brycewang-stanford/Auto-Empirical-Research-Skills / skills/18-jusi-aalto-stata-accounting-research
stata-accounting-research
UnverifiedSTATA code pattern library for empirical archival accounting research. Provides tested syntax from 126 peer-reviewed JAR (Journal of Accounting Research) replication files (2017-2025). Use when the user asks procedural questions like "How do I implement [method]?" or "Show me code for [technique]" — including: entropy balancing, propensity score matching (PSM), difference-in-differences (DiD), regression discontinuity (RDD), instrumental variables (IV), event studies (CAR/BHAR), survival analysis, Fama-MacBeth regressions, bootstrap, quantile regression, reghdfe/xtreg/areg, clustering standard errors, fixed effects, esttab/outreg2 table formatting, winsorization, leads/lags. Users can specify their variables (e.g., treatment, outcomes, controls) and receive adapted syntax. NOTE: This skill provides code patterns from published papers, not research design advice.
references/JAR_57_honigsberg.do
View on GitHubuse "C:\Users...Raw_DF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post NEW)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_DF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_DF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_DF.dta file (imported from separate do files for concision)
***Using raw dta file with ID variable, regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka dta file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_DF.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_DF.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_DF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_DF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...DF_Controls.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post remain control deregister)
winsor rateofreturn, p(.01) gen (CV_ret) //this is to correct for one obs showing monthly return over 1M. Clear error.
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_Gold.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_Gold.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_Gold.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_Gold.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_Gold.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_Gold.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...Gold_Controls.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post reg2006)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_HF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_HF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka. The code to generate the CV_sadka_HF.dta file is below.
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_HF.dta file (imported from separate do files for concision)
***Regress return on sadka permanent liquidity variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_2006.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_2006.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_HF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_HF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...HF_Controls.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***calculate fund age per month
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry remain control deregister)
drop rateofreturn
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_3period_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference using "C:\Users...CV_sadka_3Period.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_3Period.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(productreference) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_3Period.dta", replace
***Merge liquidity dta file with original raw file, keeping only the relevant control
merge m:1 productreference using "C:\Users...liquidity_3Period.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_3Period.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference using "C:\Users...CV_audit_3Period.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
g period = 2
save "C:\Users...3Period_Controls.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004).
***Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly).
***https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...DF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...DF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep productreference post adj_dec
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...DF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...DF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...DF_FLAGS.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post NEW)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink NEW
save "C:\Users...DF_KinkFlags.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...Gold_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...Gold_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...Gold_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...Gold_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...Gold_FLAGS.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post control remain deregister)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink control remain deregister
replace deregister = 0 if deregister == . //set to 1 if fund withdraws post Goldstein
replace remain = 0 if remain ==. //set to 1 if fund remains registered post Goldstein
save "C:\Users...Gold_KinkFlags.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...HF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...HF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...HF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...HF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...HF_FLAGS.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post reg2006)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink reg2006
save "C:\Users...HF_KinkFlags.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***generate key variables
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(productreference) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...3Period_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the raw data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 productreference using "C:\Users...3Period_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period, keep only relevant variables
keep productreference adj_dec
duplicates drop
save "C:\Users...3Period_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference using "C:\Users...3Period_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...3Period_FLAGS.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***classify each return into relevant bucket if that return falls from -1 to 0.5 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference control remain deregister)
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference abn_kink control remain deregister
duplicates drop
replace deregister = 0 if deregister == .
replace remain = 0 if remain ==.
save "C:\Users...3Period_KinkFlags.dta", replace
***Merge misreporting data with control variables
use "C:\Users...DF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...DF_Controls.dta"
drop _merge
***Remove disclosure-only funds and those funds that registered in response to the HF Rule
drop if ERA == 1
drop if previous_registrant==1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
rename domicilecountry country
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...DF_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...Gold_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...Gold_Controls.dta"
drop _merge
***Remove the newly registered funds that remained registered (match only withdraw and control funds)
drop if remain == 1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit deregister CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 deregister if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...Gold_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...HF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...HF_Controls.dta"
drop _merge
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
g cat1 = primarycategory
replace cat1 = "Other" if primarycategory == "Dedicated Short Bias" //Reclassify these two funds as "other" to improve possibility of match
replace cat1 = "Other" if primarycategory == "Options Strategy" //Reclassify these four funds as "other" to improve possibility of match (all other categories have at least ten funds)
egen cat=group(cat1)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...HF_PSM.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post NEW)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_DF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_DF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_DF.dta file (imported from separate do files for concision)
***Using raw dta file with ID variable, regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka dta file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_DF.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_DF.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_DF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_DF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...DF_Controls.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post remain control deregister)
winsor rateofreturn, p(.01) gen (CV_ret) //this is to correct for one obs showing monthly return over 1M. Clear error.
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_Gold.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_Gold.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_Gold.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_Gold.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_Gold.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_Gold.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...Gold_Controls.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post reg2006)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_HF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_HF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka. The code to generate the CV_sadka_HF.dta file is below.
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_HF.dta file (imported from separate do files for concision)
***Regress return on sadka permanent liquidity variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_2006.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_2006.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_HF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_HF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...HF_Controls.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***calculate fund age per month
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry remain control deregister)
drop rateofreturn
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_3period_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference using "C:\Users...CV_sadka_3Period.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_3Period.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(productreference) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_3Period.dta", replace
***Merge liquidity dta file with original raw file, keeping only the relevant control
merge m:1 productreference using "C:\Users...liquidity_3Period.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_3Period.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference using "C:\Users...CV_audit_3Period.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
g period = 2
save "C:\Users...3Period_Controls.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004).
***Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly).
***https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...DF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...DF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep productreference post adj_dec
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...DF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...DF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...DF_FLAGS.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post NEW)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink NEW
save "C:\Users...DF_KinkFlags.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...Gold_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...Gold_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...Gold_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...Gold_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...Gold_FLAGS.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post control remain deregister)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink control remain deregister
replace deregister = 0 if deregister == . //set to 1 if fund withdraws post Goldstein
replace remain = 0 if remain ==. //set to 1 if fund remains registered post Goldstein
save "C:\Users...Gold_KinkFlags.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...HF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...HF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...HF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...HF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...HF_FLAGS.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post reg2006)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink reg2006
save "C:\Users...HF_KinkFlags.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***generate key variables
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(productreference) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...3Period_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the raw data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 productreference using "C:\Users...3Period_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period, keep only relevant variables
keep productreference adj_dec
duplicates drop
save "C:\Users...3Period_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference using "C:\Users...3Period_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...3Period_FLAGS.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***classify each return into relevant bucket if that return falls from -1 to 0.5 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference control remain deregister)
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference abn_kink control remain deregister
duplicates drop
replace deregister = 0 if deregister == .
replace remain = 0 if remain ==.
save "C:\Users...3Period_KinkFlags.dta", replace
***Merge misreporting data with control variables
use "C:\Users...DF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...DF_Controls.dta"
drop _merge
***Remove disclosure-only funds and those funds that registered in response to the HF Rule
drop if ERA == 1
drop if previous_registrant==1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
rename domicilecountry country
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...DF_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...Gold_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...Gold_Controls.dta"
drop _merge
***Remove the newly registered funds that remained registered (match only withdraw and control funds)
drop if remain == 1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit deregister CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 deregister if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...Gold_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...HF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...HF_Controls.dta"
drop _merge
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
g cat1 = primarycategory
replace cat1 = "Other" if primarycategory == "Dedicated Short Bias" //Reclassify these two funds as "other" to improve possibility of match
replace cat1 = "Other" if primarycategory == "Options Strategy" //Reclassify these four funds as "other" to improve possibility of match (all other categories have at least ten funds)
egen cat=group(cat1)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...HF_PSM.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post NEW)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_DF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_DF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_DF.dta file (imported from separate do files for concision)
***Using raw dta file with ID variable, regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka dta file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_DF.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_DF.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_DF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_DF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...DF_Controls.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post remain control deregister)
winsor rateofreturn, p(.01) gen (CV_ret) //this is to correct for one obs showing monthly return over 1M. Clear error.
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_Goldstein_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_Gold.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_Gold.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_Gold.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_Gold.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_Gold.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_Gold.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...Gold_Controls.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***generate ID variable and calculate fund age per month
egen ID=group(productreference post) //used for Sadka calculation below
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry post reg2006)
rename rateofreturn CV_ret
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_HF_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference post)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference post using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference post using "C:\Users...CV_sadka_HF.dta" //the Sadka liquidity measure was supplied by Prof. Sadka. The code to generate the CV_sadka_HF.dta file is below.
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_HF.dta file (imported from separate do files for concision)
***Regress return on sadka permanent liquidity variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(ID) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_2006.dta", replace
***Merge liquidity dta file with original raw file (including ID), keeping only the relevant control
merge m:1 ID using "C:\Users...liquidity_2006.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference post
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_HF.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference post using "C:\Users...CV_audit_HF.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
drop ID
save "C:\Users...HF_Controls.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***calculate fund age per month
gsort productreference date
bysort productreference: carryforward dateaddedtotass, replace
g age=date- dateaddedtotass
***generate mean fund return, age, and net asset value over each period
collapse (mean) rateofreturn nav age, by (productreference primarycategory domicilecountry remain control deregister)
drop rateofreturn
rename nav CV_nav
rename age CV_age
replace CV_age=CV_age/365
replace CV_age=0 if CV_age<0
replace CV_nav=ln(CV_nav)
tempfile temp1
save "`temp1'"
use "C:\Users...Raw_3period_Data.dta", clear
***generate fund volatility over each period
collapse (sd) rateofreturn, by (productreference)
rename rateofreturn CV_std
***merge fund volatility with mean variables generated above
merge m:1 productreference using "`temp1'"
keep if _merge==3
drop _merge
***generate US dummy
g CV_US=1 if domicilecountry=="United States"
replace CV_US=0 if CV_US==.
***merge existing controls with dta file containing the fund's sensitivity to liquidity (sadka control)
merge 1:1 productreference using "C:\Users...CV_sadka_3Period.dta" //the Sadka liquidity measure was supplied by Prof. Sadka
keep if _merge==3
drop _merge
/* CODE TO GENERATE CV_sadka_3Period.dta file (imported from separate do files for concision)
***Regress return on sadka permanent variable provided by Prof. Sadka (line below creates a sadka.dta file with regression outputs)
statsby _b[_cons] _se[_cons] _b[variablepermanent] _se[variablepermanent] e(r2), by(productreference) saving(sadka): regress rateofreturn variablepermanent i.year, robust
***Open sadka file generated above and run the code below
rename _stat_2 se_cons
rename _stat_3 beta_sadka
rename _stat_4 se_sadka
rename _stat_5 R2
g t_sadka= beta_sadka/ se_sadka
g t_cons= beta_cons/ se_cons
drop if t_sadka==.
save "C:\Users...liquidity_3Period.dta", replace
***Merge liquidity dta file with original raw file, keeping only the relevant control
merge m:1 productreference using "C:\Users...liquidity_3Period.dta"
keep if _merge==3
drop _merge
keep beta_sadka productreference
rename beta_sadka CV_sadka
duplicates drop
save "C:\Users...CV_sadka_3Period.dta", replace
*/
***merge existing controls with dta file containing the number of audited months
merge 1:1 productreference using "C:\Users...CV_audit_3Period.dta" //as mentioned in the paper, the audit variable relies on historical data that is not available in the standard WRDS download. Please contact me directly for this code.
keep if _merge==3
drop _merge
g period = 2
save "C:\Users...3Period_Controls.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004).
***Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly).
***https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...DF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...DF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep productreference post adj_dec
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...DF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...DF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...DF_FLAGS.dta", replace
use "C:\Users...Raw_DF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post NEW)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink NEW
save "C:\Users...DF_KinkFlags.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...Gold_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...Gold_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...Gold_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...Gold_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...Gold_FLAGS.dta", replace
use "C:\Users...Raw_Goldstein_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post control remain deregister)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink control remain deregister
replace deregister = 0 if deregister == . //set to 1 if fund withdraws post Goldstein
replace remain = 0 if remain ==. //set to 1 if fund remains registered post Goldstein
save "C:\Users...Gold_KinkFlags.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***keep only funds with 30 obs per period and generate key variables
duplicates tag productreference post, gen (dups)
drop if dups<29
drop dups
egen ID=group(productreference post)
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(ID) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...HF_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the TASS data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 ID using "C:\Users...HF_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period (only for funds with data in both period), keep only relevant variables
keep adj_dec post productreference
duplicates drop
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
save "C:\Users...HF_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference post using "C:\Users...HF_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...HF_FLAGS.dta", replace
use "C:\Users...Raw_HF_Data.dta", clear
***classify each return into relevant bucket if that return falls from -1 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
g total_obs=1 //later used to keep only funds with 30 obs in each period
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period.
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference post reg2006)
***keep only funds with 30 obs in each period
keep if total_obs==30
duplicates tag productreference, gen (dups)
drop if dups<2
drop dups
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference post abn_kink reg2006
save "C:\Users...HF_KinkFlags.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***generate key variables
g december=1 if month==12
replace december=0 if december==. & month!=.
***merge the TASS data with the fund factors used by Fung and Hsieh (2004). Please see Prof. Hsieh's website for detail on these factors (the data on the website are updated monthly). https://faculty.fuqua.duke.edu/~dah7/HFRFData.htm
merge m:1 year month using "C:\Users...TrendRiskFactors.dta" //Hsieh excel file
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...BondYield.dta" //bond market factor https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
merge m:1 year month using "C:\Users...Moodys.dta" //credit spread factor https://fred.stlouisfed.org/series/DBAA https://fred.stlouisfed.org/series/DGS10
keep if _merge==3
drop _merge
g crdt_sprd=chg_moody-yld_chg
merge m:1 year month using "C:\Users...Datastream.dta" //equity oriented risk factors (S&P500 & size spread)
keep if _merge==3
drop _merge
/*
***regress each fund's return on an indicator for December, fund risk factors, and year fixed effects (line below generates a dta file called adj_december). Runs regression by ID generated above (fund-period)
statsby _b[_cons] _se[_cons] _b[december] _se[december] e(r2), by(productreference) saving(adj_december): reg rateofreturn december sizespread yld_chg chg_moody equity ptfsbd ptfsfx ptfscom i.year, robust
***open the adj_december file containing the saved regression output and run the code below
rename _stat_1 beta_cons
rename _stat_2 se_cons
rename _stat_3 beta_adjdec
rename _stat_4 se_adjdec
rename _stat_5 R2
g t_adjdec= beta_adjdec/ se_adjdec
g t_cons= beta_cons/ se_cons
drop if t_adjdec==.
save "C:\Users...3Period_adjdecember", replace
*/
***merge the adjdecember dta file generated above with the raw data, generate the misreporting flag based on the t-stat from the adjdecember dta file
merge m:1 productreference using "C:\Users...3Period_adjdecember.dta"
keep if _merge==3
drop _merge
g adj_dec=1 if t_adjdec>1.96 & t_adjdec!=.
replace adj_dec=0 if adj_dec==. & t_adjdec!=.
***keep one obs per fund in each period, keep only relevant variables
keep productreference adj_dec
duplicates drop
save "C:\Users...3Period_CookieJarFlags.dta", replace
///Merge measures of misreporting
merge 1:1 productreference using "C:\Users...3Period_KinkFlags.dta"
keep if _merge==3
drop _merge
save "C:\Users...3Period_FLAGS.dta", replace
use "C:\Users...Raw_3period_Data.dta", clear //contains data for the middle period in the extended 3-period model
***classify each return into relevant bucket if that return falls from -1 to 0.5 to 0.5
g kink1=1 if rateofreturn>0 & rateofreturn<=.50
g kink2=1 if rateofreturn<=0 & rateofreturn>-.50
g kink3=1 if rateofreturn<=-.5 & rateofreturn>-1
***sum the returns in each bucket and total obs per fund in each period. Line below generates one line of code per fund in each period
collapse (sum) kink1 kink2 kink3 total_obs, by (productreference control remain deregister)
***calculate difference between actual and expected number of obs in middle bucket
g expected=(kink1+kink3)/2
g diff=expected-kink2
***calculate standard deviation and t-stat
g pi=kink2/total_obs
g pi1=kink1/total_obs
g pi_1=kink3/total_obs
g standard=total_obs*pi*(1-pi)+(1/4)*total_obs*(pi_1+pi1)*(2-pi_1-pi1)
replace standard=sqrt(standard)
g test_stat=diff/standard
***generate misreporting flag based on t-stat
g abn_kink=1 if test_stat>1.96 & test_stat!=.
replace abn_kink=0 if abn_kink==. & test_stat!=.
***keep only relevant variables
keep productreference abn_kink control remain deregister
duplicates drop
replace deregister = 0 if deregister == .
replace remain = 0 if remain ==.
save "C:\Users...3Period_KinkFlags.dta", replace
***Merge misreporting data with control variables
use "C:\Users...DF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...DF_Controls.dta"
drop _merge
***Remove disclosure-only funds and those funds that registered in response to the HF Rule
drop if ERA == 1
drop if previous_registrant==1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
rename domicilecountry country
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...DF_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...Gold_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...Gold_Controls.dta"
drop _merge
***Remove the newly registered funds that remained registered (match only withdraw and control funds)
drop if remain == 1
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit deregister CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
egen cat=group(primarycategory)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 deregister if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...Gold_PSM.dta", replace
***Merge misreporting data with control variables
use "C:\Users...HF_FLAGS.dta", clear
merge 1:1 productreference post using "C:\Users...HF_Controls.dta"
drop _merge
***Sum measures of misreporting
g flag1=0
replace flag1=flag1+1 if abn_kink==1
replace flag1=flag1+1 if adj_dec==1
***Generate matched sample that ensures:
***(1) matched pairs have the same number of flags in the pre period,
***(2) US funds are matched to US funds (and non-US to non-US), and
***(3) matched funds have the same fund strategy (primarycategory variable).
***After these restrictions, preference is given to funds with similar propensity to be registered.
xi: probit NEW CV_ret CV_nav CV_age CV_std CV_sadka CV_audit
predict double ps
sum ps
***weight three restrictions above so that no match can be made unless all three are met
g cat1 = primarycategory
replace cat1 = "Other" if primarycategory == "Dedicated Short Bias" //Reclassify these two funds as "other" to improve possibility of match
replace cat1 = "Other" if primarycategory == "Options Strategy" //Reclassify these four funds as "other" to improve possibility of match (all other categories have at least ten funds)
egen cat=group(cat1)
g pscore1 =cat*1000+ps
g double pscore2 =CV_US*100000+pscore1
replace pscore2=flag1*10000000+pscore2
gsort productreference post
psmatch2 NEW if post==0, pscore(pscore2) caliper(100) noreplacement
gen pair = _id if _treated==0
replace pair = _n1 if _treated==1
bysort pair: egen paircount = count(pair)
save paired, replace
sort productreference _treated
g pspair=pair
replace pspair=pair[_n-1] if pspair==. & productreference==productreference[_n-1]
drop if pspair==.
g pspaircount=paircount
replace pspaircount=pspaircount[_n-1] if productreference==productreference[_n-1] & post==1
keep if pspaircount>1
g psweight=_weight
replace psweight=_weight[_n-1] if psweight==. & productreference==productreference[_n-1]
save "C:\Users...HF_PSM.dta", replace
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |