The Joint Committee's on Taxation's new website, which was rolled out this week, brings the mysterious methodologies of revenues estimates into clearer view. "Inside the JCT Revenue Estimating Process," a 26-page point point presentation by the JCT Chief of Staff reads like a pitch, with descriptions of tax models and application examples, the staff's access to 180,000 tax returns (enabling the Committee to create a "virtual world of American economic activity'), and a defense of its "Fixed GNP COnstraint" approach. A flowchart is provided on page 17 entitled "JCT Quality Control and Process," which shows the various levels of input and review that goes into each revenue estimate. If you've ever seen the JCT anaylsis letters that accompany their estimates, they are impressively sophisticated. But there is something missing from the materials that is probably important for members, staff, and the general public to know. The Committee claims that it "handle[d] 7,800 revenue estimate request in 2007" (p. 3). It published only 121 of these; only those referred to in public reports are published. So it completes some number of estimates over and above this number, but its staff of about three dozen can't possible do all that is requested of it. The website should explain:

• How many of the estimate requests it receives does it undertake?
• How does it decide which to undertake?
• How and when does it communicate its decisions regarding which to undertake?

The website is a good deal more readable and some information previously unavailable to the public is included in it. But to be fully accountable and transparent in its process and method -- to the degree it can be, given that it performs its tasks for and provides its product to members of Congress only -- is should explain how it decides which of the small fraction of the nearly 8,000 estimate requests it gets it decides to perform.