Felix Salmon has a good post today about the Treasury's new report (opens .pdf) that attempts to show the success of its mortgage modification program. He correctly points out that the graph they show is cumulative, which necessarily paints a rosy picture. After all, cumulative graphs can't go down. I thought it might be fun to show what the graph could look like if drawn differently.
Unfortunately, the raw data is not available. As a result, I had to "eye" the graph to determine the cumulative values for each period it shows. That makes my charts approximate. With that said, they can't be too far off, and they certainly should show what I want them to: that things don't look quite as nice if shown more honestly.
First, here's their cumulative bar graph:
Next, here's that broken down into non-cumulative approximate modifications extended per week:
Suddenly, their modification efforts are not showing the steady progress that the cumulative graph implies. It shows a rather volatile record where they average around 35,000 modifications per week. Also note that the pressure Washington put on servicers last week must have paid some dividends. As of 7/31/09, in just four days -- not a full week -- they saw an uptick of around 40,000. Whether the pressure-effect will fade is yet to be seen.
Here's another little chart showing the approximate percentage of the total modifications extended to date by dividing the weekly non-cumulative number by the total number as of 7/31/09.
Finally, bear in mind that all of this tracks modifications "extended," not modifications "started." That chart show a little more than half the progress. Here's the Treasury's cumulative version of that one:
I didn't bother breaking that one down, because I think those approximations would have been much harder, given the very little progress seen week-to-week. But the average here is around 20,000 modifications started per week.
Finally, Salmon also mentions that the Treasury fails to explain how their "progress" compares to the number of foreclosures occurring each week in the market. Again, he's exactly right. Without having a base to measure this "success" against, it's hard to determine what it really means.