Impact of Digital Communication on Wine Market Liquidity

Information matters: how does digital communication and social media affect liquidity and price behaviour in the wine market Pawe Oleksy coauthored with Marcin Czupryna CUE, Departmentof financial markets 11th Annual AAWE Conference Padova 2017

Motivations and research questions Motivations and research questions Motivations: Alvin Toffler: The illiterate of the 21st century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn Fast expansion of digital communication and social media in recent years has created new possibilities and challenges for information dissemination on products or services of all markets and industries, including wine Research on financial assets reveals a positive impact of information disseminated through digital communication and social media platforms on spread tightening, trading volume increase, improvement of market efficiency and diminishing information asymmetry between market participants (e.g. Blankespoor, Miller and White 2014; Ranko et al. 2015; Azur and Lo 2016) Wine economics research examines the social media adaption in the wine industry and its influence on wine consumers behaviour with apparent emphasis on marketing- and sociology-related concerns (e.g. Fiore et al. 2016, Cuomo et al. 2016, Hoffman et al. 2014) Due to the enormous increase in the amount of information available for market participants the text mining tools may be useful for analysing wine price behaviour and market liquidity Research questions: How does digital communication and social media affect liquidity and price behaviour in the wine market? Are there any textual regularities in wine market news and commentaries? Do texts disseminated via digital communication channels and social media platforms have any predictive value for wine prices?

Research Research project project Research objectives: to examine how the information disseminated through digital communication and social media platforms influences fine wine price behaviour and market liquidity to verify some textual regularities in disclosed phrases and messages and their links to changes in wine prices and liquidity Hypotheses: H1: Information dissemination through digital communication channels (including social media) improves wine pricing efficiency and wine market liquidity H2: Information style depends on the sales channel (wine exchange, auction house, wholesaler, retailer) Data: Liv-ex Blog (weekly Market insights - Talking trade , www.liv-ex.com, also disseminated via emails) Liv-ex order book (daily top 200 spreads (i.e.. bids, offers with volumes) of quoted wines; time period: XI 2016 VI 2017, SIB contracts, Liv-ex 50 index Methodology: Text mining tools (pre-processing (incl. stemming, white space elimination, lower case conversion), tagging (classification as noun, verb, adjective, etc.), POS analysis, TDM) Liquidity measures (volume-based, transaction cost-based, etc.) Correlation analysis

Digital communication and social media use by selected wine Digital communication and social media use by selected wine market participants market participants (as at 12.12.2016) (as at 12.12.2016) Liv-ex Wine-searcher Wine Spectator Christie's Wine YES NO YES* NO Blog YES YES YES YES** Likes: 4 539 Followers: 4 310 Likes: 105 037 Followers: 90 839 Followers: 250 493 Likes: 170 932 Followers: 167 534 Facebook Likes: 260 511 YES YES YES YES Tweets: 6 902 Following: 795 Followers: 7 373 Likes: 2 193 Lists: 2 Tweets 7 346 Following: 1 350 Followers: 15,4 thous. Likes: 286 Tweets 15,5 thous. Following: 25 Followers: 224 thous. Likes: 1 137 Lists 5 Tweets: 1 245 Following: 515 Followers: 3406 Likes: 889 Tweeter YES NO YES YES** LinkedIn Followers: 1 222 Followers: 2 645 Followers: 62 820 NO YES YES YES Posts: 285 Followers: 2 821 Following: 1 110 Posts: 649 Followers: 134 thous. Following: 29 Posts: 391 Followers: 3732 Following: 460 Instagram NO NO YES YES** Followers: 7 178 Following: 2 Followers: 39 830 Following: 267 Pinterest

TDM: Top 10 TDM: Top 10 words words occuring occuring in in Liv Liv- -ex ex blogs blogs Frequency (% of all words) 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% trade week wine bordeaux Stemmed words valu livex activ top vintag volum 0 50 100 150 200 250 300 Frequency (quantity)

TDM: Wine TDM: Wine critics critics in in Liv Liv- -ex ex blogs blogs Frequency of wine critics Parker's team vs. other wine critics in Liv-ex blog Robert Parker Neal Martin James Suckling 16% Decanter Panel Wine critics Antonio Galloni 74% Chris Kissack Gavin Quinney Jean-Marc Quarin Jancis Robinson James Molesworth Parker's team Other Jeff Leve Tim Atkin 0 5 10 15 20 25 30 35 Frequency (quantity)

TDM TDM wine wine regions regions frequency frequency in in text text corpora corpora 0 20 40 60 80 100 120 140 160 2,22%; 134 Bordeaux 0,56%; 34 Burgundy 0,55%; 33 Champagne 0,40%; 24 Italy/Italian 0,28%; 17 USA/California 0,10%; 6 Rh ne 0,03%; 2 Australia 0,03%; 2 Edinburgh 0,03%; 2 Spain 0,02%; 1 Chinese

Wine Wine liquidity liquidity Unit size Mean Median SD Max Min No. Bids No. Offers ?????? =???? ???? ????+ ???? 6x75 0,0309 0,0305 0,0141 0,0573 0,0048 307 413 12x75 0,0309 0,0313 0,0143 0,0620 0,0049 312 212

Liquidity vs. Part Liquidity vs. Part- -of of- -speech analysis: retrospective correlation perspective speech analysis: No. words % adj. % adj_comp % adv. % adv_comp % noun % verb % foreign % numbers Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value mean_spread_12x75 -0,33 -0,03 -0,16 0,17 -0,33 0,23 -0,14 0,34 0,12 7% 51% 89% 40% 36% 8% 22% 46% 7% -0,36 -0,09 -0,09 0,24 -0,39 0,24 -0,11 0,32 mean_spread_6x75 0,17 5% 36% 62% 64% 19% 3% 20% 55% 8% 0,13 0,08 0,03 0,30 0,09 -0,16 0,21 -0,44 total_asks_12x75 -0,06 48% 76% 68% 86% 10% 64% 39% 27% 1% -0,35 0,14 -0,07 0,04 -0,21 0,03 -0,01 0,67 total_asks_6x75 0,14 6% 44% 44% 71% 83% 25% 86% 97% 41% 0,19 0,17 -0,02 -0,13 0,17 0,11 -0,11 -0,51 total_bids_12x75 -0,37 30% 4% 37% 92% 48% 35% 57% 56% 0% 0,07 0,06 -0,16 0,07 -0,01 -0,12 -0,07 total_bids_6x75 0,14 0,05 70% 45% 74% 80% 39% 71% 97% 51% 69%

Predictive Predictive value prospective correlation perspective value of of Liv Liv- -ex blog: ex blog: No. words % adj. % adj_comp % adv. % adv_comp % noun % verb % foreign % numbers Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value Pears. p-value mean_spread_12x75 -0,56 -0,03 0,21 -0,28 -0,04 -0,08 0,01 -0,06 0,35 86% 0% 27% 13% 82% 67% 94% 75% 6% -0,50 -0,03 0,20 -0,25 -0,07 0,03 0,08 -0,04 0,28 mean_spread_6x75 87% 0% 28% 18% 70% 86% 68% 82% 13% 0,16 -0,17 -0,18 0,33 0,13 0,21 -0,07 0,06 -0,35 total_asks_12x75 34% 39% 33% 6% 49% 25% 69% 75% 5% -0,39 0,09 -0,03 -0,20 0,14 -0,29 0,07 -0,02 0,63 total_asks_6x75 61% 3% 87% 28% 46% 11% 70% 92% 0% 0,28 -0,30 0,06 0,27 -0,15 0,04 0,14 -0,19 -0,49 total_bids_12x75 9% 12% 75% 13% 41% 84% 46% 29% 0% 0,27 -0,08 -0,28 -0,02 -0,10 0,12 0,20 -0,10 -0,30 total_bids_6x75 65% 13% 12% 91% 58% 52% 28% 60% 9%

Further research Further research other digital communication and social media platforms (Twitter, Facebook, etc.) other market participants (wholesalers, retailers, producers, consumers, collectors) other trading platforms (e.g. OTC, auctions) textual differences and emotional component in wine descriptions, wine market news and commentaries advanced text mining tools .

Thank you for your attention! dr Marcin Czupryna dr Pawe Oleksy Cracow University of Economics Department of Financial Markets email: czuprynm@uek.krakow.pl Cracow University of Economics Department of Financial Markets email: oleksyp@uek.krakow.pl