Forecasting ARIMA model with 2nd differenced or 2nd order of integration

[rapti]

I am forecasting liquidity (currency in circulation) using eviews-9. unit root test shows 1st difference gives non stationary data series. correlogram indicates I need to take 2nd difference. I haven't found any such literature like this. My questions are:
1. will it give a good forecasting result? if yes, then please give the command :how to estimate equation using 2nd differenced variable for ARIMA model?
2. Is there any other ways I can find ARIMA model in first differenced ?
3. my data series is daily series (5-days). If there are some missing data; ex: holidays, and I dont want to include it in the series what should I instruct to eviews? or should I go for interpolation?