Long‐short speculator sentiment in agricultural commodity markets

Borgards, Oliver; Czudaj, Robert L. Article — Published Version Long‐short speculator sentiment in agricultural commodity markets International Journal of Finance & Economics Provided in Cooperation with: John Wiley & Sons Suggested Citation: Borgards, Oliver; Czudaj, Robert L. (2022) : Long‐short speculator sentiment in agricultural commodity markets, International Journal of Finance & Economics, ISSN 1099-1158, John Wiley & Sons, Ltd., Chichester, UK, Vol. 28, Iss. 4, pp. 3511-3528, https://doi.org/10.1002/ijfe.2605 This Version is available at: https://hdl.handle.net/10419/288138 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. http://creativecommons.org/licenses/by-nc-nd/4.0/ RESEARCH ARTICLE Long-short speculator sentiment in agricultural commodity markets Oliver Borgards 1 | Robert L. Czudaj 1,2 1 Department of Economics and Business, Chair for Empirical Economics, Chemnitz University of Technology, Chemnitz, Germany 2 Department of Mathematics, Computer Science and Statistics, Chair for Statistics and Econometrics, Ludwig-MaximiliansUniversity Munich, Munich, Germany Correspondence Robert L. Czudaj, Department of Economics and Business, Chair for Empirical Economics, Chemnitz University of Technology, Thüringer Weg 7, Chemnitz D-09126, Germany. Email: robert-lukas.czudaj@wirtschaft. tu-chemnitz.de Abstract This paper tests the hypothesis that long-short speculators are able to generate short-term investment returns based on their sentiment for 12 agricultural commodity futures. For this purpose, we dynamically model the equidirectional trading of long and short commodity futures of long-short speculators as a proxy for their market sentiment. We find evidence that the sentiment period returns are considerably positive and differ significantly from neutral sentiment periods for all commodities, which underlines the sentiment's relevance. In line with the empirical literature, we can reject the argument of price manipulation as the price continues to develop into the direction of the sentiment period although long-short speculators trade non-directionally in the following. We rather indicate the existence of a short-term time-series momentum effect, which can be robustly identified without the requirement to define an external model parameter. From the superior sentiment-based momentum returns, we conclude that long-short speculators have valuable, exclusive information, which cannot be replicated by observing their trading activity with a time lag of eight trading days. We also find that a sentiment-based momentum strategy generates significantly higher returns than the long-short speculators have realised in the 15-year sample period which we attribute to the complexity of the long-short speculators' investment strategies. KEYWORDS commodities, commitment of traders, long-short speculators, price manipulation, sentiment, time-series momentum 1|INTRODUCTION For thousands of years, people have been consuming and trading natural resources, in particular agricultural products, as a prerequisite for human and economic development. However, in the recent decades, commodity markets have changed fundamentally. Initially developed for producers and processors of the physical assets, the early agricultural commodity markets can be characterised by a lack of liquidity and transparency as investors were not able to efficiently invest in them. In the early days of commodity investing, the academic literature shares the view that investors who were speculating on the price development of commodities follow the Received: 8 June 2021 Revised: 27 January 2022 Accepted: 3 February 2022 DOI: 10.1002/ijfe.2605 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2022 The Authors. International Journal of Finance & Economics published by John Wiley & Sons Ltd. Int J Fin Econ. 2023;28:3511–3528. wileyonlinelibrary.com/journal/ijfe 3511 trading activity of hedgers (see Working, 1953,1954, 1960,1962). With the advent of the financialisation of commodity markets in the last decades, the enabling of direct investments into commodities goes hand in hand with an increasing number of complex and innovative financial products with the primary intention of portfolio diversification and inflation hedging (Domanski & Heath, 2007). In particular, the introduction of commodity indices was a significant catalyst for change in the commodity markets. As a consequence of rising volumes into passive commodity products and accompanying increasing agricultural commodity prices, the G20 announced food security as one of the world's priorities, suspecting newer speculative-oriented market participants such as commodity index traders (i.e., long-only money managers who track the performance of a commodity index) and long-short speculators (i.e., investors who speculate on increasing or decreasing commodity prices with commodity derivatives) to negatively influence commodity prices. The empirical literature on speculation in commodity markets was initiated by the political and regulatory discussion of the Master's hypothesis (Masters and White, 2008) which accused the significantly increased inflows of commodity index funds for the increase in commodity prices. Concentrated on the commodity index trader's influence, the empirical literature unambiguously rejects the Master's hypothesis and shows that they do not have a significant impact on commodity prices, but provide liquidity to the financial commodity markets (see Brunetti & Reiffen, 2014; Irwin & Sanders, 2011,2012; Maria et al., 2020; Palazzi et al., 2020; Sanders et al., 2010). The majority of the empirical literature on speculation in the commodity markets discusses the impact of financialisation as well as the role of speculators without considering a specific trader group. As a result, the largest part of the empirical literature also comes to the same conclusion (see Boyd et al., 2018; Fishe & Smith, 2019; Kim, 2015; Manera et al., 2013; Mayer et al., 2017; Often & Wisen, 2013; Wimmer et al., 2021). In the context of the increasing world population, this political and ethical discussion about the world's most essential resources remains vivid and primarily concentrates on index-tracking market players to our surprise. On the contrary, only a minor part of the empirical literature examines the role of classical speculators like hedge funds, which opportunistically exploit investment opportunities with long or short directional trades based on their information. The empirical literature predominantly confirms the prevailing view on the role of speculators also for long-short traders (see Bohl et al., 2021; Bohl & Sulewski, 2019; Brunetti et al., 2016; Buyuksahin & Robe, 2014; Miffre & Brooks, 2013). Since the investment motive of long-short traders differs considerably from that of commodity index traders, the question on the role of their information-based sentiment in this legitimate discussion about price influence remains. This motivates us to provide a comprehensive analysis of the long-short speculator's sentiment, which is the key driver in their investment process. To the best of our knowledge, this is the first study, which uses the longshort speculators' open interest as a proxy for their market sentiment to examine price dynamics as well as price influence. From the perspective of an investor, our results are highly relevant. On the one hand, they show the validity of their information and the resulting price dynamics. On the other hand, they extend the empirical literature's view on price manipulation by providing valuable insights whether their information-driven investment style has an effect on agricultural commodity prices. Hence, our motivation is to analyse if long-short speculators as the smallest trader group in the commodity markets besides producers, processors and passive investors have a measurable impact on agricultural commodity prices. In order to assess the relevance of the long-short speculator's information, we use periods of equidirectional trading defined as buying long and selling short futures contracts (long sentiment) and vice versa (short sentiment) as a proxy for their sentiment on 12 agricultural commodity futures. The open interest of futures is particularly suitable for the development of a sentiment indicator as it can be regarded as the commodity futures market's cash flow. Increasing or decreasing open interest can be a signal when certain market participants are entering or leaving the market and may give clues to market direction. In addition, our open interest based sentiment proxy also offers the benefit that we are able to rely on data on a higher frequency compared to survey based indicators, which would at best be available at a monthly frequency. Initially, we compare the commodity's log returns in long or short sentiment periods with those of neutral sentiment periods. Since equidirectional trading can impact the price itself, we also consider how the commodity price evolves subsequent to long or short sentiment periods. We can reject the argument of price manipulation, if the commodity price immediately halts or reverts back directly after the long or short sentiment period. We also test the exclusiveness of the long-short speculators information by replicating a sentimentmodelled momentum strategy based on publicly available open interest data of long-short speculators. Finally, we compare the risk–return characteristics of an exclusive and a publicly available, modelled sentiment-based trading strategy with the realised ex post returns of the longshort speculators over a 15-year sample period. 3512 BORGARDS AND CZUDAJ We contribute to the empirical literature by studying short-term price dynamics on the basis of the long-short speculators' sentiment in agricultural commodity markets. In contrast, the vast majority of the existing literature studies the speculative behaviour of commodity index traders, in particular if they have a negative impact on the agricultural commodity prices as one of the earth's most essential resources. We instead concentrate on longshort speculators representing a group of investment managers who have probably the clearest investment intentions as they predominantly enter directional trades to follow their opportunistic investment strategies (Etienne et al., 2014). We also contribute by modelling investor sentiment dynamically as a period of equidirectional trading without the requirement to define external parameters. Since it is not reasonable to quantify investor sentiment as a proxy of the long-short speculators' future price expectations with external parameters, the corresponding findings might be biased when relying on such an approach. To the best of our knowledge, this is the first study that also tests the time-series momentum effect on commodity markets by dynamically considering the long-short speculator's sentiment as the initialising formation period. Consequently, our work extends the empirical literature by illustrating price dynamics during and after varying periods of consistent trading activity for a broad range of agricultural commodities over more than a decade. We find evidence that the sentiment of long-short speculators is highly relevant in agricultural commodity markets. Defining long (short) sentiment dynamically as a period in which long-short speculators consistently buy (sell) long futures contracts and sell (buy) short futures contracts, we find that the log returns in long (short) sentiment periods are positive (negative) and significantly different from zero for all commodities. On the contrary, the log returns in neutral sentiment periods do not differ significantly from zero. Since we are able to regularly confirm these results over the entire sample period, our findings underline the directional sentiment's importance. Furthermore, since the commodity prices predominantly move further into the direction of the sentiment period, we are able to reject the argument of price manipulation because long-short speculators do not trade consistently anymore. When applying our results to the concept of time-series momentum, a long or short sentiment period would represent a formation period which is followed by a corresponding momentum period. Therefore, our results strongly support the time-series momentum effect as the momentum periods returns are significantly different from zero for most of the commodities. As a conclusion, we suppose that long-short speculators rather own valuable information in form of their investment strategies, which they use to exploit shortterm price movements. Our results also clearly show that the commodity price initially remains persistent at the beginning of a momentum period but then decays. We therefore conclude that a sentiment period can be understood as a kind of price impulse based on the valuable information of the long-short speculators. The price impulse in the form of equidirectional trading then initiates a time-series momentum effect which weakens the more the price moves away from the original equilibrium price. Moreover, we find that the replication of a sentiment-based trading strategy by observing the open interest of the long-short speculators with a time lag of eight trading days leads to considerably lower returns which are not significantly different from zero or even negative. As the time-lag represents the beginning of a directional sentiment period, our results let us to conclude that in particular the beginning of a sentiment period is a substantial but exclusive stimulus that can have a sustainable impact on the subsequent price development. Finally, we show that the returns of an exclusive sentiment-based momentum strategy considerably outperform the realised ex post returns of the long-short speculators over the entire sample period. Although surprising at first glance, the difference can be explained with the higher complexity of a long-short speculator's investment strategy. The complexity involves hedging requirements, information asymmetries, different investment horizons and negative market impact effects, which are not considered in our modelled sentiment strategy. Nevertheless, our findings provide valuable information for the risk management of individual investment managers to better assess commodity price dynamics. As we clearly provide evidence that long-short speculators do not manipulate commodity prices in line with the existing literature, our findings are an important signal that investment managers who predominantly act opportunistically, follow ethically correct investment practices. The remainder of the paper is structured as follows. Section 2reviews the existing literature on speculation in the commodity markets. Section 3outlines the data and the methodology used in this study. In Section 4, we present and discuss the results of our empirical findings. Section 5offers concluding remarks. 2|REVIEW OF THE LITERATURE The structural change of the agricultural commodity markets in the early 2000s, also changed the academic discussion fundamentally. Previously, the commodity markets were primarily hedgers-driven, in which market participants who were actively engaged in the physical commodity markets (i.e., producers and processors) were BORGARDS AND CZUDAJ 3513 hedging the prices of their commodities with futures contracts (hereinafter hedgers). In the traditional view, investors who were speculating on the price development of commodities (hereinafter speculators) follow the trading activity of hedgers (see Working, 1953,1954,1960,1962). Inthecourseofthefinancialisation of commodity markets in the first decade of the new millennium, investors and individuals could now invest in a broad range of commodities with over-the-counter (OTC) swaps, exchange-traded funds (ETF) and exchange-traded notes (ETN) without having to own the commodities themselves (Domanski & Heath, 2007). This resulted in a sharp increase in trading volume on commodity futures exchanges and established new speculative-oriented marketparticipantssuchascommodity index traders (i.e., long-only money managers who track the performance of a commodity index) and longshort speculators (i.e., investors who speculate on increasing or decreasing commodity prices with commodity derivatives). Therefore, the more recent empirical literature examines whether Working's hypothesis is still valid and whether the financial commodity markets are now speculators-driven. The emergence of the empirical literature on speculation in the commodity markets was initiated by the political and regulatory discussion of the Master's hypothesis. The Master's hypothesis is based on the hedge fund manager Michael W. Masters, who harshly supposed that the significantly increased inflows of commodity index funds are the reason for the sharp increase in commodity prices between 2007 and 2008 and their divergence from fundamental values (Masters and White, 2008). The empirical literature was enabled to examine the Master's hypothesis by the extension of the Commodity Futures Trading Commission's (CFTC) Supplemental Commitment of Traders (SCOT) report to the traditional Commitment of Traders (COT) reports. i It further splits the speculators' open interest for 12 agricultural commodity markets into commodity index traders and long-short traders starting in 2006. This now allowed for a closer examination of long-only commodity index traders and their impact on commodity prices. Sanders et al. (2010) show that the relative share of commodity index traders in open interest is stable between 2006 and 2008, concluding that the increased inflows are the response to a rising hedging demand. Irwin and Sanders (2011) criticise the underlying data as well as the methodological approaches of previous empirical studies that find a significant impact of commodity index trader's futures positions on commodity future prices. Irwin and Sanders (2012) examine the Master's hypothesis using Fama-MacBeth cross-sectional regression tests as well as Granger causality tests. In both studies, they are unable to find a direct impact of commodity index traders' futures positions on returns as well as on volatility so that they reject the Master's hypothesis as one of the first. Brunetti and Reiffen (2014) use a theoretical equilibrium model of trader behaviour to show that commodity index traders reduce hedging costs. After empirically validating their model with the commodity index trader's futures positions, they conclude that commodity index traders provide an important counterparty for hedgers by providing liquidity and do not influence commodity prices directly. Palazzi et al. (2020) use linear and non-linear regressions to find out whether speculators in general and commodity index traders in particular influence commodity futures prices or whether they follow the price movement. As a result, they find no cause and effect relationship, so that the Master's hypothesis can be rejected ultimately. Maria et al. (2020)investigate the influence of commodity index trading activity on commodity futures prices using Granger causality tests during the period 2006 to 2017. They confirm the previous empirical studies that the spike in commodity prices in 2007 to 2008 and the period thereafter is not due to speculative behaviour of commodity index traders. In summary, the empirical literature unambiguously rejects the Master's hypothesis and shows that commodity index traders do not have a significant impact on commodity prices, but provide liquidity to the financial commodity markets. The vast majority of the empirical literature on speculation in the commodity markets discusses the impact of financialisation as well as the role of speculators without considering a specific trader group from the SCOT report. Instead, they predominantly model the activity of hedgers and speculators with open interest from the traditional legacy and disaggregated COT reports. Sanders et al. (2010) note that the group of speculators (i.e., money managers and other reportables) that can be modelled from both reports cannot be assigned to a clear trader group, so that the different intentions (i.e., directional trading, hedging, index-tracking) partially overlap. However, the majority of that part of the empirical literature also comes to the same conclusion as the studies on the Master's hypothesis. Boyd et al. (2018) examine the findings of a large number of empirical studies on the role of commodity speculators and their price impact during the financialisation period. They clearly conclude that speculators provide liquidity to hedgers while finding no evidence of destabilisation as well as price distortion in commodity markets initiated by speculators. Wimmer et al. (2021) analyse more than 50 research articles that study the relationship between commodity prices and speculative behaviour using Granger causality tests. They point out that either speculative behaviour in the agricultural, energy, and metal markets cannot be detected or Granger causality tests are not able to quantify the relationship well enough. 3514 BORGARDS AND CZUDAJ Often and Wisen (2013) also investigate this relationship using Granger causality tests and show that hedgers in particular have a greater impact on prices in certain commodity markets (e.g., live cattle) than swap dealers or producers. As well as Often and Wisen (2013), Manera et al. (2013) model speculative behaviour by Working's Speculative Index (Working, 1960), which measures the excessiveness of speculation as a ratio calculated by measuring the amount by which speculation exceeds commercial hedging needs, divided by commercial open interest. Using dynamic conditional correlation (DCC) multivariate GARCH models, they show that financial speculation cannot explain the returns of five agricultural commodities. Kim (2015) examines the impact of financialisation on commodity prices as well as on their volatility as a proxy for market stability. As a result, speculators do not destabilise commodity spot prices, but tend to contribute to lower volatility and increased price efficiency associated with greater liquidity. Mayer et al. (2017) investigate the same intention with bi-directional Granger causality tests and an EGARCH volatility analysis for metal commodity markets and show that noncommercials in particular do not influence commodity prices and volatility. On the other hand, an influence in sub-samples such as booms and crises can be observed for both commercials and non-commercials, while the effect is greater for long futures positions. Fishe and Smith (2019) show that money managers including speculators tend to follow commodity prices according to their assessment of the future price development having no price-influencing impact. To our surprise, only a minor part of the empirical literature examines the role of classical speculators based on the open interest of long-short traders from the CFTC's SCOT report. Long-short speculators (i.e., hedge fund managers) use their information in the form of selection and timing strategies to enter directional long or short trades, which means that their intentions differ considerably from those of the commodity index traders (Etienne et al., 2014). By replicating various hedge fund strategies with price data of 27 commodity futures in the period from 1992 to 2011, Miffre and Brooks (2013) demonstrate that long-short speculators tend to have no significant impact on volatility as well as on the diversifiability of their commodity investments. Buyuksahin and Robe (2014)use daily open interest data of 17 commodity futures from the CFTC's non-public large trader reporting system. They show that the correlation of commodity and equity index returns increases with increasing trading activity of longshort speculators. In the context of the commodity financialisation, these traders therefore have an impact on the diversifiability of commodities, which cannot be observed for swap dealers and commodity index traders. Brunetti et al. (2016) use the same non-public data set and confirm the prevailing view of the empirical literature on the role of speculators also for long-short speculators. Showing a negative correlation between the futures positions and the volatility of crude oil, natural gas and corn, they provide evidence that long-short speculators stabilise commodity futures markets and inject liquidity into them. Brunetti et al. (2016) study long-short speculators of corn, soybeans, sugar, coffee, and wheat futures for the period from 2006 to 2017 to determine whether their trading activity has an impact on volatility and thus on price stability. Using GARCH models, they estimate conditional volatility and conclude that long-short speculators either have no impact on volatility or even reduce it. Bohl et al. (2021) examine how speculative activity affects informational efficiency of commodity futures markets and find evidence for a significant negative relation between speculative activity and the degree of informational efficiency. A subsequent analysis shows that the results are mainly driven by traditional long-short speculators while the influence of index trader is insignificant. In summary, the empirical literature largely agrees that speculators, either as a specific (i.e., commodity index traders and long-short speculators) or as a nonspecific group of money managers, do not destabilise commodity markets and provide liquidity to them as the main counterparty of hedgers. To the best of our knowledge, this is the first study which uses the long-short speculators' open interest as a proxy for their market sentiment to examine price dynamics as well as price influence and also tests the time-series momentum effect on the commodity markets by dynamically considering the long-short speculator's sentiment as the initialising formation period. 3|DATA AND METHODOLOGY 3.1 |Data We source the open interest data on the basis of the Commodity Futures Trading Commission's (CFTC, https:// www.cftc.gov) Supplement Commodity Index Trader (SCOT) report. It contains the long and short futures positions of three trader groups (non-commercials, index traders and long-short speculators) for 12 agricultural commodities during the sample period from 03 January 2006 to 29 December 2020, recorded as of Tuesday of each week. The commodity futures are wheat, corn, soybeans, soybean oil, soybean meal, cotton, cocoa, sugar, coffee, lean hogs, live cattle and feeder cattle. In addition, we use the daily close prices of the futures contract with the shortest maturity (front contract) for the same dates BORGARDS AND CZUDAJ 3515 TABLE 1 Descriptive statistics Commodity Mean Median SD Minimum Maximum Skewness Kurtosis No. obs. Long OI % Short OI % Sample period Wheat 0.00074 0.00178 0.04492 0.17625 0.16909 0.23481 0.82690 782 0.2132 0.1724 01/03/2006–12/29/2020 Corn 0.00096 0.00206 0.04228 0.25553 0.23255 0.19068 4.04113 782 0.1317 0.1113 01/03/2006–12/29/2020 Soybeans 0.00095 0.00196 0.03416 0.20049 0.12031 0.52429 2.54083 782 0.1423 0.1005 01/03/2006–12/29/2020 Soybean oil 0.00079 0.00054 0.03207 0.11597 0.14310 0.03212 1.12830 782 0.1602 0.1433 01/03/2006–12/29/2020 Soybean meal 0.00015 0.00100 0.03944 0.29350 0.14666 0.93769 8.46808 404 0.1607 0.1293 04/02/2013–12/29/2020 Cotton 0.00042 0.00044 0.04253 0.28964 0.16153 0.71166 5.63051 782 0.1923 0.1030 01/03/2006–12/29/2020 Cocoa 0.00062 0.00094 0.04017 0.16719 0.16840 0.04935 1.25533 782 0.2144 0.1450 01/03/2006–12/29/2020 Sugar 0.00008 0.00077 0.04774 0.22989 0.15835 0.08654 1.66787 782 0.1333 0.1091 01/03/2006–12/29/2020 Coffee 0.00018 0.00043 0.04432 0.14489 0.19934 0.23585 0.90913 782 0.1458 0.1575 01/03/2006–12/29/2020 Lean hogs 0.00005 0.00166 0.05437 0.24099 0.23114 0.22991 3.24713 782 0.1794 0.1260 01/03/2006–12/29/2020 Live cattle 0.00019 0.00158 0.02626 0.14223 0.11666 0.56224 3.27677 782 0.2043 0.1023 01/03/2006–12/29/2020 Feeder cattle 0.00026 0.00154 0.02418 0.11783 0.13702 0.06550 3.23120 782 0.2122 0.1660 01/03/2006–12/29/2020 Note: The table reports the mean, median, standard deviation (SD), minimum value, maximum value, skewness, kurtosis and the number of observations (No. obs.) for the 1 week close price log changes of the 12 commodities. Long and short open interest (OI) % represent the long-short speculator's mean proportion of the total open interest, respectively. 3516 BORGARDS AND CZUDAJ as the weekly open interest data in order to be able to observe futures positioning and prices at identical times. The price data was obtained from the Chicago Mercantile Exchange (CME, https://www.cmegroup.com) and the Intercontinental Exchange (ICE, https://www.theice. com). Table 1shows the descriptive statistics of the weekly close price log returns as well as the average open interest proportion of the long-short speculator's futures positioning. It shows that the log returns of most of the commodities are mildly skewed to the left, which indicates that downturns are steeper than upturns. Excess kurtosis can only be observed for a few commodities. With an average open interest proportion between 10% and 21%, the long-short speculators are the smallest of the three trader groups. 3.2 |Methodology We model the speculator's sentiment based on long and short futures contracts held by long-short speculators from the Commodity Futures Trading Commission's SCOT report. Long-short speculators use their information primarily to enter into directional, speculative trades, while commercials (i.e., producers and processors) predominantly hedge the price of their agricultural commodities and commodity index traders track the price performance of the underlying commodity index. i We therefore assume that the returns achieved by long-short speculators reflect the quality of the information incorporated in the investment strategies. We further assume that long-short speculators have particularly valuable information if they simultaneously buy long futures contracts clong (i.e., clong t>clong t1) and sell short futures contracts cshort (i.e., cshort t<cshort t1) in a period t, which we hereafter refer to as long sentiment and conversely as short sentiment. On the other hand, if they buy or sell long and short futures contracts at the same time, the investment manager sentiment is not uniform (hereafter referred to as neutral sentiment). Figure 1exemplary shows the price time series of the corn futures front contract and the long and short sentiment periods projected onto it at the respective price levels plong tand pshort t. A blue (red) outlined point marks the time twhen the long-short speculators bought long (short) futures contracts clong (cshort) and sold short (long) futures contracts compared to the previous week t1. To evaluate the quality of investment manager sentiment, we consider the log returns Δplongsentiment tand Δpshortsentiment tof the corn futures' front contract prices p for the same weekly periods t. We also define the log returns of the neutral sentiment periods Δpneutralsentiment t, which are shown as non-outlined points in Figure 1: Δplongsentiment t¼log plong t  log pfneutral;shortg tl1  if plong tland pfneutral;shortg tþ1, ð1Þ Δpshortsentiment t¼log pshort t  log pfneutral;longg tl1  if pshort tland pfneutral;longg tþ1, ð2Þ Δpneutralsentiment t¼log pneutral t  log pflong;shortg tl1  if pneutral tland pflong;shortg tþ1, ð3Þ where the parameter lrepresents the length of the consecutive sentiment periods and plong t¼clong t>clong t1and cshort t<cshort t1,ð4Þ pshort t¼clong t<clong t1and cshort t>cshort t1,ð5Þ pneutral t¼clong t>clong t1and cshort t>cshort t1 no or clong t<clong t1and cshort t<cshort t1 no : ð6Þ If investment managers have valuable information about the short-term price development of the respective commodities, we expect the log returns Δplongsentiment t (Δpshortsentiment t) to develop positively (negatively) in long (short) sentiment periods and to differ significantly from zero. We also expect that in neutral sentiment periods the log returns Δpneutralsentiment tdo not differ significantly from zero. If these observations hold, this may have two implications. On the one hand, long-short speculators may have valuable information to predict the short-term price development. Second, they may influence the price itself (price manipulation) by trading futures contracts in the same direction. The argument of price manipulation is invalidated if the price continues to increase (decrease) after one (l=1) or more consecutive (l> 1) long (short) sentiment periods. We therefore define long (short) sentimentmomentum as a long (short) sentiment period that is extended to the next short (long) sentiment period. In Figure 1, the long (short) sentiment-momentum periods are each projected as blue (red) points on the price time series of the corn futures front contract. Again, we calculate the log returns of the corresponding sentiment-momentum periods Δplongsentimentmomentum t,i as well as Δpshortsentimentmomentum t,iand compare them with the log returns of the long and short sentiment periods. The respective sentiment-momentum periods are defined as BORGARDS AND CZUDAJ 3517 Δplongsentimentmomentum t,i¼log pflong;neutralg tþi  log pfshortg tl1þi  if pflong;neutralg tland pshort tþ1, ð7Þ Δpshortsentimentmomentum t,i¼log pfshort;neutralg tþi  log pflongg tl1þi  if pfshort;neutralg tland plong tþ1, ð8Þ where iis a lag parameter defined as i=0 for the original, non-lagged sentiment-momentum periods. We expect that the log returns of the sentiment-momentum periods Δplongsentimentmomentum t,0 and Δpshortsentimentmomentum t,0 differ significantly from the sentiment periods Δplongsentiment tand Δpshortsentiment t, so that we can conclude that long-short speculators have valuable information to be able to predict the short-term price development in a sustainable way. If the log returns of the sentiment-momentum periods actually differ significantly from the ones of the sentiment periods, it is interesting from the perspective of an external trader whether he can also achieve a positive return by trading the long and short sentimentmomentum periods 1 week later after observing this trading behaviour. It should be noted that the open interest data of the SCOT report refers to a Tuesday, while the report is released on the subsequent Friday at 3:30 p.m. Eastern standard time. This means that an external trader can only open a position on Friday at the close price, so that his replicated sentiment momentum strategy has a time lag of eight trading days in total (i=8). Accordingly, we calculate the log returns of the lagged sentiment momentum periods Δplongsentimentmomentum t,8 and Δpshortsentimentmomentum t,8 with the close prices of Friday or the subsequent Monday, respectively, if Friday is an U.S. exchange holiday. An external trader would not only be able to enter a position eight trading days later, but also to close a position eight trading days later, so that we expect their log returns to be at least partially different from those of the sentiment and sentiment-momentum periods. If the lagged sentiment-momentum log returns are absolutely lower and significantly different from those in the sentiment and sentiment-momentum periods, it indicates that during the time lag a significant part of the exclusive information is priced into the commodities by the long-short speculators. Accordingly, the beginning of a sentiment period would represent a meaningful price impulse, which would have a noticeable impact on the short-term price development. FIGURE 1 Sentiment periods of long-short speculators. The blue (red) outlined points mark the end of weekly long (short) sentiment periods in which long-short speculators increase their long (short) corn futures positions and decrease their short (long) corn futures positions. The non-outlined points mark the end of weekly neutral sentiment periods in which long-short speculators increase or decrease their long and short corn futures positions simultaneously. A blue (red) non-outlined point means that the weekly period follows a recent long (short) sentiment period (sentiment momentum). All sentiment and sentiment momentum periods are projected to the price time series of the corn's front contract futures for the period from 31 December 2019 to 29 December 2020. The dark grey line marks the long-short speculator's net futures position in percent of its open interest. A net position of 1.0 (1.0) means that the long-short speculator's open interest is made up of long (short) positions only [Colour figure can be viewed at wileyonlinelibrary.com] 3518 BORGARDS AND CZUDAJ TABLE 4 Ex post long-short speculator and sentiment strategy returns Commodity Strategy Cumulated return Mean return, statistical significance Standard deviation Maximum drawdown Cocoa Ex post 1250.60 1.43387 ns 355.97 9282.55 Coffee Ex post 19,322.45 23.4142 ns 1080.04 26,431.10 Corn Ex post 14,649.42 18.76304 ns 406.58 8419.71 Cotton Ex post 39,934.16 51.14501 ns 1197.49 32,967.00 Soybean meal Ex post 2790.48 6.99488 ns 444.37 8889.60 Soybean oil Ex post 22,802.47 28.98189 *** 271.82 2572.84 Soybeans Ex post 51,518.11 66.26471 ** 942.20 15,574.81 Sugar Ex post 12,195.53 15.1218 ns 645.50 12,035.98 Wheat Ex post 15,532.32 20.82319 ns 730.33 17,367.30 Feeder cattle Ex post 24,409.29 31.45627 ns 610.22 6796.18 Lean hogs Ex post 11,260.64 14.2137 ns 604.83 20,469.20 Live cattle Ex post 18,938.92 24.19006 ns 520.24 9465.57 Cocoa Sentiment-momentum 108,630.00 138.91304 *** 1003.84 16,420.00 Coffee Sentiment-momentum 657,787.50 841.16049 *** 2370.14 29,437.50 Corn Sentiment-momentum 283,200.00 362.14834 *** 934.66 4275.00 Cotton Sentiment-momentum 388,835.00 497.23146 *** 1872.20 20,160.00 Soybean meal Sentiment-momentum 191,260.00 473.41584 *** 1417.97 14,400.00 Soybean oil Sentiment-momentum 205,860.00 263.24808 *** 752.40 5586.00 Soybeans Sentiment-momentum 592,362.50 757.4968 *** 1797.58 16,975.00 Sugar Sentiment-momentum 246,915.20 315.74834 *** 943.23 5958.40 Wheat Sentiment-momentum 251,987.50 322.23465 *** 1377.37 11,550.00 Feeder cattle Sentiment-momentum 243,592.50 311.49936 *** 1679.89 46,775.00 Lean hogs Sentiment-momentum 239,030.00 305.66496 *** 1516.52 15,210.00 Live cattle Sentiment-momentum 143,352.00 183.31458 *** 1168.43 14,260.00 Cocoa Sentiment-momentumlagged 59,120.00 75.60102 ** 1010.55 66,040.00 Coffee Sentiment-momentumlagged 104,625.00 133.79156 ns 2507.68 128,756.25 Corn Sentiment-momentumlagged 14,837.50 18.97379 ns 1002.20 23,425.00 Cotton Sentiment-momentumlagged 28,580.00 36.54731 ns 1936.83 63,650.00 Soybean meal Sentiment-momentumlagged 32,300.00 79.9505 ns 1494.31 21,980.00 Soybean oil Sentiment-momentumlagged 7740.00 9.8977 ns 796.89 28,890.00 Soybeans Sentiment-momentumlagged 25,287.50 32.33696 ns 1950.55 51,300.00 Sugar Sentiment-momentumlagged 9396.80 12.01637 ns 994.37 29,489.60 Wheat Sentiment-momentumlagged 57,275.00 73.24169 ns 1412.69 21,487.50 (Continues) BORGARDS AND CZUDAJ 3525 capitalised commodity investment fund. In summary, hedging requirements, information asymmetries, longer investment horizons and negative market impact effects may explain the lower ex post returns of the long-short speculators compared to our modelled sentiment-momentum strategy. 5|CONCLUSIONS Our paper tests the hypothesis that long-short speculators are able to generate short-term investment returns based on their sentiment. We use the equidirectional trading activity of long-short speculators as a proxy for their sentiment on 12 agricultural commodity futures. In the first step, we compare the commodity futures returns in periods with a long or short sentiment with those of a neutral sentiment to derive the sentiment's relevance. As equidirectional trading can impact the commodity price itself, we measure how the commodity price evolves directly after a long or short sentiment period. In case the commodity price continues to develop in the direction of the sentiment period, we can reject the argument of price manipulation and hypothesise that long-short speculators of agricultural commodities have valuable information. Finally, we investigate whether the replication of a sentiment-based trading strategy can be profitably applied for a trader who is only able to retroactively derive the investment manager's sentiment. We find that the log returns in long (short) sentiment periods are positive (negative) and significantly different from zero for all commodities which is not the case for each neutral sentiment period. The regular occurrence over the entire sample period shows the sentiment periods' relevance. We also find for all commodities and directions that the future prices continue to develop into the direction of the sentiment period. As a consequence, we can reject the argument of price manipulation as the price moves further in the direction of the sentiment period although the long-short speculators have partially oppositely directed open futures positions. We therefore suppose that long-short speculators have valuable information, which they use to exploit shorter-term price movements. Our results indicate the existence of a shortterm time-series momentum effect. Transferring the definition of time-series momentum to our application, a sentiment period represents the formation period, which is directly followed by the momentum period (Moskowitz et al., 2012). As both the sentiment and momentum periods are modelled dynamically on the basis of the long-short speculator's futures positions, our results do not require an external parameter which makes them robust to external market changes during the sample period. Furthermore, we provide evidence that the valuable information of the long-short speculators is exclusive which means that an external trader is not able to replicate the log returns of the sentiment-momentum periods from the long-short speculators. Finally, we conclude that our modelled, sentiment-based momentum strategy generates a significantly higher return in comparison to the realised ex post returns of the long-short speculators. The differences can be explained with the complexity of the investment managers' strategies such as hedging requirements, information asymmetries, longer investment horizons and negative market impact effects. We contribute to the empirical literature on speculation on commodity markets by studying speculative price dynamics on the basis of investor sentiment. Contrary to numerous empirical studies which concentrate on the speculative behaviour of commodity index traders TABLE 4 (Continued) Commodity Strategy Cumulated return Mean return, statistical significance Standard deviation Maximum drawdown Feeder cattle Sentiment-momentumlagged 13,812.50 17.66304 ns 1708.32 48,047.50 Lean hogs Sentiment-momentumlagged 43,872.00 56.1023 ns 1545.89 27,382.00 Live cattle Sentiment-momentumlagged 4348.00 5.5601 ns 1182.91 52,848.00 Note: The table shows the cumulated return, mean return, standard deviation and maximum drawdown of the ex post long-short speculator's futures trading as well as the sentiment-momentum and sentiment-momentum-lagged strategy in US-dollar for the period from 03 January 2006 to 29 December 2020. The ex post long-short speculator return is calculated as the futures return of all long-short speculator's long and short futures positions divided by the open interest in order to approximate the total investment result on the basis of one futures contract. In the same way, the strategies' return time series are also simulated with the trading of one long or short futures contract. The statistical significance of the mean return indicates whether the sample mean return is equal to zero. The asterisks represent the level of significance, where ***, **, * indicates that the test statistic is significant at the 1%, 5% and 10% level, respectively, while ns means that the test statistic is not significant. 3526 BORGARDS AND CZUDAJ (in particular on the Master's hypothesis), we examine the equidirectional trading effects of long-short speculators. Since long-short speculators aim to enter directional trades based on their information, their sentiment serves as an observable proxy for the value of their information. Although our modelled sentiment strategy can only be derived from the aggregated group of long-short speculators, our findings provide valuable information for the risk management of individual investment managers to better assess commodity price dynamics. Moreover, since we clearly reject the argument of manipulating commodity prices, we conclude that long-short speculators ethical correctly invest in agricultural commodities, representing the most essential food resources of our planet. As our modelled sentiment forms a valuable but not observable proxy for the long-short speculators' information, future research might also analyse its inner dynamics. Finally, we would suggest to concentrate more on the behaviour of long-short speculators as they represent the trader group of the CFTC reports that probably have the clearest investment intentions. ACKNOWLEDGEMENTS Thanks for valuable comments on a previous draft of the paper are due to one anonymous reviewer and the participants of the research seminar in Chemnitz. DATA AVAILABILITY STATEMENT Data will be made availabe online upon publication. ORCID Robert L. Czudaj https://orcid.org/0000-0002-3313-8204 ENDNOTES i The COT report provides a breakdown of each Tuesday's open interest for futures markets in which 20 or more traders hold positions equal to or above the reporting levels established by the CFTC. The legacy COT report differentiates between commercials (i.e. producers and processors of the commodities, hedgers), noncommercials (i.e. money managers, speculators) and nonreporting traders (small investors, residuals of the open interest). The disaggregated COT report further splits up commercials into the group of processors/producers and swap dealers as well as non-commercials into managed money and other reportables. See Irwin and Sanders (2012) for a detailed description of the trader groups in the legacy, disaggregated und supplemental COT reports of the CFTC. ii For a more detailed description of the different trader groups and their differences, please see Irwin and Sanders (2012) and Robe and Roberts (2021) in particular for agricultural markets. iii Mellios et al. (2016) also argue that the convenience yield has an important impact on speculation and hedging positions in commodity futures markets and the interaction among time-varying risk premia determines the magnitude and the sign of these positions. REFERENCES Barberis, N., Shleifer, A., & Vishny, R. (1998). A model of investor sentiment. Journal of Financial Economics,49(3), 307–343. Bohl, M. T., Pütz, A., & Sulewski, C. (2021). Speculation and the informational efficiency of commodity futures markets. Journal of Commodity Markets,23, 100159. Bohl, M. T., & Sulewski, C. (2019). The impact of long-short speculators on the volatility of agricultural commodity futures prices. Journal of Commodity Markets,16, 100085. Boyd, N. E., Harris, J. H., & Li, B. (2018). An update on speculation and financialization in commodity markets. Journal of Commodity Markets,10,91–104. Brunetti, C., Büyük Sahin, B., & Harris, J. H. (2016). Speculators, prices, and market volatility. Journal of Financial and Quantitative Analysis,51(5), 1545–1574. Brunetti, C., & Reiffen, D. (2014). Commodity index trading and hedging costs. Journal of Financial Markets,21, 153–180. Buyuksahin, B., & Robe, M. A. (2014). Speculators, commodities and cross-market linkages. Journal of International Money and Finance,42,38–70. Daniel, K., Hirshleifer, D., & Subrahmanyam, A. (1998). Investor psychology and security market underand overreactions. Journal of Finance,53(6), 1839–1885. De Long, J. B., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990a). Noise trader risk in financial markets. Journal of Political Economy,98(4), 703–738. De Long, J. B., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990b). Positive feedback investment strategies and destabilizing rational speculation. Journal of Finance,45(2), 379–395. Domanski, D., & Heath, A. (2007). Financial investors and commodity markets. BIS Quarterly Review,53–67. Etienne, X. L., Irwin, S. H., & Garcia, P. (2014). Bubbles in food commodity markets: Four decades of evidence. Journal of International Money and Finance,42, 129–155. Fishe, R. P., & Smith, A. (2019). Do speculators drive commodity prices away from supply and demand fundamentals? Journal of Commodity Markets,15, 100078. Grinblatt, M., & Han, B. (2005). Prospect theory, mental accounting, and momentum. Journal of Financial Economics,78(2), 311–339. Hong, H., & Stein, J. C. (1999). A unified theory of underreaction, momentum trading, and overreaction in asset markets. Journal of Finance,54(6), 2143–2184. Irwin, S. H., & Sanders, D. R. (2011). Index funds, financialization, and commodity futures markets. Applied Economic Perspectives and Policy,33(1), 1–31. Irwin, S. H., & Sanders, D. R. (2012). Testing the masters hypothesis in commodity futures markets. Energy Economics,34(1), 256–269. Jegadeesh, N., & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance,48(1), 65–91. Keenan, M. (2020). Advanced positioning, flow and sentiment analysis in commodity markets (2nd ed.). Wiley. BORGARDS AND CZUDAJ 3527 Kim, A. (2015). Does futures speculation destabilize commodity markets? Journal of Futures Markets,35(8), 696–714. Manera, M., Nicolini, M., & Vignati, I. (2013). Financial speculation in energy and agriculture futures markets: A multivariate garch approach. Energy Journal,34(3), 55–81. Maria, G. S., Giovanni, V., & Mattia, V. (2020). Did index trader and swap dealer activity produce a bubble in the agricultural commodity market? African Journal of Business Management, 14(1), 9–24. Masters, W. M., & White, A. (2008). The accidental hunt brothers: How institutional investors are driving up food and energy prices. Technical report. Mayer, H., Rathgeber, A., & Wanner, M. (2017). Financialization of metal markets: Does futures trading influence spot prices and volatility? Resources Policy,53, 300–316. Mellios, C., Six, P., & Lai, A. N. (2016). Dynamic speculation and hedging in commodity futures markets with a stochastic convenience yield. European Journal of Operational Research,250(2), 493–504. Miffre, J., & Brooks, C. (2013). Do long-short speculators destabilize commodity futures markets? International Review of Financial Analysis,30, 230–240. Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). Time series momentum. Journal of Financial Economics,104(2), 228–250. Often, E. M., & Wisen, C. H. (2013). Disaggregated commitment of traders. Journal of Applied Business Research,29(5), 1381–1400. Palazzi, R. B., Figueiredo Pinto, A. C., Klotzle, M. C., & De Oliveira, E. M. (2020). Can we still blame index funds for the price movements in the agricultural commodities market? International Review of Economics and Finance,65,84–93. Robe, M. A., & Roberts, J. S. (2021). Who holds positions in agricultural futures markets. Agricultural and Applied Economics Association 2021 Annual Meeting, August 1-3, Austin, Texas. No. 313995. Sanders, D. R., Irwin, S. H., & Merrin, R. P. (2010). The adequacy of speculation in agricultural futures markets: Too much of a good thing? Applied Economic Perspectives and Policy,32(1), 77–94. Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science,185(4157), 1124–1131. Wimmer, T., Geyer-Klingeberg, J., Hütter, M., Schmid, F., & Rathgeber, A. (2021). The impact of speculation on commodity prices: A meta-granger analysis. Journal of Commodity Markets, 22, 100148. Working, H. (1953). Futures trading and hedging. American Economic Review,43(3), 314–343. Working, H. (1954). Whose markets? Evidence on some aspects of futures trading. Journal of Marketing,19(1), 1–11. Working, H. (1960). Speculation on hedging markets. Food Research Institute Studies,1, 185–220. Working, H. (1962). New concepts concerning futures markets and prices. American Economic Review,62, 432–459. How to cite this article: Borgards, O., & Czudaj, R. L. (2023). Long-short speculator sentiment in agricultural commodity markets. International Journal of Finance & Economics,28(4), 3511–3528. https://doi.org/10.1002/ijfe.2605 3528 BORGARDS AND CZUDAJ