Since 2004, the Urban Brookings Tax Policy Center has been analyzing the tax plans of presidential candidates. We are doing the same for the 2016 hopefuls, giving prospective voters our best estimates of the revenue, economic, and distributional effects of candidates’ tax proposals.

Our goal is to elevate the tax policy debate, not to criticize candidates. We are non-partisan and have no agenda other than to help voters better understand what the candidates are proposing and what those changes would mean. We think presidential campaigns are a fantastic opportunity to educate the public about tax policy issues when people are most likely to be paying attention.

Which plans do we choose to model? Given the plethora of presidential hopefuls this year and our limited staff time, it is not any easy choice. In general, we use two criteria: Plans need to have enough detail for us to model and candidates must have a high enough level of public support. This year, we started with former Florida governor Jeb Bush largely because his plan was so well specified and raised interesting policy issues that were worth exploring.

How do we do it? We start with the tax plan descriptions provided by each candidate. These are campaign documents, not bills, so they inevitably leave out some important details that we need to score the proposals. As a result, we ask the campaigns for additional information to help us flesh out and analyze their plans. Where we don’t get the detail we need, we have to fill in the blanks. We will make any assumptions clear in our analysis and may also discuss the implications of alternative assumptions. We stand ready to revise our assumptions whenever campaigns supply us with additional details. (We posted our rules of engagement with the campaigns here.)

Once we have enough information, we run the proposals through our microsimulation tax model, a tool we have been using and refining for more than a decade. The model uses data from the IRS and many other sources. It produces revenue estimates as well as a wealth of information on how tax changes affect households of different incomes.

Our model produces a traditional revenue estimate, which doesn’t account for changes in the overall economy, but does account for how people change the composition of their spending and their reporting of income in response to changes in tax incentives. The model is calibrated to match as closely as possible the official revenue estimates—before considering macro feedback effects—that would be produced by the Joint Committee on Taxation.  Unlike JCT, however, we do not currently have a model that estimates effects on total economic output. We do discuss the likely effects of proposals on the economy in our analysis.

Like any model, ours must often rely on judgment. TPC’s choices are grounded in the academic literature as well as original research—for example, statistically linking together data from multiple sources—and our own years of experience in research and government. We document our analysis and methodology in research papers and supplemental materials posted on our website. We occasionally update our model in light of new evidence. For example, we revised our assumptions about the share of the corporate income tax borne by workers relative to owners of capital. We also regularly update our model to reflect changes in the economy and in tax collections. And we explain how our model works and the data on which it’s based. Here is a brief description for those who want to learn more about the model.

Once we’ve modeled a candidate’s plan, we write up a detailed analysis. That work is carefully reviewed by our senior staff and then sent to outside tax experts for a final look. We ask those experts, who span the political spectrum, to review our work quickly for both substance and tone.

Finally, once their review is complete and we make any necessary changes, we make our analysis and supporting data available on our website and to news organizations and we blog on it on TaxVox.

Like making good wine, this process takes time—sometimes more than even we would like. It takes a while to find out exactly what the plan does and to model its effects accurately and, as noted, we subject our work to multiple levels of internal and external review.

But the 2016 campaign will go on for another 11 months, and we feel it is most important that we get our analysis right.