[Clyde Schechter]

There is no built in feature of a command that restricts regressions to at least 10 observations, but what you can do is modify the code so that it discards results based on fewer than 10 observations.

gen Jones_Modified = . 
forval y = 1999(1) 2016 { 
    forval i = 1(1) 52 { 
        display `i' 
        display `y' 
        capture reg TA B1 B2 B3 if `i' == industry & `y' == year, noconstant
        if c(rc) == 0  {    // SUCCESSFUL REGRESSION
            if e(N) >= 10    {    // IGNORE IF < 10 OBS IN REGRESSION
                predict r if `i' == industry & `y' == year, resid 
                replace Jones_Modified = r if `i' == industry & `y' == year 
                drop r
            }
        }
        else if !inlist(c(rc), 2000, 2001) { // UNANTICIPATED ERROR
            display as error "Unexpected error encountered"
            exit c(rc)
        }
        else {    // NO OR INSUFFICIENT OBSERVATIONS; NOTIFY & PROCEED
            display "No or insufficient observations"
        }
    } 
}

[Ashley Choi]

Dear Clyde
I've tried with your comments,
run the code as below and no result

one more thing,
my data has missing total_acc & inv_ta....
it there any code to ignore the missing data when run the regression?


foreach i in `a' {
foreach x in `b' {
capture noisily reg total_acc inv_ta rev_ar s_ppe if ind_code==`i'&fyr==`x',nocons
if c(rc)==0
if e(N)>=10{
capture noisily predict aa2 if ind_code==`i'&fyr==`x',resid
capture noisily replace aa=aa2 if if ind_code==`i'&fyr==`x'
capture noisily drop aa2
}
}
}

[Yasin Seker]

Hi,Is the following code true for the m jones model? data: 2010-2017 for 35 country and 24 industry
I think that in case of more than one country, country factor should be taken into consideration in addition to firm-industry.
Thank you for your contributions.


xtset firm date, yearly
gen dREC = d.REC
gen dREV = d.REV
gen lA = l.A
gen TACC = TA/lA
gen term1 = 1/lA
gen term2 = (dREV - dREC)/ lA
gen term3 = PPEG / lA
tabulate industry, gen(d_dum)
egen ccode = group(country)
gen Jones_1995_TAC3=.
forval y = 2011(1)2017{
forval i = 1(1)35{
display `i'
display `y'
reg TACC term1 term2 term3 d_dum1-d_dum24 if `i'== ccode&`y'== date
predict residd if `i'== ccode&`y'== date
replace Jones_1995_TAC3=residd if `i'== ccode&`y'== date
drop residd
}
}

[Pedro Coelho]

Hi Yasin,

Please, read Statalist FAQ on the best practices on posting your questions. Namely, when asking for help with code, always show example data using -dataex-.

That said, I find no sense in the way you regress TACC. To do waht I think you wish to do, you would have to multiply each -term*- by each -d_dum*-.

You may try this:

  
gen Jones_1995_TAC3=.
 
gen TACC = TA/lA
gen term1 = 1/lA
gen term2 = (dREV - dREC)/ lA
gen term3 = PPEG / lA

forvalues j = 1/`=_N' {
capture quietly {
reg TACC term1 term2 term3 if industry == industry[`j'] & ccode == ccode[`j'] & date == date[`j'] & _n != `j'
if e(N) >= 8 {
            replace Jones_1995_TAC3== TACC - (_b[term1] * term1 + _b[term2] * term2 + _b[term3] * term3) in `j'
}
}
}

NOTE: 1. -_b[term1] * term1 + _b[term2] * term2 + _b[term3] * term3)- is the measure of normal accruals.
2. Here the minimum number of industry-date observations is 8.