Re-estimating the regression coefficient

[ameridian]

I'm currently studying the impact of inflation on stock return.I found a positive relationship,inflation variable having a coefficient of 1.54.So,basically,when inflation grows with 1%,stock return grows with 1.54%.I'm thinking that this relationship cannot exist economically speaking at any value of inflation.For example if inflation grows with 50%,the return should grow with 77% percent,which is highly unlikely.So,my question,is there any way i can estimate in eviews an interval or a limit over which the impact will be negative,or even 0?For example if inflation grows over 10% the coefficient becomes negative?I'm using eviews 9.5 and my regresion is using monthly data from 1997 to 2017.The model that I use is GARCH(1,1).I was thinking about using simulations but i don't know exactly how to do it and if this is the correct approach.I'm new in the field of econometrics,so if my question may seem ambiguous or wrong i apologise.

[startz]

You could do something like

Code: Select all

ls r c inf (inf>.10)*inf