How Important are p-values in an ARDL Model?

[danishussalam]

Dear All,

I am working on how infrastructure (roads, gas, electricity and telephone) affects employment in those particular sectors. My ARDL model satisfies all tests (bound-testing, normality, serial correlation, cusum, cusumsq, heteroskedasticity). The only concern is the non-significance of p-values. Can I pass policy recommendation statements such as a 1% increase in gas infrastructure will increase the employment in gas sector by let's say 12%? (assuming that the coefficients on these are not significant)?

Will really appreciate if someone can explain.