DoD IG Report D-2000-091 — the FY1999 $2.3 Trillion Audit
artifact
The $2.3 trillion everyone argues about was published 18 months before Rumsfeld said it, in an audit that admits it could not tell what the money was.
Who they are
Department of Defense Inspector General report D-2000-091, published 25 February 2000[1].
What they do
The engine holds exactly what the auditors held, and nothing more.
How it works
The auditors looked at $6.9 trillion of accounting entries made to produce the Pentagon's 1999 financial statements. Of those, $2.6 trillion were properly supported. $2.3 trillion 'was not supported by adequate audit trails or sufficient evidence to determine their validity'. And $2 trillion was never examined at all, for lack of time. The previous year had its own separate $1.7 trillion of unsupported entries. We added up the audit's own table and it matches its text to the decimal point, which settles something: the $2.3 trillion is one year's figure, not a running total built up over years.
Why it matters
Both popular readings fail here. It is not proof of theft. It is also not proof of harmless book-keeping - the auditors specifically said they could not determine what the entries were, and did not look at $2 trillion more. Three things commonly said about this report are not in it: a figure of $59.6 billion, the phrase 'journal voucher', and the argument that a $270 billion annual budget makes it all impossible - which comes from a debunking forum and uses the wrong denominator. The audit's denominator is $6.9 trillion in a single year.
The engine's record — word for word
"INTERNAL CONTROLS AND COMPLIANCE WITH LAWS AND REGULATIONS FOR THE DOD AGENCY-WIDE FINANCIAL STATEMENTS FOR FY 1999. Report No. D-2000-091. February 25, 2000. Office of the Inspector General, Department of Defense." This is the document behind the figure Rumsfeld named on 10 September 2001, published eighteen months earlier. THE AUDIT'S OWN THREE-WAY SPLIT, verbatim: "We also identified $6.9 trillion in accounting entries to financial data used to prepare financial statements... For the accounting entries, $2.3 trillion was not supported by adequate audit trails or sufficient evidence to determine their validity, $2 trillion was not reviewed because of time constraints, and $2.6 trillion were supported." On what the entries were for: "These entries were processed to force financial data to agree with various data sources, to correct errors, and to add new data." Prior year: "approximately $1.7 trillion in unsupported accounting entries" for FY1998. Movement in one component: the Army General Fund "decreased the amount of unsupported accounting entries... from $673 billion in FY 1998 to $240 billion in FY 1999". The audit's own table sums to $2,290.5B unsupported, $2,555.1B supported, $2,042.4B unreviewed, $6,888.0B total — matching its published prose to the decimal, and confirming that $2.3 trillion is a SINGLE FISCAL YEAR's unsupported slice, not a cumulative total built from prior years. THE POSITION THE ENGINE TAKES IS THE AUDIT'S OWN: the auditors could not determine the validity of $2.3 trillion of entries and did not examine a further $2 trillion at all. They declined to characterise what the entries were, and so does canon. Neither 'theft' nor 'benign churn' is established by this document. NOT HELD, and each tested: a figure of '$59.6 billion in adjustments' (zero occurrences under two independent PDF extractors and a whitespace-flattened search); the phrase 'journal voucher' (zero occurrences — it belongs to the separate Skidmore/Fitts $21 trillion work at missing_21t); and the argument that a $270 billion annual budget makes the figure impossible, which is a debunking-forum argument reproduced verbatim in structure, and which uses the wrong denominator — the audit's denominator is $6.9 trillion of accounting entries in one year.
Follow the trail
Walk this on the live map →