Change in beta coefficients while running regressions with stock and flow variables

[lal mohan kumar]

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

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

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

gen deltafixedassets=d.fixedassets/totassets

I ran

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

[lal mohan kumar]

Dear Stata members

Since I couldn't invoke any response to my earlier post, let me ask one main question from the above

Can we regress a stock variable(dependent variable) on a flow variable(independent variable) and interpret the coefficient in the case of panel data?

For instance, let the stock variable be total wealth at time t(a dependent variable which is a stock variable-wealth accumulated till time t ) and the independent variable be income during the year(flow variable).
Can we do a regression as

xtreg wealth income, fe

or should we convert wealth into a flow variable first by taking the difference and then regress it on income

gen delta_wealth=(d.wealth)/wealth
xtreg delta_wealth income, fe

Which one is more correct and intuitive?
Note that in finance, I have seen CAPEX(a flow variable) is regressed on Cash flow as well as debt(a stock variable) regressed on cash flows. But now I wonder, which one is more appropriate,
Trust I have made my question clear. If members feel that this issues are addressed elsewhere, please share that information

[lal mohan kumar]

Dear Stata members

I thought of extending this query since I think the questions which I have, would have been already encountered or will be faced by some other people. The question is with respect to the use of Stock versus Flow variables in empirics.
As I posted above, in financial economics, sometimes a flow variable is used as a dependent variable(say CAPITAL EXPENDITURES, CHANGE IN GDP GROWTH RATE), as well as stock variables(TOTAL DEBT, CASH HOLDINGS, INVENTORY). Based on the type of variables results will change and hence the implications. For instance, in the article, "Considering the Use of Stock and Flow Outcomes in Empirical Analyses:
An Examination of Marriage Data", authors show that stock and flow data yield divergent results in terms of sign and significance.

In the blog on Statistical Modeling, Causal Inference, and Social Science, one entry mentions that "there’s another case where causality exists but correlation does not", which is related to the stock and flow concept
https://statmodeling.stat.columbia.edu/2015/03/12/24887/

I strongly think that it will be better to give some advice on choosing stock versus flow variables while running data analysis.
For instance, should we use economic growth as our dependent variable or change in growth, if we want to compare the performance of political parties (left versus right)? (https://statmodeling.stat.columbia.edu/2015/05/26/stock-flow-and-two-smoking-regressions/)

Also, in financial economics, capital structure studies, use Leverage/Debt at a point of time(hence a stock variable) as a dependent variable but for investments, regression uses Capital expenditures during a time period. Why such different treatments (adjustment costs are applicable to both investments and capital structure I guess).

If anyone can elaborate on this it would be highly helpful