Three Narratives on the Changing Face of Global Commodities Market Structure

Valiante, Diego Article Three Narratives on the Changing Face of Global Commodities Market Structure Credit and Capital Markets – Kredit und Kapital Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Valiante, Diego (2015) : Three Narratives on the Changing Face of Global Commodities Market Structure, Credit and Capital Markets – Kredit und Kapital, ISSN 2199-1235, Duncker & Humblot, Berlin, Vol. 48, Iss. 2, pp. 243-308, https://doi.org/10.3790/ccm.48.2.243 This Version is available at: https://hdl.handle.net/10419/293755 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Credit and Capital Markets 2 / 2015 Three Narratives on the Changing Face of Global Commodities Market Structure Diego Valiante1* Abstract The commodity market structure has changed at an incredible pace in the last 20 years and is now subject to intense scrutiny by academics and policy-makers. Taking a long-term view of price formation, empirical findings show that interna - tional trade and finance, mostly driven by emerging markets demand, market lib - eralisations and technological developments in market infrastructure, have in - creased pro-cyclicality and interconnection among physical commodity markets. Price formation mechanisms are more sensitive to information flows. The inter - connection with the financial system is strong and so the transmission of shocks from the financial system to commodity physical and futures markets. The rise of commodity-linked financial transactions was an important contribution to those developments. WTO commitments in international trade and expansionary mone - tary policies have promoted greater financial participation and so interconnection, which is also expressed by a greater pooling of commodity returns with returns of financial indexes (also defined here as the true ‘financialisation’ process). This pa - per represents an introduction to the functioning and structure of modern com - modity markets. Three narratives emerge as key drivers of the modern global mar - ket structure: international trade, international finance and trading technology. Wandel der Marktstruktur globaler Rohstoffmärkte – Drei Erklärungsansätze Zusammenfassung Im vergangenen Jahrzehnt ist das Interesse an Rohstoffmärkten rasant angestiegen. Ein langfristiger Überblick über die Preisbildung zeigt einen Anstieg der * Head of Financial Markets and Institutions, Centre for European Policy Studies; 1, Place du Congres, 1000 Bruxelles, Belgium. Phone: 003222293914, Email: [email protected]. The author is grateful to Federico Infelise for research support and to participants to seminars in Brussels (CEPS), London (Chatham House), Geneva (FIA Burgenstock Conference), Rome (at the 2014 EFMA Annual Conference), Munster (Munster University) and two anonymous referees. The author is solely responsible for any error. Credit and Capital Markets, 48. Jahrgang, Heft 2, Seiten 243–308 Abhandlungen OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 244 Diego Valiante Credit and Capital Markets 2 / 2015 Pro-Zyklizität und wechselseitiger Abhängigkeiten von physischen Rohstoffmärkten. Diese Entwicklungen sind zurückzuführen auf den internationalen Handel und Finanzsysteme, die vor allem durch die Nachfrage der Emerging Marktes und Marktliberalisierung getrieben werden, und den technologischen Fortschritt der Marktinfrastruktur. Der Preisbildungsmechanismus reagiert heute sensibler auf Informationsflüsse. Die gestiegenen, wechselseitigen Abhängigkeiten der Finanzsysteme erhöhen die Verbreitung von Schocks auf physische Märkte und Terminmärkte von Rohstoffen. Die Ziele der WTO für den internationalen Handel und die expansive Geldmarktpolitik der Zentralbanken führen zu einer höheren Beteiligung der Finanzmärkte, welche sich im Gleichlauf von Rohstoffrenditen und Renditen von Finanzindizes erkennen lässt (und als „financialisation“ bezeichnet wird). Dieser Beitrag liefert eine Einführung in die Struk turen und Funktionsweisen von modernen Rohstoffmärkten. Dabei sind drei bedeutsame Erklärungsansätze für die neue Marktstruktur zu nennen: der internationaler Handel, das internationalen Finanzsystem und die Geldmarkt politik, und zudem auch der technologische Fortschritt Keywords: Commodities market structure, International commodities finance, Financialisation, Price formation, Futures markets JEL Classification: Q02, F61, F62, E52 I. Setting the Scene A ‘commodity’ is a good with standard quality, verifiable ex ante, which can be traded on competitive and liquid global physical markets (Clark etal. (2001)). As Table 1 suggests, commodities are search goods for which information on quality can be easily assessed before the purchase, with no need to experience the product (as it would be the case for experience goods such as ‘durables’). This implies that demand for goods with similar supply and product characteristics will be intrinsically ‘less sticky’ to price changes (i. e. high price elasticity) for search goods (commodities) rather than experience goods. These characteristics allow parties to ‘shop around’ more easily, especially for commodities with more standard quality (e. g. corn). Low costs to acquire information about product characteristics and other structural factors make these goods suitable for trade. Each commodity has its own specific characteristics, such as product properties, availability in nature, transportability, production and storage processes, substitutability, concentration of producers / users, nature of the value chain, and so on. In addition, some commodities, such as agricultural commodities like wheat and corn, are renewable and therefore have seasonal price swings, mainly due to structural supply constraints. For instance, wheat can only be harvested once a year (from May for winter wheat to mid-August for spring wheat). Cocoa plants, in contrast, be- OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 245 Credit and Capital Markets 2 / 2015 come commercially productive roughly five years after plantation and their economic life can last up to 40 years. Supply characteristics may therefore affect demand elasticity when, for instance, availability of substitute products is limited, as in the case of crude oil. Product characteristics, such as the ability to store the product over a long period, are also key elements. Notably, alternative uses, such as the production of ethanol from corn crops, and excessive dependence in the production process from energy costs, as in the smelting of alumina, allow commodities prices to influence each other’s price formation processes (again, as in the case of crude oil). 1. A Complex Marketplace Price formation in markets for physical commodities and futures contracts is the result of complex interactions between idiosyncratic factors, such as product characteristics (quality, storability or substitutability, etc.) and supply and demand factors (capital intensity, industry concentration, production facilities, average personal income level or technological developments, etc.), and exogenous factors, such as access to finance, public subsidies and interventions, and the weather. The product characteristics of the commodity itself also affect how these sets of factors impact price formation. In general, supply factors (such as capital intensity) are more important drivers of price formation Table 1 Key Characteristics Types of goods Products Quality assessment Use Information costs Ex ante Ex post Search Commodities (e. g. crude oil or rice) Yes Yes Intermediate Low Final Experience Durable goods (e. g. car) No Yes Intermediate Medium Final Credence Financial services (e. g. loan or investment advice) No No Intermediate High Final OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 246 Diego Valiante Credit and Capital Markets 2 / 2015 for energy commodities and industrial metals, while agricultural and soft commodities markets are more influenced by demand factors (such as income growth) and exogenous factors that can cause supply shocks (such as weather events or government policies). Energy commodities and industrial metals rely on a more complex market organisation with easier access to finance due to their ability to hold value (for carry trades), which may enhance pro-cyclicality with regards to shocks within the financial system (opportunity costs). Table 2 Key Drivers of Commodities Price Formation Product Characteristics Supply Factors • Quality • Storability • Renewability • Recyclability • Substitutability • (Final) usability • Production convertibility and capital intensity • Horizontal and vertical integration • Storability and transportability • Industry concentration • Geographical concentration (emerging markets) • Technological developments • Supply peaks and future trends Demand Factors Exogenous Factors • Income growth and urbanisation • Technological developments and alternative uses • Long-term habits and demographics • Economic cycle • ‘Financialisation process’ and monetary policies • Subsidies programmes • General government interventions (e. g. export bans) • The economic cycle and other macroeconomic events • Technological developments • Unpredictable events (e. g. weather) Market Organisation • Micro-structural developments (e. g. competitive setting) • Functioning of internationally recognised benchmark futures or physical prices • International trade • Futures markets infrastructure OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 247 Credit and Capital Markets 2 / 2015 a) Physical and Futures Markets The standard quality of the good makes commodities easy to sell to end users, whether consumers or industrial companies. With technological advances and trade globalisation, in recent years, small regional markets have gradually become international or global market hubs, accessible directly through physical operations run by global freight companies and trading houses, or indirectly from any place in the world through the ‘pit’ (floor) or the electronic access to a venue running trading of physically deliverable (or offset) futures contracts globally. The creation of liquid and competitive international markets has reduced transaction costs and increased chances to meet individuals’ risk profiles. This section explores the general characteristics of commodities markets and their role in coping with commercial firms’ and individuals’ choice. There are two types of commodities markets: physical and futures (derivatives) markets. The physical market is a general market (hard to point to one specific place where the trade is done) that accommodates the need to balance supply / demand disequilibria. Futures markets serve the intertemporal choice of end users by trading expectations on supply and demand patterns, which occur mainly through changes of inventory levels over a diverse time period. Futures contracts are usually negotiated on open and transparent platforms. Particular characteristics, such as seasonal production or demand, require the use of tools that can ensure sufficient time to plan business development and investments in production processes. To accommodate demand and supply, these markets should be competitive and liquid (Clark etal. (2001)), which means that they will be able to provide a market clearing price at all times, and for all quantities, within a reasonable time frame. The availability of market clearing prices for all orders sent by the buyer / seller implies a dynamic equilibrium between demand and supply. A competitive market structure would potentially increase efficiency and market liquidity over time. It is important that barriers to entry to and exit from the market are always kept fairly low, and competition authorities are able to enforce competition rules and fight monopolistic market behaviours. Particularly in commodities markets, structural supply or demand constraints may favour conditions for the development of monopolistic, oligopolistic or monopsonistic powers and, thus, for one or more counterparties to charge unfair markups on final prices. Since commodities markets are central to the global OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 248 Diego Valiante Credit and Capital Markets 2 / 2015 economy, the efficiency of their market structure should be seen as a crucial area of coordination among national supervisory bodies. aa) The Fundamental Role of Inventories Inventories are the first real barrier against market prices fluctuations. Inventories minimise the costs of adjusting production due to foreseeable (e. g. demand volatility or increases in the marginal cost of production) and unforeseeable (e. g. weather shocks) market circumstances. Inventory levels keep demand and supply in equilibrium over time. In addition, they reduce marketing costs by facilitating production and delivery schedules (Pyndick (1994), (2001)). Inventories also reduce the impact of unpredictable disruptive events, working as a buffer against exogenous factors. As a consequence, the main drivers of inventory levels may vary depending on the type of commodity. For metal (and perhaps energy) commodities, inventory levels are primarily affected by the business cycle, mainly through Gross Domestic Product (GDP) levels (Fama / French (1988)). When a peak in demand comes, inventory levels go down drastically to absorb the adjustment of production, and vice versa. For seasonal commodities such as food and agricultural commodities, however, weather changes may have important effects on inventory levels by affecting the productivity of the harvest season. In both cases, changes in the inventory levels have immediate effects on spot and futures prices, which react differently to the high or low level of inventories (Fama / French (1988)). Inventories are the response function of net demand levels. Furthermore, inventories need to be properly managed because they have explicit and implicit costs of storage that will ultimately affect production costs. If released too quickly into the market, inventories can cause excessive supply and a drop in spot and futures prices. Management of inventories is a key risk management process for commodities firms. Carrying a commodity (storage) over time has three main costs: • Costs of physical storage (and insurance). • Opportunity costs. • Costs from price risk. Storage costs can be split into three subcategories: warehousing and handling costs (load in, load out, storage), insurance, and material degra- OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 249 Credit and Capital Markets 2 / 2015 dation. Costs of storage essentially depend on the availability of warehouses, competition for them (if not owned by the commodity owner), and the nature of the commodity, which may need specific storage characteristics to limit material degradation. The storability of the commodity may be fairly limited – green coffee beans can only be stored for few months before losing their original properties, for instance. Another important cost of storage is the opportunity cost of carrying a commodity over time, which includes the interest foregone by not investing the capital in risk-free instruments instead of in the commodity. The central bank’s nominal interest rate is usually considered as point of reference to calculate foregone interest. Current and future rates of consumption, as well as price volatility, are elements that contribute to the cost of carry, but they may not be easily predicted. Finally, there is a potential cost (or benefit) if prices move against the commodity holder, in particular if the future spot price will be below expectations. In effect, expectations about spot prices are part of the storage costs internalised through futures prices. This cost can usually be efficiently hedged in the derivatives markets. b) Interaction Between Futures and Physical Markets The price interaction between futures and physical 1 markets happens in two phases: during the duration of the futures contract, and at maturity. During the duration of the futures contract, information about inventory levels and exogenous factors fuel increasing or decreasing divergence of futures prices with spot prices. When the futures price is above the spot price, i. e. the basis (difference between spot and futures price) is negative, the market is in ‘contango’. When the futures contract price is below the spot price (i. e. the basis is positive), the market is in ‘backwardation’. At maturity, the price of the futures should converge to the spot price due to the ‘commitment to deliver’ mentioned above, which does not allow arbitrage to become systematic. As inventories fall, the spot price gradually catches up with the futures price and the curve inverts into backwardation until, for one of the three reasons mentioned above, the inventory levels recover and futures prices begin to regain ground to converge at maturity. 1 The words ‘physical’ and ‘spot’ are used interchangeably in this paper. ‘Spot price’ can be pure physical trade or rolling front month futures price. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 250 Diego Valiante Credit and Capital Markets 2 / 2015 Figure 1: Futures-Spot Price Interaction Through Inventories For storable commodities, as a consequence of the storage theory (i. e. the storage process, being a response function of supply and demand, drives futures and spot prices), when the futures curve is in contango a ‘cash and carry’ trade opportunity arises. More specifically, the commodity investor will have incentives to sell the forward contract and buy the commodity directly or through a loan, if the risk-free interest rate is sufficiently low. When the futures curve is in backwardation, though, the futures price is insufficient to cover cost of storage and interest foregone for alternative investments, so the commodities investor may enter in a ‘reverse cash and carry’ trade. He / she buys a future contract and sells the commodity immediately. aa) Price Convergence An important factor in the interaction among futures and spot markets is the convergence of futures prices to the spot price. This is mainly due to the ‘commitment to deliver’ embedded in the futures contract, which ensures that futures markets are always linked to underlying physical markets. Close to delivery (maturity), markets start to discount that, if the futures price diverges at delivery, there is an opportunity of arbitrage among markets and so the market will adjust its value to the spot market. For instance, if at the delivery date the futures price is lower than the spot price, the market will buy the futures contract until the two prices become equal (taking into account costs of delivery and differ- OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 257 Credit and Capital Markets 2 / 2015 Active global trade accounts are also reflected in consumption levels, with China becoming the top global consumer of iron ore, aluminium, copper, and soybean oil in 2011. It is among the top three global consumers for crude oil (2nd), wheat (2nd), corn (2nd), sugar (3rd), and natural gas (4th). No major levels of consumption emerge for cocoa and coffee, but the Chinese weight is constantly growing over time in these markets too. For agricultural commodities, such as wheat and corn, not much has changed in the last decade in terms of consumption levels, as the population is gradually stagnating and alternative use of biofuels production is still in early development. However, China has become the top global Table 3 Top Global Exporters and China (% of Total Exports) (Author’s elaboration from World Bank) 2001 2003 2011 European Union 40.1 % 42.0 % 35.1 % United States 13.1 % 10.9 % 9.6 % Japan 5.8 % 5.6 % 4.2 % (4th) China 3.9 % (5th)5.2 % (4th)9.5 % (3rd) Source: IMF (2011, p. 4). Figure 4: Chinese Net Imports (% of World Imports) –5 0 5 10 15 20 25 30 35 –5 0 5 10 15 20 25 30 35 Food Energy Raw Materials Metals Iron ore Soybeans 2000 2009 (Net imports, in percent of world imports) 65%53% OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 258 Diego Valiante Credit and Capital Markets 2 / 2015 commodities consumer. Over time, it is unquestionable that China will need to make more efficient use of current resources. If the country does not increase its greater independence from external provision of low-cost resources, the energy-intensive nature of its manufacturing economy and its ageing population will put additional unstable pressure on commodities prices. The more China grows in size, the more its weight on commodities markets may become unsustainable (at least in the short term) if competing global players do not reduce consumption levels. This situation might be seen as an incentive to finally increase efficiency in the use of global resources, but it will take years before relevant changes may see the light. b) Freight Markets: the Backbone of International Trade Seaborne freight markets are the backbone of international trade, but the structure of freight markets presents many challenges, which has contributed as well to higher price volatility in recent years. Inelastic demand and supply exposes the market to sudden price swings and prolonged periods of instability. Figure 6 describes supply and demand interaction. As demand for seaborne freight services grows, the curve grad- Source: Author’s calculation from IMF Database, BP, OPEC, ICSG, USDA and other governmental authorities. Figure 5: Chinese Consumption as % of Global Consumption 2001–2011 / 2012 6,3% 1,1% 1,8% 14,7% 6,7% 11,1% 4,1% 22,4% 28,9% 9,0% 0% 10% 20% 30% 40% Crude oil Natural gas Corn Soybean oil Sugar 2001 2011/2012 13,0% 50,0% 41,5% 0% 20% 40% 60% Iron ore Aluminium 2001 2011/2012 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 259 Credit and Capital Markets 2 / 2015 ually shifts to the right from point a to point b, i. e. more demand causes the equilibrium to move to a level with higher quantity to be supplied at a higher market-clearing price. The growth in demand for minerals and industrial metals for construction in emerging markets from 2001 to 2007 contributed to the gradual shift from point a to point c. Among the industrial metals, iron ore production went up 82.63 %, aluminium by 56.27 % and crude steel by 63.27 %. Total global production of iron ore, steel, aluminium and copper soared by 72.8 %, on average. Eight years of steady growth in demand gradually raised prices and volatility to unsustainable levels, once the capacity of the system had reached the critical point c. Freight rates for Brazilian iron ore, for instance, reached up to 200 % of the value of the underlying commodity in the autumn of 2007 (Figure 7), to fall below 20 % of the commodity price in under six months. As a consequence of this prolonged instability, investments from financial firms flowed into the industry to build sufficient capacity and keep up with growing volumes, shifting the supply curve (Figure 6) to the right (S2), i. e. the supply capacity experienced a sudden increase that pushed prices down over a short time frame. As a result of the growing supply of dry bulk cargoes (+33.62 %) and the drop in demand in 2008, following DWT D1 D2 D3 S1 S2 d c b a a P b P c P d P P Source: Adapted from Nomikos (2012). Figure 6: Supply and Demand Interaction OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 260 Diego Valiante Credit and Capital Markets 2 / 2015 0 20 40 60 80 100 120 140 160 180 200 0% 50% 100% 150% 200% 250% Jan 06 Apr 06 Jul 06 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Jan 08 Apr 08 Jul 08 Oct 08 Jan 09 Apr 09 Jul 09 Oct 09 Jan 10 Apr 10 Jul 10 Oct 10 Jan 11 Apr 11 Jul 11 Oct 11 Jan 12 C3 as %of BrazilIron Ore FOB Total dry freight (mn DWT) Aluminiun-Copper-Iron Ore-Steel production Sources: Author’s elaboration from ICAP, UNCTADstat, WBMS, World Steel Association (WSA), LKAB.4 Figure 7: Freight Rates and Total Production / Capacity (2006 = 100) the anaemic growth of global production due to the global financial crisis initially triggered by the burst of the housing market bubble in western economies, the cost of shipping tumbled by over 93 % between June and December 2008 alone (Figure 8). Prices dropped to the equilibrium point d and may stay there for some time. Since December 2008, prices have been subject to significant swings but have never returned to the levels reached in 2008. To hedge against these highly volatile trends and exogenous factors, such as port congestion or geopolitical events, market participants are increasingly using forward contracts on underlying shipping routes, which are linked to indexes such as the BDI. These contracts are cash-settled, and over-the- counter (OTC) traded and cleared. They tend to have a high basis risk, i. e. the difference between the price of the forward and the underlying exposure, as they track an index and not the specific characteristics of the exposure. Liquidity in this market is usually concentrated in one-month to two-month contracts (Geman (2005)). 4 C3 freight rate is a dry bulk rate to ship iron ore from Brazil to China. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 261 Credit and Capital Markets 2 / 2015 Sources: Author’s elaboration from ICAP and UNCTADstats.5 Figure 8: BDI Index and Dry Freight Capacity (mn Dead Weight Tonnes, DWT) c) Moving Competition on Production Costs and the Role of Subsidies Another key fall-out of more international trade is the continuous focus of competition on production costs. Competition on production costs from new regional areas has made subsidies programmes much more expensive, contributing to a more efficient price formation coupled with higher volatility as prices begin to reflect the true underlying supply and demand factors. In some areas, such as agricultural commodities, government subsidy programmes have supported artificial prices and reduced incentives to invest in new more efficient technologies to reduce energy consumption in metal production or harvested areas for crops, for example. When subsidies have gradually become less distortive, prices have begun to discount the lack of investments in infrastructure, which puts a big constraint on the ability of supply to meet demand with the potential creation of substantial regional imbalances. More generally, growing links between commodities markets and international trade have intensified the effects of government actions such as export bans. Most notably, direct market price intervention in an open 5 The Baltic Dry Index (BDI) represents a major dry freight cost index that collects rates on major global routes, widely used across the shipping industry. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 262 Diego Valiante Credit and Capital Markets 2 / 2015 market model with international trade is unable to create incentives to tackle underlying problems of market structure. When the fiscal capacity of a country is reduced, the market has to face sudden adjustments in the flows of commodities (e. g. oversupply) with highly volatile patterns, especially for agricultural commodities for which the opportunities costs of the land are generally higher in relation to other commodities markets. For instance, in agricultural and soft commodities markets, where the opportunity costs of the land use are high (e. g. US wheat farms) or too low (e. g. sugar plantations in Brazil), public investments in new technologies for innovative applications and infrastructures, respectively, might be a preferable alternative to subsidies. They might favour more efficient allocation of the land if the market itself is unable to rebalance due to such transaction costs. 2. A Story of International Finance Over the last decade, commodities markets have increasingly improved their access to international finance. Due to accommodating monetary policies and financial deregulation, the high returns generated by growing international trade fuelled by demand emanating from emerging industrial economies have attracted the interest of financial institutions hoarding cash for what has been commonly perceived as an anti-cyclical asset class. Financial leverage appeared therefore instrumental to the development of international trade. More interaction with the financial system also means easier access to financial leverage by commodities firms, and in particular by trading companies. More specifically, greater accessibility to finance was led by the following developments: • Deregulation; • New theoretical framework in investment portfolio theories; and • Expansionary monetary and fiscal policies. Regulatory changes throughout the 1990s in the United States culminated in 1999 with the US Gramm-Leach-Bliley Act (GLBA) or the Financial Services Modernization Act6, which repealed part of the Glass- Steagall Act (1933)7 and the separation between investment and com- 6 Pub.L. 106–102, 113 Stat. 1338, enacted November 12, 1999. 7 Within the Banking Act, Pub.L. 73–66, 48 Stat. 162, enacted June 16, 1933. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 263 Credit and Capital Markets 2 / 2015 mercial banking. The GLBA, in particular, allowed combinations of different financial activities (commercial, investment and insurance), through the use of subsidiaries, within the same group. Secondly, early evidence of a supposedly counter-cyclical nature of commodities markets and their role for diversification strategies (Gorton / Rouwenhorst (2004), among others) has attracted liquidity from non-commercial passive long investors, which have contributed to the liquidity of futures markets. Finally, next sections will explore the role of expansionary monetary and fiscal policies to push new investments into commodities markets. a) The Entry of New Market Players The last decade has seen the massive entry of new financial players and the expansion of financial intermediation. Low costs of financing and lower opportunity costs (returns on alternative asset classes) have favoured storage of commodities (carry trades), especially those with a good ‘store of value’ properties such as metals. These circumstances have increased the opportunities for financial participants to enter these markets and the opportunities for commodity trading houses to use financial leverage to expand their physical interests. Firstly, an exponential growth of financial intermediation occurred, with top financial institutions at the end of 2011 holding over $5 trillion in commodities derivatives (notional), with the whole exchange-traded derivatives markets estimated around $3.5 trillion (notional).8 The business of financial institutions has developed in different directions in the last decade. The range of financial institutions is very broad and includes: brokers / dealers, private banks, commercial banks, merchant banks, insurance companies, investment managers, mutual funds, hedge funds, and private equity funds. While the direct holding of physical assets is limited to some of them, several financial institutions are involved in financing and providing trading desk services for commodities firms. To develop these activities and make them more profitable, some of these institutions have invested significant resources in physical assets, such as supply and production firms, warehouses, and logistics / transportation companies. The growing importance of finance for funding large and medium commodities businesses has led to diversification in the business model of investment banks, which have increased their investments in 8 For more data on financial institutions derivatives exposures and the size of exchange-traded derivatives, see annex. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 264 Diego Valiante Credit and Capital Markets 2 / 2015 physical commodities trading. The growth of commodities firms and their global impact has led production and risk management functions to become more interconnected. This has become a profitable business for financial institutions, as commodities firms are not always able to handle all exposures through their own internal risk management systems. There are also myriad smaller banks that provide financing services to the commodities business, on top of other financing and investment services provided in other forms than derivatives transactions. Secondly, there is a handful of global commodity trading companies that combine the offer of intermediary services for other commodity firms (in physical and financial services) and logistics in multiple commodities (typically oil, some metals and a few agricultural commodities). These firms, also due to the easy access to international finance through their strong trading arms, have also increased their exposure to physical markets over the years through the ownership of firms dedicated to production, refining, and / or logistics. The nature of trading companies, which typically invest in the most profitable areas of commodities markets through sophisticated financial instruments and financial leverage, makes their offers more diversified across commodities markets, but also exposes them to fluctuations in futures markets and the financial system (due to their leveraged positions). Easier access to international finance and so to financial leverage, due to their nature of trading houses with strong financial expertise, has boosted revenues to levels close to those of big energy firms (see annex). Trading houses trade not only with their own proprietary capital, both in the physical and the financial marketplace, but also on behalf of other firms or as a direct counterparty of other commodity firms. Finally, as mentioned above, new developments in financial markets and investment portfolio theories during recent years have paved the way to a new form of investment that spans across different asset classes. The entry of passive long investors in commodities markets is still source of great controversy in the academic literature. The following section reviews the literature and evaluates some empirical analysis. aa) The Growth and Development of Commodities Index Investing and Other Financial Players Index investing is an easy way to become exposed to a commodity without owning any underlying asset or without a commitment to deliver OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 265 Credit and Capital Markets 2 / 2015 or buy any of them with daily margin calls (on futures markets). It can be considered one of the two main types of informed trading, with some particular characteristics (Masters (2008)). A clear distinction must be made with other non-commercial trading. First, even though often fully collateralised transactions by clients, indexes offer a position across a range of commodities without using expensive margin positions in futures markets or directly owning the commodity (with their storage risks and opportunity costs). Second, investors typically take a passive long position through these instruments on a basket of commodities.9 Third, investors tend to hold these positions for a long period. This last aspect, in particular, differentiates them from classical informed traders, actively exploiting single pieces of information. There is no interest in trading the commodity, but rather in taking a position in these markets. Index investments bring important benefits to markets by offering an easily marketable exposure to an asset class with lower transaction costs than those (direct and indirect costs) involved in investing directly in futures markets or in holding the physical commodity. New players can enter markets and bring additional liquidity, increasing futures market access globally for all commodities market participants, whether physical or financial entities with an interest in physical assets. Their typically long and stable position favours those commodity firms (especially producers) that take short positions to hedge main business exposures. It also dilutes the dominant weight of the large physical players in the futures markets by also allowing small players to enter the market and take exposure. The rise of index investing in futures markets has touched upon all asset classes and grown very rapidly in commodities, reaching over $200 billion of net value in March 2013 (over $366 billion, as sum of long and short positions), according to the U.S. Commodity Futures Trading Commission (CFTC) Index Investment data. The exchange-traded side of this business, in particular, has soared in recent years, reaching more than $200 billion of assets invested in 2012. There are also a number of products tracking indexes that are offered in the OTC space, which are captured in vast amounts by the CFTC statistics (above). Markets for commodities exchange-traded products have been growing rapidly since the onset of the financial crisis and they were reinvigorated in 2012, reaching 9 As new indexes combining both long and short positions emerge (3rd generation indexes), the situation may move towards a more balanced combination of long and short positions. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 266 Diego Valiante Credit and Capital Markets 2 / 2015 a historical peak since their initial diffusion back in the early 2000s. However, most of these activities are concentrated in precious metals (in particular, gold), which may explain the nature of this type of investing as a tool to diversify investment risk in complex portfolios. The range of exchange-traded products (ETPs) is much broader and non-commodities ETPs are the biggest part of the market. Disregarding ETPs assets with exposures on precious metals, the size of ETPs in the commodities treated in this report goes down to roughly $38 billion (Figure 9). Since the fund may be unable (for costs and type of risks) to take a direct position in different futures or physical markets to replicate the return of the index (with minimal errors; so called “physical replication”), the funds can also signs an OTC swap agreement with an investment bank that ensures the perfect replication of the index in exchange of a constant flow of liquidity from investors (through the fund) to the bank (physical replication). The bank will then take exposure in the futures markets using most of the financial flows (and collateral) coming from the fund, and by rolling over their futures positions held to ensure that the index is tracked with precision over time. Figure 10 above shows the process through which investments in indexes are channelled through OTC and ETP products into futures markets, through the OTC swaps that funds sign with financial institutions. Non-commodities ETPs 1,638,700 19.718 6.746 8.803 2.627 Gold 143,442 19.133 5.256 Commodities ETPs 205,725 Industrial Materials Energy Agriculture Broad Market Other Precious Metals Silver Source: Blackrock ETP Landscape. Figure 9: Breakdown of Commodities ETPs per Underlying Exposure, Q3 2012 (US$ Million) OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 273 Credit and Capital Markets 2 / 2015 75 80 85 90 95 100 105 110 115 0 1 2 3 4 5 6 7 1994-01 1994-09 1995-05 1996-01 1996-09 1997-05 1998-01 1998-09 1999-05 2000-01 2000-09 2001-05 2002-01 2002-09 2003-05 2004-01 2004-09 2005-05 2006-01 2006-09 2007-05 2008-01 2008-09 2009-05 2010-01 2010-09 2011-05 2012-01 2012-09 0 1 2 3 4 5 6 7 $3.000 $5.000 $7.000 $9.000 $11.000 $13.000 $15.000 $17.000 1994-01 1994-10 1995-07 1996-04 1997-01 1997-10 1998-07 1999-04 2000-01 2000-10 2001-07 2002-04 2003-01 2003-10 2004-07 2005-04 2006-01 2006-10 2007-07 2008-04 2009-01 2009-10 2010-07 2011-04 2012-01 2012-10 M2 SA Interbank rate (%; rhs) US Government Debt Source: Federal Reserve and US Treasury. Figure 13: Broad Dollar Index (Inflation Adjusted)12 Devaluation and Policies, 1994–2012 Figure 13 shows how the dollar exchange rate has gradually devalued since 2002, as a result of bold cuts to nominal interest rates set by the central bank (and its effects on interbank rates) that started a prolonged 12 The Broad Dollar Index is a weighted average of the foreign exchange values of the U.S. dollar against the currencies of a large group of major U.S. trading partners including 26 countries. The index weights, which change over time, are derived from U.S. export shares and from U.S. and foreign import shares. For more details, please see http: / / www.federalreserve.gov / pubs / bulletin / 2005 / winter05_ index.pdf. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 274 Diego Valiante Credit and Capital Markets 2 / 2015 period of expansionary monetary policies in early 2000s, before attempting to correct it some years later with no success. Expansionary monetary and fiscal policies, supported by global capital imbalances, thus were a key driver for the devaluation of the dollar, which began in 2002 and has recently reached a historical low since early 1990s (Figure 13). The following section will assess what is the role of monetary policies in the growth of non-commercial positions and how non-commercial positions impact commercial ones. Notwithstanding the complex nature and implications of monetary policies, there appears to be a distinct pattern in which expansionary monetary policies may have played an important role for the growth of non-commercial (and commercial) positions, in particular via the quantity of money (M2) 13 that was injected in the system. Due to misreporting in CFTC data, only a specific sample of non-com- mercial and commercial positions for a selected contract (crude oil, WTI) can be used for a more long-term analysis (with some strong caveats). Index positions, instead, are only available from 2006, which may not offer a sufficiently long-term analysis. Among other important factors that can influence commodities prices, over the long term, the impact of monetary policies has often been unpredictable (Cooper / Lawrence (1975)), which calls for a deeper investigation into their effects across asset classes, especially for commodities markets. aa) VEC Analysis: Monetary Policies and Commercial Positions In order to investigate in more depth the relationship between non-com- mercial positions and M2, for which a simple linear combination does not fit, and a more sophisticated empirical analysis is required. The following dataset (for crude oil US futures contract on NYMEX)14 includes monthly data from January 1986 to December 2011: • Total (or only short) commercial positions (log of open interest, ‘Ln- Comm’). • Total (or only long) non-commercial positions (log of open interest, ‘LnNonComm’). 13 M2 consists of M1 (essentially, currency and similar in circulation, demand and other checkable deposits), plus savings deposits, time deposits, and money market funds, less individual retirement accounts. 14 The only contract for which CFTC data on commercial and non-commercial futures positions gives a long-term series with very limited misreporting. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 275 Credit and Capital Markets 2 / 2015 • Log of S&P 500 index, VIX index (implied volatility of S&P 500, ‘SP500’). • Log of M2 (monetary aggregate, ‘LnM2’) and the Fed interbank interest rate (here called, ‘Fed funds’ or ‘LnFedFund’). The dataset of futures positions for crude oil (commercial short and non-commercial long), despite changes to reporting criteria over the years, is the only CFTC legacy report that shows no significant jumps in the series since the beginning of data collection from CFTC in 1986, which may allow an assessment of long-term effects of monetary policies before and after the beginning of the expansionary era. As this dataset may underestimate the impact of swap dealers on non-commercial long positions, an additional empirical analysis with more granular data (available since 2006) is also run in the following section to confirm results. Moreover, this analysis uses monthly data, which do not permit the assessment of more short-term patterns. The results of this analysis, therefore, should be interpreted as an early assessment that is primarily valid over a sufficiently long time period. Table 4 Summary Statistics  LnNonComm LONG LnNonComm TOT LnComm SHORT LnComm TOT LnFedFund LnM2 LnSP500 Mean 10.682 11.394 12.612 13.296 1.033 8.448 6.579 Standard Error 0.067 0.057 0.040 0.038 0.070 0.022 0.034 Median 10.447 11.109 12.631 13.336 1.586 8.383 6.822 Standard Deviation 1.182 1.016 0.699 0.674 1.244 0.382 0.601 Sample Variance 1.398 1.031 0.488 0.454 1.547 0.146 0.361 Kurtosis –1.049 –1.046 –0.305 –0.118 1.644 –1.248 –1.286 Skewness 0.178 0.349 –0.430 –0.510 –1.650 0.284 –0.480 Range 5.379 4.151 3.277 3.163 4.947 1.344 1.990 Minimum 7.533 9.070 10.556 11.301 –2.659 7.828 5.356 Maximum 12.911 13.221 13.833 14.464 2.287 9.172 7.346 Count 312 312 312 312 312 312 306 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 276 Diego Valiante Credit and Capital Markets 2 / 2015 Variables are stationary only in first difference (integrated of first order) and cointegrated (with stationary residuals), so linear regressions may be spurious and some Granger causality tests may give misleading results. Engel and Granger (1987) showed that the use of a simple linear regression with unit-root variables (even if de-trended) can generate numerous cases of spurious regression so, provided that a cointegration relation actually exists among the variables, the estimation of this relation is indeed quite powerful in avoiding misleading conclusions. The Vector Error Correction (VEC) model might be the best model to deal with variables subject to the same stochastic trend. VEC is an extension of a Vector Autoregressive Model (VAR) for variables that are non-stationary in levels, but stationary in their first difference (first-order integration, I(1)).15 This model is particularly useful as it can take into account any relation of cointegration among two variables, i. e. they share the same stochastic trend.16 The first step checks cointegration among the variables. We run a regression of X (independent variable) on Y (dependent variable) in levels. We estimate the residuals of this regression (first step) and we test for the stationariety of the residuals through the Augmented Dickey-Fuller test (second step). If the residuals are stationary then the two variables are cointegrated17. First, a linear regression between commercial positions and M2 appears spurious, as hinted at by very high t-statistics and R-squared, as well as a very low Durbin-Watson d-statistics. Second, a test for the existence of a relationship of cointegration is performed. The Dickey-Fuller test for unit root rejects the hypothesis (of unit root), so residuals of the cointegration equation (M2 regressed on commercial positions) are stationary and thus the two variables are cointegrated. The two variables move with the same stochastic trend and adjust through a process of error correction that is described in the Annex. 15 Testing hypotheses concerning the relationship between non-stationary variables is based on OLS regressions with data that had initially been differenced (Granger / Newbald (1974)). Although this method is correct in large samples, taking into account cointegration provides more a powerful analysis tool, as it doesn’t lose information on long run equilibrium and on levels. 16 While a deterministic trend is treatable by either regressing the variable on time (trend stationary) or eliminating the seasonality, to treat a stochastic trend and make the series stationary it is possible to just differentiate the variables. 17 In this way we show that that there exists a linear combination of the 2 variables which is stationary, as the residuals u are nothing but u = y – bx. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 277 Credit and Capital Markets 2 / 2015 The Granger Theorem states that if Y and X are cointegrated, the relationship can be written as below and at least one between γ1 γ2 must be ≠ 0. (eq.1) ΔYt = a1 ΔYt–1 + b0 ΔXt + b1 ΔXt–1 + γ1(Yt–1 – Xt–1) (eq.2) ΔXt = a1 ΔXt–1 + b0 ΔYt + b1 ΔYt–1 + γ2(Yt–1 – Xt–1) γ1 and γ2 are the coefficient of the cointegrating equation. At least one of them must be statistically different from zero and with negative coefficient, as it shows how a variable, when the distance between the two variables grows, is brought back to the equilibrium and the model is then stable. Those coefficients should then be between 0 and –1. It is the speed of adjustment of the dependent variable to the equilibrium. For instance, if it is equal to 0.5 it means a 50 % movement back to equilibrium follow- Table 5 Linear Regression – Commercial Positions and M2 _cons -.4486328 .3264045 -1.37 0.170 -1.090881 .1936157 lnm2 1.62709 .0385993 42.15 0.000 1.55114 1.70304 commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 141.077867 311 .453626581 Root MSE = .26 Adj R-squared = 0.8510 Residual 20.9564659 310 .067601503 R-squared = 0.8515 Model 120.121401 1 120.121401 Prob > F = 0.0000 F( 1, 310) = 1776.90 Source SS df MS Number of obs = 312 Durbin-Watson d-statistic( 2, 312)= .1009832 _cons .0024848 .004561 0.54 0.586 -.00649 .0114596 LD. -.1069497 .055503 -1.93 0.055 -.2161641 .0022646 L1. -.0707039 .0180856 -3.91 0.000 -.1062914 -.0351165 coin1 D.coin1 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0020 Z(t) -3.909 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 310 Table 6 Augmented Dickey-Fuller Test _cons -.4486328 .3264045 -1.37 0.170 -1.090881 .1936157 lnm2 1.62709 .0385993 42.15 0.000 1.55114 1.70304 commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 141.077867 311 .453626581 Root MSE = .26 Adj R-squared = 0.8510 Residual 20.9564659 310 .067601503 R-squared = 0.8515 Model 120.121401 1 120.121401 Prob > F = 0.0000 F( 1, 310) = 1776.90 Source SS df MS Number of obs = 312 Durbin-Watson d-statistic( 2, 312)= .1009832 _cons .0024848 .004561 0.54 0.586 -.00649 .0114596 LD. -.1069497 .055503 -1.93 0.055 -.2161641 .0022646 L1. -.0707039 .0180856 -3.91 0.000 -.1062914 -.0351165 coin1 D.coin1 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0020 Z(t) -3.909 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 310 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 278 Diego Valiante Credit and Capital Markets 2 / 2015 ing a shock to the model one period later. If it is equal to 1 then there is full adjustment to the equilibrium the period after. A coefficient higher than 1 would not make much sense. The VEC analysis (described in Table 7) for the relation between the number of commercial positions in the crude oil futures market and M2 shows that the cointegration equations for both variables are statistically significant. Most notably, commercial positions react much faster to equilibrium shocks (8 % rate) compared to M2, whose coefficient is negligible. This result may indicate that commercial positions are affected by monetary policy actions much more than the other way around. The coefficient b1, which weights the impact of the cointegrated (lagged) variable on the dependent, is non-significant for M2, i. e. the lagged value of commercial position has no link with M2. The same is not true for commercial positions, as the lagged value of M2 is statistically significant. Table 7 VEC Analysis Outputs _cons .0205437 .006461 3.18 0.002 .0078301 .0332574 L1. -.0812382 .0178908 -4.54 0.000 -.1164428 -.0460336 coin1 D1. -2.367319 1.075996 -2.20 0.029 -4.484607 -.2500303 lnm2 LD. -.104876 .0548934 -1.91 0.057 -.2128924 .0031404 commTOT D.commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 2.06783957 309 .006692037 Root MSE = .0786 Adj R-squared = 0.0767 Residual 1.89063841 306 .006178557 R-squared = 0.0857 Model .177201164 3 .059067055 Prob > F = 0.0000 F( 3, 306) = 9.56 Source SS df MS Number of obs = 310 _cons .0039853 .0003435 11.60 0.000 .0033093 .0046612 L1. -.0026633 .0009751 -2.73 0.007 -.004582 -.0007446 coin1 D1. -.006499 .0029625 -2.19 0.029 -.0123284 -.0006696 commTOT LD. .0947919 .0571094 1.66 0.098 -.017585 .2071687 lnm2 D.lnm2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .005468579 309 .000017698 Root MSE = .00413 Adj R-squared = 0.0378 Residual .005211015 306 .000017029 R-squared = 0.0471 Model .000257564 3 .000085855 Prob > F = 0.0020 F( 3, 306) = 5.04 Source SS df MS Number of obs = 310 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 279 Credit and Capital Markets 2 / 2015 With this modified Granger, the conclusion is that M2 Granger-causes commercial positions and not vice-versa. We apply the same approach to non-commercial positions and M2. As shown by Output #3 (annex), non-commercial positions adjust to equilibrium with M2 at an 18 % rate. It therefore appears that are the non-com- mercial positions ‘to follow’ changes in M2. This is confirmed by the cointegrating coefficient of M2, which is not significant, hinting at the indifference of M2 towards the distance from equilibrium with non-commercial positions. Finally, the same approach is used to assess the relationship between non-commercial long positions, which represent passive speculative investments that would supposedly divert futures markets from their fundamentals, and commercial short positions (a classic commodities hedge for final users). The initial test (Output #4) confirms that the regression is spurious and residuals are stationary, so variables can be considered cointegrated. The VEC analysis (Output #5) gives some interesting results. The cointegrating equation of a non-commercial long position has a statistically significant (at 1 %) negative coefficient, which suggests that these positions react at deviations from equilibrium with commercial short positions. The opposite is not true. The cointegrating coefficient is significant at 5 %, but with a very low positive coefficient. This points to an unstable equilibrium, so we could potentially ignore it. As a result, commercial short positions Granger-cause non-commercial long. The growth of commercial players and the general interests in physical commodities markets in the last decade, with the quick and intense development of international trade, have proved fertile ground to promote the growth of non-commercial positions as a tool to provide liquidity, which could be accessed at very low costs due to accommodating monetary policies. This finding is in line with ample evidence showing, despite the potential to be harmful for price formation through herding behaviours, limited distortive effects of financial positions on commodities price formation. bb) Taking Stock from the New CFTC Disaggregated Reporting While the previous long-term price formation analysis with the legacy reports should be still valid over a long-term database (from 1986), the growth of passive investments together with other (typically long) swap dealers positions in recent years requires further analysis with the new OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 280 Diego Valiante Credit and Capital Markets 2 / 2015 CFTC reporting system that was launched in 2009 and goes back to 2006. The new reporting, therefore, disaggregates data on futures open positions in three main categories of traders (producers, swap dealers and managed money). The analysis uses the new CFTC dataset, which includes weekly data on open positions for the three most liquid futures contracts in the US (crude oil, natural gas, and corn). The analysis in the previous section is replicated by running Granger causality tests. The Dickey-Fuller test suggests that variables are not co-integrated and Granger causality tests shall not thus lead to misleading results. Different lags for each futures contract have been considered, in line with lag-order selection statistics. Table 8 confirms the results of the previous analysis but it qualifies it further. It confirms that M2 leads producers positions, which points at the potential impact of prolonged expansionary monetary policies on non-financial assets (through expansion of monetary base). However, from 2006, data for crude oil confirms an impact of the monetary base on the size of financial players’ positions in futures markets, while the impact of the monetary base only affects producers / users’ positions for natural gas and corn futures positions. Due to their constant growth in crude oil futures markets, non-commercial positions have become the main mean to transfer effects of policies and events that affect the monetary base. Most notably, the analysis on the disaggregated futures positions confirms the results of the earlier vector error correction model by ascertaining the role of producers / users position in guiding swap dealers and managed money’s long positions (and not vice versa) for the top three Table 8 Granger Causality Tests Variables Granger causality Reversed Independent → Dependent Crude oil Natural gas Corn Crude oil Natural gas Corn M2→SD / MM long Yes* No No No No Yes*** M2→Producers short No Yes* Yes* No Yes* No Producers short → SD / MM long Yes** Yes** Yes** No No No Note: *1 %, **5 %, ***10 % significance. ‘SD / MM’ stands for ‘Swap dealers / Managed money’. See also outputs in Annex. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 281 Credit and Capital Markets 2 / 2015 futures contracts (by size of open interest). Financial futures positions still complement non-financial ones and are shaped by the latter. Therefore, the nature and the role of non-commercial players’ participation in commodities markets appears benign and essential for the development of commercial positions, and thus attention should rather focus on shortterm market practices led by non-commercial players that could potentially lead to damaging herding behaviour (Boyd et al. (2013)). Shortterm price trends and market practices shall be subject to more detailed analysis, which would require more detailed information about traders’ behaviours (e. g., data on volumes by category of trader). 3. A Story of Fast-Growing Market Infrastructure Market infrastructure plays a crucial role in the development of commodities market structure and its well functioning on a global scale. Futures markets, in particular, are an essential infrastructure supporting risk management, and ultimately price formation in physical markets. Futures markets have supported the development of international trade and the consolidation of commercial participants fuelled by the opening up of international trade. Transparent and stable futures markets promote healthy interaction between the physical and financial spheres of commodities markets, which today are inextricably linked. As a result of greater interconnectedness, market infrastructure also allows faster circulation of information by increasing accessibility and so the resilience of price formation mechanisms. The size of commodities futures exchanges has more than tripled since 2004, particularly as a result of the financial crisis, which has reduced dealers’ capital commitment in OTC derivatives transactions (see table in annex) and increased the role of transparent venues as a cheaper source of liquidity for commodities users. The size of global commodities futures exchanges reached its peak in 2012, with almost 3 billion traded contracts and seven global market infrastructures of which no one is European and four of them are today Chinese companies (see Figure 14). The development of market infrastructure in recent years has been astonishing and driven by the following events: • Demutualisation; • Technological advances; and • Regulatory reforms. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 282 Diego Valiante Credit and Capital Markets 2 / 2015 0 500.000.000 1.000.000.000 1.500.000.000 2.000.000.000 2.500.000.000 3.000.000.000 3.500.000.000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Chicago Mercantile Exchange Group Dalian Commodity Exchange Shanghai Futures Exchange ICE Futures Europe and US Zhengzhou Commodity Exchange Multi Commodity Exchange of India London Metal Exchange Others Note: 2012 data for Multi Commodity Exchange of India is from 2011.18 Source: Author’s calculations from World Federation of Exchanges (WFE) Statistics and the European Capital Markets (ECMI) Institute Statistical package (2012). Figure 14: Growth of Commodity Futures Exchanges Volumes by Number of Contracts, 2002–2012 Around the early 2000s, as technological changes showed that trading venues are not natural monopolies and can stand market competition, a process of demutualisation of otherwise no-profit entities began. Demutualisation triggered a more competitive environment with for-profit entities investing to increase market share and profitability, mainly through new services to boost volumes and consolidation with other incumbent infrastructures. In commodities markets, US and Chinese exchanges are leading participants in futures market infrastructure. As shown in Figure 15, the Chicago Mercantile Exchange (CME) group is the biggest global exchange by value of open interest and number of traded contracts, but the growth of Chinese exchanges has been astonishing, and today they have a global market share of almost 50 %, as China has de facto become the major commodities consumer in the world (Figure 15). Some Chinese exchanges have become points of reference in Asia but, also due to gov- 18 ‘Others’ include: MICEX / RTS, NYSE Euronext (Europe), Bursa Malaysia Derivatives, ICE Futures Canada, Thailand Futures Exchange, Johannesburg SE, BM&FBOVESPA, ASX SFE Derivatives Trading, Korea Exchange, Buenos Aires SE, NYSE Euronext (US), Rofex, ASX Derivatives Trading, BSE India, Bursa Malaysia, Japan Exchange Group – Osaka, Tokyo Commodity Exchange (TOCOM), Tokyo Grain Exchange. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 289 Credit and Capital Markets 2 / 2015 Gruber, Joseph W. / Vigfusson, Robert J. (2013): Interest Rates and the Volatility and Correlation of Commodity Prices, Board of Governors Federal Reserve System, International Finance Discussion Papers, N. 1065, January. Irwin, Scott H. / Sanders, Dwight R. (2010): The impact of index and swap funds on commodity futures markets, OECD Food, Agriculture and Fisheries Working Papers 27, 52. Mayer, J. (2009): The Growing Interdependence Between Financial and Commodity Market, UNCTAD discussion paper n. 195. Nomikos, N. (2012): Managing Shipping Risk in the Global Supply Chain. The Case for Freight Option, Cass Business School Presentation (http: / / www.bbk. ac.uk / cfc / papers / nomikos.pdf). Panagiotidis, T. / Rutledge, E. (2007): Oil and gas markets in the UK: Evidence from a cointegrating approach, Energy Economics, 29, Issue 2, pp. 329–347. Pindyck, R. (1994): Inventories and the Short-Run Dynamics of Commodity Prices, The American Journal of Economics, Spring 1994, 25, pp. 141–59. – (2001): The dynamics of commodity and futures markets: a primer, Energy Journal 22, pp. 1–29. Ramaswamy, S. (2011): Market structures and systemic risks of exchange-traded funds, BIS Working Paper, n. 343, April. Rumbaugh, T. / Blancher, N. (2004): China: international trade and WTO accession, International Monetary Fund. Silvennoinen, A. / Thorp, S. (2010): Financialization, Crisis and Commodity Price Dynamics, Journal of International Financial Markets, Institutions and Money, 24(1), pp. 42–65. Tang, K. / Xiong, W. (2009): Index Investing and the Financialization of Commodities, Working paper, Princeton University. Valiante, D. (2013): Commodities Price Formation: Financialisation and Beyond, Centre for European Policy Studies Paperback, Brussels. Vansteenkiste, I. (2011): What is driving oil futures prices? Fundamentals versus speculation, ECB Working Paper Series, N. 1371, August. US-China Business Council (2010): Testimony of John Frisbie, President, Trade Policy Staff Committee Hearing (https: / / www.uschina.org / public / documents / 2010 / 10 / wto_commitments_testimony.pdf). World Trade Organisation (2001): WTO successfully concludes negotiations on China’s entry, Press Release, 17 September (http: / / www.wto.org / english / news_e / pres01_e / pr243_e.htm). OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 290 Diego Valiante Credit and Capital Markets 2 / 2015 Annexes Tables Growth of Exports Value ($bn) and Size, 2001–11 Value ($bn) Size 2001 2011 CAGR 2001 2011 Units Crude oil 340.1 1,475 16% 38,262.1 38,854 kbbl / day Natural Gas 82.4 368.5 16% 553.46 1073.32 bcum Iron ore 14.8 180 28% 493.1 1,072.9 mn / tonnes Wheat 19.1 47.6 10% 105.92 150.4 mn / tonnes Aluminium* 16 38.1 9% 11.1 15.87 mn / tonnes Corn 6.7 34.1 18% 74.67 117.03 mn / tonnes Coffee 5.4 28.6 18% 5.45 6.81 mn / tonnes Sugar 4 17.8 16% 21.11 31.12 mn / tonnes Soybean oil 2.9 11.1 14% 8.25 8.52 mn / tonnes Cocoa 2.6 8.8 13% 2.47 2.96 mn / tones Copper na Na na na na Na Source: Author’s calculation from World Bank, USDA, ABREE, BP, OPEC, FAO. Note: *Data on exports are estimates. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 291 Credit and Capital Markets 2 / 2015 China’s Ranking in Key Commodities Markets, 2001–2011 / 2012 Production (top 10; % tot) Consumption (top 10; % tot) Exports (top 10; % tot) Imports (top 10; % tot) 2001 2011 / 2012 2001 2011 / 2012 2001 2011 / 2012 2001 2011 / 2012 Crude oil 7th (4.4%) 5th (4.9%) 3rd (6.3%) 2nd (11.1%) no no n / a 2nd (14.9%) Natural Gas n / a (1.2%) 6th (3.1%) n / a (1.1%) 4th (4.1%) no no n / a 10th (1.2%) Iron ore n / a 2nd (22.9%) n / a (13%) 1st (50%) no no n / a 1st (60.2%) Aluminium 2nd (13.5%) 1st (41.8%) n / a 1st (41.5%) no no 5th * 10th Copper n / a 1st (26.4%) n / a 1st no no n / a 1st Wheata2nd (16%) 2nd (7.7%) 2nd (18.5%) 2nd (17.9%) no no no no Corna2nd (19%) 2nd (15%) 2nd (19.8%) 2nd (22.4%) no no no no Soybean oila 4th (12.4%) 1st (26.2%) 2nd (14.7%) 1st (28.9%) 3rd 1st no no Sugara5th (5.2%) 4th (7.2%) 5th (6.7%) 3rd (9%) no no 7th 4th Cacao no no no no no no 9th 8th Coffee no no no no no no no no *In 2003. a2012 estimate. Source: Author’s calculation from IMF Database, BP, OPEC, ICSG, USDA and other governmental authorities. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 292 Diego Valiante Credit and Capital Markets 2 / 2015 Top 12 Most Active Financial Institutions in Commodities Derivatives, by Notional / Total Assets €bn – End 2011 Notional value19 Gross value (fair value)* Total assets Revenues % Notional / Total assets % Gross / Total assets Ratio Gross / Revenues Morgan Stanley 607.07 61.60 579.00 25.02 104.85% 10.64% 246 Goldman Sachs 614.91 57.51 712.82 22.25 86.26% 8.07% 2.59 JP Morgan 859.35 90.62 1,749.42 75.07 49.12% 5.18% 1.21 Barclays 857.09 26.89 1,876.86 38.76 45.67% 1.43% 0.69 Bank of America 639.22 29.65 1,643.84 72.91 38.89% 1.80% 0.41 Credit Suisse 281.62 n / a 862.41 21.56 32.65% n / a n / a Société Générale 343.09 17.06 1,181.37 25.64 29.04% 1.44% 0.67 Deutsche Bank** 459.13 44.36 2,164.10 33.23 21.22% 2.05% 1.34 Citigroup 221.11 21.92 1,446.82 60.50 15.28% 1.52% 0.36 BNP Paribas** 156.29 13.75 1,965.28 42.38 7.95% 0.70% 0.32 Credit Agricole 69.79 8.50 1,860.00 35.13 3.75% 0.46% 0.24 HSBC 59.06 2.85 1,973.16 46.44 2.99% 0.14% 0.06 Tot. 5,167.72374.71 18,015.09 498.88 49.71%^ 3.9%^ 1.15^ Global OTC 2,5720 405 – – – – – Global ETD*** 3,585 – – – – – – Source: 2011 Annual reports, SEC K10 files, BIS (2013 update), WFE / IOMA. *Before netting adjustments. ^Weighted average (notional). “Estimates. ***Conservative estimate of value of traded futures and options contracts.21 19 Balance sheets do not provide further granularity on how this notional value can be decomposed, i. e. what kind of commodities derivatives trades (OTC or it includes estimation of exchange-traded derivatives positions in commodities). It includes precious metals. For exchange-traded futures contracts, notional value in this analysis means value of open interest. 20 Including OTC derivatives on gold and other precious metals, at the end of 2012. 21 These statistics do not include the turnover value of commodities futures and options of the London Metal Exchange, NYSE Euronext (US), Australian Securities Exchange SFE Derivatives Trading, Multi Commodity Exchange of India, Singapore Exchange, plus an undefined list of small commodities exchanges. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 293 Credit and Capital Markets 2 / 2015 Notional Value of Outstanding Commodities Futures and Options Traded OTC and on Exchange ($bn)19 Exchange-traded Over-the-counter Total 2011 2012 2011 2012 2011 2012 Futures22 3,226 (65%) 3,168 (70%) 1,745 (35%) 1,363 (30%) 4,971 4,531 Futures and options 3,585 (58%) 3,485 (62%) 2,570 (42%) 2,101 (38%) 6,155 5,584 Note: Exchange-traded data are conservative estimates derived from turnover value of futures and options contracts.2320 Value of over-the-counter positions is not daily marked-to-market. Source: Author’s estimates from WFE / IOMA, BIS, CME, LIFFE, LME, ICE, other sources. 22 Forwards and swaps for OTC transactions. 23 The statistics published by the World Federation of Exchanges and the International Options Market Association do not include the turnover value of commodities futures (forwards) and options traded on the London Metal Exchange, NYSE Euronext (US), Australian Securities Exchange SFE Derivatives Trading, Multi Commodity Exchange of India, Singapore Exchange, plus an undefined list of very small commodities exchanges. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 294 Diego Valiante Credit and Capital Markets 2 / 2015 Key Trading Companies by Total Revenues, 2003 vs. 2011 ($bn) Ownership Country Total assets Total revenues 2003 2011 2003 2011 2003–11 CAGR 1 Vitol Private Netherlands na na 61* 297.00 22%* 2 Glencore Public Switzerland 59.90** 86.16 142.34** 186.15 - 3 Trafigura Private Netherlands na na na 121.50 - 4 Noble group Public Hong Kong 1.07 17.34 4.28 80.73 44% 5 Gunvor International Private Cyprus na na na 80.00 - 6 Mercuria Private Switzerland na na na 75.00 - 7 Marubeni*** Public Japan 41 65 75.2 55.63 - 8 Xstrata Public Switzerland-UK 10.00 74.83 3.47 33.88 33% 9 Marquard & Bahls AG Private Germany 0.78 5.63 5.44 25.84 22% 10 System Capital Private Ukraine na 28.45 na 19.55 - Note: *2004 data; **2007 data; *Fiscal year ended in March 2012. Exchange rate with USD is yearly average. Source: Author’s selection from websites, annual reports and OANDA. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 295 Credit and Capital Markets 2 / 2015 Outputs of Econometric Analyses Output #1 Output #2 The Granger Theorem states that if Y and X are cointegrated, the relationship can be written as below and at least one between γ1 γ2 must be ≠ 0. (eq.1) ΔYt= a1 ΔYt – 1 + b0 ΔXt + b1 ΔXt – 1 + γ1 (Yt – 1 – Xt – 1) (eq.2) ΔXt = a1 ΔXt – 1 + b0 ΔYt + b1 ΔYt – 1 + γ2 (Yt – 1 – Xt – 1) γ1 and γ2 are the coefficient of the cointegrating equation. At least one of them must be statistically different from zero and with negative coefficient, as it shows how a variable, when the distance between the two variables grows, is brought back to the equilibrium and the model is then stable. Those coefficients should then be between 0 and –1. It is the speed of adjustment of the dependent variable to the equilibrium. For instance, if it is equal to 0.5 it means a 50% movement back to equilibrium following a shock to the model one period later. If it is equal to 1 then there is full adjustment to the equilibrium the period after. A coefficient higher than 1 would not make much sense. _cons -.4486328 .3264045 -1.37 0.170 -1.090881 .1936157 lnm2 1.62709 .0385993 42.15 0.000 1.55114 1.70304 commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 141.077867 311 .453626581 Root MSE = .26 Adj R-squared = 0.8510 Residual 20.9564659 310 .067601503 R-squared = 0.8515 Model 120.121401 1 120.121401 Prob > F = 0.0000 F( 1, 310) = 1776.90 Source SS df MS Number of obs = 312   Output #2 TheGrangerTheoremstatesthatifYandXarecointegrated,therelationshipcanbewrittenas belowandatleastonebetweenγ1γ2mustbe≠0. ΔYt=a1ΔYt‐1+b0ΔXt+b1ΔXt‐1+γ1(Yt‐1‐Xt‐1)(eq.1) ΔXt=a1ΔXt‐1+b0ΔYt+b1ΔYt‐1+γ2(Yt‐1‐Xt‐1)(eq.2) γ1andγ 2 are the coefficient of the cointegrating equation. At least one of them must be statisticallydifferentfromzeroandwithnegativecoefficient,asitshowshowavariable,when thedistancebetweenthetwovariablesgrows,isbroughtbacktotheequilibriumandthemodel isthenstable.Thosecoefficientsshouldthenbebetween0and‐1.Itisthespeedofadjustment ofthedependentvariabletotheequilibrium.Forinstance,ifitisequalto0.5itmeansa50% movementbacktoequilibriumfollowingashocktothemodeloneperiodlater.Ifitisequalto1 then there is full adjustment to the equilibrium the period after. A coefficient higher than 1 wouldnotmakemuchsense.  Durbin-Watson d-statistic( 2, 312) = .1009832 _cons .0024848 .004561 0.54 0.586 -.00649 .0114596 LD. -.1069497 .055503 -1.93 0.055 -.2161641 .0022646 L1. -.0707039 .0180856 -3.91 0.000 -.1062914 -.0351165 coin1 D.coin1 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0020 Z(t) -3.909 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 310 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 296 Diego Valiante Credit and Capital Markets 2 / 2015 Output #3    Output #3   _cons .0205437 .006461 3.18 0.002 .0078301 .0332574 L1. -.0812382 .0178908 -4.54 0.000 -.1164428 -.0460336 coin1 D1. -2.367319 1.075996 -2.20 0.029 -4.484607 -.2500303 lnm2 LD. -.104876 .0548934 -1.91 0.057 -.2128924 .0031404 commTOT D.commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 2.06783957 309 .006692037 Root MSE = .0786 Adj R-squared = 0.0767 Residual 1.89063841 306 .006178557 R-squared = 0.0857 Model .177201164 3 .059067055 Prob > F = 0.0000 F( 3, 306) = 9.56 Source SS df MS Number of obs = 310 _cons .0039853 .0003435 11.60 0.000 .0033093 .0046612 L1. -.0026633 .0009751 -2.73 0.007 -.004582 -.0007446 coin1 D1. -.006499 .0029625 -2.19 0.029 -.0123284 -.0006696 commTOT LD. .0947919 .0571094 1.66 0.098 -.017585 .2071687 lnm2 D.lnm2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .005468579 309 .000017698 Root MSE = .00413 Adj R-squared = 0.0378 Residual .005211015 306 .000017029 R-squared = 0.0471 Model .000257564 3 .000085855 Prob > F = 0.0020 F( 3, 306) = 5.04 Source SS df MS Number of obs = 310 _cons -9.729133 .4339758 -22.42 0.000 -10.58304 -8.875222 lnm2 2.500503 .0513203 48.72 0.000 2.399523 2.601483 NONcommTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 320.740798 311 1.03132089 Root MSE = .34569 Adj R-squared = 0.8841 Residual 37.0456178 310 .119501993 R-squared = 0.8845 Model 283.69518 1 283.69518 Prob > F = 0.0000 F( 1, 310) = 2373.98 Source SS df MS Number of obs = 312 _cons -.0014365 .0118329 -0.12 0.903 -.0247215 .0218485 L3D. -.0945171 .0563819 -1.68 0.095 -.2054667 .0164325 L2D. -.1436072 .0591232 -2.43 0.016 -.2599513 -.0272632 LD. -.2046474 .0607481 -3.37 0.001 -.3241889 -.0851058 L1. -.1437611 .0387549 -3.71 0.000 -.220024 -.0674983 coin2 D.coin2 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0040 Z(t) -3.709 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmente d D i c k ey-Fu l l er test f or un i t root Num b er o f o b s= 3 0 8    Output #3   _cons .0205437 .006461 3.18 0.002 .0078301 .0332574 L1. -.0812382 .0178908 -4.54 0.000 -.1164428 -.0460336 coin1 D1. -2.367319 1.075996 -2.20 0.029 -4.484607 -.2500303 lnm2 LD. -.104876 .0548934 -1.91 0.057 -.2128924 .0031404 commTOT D.commTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 2.06783957 309 .006692037 Root MSE = .0786 Adj R-squared = 0.0767 Residual 1.89063841 306 .006178557 R-squared = 0.0857 Model .177201164 3 .059067055 Prob > F = 0.0000 F( 3, 306) = 9.56 Source SS df MS Number of obs = 310 _cons .0039853 .0003435 11.60 0.000 .0033093 .0046612 L1. -.0026633 .0009751 -2.73 0.007 -.004582 -.0007446 coin1 D1. -.006499 .0029625 -2.19 0.029 -.0123284 -.0006696 commTOT LD. .0947919 .0571094 1.66 0.098 -.017585 .2071687 lnm2 D.lnm2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .005468579 309 .000017698 Root MSE = .00413 Adj R-squared = 0.0378 Residual .005211015 306 .000017029 R-squared = 0.0471 Model .000257564 3 .000085855 Prob > F = 0.0020 F( 3, 306) = 5.04 Source SS df MS Number of obs = 310 _cons -9.729133 .4339758 -22.42 0.000 -10.58304 -8.875222 lnm2 2.500503 .0513203 48.72 0.000 2.399523 2.601483 NONcommTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 320.740798 311 1.03132089 Root MSE = .34569 Adj R-squared = 0.8841 Residual 37.0456178 310 .119501993 R-squared = 0.8845 Model 283.69518 1 283.69518 Prob > F = 0.0000 F( 1, 310) = 2373.98 Source SS df MS Number of obs = 312 _cons -.0014365 .0118329 -0.12 0.903 -.0247215 .0218485 L3D. -.0945171 .0563819 -1.68 0.095 -.2054667 .0164325 L2D. -.1436072 .0591232 -2.43 0.016 -.2599513 -.0272632 LD. -.2046474 .0607481 -3.37 0.001 -.3241889 -.0851058 L1. -.1437611 .0387549 -3.71 0.000 -.220024 -.0674983 coin2 D.coin2 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0040 Z(t) -3.709 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmente d D i c k ey-Fu l l er test f or un i t root Num b er o f o b s= 3 0 8 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 297 Credit and Capital Markets 2 / 2015    Output #4    _cons .0255268 .0171685 1.49 0.138 -.0082566 .0593101 L1. -.1832619 .0365718 -5.01 0.000 -.255226 -.1112978 coin2 D1. -2.868 2.843853 -1.01 0.314 -8.463982 2.727982 lnm2 LD. -.1301126 .0564883 -2.30 0.022 -.2412674 -.0189579 NONcommTOT D.NONcommTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 15.394331 309 .049819841 Root MSE = .20996 Adj R-squared = 0.1152 Residual 13.4892931 306 .044082657 R-squared = 0.1237 Model 1.90503788 3 .635012625 Prob > F = 0.0000 F( 3, 306) = 14.41 Source SS df MS Number of obs = 310 _cons .0037679 .0003417 11.03 0.000 .0030955 .0044402 L1. .0000999 .0007302 0.14 0.891 -.0013371 .0015368 coin2 D1. -.0011387 .0011296 -1.01 0.314 -.0033616 .0010841 NONcommTOT LD. .1317168 .0567432 2.32 0.021 .0200606 .243373 lnm2 D.lnm2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .005468579 309 .000017698 Root MSE = .00418 Adj R-squared = 0.0107 Residual .005357756 306 .000017509 R-squared = 0.0203 Model .000110823 3 .000036941 Prob > F = 0.0990 F( 3, 306) = 2.11 Source SS df MS Number of obs = 310 _cons -7.737542 .613394 -12.61 0.000 -8.944485 -6.5306 commSHORT 1.460438 .048561 30.07 0.000 1.364887 1.555989 NONcommLONG Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 434.85012 311 1.3982319 Root MSE = .59838 Adj R-squared = 0.7439 Residual 110.99847 310 .358059581 R-squared = 0.7447 Model 323.851649 1 323.851649 Prob > F = 0.0000 F( 1, 310) = 904.46 Source SS df MS Number of obs = 312 _cons -.0039797 .018784 -0.21 0.832 -.0409422 .0329828 L2D. -.1623107 .0557445 -2.91 0.004 -.2720032 -.0526182 LD. -.1722922 .057707 -2.99 0.003 -.2858464 -.058738 L1. -.137106 .0341737 -4.01 0.000 -.2043521 -.0698599 coin5 D.coin5 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0013 Z(t) -4.012 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 309 Output #4    Output #4    _cons .0255268 .0171685 1.49 0.138 -.0082566 .0593101 L1. -.1832619 .0365718 -5.01 0.000 -.255226 -.1112978 coin2 D1. -2.868 2.843853 -1.01 0.314 -8.463982 2.727982 lnm2 LD. -.1301126 .0564883 -2.30 0.022 -.2412674 -.0189579 NONcommTOT D.NONcommTOT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 15.394331 309 .049819841 Root MSE = .20996 Adj R-squared = 0.1152 Residual 13.4892931 306 .044082657 R-squared = 0.1237 Model 1.90503788 3 .635012625 Prob > F = 0.0000 F( 3, 306) = 14.41 Source SS df MS Number of obs = 310 _cons .0037679 .0003417 11.03 0.000 .0030955 .0044402 L1. .0000999 .0007302 0.14 0.891 -.0013371 .0015368 coin2 D1. -.0011387 .0011296 -1.01 0.314 -.0033616 .0010841 NONcommTOT LD. .1317168 .0567432 2.32 0.021 .0200606 .243373 lnm2 D.lnm2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .005468579 309 .000017698 Root MSE = .00418 Adj R-squared = 0.0107 Residual .005357756 306 .000017509 R-squared = 0.0203 Model .000110823 3 .000036941 Prob > F = 0.0990 F( 3, 306) = 2.11 Source SS df MS Number of obs = 310 _cons -7.737542 .613394 -12.61 0.000 -8.944485 -6.5306 commSHORT 1.460438 .048561 30.07 0.000 1.364887 1.555989 NONcommLONG Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 434.85012 311 1.3982319 Root MSE = .59838 Adj R-squared = 0.7439 Residual 110.99847 310 .358059581 R-squared = 0.7447 Model 323.851649 1 323.851649 Prob > F = 0.0000 F( 1, 310) = 904.46 Source SS df MS Number of obs = 312 _cons -.0039797 .018784 -0.21 0.832 -.0409422 .0329828 L2D. -.1623107 .0557445 -2.91 0.004 -.2720032 -.0526182 LD. -.1722922 .057707 -2.99 0.003 -.2858464 -.058738 L1. -.137106 .0341737 -4.01 0.000 -.2043521 -.0698599 coin5 D.coin5 Coef. Std. Err. t P>|t| [95% Conf. Interval] MacKinnon approximate p-value for Z(t) = 0.0013 Z(t) -4.012 -3.455 -2.878 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 309 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 298 Diego Valiante Credit and Capital Markets 2 / 2015 Output #5 Output #5   Output #6  _cons -.0038412 .0190743 -0.20 0.841 -.0413745 .033692 1 L1. -.1598455 .0332285 -4.81 0.000 -.2252308 -.0944602 coin5 D1. 1.753686 .1919461 9.14 0.000 1.375985 2.131387 commSHORT LD. -.1039714 .0501897 -2.07 0.039 -.202732 -.0052108 NONcommLONG NONcommLONG Coef. Std. Err. t P>|t| [95% Conf. Interval] D. Total 47.5573066 309 .153907141 Root MSE = .33408 Adj R-squared = 0.2748 Residual 34.153502 306 .111612752 R-squared = 0.2818 Model 13.4038046 3 4.46793487 Prob > F = 0.0000 F( 3, 306) = 40.03 Source SS df MS Number of obs = 310 _cons .0092288 .0049606 1.86 0.064 -.0005325 .0189901 L1. .018442 .0089171 2.07 0.039 .0008952 .0359888 coin5 LD. .0294207 .0146473 2.01 0.045 .0005981 .0582432 D1. .120293 .0131884 9.12 0.000 .0943413 .1462448 NONcommLONG LD. -.1805889 .0559078 -3.23 0.001 -.2906028 -.0705751 commSHORT D.commSHORT Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 3.03075463 309 .009808267 Root MSE = .08687 Adj R-squared = 0.2307 Residual 2.30139603 305 .007545561 R-squared = 0.2407 Model .729358602 4 .182339651 Prob > F = 0.0000 F( 4, 305) = 24.17 Source SS df MS Number of obs = 310 Prob > F = 0.0013 F( 2, 349) = 6.80 ( 2) L2D.LnSp500 = 0 ( 1) LD.LnSp500 = 0 . test dl1.LnSp500 dl2.LnSp500 _cons .0010496 .0009388 1.12 0.264 -.0007967 .002896 L2D. .0584193 .0362431 1.61 0.108 -.0128632 .1297017 LD. .1234158 .035819 3.45 0.001 .0529676 .1938639 LnSp500 L2D. .1405412 .0517441 2.72 0.007 .0387716 .2423107 LD. .0747558 .0530476 1.41 0.160 -.0295773 .179089 lnindexpos~n lnindexpos~n Coef. Std. Err. t P>|t| [95% Conf. Interval] D. Total .116236705 353 .000329282 Root MSE = .01754 Adj R-squared = 0.0659 Residual .107345758 349 .000307581 R-squared = 0.0765 Model .008890947 4 .002222737 Prob > F = 0.0000 F( 4, 349) = 7.23 Source SS df MS Number of obs = 354 D_lnindexposition ALL 13 275 2 0 001 D_lnindexposition D.LnSp500 13.275 2 0.001 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests . vargranger OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 305 Credit and Capital Markets 2 / 2015 (c) 2002–2011 (c) 2002‐2011  D_LnVix ALL 6.0271 3 0.110 D_LnVix D.lnnoncomm 6.0271 3 0.110 D_lnnoncomm ALL 2.1708 3 0.538 D_lnnoncomm D.LnVix 2.1708 3 0.538 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests . vargranger _cons -.0002015 .0049101 -0.04 0.967 -.0098251 .009422 1 L3D. .0339411 .0437054 0.78 0.437 -.0517198 .119602 1 L2D. -.0029362 .044963 -0.07 0.948 -.091062 .085189 6 LD. -.2379383 .0436259 -5.45 0.000 -.3234436 -.15243 3 LnVix L3D. .1359666 .0659793 2.06 0.039 .0066495 .265283 8 L2D. -.0901239 .0659167 -1.37 0.172 -.2193183 .039070 5 LD. .041004 .0659733 0.62 0.534 -.0883013 .170309 3 lnnoncomm D_LnVix _cons .0034873 .0032461 1.07 0.283 -.002875 .009849 5 L3D. .0000848 .0288939 0.00 0.998 -.0565463 .056715 8 L2D. -.0152794 .0297253 -0.51 0.607 -.0735399 .042981 2 LD. .0347707 .0288414 1.21 0.228 -.0217574 .091298 8 LnVix L3D. -.0846017 .0436194 -1.94 0.052 -.1700941 .000890 8 L2D. -.0724849 .043578 -1.66 0.096 -.1578961 .012926 4 LD. .0409936 .0436154 0.94 0.347 -.044491 .126478 2 lnnoncomm D_lnnoncomm Coef. Std. Err. z P>|z| [95% Conf. Interval] D_LnVix 7 .112564 0.0697 39.0216 0.0000 D_lnnoncomm 7 .074417 0.0194 10.30241 0.1125 Equation Parms RMSE R-sq chi2 P>chi2 Det(Sigma_ml) = .0000682 SBIC = -3.74864 1 FPE = .000072 HQIC = -3.81820 5 Log likelihood = 1020.311 AIC = -3.86299 9 Sample: 483 - 1003 No. of obs = 52 1 Vector autoregression . var d.lnnoncomm d.LnVix if tin(483,1003), lags(1/3) OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 306 Diego Valiante Credit and Capital Markets 2 / 2015 Output #9 (a) 1992–2011 Output #9 (a) 1992‐2011   D_LnVix ALL 9.5235 5 0.090 D_LnVix D.lnnocomlong 9.5235 5 0.090 D_lnnocomlong ALL 7.6402 5 0.177 D_lnnocomlong D.LnVix 7.6402 5 0.177 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests . vargranger _cons .0004232 .0033451 0.13 0.899 -.0061332 .0069795 L5D. -.0554288 .0315983 -1.75 0.079 -.1173603 .0065027 L4D. -.0440966 .0324382 -1.36 0.174 -.1076742 .0194811 L3D. -.0019844 .0324943 -0.06 0.951 -.065672 .0617032 L2D. -.0546619 .032335 -1.69 0.091 -.1180373 .0087135 LD. -.230547 .0316048 -7.29 0.000 -.2924912 -.1686028 LnVix L5D. -.0045648 .0177741 -0.26 0.797 -.0394015 .0302718 L4D. .0517498 .0178108 2.91 0.004 .0168414 .0866583 L3D. -.0010107 .0177692 -0.06 0.955 -.0358377 .0338164 L2D. -.0140857 .0184874 -0.76 0.446 -.0503202 .0221489 LD. .0184123 .0184551 1.00 0.318 -.017759 .0545836 lnnocomlong D _LnVix _cons .0037165 .0056929 0.65 0.514 -.0074414 .0148744 L5D. .021636 .0537757 0.40 0.687 -.0837625 .1270344 L4D. -.0575989 .055205 -1.04 0.297 -.1657988 .050601 L3D. .0172261 .0553005 0.31 0.755 -.0911609 .125613 L2D. -.1252399 .0550294 -2.28 0.023 -.2330955 -.0173842 LD. -.0308078 .0537867 -0.57 0.567 -.1362279 .0746123 LnVix L5D. -.0232346 .030249 -0.77 0.442 -.0825215 .0360523 L4D. -.0120244 .0303113 -0.40 0.692 -.0714336 .0473847 L3D. -.1167583 .0302406 -3.86 0.000 -.1760289 -.0574878 L2D. -.0078661 .0314628 -0.25 0.803 -.069532 .0537999 LD. .0639396 .0314079 2.04 0.042 .0023813 .1254979 lnnocomlong D _lnnocoml~g Coef. Std. Err. z P>|z| [95% Conf. Interval] D_LnVix 11 .106096 0.0644 68.59819 0.0000 D_lnnocomlong 11 .18056 0.0286 29.33622 0.0011 Equation Parms RMSE R-sq chi2 P>chi2 Det(Sigma_ml) = .0003583 SBIC = -2.10616 FPE = .0003744 HQIC = -2.173248 Log likelihood = 1125.873 AIC = -2.214389 Sample: 7 - 1003 No. of obs = 997 Vector autoregression OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 Three Narratives on the Changing Face 307 Credit and Capital Markets 2 / 2015 (b) 1992–2001 (b) 1992‐2001  D_LnVix ALL 8.5082 5 0.130 D_LnVix D.lnnocomlong 8.5082 5 0.130 D_lnnocomlong ALL 8.3867 5 0.136 D_lnnocomlong D.LnVix 8.3867 5 0.136 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests . vargranger _cons .0011988 .0044394 0.27 0.787 -.0075022 .0098998 L5D. -.0495701 .0454865 -1.09 0.276 -.1387221 .0395819 L4D. -.1089294 .0463863 -2.35 0.019 -.1998449 -.0180139 L3D. -.0459442 .0468118 -0.98 0.326 -.1376936 .0458053 L2D. -.1361711 .0462092 -2.95 0.003 -.2267393 -.0456028 LD. -.2117214 .0455857 -4.64 0.000 -.3010678 -.122375 LnVix L5D. -.0028873 .0180473 -0.16 0.873 -.0382594 .0324847 L4D. .0455337 .0180153 2.53 0.011 .0102243 .0808431 L3D. -.0215225 .0179693 -1.20 0.231 -.0567417 .0136967 L2D. -.0117164 .0188911 -0.62 0.535 -.0487424 .0253095 LD. .0146031 .0189095 0.77 0.440 -.0224589 .051665 lnnocomlong D_LnVix _cons .0007539 .0106453 0.07 0.944 -.0201105 .0216184 L5D. .0256494 .1090738 0.24 0.814 -.1881313 .2394302 L4D. -.0969172 .1112314 -0.87 0.384 -.3149267 .1210922 L3D. .1040774 .1122517 0.93 0.354 -.115932 .3240867 L2D. -.2388437 .1108066 -2.16 0.031 -.4560206 -.0216667 LD. -.1117488 .1093117 -1.02 0.307 -.3259957 .1024982 LnVix L5D. -.0403679 .0432763 -0.93 0.351 -.1251878 .044452 L4D. -.0300629 .0431996 -0.70 0.486 -.1147326 .0546069 L3D. -.1290661 .0430892 -3.00 0.003 -.2135194 -.0446128 L2D. -.017854 .0452997 -0.39 0.693 -.1066398 .0709318 LD. .0240137 .0453438 0.53 0.596 -.0648585 .1128858 lnnocomlong D_lnnocoml~g Coef. Std. Err. z P>|z| [95% Conf. Interval] D_LnVix 11 .098069 0.0782 40.48869 0.0000 D_lnnocomlong 11 .235164 0.0397 19.69592 0.0323 Equation Parms RMSE R-sq chi2 P>chi2 Det(Sigma_ml) = .000506 SBIC = -1.628684 FPE = .0005549 HQIC = -1.745322 Log likelihood = 456.2837 AIC = -1.820896 Sample: 7 - 483 No. of obs = 477 Vector autoregression OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15 308 Diego Valiante Credit and Capital Markets 2 / 2015 (c) 2002–2011 (c) 2002‐2011     D_LnVix ALL 7.8948 3 0.048 D_LnVix D.lnnocomlong 7.8948 3 0.048 D_lnnocomlong ALL 4.1631 3 0.244 D_lnnocomlong D.LnVix 4.1631 3 0.244 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests . vargranger _cons -.0004674 .0049074 -0.10 0.924 -.0100857 .009150 8 L3D. .0405226 .0436966 0.93 0.354 -.0451211 .126166 2 L2D. -.0076323 .0448613 -0.17 0.865 -.0955588 .080294 1 LD. -.2366194 .0435762 -5.43 0.000 -.3220272 -.151211 6 LnVix L3D. .1216999 .0459984 2.65 0.008 .0315446 .211855 1 L2D. -.0678971 .0473205 -1.43 0.151 -.1606436 .024849 4 LD. .0335107 .0460577 0.73 0.467 -.0567608 .123782 1 lnnocomlong D_LnVix _cons .0052598 .0046603 1.13 0.259 -.0038744 .014393 9 L3D. -.037147 .0414971 -0.90 0.371 -.1184799 .044185 9 L2D. -.0601805 .0426032 -1.41 0.158 -.1436813 .023320 2 LD. .041696 .0413828 1.01 0.314 -.0394129 .122804 9 LnVix L3D. -.0746152 .0436831 -1.71 0.088 -.1602326 .011002 2 L2D. -.0183554 .0449387 -0.41 0.683 -.1064336 .069722 8 LD. .2348404 .0437394 5.37 0.000 .1491127 .320568 1 lnnocomlong D_lnnocoml~g Coef. Std. Err. z P>|z| [95% Conf. Interval] D_LnVix 7 .112365 0.0730 41.00625 0.0000 D_lnnocomlong 7 .106709 0.0650 36.19711 0.0000 Equation Parms RMSE R-sq chi2 P>chi2 Det(Sigma_ml) = .0001396 SBIC = -3.032894 FPE = .0001473 HQIC = -3.102458 Log likelihood = 833.8593 AIC = -3.147252 Sample: 483 - 1003 No. of obs = 521 Vector autoregression . var d.lnnocomlong d.LnVix if tin(483,1003), lags(1/3) OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.48.2.243 | Generated on 2023-01-16 13:25:15