> For the complete documentation index, see [llms.txt](https://research.ambitiousimpact.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://research.ambitiousimpact.com/ambitious-impact-eval/cost-effectiveness-modelling/how-much-will-things-cost.md). # How much will things cost? * A lot of models out there have very sophisticated benefit modelling… and not much to speak of in terms of costings. Pay equal attention to costs and benefits. ## **Approaches to costing** | **Ingredients** | Bottom-up | Break down the intervention into components needed, and figure out how much they cost. | | -------------------------------------------------------------- | --------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | **Literature-based** | Top-down | Use costs stated in studies in the academic literature. | | **Reference-case** | Top-down | If the intervention is very similar to that delivered by existing organizations, it can be quicker to use their costings if the budget is | | **Guesswork** | Top-down | Judgment-based approach to roughly estimate costs for an intervention. | | Fair for BOTECs and cases where none of the above have worked. | | | If you have a lot of time and getting costings as close to real life as possible is a top priority, the following flow chart may be useful {% code overflow="wrap" %} ```mermaid graph LR A[1. Guesswork
What is a wide ballpark guess for what this should cost?] ---> B[2. Ingredients] ----> C[3. Does the ingredients estimate align with the ballpark guess?]--- D[Yes] D --->G[4. Does the ingredients estimate align with the costs paid by other actors for similar interventions?] G ---H[Yes] -->J[Done] G---I[No] -->F C---E[No]-->F[Double check ingredients, and go back to step 2 ] ``` {% endcode %} * Generally the most reliable method is an ingredients approach * A key exception may be a very popular and often used model where you have a lot of reliable data on real-life costings. In those cases re-inventing the wheel has lower returns. Stick to the reliable estimates you’ve found, and skip the ingredients step. * There may be good reasons why something does not align with existing costings, in particular if you are mapping out an innovative approach aimed at reducing costs (duh). ### Ingredients approach * “We can think of this like breaking down and evaluating the cost for each ingredient that goes into baking a cake” (International Rescue Committee, 2025, p.4). * “Program ingredients are the resources used to deliver program’s activities. They usually align with budget line items but can also include items not listed in the budget, such as donated goods. Knowing the total cost, unit cost, and number of units for each ingredient helps us see which resources drive most of the costs and how they contribute to differences in cost-efficiency or cost-effectiveness across programs.” (International Rescue Committee, 2025, p.4). * To carry out an ingredients approach costing break down an intervention to its core activities and cost categories, and include all elements needed to fulfil them. ### Literature-based * Carrying out a literature review to identify unit costs, program costs, or cost-effectiveness analyses to use as inputs in your model. * In my experience costs from pilots or RCTs are usually higher than costs at scale or by direct implementers ### Reference case * Not too different from literature-based approaches. * Reference cases can be a similar or equal program in a different country/year (see below on standardizing and adjusting) * You can use reference cases to either * estimate the overall sticker price of an intervention (per unit, or total, etc.) * find unit costs for shared ingredients or activities ### Guesswork * What it says on the tin. * Guesswork is similar to reference-cases, as you will usually be approaching a guess based on reference-class thinking. * See best practice on judgment and forecasting to get better at guesswork in models. * Use sparingly and for guidance only. Be transparent about why your guess is what it is. ## Standardizing and adjusting inputs * There are two key aspects that you should think about when carrying out costing * Time * Location * The priority here is **consistency across models.** There are various approaches to handling time and location differences in the literature. * The order in which you convert things to a unified time and location changes the outcome. * The approaches you use to align costings (e.g., inflation v. GDP deflator) * Choose the approach that most makes sense, and stick to it. #### Example approach (based on J-PAL) * Simplified approach attempting to get all costs into same currency and reported in one common year. * I am not 100% of how technically correct this is - if any economists are gasping right now do email me. ```mermaid graph LR A[Is the cost data in USD?] --NO-->B[Convert to USD using exchange rate for the year] A --YES---C[Is the USD cost data from the year you are reporting costs in?] B---C C--YES-->D[Report
noting currency and year] C--NO-->E[Adjust using a GDP Deflator] E--->D ``` * To convert from a CURRENCY to USD → * A GDP deflator tool can be built relatively easily. See below * Note this is for forward looking CEAs. ## Other best practice * Use standard assumptions across models to improve comparability * Time-discounting is also important (see here) ## References International Rescue Committee. (2025). *Cost Analysis Methods at the International Rescue Committee*. Dhailiwal, I. D., Duflo, E. D., Glennerster, R., & Tulloch, C. (2013). *Comparative cost-effectiveness analysis to inform policy in developing countries: A general framework with applications for education*. University of Chicago Press. J-PAL. (2020). *Conducting cost-effectiveness analysis (CEA)*. The Abdul Latif Jameel Poverty Action Lab (J-PAL). --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://research.ambitiousimpact.com/ambitious-impact-eval/cost-effectiveness-modelling/how-much-will-things-cost.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.