Determinants of Agricultural and Mineral

oppositemincedManagement

Oct 28, 2013 (3 years and 10 months ago)

67 views

Determinants

of Agricultural and Mineral

Commodity Prices



Jeffrey A. Frankel,


Harvard University, &


Andrew K. Rose,


University of California, Berkeley




Reserve Bank of Australia, August 2009.


2

Determination of Prices for Oil and Other
Mineral & Agricultural Commodities


Predominantly microeconomic.


Still, difficult to ignore macroeconomic
influences sometimes.


Examples: many commodity prices
move far in same direction at the
same time:


The decade of the 1970s.


The decade of the 2000s.



3



Increase in
oil price
can
be explained by “peak oil”
fears, a risk premium on
Gulf instability, or political
developments in Russia,
Nigeria or Venezuela...





Some
farm prices

might be explained by
drought in Australia,
shortages in China, or
ethanol subsidies in the US.





4

But it
Cannot be Coincidence
that
almost all commodity prices rose
together during much of the decade,
and peaked abruptly in mid
-
2008.

Our Innovation


Combine
Macro and Micro
Determinants
of Commodity Prices


Hope: Get macro swings nested inside
well
-
grounded micro model


Need Good Micro Data on Determinants
of Individual Commodities

5

6

Three “Aggregate” Theories Explain the
Recent Rise of Commodity Prices

1.
Destabilizing Speculation
.


Storability & Homogeneity



=> Asset
-
like Speculation


2.
Monetary:


Low Real Interest Rates


Or High Expected Inflation

3.
Global Demand Growth.


Actual/Future Growth (China …)



Issues Exist with All Three
“Explanations”


In Theory, Speculation may be Stabilizing



Empirical Issues with All Three Theories

7

8

Counter
-
Evidence to Claims of
Destabilizing Speculation

1. Futures price of oil initially lagged behind
spot price.

2. High volume of trading


net short position

3. Commodities that lack futures markets are
as volatile as those that have them.

4. Historical efforts to ban speculative futures
markets have failed to reduce volatility.

9

Monetary Explanation

1.
Some argue that high prices for oil & other
commodities in the 1970s were not exogenous,

but rather a result of easy monetary policy.



Perhaps inflation directly raises commodity prices? Commodities
may be an inflation hedge.

2.
Conversely, a rise in US real interest rates in the
early 1980s
.
helped drive commodity prices down
.


3.
The Fed cut real interest rates sharply,2001
-
04,

and again in 2008
-
09. Did this help push prices
first up, then down?





10

High Interest Rates in Theory

1.
Lower inventory demand;
and

2.
Encourage faster pumping of oil,

mining of deposits, harvesting of crops, etc.,


because owners can invest the proceeds at interest
rates higher than the return to saving the reserves.

3.
Both channels


fall in demand and rise in
supply


work to lower commodity price.


11

But … Counter
-
arguments Exist


Inventories of oil & other commodities said to
be low in 2008, contrary to the theory
(Krugman, Kohn)



Perhaps inventory numbers


do not capture all inventories, or


are less relevant than (larger) reserves.


King of Saudi Arabia (2008):

“we might as well leave the reserves

in the ground for our grandchildren.”



How Important are Monetary Effects?

12

Global Boom Theory Reasonable?


Sub
-
prime Mortgage Crisis

hit US, August 2007.


Thereafter, Growth Forecasts Fell Globally


But Commodity Prices did not
Decline
; their
rise actually
Accelerated
.

Quick Peek at Aggregate Data: Little


``

13

But Perhaps Too Macro?


Need to Control for Micro Determinants
of Commodity Prices


Our Objective:
Integrate

Micro and Macro
Commodity Price Determination


Theory


Empirical Estimation

14

15

“Overshooting” Theory of
Real Commodity Prices


s

≡ the spot price,


S ≡ its long run equilibrium,


p
≡ the economy
-
wide price index,


q ≡ s
-
p
, the real price of the commodity,
and


Q

the long run equilibrium real price of
the commodity;


all in log form.

16

Derive Relationship for Real
Commodity from Two Equations:



Regressive Expectations (can be Rational):


E (Δs) =
-

θ (q
-
Q)

+
E(
Δp
)



“Arbitrage
-
like” condition links Inventories & Bonds:


E Δs + c =
i




where
c ≡ cy


sc


rp

.



cy


convenience yield from holding the stock (e.g., the insurance value of
having an assured supply of a critical input in the event of a disruption)


sc



storage costs (e.g., rental rate on oil tanks, etc.)


rp




E Δs


(f
-
s)


risk premium,

>0 if being long in commodities is risky, and


i



the interest rate

17

Combining:



q
-

Q =
-

(1/
θ
) (i
-

E(
Δ
p
)


c)



This inverse relationship between q & r
has already been somewhat studied


Event studies
(monetary announcements)


Regressions of
q
against
r

in Frankel (2008):


Significant for half of the individual commodities


and in a panel study


and for various aggregate commodity price indices


But much is left out of this equation.


Especially variation in
c



18

Observable Manifestations of
Convenience Yield, Storage Costs, &
Risk Premium (c)

1. Inventories

Storage costs rise with inventory


Measured with World inventories where possible, US
otherwise


Could also estimate an inventory equation


19

Other Determinants

2. Real GDP


Transactions Demand for Inventories,
determinant of convenience yield
cy


Measured with real World GDP,


Also try OECD output gap, de
-
trend, G
-
7, IP …


3. The spot
-
futures spread,
s
-
f


High spread (“normal backwardation:) signifies
low speculative return, hence negative effect on
inventory demand and prices


Measurement more straightforward

20

Uncertainty Measures

4. Medium
-
term volatility

(σ)


Volatility a determinant of convenience yield,
and so of commodity prices


May also be determinant of risk premium


Measured as standard deviation of spot price


Can also extract implicit forward
-
looking expected
volatility from options prices


21

5. Risk

(political, financial, & economic)


Theoretical effect ambiguous:


Risk a determinant of
cy

(fear of

supply disruption), should have

a
positive

effect on commodity prices


Also a determinant of
rp
, risk

premium, should have a
negative

effect on prices


Measured (e.g., for oil) by weighted average of
(inverse) political risk for 12 top (oil) producers


Data availability issues; hence not always included



22

Complete Equation



q = Q
-

(1/
θ
) r
+

(1/
θ
)

γ
(Y) + (1/
θ

(
σ
)



-

(1/
θ
)
Φ (
INVENTORIES)
-
δ(
s
-
f
)


Objective: Determine (log) real commodity
price


3 Micro determinants


Volatility; spread; inventories


2 Macro determinants


World GDP; real interest rates


23

Estimation Strategy


Gather, use dis
-
aggregated data on 11
commodity panel


Annual data from 1960s to 2008


Commodities, span, frequency chosen to
maximize data availability

24

Booms around 1974
-
75 and 2008

25

Table 3a: Panel Results,

for log

real commodity prices,


Ln
(World
Real GDP)

Volatility

Spot
-
Futures
Spread

Inven
-
tories


Real
US

interest
rate

.60

2.29**

-
.003*

-
.15**

-
.01

(.27)

(.40)

(.001)

(.02)

(.01)

** (*) => significantly different from zero at .01 (.05) significance level.


Robust standard errors in parentheses; Intercept & trend included, not reported.

Results Seem Sensible


Micro Factors all “correctly” signed


Statistically significant


Macro Factors correctly signed


World GDP: statistically marginal effect


Real Interest Rate
consistently unreliable


Biggest Disappointment

26

Results Also Robust


Results insensitive to exact econometric
specification, model of world activity


Many variants reported in Table 3a


Results from first
-
differences in Table 3b


Possibly relevant because of (lack of) co
-
integration


27

Reasonable Fit to Data

28

29

Table 4: To Look for Bandwagon
Expectations, Add Lagged Rate
of Commodity Price Rise


Ln
(World

Real GDP)

Volatility

Spot
-
Futures
Spread

Inven
-
tories


Real
US

interest
rate

Lag of
Nominal
Price

Growth

.50

1.84**

-
.004**

-
.13**

.00

.0061**

(.27)

(.40)

(.001)

(.02)

(.01)

(.0005)

** (*) => significantly different from zero at .01 (.05) significance level.


Robust standard errors in parentheses; Intercept & trend included, not reported.

Bandwagon Effects!


Commodity Prices Positively,
Significantly affected by Lagged Growth
in
Nominal

Commodity Price


Small but Insensitive Effect


Another Inefficiency in Commodity
Markets?


Helps Explain Recent Run
-
Up (somewhat)

30

31

Table 5: To Look for Another
indicator of Monetary Ease,

Add Aggregate Inflation


Ln
(World

Real GDP)

Volatility

Spot
-
Futures
Spread

Inven
-
tories


Real
US

interest
rate

Inflation

-
2.11**

2.12**

-
.003**

-
.14**

.02

.082**

(.61)

(.27)

(.001)

(.02)

(.01)

(.015)

** (*) => significantly different from zero at .01 (.05) significance level.


Robust standard errors in parentheses; Intercept & trend included, not reported.

Inflation Effects!


Commodity Prices Positively,
Significantly affected by Inflation


Again: Robust Results, but Small


Probably negligible effect for conduct of
monetary policy


Hedge against Inflation?


Doesn’t Explain Recent Run
-
Up

32

33

Other Tests: Indices


Construct Commodity Price Indices


Use 6 Weighting Schemes


Dow
-
Jones/AIG; S&P GCSI; CRB
Reuters/Jefferies;
Grilli
-
Yang; Economist; Equal


3 Different Periods of Time


Data availability => longer span has fewer
commodities


Similar (Weaker) Results


Micro work OK; poor real interest rate results


34

Other Tests: Hi
-
Tech



Unit root tests



Philips
-
Perron on individual commodities


Panel unit
-
root tests



Co
-
integration tests


Johansen on individual commodities


Panels too


Vector error correction results

35

Overall Model Performance


The commodity
-
specific explanatory factors
work (surprisingly) well:


Inventory holdings


Spot
-
futures spread


Volatility


Macroeconomic variables work (surprisingly)
poorly:


Economic activity


(Especially) Real interest rates

36

Possible Extensions


Survey data as direct measure of
expectations


Higher Frequency data (on fewer
commodities, shorter time
-
span)


Modeling non
-
linearities


Estimate simultaneous system in
inventories, expectations, and commodity
prices, tied directly to the theory

Conclusion


Model works reasonably:


Micro determinants work well


Macro phenomena
not
as important


Real growth raises real commodty prices


As does inflation


But real interest rate channel fails here.


Evidence of Bandwagon Effects


“Speculative Bubble” possible in Commodities


Helps explains 2007
-
9 boom and bust?


37

38

Appendices


Graphs of data


Why American interest rates?


Commodity
-
specific Results


Full Panel Results

39

40

41

42

43

Why American Real
Interest Rates?


Assume commodity markets integrated


If so, denomination doesn’t matter


Data availability issues for G
-
3/G
-
7
interest rates


Inevitable EMU issues

44

Table 2a: Commodity
-
Specific Results

Real World
GDP (+)

Volatility

+

Spot
-
Future
Spread (
-
)

Inventories

(
-
)

Real Interest
Rate (
-
)

Corn

1.53*

(.69)

1.52

(.89)

-
.003

(.003)

-
.18

(.17)

-
.01

(.02)

Copper

.03

(.68)

1.92**

(.54)

-
.005

(.003)

-
.21**

(.06_

-
.03

(.01)

Cotton

.66

(.85)

1.07

(.57)

-
.002

(.002)

-
.12

(.14)

.01

(.01)

Cattle

7.37**

(1.03)

-
.65

(.34)

-
.007

(.002)

2.37**

(.48)

-
.06**

(.01)

Hogs

-
.57

(1.64)

.64

(.71)

-
.004*

(.002)

.18

(.31)

-
.03**

(.01)

Oats

2.66**

(.71)

3.28

(1.69)

-
.006**

(.002)

-
.59**

(.11)

-
.02

(.01)

Oil

.05

(8.60)

.57

(1.69)

-
.003

(.003)

-
2.52

(5.02)

-
.01

(.07)

Platinum

1.22

(2.17)

1.78*

(.87)

.002

(.002)

-
.21**

(.03)

.08**

(.01)

Silver

2.69

(2.13)

3.32**

(.73)

.003

(.003)

-
.37*

(.18)

.01

(.03)

Soybeans

1.94**

(.70)

2.68**

(.55)

-
.001

(.002)

-
.05

(.07)

-
.01

(.01)

Wheat

-
5.98*

(2.79)

1.90**

(.47)

.008*

(.003)

-
1.42**

(.27)

.03

(.02)

45

Full Panel Results
Table 3a: Levels

Real World
GDP (+)

Volatility

(+)

Spot
-
Future
Spread

(
-
)

Inventories

(
-
)

Real Interest
Rate (
-
)

Basic

.60

(.27)

2.29**

(.40)

-
.003*

(.001)

-
.15**

(.02)

-
.01

(.01)

Drop Fixed Effects

.56

(.31)

2.65

(1.40)

-
.023**

(.006)

-
.20**

(.03)

.02

(.04)

Substitute Time
Effects

n/a

2.32

(1.80)

-
.026**

(.007)

-
.20**

(.01)

n/a

Time and Fixed
Effects

n/a

1.61**

(.29)

-
.002*

(.001)

-
.13**

(.01)

n/a

Drop Spread

.58

(.30)

2.36**

(.38)

n/a

-
.15**

(.02)

-
.01

(.01)

Growth (not log) of
World GDP

-
.01

(.01)

2.36**

(.40)

-
.003

(.001)

-
.15**

(.02)

-
.00

(.01)

OECD Output Gap

.01

(.01)

2.34**

(.44)

-
.002*

(.001)

-
.15**

(.02)

-
.01

(.01)

HP
-
Filtered GDP

2.35

(1.47)

2.32**

(.43)

-
.003*

(.001)

-
.14**

(.02)

-
.01

(.01)

Add Quadratic
Trend

.48

(.40)

2.30**

(.40)

-
.003*

(.001)

-
.15**

(.02)

-
.01

(.01)

46

Table 3b: Panel Results, First
-
Differences

Real

World GDP

+

Volatility

+

Spot
-
Future Spread

-

Inventories

-

Real Interest Rate

-

Basic

.03**

(.01)

.75**

(.24)

-
.002**

(.001)

-
.10*

(.05)

.00

(.01)

Drop Fixed Effects

.03**

(.01)

.78**

(.17)

-
.002**

(.001)

-
.11**

(.04)

.00

(.01)

Substitute Time
Effects

n/a

.55**

(.19)

-
.002**

(.001)

-
.08*

(.04)

n/a

Time and Fixed
Effects

n/a

.53**

(.18)

-
.002**

(.001)

-
.07

(.04)

n/a

Drop Spread

.04**

(.01)

-
.0020**

(.0005)

-
.10

(.05)

-
.00

(.01)

OECD Output Gap

.03**

(.01)

.77*

(.25)

-
.0018**

(.0005)

-
.12*

(.04)

.01

(.01)

HP
-
Filtered GDP

4.91**

(.97)

.78*

(.23)

-
.002**

(.001)

-
.12*

(.04)

.01

(.01)

Add Quadratic Trend

.03**

(.01)

.75**

(.24)

-
.002**

(.001)

-
.10*

(.05)

.00

(.01)

47

Table 4: Panel Results, Bandwagons

Real World
GDP (+)

Volatility

(+)

Spot
-
Future
Spread
(
-
)

Inventories

(
-
)

Real Interest
Rate
(
-
)

Lagged Price

Change (+)

Basic

.50

(.27)

1.84**

(.40)

-
.004**

(.001)

-
.13**

(.02)

.00

(.01)

.0061**

(.0005)

Drop Fixed Effects

.57

(.31)

1.92

(1.42)

-
.025**

(.006)

-
.19**

(.03)

.04

(.04)

.0104*

(.0044)

Substitute Time
Effects

n/a

1.84

(1.90)

-
.028**

(.007)

-
.19**

(.01)

n/a

.0101

(.0067)

Time and Fixed
Effects

n/a

1.37**

(.28)

-
.004**

(.001)

-
.12**

(.01)

n/a

.0050**

(.0008)

Drop Spread

.48

(.32)

2.01**

(.37)

-
.14**

(.02)

-
.00

(.01)

.0053**

(.0005)

Growth (not log) of
World GDP

-
.01

(.01)

1.90**

(.40)

-
.005**

(.001)

-
.13**

(.02)

.01

(.01)

.0061**

(.0005)

OECD Output Gap

-
.00

(.01)

1.90**

(.43)

-
.004**

(.001)

-
.13**

(.02)

.01

(.01)

.0063**

(.0005)

HP
-
Filtered GDP

-
.71

(1.58)

1.92**

(.42)

-
.004**

(.001)

-
.13**

(.02)

.01

(.01)

.0062**

(.0005)

Add Quadratic
Trend

.26

(.37)

1.85**

(.41)

-
.004**

(.001)

-
.13**

(.02)

.01

(.01)

.0062**

(.0005)

Drop post
-
2003 data

1.21**

(.28)

1.26

(.58)

-
.004**

(.001)

-
.11**

(.04)

.01

(.01)

.0049**

(.0005)

With AR(1)
Residuals

2.08*

(.81)

.89**

(.13)

-
.0033**

(.00004)

-
.10**

(.03)

.00

(.01)

.0031**

(.0004)

48

Table 5: Panel Results,
Adding Inflation

Real

World
GDP

+

Volatility

+

Spot
-
Future
Spread

-

Inventories

-

Real Interest
Rate

-

Inflation

Basic

-
2.11**

(.61)

2.12**

(.27)

-
.0032**

(.0007)

-
.14**

(.02)

.019

(.012)

.082**

(.015)

Drop Fixed
Effects

.70*

(.32)

2.25

(1.43)

-
.023**

(.006)

-
.19**

(.03)

.040

(.038)

.075

(.041)

Drop Spread

-
2.04**

(.63)

2.21**

(.26)

-
.15**

(.02)

.015

(.012)

.079**

(.015)

Growth (not log)
of World GDP

.02

(.01)

2.01**

(.32)

-
.0027**

(.0007)

-
.15**

(.02)

.006

(.011)

.058**

(.010)

OECD Output
Gap

-
.00

(.01)

2.09**

(.28)

-
.0030**

(.0007)

-
.15**

(.02)

.014

(.012)

.083**

(.014)

HP
-
Filtered
GDP

.19

(1.64)

2.03**

(.33)

-
.0031**

(.0008)

-
.15**

(.02)

.005

(.013)

.051**

(.009)

Add Quadratic
Trend

-
2.47**

(.76)

2.14**

(.27)

-
.0032**

(.0006)

-
.14**

(.02)

.017

(.011)

.085**

(.015)

49

50