*************************************************************************************************** Doit3: starting new run If bankrupt, interpreting as before reported date No stocks. gamma is mean. merging all rounds together All industries together Using round to IPO/Acquisition returns. Returns overlap for same company! Doing model with k, bankruptcy minage parameter 0.25 using all rounds, even those that take less than 60 days keeping all returns, even 4 huge There are data points in this estimation before cleaning = 16852.00 elminated observations with unknown fate: 0 Eliminated observations with >100% or <0% shareholder value 99.00 Eliminated observations with missing round date 0 Note. Beta procedure -- cannot handle missing round date, so eliminating those. Eliminating observations that start in second quarter 2000. Need one quarter lag for probability. 107.00 If out date is before or equal to begin date, treat out date as missing. This affects 25.00 Relabeling all ipo, acquisition or new round with zero return as out of business. This affects 37.00 deleting observations with log annualized return greater than 15 4.00 Remaining data: 16642.00 Note: following refers to round, not company. Round may end in another round, though company eventually goes public Percent bankrupt 9.19 Percent ipo 21.20 Percent acquired 20.41 Percent with subsequent round 0 Percent Private 45.49 Percent Ipo registered 3.71 Percent fate unknown 0 0 Percent of bankrupt have good data. Using this parameter in simulation Note: good means round date and exit date not -99 95.09 Percent of new valuation (ipo,acquired,new round) that have good data. Using this parameter in simulation. Note: good means final date not -99, return not missing or 0 47.66 findlk3 WARNING: lnV is past grid. Make grid bigger. i x 6341.00 7456.00 2.00 8.00 19950407.00 1.00 3.00 19960412.00 1.00 4144.00 1368.38 0 0 0 Directional Iteration Func-count f(x) Step-size derivative 1 5 59948.1 0.001 -4.26e+008 2 14 59948.1 1e-011 -4.26e+008 3 23 57535.9 6.36696e-006 -2.97e+007 4 32 54834.6 1.28654e-005 -8.38e+006 5 42 54743.9 3.35213e-005 -598 6 52 54611.8 0.00132003 -1.02e+004 7 61 54316.7 1.23071 37 8 70 54276 1.57634 -0.437 9 79 54249 2.64359 -0.289 10 88 54242.9 0.877628 -0.012 11 97 54240.6 1.36495 0.0453 12 107 54239.2 1.77265 -0.0712 13 116 54239.1 0.397511 -0.223 14 125 54239 0.524783 -0.137 15 137 54239 0.000110675 -0.754 Optimization terminated successfully: Search direction less than 2*options.TolX initial and final parameters 0.01 0 0.90 0.10 1.00 3.00 0 0.04 NaN 0.52 0.18 0.68 5.26 NaN chi2 tests. NOTE NO MINIMIZATION OVER OTHER PARAMETERS YET test for gamma = 0, and all other parameters unchanged. chi2 value, chi2(1) and chi2(#pars) 865.61 3.84 11.07 looking for most influential observations 30 most influential observations and their contrib to lkly co. no. round date post value exit date exit typ post value gross ret age(mo) lkly contrib 5689.00 19960815.00 750.00 19990630.00 2.00 25.00 0.03 34.49 15.76 6146.00 19941115.00 6.40 19950302.00 1.00 88.60 10.89 3.57 15.80 2083.00 19960410.00 51.50 19960816.00 1.00 677.90 11.59 4.20 15.80 998.00 19900101.00 28.30 19971107.00 1.00 24.80 0.05 94.20 15.85 4191.00 19900207.00 16.80 19990518.00 2.00 1.50 0.05 111.36 16.34 3859.00 19990115.00 43.55 19990721.00 1.00 1914.60 37.31 6.20 16.72 1985.00 19990322.00 140.00 19990629.00 1.00 86.20 0.12 3.23 16.95 1269.00 19920814.00 7.20 19961001.00 2.00 0.50 0.03 49.57 17.02 1269.00 19920201.00 6.70 19961001.00 2.00 0.50 0.03 56.00 17.18 2773.00 19950817.00 16.50 19990607.00 2.00 0.41 0.02 45.67 17.28 6613.00 19950726.00 15.00 19970815.00 2.00 4.00 0.02 24.64 17.41 1531.00 19990401.00 212.00 19990701.00 1.00 3034.10 13.94 3.00 17.52 4471.00 19981115.00 2.40 19990923.00 1.00 618.00 98.61 10.26 17.56 1145.00 19951222.00 31.60 19980914.00 2.00 0.62 0.02 32.74 17.89 1765.00 19970531.00 108.00 19980626.00 1.00 104.00 0.03 12.84 18.34 1313.00 19990501.00 122.00 19991101.00 2.00 6900.00 56.56 6.00 19.32 4636.00 19990219.00 159.00 19990505.00 1.00 2970.20 16.08 2.54 19.40 6533.00 19990917.00 120.00 20000201.00 1.00 82.90 0.04 4.47 19.65 2394.00 19971212.00 117.00 19980318.00 1.00 2160.30 17.80 3.20 20.39 6787.00 19960321.00 15.15 19971007.00 1.00 38.20 0.01 18.54 21.47 6787.00 19950321.00 12.00 19971007.00 1.00 38.20 0.01 30.54 21.78 5762.00 19991005.00 38.00 20000306.00 2.00 805.20 21.19 5.03 22.50 6787.00 19930219.00 2.20 19971007.00 1.00 38.20 0.01 55.61 23.40 3567.00 19931201.00 150.00 19990402.00 2.00 1.25 0.01 64.03 23.65 6787.00 19961115.00 20.34 19971007.00 1.00 38.20 0.01 10.74 24.83 3567.00 19940214.00 260.00 19990402.00 2.00 1.25 0.00 61.61 25.75 7456.00 19950407.00 3.00 19960412.00 1.00 4144.00 1368.38 12.16 27.14 6787.00 19970327.00 26.30 19971007.00 1.00 38.20 0.01 6.34 27.43 3966.00 19891024.00 10.00 19901001.00 1.00 22.80 0.00 11.24 29.65 5488.00 19960116.00 28.80 19960627.00 1.00 1607.10 54.61 5.36 33.87 6787.00 19970627.00 40.20 19971007.00 1.00 38.20 0.01 3.34 36.75 Table . Maximum Likelyhood estimates A. Parameter Estimates and Standard Errors Parameter estimate std err mu % 17.52 0.53 sig % 103.26 0.70 k % 18.27 0.04 a 0.68 0.00 b 5.26 0.02 pim * 100 NaN NaN c 0.95 d 0.48 log likelyhood -54239.03 gamma and sigma are reported on an annualized basis - 4*gamma and 2*sigma. pim is times 100 Implied mean and sd of return levels 77.50 131.96 NOTE: Doit3 is incrementing tablerow *************************************************************************************************** Doit3: starting new run If bankrupt, interpreting as before reported date No stocks. gamma is mean. merging all rounds together All industries together Using round to IPO/Acquisition returns. Returns overlap for same company! Doing model with k, bankruptcy minage parameter 0.25 using all rounds, even those that take less than 60 days keeping all returns, even 4 huge There are data points in this estimation before cleaning = 16852.00 elminated observations with unknown fate: 0 Eliminated observations with >100% or <0% shareholder value 99.00 Eliminated observations with missing round date 0 Note. Beta procedure -- cannot handle missing round date, so eliminating those. Eliminating observations that start in second quarter 2000. Need one quarter lag for probability. 107.00 If out date is before or equal to begin date, treat out date as missing. This affects 25.00 Relabeling all ipo, acquisition or new round with zero return as out of business. This affects 37.00 deleting observations with log annualized return greater than 15 4.00 Remaining data: 16642.00 Note: following refers to round, not company. Round may end in another round, though company eventually goes public Percent bankrupt 9.19 Percent ipo 21.20 Percent acquired 20.41 Percent with subsequent round 0 Percent Private 45.49 Percent Ipo registered 3.71 Percent fate unknown 0 0 Percent of bankrupt have good data. Using this parameter in simulation Note: good means round date and exit date not -99 95.09 Percent of new valuation (ipo,acquired,new round) that have good data. Using this parameter in simulation. Note: good means final date not -99, return not missing or 0 47.66 findlk3 WARNING: lnV is past grid. Make grid bigger. i x 6341.00 7456.00 2.00 8.00 19950407.00 1.00 3.00 19960412.00 1.00 4144.00 1368.38 0 0 0 Directional Iteration Func-count f(x) Step-size derivative Error in ==> c:\aaj\offroad\revision\findlk3.m On line 92 ==> [trash, logVindx] = min(abs(logV-lvgrid)); % find nearest value gridpoint Error in ==> C:\MATLAB6p5\toolbox\optim\private\fminusub.m On line 89 ==> f = feval(funfcn{3},x,varargin{:}); Error in ==> C:\MATLAB6p5\toolbox\optim\fminunc.m On line 229 ==> [x,FVAL,GRAD,HESSIAN,EXITFLAG,OUTPUT] = fminusub(funfcn,x,verbosity,options,defaultopt,f,GRAD,HESS,varargin{:}); Error in ==> c:\aaj\offroad\revision\doit3.m On line 568 ==> [tparfin,lklyfin,exitflag,outputstr,grd,hessb] = fminunc(@findlk3,tpar0,options,... Error in ==> c:\aaj\offroad\revision\mlest3.m On line 95 ==> headtxt = ' All, baseline ' ; delete ipoal1.txt ; diary ipoall.txt ; doit3; save base1; % all together S&P clear all close all uiimport -file uiimport -file help memory MEMORY Help for memory limitations. If the OUT OF MEMORY error message is encountered, there is no more room in memory for new variables. You must free up some space before you may proceed. One way to free up space is to delete some variables (see CLEAR). Another is to issue the command PACK (See PACK). PACK compresses the data in memory, opening up larger contiguous blocks. Here are some additional system specific tips: Windows: Increase virtual memory by using System in the Control Panel. Unix: Ask your system manager to increase your Swap Space. *************************************************************************************************** Doit3: starting new run If bankrupt, interpreting as before reported date No stocks. gamma is mean. merging all rounds together All industries together Using round to IPO/Acquisition returns. Returns overlap for same company! Doing model with k, bankruptcy minage parameter 0.2500 using all rounds, even those that take less than 60 days keeping all returns, even 4 huge There are data points in this estimation before cleaning = 16852 elminated observations with unknown fate: 0 Eliminated observations with >100% or <0% shareholder value 99 Eliminated observations with missing round date 0 Note. Beta procedure -- cannot handle missing round date, so eliminating those. Eliminating observations that start in second quarter 2000. Need one quarter lag for probability. 107 If out date is before or equal to begin date, treat out date as missing. This affects 25 Relabeling all ipo, acquisition or new round with zero return as out of business. This affects 37 deleting observations with log annualized return greater than 15 4 Remaining data: 16642 Note: following refers to round, not company. Round may end in another round, though company eventually goes public Percent bankrupt 9.1876 Percent ipo 21.1994 Percent acquired 20.4122 Percent with subsequent round 0 Percent Private 45.4873 Percent Ipo registered 3.7135 Percent fate unknown 0 0 Percent of bankrupt have good data. Using this parameter in simulation Note: good means round date and exit date not -99 95.0948 Percent of new valuation (ipo,acquired,new round) that have good data. Using this parameter in simulation. Note: good means final date not -99, return not missing or 0 47.6601 findlk3 WARNING: lnV is past grid. Make grid bigger. i x 6341 1.0e+007 * 0.0007 0.0000 0.0000 1.9950 0.0000 0.0000 1.9960 0.0000 0.0004 0.0001 0 0 0 Error in ==> c:\aaj\offroad\revision\findlk3.m On line 66 ==> newbeg = ddate(x(il,4)); % find beginning date of this observation */ Error in ==> C:\MATLAB6p5\toolbox\optim\fminunc.m On line 149 ==> f = feval(funfcn{3},x,varargin{:}); Error in ==> c:\aaj\offroad\revision\doit3.m On line 568 ==> [tparfin,lklyfin,exitflag,outputstr,grd,hessb] = fminunc(@findlk3,tpar0,options,... Error in ==> c:\aaj\offroad\revision\mlest3.m On line 95 ==> headtxt = ' All, baseline ' ; delete ipoal1.txt ; diary ipoall.txt ; doit3; save base1; % all together S&P mlest3 *************************************************************************************************** Doit3: starting new run If bankrupt, interpreting as before reported date No stocks. gamma is mean. merging all rounds together All industries together Using round to IPO/Acquisition returns. Returns overlap for same company! Doing model with k, bankruptcy minage parameter 0.2500 using all rounds, even those that take less than 60 days keeping all returns, even 4 huge There are data points in this estimation before cleaning = 16852 elminated observations with unknown fate: 0 Eliminated observations with >100% or <0% shareholder value 99 Eliminated observations with missing round date 0 Note. Beta procedure -- cannot handle missing round date, so eliminating those. Eliminating observations that start in second quarter 2000. Need one quarter lag for probability. 107 If out date is before or equal to begin date, treat out date as missing. This affects 25 Relabeling all ipo, acquisition or new round with zero return as out of business. This affects 37 deleting observations with log annualized return greater than 15 4 Remaining data: 16642 Note: following refers to round, not company. Round may end in another round, though company eventually goes public Percent bankrupt 9.1876 Percent ipo 21.1994 Percent acquired 20.4122 Percent with subsequent round 0 Percent Private 45.4873 Percent Ipo registered 3.7135 Percent fate unknown 0 0 Percent of bankrupt have good data. Using this parameter in simulation Note: good means round date and exit date not -99 95.0948 Percent of new valuation (ipo,acquired,new round) that have good data. Using this parameter in simulation. Note: good means final date not -99, return not missing or 0 47.6601 findlk3 WARNING: lnV is past grid. Make grid bigger. i x 6341 1.0e+007 * 0.0007 0.0000 0.0000 1.9950 0.0000 0.0000 1.9960 0.0000 0.0004 0.0001 0 0 0 Directional Iteration Func-count f(x) Step-size derivative 1 5 59948.1 0.001 -4.26e+008 Error in ==> c:\aaj\offroad\revision\findlk3.m On line 209 ==> ageindx = floor((ddate(outdat)-1987)*4)+1 - begdatinx; Error in ==> C:\MATLAB6p5\toolbox\optim\private\fminusub.m On line 302 ==> f = feval(funfcn{3},x,varargin{:}); Error in ==> C:\MATLAB6p5\toolbox\optim\fminunc.m On line 229 ==> [x,FVAL,GRAD,HESSIAN,EXITFLAG,OUTPUT] = fminusub(funfcn,x,verbosity,options,defaultopt,f,GRAD,HESS,varargin{:}); Error in ==> c:\aaj\offroad\revision\doit3.m On line 568 ==> [tparfin,lklyfin,exitflag,outputstr,grd,hessb] = fminunc(@findlk3,tpar0,options,... Error in ==> c:\aaj\offroad\revision\mlest3.m On line 95 ==> headtxt = ' All, baseline ' ; delete ipoal1.txt ; diary ipoall.txt ; doit3; save base1; % all together S&P mlest3 *************************************************************************************************** Doit3: starting new run If bankrupt, interpreting as before reported date No stocks. gamma is mean. merging all rounds together All industries together Using round to IPO/Acquisition returns. Returns overlap for same company! Doing model with k, bankruptcy minage parameter 0.2500 using all rounds, even those that take less than 60 days keeping all returns, even 4 huge There are data points in this estimation before cleaning = 16852 elminated observations with unknown fate: 0 Eliminated observations with >100% or <0% shareholder value 99 Eliminated observations with missing round date 0 Note. Beta procedure -- cannot handle missing round date, so eliminating those. Eliminating observations that start in second quarter 2000. Need one quarter lag for probability. 107 If out date is before or equal to begin date, treat out date as missing. This affects 25 Relabeling all ipo, acquisition or new round with zero return as out of business. This affects 37 deleting observations with log annualized return greater than 15 4 Remaining data: 16642 Note: following refers to round, not company. Round may end in another round, though company eventually goes public Percent bankrupt 9.1876 Percent ipo 21.1994 Percent acquired 20.4122 Percent with subsequent round 0 Percent Private 45.4873 Percent Ipo registered 3.7135 Percent fate unknown 0 0 Percent of bankrupt have good data. Using this parameter in simulation Note: good means round date and exit date not -99 95.0948 Percent of new valuation (ipo,acquired,new round) that have good data. Using this parameter in simulation. Note: good means final date not -99, return not missing or 0 47.6601 findlk3 WARNING: lnV is past grid. Make grid bigger. i x 6341 1.0e+007 * 0.0007 0.0000 0.0000 1.9950 0.0000 0.0000 1.9960 0.0000 0.0004 0.0001 0 0 0 Directional Iteration Func-count f(x) Step-size derivative 1 5 59948.1 0.001 -4.26e+008 2 14 59948.1 1e-011 -4.26e+008 3 23 57535.9 6.36696e-006 -2.97e+007 4 32 54834.6 1.28654e-005 -8.38e+006 5 42 54743.9 3.35213e-005 -598 6 52 54611.8 0.00132003 -1.02e+004 7 61 54316.7 1.23071 37 8 70 54276 1.57634 -0.437 9 79 54249 2.64359 -0.289 10 88 54242.9 0.877628 -0.012 11 97 54240.6 1.36495 0.0453 12 107 54239.2 1.77265 -0.0712 13 116 54239.1 0.397511 -0.223 14 125 54239 0.524783 -0.137 15 137 54239 0.000110675 -0.754 Optimization terminated successfully: Search direction less than 2*options.TolX initial and final parameters 0.0100 0 0.9000 0.1000 1.0000 3.0000 0 0.0438 NaN 0.5163 0.1827 0.6815 5.2627 NaN chi2 tests. NOTE NO MINIMIZATION OVER OTHER PARAMETERS YET test for gamma = 0, and all other parameters unchanged. chi2 value, chi2(1) and chi2(#pars) 865.6108 3.8415 11.0705 looking for most influential observations 30 most influential observations and their contrib to lkly co. no. round date post value exit date exit typ post value gross ret age(mo) lkly contrib 5689.00 19960815.00 750.00 19990630.00 2.00 25.00 0.03 34.49 15.76 6146.00 19941115.00 6.40 19950302.00 1.00 88.60 10.89 3.57 15.80 2083.00 19960410.00 51.50 19960816.00 1.00 677.90 11.59 4.20 15.80 998.00 19900101.00 28.30 19971107.00 1.00 24.80 0.05 94.20 15.85 4191.00 19900207.00 16.80 19990518.00 2.00 1.50 0.05 111.36 16.34 3859.00 19990115.00 43.55 19990721.00 1.00 1914.60 37.31 6.20 16.72 1985.00 19990322.00 140.00 19990629.00 1.00 86.20 0.12 3.23 16.95 1269.00 19920814.00 7.20 19961001.00 2.00 0.50 0.03 49.57 17.02 1269.00 19920201.00 6.70 19961001.00 2.00 0.50 0.03 56.00 17.18 2773.00 19950817.00 16.50 19990607.00 2.00 0.41 0.02 45.67 17.28 6613.00 19950726.00 15.00 19970815.00 2.00 4.00 0.02 24.64 17.41 1531.00 19990401.00 212.00 19990701.00 1.00 3034.10 13.94 3.00 17.52 4471.00 19981115.00 2.40 19990923.00 1.00 618.00 98.61 10.26 17.56 1145.00 19951222.00 31.60 19980914.00 2.00 0.62 0.02 32.74 17.89 1765.00 19970531.00 108.00 19980626.00 1.00 104.00 0.03 12.84 18.34 1313.00 19990501.00 122.00 19991101.00 2.00 6900.00 56.56 6.00 19.32 4636.00 19990219.00 159.00 19990505.00 1.00 2970.20 16.08 2.54 19.40 6533.00 19990917.00 120.00 20000201.00 1.00 82.90 0.04 4.47 19.65 2394.00 19971212.00 117.00 19980318.00 1.00 2160.30 17.80 3.20 20.39 6787.00 19960321.00 15.15 19971007.00 1.00 38.20 0.01 18.54 21.47 6787.00 19950321.00 12.00 19971007.00 1.00 38.20 0.01 30.54 21.78 5762.00 19991005.00 38.00 20000306.00 2.00 805.20 21.19 5.03 22.50 6787.00 19930219.00 2.20 19971007.00 1.00 38.20 0.01 55.61 23.40 3567.00 19931201.00 150.00 19990402.00 2.00 1.25 0.01 64.03 23.65 6787.00 19961115.00 20.34 19971007.00 1.00 38.20 0.01 10.74 24.83 3567.00 19940214.00 260.00 19990402.00 2.00 1.25 0.00 61.61 25.75 7456.00 19950407.00 3.00 19960412.00 1.00 4144.00 1368.38 12.16 27.14 6787.00 19970327.00 26.30 19971007.00 1.00 38.20 0.01 6.34 27.43 3966.00 19891024.00 10.00 19901001.00 1.00 22.80 0.00 11.24 29.65 5488.00 19960116.00 28.80 19960627.00 1.00 1607.10 54.61 5.36 33.87 6787.00 19970627.00 40.20 19971007.00 1.00 38.20 0.01 3.34 36.75 Table . Maximum Likelyhood estimates A. Parameter Estimates and Standard Errors Parameter estimate std err mu % 17.52 0.53 sig % 103.26 0.70 k % 18.27 0.04 a 0.68 0.00 b 5.26 0.02 pim * 100 NaN NaN c 0.95 d 0.48 log likelyhood -54239.03 gamma and sigma are reported on an annualized basis - 4*gamma and 2*sigma. pim is times 100 Implied mean and sd of return levels 77.50 131.96 NOTE: Doit3 is incrementing tablerow