Changed DivYieldST from VW to EW

Incorrectly assigned previously

Author: tomz23

SignalDoc.csv

@@ -50,7 +50,7 @@ DelNetFin,Predictor,1_clear,1_good,Richardson et al.,2005,Change in net financia
 DivInit,Predictor,1_clear,2_fair,"Michaely, Thaler and Womack",1995,Dividend Initiation,JF,discrete,Event,payout indicator,1964,1988,DivInit,t=3.4 in event study,"3 init, to Day 254",event study 12 months,1,0.625,3.37,EW,NA,NA,1,12,shrcd <= 11,"We deviate from OP in not imposing an NYSE/AMEX requirement.  This allows our portfolios to have a reasonable number of stocks.  We also ""hold"" for 6 months rather than 12, since most of the returns come in the first 6 months.","Keep only distcd 2nd digit = 2 or 3.  Define dividend initiation as having paid a dividend in month t (divamt > 0), and not having paid a dividend in the last 24 months.  DivInit is equal to 1 if a dividend was initiated in the past 6 months, and 0 for all other stocks."
 DivOmit,Predictor,1_clear,2_fair,"Michaely, Thaler and Womack",1995,Dividend Omission,JF,discrete,Event,payout indicator,1964,1988,DivOmit,t=6 in event study,"3 omit, to Day 254",event study 12 months,-1,0.916666667,6.33,EW,NA,NA,1,12,shrcd <= 11,"We deviate from OP in not imposing an NYSE/AMEX requirement.  This allows our portfolios to have a reasonable number of stocks.  OP's returns are short DivOmit and long EW crsp, which probably pushes up their t-stat.  Unlike DivInit, we ""hold"" for only 2 month because Table 3 shows that the DivOmit performance is highly concentrated early in the event.  Very few stocks in the omission portfolio.","Keep only distcd 2nd digit = 2 or 3.  Define firms as quarterly, semi-annual, or annual payers based on payment history.  Define a consistent payer as a firm that paid quarterly or semi-annual dividends for the past 18 months, or annual dividends for the past 2 years.  An omission is the first month where a consistent payer failed to pay a dividend over the past quarter/6-months/year.  Finally, DivOmit = 1 if there is an omission in the past 2 months and 0 otherwise"
 DivSeason,Predictor,1_clear,1_good,Hartzmark and Salomon,2013,Dividend seasonality,JFE,discrete,Event,payout indicator,1927,2011,DivSeason,t=16 in long-short,2B long (1) short (2),LS port,1,0.36,16.19363096,EW,NA,NA,1,12,abs(prc)>5,NA,"Drop if 3rd digit of distcd = 2 or >= 6.  Assign DivSeason = 0 if there was a dividend paid in the last 12 months.  Replace DivSeason = 1 if the third digit of disctcd is 3, 0, or 1, and a positive dividend was paid 2, 5, 8, or 11 months ago, if the third digit is 4 and a dividend was paid 5 or 11 months ago, or if the third digit is 5 and a dividiend was paid 11 months ago."
-DivYieldST,Predictor,1_clear,2_fair,Litzenberger and Ramaswamy,1979,Predicted div yield next month,JFE,discrete,Accounting,valuation,1936,1977,DivYieldST,t=6 in mv reg,1,mv reg,1,NA,6.3,VW,NA,NA,NA,NA,NA,"HXZ cite Litzenberger and Ramaswamy (LR) for their Dp predictor, but Dp is an annual dividend yield that is closer to Keim (1985), which shows very weak predictability.  LR is actually similar to DivSeason (Hartzmark and Solomon), and we follow LR.  LR uses a badly behaved regression with 75% of their div yield variable = 0, so we are flexible in our approach to mimic their results.  Also the paper is old, mostly theory, and provides little detail on their data handling.  Clear is a judgment call.","Using CRSP distributions, keep only distcd beginning in 12 and if the third digit of distcd is 3, 4, or 5.  Also keep only if stock paid a dividend in the last 12 months.  Define Ediv1 = div 2 months ago if distcd's 3rd digit is 0, 1, or 3.  Define Ediv1 = div 5 months ago if the distcd 3rd digit is 4.  Define Ediv = div 11 months ago if the distcd 3rd digit is 5.  Define Edy1 as Ediv1/abs(prc).  Finally, discretize Edy1 to smooth around the huge mass at 0 as follows: DivYieldST = 0 if Edy1 = 0, 1 if Edy is between 0 and 0.005, 2 if Edy1 is between 0.005 and 0.010, and 3 if Edy > 0.010."
+DivYieldST,Predictor,1_clear,2_fair,Litzenberger and Ramaswamy,1979,Predicted div yield next month,JFE,discrete,Accounting,valuation,1936,1977,DivYieldST,t=6 in mv reg,1,mv reg,1,NA,6.3,EW,NA,NA,NA,NA,NA,"HXZ cite Litzenberger and Ramaswamy (LR) for their Dp predictor, but Dp is an annual dividend yield that is closer to Keim (1985), which shows very weak predictability.  LR is actually similar to DivSeason (Hartzmark and Solomon), and we follow LR.  LR uses a badly behaved regression with 75% of their div yield variable = 0, so we are flexible in our approach to mimic their results.  Also the paper is old, mostly theory, and provides little detail on their data handling.  Clear is a judgment call.","Using CRSP distributions, keep only distcd beginning in 12 and if the third digit of distcd is 3, 4, or 5.  Also keep only if stock paid a dividend in the last 12 months.  Define Ediv1 = div 2 months ago if distcd's 3rd digit is 0, 1, or 3.  Define Ediv1 = div 5 months ago if the distcd 3rd digit is 4.  Define Ediv = div 11 months ago if the distcd 3rd digit is 5.  Define Edy1 as Ediv1/abs(prc).  Finally, discretize Edy1 to smooth around the huge mass at 0 as follows: DivYieldST = 0 if Edy1 = 0, 1 if Edy is between 0 and 0.005, 2 if Edy1 is between 0.005 and 0.010, and 3 if Edy > 0.010."
 dNoa,Predictor,1_clear,1_good,"Hirshleifer, Hou, Teoh, Zhang",2004,change in net operating assets,JAE,continuous,Accounting,investment,1964,2002,dNoa,t=8.9 in mv reg,7B DeltaNOA,mv reg,-1,NA,8.85,EW,NA,NA,1,6,NA,Seems to be monthly.  Table 4 footnote says decile portfolios formed monthly for NOA.  Interestingly they have a minimum 4 month lag.,"12-month growth in Net Operating Assets scaled by lagged total assets (at). Net Operating assets are operating assets minus operating liabilities. Operating assets are total assets (at) minus cash- and short-term investments (che), operating liabilities are total assets minus long-term debt (dltt), minority interest (mib), deferred charges (dlc), book equity (ceq) and preferred stock (pstk), all items (except at and ceq) replaced with 0 if missing."
 DolVol,Predictor,1_clear,1_good,"Brennan, Chordia, Subra",1998,Past trading volume,JFE,continuous,Trading,volume,1966,1995,VolumeDol,t=2.9 in regression,6A,mv reg,-1,NA,2.86,EW,NA,NA,1,12,NA,"OP is an exploration style paper, and DolVol is just one predictor.  Tables 4-6 have related results, but Table 6 is the simplest.  Table 6A has t=2.86 for NYSE subsample, Table 6B has t=2.6 for NASDAQ subsample, we write down 2.86, but use NYSE and NASDAQ in our ports.",Log of two-month lagged trading volume (vol) times two-month lagged price (prc).
 EarningsConsistency,Predictor,1_clear,2_fair,Alwathainani,2009,Earnings consistency,BAR,continuous,Accounting,earnings growth,1971,2002,EarnCons,t=2.7 in complicated LS port,11A CLG-CHG,LS port,1,0.360833333,2.67,EW,NA,NA,12,6,abs(prc)>1,"Could not access OP, so we used the Alwathainani's dissertation from VCU.  We follow MP, which is simpler than OP.","Average earnings growth over previous 48 months. Earnings growth is defined as EPS (epspx) minus EPS 12 months ago divided by average EPS 12 and 24 months ago. Exclude if price less than 5, absolute value of 12 month earnings growth greater 600%, or earnings growth and earnings growth 12 months ago have different signs."