetary policy.53 NGDP.52 Figure 15 shows the sticky forecast series in level

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 29 FIGURE 16. STICKY FORECAST GAP

NGDP Deviation from Sticky Forecast Path 6%

–2% Percent –4%

–6%

–8%

–10% 1997 2000 2003 2006 2009 2012 2015 2018

Source: FRED database and author’s calculations.

form, and figure 16 shows the percent difference however, that had the Fed followed equation (7), between it and actual NGDP. The sticky forecast the economy would have responded very differ- Level Gap gap in figure 16 is what enters the NGDPt ently because this rule is targeting the level of term in equation (7). It is worth noting that this NGDP. As mentioned earlier, a credible NGDP setup requires no knowledge of any natural rate level target is likely to prevent a ZLB experience variables. It requires only forecasts and the target from occurring in the first place. growth rate of NGDP. Still, figure 14 shows that an NGDP rule cre- Figure 14 reveals that this rule would have ates reasonable policy prescriptions given exist- created a target interest rate path that fits the ing economic conditions. It implies that it should standard narrative of US monetary policy. That not be too hard for the Fed to transition to such a is, the rule indicates fairly reasonable monetary rule since it would not be too radically diff erent.54 policy during the Great Moderation period, but Moreover, as Joe Gagnon and Jeffrey Frankel note, then effectively too tight monetary policy when not only could NGDPLT be incorporated fairly the ZLB hit in late 2008. It also indicates that the easily by the Fed, but it should also make the Fed’s tightening of policy that started in 2015 was Fed’s existing tools more effective.55 Implement- a bit premature. ing NGDPLT is a functionally feasible and desir- The counterfactual interest rate path in fig- able objective.56 ure 14 takes the economy as given. It is likely,

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 30 IV. CONCLUSION

This paper has addressed the basic facts and are as follows. First, as suggested by Frankel, the fears of NGDPLT. It has shown that this frame- Federal Open Market Committee’s Summary of work provides a firm nominal anchor, is a Economic Projections should have as its top line velocity-adjusted money supply target, is a work- an NGDP forecast.57 The Board of Governors around to the supply shock and incomplete could also add an explicit NGDP rule to its web- financial market problems, is an anti-ZLB tool, site along with the existing benchmark mone- and is a rules-based approach to monetary policy. tary policy rules. FOMC officials could also start The paper has also shown that the standard con- citing real-time indicators of NGDP in their cerns about NGDPLT, while understandable, speeches and interviews. These moves would have viable fixes. introduce and condition the public to NGDP The real challenge, then, is likely to be the thinking and build up credibility for an eventual transition to an NGDP level target rather than the NGDP level target. After some time of following framework itself. While a thorough discussion of these and other similar conditioning steps, the the transition process to NGDPLT is beyond the Fed could then announce a transition to an NGDP scope of this paper, some basic transition steps level target.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 31 APPENDIX. SUPPLY SHOCKS IN A MONETARY POLICY–PHILLIPS CURVE MODEL

PC This appendix outlines a graphical model that PCPC : : πt = Etπt+1 + φΔy!t + et illustrates how an NGDP target better handles MP : π + Δy = NGDPTarget temporary and permanent supply shocks than t t P P FE does an inflation or a price level target. The FEE : : Δyt = Δyt−1 + et model is a version of the dynamic Aggregate

Demand–Aggregate Supply (AD-AS) framework In the PC equation, πt is the current infla- found in the Cowen-Tabarrok macroeconomic tion rate, Etπt+1 is the expected inflation rate next textbook that has been modified to better reflect period, Δy!t is the change in the output gap, and PC the goals of NGDP targeting and the thinking of the PC shock, et , is an inverse temporary supply modern central bankers.58 The model here makes shock.60 The shock is temporary since it does not two changes. First, there is a monetary policy carry over to the next period. The change in the P (MP) curve rather than an AD curve; and second, output gap is defined as Δy!t = Δyt − Δyt , where P there is a Phillips Curve (PC) instead of a short- Δyt is the change in the log of real GDP and Δyt run aggregate supply curve. Consequently, this is the change in the log of potential real GDP. framework is called the MP-PC model. The The MP equation assumes the NGDP target details are sketched out below, followed by the is credible and thus the monetary authority can two applied scenarios of temporary supply easily adjust the NGDP growth rate—π t + Δyt—to shocks and permanent supply shocks in a graphi- its targeted value. The FE equation shows that the P cal format. growth of potential real GDP, Δyt , is equal to last period’s potential real GDP plus a permanent FE supply shock, et . Given that monetary policy is I. THE MP-PC MODEL credible, expected inflation becomes the differ- The MP-PC model is what Blanchard would call a ence between the targeted NGDP growth rate toy model, but it illustrates clearly the supply and the expected potential real GDP growth: shock challenges faced by central banks.59 This Target P model also demonstrates nicely how NGDP tar- Etπt+1 = NGDP − EtΔyt+1. (A1) geting is a work-around to this problem. The model consists of a PC equation, an MP equation, Figure A1 illustrates this model in graphical and a full-employment (FE) equation: form with the assumption that the current and expected future value of potential real GDP

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 32 FIGURE A1. EQUILIBRIUM IN MP-PC MODEL PC rarily, since the shock, et , does not carry over to p the next period or change the expected inflation ∆yt

πt rate. The left side of figure A1 shows specifically ~ PC what happens when there is a negative version of πt = 2% + φΔyt + et this shock. The PC shifts inward to point b, temporarily raising inflation to 3 percent and temporarily lowering real GDP growth to 1 percent. The NGDP-targeting central bank allows this tempo- 2% a rary change in the composition of the NGDP because the overall targeted growth rate of Δngdpt = Δyt + πt = 4% 4 percent is still maintained. Now imagine the monetary authority had instead been targeting inflation at 2 percent. The

2% ∆yt initial equilibrium at point a would be fine, but not the new one at point b. Here, inflation is above Note: PC = Phillips Curve; MP = monetary policy. target. The inflation-targeting central bank would have to respond as seen in the bottom left panel growth rate is 2 percent. The central bank is also of figure A2. There, the central bank lowers assumed to be targeting NGDP at 4 percent. That NGDP growth—an inward shift of the MP curve— results in a current and expected inflation rate of until the inflation target is reached at point c. 2 percent. Short-run and long-run equilibrium is This response, however, further contracts an at point a, where the PC, MP, and FE curves all already weakened economy. intersect each other. The central bank can shift the Conversely, consider a temporary supply MP curve by changing the targeted NGDP growth shock that is positive. As seen on the right-hand rate. For a given NGDP target, any point on the MP side of figure A2, the PC shifts outward to point b, curve reflects a combination ofπ t + Δyt that sums temporarily lowering inflation to 1 percent and to the target NGDP growth rate. In the case of fig- temporarily raising real GDP growth to 3 percent. ure A1, any combination sums to 4 percent. Again, the NGDP-targeting central bank allows Shifts in the PC are caused by changes in this temporary change in the composition of the either the expected inflation rate or a temporary NGDP because the overall targeted growth rate of supply shock. Shifts in the FE curve are created by 4 percent is still maintained. permanent supply shocks. The two types of supply An inflation-targeting central bank, on the shocks and what they mean for NGDP targeting other hand, would respond to the positive supply versus inflation targeting are considered next. shock as seen in the bottom right panel of figure A2. There, the central bank raises NGDP growth— an outward shift of the MP curve—until the infla- II. A TEMPORARY SUPPLY SHOCK tion target is reached at point c. This response, Figure A2 shows how a temporary supply shock however, further stimulates an already strength- will affect this economy. Starting in equilibrium, ened economy. this supply shock shifts the PC, but only tempo-

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 33 FIGURE A2. TEMPORARY SUPPLY SHOCKS: NGDPT TARGETING VERSUS INFLATION TARGETING

Economy in Equilibrium

p ∆yt πt ~ PC πt = 2% + φΔyt + et

2% a

Δngdpt = Δyt + πt = 4%

2% ∆yt

A Temporary Negative Supply Shock A Temporary Positive Supply Shock 4% NGDP Target 4% NGDP Target

p p ∆yt ∆yt πt πt ~ PC ~ PC πt = 2% + φΔyt + et πt = 2% + φΔyt + et 3% b

2% a 2% a

Δngdpt = 4% 1% b Δngdpt = 4%

1% 2% ∆yt 2% 3% ∆yt

2% Inﬂation Target 2% Inﬂation Target

p p ∆y ∆yt t πt πt ~ PC ~ PC πt = 2% + φΔyt + et πt = 2% + φΔyt + et 3% b

a 2% c a 2% c Δngdpt = 6.5% Δngdpt = 2% 1% b

0% 1% 2% ∆yt 2% 3.5% ∆yt

This discussion highlights that inflation tar- including ones caused by supply shocks. Ideally, geting, in general, leads to increased volatility in monetary authorities would ignore changes in real GDP in response to temporary supply inflation caused by supply shocks and respond shocks. It does so because it causes the central only to changes in inflation caused by demand bank to worry about all inflation movements, shocks. This is hard to know in real time, as noted

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 34 FIGURE A3. PERMANENT SUPPLY SHOCKS: FIXED VERSUS VARYING NGDP TARGET

Economy in Equilibrium p ∆yt πt ~ PC πt = 2% + φ∆yt + et

2% a

∆ngdpt = ∆yt + πt = 4%

2% ∆yt

A Permanent Negative Supply Shock A Permanent Positive Supply Shock Fixed 4% NGDP Fixed 4% NGDP

p p ∆y ∆y t π t π ~ PC t t πt = 3% + φ∆yt + et ~ PC πt = 1% + φ∆yt + et 3% b

2% a 2% a

∆ngdpt = 4% 1% b ∆ngdpt = 4%

∆y 1% 2% t 2% 3% ∆yt

Varying NGDP Target Varying NGDP Target (Aiming for 2% πt) (Aiming for 2% πt) p p ∆y ∆y t π t π ~ PC t t πt = 3% + φ∆yt + et ~ PC πt = 1% + φ∆yt + et 3% b

c 2% 2% ∆ngdpt = 5.5%

1% b

∆ngdpt = 3%

0% 1% ∆yt 3% 3.5% ∆yt previously. Moreover, even if a central bank could III. A PERMANENT SUPPLY SHOCK divine the sources of inflation movements and Figure A3 considers the case of the permanent respond only to demand shocks, this approach FE supply shock, et . Starting in equilibrium, this would amount to an NGDP target. So why not shock shifts the FE curve as it changes the just do NGDP targeting? potential real GDP growth rate. The left side of

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 35 figure A3 shows a negative version of this shock. been ignoring a permanent supply shock, it is It shifts the FE curve inward to point b and low- likely that inflation expectations will have ers potential real GDP growth to 1 percent. The already adjusted to the new level of potential real negative supply shock also shifts the PC inward GDP growth. If so, the PC will have already since the shock causes the expected inflation rate shifted. to rise to 3 percent for the given NGDP target of This is the story depicted at the top left panel P 4 percent. That is, Et πt+1 rises as EtΔyt+1 declines of figure A3. Both the FE curve and the PC have via equation (A1). adjusted, and the 4 percent NGDP target is left This change in the expected inflation rate unchanged. As a result, the trend inflation rate occurs immediately in the model since it is under- changes to 3 percent. The bottom left panel stood that the negative supply shock is perma- shows what happens if the central bank decides nent. In practice, the updating of inflation expec- to adjust its NGDP target so that the trend infla- tations may take a while since it is not clear in real tion rate stays at the original 2 percent. This time whether a supply shock is temporary or per- requires lowering the target NGDP growth manent. The PC’s shift inward will occur eventu- rate—an inward shift of the MP curve—until the ally, however, for a fixed NGDP target. 2 percent inflation is hit. Given the new PC with The question then becomes whether the 3 percent expected inflation, this lowering of central bank should adjust its NGDP target in NGDP growth leads to a further contraction of response to a permanent supply shock in order to real GDP. keep expected inflation constant. Doing so, how- The right panels of figure A3 tell a similar ever, is difficult in practice for three reasons. First, story for a positive permanent supply shock. an NGDP-targeting central bank wants to ignore This scenario also assumes that by the time the temporary supply shocks for the reasons laid out central bank is certain that the supply shock is above. Second, it is impossible in real time for a permanent, inflation expectations and the PC central bank to know whether a supply shock is have already adjusted. Changing the NGDP tar- temporary or permanent. Consequently, if mon- get growth rate at that point further stimulates etary authorities attempt to ignore temporary an already strengthened economy. supply shocks, they are likely to ignore perma- In practice, then, the model suggests that to nent supply shocks as well. Third, by the time minimize real GDP volatility, it is best to ignore the central bank realizes it has inadvertently both temporary and permanent supply shocks.

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 36 NOTES

Introduction 2011 to adopt an NGDP level target at around the same time the prominent monetary theo- 1. See, for example, Christina Romer, “Dear Ben: rist Michael Woodford made the case for it at It’s Time for Your Volker Moment,” New York an important conference for central bankers. Times, October 29, 2011; Scott Sumner, “Re- Romer, “Dear Ben”; Woodford, “Methods of targeting the Fed,” National Affairs 9 (Fall Policy Accommodation.” Wall Street firms such 2011); Michael Woodford, “Methods of Policy as Goldman Sachs also became interested in Accommodation at the Interest-Rate Lower this new framework and added momentum to Bound” (Proceedings from Kansas City Federal the NGDP targeting conversation. Jan Hatzius Reserve Economic Symposium, Jackson Hole, et al., “The Case for a Nominal GDP Level WY, August 30–September 1, 2012); Jeffrey Target,” Goldman Sachs US Economics Analyst Frankel, “Inflation Targeting Is Dead: Long Live 11/41 (October 24, 2011). Selgin notes, however, Nominal GDP Targeting,” VoxEU, June 19, 2012; that NGDP targeting has a long history with and David Beckworth, “Inflation Targeting: A the notion of a total spending target going back Monetary Regime Whose Time Has Come and to at least 1837. George Selgin, “Some ‘Serious’ Gone” (Mercatus Research, Mercatus Center at Theoretical Writings That Favor NGDP George Mason University, Arlington, VA, 2014). Targeting,” Alt-M, June 19, 2018. 2. See, for example, Michael T. Kiley and John M. 4. The search criteria used for NGDP target- Roberts, “Monetary Policy in a Low Interest ing were as follows: “nominal gdp targeting” Rate World,” Brookings Papers on Economic or “nominal gdp target” or “ngdp targeting” or Activity (Spring 2017); Lawrence H. Summers, “ngdp target” or “nominal income targeting” Why the Fed Needs a New Monetary Policy or “nominal income target.” For price level target- Framework (Washington, DC: Brookings, ing, the search criteria used were “price level tar- 2018); and Ben Bernanke, Michael Kiley, and geting” or “price level target.” John Roberts, “Monetary Policy Strategies for 5. For a recent critique, see, for example, Charles a Low-Rate Environment” (FEDS Working Goodhart, Melanie Baker, and Jonathan Paper No. 2019-009, Federal Reserve Staff, Ashworth, “Monetary Targetry: Might Carney February 13, 2019). Make a Difference?,” VoxEU, January 22, 2013. 3. The renewed interest began with blogging, spread to more traditional news media, and I. Facts about Nominal GDP then made its way to influential academics. Level Targeting Derek Thompson, “The Blogger Who Saved the Economy,” The Atlantic, September 14, 2012. 6. Technically, NGDP equals consumption spend- Former Council of Economic Advisers chair ing, investment spending, government pur- Christina Romer, for example, urged the Fed in chases, and net exports, while NGDI equals

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 37 wages, rental income, interest income, and “Monetary-Policy Rules and the Great Inflation,” profits. American Economic Review 92, no. 2 (2002): 7. To further illustrate this point, imagine the oppo- 115–20; Orphanides, “The Unreliability site case where the public expects total dollar of Output-Gap Estimates in Real Time,” spending to sharply fall in the future and cause a Review of Economics and Statistics 84, no. 4 recession. This expectation would drive up the (2002); Orphanides, “Monetary Policy Rules, demand for money in the present and make the Macroeconomic Stability, and Inflation: A View expected spending collapse a self-fulfilling event. from the Trenches,” Journal of Money, Credit, Conversely, an expected boom where prices are and Banking 36, no. 2 (2004): 151–75. Negative expected to soar would cause the public to spend supply shocks lowered the productive capac- their money balances in the present, creating ity of the economy in the 1970s, but policymak- another self-fulfilling outcome. An NGDP level ers were slow to realize it. Consequently, they target minimizes these swings by managing the assumed there was more spare capacity in the public’s expectations of the future path of total economy than existed and ran the economy too dollar spending. hot. Beckworth and Hendrickson show how such 8. To be clear, most money is created by banks and output gap confusion continues to plague Fed other financial firms when they make loans. So policy and how NGDPLT would solve this prob the Fed cannot directly adjust the money sup- lem. David Beckworth and Josh Hendrickson, ply in response to changes in money demand. “Nominal GDP Targeting versus the Taylor Rule What it can do is adjust monetary policy to on an Even Playing Field,” Journal of Money, influence how much spending households and Credit, and Banking (forthcoming). Along these businesses will wish to make. This influence, in lines, Garín, Lester, and Sims show how an NGDP turn, will affect both the demand for money and target tends to outperform an inflation target how much money is created by banks. in the presence of supply shocks. Julio Garín, 9. Joshua R. Hendrickson, “An Overhaul of Federal Robert Lester, and Eric Sims, “On the Desirability Reserve Doctrine: Nominal Income and the Great of Nominal GDP Targeting,” Journal of Economic Moderation,” Journal of Macroeconomics 34, Dynamics and Control 69 (August 2016): 21–44. no. 2 (2012): 304–17. These periods correspond to 13. Evan Koenig, “Like a Good Neighbor: Monetary the Great Inflation andGreat Moderation peri- Policy, Financial Stability, and the Distribution of ods, which are affiliated with different inflation Risk,” International Journal of Central Banking regimes for the Fed. 9, no. 2 (June 2013): 57–82; Kevin Sheedy, “Debt 10. The data come from the Center for Financial and Incomplete Financial Markets,” Brookings Stability. Papers on Economic Activity 45, no. 1 (2014): 301– 11. George Selgin, David Beckworth, and Berrak 61; Costas Azariadis, James Bullard, Aarti Singh, Bahadir, “The Productivity Gap: Monetary and Jacek Suda, “Incomplete Credit Markets Policy, the Subprime Boom, and the Post-2001 and Monetary Policy” (Federal Reserve Bank of Productivity Surge,” Journal of Policy Modeling St. Louis Working Paper, 2016); James Bullard and 37, no. 2 (2015): 189–207. Riccardo DiCecio, “Optimal Monetary Policy for 12. David Beckworth, “The Monetary Policy Origins the Masses” (Federal Reserve Bank of St. Louis of the Eurozone Crisis,” International Finance Working Paper, 2018). 20, no. 2 (2017): 114–34. Orphanides shows how 14. David Beckworth, “The Financial Stability Case supply-side confusion contributed to the high for a Nominal GDP Target,” Cato Journal 39, inflation of the 1970s. Athanasios Orphanides, no. 2 (2019): 419–47. “Activist Stabilization Policy and Inflation: The 15. I also checked the strength of these relation- Taylor Rule in the 1970s” (Federal Reserve Board ships (excluding Greece) and found that in all of Governors Working Paper, 2000); Orphanides, but the equity premium case, the relationships

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 38 remained statistically and econom ically signifi- University of Chicago Press, 1999); Alex Nikolsko- cant. Beckworth, “Financial Stability Case for a Rzhevskyy, David Papell, and Ruxandra Prodan, Nominal GDP Target.” “Deviations from Rules-Based Policy and Their 16. Beckworth, “Financial Stability Case for a Effects,”Journal of Economic Dynamics and Nominal GDP Target.” Control 49 (2014): 4–17. 17. Per the Fisher equation, the real interest rate 25. Taylor, “Discretion vs. Policy Rules in Practice.” equals the nominal interest rate minus expected 26. Bennett McCallum, “The Case for Rules in inflation. Consequently, if the nominal interest the Conduct of Monetary Policy: A Concrete rate and inflation are both near zero, then the Example,” Federal Reserve Bank of Richmond real interest rate cannot fall to the natural inter- Economic Review (September/October 1987): est rate level, and markets fail to clear. 10–18; McCallum, “Robustness Properties of a 18. Kiley and Roberts, “Monetary Policy in a Low Rule for Monetary Policy,” Carnegie-Rochester Interest Rate World.” Conference Series on Public Policy 29 (1988): 19. The only restriction imposed on this VAR is that 173–203. core personal consumption expenditure (PCE) 27. Friedman, “The Role of Monetary Policy.” inflation cannot influence NGDP or the out- 28. Garín, Lester, and Sims, “On the Desirability put gap. All other interactions are allowed. This of Nominal GDP Targeting”; Beckworth and restriction is imposed for several reasons. First, Hendrickson, “Nominal GDP Targeting versus the PCE deflator is already in NGDP, so this is the Taylor Rule on an Even Playing Field.” one way to increase degrees of freedom. Second, 29. Hendrickson, “Overhaul of Federal Reserve the restriction means the output gap is being Doctrine.” solely determined by nominal demand shocks and past lags of itself. These interactions are then II. Fears about Nominal GDP allowed to feed into the inflation, creating a kind Level Targeting of reduced-form Phillips Curve. 30. Jeffrey Frankel, “Should the Fed Be 20. John Taylor, “Discretion vs. Policy Rules in Constrained?,” Cato Journal 39, no. 2 (2019): Practice,” Carnegie-Rochester Conference Series 461–70. on Public Policy 39 (1993): 197. 31. George Selgin, Less Than Zero: The Case for 21. Finn Kydland and Edward Prescott, “Rules Rather a Falling Price Level in a Growing Economy, Than Discretion: The Inconsistency of Optimal 2nd ed. (Washington, DC: Cato Institute, 2018). Plans,” Journal of Politi cal Economy 85, no. 3 32. Studies that show at least some endogeneity of the (1977): 473–91; Robert Barro and David Gordon, potential real GDP to the business cycle, espe- “Rules, Discretion, and Reputation in a Model of cially at the ZLB, include Dave Reifschneider, Monetary Policy,” Journal of Monetary Economics William Wascher, and David Wilcox, 12, no. 1 (1983): 101–21. “Aggregate Supply in the United States: Recent 22. Milton Friedman, “The Role of Monetary Policy,” Developments and Implications for the Conduct American Economic Review 58, no. 1 (1968): 1–17. of Monetary Policy,” IMF Economic Review 23. Athanasios Orphanides, “Fear of Liftoff: 63, no. 1 (2015): 71–109; and J. W. Mason, What Uncertainty, Rules, and Discretion in Recovery? The Case for Continued Expansionary Monetary Policy Normalization,” Federal Policy at the Fed (New York: Roosevelt Institute, Reserve Bank of St. Louis Review 97, no. 3 2017). (2015): 173–96; Mark Calabria, “Behavioral 33. Orphanides, “Activist Stabilization Policy and Economics and Fed Policymaking,” Cato Inflation”; Orphanides, “Monetary-Policy Journal 36, no. 3 (2016): 573–87. Rules and the Great Inflation”; Orphanides, 24. John Taylor, “A Historical Analysis of Monetary “Unreliability of Output-Gap Estimates in Real Policy Rules,” in Monetary Policy Rules (Chicago: Time”; Orphanides, “Monetary Policy Rules.”

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 39 34. Beckworth and Hendrickson, “Nominal GDP of Oil Price Shocks,” Brookings Papers on Targeting versus the Taylor Rule on an Even Economic Activity 28, no. 1 (1997): 91–157. Playing Field.” If a Taylor Rule uses the unem- Ironically, the Bernanke Fed fell prey to this ployment gap instead of the output gap, the infor- same tendency in 2008. Though there were mation requirement is lessened since the unem- many factors that caused the Great Recession, ployment rate is released monthly. Still, even the Fed deciding not to ease between April and this approach requires an estimate of the natu October 2008 because of inflationary pres- ral rate of unemployment, which has undergone sures and the Fed signaling that it would actu- significant revisions over the past decade. Adam ally tighten policy during this time were con- Ozimek and Michael Ferlez, “The Fed’s Mistake,” tributing factors to the downturn. For more Moody’s Analytics Regional Financial Review on this view, see Robert Hetzel, “Monetary (August 2018): 25–31. Policy in the 2008–2009 Recession,” Richmond 35. Boragan Aruoba et al., “Improving GDP Mea Federal Reserve Bank Economic Quarterly 95, surement: A Measurement-Error Perspective” no. 2 (2009): 201–33; and David Beckworth and (Philadelphia Fed Working Paper Series, 2013), Ramesh Ponnuru, “Subprime Reasoning on 13–16. Housing,” New York Times, January 27, 2016. 36. Aditya Aladangady et al., “From Transactions 41. George Selgin, David Beckworth, and Berrak Data to Economic Statistics: Constructing Real- Bahadir, “The Productivity Gap: Monetary Time, High-Frequency, Geographic Measures of Policy, the Subprime Boom, and the Post-2001 Consumer Spending” (Paper presented at NBER Productivity Surge,” Journal of Policy Modeling “Big Data for 21st Century Economics Statistics” 37, no. 2 (2015): 189–207. Conference, Washington, DC, March 15–16, 42. The impulse response functions come from an 2019). estimated vector autoregression where long- 37. Scott Sumner, “Using Futures Instrument Prices run restrictions are imposed such that only to Target Nominal Income,” Bulletin of Economic productivity shocks can permanently influ- Research 41, no. 2 (1989): 157–62; Sumner, ence productivity in the long run. This replica- “A Market-Driven Nominal GDP Targeting tion uses PCE inflation and the Congressional Regime” (Mercatus Research, Mercatus Center Budget Office’s output gap. It closely follows at George Mason University, Arlington, VA, Selgin, Beckworth, and Bahadir, “Productivity 2013). Gap.” The sample only runs through 2008Q3 38. For example, Selgin, Less Than Zero. since after that time the ZLB becomes binding. 39. New Keynesian models (NKMs) frame this dis- 43. Selgin, Beckworth, and Bahadir, “Productivity cussion differently. They show that only in spe- Gap.” cial cases does stabilizing inflation also stabilize 44. Michael Bordo and David Wheelock, “Monetary output. Such “divine coincidences” are rare and Policy and Asset Prices: A Look Back at Past U.S. do not depend on the absence of supply shocks in Stock Market Booms,” Federal Reserve Bank the NKM but rather on the absence of real rigidi- of St. Louis Review 86, no. 6 (November/ ties. Kim shows, however, that this NKM result December 2004): 19–44; Bordo and Wheelock, is not robust to different forms of the produc- “Stock Market Booms and Monetary Policy in tion function. Bae-Gem Kim, “Supply Shocks and the Twentieth Century,” Federal Reserve Bank the Divine Coincidence,” Economic Letters 145 of St. Louis Review 89, no. 2 (March/April 2007): (2016): 210–13. Once this issue is fixed, the NKM 91–122; William R. White, “Should Monetary shows that this inflation-output tradeoff can exist Policy Lean or Clean?” (Globalization and because of supply shocks alone. Monetary Policy Institute Working Paper No. 40. Ben Bernanke, Mark Gertler, and Mark Watson, 34, Federal Reserve Bank of Dallas, April 2009); “Systematic Monetary Policy and the Effects Lawrence Christiano et al., “Monetary Policy and

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 40 Stock Market Booms” (Proceedings from Kansas lic makes many economic decisions based on a City Federal Reserve Economic Symposium, forecast of their nominal incomes. For example, Jackson Hole, WY, 2010). households may take out a 30-year mortgage 45. Christiano et al., “Monetary Policy and Stock based on an implicit forecast of their nominal Market Booms”; Selgin, Beckworth, and income over this horizon. The actual realiza- Bahadir, “Productivity Gap.” tion of nominal income may turn out to be very 46. All else equal, a rise in current and expected pro- different than expected, but the households ductivity growth should raise the natural real may not be able to quickly adjust their plans rate of interest. given sticky debt contracts and other commit- 47. Along these lines, Fackler and McMillan show ments that constrain them. Therefore, the con- that NGDPLT would have outperformed a sequences of previous forecasts are often bind- price level target in terms of minimizing a loss ing on them and slow to change even if their function in the period leading up to the Great nominal income forecasts have been updated. Recession. James Fackler and Doug McMillan, Second, in addition to these old forecasts and “Nominal GDP versus Price Level Targeting: decisions whose influence lingers, new fore- An Empirical Evaluation” (University of casts and new decisions are being made each Kentucky Working Paper, Lexington, KY, quarter for subsequent periods that will also January 2019). have lingering effects. Together, this means 48. Selgin, Beckworth, and Bahadir, “Productivity future periods have many overlapping and dif Gap.” ferent forecasts applied to them that only grad- ually adjust. See Beckworth, “A Practical Guide III. Functionality: How to Implement to Nominal GDP,” for more details on how it is Nominal GDP Level Targeting measured. 49. Lars Svensson, “Inflation Forecast Targeting: 54. One can also target NGDP using a reaction Implementing and Monitoring Inflation Targets,” function with the monetary base as the instru- European Economic Review 41, no. 6 (1997): 1111– ment, such as in the famous McCallum rule. 46; Svensson, “What Is Wrong with Taylor Rules? McCallum, “The Case for Rules in the Conduct Using Judgment in Monetary Policy through of Monetary Policy.” The Fed’s paying of Targeting Rules,” Journal of Economic Literature interest on excess reserves complicates this 41 , no. 2 (2003). approach, but I show how one can still imple- 50. James Tobin, “Stabilization Policy Ten Years After,” ment a McCallum rule by looking at the non– Brookings Papers on Economic Activity 11, no. 1 excess reserve portion of the monetary base. (1980): 51. Beckworth, “Nominal GDP as the Stance of 51. Sumner, “Using Futures Instrument Prices to Monetary Policy.” Belongia and Ireland simi- Target Nominal Income”; Sumner, “A Market- larly show how to target NGDP using the Driven Nominal GDP Targeting Regime.” monetary base and Divisia monetary aggre- 52. As constructed in David Beckworth, “Nominal gates. Michael Belongia and Peter Ireland, “A GDP as the Stance of Monetary Policy: A ‘Working’ Solution to the Question of Nominal Practical Guide” (working paper, December 1, GDP Targeting,” Macroeconomic Dynamics 19, 2018). no. 3 (2015): 508–34. 53. The time frame of five years is chosen for 55. Joe Gagnon, “What Have We Learned about the sticky forecast path for NGDP since it Central Bank Balance Sheets and Monetary is assumed that all constraints created by Policy?” Cato Journal 39, no. 2 (2019): 407–17; decisions based on the forecast can be fully Frankel, “Should the Fed Be Constrained?” unwound within five years. There are two 56. Michel notes it would also make life easier for motivations for this measure. First, the pub- the Fed since it would narrow the Fed’s focus

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 41 down to one objective. Norbert Michel, Give 59. Oliver Blanchard, “On the Need for (At Least) the Fed a Single Mandate: Monetary Neutrality Five Classes of Macro Models,” Peterson (Washington, DC: Heritage Foundation, 2019). Institute for International Economics, April 10, Congress would need to ratify a change to a sin- 2017. gle mandate approach such as NGDPLT. 60. The PC can be derived from a Short-Run Aggregate Supply Curve, and consequently, IV. Conclusion PC et ,can be viewed as an inverted supply shock. 57. Frankel, “Should the Fed Be Constrained?” Also, note that to be consistent with the graphical representation of the PC, the parameter φ is Appendix assumed to be nonlinear: φ = λ ΔY! 2,, with λ > 1. 58. Tyler Cowen and Alex Tabarrok, Modern Princi ples: Macroeconomics, 4th ed. (New York: Macmillan Learning, 2018).

MERCATUS CENTER AT GEORGE MASON UNIVERSITY 42

Facts, Fears, and Functionality of NGDP Level Targeting: a Guide to a Popu­Lar Framework for Monetary Policy David Beckworth

Facts, Fears, and Functionality of NGDP Level Targeting: a Guide to a PopuLar Framework for Monetary Policy David Beckworth