Do OPEC Members Know Something the Market Doesnt

Do OPEC Members Know Something the Market Doesn’t?

“Fair Price” Pronouncements and the Market Price of Oil Celso Brunetti Bahattin Büyükşahin Michel Robe Kirsten Soneson 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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Background Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Some Quotes 4 “ ‘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. Do. E, Feb. 2010 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Does OPEC have Information the Market Doesn’t? 5 � 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. 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 7 � 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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Related Literature 8 � 1. OPEC’s ability to influence the price of crude oil Efficiency of OPEC’s “official prices” (1970’s to mid-1980 s): � 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, En. J-04; Horan et al, En. J-04; Wang et al, JFut. M-08; Demirer & Kutan, EE-10. � 2. Efficiency of energy-derivatives markets Do energy futures market react to OPEC conference news? � Draper, JFut. M-84; Deaves & Krinsky, JFut. M-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) Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Our Paper 9 �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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Our Findings in a Simple Picture 10 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Our Findings in Words 11 �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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

II. Data & Graphical Analysis Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

“Fair Price” Series 13 �“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 (1 st, 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 � Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Market Price Series 14 �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 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Q 1. Do “OPEC” Officials Agree with the Current Market Price? 15 145 WTI Nearby Futures Price 130 Agree with Price 115 Disagree with Price 100 Cannot Tell 85 70 55 40 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011 Jul-10 Jan-10 Jul-09 Jan-09 Jul-08 Jan-08 Jul-07 Jan-07 Jul-06 Jan-06 Jul-05 Jan-05 Jul-04 Jan-04 Jul-03 Jan-03 Jul-02 Jan-02 Jul-01 Jul-00 Jan-00 10 Jan-01 25

Explanation of Graphical Analysis 1 16 � “Fair price” statements subdivided into 3 groups: Official agrees with the market price 17 times (green triangles). � Official disagrees with the market price 25 times (red dots). � Mostly after mid-2009 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). Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Q 2. Heterogeneity within OPEC? 17 145 WTI Nearby Futures Price 130 VILAN 115 Saudi Arabia 100 Other OPEC/OPEC Consensus 85 70 55 40 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011 Jul-10 Jan-10 Jul-09 Jan-09 Jul-08 Jan-08 Jul-07 Jan-07 Jul-06 Jan-06 Jul-05 Jan-05 Jul-04 Jan-04 Jul-03 Jan-03 Jul-02 Jan-02 Jul-01 Jul-00 Jan-00 10 Jan-01 25

Explanation of Graphical Analysis 2 18 � 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 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

10 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011 Sep-10 May-10 Jan-10 Sep-09 Apr-09 Dec-08 Aug-08 Apr-08 Dec-07 Aug-07 Apr-07 Nov-06 Jul-06 Mar-06 Nov-05 Jul-05 Mar-05 Nov-04 Jun-04 Feb-04 Oct-03 Jun-03 Feb-03 Oct-02 Jun-02 Jan-02 Sep-01 May-01 115 Jan-01 130 Sep-00 May-00 Jan-00 Q 3. Which moves first – “fair price” or market price? 19 145 WTI Nearby Futures Price Fair Price Trend 100 85 70 55 40 25

Explanation of Graphical Analysis 3 20 �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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

III. Formal Methodologies Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

1 st Methodology - Event Study 22 �Event Study (see e. g. Mac. Kinlay, 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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Event Study continued 23 �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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Event Study continued 24 �Non-parametric tests: Humpage, JMCB-99; Fatum and Hutchinson, EJ-03. � Criterion (2): “Smoothing Criterion” “Success” = “the ‘fair price’ pronouncement smoothes prior marketprice 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%. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

2 nd Methodology - Kalman Filter 25 � 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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

3 rd Methodology – ADL Model 26 � 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. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

IV. Results Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

1. Event Study Analyses & Results 28 � 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 postpronouncement 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 – Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Event Study Tables 29 2 -Day Returns; No Concomitant Production News Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011 2 -Day Returns Amid News of Production Changes

Event Study Tables cont. 30 5 -Day Returns; No Concomitant Production News Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011 5 -Day Returns Amid News of Production Changes

“Fair Price” Events, Production News, and Median Returns 31 Production News and Median Returns after all “Fair Price” Events, 2000 -2009 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

Different regimes? 31 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

10 Jan-00 May-00 Sep-00 Jan-01 May-01 Sep-01 Jan-02 Jun-02 Oct-02 Feb-03 Jun-03 Oct-03 Feb-04 Jun-04 Nov-04 Mar-05 Jul-05 Nov-05 Mar-06 Jul-06 Nov-06 Apr-07 Aug-07 Dec-07 Apr-08 Aug-08 Dec-08 Apr-09 Sep-09 Jan-10 May-10 Sep-10 Maybe – but Results are Robust 31 145 130 115 WTI Nearby Futures Price 100 85 70 55 40 25 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

2. ADL Analyses & Results 34 �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 � R 1: Actual and forecasted prices are nearly indistinguishable. i. e. , ADL model produces good one-step-ahead forecasts. � R 2: “Fair price” outside confidence interval in H 2 -2008/Q 1 -2009 YET– “fair price” levels in that period were worse forecasters of 1 -month-ahead prices than were market-based forecasts. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

ADL Graph 35 ADL-based Market Price Forecasts vs. “Fair Price” Levels (monthly) Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

ADL Analyses & Results 34 �Result #2 – beyond “Model-free” specification Add “fair price” surprise � Either � Or Fair. Price(t) – Market. Price(t) Fair. Price (t) – Fair. Price(t-1) to ADL model of oil market returns and see if it helps explain oil-market returns R 3: Statistically speaking, we again find that adding the fair price information does not help. Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

V. Conclusions Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

No New Information 37 � “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? Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

The Road Ahead 37 �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 Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011

The End Brunetti, Büyükşahin, Robe & Soneson – USAEE 2011