UNIVERSIT OF CALIFORNIA Economics 134 DEPARTMENT OF ECONOMICS Spring 2018 Professor David Romer LECTURE 10 THE ZERO LOWER BOUND IN THE IS-MP-IA FRAMEWORK FEBRUAR 21, 2018 I. INTRODUCTION II. THE IS-MP-IA MODEL EXTENDED A. Assumptions 1. The nominal interest rate can t be negative 2. Expected inflation depends on actual inflation 3. Discussion B. The AD Curve 1. Where we are headed 2. The IS-MP diagram 3. Deriving the AD curve C. A Little Bit about the Case of Money Targeting III. EXAMPLES A. A Large, Long-Lasting Fall in Planned Expenditure 1. The initial situation 2. The shock 3. Aside: Why doesn t the AD curve shift left by the same amount at each inflation rate? 4. The dynamics of the economy 5. What happens when there is a rebound in planned expenditure 6. How seriously should we take this? B. The Case of Anchored Expectations 1. Overview 2. A model of anchored expectations 3. The effects of a large, long-lasting fall in planned expenditure 4. A concern: how long can this last?
Economics 134 Spring 2018 David Romer LECTURE 10 The Zero Lower Bound in the IS-MP- IA Framework February 21, 2018
Announcements Problem Set 2 is being distributed. It is due at the beginning of lecture a week from today (Feb. 28). Optional problem set work session: Monday, Feb. 26, 6:45 8:15, in 597 Evans Hall. A packet of sample exam questions is also being distributed.
Announcements (cont.) For next time, you do not need to read the paper by Temin and Wigmore. My upcoming office hours: This week: Usual time: Thursday (2/22), 4 5:30. Next week and the week after: Monday (2/26 and 3/5), 3:30 5:00.
Economics 134 Spring 2018 David Romer LECTURE 9 The Conduct of Postwar Monetary Policy (concluded)
Bad Idea: Inflation Responds Little to Slack π π 0 π 1 IA 0 IA 1 0 AD IA will shift down only very slowly in response to <.
What Policies Are Likely to Be Followed If Policymakers Believe Inflation Responds Little to Slack? π π 0 IA 0 0, 1 AD 0 AD 1 No reason to have <. Result: Inflation doesn t fall.
What If Policymakers Believe Inflation Responds Little to Slack and Have an Overly Optimistic Estimate of? π π 1 π 0 IA 1 IA 0 AD 1 actual believed No reason to have < believed. Result: Inflation rises.
If It Is Monetary Policymakers Who Have These Ideas, What Will Be Going on in IS-MP? r MP 0 MP 1 IS 0 actual believed Fed shifts MP down to get = believed.
How Were Ideas Reflected in Monetary Policy Choices in the Early and Late 1970s? No reason to for contractionary policy because they thought it wouldn t curb inflation. Unrealistic estimates of the natural rate led to expansionary policy. Fed officials pushed for other policies to control inflation, such as price controls.
Figure 2 Inflation Rate 20 Eccles Martin Burns Volcker Greenspan 15 10 5 0-5 Jan-34 Jan-37 Jan-40 Jan-43 Jan-46 Jan-49 Jan-52 Jan-55 Jan-58 Jan-61 Jan-64 Jan-67 Jan-70 Jan-73 Jan-76 Jan-79 Jan-82 Jan-85 Jan-88 Jan-91 Jan-94 Jan-97 Jan-00 Jan-03 Percent
What Does Romer and Romer s Analysis Suggest about a Question We Discussed Early in the Course? Why did the rise of stabilization policy not cause the economy to quickly become much more stable? Romer and Romer s analysis provides support for the the tools were used badly hypothesis.
The Unemployment Rate after Romer & Romer Dates
20 The CPI Inflation Rate after Romer & Romer Dates 15 Percent 10 5 0-5 Jan-47 May-49 Sep-51 Jan-54 May-56 Sep-58 Jan-61 May-63 Sep-65 Jan-68 May-70 Sep-72 Jan-75 May-77 Sep-79 Jan-82 May-84 Sep-86 Jan-89 May-91 Sep-93 Jan-96 May-98 Sep-00 Jan-03 May-05 Sep-07 Jan-10
Interpreting Regression Results Example: Taylor s Estimates of the Pre-Volcker Monetary Policy Rule Note: Numbers in parentheses are t-statistics (coefficient estimate divided by the standard error).
Interpreting Regression Results Example (cont.) For the 1960 1979 sample, Taylor finds a coefficient on inflation = 0.813, with t-statistic = 12.9. Since the t-statistic is >> 2, we can reject the hypothesis that the coefficient is 0. But what is the 2-standard error confidence interval? Can we reject the hypothesis that the coefficient is 1?
Interpreting Regression Results Example (cont.) Coefficient on inflation = 0.813, t-statistic = 12.9. t-statistic coefficient/standard error, so standard error = coefficient/t-statistic. So: standard error = 0.813/12.9 = 0.063. The two-standard error confidence interval is from 2 standard errors below point estimate to 2 standard errors above. So: 2-standard error confidence interval = (0.687,0.939). 1 is outside this confidence interval, so we can reject ( at the 5% level ) the hypothesis that the coefficient is 1.
Economics 134 Spring 2018 David Romer LECTURE 10 The Zero Lower Bound in the IS-MP- IA Framework
I. INTRODUCTION
II. THE IS-MP-IA MODEL EXTENDED
Key Assumptions: 1 The nominal interest rate cannot be negative The central bank would like to set r = r(,π). Since the real interest rate, r, equals i π e, this means that r cannot be less than 0 π e. Thus: r = r(, π) e 0 π if r(, π) otherwise + π e 0
Key Assumptions: 2 Expected inflation is an increasing function of actual inflation. That is, π e = π e (π), where π e (π) is an increasing function.
One Comment Before We Proceed We will continue to use the usual IS-MP-IA model (that is, the model without the zero lower bound) in cases where it is appropriate.
Where We Are Headed: The Aggregate Demand Curve Accounting for the Zero Lower Bound π AD
The IS and MP Curves Accounting for the Zero Lower Bound: Step 1 r r(,π) 0 π e (π) IS
The IS and MP Curves Accounting for the Zero Lower Bound: Step 2 r MP 0 π e (π) IS
r Deriving the AD Curve MP(π 0 ) 0 π e (π 0 ) π π 0 IS 0 0
0 π e (π 0 π e 2 ) (π 1 ) 0 π e (π 0 ) Deriving the AD Curve r MP(π 0 ) MP(π 1 ) MP(π 2 ) π π 0 π 1 π 2 IS 0 π 0 > π 1 > π 2 0 1 2
Deriving the AD Curve (continued) r MP(π 2 ) 0 π e (π 2 ) π IS 0 π 2 2
Deriving the AD Curve (continued) r 0 π e (π 3 ) 0 π e (π 2 ) MP(π 2 ) MP(π 3 ) π IS 0 π 2 π 3 π 2 > π 3 3 2
Deriving the Aggregate Demand Curve: Conclusion π AD
A Little Bit about the Case of Money Targeting Continue to assume that expected inflation is lower when actual inflation is lower. Suppose that at some inflation rate, π 0, the nominal interest rate is zero. Thus the real interest rate is 0 π e (π 0 ). Now consider lower inflation, π 1 (so π 0 > π 1 ). The lowest possible real interest rate is 0 π e (π 1 ), which is higher than the real interest rate at π 0, 0 π e (π 0 ). Thus, r must be higher. That is, it is still true that when the economy is at the zero lower bound, lower inflation raises r.
III. EXAMPLES
Example: A Large, Long-Lasting Fall in Planned Expenditure r MP(π 0 ) 0 π e (π 0 ) π AD 1 IS 1 AD 1 0 (= ) 0 IS 0 π 0, π 1 IA 0, IA 1 1 0 (= )
Why Doesn t the AD Curve Shift Left by the Same Amount at Each Inflation Rate? π AD if r = r(,π) AD if r = 0 π e (π)
Why Doesn t the AD Curve Shift Left by the Same Amount at Each Inflation Rate? (continued) r r r = r(,π) 0 π e (π) IS 1 IS 0 IS 1 IS 0 1 0 1 0 A given shift of the IS curve causes a bigger fall in (at a given π) if r = 0 π e (π) than if r = r(,π).
Why Doesn t the AD Curve Shift Left by the Same Amount at Each Inflation Rate? (continued) π AD if r = r(,π) AD if r = 0 π e (π)
Why Doesn t the AD Curve Shift Left by the Same Amount at Each Inflation Rate? (concluded) π AD 1 AD 0
A Large, Long-Lasting Fall in Planned Expenditure (cont.) r MP(π 1 ) MP(π 2 ) 0 π e (π 2 ) 0 π e (π 1 ) π AD 1 IS 1 π 1 IA 1 π 2 IA 2 Note: Because inflation does not respond immediately to shocks, π 1 = π 0 (and so IA 1 is the same as IA 0 ). 2 1
The Effects of a Large Rebound in Planned Expenditure r 0 π e ( π 2 ) MP(π 2 ) IS 1 IS 3 π AD 1 π2 IA 2 2 AD 3 4
How Seriously Should We Take This? The main message, which we should take very seriously: When the economy is at the zero lower bound, a key force keeping the economy stable is inoperative.
Example 2: Anchored Expectations Inflation fell less in the Great Recession and the (subsequent period of continued high unemployment) than in previous recessions.
A Model of Anchored Expectations Two influences on inflation: As usual, below-normal output acts to make firms raise price and wages by less than before. This works to push inflation down. Firms expectations of inflation act to move inflation toward π*. When actual inflation is below π*, this works to push inflation up.
Revisiting a Large, Long-Lasting Fall in Planned Expenditure r MP(π*) 0 π e (π*) IS 0 π AD 0 0 (= ) π* IA 0 0 (= )
A Large, Long-Lasting Fall in Planned Expenditure (cont.) r MP(π*) 0 π e (π*) π AD 1 IS 1 π* IA 1 1
A Large, Long-Lasting Fall in Planned Expenditure (concluded) With anchored expectations, inflation can stabilize at a level below π* where the upward _ pull from π* and the downward pull from < 0 balance.