I need to forecast repeated cross sectional data (survey data) with

(dep. var.) monthly electricity consumption = income (ind. var.)

I need to find the coefficients and multiply them with projected incomes to create forecasts for 30 years.
However, when I included a time series dummy in the regression above for fun, I realized that the effect of income diminishes over time to a large extent and I do not know how to model this.
I get values for the past years for the dummy but how can I forecast this?

Does anyone have an idea? Thank you in advance.