TREND function results jump dramatically more than they should on next set of data?

OK data and statistics gurus, could use some experience on this one...

I have created a mock up of the data I'm working with trying to forecast (bad use of the word since using the TREND function instead of the FORECAST function I know ). There are four variables to contend with. Number of people, Year, Percentage of the day worked, and Amount paid per percentage of the day worked. I added three columns and coded the data of the Percentage of the day worked into three separate columns as you can see when you open the spreadsheet.

e.g
100% = 1 0 0
75% = 0 1 0
50% = 0 0 0
25% = 0 0 1


All the data is current up to Line 1146 where I start using the TREND function. The problem is, the amounts each year change slightly, by about 13 points or so, but then the moment the TREND formula starts, the data jumps by over 100 points. You'll see what I mean when you look at the sheet.

Any ideas why this is happening? Is there a better/more accurate way to forecast (has to be a better word here) this?

Thanks for looking!