# Analysis of the Impact of COVID-19 Pandemic on the Intraday Efficiency of Agricultural Futures Markets

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Data Description

## 3. Multifractal Detrended Fluctuation Analysis (MF-DFA)

## 4. Empirical Findings

## 5. Conclusions and Discussion

## Author Contributions

## Funding

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Akyildirim, Erdinc, Oguzhan Cepni, Linh Pham, and Gazi Salah Uddin. 2022. How connected is the agricultural commodity market to the news-based investor sentiment? Energy Economics 113: 106174. [Google Scholar] [CrossRef]
- Alam, Md Mahmudul, Haitian Wei, and Abu N. M. Wahid. 2021. COVID-19 outbreak and sectoral performance of the Australian stock market: An event study analysis. Australian Economic Papers 60: 482–95. [Google Scholar] [CrossRef]
- Ali, Haider, Faheem Aslam, and Paulo Ferreira. 2021. Modeling Dynamic Multifractal Efficiency of US Electricity Market. Energies 14: 6145. [Google Scholar] [CrossRef]
- Ali, Jabir, and Kriti Bardhan Gupta. 2011. Efficiency in agricultural commodity futures markets in India: Evidence from cointegration and causality tests. Agricultural Finance Review 71: 162–78. [Google Scholar] [CrossRef]
- Ali, Mohsin, Nafis Alam, and Syed Aun R. Rizvi. 2020. Coronavirus (COVID-19)—An epidemic or pandemic for financial markets. Journal of Behavioral and Experimental Finance 27: 100341. [Google Scholar] [CrossRef]
- Alvarez-Ramirez, Jose, Jesus Alvarez, and Eduardo Rodriguez. 2008. Short-term predictability of crude oil markets: A detrended fluctuation analysis approach. Energy Economics 30: 2645–56. [Google Scholar] [CrossRef]
- Anagnostidis, Panagiotis, Christos Varsakelis, and Christos J Emmanouilides. 2016. Has the 2008 financial crisis affected stock market efficiency? The case of Eurozone. Physica A: Statistical Mechanics and its Applications 447: 116–28. [Google Scholar] [CrossRef]
- Apergis, Emmanuel, and Nicholas Apergis. 2020. Can the COVID-19 pandemic and oil prices drive the US Partisan Conflict Index. Energy Research Letters 1: 13144. [Google Scholar] [CrossRef]
- Ashraf, Badar Nadeem. 2020. Stock markets’ reaction to COVID-19: Cases or fatalities? Research in International Business and Finance 54: 101249. [Google Scholar] [CrossRef]
- Aslam, Faheem, Francisca Nogueiro, Mariana Brasil, Paulo Ferreira, Khurram Shahzad Mughal, Beenish Bashir, and Saima Latif. 2020a. The footprints of COVID-19 on Central Eastern European stock markets: An intraday analysis. Post-Communist Economies 33: 751–69. [Google Scholar] [CrossRef]
- Aslam, Faheem, Ijaz Ali, Fahd Amjad, Haider Ali, and Inza Irfan. 2022a. On the inner dynamics between Fossil fuels and the carbon market: A combination of seasonal-trend decomposition and multifractal cross-correlation analysis. Environmental Science and Pollution Research, 1–19. [Google Scholar] [CrossRef] [PubMed]
- Aslam, Faheem, Paulo Ferreira, Fahd Amjad, and Haider Ali. 2021a. The Efficiency of Sin Stocks: A Multifractal Analysis of Drug Indices. The Singapore Economic Review, 1–22. [Google Scholar] [CrossRef]
- Aslam, Faheem, Paulo Ferreira, Haider Ali, and Ana Ercília José. 2022b. Application of Multifractal Analysis in Estimating the Reaction of Energy Markets to Geopolitical Acts and Threats. Sustainability 14: 5828. [Google Scholar] [CrossRef]
- Aslam, Faheem, Paulo Ferreira, Haider Ali, and Sumera Kauser. 2021b. Herding behavior during the COVID-19 pandemic: A comparison between Asian and European stock markets based on intraday multifractality. Eurasian Economic Review 12: 333–59. [Google Scholar] [CrossRef]
- Aslam, Faheem, Paulo Ferreira, Khurrum Shahzad Mughal, and Beenish Bashir. 2021c. Intraday volatility spillovers among european financial markets during COVID-19. International Journal of Financial Studies 9: 5. [Google Scholar] [CrossRef]
- Aslam, Faheem, Saqib Aziz, Duc Khuong Nguyen, Khurrum S. Mughal, and Maaz Khan. 2020b. On the efficiency of foreign exchange markets in times of the COVID-19 pandemic. Technological Forecasting and Social Change 161: 120261. [Google Scholar] [CrossRef]
- Aslam, Faheem, Tahir Mumtaz Awan, Jabir Hussain Syed, Aisha Kashif, and Mahwish Parveen. 2020c. Sentiments and emotions evoked by news headlines of coronavirus disease (COVID-19) outbreak. Humanities and Social Sciences Communications 7: 23. [Google Scholar] [CrossRef]
- Aslam, Faheem, Yasir Tariq Mohmand, Paulo Ferreira, Bilal Ahmed Memon, Maaz Khan, and Mrestyal Khan. 2020d. Network analysis of global stock markets at the beginning of the coronavirus disease (COVID-19) outbreak. Borsa Istanbul Review 20: S49–S61. [Google Scholar] [CrossRef]
- Barrafrem, Kinga, Daniel Västfjäll, and Gustav Tinghög. 2020. Financial well-being, COVID-19, and the financial better-than-average-effect. Journal of Behavioral and Experimental Finance 28: 100410. [Google Scholar] [CrossRef]
- Barro, Robert J., José F. Ursúa, and Joanna Weng. 2020. The Coronavirus and the Great Influenza Pandemic: Lessons from the “Spanish Flu” for the Coronavirus’s Potential Effects on Mortality and Economic Activity. Munich: Munich Society for the Promotion of Economic Research—CESifo GmbH. [Google Scholar]
- Cajueiro, Daniel O., Periklis Gogas, and Benjamin M. Tabak. 2009. Does financial market liberalization increase the degree of market efficiency? The case of the Athens stock exchange. International Review of Financial Analysis 18: 50–57. [Google Scholar] [CrossRef]
- Cao, Guangxi, and Wei Xu. 2016. Nonlinear structure analysis of carbon and energy markets with MFDCCA based on maximum overlap wavelet transform. Physica A: Statistical Mechanics and Its Applications 444: 505–23. [Google Scholar] [CrossRef]
- Caraiani, Petre. 2012. Evidence of multifractality from emerging European stock markets. PLoS ONE 7: e40693. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Chen, Jilong, Liao Xu, and Hao Xu. 2022. The impact of COVID-19 on commodity options market: Evidence from China. Economic Modelling 116: 105998. [Google Scholar] [CrossRef] [PubMed]
- Consuegra, Meliyara, and Javier Garcia-Verdugo. 2017. Measuring the functional efficiency of agricultural futures markets. Australian Journal of Agricultural and Resource Economics 61: 232–46. [Google Scholar] [CrossRef] [Green Version]
- Corbet, Shaen, Charles Larkin, and Brian Lucey. 2020. The contagion effects of the COVID-19 pandemic: Evidence from gold and cryptocurrencies. Finance Research Letters 35: 101554. [Google Scholar] [CrossRef]
- Czech, Katarzyna, Michał Wielechowski, Pavel Kotyza, Irena Benešová, and Adriana Laputková. 2020. Shaking stability: COVID-19 impact on the Visegrad Group countries’ financial markets. Sustainability 12: 6282. [Google Scholar] [CrossRef]
- Devpura, Neluka, and Paresh Kumar Narayan. 2020. Hourly oil price volatility: The role of COVID-19. Energy Research Letters 1: 13683. [Google Scholar] [CrossRef]
- Domino, Krzysztof. 2011. The use of the Hurst exponent to predict changes in trends on the Warsaw Stock Exchange. Physica A: Statistical Mechanics and its Applications 390: 98–109. [Google Scholar] [CrossRef]
- Drożdż, Stanisław, Jarosław Kwapień, Paweł Oświȩcimka, and Rafał Rak. 2010. The foreign exchange market: Return distributions, multifractality, anomalous multifractality and the Epps effect. New Journal of Physics 12: 105003. [Google Scholar] [CrossRef]
- Drożdż, Stanisław, Paweł Oświȩcimka, Andrzej Kulig, Jarosław Kwapień, Katarzyna Bazarnik, Iwona Grabska-Gradzińska, Jan Rybicki, and Marek Stanuszek. 2016. Quantifying origin and character of long-range correlations in narrative texts. Information Sciences 331: 32–44. [Google Scholar] [CrossRef]
- Fama, Eugene F. 1970. Efficient market hypothesis: A review of theory and empirical work. Journal of Finance 25: 28–30. [Google Scholar] [CrossRef]
- Farjah, Ebrahim. 2019. Proposing an efficient wind forecasting agent using adaptive MFDFA. Journal of Power Technologies 99: 152. [Google Scholar]
- Garcin, Matthieu, Jules Klein, and Sana Laaribi. 2020. Estimation of time-varying kernel densities and chronology of the impact of COVID-19 on financial markets. arXiv arXiv:2007.09043. [Google Scholar]
- Goodell, John W. 2020. COVID-19 and finance: Agendas for future research. Finance Research Letters 35: 101512. [Google Scholar] [CrossRef]
- Heng, Yan, Marisa Zansler, and Lisa House. 2020. Orange Juice Consumers Response to the COVID-19 Outbreak: FE1082, 06/2020. EDIS 2020: 4. [Google Scholar]
- Ihlen, Espen A. F., and Beatrix Vereijken. 2013. Multifractal formalisms of human behavior. Human Movement Science 32: 633–51. [Google Scholar] [CrossRef]
- Jafari, G. Reza, Pouria Pedram, and Leila Hedayatifar. 2007. Long-range correlation and multifractality in Bach’s inventions pitches. Journal of Statistical Mechanics: Theory and Experiment 2007: P04012. [Google Scholar] [CrossRef] [Green Version]
- Kantelhardt, Jan W., Stephan A. Zschiegner, Eva Koscielny-Bunde, Shlomo Havlin, Armin Bunde, and H. Eugene Stanley. 2002. Multifractal detrended fluctuation analysis of nonstationary time series. Physica A: Statistical Mechanics and its Applications 316: 87–114. [Google Scholar] [CrossRef] [Green Version]
- Kelty-Stephen, Damian G. 2017. Threading a multifractal social psychology through within-organism coordination to within-group interactions: A tale of coordination in three acts. Chaos, Solitons & Fractals 104: 363–70. [Google Scholar]
- Kim, Hongseok, Gabjin Oh, and Seunghwan Kim. 2011. Multifractal analysis of the Korean agricultural market. Physica A: Statistical Mechanics and its Applications 390: 4286–92. [Google Scholar]
- Kim, Irene, and Mohan Venkatachalam. 2011. Are sin stocks paying the price for accounting sins? Journal of Accounting, Auditing & Finance 26: 415–42. [Google Scholar]
- Kumar, Sunil, and Nivedita Deo. 2009. Multifractal properties of the Indian financial market. Physica A: Statistical Mechanics and Its Applications 388: 1593–602. [Google Scholar] [CrossRef]
- Laib, Mohamed, Jean Golay, Luciano Telesca, and Mikhail Kanevski. 2018. Multifractal analysis of the time series of daily means of wind speed in complex regions. Chaos, Solitons & Fractals 109: 118–27. [Google Scholar]
- Laing, Timothy. 2020. The economic impact of the Coronavirus 2019 (COVID-2019): Implications for the mining industry. The Extractive Industries and Society 7: 580–82. [Google Scholar] [CrossRef]
- Li, Zhihui, and Xinsheng Lu. 2011. Multifractal analysis of China’s agricultural commodity futures markets. Energy Procedia 5: 1920–26. [Google Scholar] [CrossRef] [Green Version]
- Lu, Ran, and Hongjun Zeng. 2022. VIX and major agricultural future markets: Dynamic linkage and time-frequency relations around the COVID-19 outbreak. Studies in Economics and Finance. ahead-of-print. [Google Scholar] [CrossRef]
- Maier, Benjamin F., and Dirk Brockmann. 2020. Effective containment explains subexponential growth in recent confirmed COVID-19 cases in China. Science 368: 742–46. [Google Scholar] [CrossRef]
- Makowiec, Danuta, Aleksandra Dudkowska, Rafał Gałaska, and Andrzej Rynkiewicz. 2009. Multifractal estimates of monofractality in RR-heart series in power spectrum ranges. Physica A: Statistical Mechanics and Its Applications 388: 3486–502. [Google Scholar] [CrossRef]
- Mandelbrot, Benoit B. 1971. When can price be arbitraged efficiently? A limit to the validity of the random walk and martingale models. The Review of Economics and Statistics 53: 225–36. [Google Scholar] [CrossRef]
- Mandelbrot, Benoit B. 1982. The Fractal Geometry of Nature. New York: WH Freeman, vol. 1. [Google Scholar]
- Mandelbrot, Benoit B. 1997. The variation of the prices of cotton, wheat, and railroad stocks, and of some financial rates. In Fractals and Scaling in Finance. Berlin/Heidelberg: Springer, pp. 419–43. [Google Scholar]
- Mandelbrot, Benoit B., and James R. Wallis. 1969. Robustness of the rescaled range R/S in the measurement of noncyclic long run statistical dependence. Water Resources Research 5: 967–88. [Google Scholar] [CrossRef]
- Mandelbrot, Benoit. 1967. The variation of some other speculative prices. The Journal of Business 40: 393–413. [Google Scholar] [CrossRef]
- McKibbin, Warwick, and David Vines. 2020. Global macroeconomic cooperation in response to the COVID-19 pandemic: A roadmap for the G20 and the IMF. Oxford Review of Economic Policy 36: S297–S337. [Google Scholar] [CrossRef]
- Ming, Wen, Zhengqing Zhou, Hongshan Ai, Huimin Bi, and Yuan Zhong. 2020. COVID-19 and air quality: Evidence from China. Emerging Markets Finance and Trade 56: 2422–42. [Google Scholar] [CrossRef]
- Mishra, Sibanjan. 2019. Testing Martingale Hypothesis Using Variance Ratio Tests: Evidence from High-frequency Data of NCDEX Soya Bean Futures. Global Business Review 20: 1407–22. [Google Scholar] [CrossRef]
- Mnif, Emna, Anis Jarboui, and Khaireddine Mouakhar. 2020. How the cryptocurrency market has performed during COVID 19? A multifractal analysis. Finance Research Letters 36: 101647. [Google Scholar] [CrossRef] [PubMed]
- Muzy, Jean-François, Emmanuel Bacry, Rachel Baile, and Philippe Poggi. 2008. Uncovering latent singularities from multifractal scaling laws in mixed asymptotic regime. Application to turbulence. EPL (Europhysics Letters) 82: 60007. [Google Scholar] [CrossRef] [Green Version]
- Nagy, Zoltan, Peter Mukli, Peter Herman, and Andras Eke. 2017. Decomposing multifractal crossovers. Frontiers in Physiology 8: 533. [Google Scholar] [CrossRef]
- Nicola, Maria, Zaid Alsafi, Catrin Sohrabi, Ahmed Kerwan, Ahmed Al-Jabir, Christos Iosifidis, Maliha Agha, and Riaz Agha. 2020. The socio-economic implications of the coronavirus pandemic (COVID-19): A review. International Journal of Surgery 78: 185–93. [Google Scholar] [CrossRef]
- Njindan Iyke, Bernard. 2020. The disease outbreak channel of exchange rate return predictability: Evidence from COVID-19. Emerging Markets Finance and Trade 56: 2277–97. [Google Scholar] [CrossRef]
- Oh, Gabjin, Seunghwan Kim, and Cheoljun Eom. 2010. Multifractal analysis of Korean stock market. Journal of the Korean Physical Society 56: 982–85. [Google Scholar] [CrossRef]
- Peng, Chung-Kang, Sergey V. Buldyrev, Shlomo Havlin, Michael Simons, H. Eugene Stanley, and Ary L. Goldberger. 1994. Mosaic organization of DNA nucleotides. Physical Review E 49: 1685. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Pleşoianu, Anita, Alexandru Todea, and Răzvan Căpuşan. 2012. The informational efficiency of the Romanian stock market: Evidence from fractal analysis. Procedia Economics and Finance 3: 111–18. [Google Scholar] [CrossRef] [Green Version]
- Podobnik, Boris, and H. Eugene Stanley. 2008. Detrended cross-correlation analysis: A new method for analyzing two nonstationary time series. Physical Review Letters 100: 084102. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Rizvi, Syed Aun R., Ginanjar Dewandaru, Obiyathulla I Bacha, and Mansur Masih. 2014. An analysis of stock market efficiency: Developed vs Islamic stock markets using MF-DFA. Physica A: Statistical Mechanics and Its Applications 407: 86–99. [Google Scholar] [CrossRef]
- Rosas, Alexandre, Edvaldo Nogueira Jr., and José F Fontanari. 2002. Multifractal analysis of DNA walks and trails. Physical Review E 66: 061906. [Google Scholar] [CrossRef] [Green Version]
- Sadowski, Adam, Zbigniew Galar, Robert Walasek, Grzegorz Zimon, and Per Engelseth. 2021. Big data insight on global mobility during the COVID-19 pandemic lockdown. Journal of Big Data 8: 78. [Google Scholar] [CrossRef]
- Sifat, Imtiaz, Abdul Ghafoor, and Abdollah Ah Mand. 2021. The COVID-19 pandemic and speculation in energy, precious metals, and agricultural futures. Journal of Behavioral and Experimental Finance 30: 100498. [Google Scholar] [CrossRef]
- Sipra, Hassaan, Faheem Aslam, Jabir Hussain Syed, and Tahir Mumtaz Awan. 2021. Investigating the Implications of COVID-19 on PM2. 5 in Pakistan. Aerosol and Air Quality Research 21: 200459. [Google Scholar] [CrossRef]
- Stanley, H. Eugene, and Paul Meakin. 1988. Multifractal phenomena in physics and chemistry. Nature 335: 405–09. [Google Scholar] [CrossRef]
- Stosic, Tatijana, Salman Abarghouei Nejad, and Borko Stosic. 2020. Multifractal analysis of Brazilian agricultural market. Fractals 28: 2050076. [Google Scholar] [CrossRef]
- Štreimikienė, Dalia, Tomas Baležentis, Artiom Volkov, Erika Ribašauskienė, Mangirdas Morkūnas, and Agnė Žičkienė. 2022. Negative effects of covid-19 pandemic on agriculture: Systematic literature review in the frameworks of vulnerability, resilience and risks involved. Economic Research-Ekonomska Istraživanja 35: 529–45. [Google Scholar] [CrossRef]
- Subramaniam, Arvind R., Ilya A. Gruzberg, and Andreas W. W. Ludwig. 2008. Boundary criticality and multifractality at the two-dimensional spin quantum Hall transition. Physical Review B 78: 245105. [Google Scholar] [CrossRef] [Green Version]
- Telesca, Luciano, Vincenzo Lapenna, and Maria Macchiato. 2005a. Multifractal fluctuations in seismic interspike series. Physica A: Statistical Mechanics and Its Applications 354: 629–40. [Google Scholar] [CrossRef] [Green Version]
- Telesca, Luciano, Vincenzo Lapenna, and Maria Macchiato. 2005b. Multifractal fluctuations in earthquake-related geoelectrical signals. New Journal of Physics 7: 214. [Google Scholar] [CrossRef]
- Topcu, Mert, and Omer Serkan Gulal. 2020. The impact of COVID-19 on emerging stock markets. Finance Research Letters 36: 101691. [Google Scholar] [CrossRef]
- Udovichenko, Vladimir, and Peter Strizhak. 2002. Multifractal properties of copper sulfide film formed in self-organizing chemical system. Theoretical and Experimental Chemistry 38: 259–62. [Google Scholar] [CrossRef]
- Umar, Zaghum, Francisco Jareño, and Ana Escribano. 2021. Agricultural commodity markets and oil prices: An analysis of the dynamic return and volatility connectedness. Resources Policy 73: 102147. [Google Scholar] [CrossRef]
- Wang, Jian, Wei Shao, and Junseok Kim. 2020a. Analysis of the impact of COVID-19 on the correlations between crude oil and agricultural futures. Chaos, Solitons & Fractals 136: 109896. [Google Scholar]
- Wang, Yating, Donghao Zhang, Xiaoquan Wang, and Qiuyao Fu. 2020b. How does COVID-19 affect China’s insurance market? Emerging Markets Finance and Trade 56: 2350–62. [Google Scholar] [CrossRef]
- Wang, Yudong, and Li Liu. 2010. Is WTI crude oil market becoming weakly efficient over time?: New evidence from multiscale analysis based on detrended fluctuation analysis. Energy Economics 32: 987–92. [Google Scholar] [CrossRef]
- Zhang, Dayong, Min Hu, and Qiang Ji. 2020. Financial markets under the global pandemic of COVID-19. Finance Research Letters 36: 101528. [Google Scholar] [CrossRef] [PubMed]
- Zhang, Yue-Jun, and Shu-Jiao Ma. 2021. Exploring the dynamic price discovery, risk transfer and spillover among INE, WTI and Brent crude oil futures markets: Evidence from the high-frequency data. International Journal of Finance & Economics 26: 2414–35. [Google Scholar]

**Figure 3.**MFDFA results for cocoa. (Panel

**A**) Fluctuation functions for q = [−30,0,+30]; (Panel

**B**) Generalized Hurst exponent depending on q; (Panel

**C**) Mass exponent t(q); (Panel

**D**) Multifractal spectrum. On the left is the information for the period before the declaration of the COVID-19 pandemic and on the right is the information for the period after the declaration of the COVID-19 pandemic.

**Figure 4.**MFDFA results for coffee, cotton, and orange Juice. (Panel

**A**) Fluctuation functions for q = [−30,0,+30]; (Panel

**B**) Generalized Hurst exponent depending on q; (Panel

**C**) Mass exponent t(q); (Panel

**D**) Multifractal spectrum. On the left is the information for the period before the declaration of theCOVID-19 pandemic and on the right is the information for the period after the declaration of the COVID-19 pandemic.

**Figure 5.**MFDFA results for soybean and sugar. (Panel

**A**) Fluctuation functions for q = [−30,0,+30]; (Panel

**B**) Generalized Hurst exponent depending on q; (Panel

**C**) Mass exponent t(q); (Panel

**D**) Multifractal spectrum. On the left is the information for the period before the declaration of the COVID-19 pandemic and on the right is the information for the period after the declaration of the COVID-19 pandemic.

**Figure 6.**The multifractal spectra for agricultural commodities before and after the declaration of the pandemic.

S. No. | Commodity | Number Obs. Before Pandemic (1 August 2019 to 10 March 2020) | Number Obs. After Pandemic (11 March 2020 to 25 September 2020) |
---|---|---|---|

1 | US Cocoa | 8986 | 4704 |

2 | US Coffee | 5385 | 4999 |

3 | US Cotton | 10,143 | 9567 |

4 | US Orange Juice | 3460 | 3314 |

5 | US Soybean | 9414 | 8961 |

6 | London Sugar | 5421 | 5014 |

Q | Before COVID-19 Outbreak | After COVID-19 Outbreak | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|

Cocoa | Coffee | Cotton | Orange Juice | Soybean | Sugar | Cocoa | Coffee | Cotton | Orange Juice | Soybean | Sugar | |

−30 | 0.7302 | 0.6568 | 0.6588 | 0.7765 | 0.7007 | 0.6219 | 0.6424 | 0.6658 | 0.7878 | 0.7983 | 0.7559 | 0.7346 |

−29 | 0.7292 | 0.6556 | 0.6578 | 0.7754 | 0.6996 | 0.6208 | 0.6414 | 0.6648 | 0.7868 | 0.7971 | 0.7547 | 0.7336 |

−28 | 0.7281 | 0.6544 | 0.6567 | 0.7741 | 0.6985 | 0.6196 | 0.6403 | 0.6636 | 0.7857 | 0.7959 | 0.7535 | 0.7324 |

−27 | 0.7269 | 0.653 | 0.6555 | 0.7728 | 0.6973 | 0.6184 | 0.6391 | 0.6624 | 0.7846 | 0.7945 | 0.7522 | 0.7312 |

−26 | 0.7256 | 0.6516 | 0.6543 | 0.7714 | 0.696 | 0.617 | 0.6379 | 0.661 | 0.7833 | 0.7931 | 0.7507 | 0.7299 |

−25 | 0.7243 | 0.6501 | 0.6529 | 0.7698 | 0.6946 | 0.6156 | 0.6366 | 0.6596 | 0.782 | 0.7915 | 0.7492 | 0.7285 |

−24 | 0.7228 | 0.6485 | 0.6515 | 0.7682 | 0.6931 | 0.6141 | 0.6352 | 0.6581 | 0.7805 | 0.7898 | 0.7475 | 0.7269 |

−23 | 0.7212 | 0.6467 | 0.6499 | 0.7664 | 0.6915 | 0.6124 | 0.6337 | 0.6565 | 0.779 | 0.788 | 0.7457 | 0.7253 |

−22 | 0.7195 | 0.6448 | 0.6482 | 0.7644 | 0.6898 | 0.6106 | 0.6321 | 0.6547 | 0.7773 | 0.786 | 0.7437 | 0.7235 |

−21 | 0.7176 | 0.6427 | 0.6464 | 0.7623 | 0.688 | 0.6086 | 0.6303 | 0.6528 | 0.7754 | 0.7838 | 0.7416 | 0.7216 |

−20 | 0.7155 | 0.6404 | 0.6444 | 0.76 | 0.6859 | 0.6065 | 0.6284 | 0.6507 | 0.7734 | 0.7814 | 0.7392 | 0.7195 |

−19 | 0.7133 | 0.6379 | 0.6422 | 0.7574 | 0.6837 | 0.6042 | 0.6264 | 0.6484 | 0.7713 | 0.7787 | 0.7366 | 0.7172 |

−18 | 0.7108 | 0.6352 | 0.6399 | 0.7546 | 0.6813 | 0.6016 | 0.6241 | 0.6459 | 0.7688 | 0.7758 | 0.7338 | 0.7147 |

−17 | 0.7081 | 0.6323 | 0.6373 | 0.7515 | 0.6786 | 0.5988 | 0.6217 | 0.6432 | 0.7662 | 0.7725 | 0.7307 | 0.7119 |

−16 | 0.7051 | 0.629 | 0.6344 | 0.748 | 0.6757 | 0.5957 | 0.6191 | 0.6401 | 0.7632 | 0.7688 | 0.7272 | 0.7089 |

−15 | 0.7017 | 0.6254 | 0.6312 | 0.7441 | 0.6724 | 0.5923 | 0.6162 | 0.6368 | 0.7599 | 0.7646 | 0.7233 | 0.7055 |

−14 | 0.6979 | 0.6214 | 0.6277 | 0.7398 | 0.6688 | 0.5886 | 0.613 | 0.6331 | 0.7562 | 0.7599 | 0.719 | 0.7017 |

−13 | 0.6937 | 0.6169 | 0.6237 | 0.7348 | 0.6648 | 0.5843 | 0.6095 | 0.629 | 0.752 | 0.7545 | 0.7141 | 0.6975 |

−12 | 0.6889 | 0.612 | 0.6193 | 0.7292 | 0.6603 | 0.5796 | 0.6057 | 0.6244 | 0.7472 | 0.7483 | 0.7085 | 0.6927 |

−11 | 0.6834 | 0.6064 | 0.6142 | 0.7227 | 0.6552 | 0.5743 | 0.6015 | 0.6193 | 0.7417 | 0.7411 | 0.7022 | 0.6873 |

−10 | 0.6771 | 0.6002 | 0.6085 | 0.7152 | 0.6494 | 0.5682 | 0.597 | 0.6135 | 0.7352 | 0.7327 | 0.6949 | 0.6811 |

−9 | 0.6699 | 0.5933 | 0.602 | 0.7066 | 0.6429 | 0.5615 | 0.592 | 0.607 | 0.7277 | 0.7226 | 0.6865 | 0.674 |

−8 | 0.6617 | 0.5857 | 0.5946 | 0.6964 | 0.6353 | 0.5538 | 0.5867 | 0.5996 | 0.7188 | 0.7107 | 0.6768 | 0.6659 |

−7 | 0.6521 | 0.5773 | 0.586 | 0.6844 | 0.6267 | 0.5453 | 0.581 | 0.5914 | 0.7081 | 0.6964 | 0.6655 | 0.6566 |

−6 | 0.6412 | 0.5682 | 0.576 | 0.6702 | 0.6169 | 0.5359 | 0.5753 | 0.5823 | 0.6953 | 0.6792 | 0.6524 | 0.6461 |

−5 | 0.629 | 0.5587 | 0.5646 | 0.6536 | 0.6056 | 0.5262 | 0.5697 | 0.5724 | 0.6797 | 0.6587 | 0.6373 | 0.6343 |

−4 | 0.6158 | 0.549 | 0.5517 | 0.6343 | 0.5929 | 0.5169 | 0.5649 | 0.562 | 0.661 | 0.635 | 0.6203 | 0.6217 |

−3 | 0.6021 | 0.5393 | 0.5373 | 0.6122 | 0.5785 | 0.509 | 0.5611 | 0.5517 | 0.6389 | 0.6098 | 0.6018 | 0.609 |

−2 | 0.5884 | 0.5298 | 0.5215 | 0.587 | 0.5624 | 0.5032 | 0.5586 | 0.5425 | 0.6137 | 0.5868 | 0.5826 | 0.5968 |

−1 | 0.5748 | 0.5203 | 0.5032 | 0.5572 | 0.5443 | 0.4982 | 0.5569 | 0.5346 | 0.5862 | 0.5713 | 0.5633 | 0.5852 |

0 | 0.5612 | 0.51 | 0.4807 | 0.5195 | 0.5242 | 0.491 | 0.5553 | 0.5275 | 0.5577 | 0.5694 | 0.5446 | 0.5736 |

1 | 0.5474 | 0.4976 | 0.4525 | 0.4695 | 0.5027 | 0.4796 | 0.5531 | 0.5203 | 0.5302 | 0.5842 | 0.528 | 0.5611 |

2 | 0.5332 | 0.4816 | 0.4194 | 0.4059 | 0.4804 | 0.4645 | 0.55 | 0.5111 | 0.5049 | 0.6079 | 0.5144 | 0.5472 |

3 | 0.519 | 0.462 | 0.3846 | 0.3398 | 0.4585 | 0.4468 | 0.5461 | 0.4988 | 0.4816 | 0.6268 | 0.5039 | 0.5325 |

4 | 0.5051 | 0.4408 | 0.3525 | 0.2855 | 0.438 | 0.4283 | 0.5414 | 0.484 | 0.4603 | 0.636 | 0.4952 | 0.5177 |

5 | 0.4917 | 0.4203 | 0.3255 | 0.2458 | 0.4198 | 0.4106 | 0.5361 | 0.4686 | 0.4412 | 0.6377 | 0.4874 | 0.5038 |

6 | 0.4793 | 0.4022 | 0.3036 | 0.2173 | 0.404 | 0.3949 | 0.5307 | 0.4543 | 0.4246 | 0.6352 | 0.4799 | 0.4913 |

7 | 0.4681 | 0.3867 | 0.2862 | 0.1963 | 0.3907 | 0.3814 | 0.5255 | 0.4417 | 0.4106 | 0.6306 | 0.4728 | 0.4802 |

8 | 0.4582 | 0.3738 | 0.2722 | 0.1804 | 0.3794 | 0.3698 | 0.5205 | 0.4308 | 0.3987 | 0.6253 | 0.4661 | 0.4706 |

9 | 0.4495 | 0.363 | 0.2609 | 0.1679 | 0.3698 | 0.3601 | 0.5158 | 0.4215 | 0.3886 | 0.6199 | 0.46 | 0.4622 |

10 | 0.4418 | 0.354 | 0.2515 | 0.1578 | 0.3616 | 0.3517 | 0.5115 | 0.4135 | 0.38 | 0.6147 | 0.4545 | 0.455 |

11 | 0.4351 | 0.3463 | 0.2437 | 0.1495 | 0.3546 | 0.3446 | 0.5076 | 0.4066 | 0.3726 | 0.6099 | 0.4495 | 0.4488 |

12 | 0.4293 | 0.3397 | 0.2371 | 0.1426 | 0.3486 | 0.3384 | 0.5041 | 0.4006 | 0.3662 | 0.6055 | 0.445 | 0.4433 |

13 | 0.4241 | 0.3341 | 0.2314 | 0.1367 | 0.3433 | 0.3331 | 0.5008 | 0.3954 | 0.3607 | 0.6015 | 0.441 | 0.4385 |

14 | 0.4196 | 0.3291 | 0.2265 | 0.1316 | 0.3386 | 0.3284 | 0.4979 | 0.3909 | 0.3558 | 0.5978 | 0.4374 | 0.4343 |

15 | 0.4155 | 0.3248 | 0.2222 | 0.1271 | 0.3345 | 0.3242 | 0.4951 | 0.3868 | 0.3515 | 0.5945 | 0.4341 | 0.4305 |

16 | 0.4119 | 0.3209 | 0.2184 | 0.1232 | 0.3309 | 0.3205 | 0.4926 | 0.3832 | 0.3477 | 0.5914 | 0.4312 | 0.4272 |

17 | 0.4086 | 0.3175 | 0.215 | 0.1197 | 0.3276 | 0.3172 | 0.4903 | 0.38 | 0.3443 | 0.5887 | 0.4285 | 0.4242 |

18 | 0.4057 | 0.3145 | 0.212 | 0.1166 | 0.3246 | 0.3143 | 0.4882 | 0.3771 | 0.3412 | 0.5862 | 0.4261 | 0.4215 |

19 | 0.4031 | 0.3117 | 0.2093 | 0.1138 | 0.3219 | 0.3116 | 0.4863 | 0.3745 | 0.3384 | 0.5839 | 0.4238 | 0.419 |

20 | 0.4006 | 0.3092 | 0.2069 | 0.1113 | 0.3195 | 0.3092 | 0.4845 | 0.3721 | 0.3359 | 0.5817 | 0.4218 | 0.4168 |

21 | 0.3984 | 0.3069 | 0.2047 | 0.109 | 0.3173 | 0.307 | 0.4828 | 0.3699 | 0.3336 | 0.5798 | 0.4199 | 0.4147 |

22 | 0.3964 | 0.3049 | 0.2026 | 0.1069 | 0.3152 | 0.305 | 0.4812 | 0.3679 | 0.3315 | 0.578 | 0.4181 | 0.4129 |

23 | 0.3945 | 0.303 | 0.2008 | 0.105 | 0.3134 | 0.3031 | 0.4798 | 0.3661 | 0.3296 | 0.5763 | 0.4165 | 0.4111 |

24 | 0.3928 | 0.3012 | 0.1991 | 0.1032 | 0.3116 | 0.3014 | 0.4784 | 0.3644 | 0.3278 | 0.5748 | 0.415 | 0.4096 |

25 | 0.3912 | 0.2996 | 0.1975 | 0.1016 | 0.3101 | 0.2998 | 0.4771 | 0.3629 | 0.3262 | 0.5733 | 0.4137 | 0.4081 |

26 | 0.3897 | 0.2981 | 0.196 | 0.1001 | 0.3086 | 0.2984 | 0.476 | 0.3615 | 0.3247 | 0.572 | 0.4124 | 0.4067 |

27 | 0.3884 | 0.2967 | 0.1947 | 0.0987 | 0.3072 | 0.297 | 0.4748 | 0.3601 | 0.3233 | 0.5707 | 0.4112 | 0.4054 |

28 | 0.3871 | 0.2954 | 0.1934 | 0.0974 | 0.3059 | 0.2957 | 0.4738 | 0.3589 | 0.322 | 0.5696 | 0.41 | 0.4043 |

29 | 0.3859 | 0.2942 | 0.1922 | 0.0962 | 0.3047 | 0.2946 | 0.4728 | 0.3577 | 0.3207 | 0.5685 | 0.409 | 0.4031 |

30 | 0.3848 | 0.2931 | 0.1911 | 0.0951 | 0.3036 | 0.2935 | 0.4719 | 0.3566 | 0.3196 | 0.5674 | 0.408 | 0.4021 |

Delta H | 0.3454 | 0.3637 | 0.4677 | 0.6814 | 0.3971 | 0.3284 | 0.1705 | 0.3092 | 0.4682 | 0.2309 | 0.3479 | 0.3325 |

Delta α | 0.4087 | 0.4324 | 0.531 | 0.75 | 0.4624 | 0.3942 | 0.2291 | 0.3751 | 0.5341 | 0.2982 | 0.4136 | 0.3977 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Aslam, F.; Ferreira, P.; Ali, H.
Analysis of the Impact of COVID-19 Pandemic on the Intraday Efficiency of Agricultural Futures Markets. *J. Risk Financial Manag.* **2022**, *15*, 607.
https://doi.org/10.3390/jrfm15120607

**AMA Style**

Aslam F, Ferreira P, Ali H.
Analysis of the Impact of COVID-19 Pandemic on the Intraday Efficiency of Agricultural Futures Markets. *Journal of Risk and Financial Management*. 2022; 15(12):607.
https://doi.org/10.3390/jrfm15120607

**Chicago/Turabian Style**

Aslam, Faheem, Paulo Ferreira, and Haider Ali.
2022. "Analysis of the Impact of COVID-19 Pandemic on the Intraday Efficiency of Agricultural Futures Markets" *Journal of Risk and Financial Management* 15, no. 12: 607.
https://doi.org/10.3390/jrfm15120607