Cross sectional regression modified Jones and estimates of coefficient

[Matteo Ascione]

Dear Statalisters,

I would much appreciate your help with the following problem,

i have to run a modified jones cross sectional regression industry/year (industry are 31 and years are from 2011 to 2012) on comparables and i need the estimate of bx1 bx2 bx3 of the regression(Dependent variable is Accruals and x1 is attivo, x2 is term2 and x3 is term3) because i use them to calculate non-discretionary accruals of my sublesampe
my dataset is format as panel like this
Società . industry . year . accruals . attivo term 2 . term 3
1 1 2011 . 0.045 . 0.0001 -0.023 0.18
1 1 . 2012 . 0.01 . 0.0034 . 0.001 . 0.023
1 . 1 . 2013 . 0.02 . 0.045 . 0.03 . 0.30
2 . 1 . 2011 . 0.02 . 0.03 . 0.064 . 0.11

I don t need a minimun number of observation becuase on excell i ve already selected industries with more than 10 observation
Thank you so much for your help

[Carlo Lazzaro]

Matteo:
welcome to this forum.
Unfortunately, I'm not familiar with the modified jones approach (by the way: being this an interdisciplinary forum, it would be useful to post some references along with your query, Thanks).
That said, I would be concerned with your upsteraem approach of selecting industries with more than ten observation. Have you considered possible biases in your results due to this methodological choice?

[Matteo Ascione]

Hi carlo,
thanks for your answer
Cross sectional modified jones approach states that societies in the same industries/year have same non discretionary accruals; The analysis is divided in two parts: first one is to run the corss section regression on Comparables ( same industries/year), the second one is to use the estimates of b1 b2 b3 (of previous regression) in my sample cluster analysed.
Comparables and my cluster has no common societies because the hypothesis is the absense of earnings managment in comparables: as consequence, comparables total accruals is a proxy for non discretionary accruals for my cluster (where it should be high level of earnings management). Indeed, i need the estimates of b1 b2 b3 to calculate NDA in my sample and then find discretionay ones.
What concerned about my upstream approch, i meant that in my cluster i ve selected just industries that have at least 10 comparables observations
Ps comparables are old listed companies meanwhile my cluser is composed by really young listed companies.
(i m sorry for my english but this topic is hard to discuss even in italian)

the cross regression i need to run is accruals attivo term2 term3, and i need the estimates of betas of indipendent varables for each year/industries and i don t know how to do it on stata

[Carlo Lazzaro]

Matteo:
unfortunately, -search jones- and -search modified jones- give back no community-contributed programmes.
Let's hope that interested listers skilled on this topic will chime in.