Wha Do Shor Sellers Know? Boehmer Jones & Zhang D I S C U S S I O N B Y A D A M V. R E E D U N C C H A P E L H I L L
Background: Shors are Profable Trades Asquh & Meulbroek (996) Shor sales uncondonally predc lower fuure reurns Asquh Pahak & Rer (2005) Shor sales do predc fuure reurns bu hs effec s concenraed n ceran socks and suaons Boehmer Jones and Zhang (2008) Shor sales predc lower fuure reurns over shor perods
So Why Are They More Profable? News? Asquh Pahak & Rer (2005) Shor sales do predc fuure reurns bu hs effec s concenraed n ceran socks and suaons Karpoff & Lou (2009) Shor sellers denfy fnancal msconduc before s revealed by resaemens or SEC acons Angel Ferr & Chrsophe (2004) Shor-sellng doesn usually ncrease before earnngs announcemens bu when does correcly predcs reurns
Wha s he Queson n Ths Paper? Tradonal queson: Do shor sellers rade around earnngs evens? (e.g. Chrsophe Ferr and Angel (2004) Daske Rchardson and Tuna (2005)) Ths paper s quesons: How much of her profably s comng from earnngs and analys news? Whch shor sellers have he mos profable rades around hese evens? Jons wo oher noable papers: Fox Glosen and Telock (200) Engelberg Reed and Rnggenberg (200) Boh use all news evens. Boh asses he percenage of shors advanage ha comes from news
Wha do hey do? Man expermen s: r k b0 ( b c0d ) Los of perurbaons (Grea Daa!): Reurn wndow {[+] [+5] [+0] [+20]} Dummy Varable Evens {earnngs announcemen analys recommendaon changes analys forecas changes and combnaons} Caegores of Shorsellers {all ndv. ns. np ns. p prop p. oher} Oher vews Cross Seconal Approach Facor Tmng shor 5 X e
Wha do hey fnd? Confrm Boehmer Jones & Zhang (2008) resul: One SD ncrease n shorng leads o 0% decrease n nex week s reurn. Fnd abou one-quarer of underperformance can be arbued o earnngs and analys-relaed news releases. (News*Abre w/ News) + (No News*Abre w/o News) = Abre.6%*(3.36 +3.84) + (-.6%)*(3.36) = 3.80 bp/day 22% for earnngs or analys announcemen 24% for nsuons around earnngs analys announcemen or analys recommendaon change. 43% for all news releases for ndvduals.
Economercs k e X shor d c b shor b r 5 0 5 0 Man expermen s: Suppose he rue model s: In oher words he dummy varable s no ncluded alone. So error erm s: Unless wll be correlaed wh d volang he OLS assumpon ha he ndependen varables are uncorrelaed wh he error erm. k X shor d c shor b d z b r 5 0 5 2 0 d z e e z 0
So s z =0? z s measurng he response of reurns o earnngs and analys news when shor sellng s zero. Good reason o hnk ha reurns are dfferen on earnngs and analys news days. Le s See: Use Dow-Jones News Daabase 2005-2007 d s earnngs news...no acual earnngs. All correlaons wh d are sascally sgnfcan a he 5% level. d shor ;-5-0.046 d re ; +2-0.00647 re ; +20 0.00336
Replcaon of Table 4 Orgnal Specfcaon r +k = b 0 + b *shor ;-5- + c 0 *d *shor ;-5- + γln(mc m- ) + e 2 Day Reurn % ( +) 20 Day Reurn % ( +20) Esmae -sa Esmae -sa nercep [b 0 ] 0.48 2.2 4.0944 35.34 shor [b ] -0.285-6.55 -.6878-3.40 d *shor [c 0 ] -0.3250-4.77 0.237 0.63 LN(MC) [γ] -0.079-7.30-0.57-2.69
Replcaon of Table 4 Revsed Specfcaon r +k = b 0 + b *shor ;-5- + z *d + c 0 *d *shor ;-5- + γln(mc m- ) + e 2 Day Reurn % ( +) 20 Day Reurn % ( +20) Esmae -sa Esmae -sa nercep [b 0 ] 0.4849 2.29 4.088 35.9 shor [b ] -0.3024-6.83 -.680-2.47 d [z ] -0.0675 -.97 0.225 2.22 d *shor [c 0 ] -0.042-0.27-0.8020 -.74 LN(MC) [γ] -0.079-7.26-0.574-2.72
Commen: Endogeney/Reurn Conrols Is he rec. changes/reurn relaonshp endogenous? Suppose here s an unexpeced even ha drves boh shor erm reurns and analys recommendaon changes Then he news may be causng he shorng and he reurn no he analys recommendaon. Especally mporan b/c much of he acon s comng from recommendaon changes. Esmae of -0.74 on c 2 vs -0.0537 on c and -0.005 on c 3
Oher Quesons The paper could do more wh accoun ypes. I mss he oher coeffcen esmaes. Dae Range: Oc. 2000-Sep. 2005 Does no cover recen perod Does no cover new shor sellng regulaory regme Tha could be a good hng
Overall Nce daa used houghfully o answer a new queson. We learn ha much of shor sellers advanage comes from earnngs and analys news. Some ssues and quesons Bu he overall pcure of why do shor sellers have an advanage s an mporan and unanswered queson.