Dear Stata members
My dependent variable is, say, fixed assets scaled by total assets. It is a stock variable. My independent variable is economic policy uncertainty (developed by Baker et al., 2016). As alternative measures of investment, I have another measure: - Cash spend for investment activities. Cash spend on investment activities is a flow variable, as it is a yearly item. However, this variable has predominantly -ve values, as -ve values denote money outflow (+ve values denote cash received from assets sales). I scaled this variable, too with total assets.
First I ran the regression as follows
Code:
reg fixedassets_totalassets log(epu)
Here I got a positive coefficient, which implies, higher the uncertainty, the investments in fixed assets increase
Then I ran
Code:
reg cashspend_totalassets log(epu)
Here I got a +ve coefficient, which implies, higher the uncertainty, the less cash will be spent (as spending is indicated as negative).

Also to create a flow variable from fixed assets, I generated

Code:
gen deltafixedassets=d.fixedassets/totassets
I ran
Code:
reg deltafixedassets log(epu)
I got -ve coefficient which implies higher the uncertainty, less addition made to fixed assets.

Are my results(signs) consistent?
Have I interpreted the coefficients correctly?
If so, why stock variables respond differently than their flow counterparts?

Can someone help me with this