This document analyzes differences in how monetary policy transmission mechanisms impact output composition in Australia compared to the Euro area and US. It uses VAR models and impulse response analysis on quarterly data from 1982-2007 to examine responses of consumption, investment, housing investment, durable consumption, and other GDP components to interest rate changes. Results show Australia's investment channel is stronger, while the Euro area and US channels are similar. This is largely attributable to stronger housing investment responses in Australia and the US, and Australia having a higher investment share of GDP than the US.
3. Motivation
Concensus: i ↑−→ Y ↓ (output, aggregate demand)
Two main channels:
- Investment channel: i ↑−→ I ↓−→ Y ↓
- Consumption channel: i ↑−→ C ↓−→ Y ↓
Deeper look: Which channel is more important
(dominant)?
Consumption smoothing and Investment volatility: investment
channel might be more sensitive to interest rate changes.
Answers can be informative for the theory on the channels of
the monetary policy transmission mechanism, and also
potentially help central bankers/policy makers in monitoring
the economy.
4. Literature
Angeloni, I, Kashyap, A, Mojon, B and Terlizzese, D 2003,
‘The Output Composition Puzzle: A Difference in the
Monetary Transmission Mechanism in the Euro Area and
U.S.’, Journal of Money, Credit and Banking, 35(6):
- European countries (1970 - 2000): I>C
- US (1965 - 2001): C>I “Output Composition Puzzle”
Fujiwara, I 2004, ‘Output composition of the monetary policy
transmission mechanism in Japan’, The B.E. Journal of
Macroeconomics, 0(1): I>C (1980 - 2003)
Reasons: most possible: structure of housing markets (i
affects C and I in housing markets)
If so, we can guess: for Australia: C>I, or at least closer to
US rather than Japan or Euro area.
5. The VAR models
Christiano, Eichenbaum and Evans (2001): 10 variables
(consumption, investment, other GDP components, CPI, real
wages, labour productivity, policy rate, the profit-to-GDP
ratio, money and share price index)
Erceg and Levin (2002): 6 variables (consumption,
investment, other GDP components, CPI, commodity prices,
and policy rate)
Generalized Erceg and Levin (2002): add 2 variables (bond
yields and money)
Peersman and Smets (2003): 7 endogenous variables
(consumption, investment, other GDP components,
commodity prices, money, policy rate, and the real effective
exchange rate), 3 exogenous variables (the US federal funds
rate, the US GDP, and the US price index) for Australia and
the Euro area; oil prices for the US
6. Data
Australia, the Euro area, the US
Quarterly data for the period 1982Q3–2007Q4
Log form
1 or 2 lag structure
Exogenous variables for all VARs for Australia
7. Output composition (Consumption and Investment
channels)
Results are based on 2 indicators:
Impulse Responses: Proportional effect
Contribution: Size effect
11. Proportional effect
Investment reacts stronger at peak, but comes back faster
Consumption stays significant for longer period
Investment responds relatively strongest in Australia, then US,
then Euro area
12. Calculation of Size contribution
1. Step 1
C × weightC
= C1
C × weightC + I × weightI + G × weightG
I × weightI
= I1
C × weightC + I × weightI + G × weightG
2. Step 2
C1
= C2
C1 + I1
I1
= I2
C1 + I1
So C2 + I2 = 1. Normally C2 , I2 ∈ (0, 1).
If C2 > 0.5 ⇒ C > I.
If C2 < 0.5 ⇒ I > C.
13. Shares of GDP components
Country/Region
Australia
Euro area
US
Consumption
0.54
0.57
0.65
Investment
0.17
0.21
0.16
Other GDP components
0.29
0.22
0.19
15. Size contribution results
Consumption channel is much weaker in Australia (Investment
channel is much stronger).
In the Euro area and the US: not clear whether investment or
consumption channel is dominant.
So what are the main reasons of the differences between Australia
and the two comparators?
Might be the housing investment (the
more-interest-rate-sensitive component of investment)?
To check: decompose investment into housing and
non-housing investment
Use the same VARs, the same techniques (proportional and
size effect)
16. Housing and non-housing investment: impulse responses
Australia
Housing
Non-housing
Euro area
Housing
Non-housing
US
Housing
Non-housing
17. Shares of housing and non-housing investment
Country
/Region
Australia
Euro area
US
Housing
In GDP (In investment)
0.06 (35%)
0.075 (36%)
0.05 (31%)
Non-housing
In GDP (In investment)
0.11 (65%)
0.135 (64%)
0.11 (69%)
19. Housing contribution results
Housing contribution is higher in Australia and the US.
Housing contribution is much lower in the Euro area.
Therefore, the difference in housing investment responses
relatively to non-housing investment is likely the key reason
behind the difference between Australia and the Euro area.
Remain: between Australia and the US?
Next: decompose consumption into durable consumption
(more sensitive to interest rate changes) and non-durable
consumption (for Australia and the US).
Use the same VARs, the same techniques (proportional and
size effect).
20. Durable and non-durable consumption: impulse responses
Australia
Durable
Non-durable
US
Durable
Non-durable
21. Shares of durable and non-durable consumption
Country
/Region
Australia
US
Durable consumption
In GDP (In consumption)
0.04 (7.5%)
0.055 (8.5%)
Non-durable consumption
In GDP (In consumption)
0.54 (92.5%)
0.595 (91.5%)
23. Durable and non-durable contribution results
Durable contribution is around 0.3–0.4 in both Australia and
the US (though the band is much narrower in the US).
Therefore, the difference in the shares of consumption and
investment in total GDP is the main reason behind the output
compostion difference between Australia and the US.
Country/Region
Australia
US
Consumption
0.54
0.65
Investment
0.17
0.16
Other GDP components
0.29
0.19
24. Conclusion
Investment channel is stronger compared to consumption
channel in Australia.
In the Euro area and the US, the two channels are basically
not different when shares are taken into account.
The reasons behind:
Between Australia and the Euro area: the housing investment
responses.
Between Australia and the US: the shares in the total GDP.