Abstract
Past projects can help a development team understand the work ahead. They can reveal relevant approval paths, recurring questions, and the sequence of agency decisions. Their value depends on how the record is assembled and interpreted. A list of completed approvals does not describe every project that entered the process, and a reported entitlement duration may omit work performed before an application was deemed complete. Drawing on official reporting definitions, statistical methods, and Prebreak's research architecture, this report proposes a framework for selecting and presenting entitlement comparables. The objective is decision support that makes its evidence and limits visible, rather than precision unsupported by the underlying record.
Method and scope
This report combines a review of California Department of Housing and Community Development reporting guidance, established methods for incomplete time-to-event data, and Prebreak's implementation and research instructions at revision d468d56. The statistical literature is used to explain general measurement principles; its application to entitlement is our methodological interpretation. No new statewide dataset was assembled, no outcome model was fitted, and no accuracy or software-savings result is reported here. The worked example is entirely synthetic.
This third report complements Prebreak's perspectives on approval pathways and research architecture by examining the evidence used to compare one project with another.
1. Start with the question the record can answer
A comparable should be selected for a decision. A developer considering an approval strategy may need examples of the same application type. A team preparing a submission may need to understand which studies appeared in relevant cases. A financing discussion may require a clearly defined stage duration. These are different questions, even when they draw on the same public records.
HCD describes one primary purpose of its Annual Progress Report data as tracking local progress toward Regional Housing Needs Allocation goals, with issued building permits as the key benchmark. It provides downloadable records and a dashboard. That makes APR data a valuable starting point for housing activity research, but its reporting purpose is not a calibrated prediction of an individual project's chance of approval. 1
A useful comparison therefore begins by naming the unit of analysis: an application, a project, a housing unit, a phase, or an agency action. The research protocol should resolve repeated records and changing identifiers before counting cases. Otherwise, a single project appearing at several stages can receive more weight than the analyst intended. This is a proposed data-preparation requirement, not a claim that the current pipeline fully resolves every such relationship.
2. Define the clock before quoting the duration
HCD's APR instructions define the submitted-application date by reference to the jurisdiction's completeness determination. If the initial submission was incomplete, the reported date is when it became complete. A duration beginning at that date does not include all earlier application preparation or correction work. 2, pp. 5, 9–10
That distinction is material to Prebreak's coordination thesis. A project team may spend time resolving assumptions, replacing drawings, and collecting missing information before the statistical clock begins. Measuring only the later stage cannot establish whether a tool improved that earlier work. This is an implication of the field definition, not an empirical estimate of how much time is omitted.
For any timeline, a report should identify its start event, end event, observation date, sample, and exclusions. “Application complete to entitlement” and “first design work to construction” should never appear as interchangeable measures. If the underlying dates cannot be reconciled, preserve the uncertainty rather than producing a precise duration from an ambiguous match.
3. Keep the reporting population visible
HCD's Table A provides approved, pending, disapproved, and withdrawn status options. Table A2 records housing activity involving an entitlement, building permit, or occupancy/completion milestone. A2 should not be described as an exhaustive register of all attempts to obtain approval, nor simply as a table containing only entitlement approvals. 2, pp. 11–12
Prebreak's reviewed timeline query specifically selects records with a populated entitlement approval date and summarizes matched application and entitlement dates. That selection describes the observed durations of the matched records. It cannot, on its own, establish the probability that another application will be approved. The implementation also returns cautions with its record summaries. Those cautions are a useful design choice, but claims about the extent of missing outcomes require a separately documented data-quality study. 3
The distinction between a reporting system and a particular query matters. An approved-only sample may arise from an analyst's filter even when the source system supports other statuses. Evaluating the data requires inspecting both the source definitions and the extraction logic.
4. Pending projects are unfinished observations
A project still under review at the observation date has an elapsed duration but no observed entitlement date. In time-to-event terminology, a still-eligible case whose future event time is not yet observed can be right-censored. NIST explains censoring through reliability studies; the same measurement concept is useful when thinking about an incomplete project history. The application to entitlement still requires a defensible study design. 4
Consider four hypothetical applications measured from a common start definition:
| Synthetic application | Observed status | Recorded time | Interpretation |
|---|---|---|---|
| A | Entitled | 8 months | A completed entitlement duration |
| B | Entitled | 10 months | A completed entitlement duration |
| C | Still pending | 14 months so far | Elapsed follow-up, not a final duration |
| D | Withdrawn | Withdrawal at 6 months | A different observed outcome |
The two completed approvals have a conventional median of nine months. That calculation is correct for those two observations. It is not the typical completion time of all four applications, and it does not determine the eventual outcome of application C. Counting two approvals out of four also fails to establish a new project's eventual approval probability while follow-up and outcome definitions remain unresolved.
Table 1. Original synthetic illustration. These are invented teaching records, not Prebreak customer results, an APR extract, or a software benchmark.
Withdrawal and disapproval need their own treatment. Depending on the unit and endpoint, they may end an application episode, while an appeal or a new submission begins another transition. Statistical literature distinguishes competing events from ordinary censoring and shows why treating them identically can distort cumulative outcome estimates. The appropriate model depends on the question; this report does not propose applying a clinical model to planning records without validation. 5
5. Match the approval path before widening the geography
Nearby projects can be poor comparables if they followed materially different rules. California's official guidance distinguishes ministerial and discretionary decisions and explains that housing projects may qualify for exemptions or streamlining. A historical discretionary case is therefore not sufficient evidence for how a different, currently eligible ministerial path will operate. 6
Prebreak's research instructions prioritize the subject jurisdiction and relevant entitlement type, then permit a wider search when local evidence is sparse. This is a research heuristic, not a demonstrated predictive ranking. The instructions also call for attention to changes in the legal or decision-making environment. 3
Our proposed comparison record makes the rationale inspectable:
| Dimension | What to document |
|---|---|
| Approval path | Application type and the basis for treating it as relevant |
| Decision context | Responsible jurisdiction and body, plus material changes since the case |
| Project scope | Use, scale, requested deviations, and other characteristics important to the question |
| Time period | Decision dates and whether the governing framework remains comparable |
| Outcome evidence | Dated determination or other authoritative record, including unresolved appeals when relevant |
| Selection rationale | Why the case was included, where it differs, and why the search widened |
A small, well-explained set can be useful for identifying questions. It should not be presented as a statistically representative sample merely because it is the best material available.
6. Deliver a case record the team can use
The proposed output is an evidence-backed case record: what was proposed, which path was pursued, what the agency decided, which conditions or issues mattered, and why this case is relevant to the current project. Each material statement should remain linked to its supporting source and date.
This makes the comparison useful beyond an initial report. A team can revisit the rationale when its proposal changes, distinguish a procedural requirement from a negotiable condition, and identify the next question for staff or counsel. Historical similarity supports investigation; it does not override current law or replace a project-specific assessment.
The product opportunity is the disciplined transformation of fragmented records into reviewable decision context. Demonstrating that opportunity requires measuring the quality and usefulness of the resulting records, not simply the number of documents retrieved.
Evaluation agenda and limitations
A first evaluation could use a permission-cleared set of applications whose histories are independently reconstructed by qualified reviewers. It should assess project matching, date interpretation, source support, pathway classification, and preservation of unresolved outcomes. A separate evaluation could test whether practitioners find the selected cases relevant to an actual decision.
If forecasting is later studied, the protocol should fix an observation date, prevent the use of future information, document follow-up, and compare performance with a suitable baseline. Publishing ranges does not remove the need for these checks: an interval calculated from a selected sample still inherits that sample's limitations.
This report establishes a methods framework, not a validated forecast or a finding about statewide denial rates. The practical standard is straightforward: every comparison should explain what happened, why it is relevant, and what remains unknown. That is how public records become a useful part of entitlement planning.
References
Reference 1
California Department of Housing and Community Development. Annual Progress Reports. Reporting purpose and public data access. Accessed September 10, 2026.
Reference 2
California Department of Housing and Community Development. Housing Element Annual Progress Report Instructions. Current posted instructions, especially printed pages 5 and 9–12. Accessed September 10, 2026. Definitions should be reconciled with the instructions applicable to each historical reporting year.
Reference 3
Prebreak. Internal implementation and research-instruction audit, revision d468d56, September 10, 2026. Sources include the APR query tool, peer-jurisdiction guidance, and California decision paradigm. Implementation and design evidence, not independently measured predictive performance.
Reference 4
NIST/SEMATECH. e-Handbook of Statistical Methods, §8.1.3.1: Censoring. General statistical reference; entitlement application is the authors' interpretation. Accessed September 10, 2026.
Reference 5
Austin, P. C., Lee, D. S., and Fine, J. P. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Circulation 133(6), 601–609 (2016). doi:10.1161/CIRCULATIONAHA.115.017719. Statistical methods reference, not a study of housing applications.
Reference 6
California Governor's Office of Land Use and Climate Innovation. Site Check and Other CEQA Housing Resources. Ministerial/discretionary distinctions and housing pathways. Accessed September 10, 2026.