Vance Ginn Sam Houston State University Fall 2012 - - PowerPoint PPT Presentation

SLIDE 1

Vance Ginn

Sam Houston State University Fall 2012 vance.ginn@shsu.edu

SLIDE 2

A higher oil price has preceded 10 out of the last 11

recessions, making it a (arguably) leading indicator of business cycles.

The Energy Information Administration (2012) projects

that by 2035 the nominal oil price will remain volatile and could reach $235 per barrel.

• This compares with a range of $85 to $110 per barrel in 2011.

Recent evidence finds that the underlying source of oil

price disruptions matters (Kilian 2009)

Since the 1980s, the primary sources of oil price

shocks have been from factors related to the following:

• Oil demand—increased oil consumption in China and India
• Expectations of future oil supply disruptions (i.e. precautionary

demand)—instability in the Middle East

• I define these collectively as "global demand for oil shocks"

SLIDE 3

Other economic models that assume (explicitly or

implicitly) that an oil price disturbance is from a shifting supply curve, which I define as an “oil price shock,” are not well specified after 1980.

After estimating a four-variable VAR with quarterly

data from 1974:1-2011:3 and constructing a structural VAR with long-run restrictions, I compare the effects from identified shocks on macroeconomic variables.

I find that the effects are greater from a global demand

for oil shock than those from an oil price shock.

Therefore, policymakers should not necessarily be

concerned with an oil price shock driven by an oil supply disruption, but how their policies influence, or react, to global demand for oil shocks.

3

SLIDE 4

Three primary transmission mechanisms for oil price

disruptions through the macroeconomy:

• Aggregate supply:

Kim & Loungani (1992)—effect on producers Rotemberg & Woodford (1996)—imperfect competition Finn (2000)—perfect competition

• Aggregate demand:

Edelstein & Kilian (2009)—reduction in purchasing power Mehra & Peterson (2005)—decline in durable goods consumed Ramey & Vine (2011)—fewer automobiles purchased

• Term structure of interest rates:

Bernanke, Gertler, & Watson (1997)—systematic monetary policy

These mechanisms may be less relevant because of the

declining effects from oil price shocks since the 1980s (Barsky & Kilian, 2004; Blanchard & Galí, 2007)

4

SLIDE 5

Depending on whether oil disruptions are supply or

demand driven, the macroeconomic effects of an “oil- related shock” could be quite different (Kilian, 2009)

Supply driven “oil price shocks”:

• Shapiro & Watson (1988)
• Bernanke, Gertler, and Watson (1997)—factors other than oil

price shocks caused the 1970's recessions

• Hamilton and Herrera (2004)—exogenous oil price shocks were

to blame for these recessions

Demand driven “global demand for oil shocks”:

• Kilian (AER, 2009)—causes after 1980
• Baumeister and Peersman (2009)—global economic disruptions,

an increasing number of substitutes for oil, and more energy- efficient technologies

• Hamilton (2009) & Kilian (Comment, 2009)—global economic

slowdown in 2008

5

SLIDE 6

Since the seminal paper by Sims (1980), structural

VAR models have been the workhorse for separating the effects of shocks with few identifying restrictions.

I impose long-run restrictions similar to the

approaches by:

• King & Morley (2007)—study of the long-run dynamics of

the Phillips Curve relationship

• Valcarcel (2011)—analyze the link between the stock

market and macroeconomic aggregates

By imposing these long-run restrictions, I attempt

to disaggregate structural shocks.

6

SLIDE 7

I define a stationary vector of quarterly variables

as: xt = [Δut, reat or Δot, Δyt, Δπt]'

• Δut is the first difference of the average monthly

unemployment rate

• reat denotes the average of the detrended monthly index
• f global "real economic activity" constructed by Kilian

(2009b)

• Δot represents the first difference of the average monthly

real oil price

• Δyt is the first difference of the log of real GDP
• Δπt refers to the first difference of the average monthly

annualized rate of headline CPI inflation

• Δπ't represents the core CPI inflation.

7

SLIDE 8

I consider a reduced-form stationary VAR model: I estimate the reduced form VAR model, impose

fully recursive long-run identifying restrictions to A(1), and assume the structural disturbances (εt) are orthogonal.

I present the structural VAR representation as:

8

SLIDE 9

Quarterly data are for the sample from 1974:1 to

2011:3, the period when all variables were available.

• Monthly average of the level of the unemployment rate

(u)

• Log of real gross domestic product (y)
• Headline (π) and core (π') annualized inflation rates for

the CPI

• Real oil price is the refiner’s acquisition cost of imported

crude oil deflated by the headline CPI (o)

• Index of global real economic activity (rea) from Kilian

(n.d.)

Proxy of global aggregate demand for commodities from various bulk dry cargo freight rates. Kilian deflates this index by the headline CPI, linearly detrends it, and deviates it from the long-run trend

9

SLIDE 10

10

2% 4% 6% 8% 10% 12% 2% 4% 6% 8% 10% 12% 1975 1980 1985 1990 1995 2000 2005 2010

Unemployment Rate

• 80
• 40

40 80

• 80
• 40

40 80 1975 1980 1985 1990 1995 2000 2005 2010

Real Economic Activity

$0 $10 $20 $30 $40 $50 $60 $0 $10 $20 $30 $40 $50 $60 1975 1980 1985 1990 1995 2000 2005 2010

Real Oil Price

$4 $6 $8 $10 $12 $14 $4 $6 $8 $10 $12 $14 1975 1980 1985 1990 1995 2000 2005 2010

Real Gross Domestic Product

• 4%

0% 4% 8% 12% 16%

• 4%

0% 4% 8% 12% 16% 1975 1980 1985 1990 1995 2000 2005 2010

Annualized Headline CPI Inflation

0% 4% 8% 12% 16% 0% 4% 8% 12% 16% 1975 1980 1985 1990 1995 2000 2005 2010

Annualized Core CPI Inflaiton

SLIDE 11

Considering that a VAR model is sensitive to the order

• f integration for each of the variables and a unit root

is one type of non-stationary series, I check each of the variables for a unit root.

In regression: Tests: u rea

• y

π π'

Constant + trend: ADF

• 1.84

(0.68)

• 3.76

(0.02)**

• 1.14

(0.92)

• 1.52

(0.82)

• 2.44

(0.35)

• 2.24

(0.47) PP

• 2.06

(0.56)

• 4.23

(0.00)**

• 1.37

(0.87)

• 1.09

(0.93)

• 2.75

(0.22)

• 3.36

(0.06)* Constant: ADF

• 2.37

(0.15)

• 3.63

(0.01)**

• 1.25

(0.65)

• 1.99

(0.29)

• 2.34

(0.16)

• 1.95

(0.31) PP

• 2.07

(0.26)

• 4.10

(0.00)**

• 1.52

(0.52)

• 1.20

(0.68)

• 2.39

(0.15)

• 1.67

(0.44)

In regression: Tests: Δu Δrea Δo Δy Δπ Δπ'

Constant: ADF

• 5.05

(0.00)**

• 7.21

(0.00)**

• 6.96

(0.00)**

• 4.87

(0.00)**

• 4.65

(0.00)**

• 5.01

(0.00)** PP

• 5.52

(0.00)**

• 11.10

(0.00)**

• 9.42

(0.00)**

• 7.96

(0.00)**

• 8.59

(0.00)**

• 7.57

(0.00)**

SLIDE 12

12

• .4
• .3
• .2
• .1

.0 .1 5 10 15 20 25 30 35 40

(a) Accumulated Response of Unemployment Rate

4 8 12 16 20 5 10 15 20 25 30 35 40

(b) Accumulated Response of Real Oil Price Response of Real Economic Activity

• .4
• .2

.0 .2 .4 .6 5 10 15 20 25 30 35 40

(c) Accumulated Response of Output

.0 .2 .4 .6 .8 5 10 15 20 25 30 35 40

Oil Price Shock with Headline Inflation Oil Price Shock with Core Inflation Global Demand for Oil Shock with Headline Inflation Global Demand for Oil Shock with Core Inflation (d) Accumulated Response of Inflation

SLIDE 13

Real Economic Activity & Core Inflation Real Oil Price & Core Inflation

(a) Unemployment Rate (b) Real Economic Activity Period s.e. NRU GDO AS AD Period s.e. NRU GDO AS AD 1 0.2 39.8 40.7 19.2 0.4 1 10.6 38.8 53.5 4.6 3.1 4 0.3 42.0 41.3 12.0 4.7 4 19.4 36.0 61.1 1.4 1.5 8 0.3 40.8 41.6 11.6 5.9 8 22.3 38.6 58.8 1.1 1.5 16 0.3 42.1 41.2 11.0 5.7 16 22.9 40.4 57.1 1.1 1.4 40 0.3 42.2 41.1 10.9 5.7 40 23.0 40.6 56.9 1.1 1.4 (c) Output (d) Core Inflation Period s.e. NRU GDO AS AD Period s.e. NRU GDO AS AD 1 0.7 44.8 35.1 19.4 0.7 1 0.4 6.5 2.2 0.3 90.9 4 0.8 44.2 31.3 15.3 9.2 4 0.5 6.4 17.0 2.2 74.5 8 0.8 42.4 33.4 14.1 10.1 8 0.5 8.2 17.2 3.2 71.4 16 0.8 43.2 33.2 13.6 10.0 16 0.5 8.4 18.6 3.4 69.6 40 0.8 43.3 33.2 13.5 10.0 40 0.5 8.5 18.6 3.4 69.6 (a) Unemployment Rate (b) Real Oil Price Period s.e. NRU OP AS AD Period s.e. NRU OP AS AD 1 0.2 66.0 5.7 27.8 0.6 1 13.3 2.0 95.1 0.1 2.9 4 0.3 76.4 4.7 15.6 3.3 4 14.8 3.2 92.5 0.6 3.7 8 0.3 72.0 7.4 16.9 3.7 8 14.9 3.6 91.5 0.6 4.3 16 0.3 71.9 7.5 17.0 3.7 16 15.0 3.6 91.4 0.7 4.3 40 0.3 71.9 7.5 17.0 3.7 40 15.0 3.6 91.4 0.7 4.3 (c) Output (d) Core Inflation Period s.e. NRU OP AS AD Period s.e. NRU OP AS AD 1 0.7 78.2 8.8 12.1 0.9 1 0.4 3.2 8.1 0.0 88.6 4 0.8 75.4 7.6 9.2 7.8 4 0.5 10.0 12.7 4.3 73.1 8 0.9 72.0 9.6 10.2 8.2 8 0.5 14.5 12.5 5.5 67.6 16 0.9 71.7 9.6 10.2 8.5 16 0.5 14.2 12.5 6.0 67.3 40 0.9 71.7 9.6 10.2 8.5 40 0.5 14.2 12.5 6.0 67.3

13

SLIDE 14

The macroeconomic impacts from a global demand for

• il shock last longer and explain larger shares of

economic volatility than an “oil price shock.”

These results support Bernanke's (2012) view that

monetary policy should be passive to the temporary economic effects from an oil supply shock.

It appears that policy choices should be different

depending on the underlying cause.

Therefore, I provide evidence that affirmatively answers

the question asked in this paper; the economic effects from a global demand for oil shock are greater than those from an oil supply shock.

14

SLIDE 15

Is the Fed able to directly affect global demand for

• il and therefore indirectly affect the price of oil?
• Frankel (2008)—overshooting oil price effect
• Frankel and Rose (2010)—little significance of this effect.

Considering the following:

1.U.S. is the world's largest economy 2.U.S. is the largest consumer of oil 3.Other central banks tend to follow the Fed's actions

It is likely that the impact on global demand from

Fed policies would magnify the effects of a global demand for oil shock.

• Barsky and Kilian (2004) and Anzuini, Pagano, and Pisani

(2012) have discussed this possible channel.

15