Hi Hana, just as one possible approach, at one point in the past when I needed lag variables: I registered an orca table representing the previous year's table (e.g. "establishments_previous_year"), and updated this previous-year table each simulation year by using an orca step to make a copy of the relevant table (orca.add_table("establishments_previous_year", df) prior to the next year's simulated changes. And then registered columns on that previous-year table to serve as lag variables. Curious about your planned approach and any approaches others have taken.