Monetary Policy Transmission Revisited

# Monetary Policy Transmission Revisited
## RBNZ Roundtable
### Alex Li
### 18 November, 2019

---

# Economic Reality

---

# Economic Reality

## NZ Wealth Distribution

Source: Statistics New Zealand, Unit: .hi[1000 NZD], Year: .hi[2015]

---

# Economic Reality

## Household Balance Sheet

__Differences__ across groups of agents or regions in __four aspects__:

- .hi[Nominal-term assets/liability]: Short-term Deposits, Long-term Bonds, (Most) Mortgages...

- .hi[Real-term assets/liability]: Stocks, Inflation-Indexed Government Bonds, Price-level Adjusted Mortgages....

- .hi[Unearned income]: Government Transfers, Pensions, Social Security, Unemployment benefits, Alimony, and Child Support...

- .hi[Wage income]: Salaries, Bonuses, Wages, Commissions, and Tips...

---
# Economic Reality

## Fiscal Policy

New Zealand counts all public debt owed by .hi[all levels of government] in its national debt figure. (In millions)

---
# Economic Reality

## Consumer Confidence Index

Confidence index fluctuates __across time__.

---
# Economic Reality

## New Perspectives

__Revisit the monetary transmission mechanism__ through the lens of Heterogeneous-agent New Keynesian .hi[(HANK)] models.

- .hi[New Keynesian] models with .hi[extensions] on the household side. This is a rich environment where there are differences across the agents.

- This type of models can incorporate the .hi[previously-mentioned economic reality]

- It provides a more comprehensive framework for .hi[policy analysis and help inform forecasts].

---
# Economic Reality

## My Research
  
- .hi[My research] explores the implication of .hi[bounded rationality] based on a HANK model. Households do not know how exactly the economy works.

- The key difference is in the aspect of .hi[expectation formation].

- If the agents have __adaptive expectations__: Agent-based Computational economics (ACE) 
 
 - If the agents have __rational expectations__: Dynamic Stochastic General Equilibrium (DSGE)

---
class: inverse, center, middle

# Transmission Mechanism Review

---
# Transmission Mechanism Review

## Overview

---
# Transmission Mechanism Review

## Overview

- Transmission mechanism from .hi-purple[monetary policy] to .hi[household consumption].

- Different __Channels__ for the transmission.

- __Representative__ agents stories.

- __Heterogeneous__ agents stories.

- __Bounded Rationality__ Expectations.
 
---
# Transmission Mechanism Review

## Direct v.s Indirect Effects

### .hi[Direct Effects]:

- Effects that operate .hi-slate[without] change in household disposable .hi-slate[income].

- .hi-purple[Intertemporal Substitution Effect]

- .hi-purple[Asset Revaluation Effect]
  
--

### .hi[Indirect Effects]:

- Effects that arise from .hi-slate[changes in household disposable income].

- .hi-purple[Labor Income], .hi-purple[Fiscal Response],

---

# Transmission Mechanism Review

## Consumption Channel

---

# Transmission Mechanism Review

## Potential Topics

There are a few potential topics HANK models can address

- Stimulate the economy: .hi[Small Persistent OCR Cuts] v.s. .hi-turquoise[Large but Transitory Cuts]

- .hi[Inflation-activity tradeoff] for the monetary authority depends on the .hi-turquoise[adjustment rule of fiscal policy].

- Well-speicified structural models can .hi[offer guidance] on .hi-turquoise[previously-unused] policy instruments.

- .hi[Dual Mandate] acknowledge the impact monetary policy has on employment and economic activity through the .hi[labor market].

---

# Transmission Mechanism Review

## Literature

---

# Representative Agent

---

# Representative Agent

## Definition

A .hi[typical decision-maker] of a certain type (typical consume or firm). These models are .hi[easy to use].

The assumption is good when:

- All agents of the same type are .hi-turquoise[identical].

- Agents' choices are .hi-turquoise[equivalent to the decision of one individual].

---

# Representative Agent

## Implications for Transmission

- Households are .hi[permanent income consumers].

- Consumption .hi-turquoise[smoothing]. They are .hi-turquoise[not responsive] to transitory income changes. 
 
 - The representative-agent models implies the .hi-turquoise[marginal propensity to consume (MPC) is low]. 
  
--

- Monetary transmission mechanism: .hi[the indirect effect is not important]. Households weakly respond to:

- .hi-turquoise[Fiscal response].
  
 - .hi-turquoise[Higher labor income].

---

# Representative Agent

## Implications for Transmission

---

# Representative Agent

## Comments on TANK

- __Two-Agent New Keynesian (TANK)__ models are widely used to account for the hand-to-mouth households.

- However, the distribution of wealth/income is __exogenously imposed__ by the modeler.

---

# Heterogeneous Agent

---

# Heterogeneous Agent

## Definition

Heterogeneous models contain __different economic agents__, which can be modeled in terms of

- .hi[Idiosyncractic productivity] shocks: Wealth Distribution, Labor Income Distribution

- .hi[Asset Portfolio]: Nominal/Real Assets, Long/Short-term Assets

- .hi[Other ways to model heterogeneity...]
 
HANK models usually assume .hi[incomplete markets] and .hi[binding borrowing constraint]

---
# Heterogeneous Agent

## Implications for Transmission

- Generate a sizable fraction of (poor and wealthy) .hi[hand-to-mouth households] as in data.

- These households are not responsive to interest rate cuts. 
  
--

- Households are .hi[no longer permanent-income] consumers.

- Interest rate cuts lead to strong offsetting income effects on consumption. 
 
---
# Heterogeneous Agent

## Implications for Transmission

- Monetary policy is effective only to the extent that it generates a .hi[general equilibrium response] in household disposable income.

- Higher __Labor Income__;
 
 - Higher __Government Net Transfers__;
 
 - Higher __Dividends__;
 
 - ...

---

# Heterogeneous Agent

## Implications for Transmission

---

# Heterogeneous Agent

## Overall Response

- The __overall response__ of aggregate consumption to a cut in interest rates may be .hi[larger or smaller] than in RANK models, depending on the .hi[fiscal reaction to the monetary expansion].

- Higher .hi-turquoise[Government Spending]

- Higher .hi-turquoise[Transfer]

- Lower .hi-turquoise[Taxation]
 
  - Lower .hi-purple[Government debt]

---

# Heterogeneous Agent

## Policy Takeaway for RBNZ

__Kaplan, Moll, and Violante (2018)__. .hi-purple[Fiscal reaction] to monetary expansion matters to the overall response of aggregate consumption. A combination of the following four can happen:

- Study the Fiscal-Monetary Interactions in New Zealand

- If the fiscal policy responds through changes in government debt, .hi[a rate cut is less potent]. 
  
  - if the fiscal policy responds through changes in government spending, debt, transfers, taxation, .hi[a rate cut is more potent].

---

# Heterogeneous Agent

## Policy Takeaway for RBNZ

Monetary policy designs to improve the potency of monetary effects? __Kaplan, Moll, and Violante (2018)__

A .hi[transitory large] change in interest is .hi-purple[more potent] than a .hi[persistent small] change, __given the accumulative changes are the same__.

- When shocks are persistent and small, a large portion of the interest rate cut, and the associated relaxation of the government budget constraint and increase of labor income, __happen in the future__.

- Hand-to-mouth households __do not respond__ much to future changes.

---
# Heterogeneous Agent

## Policy Takeaway for RBNZ

__Auclert (2019)__ offers a decomposition method for the monetary transmission mechanism. A monetary shock triggers the following things to change. Each household has .hi[different exposures] to each of the following change

- .hi[Nominal prices] rise.

- .hi[Real discount rates] rise.

- .hi[Unearned and earned income] rise.

---
# Heterogeneous Agent

## Policy Takeaway for RBNZ

- Joint consumption, income and asset __data__.

- A better empirical understanding of __household balance sheet__.

- Better understanding the __winners and losers__ from interest rate changes.

---

# Bounded Rationality

---

# Bounded Rationality

## Rational Expectations

Heterogeneous-agent models are solved under __rational expectations__. They make the following strong assumptions. Agents know:

- .hi[Structure of the economy].

- .hi[Current distribution] of wealth, labor productivity, and other individual states.

- .hi[How distribution evolves] across time and affects the interest rate and wage.

---

# Bounded Rationality

## Concept

HANK with Local Rationality by Li (2019)

- .hi[Local rationality] is a novel bounded rationality concept developed by Evans, Li, and McGough (2019)

- Agents are locally rational in the sense that they make .hi[optimal decisions] .hi-purple[without any OCR changes].

- Agents learn from their past experience to respond to .hi-purple[OCR changes].

---

# Bounded Rationality

## Model

- To accommodate __aggregate shocks__, agents .hi[estimate econometric models to forecast] .hi-purple[future states] and make choices which are optimal up to their forecasts.

- Both __households and firms__ need to make forecasts about the future to make decisions about today.

- __Households__ decision on consumption: Marginal utilities from consumption
  
  - __Firms__ decision on price: Inflation

---

# Bounded Rationality

## Impulse Response Functions

<div class="col2">
.hi-purple[Consumption]

.hi-purple[Inflation]
<center>
 <img src="figures/inflation.png" width="410" height="308">
</center> 
</div>

---

# Bounded Rationality

## Impulse Response Functions

<div class="col2">
.hi-purple[Wage]

.hi-purple[Gini Index (Inequality)] 
<center>
 <img src="figures/gini.png" width="400" height="300">
</center> 
</div>

---

# Bounded Rationality

## Policy Implication

- Expectation adjustments .hi[across households]. Consider a simple setting:

- .hi[Asset-poor households] are more sensitive to the labor income change. Higher expectation adjustment

- .hi[Asset-wealthy households] are more sensitive to interest rate change. 
  
---
class: inverse, center, middle

# Thank You!

---
exclude: true