do opec members know something the market doesn’t?
TRANSCRIPT
Do OPEC Members Know Something the Market Doesn’t?
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
“Fair Price” Pronouncements and the Market Price of Oil
This presentation reflects the authors’ opinions only – not those of the International Energy Agency (IEA), its member governments, the U.S. Commodities Futures Trading Commission (CFTC), the Commissioners, or other staff at either institution.
Celso Brunett
i
Bahattin Büyükşahi
n
Michel Robe
Kirsten Soneson
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Background
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
“ ‘We expected at the start of the year oil prices between $75 and $80 a barrel, this is a fair price… Oil prices (…) might rise reasonably’, (Saudi Arabia’s King Abdullah) said. On Thursday, U.S. oil crude futures rose $1.38 to $78.05.”
Reuters, December 26, 2009
“WTI (prices) stabilized between $70 to $80 per barrel since the middle of last year. This range is consistent with the ‘fair price’ range for crude oil proposed by King Abdullah of Saudi Arabia at the beginning of 2009.”EIA Short-Term Energy Outlook, U.S. DoE, Feb. 2010
4
Some Quotes
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Does OPEC have Information the Market Doesn’t?
As these quotes illustrates, news agencies, market commentators and even oil analysts often make references to OPEC pronouncements about crude oil prices.
During the oil price surge of 2008, pundits and politicians instead used “fair price” statements by OPEC-linked officials as evidence that sky-high prices were “not caused by physical supply and demand factors”.
Implicit in such views is an implicit assumption that OPEC-related “fair price” pronouncements contain information not already reflected in market prices.
5
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
I. Our Paper
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
OPEC “Fair Price” Announcements’ Impact on the Futures Market Price of Crude Oil
OPEC members, individually or collectively, often comment on what the “fair price” for crude oil should be We construct a sample of all fair price pronouncements in 2000-
10.
No definition of “fair price” in classical economics We define the “fair price” to be the price that a particular OPEC
producer would like to see prevail in the market.
We investigate whether “fair price” statements contain any new information (i.e., news content not already absorbed in the market) & the price effect of such pronouncements.
7
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Related Literature
1. OPEC’s ability to influence the price of crude oil Efficiency of OPEC’s “official prices” (1970’s to mid-1980s):
E.g., Gately et al, EER-77; Verleger, REStat-82; Hubbard, QJE-86, ... Impact of production-related OPEC conferences
announcements do they influence market prices, returns, or implied volatility? Loderer, JF-85; Wirl & Kujundzic, EnJ-04; Horan et al, EnJ-04; Wang
et al, JFutM-08; Demirer & Kutan, EE-10.
2. Efficiency of energy-derivatives markets Do energy futures market react to OPEC conference news?
Draper, JFutM-84; Deaves & Krinsky, JFutM-92.
3. Impact of verbal interventions on markets Extant work focuses on communications in forex (see Blinder
et al, JEL-08) or on interest rate markets (e.g., Fratzscher, EJ-08, JIE-09)
8
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Our Paper
Contribution to the empirical literature
We focus on OPEC statements related to prices, not on statements related to production.
Whereas OPEC conferences and meetings are publicized ahead of time (and are thus predictable), “fair price” pronouncements are not timed with any predictability, and are thus not likely to be forecasted ahead of time.
Methodological contributions Adapt Kalman-filter and ADL methodologies to oil issues Enhance the event-study methodology used in extant oil literature.
9
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Our Findings in a Simple Picture
10
Jan-0
0
Jul-00
Jan-0
1
Jul-01
Jan-0
2
Jul-02
Jan-0
3
Jul-03
Jan-0
4
Jul-04
Jan-0
5
Jul-05
Jan-0
6
Jul-06
Jan-0
7
Jul-07
Jan-0
8
Jul-08
Jan-0
9
Jul-09
Jan-1
0
Jul-10
0
20
40
60
80
100
120
140
160
0
500
1000
1500
2000
2500
3000
WTI Nearby Futures Price
Fair Price (range bars in black)
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Our Findings in Words
If “fair price” pronouncements contained news for non-OPEC market participants, e.g., signals of physical market conditions e.g., signals of a country’s or OPEC’s strategies
Then, we should observe an impact on crude oil futures prices.
In fact, we find that “fair price” announcements add little to pre-existing information in the market have little impact on the futures market price of crude oil.
11
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
II. Data & Graphical Analysis
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
“Fair Price” Series
“Fair Price” Data – Hand-collected time series of quotes made by OPEC
officials or senior government officials from OPEC countries: Source: news articles in the Lexis-Nexis Academic and Westlaw databases
Search Criteria: “fair price” or “just price”; Over 1,000 articles returned over a 10-year period Total of 120+ events Filtering leaves 78 unique observations (1st, March 2000; last, Nov. 2010)
Information compiled (cross-checked with industry newsletters) Release date of the story Statement date (when the official stated the “fair price”) Country of origin of the OPEC official “Fair price” level or range When available –
whether the official agreed with the current market price of oil the benchmark crude referenced
13
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Market Price Series
Market Price Data – West Texas Intermediate (WTI) light sweet crude oil futures
contract traded on the NYMEX. World’s most liquid crude oil contract. Settlement price of the nearby-futures contract is used
= “closest-to-delivery contract with the highest open interest”. Sample period – January 2000 to December 2010.
Control Variables – Production announcements
Dates Direction of production announcement Only 13 of the 78 events had production announcements within 5
days of (before or after) the “fair price” pronouncement. Survey of economists regarding production expectations
14
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Q1. Do “OPEC” Officials Agree with the Current Market Price?
15
Jan-0
0
Jul-00
Jan-0
1
Jul-01
Jan-0
2
Jul-02
Jan-0
3
Jul-03
Jan-0
4
Jul-04
Jan-0
5
Jul-05
Jan-0
6
Jul-06
Jan-0
7
Jul-07
Jan-0
8
Jul-08
Jan-0
9
Jul-09
Jan-1
0
Jul-10
10
25
40
55
70
85
100
115
130
145
WTI Nearby Futures Price
Agree with Price
Disagree with Price
Cannot Tell
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Explanation of Graphical Analysis 1
“Fair price” statements subdivided into 3 groups:
Official agrees with the market price 17 times (green triangles). Mostly after mid-2009
Official disagrees with the market price 25 times (red dots). Mostly before mid-2009
Official does not say 36 times (purple squares).
The graph suggests that, even when they are not explicit about it, officials often seemed to disagree with market prices before 2009
• From Summer 2003 to Summer 2008 – market price was viewed as “too high” • From Fall 2008 to Spring 2009 – market price was viewed as “too low”
(market price collapsed yet “fair price” remained in the $75-100 range).
16
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Q2. Heterogeneity within OPEC?
17
Jan-0
0
Jul-00
Jan-0
1
Jul-01
Jan-0
2
Jul-02
Jan-0
3
Jul-03
Jan-0
4
Jul-04
Jan-0
5
Jul-05
Jan-0
6
Jul-06
Jan-0
7
Jul-07
Jan-0
8
Jul-08
Jan-0
9
Jul-09
Jan-1
0
Jul-10
10
25
40
55
70
85
100
115
130
145
WTI Nearby Futures Price
VILAN
Saudi Arabia
Other OPEC/OPEC Consensus
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Explanation of Graphical Analysis 2
Idea: Country-specific factors may influence which price an individual cartel member would like to see prevail in the market.
OPEC Member Classification - OPEC “hawks” (red): Venezuela, Algeria, Iran and Libya
Tend to advocate higher prices than current market prices? Saudi Arabia (green):
With ample oil reserves, tends to advocate lower prices? Others (purple): all remaining individual member countries or a “consensus” by
OPEC
The figure suggests that “hawks” do not make “fair price” statements more bullish than the other groups, yet: Hawks more vocal in support of higher prices after price collapse at the end of 2008 Venezuela –
Favored production cuts through most of sample period Exception – constant at $100 as “fair price” during price spike and crash of 2008
18
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Q3. Which moves first – “fair price” or market price?
19
Jan-0
0
May-
00
Sep-0
0
Jan-0
1
May-
01
Sep-0
1
Jan-0
2
Jun-0
2
Oct
-02
Feb-0
3
Jun-0
3
Oct
-03
Feb-0
4
Jun-0
4
Nov-
04
Mar-
05
Jul-05
Nov-
05
Mar-
06
Jul-06
Nov-
06
Apr-
07
Aug-0
7
Dec-
07
Apr-
08
Aug-0
8
Dec-
08
Apr-
09
Sep-0
9
Jan-1
0
May-
10
Sep-1
0
10
25
40
55
70
85
100
115
130
145
WTI Nearby Futures Price
Fair Price
Fair Price Trend
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Explanation of Graphical Analysis 3
We ask whether “fair price” statements lead or lag market prices.
The figure strongly suggests that “fair price” statements lag market prices Interpretation: they contain no new information.
We now turn to more formal analyses to test this apparent relationship between the two time series.
20
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
III. Formal Methodologies
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
1st Methodology - Event Study
Event Study (see e.g. MacKinlay, JEL-97)
Our interest – Arrival of “true news” (information content into the market) Compare the nearby oil futures price before & after the event
We use two non-parametric tests of statistical significance
Event and event window – Event = “fair price” statement date. The event separates all preceding and succeeding observations
captures the normal performance and after-announcement performance of the price of oil.
We use windows of 2, 3, 4, and 5 days for robustness.
22
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Event Study continued
Non-parametric tests: Humpage, JMCB-99; Fatum and Hutchinson, EJ-03.
Criterion (1): “Direction Criterion” “Success” = “post-announcement market price moves in the direction
suggested by ‘fair price’ level”. Formally:
To test if the direction of futures prices is random or systematic following the event, we use a non-parametric sign test: • Null : the direction is random.• Equal probability that the post-event return is
positive or negative. X+ stands for the # of successes.
23
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Event Study continued
Non-parametric tests: Humpage, JMCB-99; Fatum and Hutchinson, EJ-03.
Criterion (2): “Smoothing Criterion” “Success” = “the ‘fair price’ pronouncement smoothes prior
market-price movements”. Formally: if it is successful according to “direction criterion” (1) or
if :
Judging success on the “smoothing criterion” is only meaningful if the official marking the statement is “leaning against the wind”. • i.e., he is trying to reverse or to slow a prior market price trend.
First case: signal = “prices are too high or increasing too fast”.Second case: signal = “prices are too low or should not be falling”.
• Broad “smoothing criterion” – 75%; stricter “reversal criterion” – 50%.
24
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
2nd Methodology - Kalman Filter
Question – How close is the “fair price” to the “true” (i.e., filtered) market price of crude oil? Filtering lets us separate the “true price” from the noise component in the
market price to answer the above question.
Results – Actual price and the forecasted price are nearly identical Until mid-2008, “fair price” levels fall within the confidence band
around the forecast - meaning “fair prices” reflect the same information as do market prices.
In mid-2008, “fair prices” trickled outside the band. But… “Fair price” statements during this period were worse forecasters
(of one-month ahead prices) than were market-price-based forecasts. Bottom line: “Fair price” statements contain no useful information
for improving market-price forecasts.
25
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
3rd Methodology – ADL Model
Auto-Regressive Distributed Lag (ADL) process allows us to forecast crude oil futures prices from past market price
information.
If “fair price” announcements do not provide any additional news to the futures market, then the forecasted price and the “fair price” should be very close.
Two versions: “market only” vs. “market + FP”
We use weekly data (every Wednesday) and monthly data (the last observation of every month).
Rolling windows are 1 year (52 weeks) and 3 years (36 months). One-step-ahead forecasts with optimal lag length (q) selected by the Akaike
Information Criterion (AIC) and likelihood ratio tests.
26
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
IV. Results
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
1. Event Study Analyses & Results
Results –
“Fair price” pronouncements are not successful (either at affecting the direction of the market or at slowing down the pace of prior market price movements).
We cannot reject (at any statistical significance level) the hypothesis that the post-pronouncement return is random.
Robust results regardless of: Divisions based on a concomitant OPEC oil-production announcement Who is making the statements Different event-window lengths (2- vs. 5- day) Different time periods (SPARE)
Tables –
Panel A - “direction criterion” Panel B - reversals (“leaning against the wind”) Panel C - “smoothing criterion”
Robust –
28
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Event Study Tables
2-Day Returns; No Concomitant Production News 2-Day Returns Amid News of Production Changes
29
Panel A
Nonparametric Sign Test – Direction Criteria: No production cut or increase (2 Day)
# of Events # of Success z-value p-value
FP>P- 22 13 0.8528 0.1969 FP<P- 43 15 -1.9825 0.9763 Total 65 28 -1.1163 0.8679
Panel B
Reversal Criteria Test – Leaning Against the Wind: No production cut or increase (2 Day)
# of Events # of Success z-value p-value
FP>P- 11 7 0.9045 0.1829 FP<P- 21 5 -2.4004 0.9918 Total 32 12 -1.4142 0.9214
Panel C
Smoothing Criteria Test - Leaning Against the Wind: No production cut or increase (2 Day)
# of Events # of Success z-value p-value
FP>P- 11 10 0.7035 0.2409 FP<P- 21 12 -1.0911 0.8624 Total 32 22 -0.4714 0.6813
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Event Study Tables cont.
5-Day Returns; No Concomitant Production News
5-Day Returns Amid News of Production Changes
30
Panel A
Nonparametric Sign Test – Direction Criteria: No production cut or increase (5 Day)
# of Events # of Success z-value p-value
FP>P- 21 10 -0.2182 0.5864 FP<P- 44 24 0.6030 0.2732 Total 65 34 0.3721 0.3549
Panel B
Reversal Criteria Test – Leaning Against the Wind: No production cut or increase (5 Day)
# of Events # of Success z-value p-value
FP>P- 8 4 0.0000 0.5000 FP<P- 18 9 0.0000 0.5000 Total 26 13 0.0000 0.5000
Panel A
Smoothing Criteria Test – Leaning Against the Wind: No production cut or increase (5 Day)
# of Events # of Success z-value p-value
FP>P- 8 7 0.4714 0.3187 FP<P- 18 14 0.1571 0.4376 Total 26 21 0.3922 0.3474
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
“Fair Price” Events, Production News, and Median Returns
Production News and Median Returns after all “Fair Price” Events, 2000-2009
31
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Different regimes?
31
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
Maybe – but Results are Robust
31
Jan-0
0M
ay-
00
Sep-0
0Ja
n-0
1M
ay-
01
Sep-0
1Ja
n-0
2Ju
n-0
2O
ct-0
2F
eb-0
3Ju
n-0
3O
ct-0
3F
eb-0
4Ju
n-0
4N
ov-
04
Mar-
05
Jul-05
Nov-
05
Mar-
06
Jul-06
Nov-
06
Apr-
07
Aug-0
7D
ec-
07
Apr-
08
Aug-0
8D
ec-
08
Apr-
09
Sep-0
9Ja
n-1
0M
ay-
10
Sep-1
0
10
25
40
55
70
85
100
115
130
145WTI Nearby Futures PriceFair Price
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
2. ADL Analyses & Results
Analyses – We use monthly (rather than weekly) data
account for possibility that “fair price” pronouncements forecast supply and demand conditions over longer period of time than a few days.
Result #1 – “Model-free” specification We use past market prices only
R1: Actual and forecasted prices are nearly indistinguishable. i.e., ADL model produces good one-step-ahead forecasts.
R2: “Fair price” outside confidence interval in H2-2008/Q1-2009 YET– “fair price” levels in that period were worse
forecasters of 1-month-ahead prices than were market-based forecasts.
34
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
ADL Graph
ADL-based Market Price Forecasts vs. “Fair Price” Levels (monthly)
35
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
ADL Analyses & Results
Result #2 – beyond “Model-free” specification Add “fair price” surprise
Either FairPrice(t) – MarketPrice(t) Or FairPrice (t) – FairPrice(t-1)
to ADL model of oil market returns and see if it helps explain oil-market returns
R3: Statistically speaking, we again find that adding the fair price information does not help.
34
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
V. Conclusions
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
No New Information
“Fair price” statements seem to have little influence on crude oil futures prices.
No evidence of price reversals (not even evidence of slowing down) when “fair price” announcements “lean against the wind” (differ from the current market trend).
“Fair price” events supply no extra news to oil futures market participants and no new information content to the market.
Going forward – Should commentators, reporters, and policy makers treat “fair price” pronouncements as informative when attempting to explain market prices or making policy decisions?
37
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
The Road Ahead
Not so fast?
Where next? Testable theoretical model of “cheap talk” (price
statements) + costly signaling (output levels)?
“If Saudi Arabia wants to stabilize oil prices, (it) will have to back its verbal commitment to a target of $70-80 per barrel with adjustments in production. Verbal interventions alone will not be enough to anchor expectations or market prices (…) Saudi officials admit (…) expectations management on its own is not enough to stabilize prices.”Oil and Gas News, Dec. 28, 2010
37
Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011
The End