Inputs

Every assumption, calibration choice and measured series behind the four-quarter forecast. Nothing here is re-estimated when the site serves.

Download assumptions.csv observations.csv (full history, model units) measurement_mapping.csv key_items.csv (forecast, all regimes)

Scenario

Input	Value	Unit	Note
Data vintage	2026-09-03		Local panel; nothing is downloaded or re-estimated when the site serves
Nowcast quarter	2026Q3		CPI, wages and unemployment observed through 2026Q2, GDP and fiscal ratios through 2026Q1; 2026Q3 has partial financial data only
Forecast horizon	2026Q4–2027Q3	quarters	Four quarters after the nowcast
Inflation target	4	% y/y	Fixed exactly (3.9221 annualised log pp); no target shock
Supply shock	1.5	annualised log pp	Direct headline effect in 2026Q3 via a fuel-price wedge, held in 2026Q4, reversed in 2027Q1; fully anticipated from 2026Q3
Fuel CPI weight	4.64	% of CPI	Released CBR calibration

Supply-shock schedule

Quarter	Direct headline inflation, annualised log pp	Headline price-level wedge, log points	Fuel price wedge, log points
2026Q3	1.500	0.375	8.082
2026Q4	0.000	0.375	8.082
2027Q1	−1.500	0.000	0.000
2027Q2	0.000	0.000	0.000
2027Q3	0.000	0.000	0.000

Measured data, latest quarters

Readable units; the fit uses 100·log levels (see observations.csv)

Series	Unit	2024Q4	2025Q1	2025Q2	2025Q3	2025Q4	2026Q1	2026Q2	2026Q3	Last observed
Unemployment rate	%	2.30	2.37	2.25	2.17	2.17	2.17	2.15	—	2026Q2
Real wages	% y/y	9.72	3.72	5.80	5.96	5.37	8.97	5.32	—	2026Q2
Headline CPI (SA)	% y/y	9.04	10.06	9.69	8.25	6.62	5.89	5.65	—	2026Q2
Core CPI (SA)	% y/y	8.49	9.48	8.92	8.03	6.18	5.13	4.87	—	2026Q2
Fuel CPI (SA)	% y/y	1.38	1.68	1.69	1.57	1.75	1.63	2.07	—	2026Q2
Real GDP (SA)	% y/y	4.46	1.65	1.38	0.86	0.34	0.06	—	—	2026Q1
RUONIA	%	20.19	21.03	20.66	18.05	16.28	15.28	14.28	14.04	2026Q3
RUB per USD	RUB	99.61	92.42	80.94	80.56	79.89	78.33	74.51	80.81	2026Q3
Foreign policy-rate proxy	%	3.97	3.64	3.38	—	—	—	—	—	2025Q2
Brent oil price	USD/bbl	74.66	75.87	68.07	69.03	63.65	80.72	102.63	87.36	2026Q3
2-year OFZ yield	%	20.29	17.44	16.49	13.75	14.45	14.43	13.43	14.53	2026Q3
5-year OFZ yield	%	18.19	16.26	15.71	14.00	14.86	14.77	14.45	15.65	2026Q3
Federal oil and gas revenue	% of GDP, 4q	5.50	5.27	4.90	4.46	3.96	3.37	—	—	2026Q1
Federal non-oil revenue	% of GDP, 4q	12.64	12.72	13.02	13.24	13.44	13.54	—	—	2026Q1
Federal expenditure	% of GDP, 4q	19.86	20.48	20.93	21.26	20.03	20.69	—	—	2026Q1

A quarter can rest on incomplete monthly coverage. Unemployment is not seasonally adjusted. Year-on-year rates are computed from quarterly levels.

Measurement mapping

Model variable	Source	Sample	Error SD	Note
`UR`	U_OBS	2015Q1–2026Q2	0.1	Percentage rate, NSA; seasonal/definition risk remains
`WR`	W_OBS - CPI_OBS	2015Q1–2026Q2	0.5	100 log nominal SA wage minus 100 log SA CPI; equilibrium/gap remain latent
`LCPI`	CPI_OBS	2015Q1–2026Q2	0.1	100 log SA headline CPI
`LCPI_C`	CPIC_OBS	2015Q1–2026Q2	0.15	100 log SA core CPI; component definition needs forecast-stage review
`LCPI_FUEL`	100 log CPI_FUEL_SA_LEVEL	2015Q1–2026Q2	0.5	100 log SA fuel CPI; source component weight retained
`Y`	Y_OBS	2015Q1–2026Q1	0.5	100 log SA real GDP
`RS`	RUONIA_OBS	2015Q1–2026Q3	0.1	Annual percentage rate; policy-rate proxy
`LS_USD`	S_OBS	2015Q1–2026Q3	0.1	100 log RUB per USD; not the basket rate
`RSF`	55% US/45% euro-area policy-rate proxy	2015Q1–2025Q2	0.2	55% US/45% euro-area policy-rate proxy
`LPOIL_USD`	Brent USD oil-price proxy	2015Q1–2026Q3	1	Brent USD oil-price proxy
`RSG2`	panel:OFZ_2Y, matching maturity	2015Q1–2026Q3	0.2	panel:OFZ_2Y, matching maturity
`RSG5`	panel:OFZ_5Y, matching maturity	2015Q1–2026Q3	0.2	panel:OFZ_5Y, matching maturity
`REV_O_ARAT`	100 * rolling-four-quarter BUD_REV_OILGAS / nominal GDP	2015Q1–2026Q1	0.5	100 * rolling-four-quarter BUD_REV_OILGAS / nominal GDP
`REV_NO_FED_ARAT`	100 * rolling-four-quarter BUD_REV_NONOIL / nominal GDP	2015Q1–2026Q1	0.5	100 * rolling-four-quarter BUD_REV_NONOIL / nominal GDP
`EXP_FED_ARAT`	100 * rolling-four-quarter BUD_EXP_TOTAL / nominal GDP	2015Q1–2026Q1	0.5	100 * rolling-four-quarter BUD_EXP_TOTAL / nominal GDP

FX regimes

Input	Value	Note
Switch	2026Q4	Unexpected permanent change beginning Q4 2026
Common history	2026Q3	Identical Q3 2026 filtered posterior in all regimes; no re-estimation
Current fitted setup	smoothing 0; fiscal FX channel on; no jump	The fitted model with no change to its FX equation. Capital-control weight w_cc = 0.75 as calibrated.
Fiscal FX channel off	smoothing 0; fiscal FX channel off; no jump	From 2026Q4, new fiscal oil-revenue terms no longer feed the FX-specific risk premium. Other fiscal and premium channels remain.
Moderate FX smoothing	smoothing 0.5; fiscal FX channel on; no jump	From 2026Q4 the UIP equation gains support of −0.5 × the quarterly change in 100·log RUB/USD. Illustrative smoothing, not an intervention volume.
Strong FX smoothing	smoothing 2; fiscal FX channel on; no jump	As moderate smoothing, with coefficient 2.0.
95, then constant real basket	smoothing 0; fiscal FX channel on; jump to 95 RUB/USD	In 2026Q4 RUB/USD jumps to 95; the model's native real basket exchange rate is then held at its achieved level. Not a nominal peg or a US-CPI real anchor.
Limitations		FX support is an equation wedge without calibrated reserves, funding or welfare cost. Fiscal premium decomposition depends on latent fiscal states.

Oil scenario

Input	Value	Unit	Note
Slider range	40–130	USD/bbl, average 2026Q4–2027Q3	Average Brent price over 2026Q4–2027Q3, set by a parallel log shift of the fitted model's own Brent path
Model oil path average	82.19	USD/bbl	Geometric mean of the fitted Brent path, 2026Q4–2027Q3; the slider default
Carrier	LTOT_GAP		Innovations to the released oil terms-of-trade gap (LTOT_GAP), delivered through four announcement states in 2026Q4
Announcement	2026Q4		Announced unexpectedly in 2026Q4; the whole 2026Q4–2027Q3 path is then known to all agents
After 2027Q3			No further announcements; LTOT_GAP decays at rho_tot_gap = 0.8 per quarter
Unchanged			Equilibrium oil price, foreign output and prices, EUR/USD, the fiscal base oil price and all other equations follow the fitted model
Uncertainty			Hosted CPI intervals are the regime intervals shifted with the central path; conditioning on a known oil path is not reflected in their width
Limitations			Linear in log Brent: no budget-rule, sanctions-discount or other non-linearities. Oil moves only through its cyclical gap; a permanent oil price change would move the fiscal base price and equilibrium terms of trade instead.

Labour calibration

Input	Value	Unit	Note
NAIRU anchor	4.5	%	Calibrated centre, not an empirical estimate
NAIRU persistence	0.9	per quarter	Stationary around the anchor
NAIRU innovation SD	0.025	pp	Drift innovations disabled
Unemployment-gap persistence	0.7	per quarter
Okun slope	0.45	pp per 1% output gap	Exploratory; not a CBR estimate
Bargaining tightness	1.5	% wage per 1pp shortfall	Exploratory; not a CBR estimate
Hiring-cost tightness	0.375	% cost per 1pp shortfall	Exploratory; not a CBR estimate
Unemployment-gap shock SD	0.15	pp

Estimation

Input	Value	Note
Historical fit	2015Q1–2026Q3	Kalman filtered states; source coefficients fixed except grid-selected labor slopes
Selected labour multiplier	1.5	Prior selected labor multiplier 1.5 held fixed; NAIRU repair refits states only
Omitted foreign observations	YF, LCPIF, LCPIF_C	Remain latent in the fit
Diffuse initial variance	10000	Unit-root directions; stationary directions use unconditional moments

Supplementary shock SDs

Input	Value	Unit	Note
RES_EXP_DEBT_FED_ARAT	0.1	model units	Not supplied by the CBR release
RES_REV_WF_FED_ARAT	0.25	model units	Not supplied by the CBR release
RES_X_O_GAP	1	model units	Not supplied by the CBR release
10 seasonal reporting shocks	0	model units	RES_DOT4_M_SF, RES_DOT4_W_SF, RES_DOT4_X_SF, RES_DOT4_Y_SF, RES_PIE4_SF, RES_PIE_C4_SF, RES_PIE_FUEL4_SF, RES_PIE_SREG4_SF, RES_PIE_VEG4_SF, RES_PIE_VOL4_SF

Provenance

Input	Value	Note
Quarterly panel SHA-256	`816112857df01ac32c227d2b37b3ad3bb10dcc15e7c3f57c55ed0311fe30b53d`
Fitted posterior SHA-256	`f4b66bb9c7493ddb3d341cabb830b127456f2a5170cc57e4f752233d46f004ab`
Oil multipliers SHA-256	`383ef0133bb4081d545ef9674c9f37d5909bac58c01f2b585ea957796981d9f9`
CBR source model/external_model.mod	`88e2629110f3c533f7ce60225d85f9510a61b8ca7193191a3072bf4ea117d1ee`	SHA-256
CBR source model/parameters_main.mod	`1f48bf59ce710057902dbd99ce4972d67b5bac96498ea84e7d81dc5ef680f840`	SHA-256
CBR source model/qpm_fiscal_block.mod	`05a76884f1b1b31616fca75b1236287f752fbd9b8c42952c06e10ae808635633`	SHA-256
CBR source model/qpm_model.mod	`845a8c857f0bff9b1fa4fdcea24cf3a491a4d425298a1117ad3c1f22092d8e0d`	SHA-256
CBR source model/qpm_term_structure.mod	`1398314d4a6fec3997c72ab452d244743890eb23e3a9eff789b479be175acde2`	SHA-256
CBR source proc/get_ss_growth_rates.m	`2912e75edf7ad952ccc9c0cc266583c842b22d44300d273038cab42bcb2cfdf1`	SHA-256
Native MATLAB/IRIS parity	not verified
Release	cb-model2-keyitems-20260910T101525Z