How much time does regulation really give?

Regulation gives organisations time in at least three different ways. A draft may provide early warning. A consultation deadline may provide time to respond. Publication of the final instrument may begin a separate period in which policies, systems, contracts, controls, and staff must be changed before the rule applies. Treating those periods as one number would obscure the operational question the research is meant to answer.

Research status. This is a protocol and open methodology draft. It contains no findings. Monitorscape will publish results only after the date semantics, source coverage, lifecycle links, exclusions, and validation sample described below have been completed.

The operational problem

Regulatory monitoring often answers what changed? Regulatory operations must also answer how much usable time remains? The second question is harder because the start and end of the clock depend on the decision being made.

A public-affairs team can act when a draft first becomes visible. A compliance team may need the adopted text before it can finalise controls. A regulated organisation may face an effective date, a transitional provision, or a later sector-specific deadline. These moments can be separated by months, compressed into days, or occur on the same date.

Romanian law itself demonstrates why the intervals must be separated. Article 7 of Law no. 52/2003 on decisional transparency requires an announcement about a draft normative act at least 30 working days before it is submitted for endorsement and a written-comment period of at least 10 calendar days. The same article permits earlier adoption when an urgent or exceptional situation requires an immediate solution.

The clock changes again after adoption. Article 78 of the Romanian Constitution states that a law enters into force three days after publication or on a later date in its text. Article 12 of Law no. 24/2000 distinguishes laws and ordinances issued under an enabling law, emergency ordinances, and other normative acts. Emergency ordinances may enter into force on publication, subject to their prior submission to the competent chamber, while other instruments may also apply on publication unless they specify a later date.

This legal heterogeneity is part of the phenomenon being measured. It is not noise to be averaged away.

Four windows, not one

The study models a regulatory lifecycle with four event dates. A date enters the analysis only when its meaning and provenance are known.

Event	Operational definition	Minimum evidence
T₀ — first public disclosure	The earliest verified date on which the specific proposal or draft was publicly accessible through an official institutional surface.	Dated official page, announcement, register entry, or source document.
T₁ — consultation close	The stated deadline for written comments, or the documented close of a consultation when no deadline is available.	Source deadline preserved as a date, with the source URL or document.
T₂ — final publication	Publication of the adopted normative instrument in the authoritative venue applicable to that instrument.	Official publication identifier and publication date.
T₃ — analytical endpoint	The declared endpoint for the analysis: either whole-instrument entry into force or the application date of a defined obligation. The two endpoint types are not pooled.	Explicit provision in the final text or a documented default rule appropriate to the instrument type.

From these events we derive four measures:

Wconsult = T1 − T0   consultation opportunity
Wvisible = T2 − T0   public visibility before final publication
Wpost    = T3 − T2   post-publication preparation interval
Wtotal   = T3 − T0   total observable notice-to-effect interval

The primary operational measure is Wtotal, because it represents the maximum observable preparation period available to an organisation that detected the first public signal. It is not calculated unless T₀ and T₃ are both supported.

The other windows explain where time was available. Two instruments can have the same total window while creating very different operating conditions: one may spend most of its lifecycle in consultation and then apply immediately; another may move quickly to publication but include a long transition period.

Durations are reported in calendar days unless a measure is explicitly labelled as working days. Date-only fields use endpoint subtraction, so events on the same date have a duration of zero; the study does not infer hour-level precision. Testing a procedural requirement expressed in working days uses a versioned Romanian holiday calendar and a separately documented counting rule.

Effective dates require instrument-specific rules

The final date cannot be imputed with one universal “publication plus three days” rule. The instrument must first be classified, then checked for an explicit entry-into-force clause, transitional provision, or obligation-specific date. Only when the relevant legal default is unambiguous may the study derive T₃, and every derived T₃ will retain the rule used to generate it.

The first findings paper will use whole-instrument entry into force as its T₃ endpoint. Obligation-level windows will be analysed separately only after the relevant population of obligations and the rule for selecting them have been specified. If an instrument contains multiple obligation dates, each coded milestone will retain its provision and scope. The report will not describe the full instrument as having a single compliance deadline when only one provision has been dated.

Why Romania is a useful first case

Romania combines formal consultation rules with multiple legislative and executive pathways. That makes it useful for studying the difference between a procedural opportunity and the preparation time organisations actually observe.

The OECD’s 2025 country profile for Romania reports that Law no. 52/2003 requires ministries to publish draft laws and subordinate regulations for comments, while the minimum comment period remains 10 days. It also reports that executive-initiated proposals represent about 39% of Romanian primary laws in the indicator scope. This means an executive-only dataset cannot stand in for the whole legislative system, and a parliamentary-only dataset cannot describe the upstream executive process.

The study therefore begins with two cohorts:

1. Executive cohort: draft normative acts and related institutional actions observed on ministries, agencies, regulators, and other executive sources monitored by Monitorscape.
2. Parliamentary cohort: legislative proposals and dated procedures observed across the Senate and Chamber of Deputies lifecycle.

The cohorts will be reported separately. A proposal may move from an executive consultation into Parliament and eventually into the Official Gazette, but that connection will be used only when a stable identifier, official reference, or reviewed match supports it.

What Monitorscape records can support

Monitorscape retains a longitudinal trace rather than only the latest status. Executive records include the issuing institution, instrument type, publication state, source link, and dated actions. Parliamentary records include chamber identifiers, procedure dates, event dates, deadlines, procedural text, chamber, urgency indicators, and publication references when available.

Three kinds of date must remain distinct:

• Source event date: the date an institution assigns to an action, procedure, deadline, or publication.
• First observation date: the first time Monitorscape captured the record or event.
• System modification date: the time the stored record was refreshed or changed.

The study uses a source event date for event-time analysis. A first observation date can support a separate detection latency analysis, but it cannot silently replace an absent source date. A modification timestamp describes the data pipeline, not the regulatory lifecycle.

This distinction prevents a common measurement error. If an institution publishes an undated page on 10 May and Monitorscape first observes it on 12 May, 12 May is evidence that the page existed by that date. It is not evidence that the public consultation began on 12 May.

Cohort construction

The study population will be fixed before outcome distributions are calculated. The pre-analysis specification will record the observation period, covered institutions, source interruptions, and the query or data release used to produce the cohort.

Inclusion rules

A record is eligible when it:

1. concerns a proposed or adopted act with general normative effect;
2. has a stable Monitorscape identity and an official source reference;
3. belongs to an institution and source surface with audited coverage during the observation period; and
4. contains at least one verified lifecycle event relevant to a defined window.

Eligibility for a particular measure is stricter than eligibility for the study. For example, a record with T₀ and T₂ can contribute to Wvisible but cannot contribute to Wtotal until T₃ is known.

Exclusion rules

The main cohort will exclude individual administrative decisions, appointments, procurement notices, meeting announcements without a normative instrument, undated secondary commentary, and records that cannot be resolved to a stable source object. Duplicate pages and corrected republications will be consolidated under a documented precedence rule.

Still-active proposals are not treated as failed data. They remain right-censored at the study cut-off because their later outcome is not yet observed. Withdrawn and rejected proposals are retained as competing terminal outcomes and reported separately. Removing any of these records would bias the study toward instruments that reached publication.

Urgent pathways

Urgency is an analytical stratum, not an exclusion. Records using an emergency ordinance, an urgent parliamentary procedure, or the exception in Article 7(13) of Law no. 52/2003 will be identified where the official record supports the classification. Their timing distributions will be shown separately before any pooled estimate is considered.

Linking one proposal across institutions

The unit of analysis is a regulatory lifecycle, while the source data arrives as pages, documents, actions, and procedures. Linking those objects is the study’s largest source of potential error.

The linkage process uses a hierarchy:

1. Deterministic identifiers: Official Gazette number, parliamentary proposal number, notification identifier, or an explicit official cross-reference.
2. Exact structured references: an official page that names the preceding or subsequent instrument.
3. Reviewed candidate match: normalised title, institution, instrument type, subject, and compatible dates produce a candidate that a reviewer accepts with recorded evidence.
4. Unlinked: similarity without sufficient evidence remains unresolved and does not enter cross-stage estimates.

Automated title similarity can propose a match; it cannot make the match authoritative. Amendments, omnibus instruments, renamed drafts, and multiple proposals with similar titles make title-only matching especially unsafe.

The final release will report linkage yield and reviewed precision. A timing distribution without the number of unresolved lifecycles would give a false impression of completeness.

Analysis plan

The study is descriptive. It measures the distribution and composition of observable preparation time; it does not infer that a longer interval caused better compliance.

For every window, the report will publish:

• eligible records and records with both endpoints;
• median, interquartile range, 10th and 90th percentiles;
• the empirical cumulative distribution, so readers can see the share below operational thresholds such as 7, 14, 30, and 90 days;
• a survival estimate that retains active right-censored lifecycles and treats withdrawal or rejection as competing outcomes where appropriate;
• missingness for each endpoint and the reason it is missing;
• results by instrument type, institution class, policy area, urgent pathway, and calendar period when cell sizes permit; and
• sensitivity analyses using alternative defensible definitions of T₀ and T₃.

Means will be secondary because a small number of very long procedures can dominate them. No institution ranking will be published until differences in instrument mix, source coverage, and censoring have been assessed.

Pre-specified comparisons

The first report will examine three questions:

1. How often does the legally operative publication-to-effect window leave fewer than 3, 7, 14, or 30 calendar days?
2. How much of the total observable window occurs before the final text is available?
3. How do ordinary, urgent, executive, and parliamentary pathways differ in both the amount and composition of preparation time?

These are comparisons, not assumptions about what the results will show.

Validation and quality controls

Before analysis, Monitorscape will draw a stratified validation sample across institutions, years, instrument types, and apparent window lengths. Two reviewers will independently verify event meaning and source evidence. Disagreements will be adjudicated and used to refine the coding guide.

Automated checks will flag:

• impossible event order, such as an effective date before final publication;
• a consultation deadline before public disclosure;
• conflicting dates across an HTML page, attachment, and institutional register;
• a derived effective date where the instrument specifies a different one;
• duplicate publication identifiers assigned to unrelated lifecycles; and
• source events that use a scrape or modification timestamp as their regulatory date.

The quality appendix will publish agreement rates, error categories, missingness by source, and corrections made after validation. Source pages can change or disappear, so the internal evidence package will retain the source URL, observation time, extraction method, and a content fingerprint where permitted.

How to interpret the eventual results

A longer preparation window is not automatically better regulation. Long delays can reflect unresolved policy, while a short window can be appropriate for a narrow or beneficial change. The measure describes time availability; it does not measure legal quality, policy merit, administrative burden, or actual organisational readiness.

The first public signal also represents observable notice, not proof that every affected organisation became aware of the proposal. Monitorscape’s detection time can be measured separately from the institution’s publication time, but neither establishes when an individual firm began implementation.

The OECD’s best-practice principles for regulatory impact assessment place implementation, enforcement, compliance strategy, stakeholder engagement, and monitoring within the regulatory policy cycle. This study contributes one operational measure to that broader assessment. It should be read alongside evidence on impact analysis, consultation quality, clarity, proportionality, and enforcement.

Publication gate and planned release

The findings paper will be released only when the following conditions are met:

• cohort and source coverage are frozen and documented;
• event-date definitions pass the validation sample;
• lifecycle linkage precision and unresolved-match rates are reported;
• effective-date extraction distinguishes explicit provisions from legal defaults;
• urgent and censored pathways are identifiable;
• aggregate results can be reproduced from a versioned analysis package; and
• disclosure review confirms that every public row and source link can be shared lawfully.

The release package will contain the findings article, technical appendix, data dictionary, aggregate result tables, reproducible calculation code, source-coverage matrix, and correction policy. Where source licensing permits, it will also include a row-level research extract with provenance and uncertainty fields.

Further reading

1. Law no. 52/2003 on decisional transparency in public administration, consolidated text, Romanian Legislative Portal.
2. Law no. 24/2000 on legislative drafting technique, consolidated text, Romanian Legislative Portal.
3. Constitution of Romania, Article 78, Romanian Legislative Portal.
4. Better Regulation Practices across the European Union 2025: Romania, OECD.
5. Regulatory Impact Assessment: Best Practice Principles for Regulatory Policy, OECD, 2020.