Calculate a value based on its previous value in the same column

There are probably more efficient ways to do this. But this should work for you. It relies on grouping the table into sections and generating an output list for each section. A nested zero based index is then used to extract the desired output from the index position of the list.

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("fdFBDoAgDETRu7AmhlaoeBbj/a+hjYmK6XcBizeQNNNtS1q0pZzqecSvMhVpac934idyA1/AO/gKLoUCoUCDwMeMPrjT+xm8gkcVGVRkUJH9VORzar8Snd/JM+noFbyBG/gC3sFX8NcyP4FQoEEwLnN0eh9VNC7TfT8A", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Year = _t, Month = _t, ValueStart = _t, Rate = _t]),
    #"Changed Type" = 
    Table.TransformColumnTypes(
        Source,
        {
            {"Year", Int64.Type}, {"Month", Int64.Type}, {"ValueStart", Int64.Type}, {"Rate", type number}
        }
    ),
    //this uses the ValueStart column to group the the table into sections that can be indexed and can have the list generate fuction applied.
    //it would be more stable if a more reliable grouping column was used.
    #"Filled Down" = 
    Table.FillDown(
        #"Changed Type",
        {"ValueStart"}
    ),
    //group the table by ValueStart and Rate, getting a nested table row count and all rows in the table
    #"Grouped Rows" = 
    Table.Group(
        #"Filled Down", 
        {"ValueStart", "Rate"}, 
        {
            {"Count", each Table.RowCount(_), Int64.Type}, 
            {"AllRows", each _, type table [Year=nullable number, Month=nullable number, ValueStart=nullable number, Rate=nullable number]}
        }
    ),
    //use the List.Generate function to create a list of outputs 
    Custom2 = 
    Table.AddColumn(
        #"Grouped Rows", 
        "list", 
        each 
        List.Generate(
            ()=> [
                    x=[ValueStart],             //initial x value is the ValueStart value for the row
                    y=[Rate],                   //y is always the Rate for the row
                    z=[Count]                   //z is the number of rows in the nested table, i.e., the number of output rows we need
                ], 
                each [z] > 0,                   //iterates as long as z is > 0 
                each [
                        x= [x] - ([x]*[y]),     //for every iteration, x is set to the previous value of x minus the previous value of x times the rate
                        y=[y],                  //y never changes
                        z = [z]-1               //z is decreased by one for every iteration
                    ], 
                each [x]                        //return the value of x to the list
        ), 
        type list
    ),
    //add an index column to the nested AllRows table. IMPORTANT that this starts at 0 for list indexing reasons.
    Custom1 = 
    Table.TransformColumns(
        Custom2, 
        {
            {"AllRows", each Table.AddIndexColumn(_, "Index", 0, 1, Int64.Type), type table [Year=nullable number, Month=nullable number, Index=nullable number]}
        }
    ),
    //expand the needed columns from the nested table to new rows
    #"Expanded AllRows" = 
    Table.ExpandTableColumn(
        Custom1, 
        "AllRows", 
        {"Year", "Month", "Index"}, 
        {"Year", "Month", "Index"}
    ),
    //add a column that displays the value from the list at the current row's index position. 
    Custom3 = 
    Table.AddColumn(
        #"Expanded AllRows", 
        "Output", 
        each [list]{[Index]}, 
        type number
    ),
    //sort the needed columns
    Custom4 = 
    Table.SelectColumns(
        Custom3, 
        {"Year", "Month", "ValueStart", "Rate", "Output"}
    )
in
    Custom4