How would you regress this without employing an independent variable?

[Thomas Lehnsherr]

Analyzing the value of deals in joint ventures. Made two samples: domestic joint ventures per year, international joint ventures. (In a group of countries.)
The main focus would be weather the international deal values are different than the domestic ones.

Should I run two regression models or one?
Dependent: deal value
Control variables: inflation, GDP PPP etc.
Independent: *this would be weather domestic or international*

So maybe Model 1 would measure the effect of the controls on the deal value in terms of domestic deals, while Model 2 the same for international deals? Hence not using the independent variable, only the dependent + control.

[Jeff Wooldridge]

So you have two values per country, right? Then you should set it up as one data set with a country identifier and two variables per country: one for domestic sales the other international. Difference the variables within country. If you compute the average then that is the estimated difference in means without controls and you obtain the usual standard error. You can then use the difference as the dependent variable.

[Joro Kolev]

It seems from your description that you just need to regress deal value on a dummy for domestic/international and the rest of the controls. The dummy for domestic/international will tell you the difference in value between domestic and foreign deals.

[Jeff Wooldridge]

Joro's suggestion works, too, but then you have to compute a proper standard error to account for the within-country correlation. I was thinking of this as a "matched pairs" analysis. By differencing, you turn it into a cross section and then you just have to make standard errors robust to heteroskedasticity (if you have covariates).

[Thomas Lehnsherr]

Dear Jeff,

I have it like this now: M&A volume (domestic) M&A number (domestic) M&A volume (cross-border) M&A number (cross-border) M&A volume (all) M&A number (all)

So should I make the below calculation?
M&A volume (domestic) without dummy + M&A volume (cross-border) and here I add "1" as a dummy for instance?

And also the same for deal numbers.

[Joro Kolev]

Thomas, this is all too abstract, and I do not follow at all neither what your data is like, nor what you want to do with your data.

Can you not post a small subsample of your data, using -dataex-, and explain with reference to this subsample what you want to measure/estimate?

Dear Jeff,

I have it like this now: M&A volume (domestic) M&A number (domestic) M&A volume (cross-border) M&A number (cross-border) M&A volume (all) M&A number (all)

So should I make the below calculation?
M&A volume (domestic) without dummy + M&A volume (cross-border) and here I add "1" as a dummy for instance?

And also the same for deal numbers.

[Thomas Lehnsherr]

* Example generated by -dataex-. For more info, type help dataex
clear
input str7 Codes double(Inflationconsumerpricesannu MAvolumedomestic) byte MAnumberdomestic double MAvolumecrossborder byte MAnumbercrossborder double MAvolumeall byte MAnumberall double FDIIndex
"ALB1997"  33.1802743753956       . .        . .        . .    0
"ALB1998"  20.6428588670597       . .        . .        . .    .
"ALB1999"  .389437653561604       . .        . .        . .    .
"ALB2000" .0500181363468265       . .        . .        . .    .
"ALB2001"  3.10758827031434       . .        . .        . .    .
"ALB2002"   7.7705258343155       . .        . .        . .    .
"ALB2003"  .484002611818489       . .        . .        . .    0
"ALB2004"  2.28001916938101       . .        . .        . .    .
"ALB2005"  2.36658195679796       . .        . .        . .    .
"ALB2006"  2.37072831904283       . .        . .        . .    0
"ALB2007"  2.93268248162319       . .   120000 1   120000 1    .
"ALB2008"  3.36313757366391       . .        . .        . .    .
"ALB2009"  2.23139683475867       . .    48200 1    48200 1    .
"ALB2010"  3.62233541062328       . .        . .        . .    0
"ALB2011"  3.42912324722163       . .        . .        . .    0
"ALB2012"  2.03159593996543       . .        . .        . .    0
"ALB2013"  1.93761754903872       . .        . .        . .    0
"ALB2014"  1.62586504402605       . .        . .        . .    0
"ALB2015"  1.89617402592348       . .        . .        . .    0
"ALB2016"  1.27543168367409       . .        . .        . .    0
"ALB2017"  1.98666133171194       . .        . .        . . .057
"ALB2018"  2.02805963071135       . .        . .        . . .057
"ALB2019"  1.41109078954244       . .    50000 1    50000 1 .057
"ALB2020"                 0       . .        . .        . . .057
"AND1997"                 .       . .        . .        . .    .
"AND1998"                 .       . .        . .        . .    .
"AND1999"                 .       . .        . .        . .    .
"AND2000"                 .       . .        . .        . .    .
"AND2001"                 .       . .        . .        . .    .
"AND2002"                 .       . .        . .        . .    .
"AND2003"                 .       . .        . .        . .    .
"AND2004"                 .       . .        . .        . .    .
"AND2005"                 .       . .        . .        . .    .
"AND2006"                 .       . .        . .        . .    .
"AND2007"                 .       . .        . .        . .    .
"AND2008"                 .       . .        . .        . .    .
"AND2009"                 .       . .        . .        . .    .
"AND2010"                 .       . .        . .        . .    .
"AND2011"                 .       . .        . .        . .    .
"AND2012"                 .       . .        . .        . .    .
"AND2013"                 .       . .        . .        . .    .
"AND2014"                 .       . .        . .        . .    .
"AND2015"                 .       . .        . .        . .    .
"AND2016"                 .       . .        . .        . .    .
"AND2017"                 .       . .        . .        . .    .
"AND2018"                 .       . .        . .        . .    .
"AND2019"                 .       . .        . .        . .    .
"AND2020"                 .       . .        . .        . .    .
"ARM1997"   13.960764121791       . .        . .        . .    0
"ARM1998"   8.6724863238516       . .        . .        . .    .
"ARM1999"  .648245760470397       . .        . .        . .    .
"ARM2000" -.790883768934766       . .        . .        . .    .
"ARM2001"  3.14590464685019       . .        . .        . .    .
"ARM2002"  1.06004929341667       . .        . .        . .    .
"ARM2003"  4.72155336605342       . .        . .        . .    0
"ARM2004"  6.96126135875792       . .        . .        . .    .
"ARM2005"  .638928002410126       . .        . .        . .    .
"ARM2006"  2.89235662459002       . .   381900 1   381900 1    0
"ARM2007"  4.40736089644519       . .    38600 1    38600 1    .
"ARM2008"  8.94995335353386       . .        . .        . .    .
"ARM2009"  3.40676682683799       . .        . .        . .    .
"ARM2010"  8.17636138473956       . .        . .        . .    0
"ARM2011"   7.6500080785929       . .        . .        . .    0
"ARM2012"  2.55802007757907       . .        . .        . .    0
"ARM2013"  5.78966778544654       . .        . .        . .    0
"ARM2014"  2.98130868933673       . .        . .        . .    0
"ARM2015"  3.73169119261695       . .        . .        . .    0
"ARM2016" -1.40360755900906       . .        . .        . .    0
"ARM2017"  .969553268816231       . .        . .        . .    0
"ARM2018"  2.52023382001631       . .        . .        . . .019
"ARM2019"  1.44344660770702       . .  7113.77 1  7113.77 1 .019
"ARM2020"                 0       . .        . .        . . .019
"AUT1997"  1.30598307237497       . .        . .        . . .158
"AUT1998"  .922465840847906       . . 142727.3 1 142727.3 1    .
"AUT1999"  .568990766899814       . . 883378.5 2 883378.5 2    .
"AUT2000"  2.34486455542566       . . 128205.1 1 128205.1 1    .
"AUT2001"  2.64999922285055       . .     2000 1     2000 1    .
"AUT2002"  1.81035953871319  730000 3        . .   730000 3    .
"AUT2003"  1.35555673319826       . .    15900 1    15900 1 .149
"AUT2004"  2.06120354001778   12000 1   213000 1   225000 2    .
"AUT2005"  2.29913882222146       . .    44780 1    44780 1    .
"AUT2006"  1.44154696188379 1395000 2        . .  1395000 2 .149
"AUT2007"  2.16855599824037  110170 2        . .   110170 2    .
"AUT2008"  3.21595022953491 1832544 1    90300 1  1922844 2    .
"AUT2009"  .506308350120954   62585 3        . .    62585 3    .
"AUT2010"  1.81353628033681       . .        . .        . . .106
"AUT2011"  3.28658197365512       . .        . .        . . .106
"AUT2012"  2.48567352860068       . .        . .        . . .106
"AUT2013"  2.00015891157196  390000 1  1300000 1  1690000 2 .106
"AUT2014"  1.60580414549751 1042091 1        . .  1042091 1 .106
"AUT2015"  .896565315628228       . .        . .        . . .106
"AUT2016"  .891592367302732       . .        . .        . . .106
"AUT2017"  2.08126858275521   95000 1        . .    95000 1 .106
"AUT2018"  1.99838187702266 1900000 1        . .  1900000 1 .106
"AUT2019"  1.53089553422702       . .        . .        . . .106
"AUT2020"                 0       . .        . .        . . .106
"AZE1997"   3.6743480968473       . .        . .        . .    0
"AZE1998" -.772697886033187       . .        . .        . .    .
"AZE1999" -8.52517000536062       . .        . .        . .    .
"AZE2000"  1.80500303704358       . .        . .        . .    .
end

I think this dataex didn't work properly here, but anyway. So I'm trying to measure if there's a difference in the "Percentage of domestic M&A" and "Percentage of cross-border M&A". So the dummy would be weather it's cross-border or not. I also have the absolute values in euro. The control variables are GDP growth (%), FDI Regulatory Index, and then further control variables would be "ICT investment percentage" and "Internet access at home". As the industry is telecom.

[Joro Kolev]

And what is this Codes variable? It seems to be constructed from some letter abbreviation (which denote what a country?) and a year?

And then you want to measure whether the percentage of domestic M&A are different from foreign per country? Or overall ? Or per year?

My suggestion with the dummy for domestic is out, I thought that your data is at the transaction level.

Note that the percentage domestic and percentage foreign sum to 100.

So most probably you just need to calculate the percentage of domestic M&As and regress it on your explanatory variables.

[Thomas Lehnsherr]

Codes: yes, countrycode+year, e.g. GBR2002
Goal: domestic investments are different or not compared to international ones.

My data is coded as for country per year. So it might happen, than in a given year and country there's no transaction, or there's multiple ones. I measured the (1) aggregated volume for a given year for a given country, and also (2) the number of transations in that year. E.g. In 2002 in Great Britain 4 deals, 2 million EUR in sum. (For international) And same method for domestic deals.

Percentage: Yes, I thought so. So the dependent variables would be PercentageOfInternationalVolume. And then come the control variables.

Do you think this is a correct method?