Investisseurs indiciels et comportements spéculatifs: l exemple des marchés dérivés agricoles Steve Ohana, ESCP Europe et Riskelia EIFR 27 février 2013
Outstanding commodity price fluctuations by their amplitude, duration and synchronization Source: Riskelia Historical perspective on commodity prices: Helbling, T. 2008. The Current Commodity Price Boom in Perspective. In International Monetary Fund, World Economic Outlook. Washington, D.C.
The traditional role of speculators in commodity derivatives markets Hedgers Speculators (Hedge funds) Hedgers transfer price risk to speculators Speculators apply different strategies to exploit predictable patterns and systematic risk premiums in commodities derivatives markets It is generally agreed that they provide liquidity to hedgers and improve price discovery
The emergence of index speculation Long/short Select maturities & commodities Hedgers Speculators Hold proprietary information Exploit sytematic risk premiums Exploit predictable patterns Index investors Asset allocators Pension funds Insurance companies etc.. Long only Don t select mat. nor comm. Do not trade based on fund. Research of diversification Research of long term performance
The 5 main commodity indices
The index investment industry COMMODITY INDEX INVESTING AND COMMODITY FUTURES PRICES Hans R. Stoll and Robert E. Whaley
The debate on the «speculative impact» Theoretical approaches Krugman s chinese wall between spot and paper markets Spot market The price is determined only by supply and demand Speculative storage could play a role No impact Paper markets Traders «bet» on future spot price outcome How can the bet influence the outcome? http://krugman.blogs.nytimes.com/2008/06/23/speculative-nonsenseonce-again/
The debate on the «speculative impact» Theoretical approaches Babusiaux, Pierru and Lasserre (2011) 2 Traders execute The C&C arbitrage Spot market The spot price follows the trend set by the paper market The spot price becomes higher than justified by fundamentals This anomaly takes time to be corrected because supply and demand are inelastic in the short term 4 At some point, new production drowns the market while consumption is destroyed Inventories increase The price starts dropping 1 Paper markets Imagine uninformed long investors arrive in mass, we need to balance out the new long investment positions with new short commercial positions (producers ) the forward price is driven higher, away from fundamentals 3 The increase in the spot price attracts new uninformed investors and the loop goes on 5 Investors withdraw en masse, triggering a price collapse
Empirical approaches CFTC data Irwin and Sanders, 2012
Empirical approaches to the problem The visual correlation between index flows and prices Important reference: Masters (2008)
Empirical approaches to the problem Correlation is not causation Other variables (fundamentals, liquidity, dollar ) Index flows Prices Approaches found in the literature to bypass the endogeneity problem: - Granger causality Irwin and Sanders, 2010 Büyüksahin and Harris, 2011 - Fundamental control variables Singleton (2011) - Exogenous index flows variables Hendersen, Pearson, Wang (2012)
Empirical approaches to the problem Temporal precedence (Granger causality) is not causation either Flows can precede prices without causing them Flows can cause prices without preceding them Index flows Index flows Week t Other variables Liquidity, dollar t and t +1 t and t+1 Prices Prices changes Week t A third variable could cause both flows and prices The causal link could be contemporaneous
Empirical approaches Is it possible to disentangle «fundamental factors» from «herding factors» in commodities prices? Supply/demand fundamentals Harvests, demand «Fundamental financial variables» (liquidity) «Non fundamental financial variables» (Bandwagon effect) Observed market prices Structural models to estimate the «bandwagon effect» or «overshooting» (undershooting) Frankel and Rose, 2010; Lombardi and Robays, 2011; Morana, 2012; Juvenaly and Petrella, 2012
Empirical approaches to the problem Bubble «tests» Sornette (2009) identifies a bubble on oil prices in 2008 Emketer et al. (2012) identify bubbles on grains prices Gilbert (2012) and Liu et al. (2012) reach opposite conclusions
Empirical approaches to the problem Where does the integration trend come from? Network of risky assets Risky assets correlation network in the period 2007-2010. Yellow lines correspond to correlations over 25% and below 50%, green lines to correlations over 50% and below 75% and red lines to correlations above 75%. The correlations are measured on daily prompt-month price returns. Period 2002-2006 Average correlation : 22% % of correlations above 50% : 9% Source: Riskelia
Empirical approaches to the problem Where does the integration trend come from? Network of risky assets Risky assets correlation network in the period 2007-2010. Yellow lines correspond to correlations over 25% and below 50%, green lines to correlations over 50% and below 75% and red lines to correlations above 75%. The correlations are measured on daily prompt-month price returns. Period 2007-2012 Average correlation: 43% % of correlations above % : 9% Source: Riskelia
Empirical approaches to the problem Integration and index investment Tang, K., Xiong, W., 2012. Index investment and financialization of commodities. Financial Analyst Journal 68, 54-74
Empirical approaches to the problem Integration and algorithmic/high frequency trading Bicchetti, D., Maystre, N., 2012. The synchronized and long-lasting structural change on commodity markets: evidence from high frequency data. United Nations Conference on Trade and Development UNCTAD White Paper
A direct proof of index investors impact on calendar spreads at the moment of the roll Mou Y., 2011. Limits to arbitrage and commodity index investment; frontrunning the Goldman roll. Columbia University Working Paper
Our approach Contemporaneous relations between weekly index flows and prices returns for the 12 agricultural commodities covered by the Supplemental Report Contemporaneous relations between index flows and hedgers flows (are index flows balanced by hedgers or hedge funds?) We alleviate the endogeneity problem
Our approach Contemporaneous relations between index flows and prices for the 12 agricultural commodities covered by the Supplemental Report Contemporaneous relations between index flows and hedgers flows (are index flows balanced by hedgers or hedge funds?) We alleviate the endogeneity problem Index investor s positions in individual agricultural markets can be broken down into three distinct components, ranked by decreasing level of exogeneity to individual agricultural markets: Index investors investment into generalist commodity indices (consisting of baskets of agriculture, energy and metal contracts) Index investors investment into general agricultural commodity indices (consisting of baskets of agricultural contracts only) Idiosyncratic index investors investment into single-commodity indices (may lead to overestimate the impact of index flows on prices if flows are trend-following or informed) Fourth component: periodic rebalancings to maintain the weights constant in the basket (may lead to underestimate the impact of index flows on prices)
Our approach Management of endogenity problems: Index flows are plausibly exogenous to agricultural markets because the main commodity indices have lower correlation to agricultural prices We project index flows on more generalist index flows (flows to the main generalist ETFs, aggregate flows to the 12 agricultural commodities) -> Two Stage Least Squares regression We introduce control variables in the price model (liquidity, dollar, perceived inventory level inferred from forward curves)
12 US-traded agricultural contracts Open Interests (in number of lots) Index investors and hedge funds in % of OI CIT HF Mean Std Mean Std Wheat (CBOT) 38% 4% -6% 6% Bean Oil (CBOT) 24% 4% 5% 10% Corn (CBOT) 22% 4% 7% 5% Soybean (CBOT) 24% 3% 7% 8% Feeder Cattle (CME) 23% 5% 10% 13% Lean Hogs (CME) 39% 5% 1% 9% Live Cattle (CME) 36% 5% 8% 8% Kansas Wheat (KCBT) 23% 5% 13% 11% Cocoa (ICE US) 14% 4% 8% 12% Coffee (ICE US) 25% 5% 5% 9% Cotton (ICE US) 28% 7% 6% 10% Sugar (ICE US) 22% 5% 7% 5% Average 26% 5% 6% 9%
Our liquidity variable Guilleminot, B., Ohana, S., 2012a., A new financial stress indicator: construction, properties and applications, Working Paper
Correlations between weekly flows/inventory changes/price returns RA Dollar Cycl. Com. Inv. Proxy Grains Inv. Proxy Grains prices Global CIT Grains Global HF Grains RA 100% Dollar 32% 100% Cycl. Com. Inv. Proxy 3% 22% 100% Grains Inv. Proxy 8% 8% 16% 100% Grains prices -30% -37% -28% -71% 100% Global CIT Grains -14% -26% -8% -2% 20% 100% Global HF Grains -21% -20% -12% -55% 65% 13% 100% Hedge funds are sensitive to fundamentals and prices Index flows are sensitive to liquidity but not to fundamentals (as could be expected)
Trends and HF positioning for wheat (top) and corn (bottom)
Three main generalist ETFs
Flows to the three main ETFs and index flows towards the 12 agricultural contracts The correlation between weekly index flows is 40%
Methodology 2SLS regression for commercial flows
HF trade in sync with CIT, so that hedgers have a double burden.meat markets are an exception
Methodology 2SLS regression for price returns
CITs have a significant impact for a number of commodities.meat markets are again an exception
The tendency of HF to imitate CIT amplifies the CIT s impact (in absolute value)
Methodology what happens when liquidity dries up
CIT s impact is indeed increased in periods of liquidity stress meat markets are again an exception (- β2)
Conclusion We contribute to the debate in several respects: We shed light on HF and CIT s motives in agricultural markets We alleviate the endogeneity problems in the estimation of the CIT s impact We relate the CIT s impact to the behavior of HF: the impact is increased for those commodities where HF imitate CITs We document an aggravated market impact in degraded liquidity periods, where HF align even more with CITs Meat markets stand out by their resilience to index speculation and liquidity stresses
Conclusion Implications and opened questions: The interaction of uninformed and trend-chasing investors may provoke artifical fluctuations Why don t hedge funds trade directionnally against index investors? (as they do e.g. in calendar spreads when index investors roll over their positions) Is it the same as «going against a bubble»? Would limits on CIT positions change the HF behavior? How to explain the resilience of meat markets to index speculation? What are the characteristics of hedge funds trading in meat markets?
Annex
Determinants of CIT flows
Determinants of HF flows
Relation to Past trends
First stage linear regression