A purchase, a contract or a budget line is often put to its decision maker as a short table: so many of one thing at one price, so many of another at another, and a total underneath. This paper argues that, in one designed set of decision records, the total of such a table does one job exactly and is then asked to do several more. The multiplication is correct and has been replayed. The records make its total into the figure that a budget, a ceiling or a tender will hold, while the doubts they themselves name lie in the quantities and prices the multiplication starts from, and while several of their questions ask for something that a sum of products cannot give.
The question and the contribution
The records come from AEGIS, the safety and protection studio of the RISCON family. AEGIS publishes authored decision cases: synthetic settings, written by an author, that pose a question to a named role, offer three options, mark the author's recommended option, list the inputs still missing, and say when the question would be looked at again. A programme of 196 of these cases goes further. Each of its cases executes one small calculation, keeps a hash of its input, and records that a second run gave the same output. One of its 12 rules multiplies, line by line, a quantity by a unit cost and adds the products. Seventeen cases in the programme run this rule, and they are this paper's material. They are synthetic, and nothing in them is a fact about any real budget, purchase or contract.
The question is put to the record set, not to the world. When a decision record prices a purchase, a contract or a budget line by an executed calculation of quantity times unit price, summed over lines, what does the calculation settle, what does the record do with its total, and what stays outside the lines?
The paper's contribution is a close reading of how one designed record turns a reproducible total into a reference figure. It shows that 12 of the seventeen recommendations carry the run's total into a budget, a procurement total, a ceiling, a reserve price, a baseline or a contract; that the reasons for those recommendations can describe the inputs as fixed while the runs describe the same inputs as assumed; that the records guard their separate lines against a single averaged price; and that the budget a total is compared with, the costs a line leaves out, and the basis that the allocation questions ask for all stay outside the calculation. None of this says what any purchase is worth or what any budget should be.
Five debates the records touch
Five bodies of work bear on records of this shape, and each gives this paper a question to put to them.
The first is about the line item itself. Yilin Hou, Line Item and Budget: Towards a Theory of Line Item in Budgeting (2023), treats the line item as an important, if informal, invention that paved the way for accounting and budgeting, and defines it as an information unit: the basic unit in the writing system of accounting and in the arrangement of information in accounts and budgets. The working paper notes that the line-item format has been mostly blamed and criticized since the budget movement at the turn of the twentieth century advocated exhaustive lists of line items, and argues that its advantages, convenience and conciseness explain why it has survived the program and performance budgeting reforms. A run of the AEGIS rule is a small line-item table, and the question is what the records let a line carry and what they refuse to let it lose.
John Wanat, Bases of Budgetary Incrementalism (1974), approaches budgets from another side. Testing the theory of budgetary incrementalism, the essay finds that strategic or gaming explanations are not warranted by correlational analysis, and explains the steady growth of requests through concepts in the vocabulary of the budget participants themselves, distinguishing mandatory requests from programmatic ones. This paper borrows only that attention to vocabulary. The records have a vocabulary of their own for a total: it is reflected in the draft budget as it stands, fixed as a ceiling, made the basis of the reserve price. The question is what those words do to a figure the same records call assumed.
The second debate is about numbers that become starting points. Amos Tversky and Daniel Kahneman, Judgment under Uncertainty: Heuristics and Biases (1974), describe adjustment from an anchor as a heuristic usually employed in numerical prediction when a relevant value is available, economical and usually effective but a source of systematic and predictable errors. Bent Flyvbjerg, Curbing Optimism Bias and Strategic Misrepresentation in Planning: Reference Class Forecasting in Practice (2007), documents that inaccurate projections of costs and demand are a major problem in planning, and presents reference class forecasting, which bases projections on the actual performance of a reference class of comparable actions. Neither work is about the AEGIS records, and this paper tests neither on them. It asks a narrower question that they suggest: when a record fixes an assumed total and then plans to record how actual prices depart from it, which figure becomes the reference, and what is the record collecting?
The third is about what a price leaves out. Lisa M. Ellram, Total Cost of Ownership (1995), presents an approach that looks beyond the price of a purchase to include many other purchase-related costs. Eric A. Greenleaf, Eric J. Johnson, Vicki G. Morwitz and Edith Shalev, The Price Does Not Include Additional Taxes, Fees, and Surcharges: A Review of Research on Partitioned Pricing (2015), review research on prices divided into a base price and one or more mandatory surcharges, and note concerns that buyers do not always attend to all the price information and underestimate totals. The AEGIS records are not about consumers, and the review's findings are not carried over to them. The question is narrower: where do the records put the maintenance, the allowances and the equipment that their own words say a line's price does not hold?
The fourth is about allocation. Eric Noreen, Conditions under Which Activity-Based Cost Systems Provide Relevant Costs (1991), contrasts traditional systems, which assign costs on one allocation basis proportional to volume, with systems that use several cost drivers, and finds the conditions under which the second give relevant costs quite stringent. Some of the seventeen records ask how unit prices should be allocated: by share of quantity, by share of cost, by year, by the number of runs a vehicle makes. The question is what a multiplication of quantity by price can say to a question about the basis of allocation.
The fifth is about labels. Richard H. Thaler, Mental Accounting Matters (1999), describes how the sources and uses of funds are labelled, how expenditures are grouped into categories and sometimes constrained by budgets, how often accounts are evaluated and how narrowly or broadly they are defined, each of which violates the principle that money is fungible. The records count money in four different units and over a month, a year or no stated period. The question is what the unit and the period written beside a total do to what the total says.
The debate between the debates
These works pull in different directions. Hou argues that the line item's convenience and conciseness explain its survival; Noreen finds that moving beyond a single volume basis gives relevant costs only under stringent conditions; Ellram, and Greenleaf and co-authors, ask what lies beyond the price a line shows. Tversky and Kahneman are concerned with the figure a judgment starts from, Flyvbjerg with the actual performance of comparable actions on which a projection can be based. Wanat and Thaler, one from budgeting and one from psychology, both look at the words and categories through which money is handled. This paper does not settle these disagreements. It uses them to read a record set that is exact about its arithmetic and plain about what it does not know.
How the paper proceeds
Chapter 1 gives a short account of the data and method and then two case sections: what the total becomes, and what stays outside the lines. An appendix gives the records read, word for word in this study's translation. The conclusion returns to the debates, states the contribution again, and names the limits of a reading built on seventeen authored synthetic cases.