Bump hosted SVAR estimation window 2023Q2 -> 2025Q1

docs/index.html

@@ -106,7 +106,7 @@ <h2>Five model classes, side by side.</h2>
           <tr><th scope="row"><strong>Question class</strong></th><td>Selected near-term UK fiscal and macro scenarios</td><td>What drove the UK economy at the latest model-data vintage; short-run forecasts and revision narratives</td><td>US macro dynamics: monetary and fiscal shocks, quarter by quarter</td><td>Long-run incentives: labour supply, saving, the capital stock over decades</td><td>What a specific family pays and receives; distributional effects</td></tr>
           <tr><th scope="row"><strong>Method</strong></th><td>Structural macroeconometric emulator — Gauss–Seidel over 372 equations per quarter</td><td>Bayesian structural VAR — posterior sampling with zero + sign restrictions naming the shocks</td><td>Large-scale macroeconometric model — 284 equations solved under VAR expectations</td><td>Dynamic general equilibrium (OLG) — steady-state root-find and transition-path iteration</td><td>Static tax-benefit microsimulation — direct rule evaluation, no behavioural response</td></tr>
           <tr><th scope="row"><strong>Country</strong></th><td>UK</td><td>UK</td><td>US</td><td>UK (OG-UK)</td><td>UK &amp; US</td></tr>
-          <tr><th scope="row"><strong>Data vintage</strong></th><td>OBR <em>Economic and Fiscal Outlook</em>, March 2026</td><td>ONS, BoE and FRED series — coefficients estimated through 2023Q2; conditioned through the 2026Q1 data edge</td><td>The Fed's <code>model.xml</code> and LONGBASE database, April 2026</td><td>ONS, OBR and BoE national accounts as calibration targets; OG-UK 0.3.2 (from GitHub)</td><td>UK &amp; US statute; enhanced FRS 2023–24 microdata for UK population runs</td></tr>
+          <tr><th scope="row"><strong>Data vintage</strong></th><td>OBR <em>Economic and Fiscal Outlook</em>, March 2026</td><td>ONS, BoE and FRED series — coefficients estimated through 2025Q1; conditioned through the 2026Q1 data edge</td><td>The Fed's <code>model.xml</code> and LONGBASE database, April 2026</td><td>ONS, OBR and BoE national accounts as calibration targets; OG-UK 0.3.2 (from GitHub)</td><td>UK &amp; US statute; enhanced FRS 2023–24 microdata for UK population runs</td></tr>
           <tr><th scope="row"><strong>Evidence class</strong> (<a href="/validation/">numbers →</a>)</th><td>Validated for selected scenarios; published anchor is not out-of-sample validation</td><td>Validated replication for selected outputs</td><td>Validated software replication; substantive comparisons are approximate</td><td>Research prototype; calibrated counterfactual with no ground truth</td><td>Deterministic rule calculation; population estimates add survey uncertainty</td></tr>
           <tr><th scope="row"><strong>Reform scoring</strong></th><td>Yes — shocks to exogenous model variables; static-costing bridge from PolicyEngine reforms</td><td>No — the baseline/conditioning member: it reads the economy reforms are scored against</td><td>No — shock experiments only (funds-rate or fiscal shocks under VAR expectations); <code>score_reform</code> refuses <code>model="frbus"</code>, as no PolicyEngine reform bridge exists</td><td>Yes — a PolicyEngine policy translated into estimated tax functions</td><td>Yes — a PolicyEngine parameter + a new value, applied directly to households</td></tr>
           <tr><th scope="row"><strong>Cannot establish</strong></th><td>Arbitrary reform incidence or borrowing through the current adapter</td><td>The causal effect of a specific statutory reform</td><td>PolicyEngine reform effects, anticipated-policy paths, or MCE scenarios</td><td>A short-run forecast or independently validated reform effect</td><td>GDP, inflation, interest rates, or macro feedback</td></tr>

index.html

@@ -136,7 +136,7 @@ <h3>OBR macroeconometric model</h3>
         <h3>Bank of England structural VAR model</h3>
         <p>
           Diagnose the latest model-data vintage and forecast GDP and inflation
-          with uncertainty ranges. Coefficients are estimated through 2023Q2;
+          with uncertainty ranges. Coefficients are estimated through 2025Q1;
           the model does not score reforms.
         </p>
         <span class="strategy-paper-link mono">read the model →</span>

integration/src/policyengine_macro/capabilities.py

@@ -76,9 +76,9 @@
         "runtime": "minutes per estimation and identification run",
         "uncertainty": "posterior 68% and 90% intervals",
         "status": "validated replication for selected outputs",
-        "estimation_sample": "1992Q1-2023Q2",
+        "estimation_sample": "1992Q1-2025Q1",
         "data_edge": "2026Q1",
-        "data_vintage": "2026Q1 conditioning data; estimation ends 2023Q2",
+        "data_vintage": "2026Q1 conditioning data; estimation ends 2025Q1",
     },
     "frb-us": {
         "display_name": "Federal Reserve FRB-US implementation",

integration/src/policyengine_macro/core.py

@@ -725,17 +725,24 @@ def _covid_dummies(index) -> np.ndarray:
 # Estimation window for the hosted SVAR (start of Great-Moderation sample;
 # end held fixed so hosted results only move on a deliberate refresh). The
 # provenance strings are derived from the actual df_est index, never retyped.
+#
+# 2026-07-24 refresh: bumped 2023Q2 -> 2025Q1 (126 -> 133 observations) to match
+# the sample boe-var-model's production replication now runs on. Held one
+# quarter short of the 2026Q1 data edge, matching the repo, so the forecast
+# origin sits outside the estimation sample.
 _SVAR_EST_START = "1992Q1"
-_SVAR_EST_END = "2023Q2"
-
-# Default posterior draws for hosted calls. Measured on the 2026Q1 vintage
-# (Apple Silicon, single process): draws=500 -> 42 accepted, ESS 15.5, ~23s;
-# 1000 -> 76 accepted, ESS 26.5, ~44s; 2000 -> 165 accepted, ESS 63.6,
-# ~117s. Acceptance is ~8% and ESS ~3% of draws, so 2000 is the smallest
-# default that clears the 100-accepted-draws reliability threshold at
-# roughly two minutes of first-call runtime (results are cached in-process).
-# Full ESS >= 100 needs ~3500 draws (~3.5 min); a `warnings` entry flags the
-# residual ESS shortfall honestly instead of hiding it.
+_SVAR_EST_END = "2025Q1"
+
+# Default posterior draws for hosted calls. Re-measured on the 1992Q1-2025Q1
+# estimation sample (seed 0), since the longer sample tightens the posterior
+# and lowers the sign-restriction acceptance rate: draws=500 -> 35 accepted,
+# ESS 13.6; 1000 -> 69, ESS 31.5; 2000 -> 135, ESS 65.3; 3500 -> 233,
+# ESS 101.4. Acceptance is ~6.8% and ESS ~3.3% of draws (was ~8% and ~3% on
+# the old 2023Q2 sample), so 2000 remains the smallest default that clears
+# the 100-accepted-draws reliability threshold -- 1000 would not (69).
+# Full ESS >= 100 still needs ~3500 draws; a `warnings` entry flags the
+# residual ESS shortfall honestly instead of hiding it. First-call runtime is
+# a couple of minutes end to end and results are cached in-process.
 _SVAR_DEFAULT_DRAWS = 2000
 
 
@@ -2705,7 +2712,7 @@ def svar_summary() -> dict:
         "provenance": _provenance(
             model_id="boe-svar",
             distribution="boe-var-model",
-            data_vintage="estimation through 2023Q2; conditioning data through 2026Q1",
+            data_vintage="estimation through 2025Q1; conditioning data through 2026Q1",
             baseline="committed replication and forecast artefacts",
         ),
     }

integration/src/policyengine_macro/mcp_server.py

@@ -290,7 +290,7 @@ def frbus_summary() -> dict:
 def forecast_uk(horizons: int = 12, draws: int = 2000) -> dict:
     """Forecast UK YoY GDP growth and CPI inflation with the UK SVAR model.
 
-    Estimates a Bayesian VAR (1992Q1-2023Q2 sample, sign-identified structural
+    Estimates a Bayesian VAR (1992Q1-2025Q1 sample, sign-identified structural
     shocks) and simulates the predictive distribution from the latest data
     quarter. Returns, per future quarter, the median and 68%/90% bands for YoY
     GDP growth and YoY CPI inflation, both in percent. The response includes a
@@ -299,11 +299,11 @@ def forecast_uk(horizons: int = 12, draws: int = 2000) -> dict:
 
     Args:
         horizons: Forecast horizon in quarters (default 12 = 3 years).
-        draws: Posterior draws (default 2000: ~165 accepted draws, ESS ~64,
-            about two minutes of runtime on first call; ~3500 draws reaches
-            ESS >= 100 in ~3.5 minutes; 500 responds in ~25s but yields ESS
-            ~15 and a warning). Results are cached in-process, so repeated
-            calls with the same (horizons, draws) are instant.
+        draws: Posterior draws (default 2000: ~135 accepted draws, ESS ~65,
+            a couple of minutes on first call; ~3500 draws reaches ESS >= 100;
+            500 is faster but yields ~35 accepted, ESS ~14, and a warning).
+            Results are cached in-process, so repeated calls with the same
+            (horizons, draws) are instant.
     """
     return core.svar_forecast(horizons=horizons, draws=draws)
 

integration/tests/test_site_contract.py

@@ -14,10 +14,25 @@ def _read(path: str) -> str:
 
 
 def test_svar_site_uses_runtime_estimation_endpoint():
+    """The published estimation endpoint must match the one the server actually
+    estimates on.
+
+    Derived from ``_SVAR_EST_END`` rather than hardcoded: the point of this
+    guard is to catch the site drifting away from the runtime, and a literal
+    makes it fail on the refresh it is supposed to permit while still missing
+    drift in the other direction."""
+    from policyengine_macro.core import _SVAR_EST_END
+
     pages = _read("svar/index.html") + _read("docs/index.html")
-    assert "estimated through 2023Q2" in pages
-    assert "estimation sample to 2025Q1" not in pages
-    assert "Estimation to 2025Q1" not in pages
+    assert f"estimated through {_SVAR_EST_END}" in pages
+
+    # No other quarter may be presented as the estimation endpoint.
+    stale = {
+        q
+        for q in re.findall(r"estimat\w*\s+(?:through|to)\s+(\d{4}Q[1-4])", pages)
+        if q != _SVAR_EST_END
+    }
+    assert not stale, f"site quotes stale estimation endpoint(s): {sorted(stale)}"
 
 
 def test_og_is_not_labelled_as_us_model():

models/index.html

@@ -52,7 +52,7 @@ <h1 class="reveal page-title" style="--d:1">Choose the model that fits the quest
     <div class="band-head"><span class="kicker mono">all models</span><h2>Inspect the model before you use it.</h2><p class="section-intro">Each guide covers inputs, outputs, method, validation, limitations, and how to run the model.</p></div>
     <div class="strategy-grid">
       <a class="strategy-card" href="/obr/"><header><span class="card-id mono">obr-macro</span><span class="tag-row"><span class="tag tag-country">UK</span><span class="tag tag-hosted">hosted</span></span></header><h3>OBR macroeconometric model</h3><p><strong>Selected near-term scenarios.</strong> Trace GDP, consumption, and investment against the official forecast. Validated for selected shocks; borrowing is not yet returned.</p><span class="strategy-paper-link mono">open model →</span></a>
-      <a class="strategy-card" href="/svar/"><header><span class="card-id mono">boe-svar</span><span class="tag-row"><span class="tag tag-country">UK</span><span class="tag tag-hosted">hosted</span></span></header><h3>Bank of England structural VAR</h3><p><strong>UK conditions and forecasts.</strong> Identify shocks and forecast GDP and inflation with uncertainty ranges. Estimated through 2023Q2; conditioned through 2026Q1.</p><span class="strategy-paper-link mono">open model →</span></a>
+      <a class="strategy-card" href="/svar/"><header><span class="card-id mono">boe-svar</span><span class="tag-row"><span class="tag tag-country">UK</span><span class="tag tag-hosted">hosted</span></span></header><h3>Bank of England structural VAR</h3><p><strong>UK conditions and forecasts.</strong> Identify shocks and forecast GDP and inflation with uncertainty ranges. Estimated through 2025Q1; conditioned through 2026Q1.</p><span class="strategy-paper-link mono">open model →</span></a>
       <a class="strategy-card" href="/frb-us/"><header><span class="card-id mono">frb-us</span><span class="tag-row"><span class="tag tag-country">US</span><span class="tag tag-hosted">hosted</span></span></header><h3>Federal Reserve macroeconomic model</h3><p><strong>Selected US shocks.</strong> Run raw monetary and fiscal levers under VAR expectations. No reform bridge or model-consistent expectations.</p><span class="strategy-paper-link mono">open model →</span></a>
       <a class="strategy-card" href="/olg/"><header><span class="card-id mono">psl-og</span><span class="tag-row"><span class="tag tag-country">UK</span><span class="tag tag-local">local</span><span class="tag tag-experimental">experimental</span></span></header><h3>PSL overlapping generations</h3><p><strong>Long-run incentives and capital.</strong> Run local steady-state reform analysis; transition paths are available from the package. Calibrated, not outcome-validated.</p><span class="strategy-paper-link mono">open model →</span></a>
       <a class="strategy-card" href="/pe/"><header><span class="card-id mono">pe-microsim</span><span class="tag-row"><span class="tag tag-country">UK</span><span class="tag tag-country">US</span><span class="tag tag-hosted">hosted</span></span></header><h3>PolicyEngine microsimulation</h3><p><strong>Households and distribution.</strong> Calculate taxes, benefits, net income, and population impacts. Population results depend on survey data; no macro feedback.</p><span class="strategy-paper-link mono">open model →</span></a>

svar/index.html

@@ -137,7 +137,7 @@ <h2>Eight variables, six named shocks.</h2>
     <div class="prose">
       <p>
         A VAR in eight quarterly variables — three global, five UK — estimated
-        over the inflation-targeting era — <strong>1992Q1–2023Q2</strong> in
+        over the inflation-targeting era — <strong>1992Q1–2025Q1</strong> in
         both the paper replication and the hosted adapter. The hosted forecast
         conditions on observations through <strong>2026Q1</strong>, but does
         not re-estimate its coefficients on those later observations. It is estimated in
@@ -178,7 +178,7 @@ <h2>Eight variables, six named shocks.</h2>
         so medians and bands are correct but noisier than the paper's. The
         hosted forecast conditions on the current vintage — data through
         2026Q1 — with 6,000 draws (430 accepted, ESS 197.9), while retaining
-        the 2023Q2 coefficient-estimation endpoint.
+        the 2025Q1 coefficient-estimation endpoint.
       </p>
     </div>
   </section>
@@ -368,7 +368,7 @@ <h2>Where this departs from the Bank's model.</h2>
         <tbody>
           <tr>
             <th scope="row"><strong>Frozen data edge</strong></th>
-            <td>Coefficient estimation to 2023Q2; conditioning data to 2026Q1 (refreshed July 2026). Results shift with data revisions; the replication claims are scored on the paper's original sample.</td>
+            <td>Coefficient estimation to 2025Q1; conditioning data to 2026Q1 (refreshed July 2026). Results shift with data revisions; the replication claims are scored on the paper's original sample.</td>
           </tr>
           <tr>
             <th scope="row"><strong>Proxied world aggregates</strong></th>