## 1. Russia’s exports composition and competitiveness

## Source details

**Canonical URL:** [1. Russia’s exports composition and competitiveness](https://www.imf.org/-/media/files/publications/cr/2017/cr17198.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2017/cr17198.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2017/cr17198.pdf.json)

---

### A. Introduction and summary
- Lower commodity prices can trigger real depreciation and potentially unwind Dutch disease, but macro-level recovery of Russian tradable sectors has not materialized.
- Two factor sets explaining muted non-commodity export response:
  - (i) disruptiveness of sudden terms-of-trade driven devaluations; and
  - (ii) external demand and access to external markets.
- Key observations:
  - The recent oil price collapse was primarily driven by supply factors; global demand remained reasonably strong over the last two years, yet Russia’s main trading partners (CIS) performed relatively poorly—markets where Russia has a high concentration of non-commodity exports.
  - Long-term trends in diversification have stalled and have not been accompanied by a move into more sophisticated products.
- Policy recommendations (Section F):
  - Insulate the non-commodity sector from oil price volatility.
  - Structural reforms to support reallocation of resources across sectors.
  - Improve penetration of global markets.
  - Ensure the financial system can support reallocation even during periods of stress.

### B. Export elasticities during terms-of-trade driven depreciations
- Methodology:
  - Elasticities of manufacturing exports w.r.t. REER and trading partner growth estimated via standard panel regression on first differences of log manufactured exports, data averaged over three year periods (interpreted as medium-term elasticities).
  - REER lagged one period.
  - Export Commodity Price Index (ECPI) from Gruss (2014) used to interact with ∆ Ln REER to estimate elasticities separately when ECPI is rising versus falling.
- Empirical findings:
  - Baseline (61 advanced and emerging markets): REER elasticity is negative and external demand elasticity is positive, both significant at the 1 percent level.
  - When ECPI is falling, elasticity of manufacturing exports w.r.t. REER is close to zero.
  - For commodity-exporting countries (commodity exports >20 percent of all export or >10 percent of GDP):
    - Elasticity w.r.t. REER is nearly twice as large when commodity prices are on the rise than when on the decline.
    - Elasticity during commodity price declines is not statistically significant from zero.
  - Extreme swings (annualized ECPI swings >2.5 percent, 90th percentile): differences between upswings and downswings are more dramatic in commodity exporters.
- Mechanisms explaining asymmetric elasticity:
  - Learning-by-doing externalities (Krugman (1987)): commodity booms can drive marginal tradable industries out; subsequent depreciation may not restore capacity.
  - Hysteresis and sunk costs (Krugman (1989)): uncertainty about exchange rates delays fixed exporting costs and re-entry after depreciation.
  - Commodity–banking channel: linkages between commodity and non-commodity sectors can stress the financial sector during commodity-price driven depreciation, restricting lending needed by tradables.

### C. Banking stress and non-performing loans (NPLs)
- Stylized evidence from the recent oil shock:
  - Russian banks experienced an increase in NPLs of a scale uncommon among the 20 largest EMs; a few countries with equally large oil-to-exports ratios showed similar asset-quality deterioration.
  - NPL evolution (p.p. change since 2014 Q1) highlights notable deterioration for Russia relative to peers.
- Consequences:
  - Lending becomes restricted and expensive just as non-commodity tradable industries need to invest to take advantage of depreciation.
  - Established manufacturers may finance expansions internally, but tight credit can prevent new entry.
- Supporting empirical studies:
  - Kinda et al. (2016): negative commodity-price shocks weaken financial sectors of commodity-exporting EMs; larger shocks yield higher NPLs, higher bank costs, reduced bank profits, liquidity, and provisions.
  - Agarwal et al. (2016): for 46 commodity-dependent LICs, falling commodity prices reduce bank lending via deteriorated bank capitalization.
- Data caveat:
  - NPL increases in Russia (and India) may partially reflect large-scale banking cleanups; in some oil-rich countries deterioration may be obscured by loan restructurings (evergreening).

### D. Global demand, commodity prices, and Russia’s divergence (2014–16)
- Trading partner growth is positively and statistically significantly correlated with the export commodity price index for 29 commodity-exporting countries.
- The recent oil price drop was linked primarily to supply factors; depressed demand did not prevent some countries from taking advantage of improved price competitiveness.
- Focusing on 2014–16, trading partner growth showed little correlation with commodity prices (the 2014–16 regression line is virtually flat) — trading partner growth stayed roughly at long-time averages.
- Russia’s divergence:
  - Russia’s 2014–2016 data points lie some 1–1½ percentage point below the corresponding regression line.
  - Russia is at the bottom of the distribution of trading partner growth among major EMs.
  - Russia’s manufacturing exports performed relatively poorly in 2013–15.
- Common pattern:
  - All EM commodity exporters are well below the regression line—non-commodity tradable industries of commodity exporters face an uphill battle during commodity price collapses even controlling for trading partner growth.

### E. Russia’s export geography, market access, and trade restrictiveness
- Export geography:
  - CIS markets account for just over ½ percent of global GDP but absorbed 12 percent of Russia exports in 2013 and some 28 percent of manufacturing exports.
  - Growth in CIS countries is strongly correlated with Russia’s growth.
  - Russia’s manufactured exports registered particularly large drops on CIS markets; excluding Ukraine, manufactured exports to other CIS members decreased by over 30 percent (more than twice the drop on non-CIS markets).
  - Russia’s growth is more correlated with that of its export partners than its trade openness would suggest.
- Market access and regional agreements:
  - Russia has RTAs only with other EAEU members and Serbia; the 1994 CIS FTA was never ratified by Russia; Russia terminated the 2011 CIS FTA in 2016.
  - By RTA partners’ share in world GDP, Russia is last among the twenty top economies considered.
  - Ongoing efforts to establish bilateral agreements (e.g., with India and Vietnam) and planned RTAs could support exporters and integration into global value chains.
- Trade restrictiveness:
  - Using MA-OTRI (Kee et al. (2009)), Russia faces relatively low tariff and overall trade restrictions on its external markets.
  - Caveat: MA-OTRI may reflect selection bias driven by markets and export baskets.

### F. Structural transformation: margins of export growth and export sophistication
- Export growth decomposition (Zahler (2007)):
  - Over the medium term, Russia’s manufacturing exports grew in a balanced way: new products contributed 14 percent of total exports growth (top quintile).
  - Russia managed to grow exports by introducing products to new markets despite market access constraints.
  - During 2013–2015 (falling oil prices), commodity price-driven ToT shocks were not conducive to structural change: neither Russia nor other commodity exporters introduced new products; few made significant market gains.
- Export sophistication (EXPY):
  - EXPY decreased significantly between 2001 and 2015 for Russia and most commodity exporters.
  - EXPY recomputed for manufactured exports only: Russia registered negative evolution in sophistication of the manufactured exports basket.
  - Interpretation: structural transformation in commodity-exporting countries is difficult (consistent with Krugman (1987)).

### G. Outlook and policy implications for export competitiveness
- Factors blunting competitiveness effects when REER shock driven by unfavorable commodity prices:
  - Stress from worsened ToT impedes reallocation to the non-commodity tradable sector via uncertainty and banking sector weaknesses.
  - ToT-driven depreciations usually coincide with reduced trading partner growth, lowering demand for non-commodity exports.
  - Low trading partner growth reduces incentives to introduce new products, limiting structural transformation.
- Non-commodity tradable sector suffers from overvalued exchange rate during booms, but busts do not enable rapid reversal.
- Policy recommendations for Russia (to offset structural handicaps):
  - Attenuate effects of commodity price swings on the non-commodity sector through highly counter-cyclical fiscal policy — the new mechanism introduced by the Ministry of Finance is noted as welcome.
  - Ensure product and labor market regulations support resource reallocation in response to price signals; may require strengthening social safety net.
  - Strengthen regional and multilateral trade relations to improve market penetration and integration into global value chains.
  - Ensure the financial system is healthy to reallocate credit to new sectors even during external stress.

### H. Annex II — Export Growth Decomposition: methodology (high-level)
- Decomposition separates intensive margin (surviving product-destination combinations) and extensive margin (new products and/or new destinations) following Zahler (2007).
- PD matrix conceptualization: ~200 columns (countries) × ~5000 rows (products, HS).
- Interpretation highlights:
  - Sum of surviving PDs and extinct PDs = intensive margin.
  - Other categories = extensive margin with product and destination sub-dimensions.
  - Longer intervals increase the relative importance of the extensive margin.

### I. Box 1 — Russia’s previous fiscal frameworks, benchmarking, and simulations (key findings)
- History of oil-related funds:
  - Oil Stabilization Fund (OSF) introduced 2004; OSF reached US$157 billion at end-2007.
  - OSF abolished in 2008 and split into Reserve Fund (initial US$25 billion) and National Welfare Fund (initial US$32 billion).
  - Balances after decade of high oil prices: NWF US$73 billion; Reserve Fund declined from US$125 billion in early 2008 to US$16 billion as end-2016.
- 2008 fiscal rule:
  - Targeted long-term non-oil fiscal deficit of 4.7 percent of GDP to be achieved by 2011; suspended during global financial crisis and abolished in 2012.
- 2013 expenditure rule (abandoned 2015) weaknesses:
  - Expenditure ceiling tied to oil revenues measured at a benchmark price; benchmark adjusted slowly (example: 3-year moving average yielded ~US$85pb vs actual US$42pb in 2016).
- Modifications proposed to strengthen fiscal rule:
  - Use futures oil prices in benchmark to allow faster adjustment (with caveats).
  - Target a surplus informed by long-term fiscal considerations.
  - Add a spending growth cap (not exceed estimated long-term real growth) to avoid pro-cyclicality.
  - Strengthen escape clauses to cap withdrawals from the reserve fund and complement with borrowing constraints.
  - Consider targeting a structural non-oil balance once sufficient data are compiled.
- Determining a benchmark oil price (Box 2):
  - Forecasting oil prices is difficult; futures help but may overweight current prices.
  - Proposed US$40 pb (50-year average) reduces pro-cyclicality but may leave observed prices far from benchmark for long periods.
  - Governance: independent committee recommended to set benchmark and review the formula.
- Simulations (FSGM counterfactual 2010–16 and scenario projections):
  - Rules evaluated:
    - Old rule (authorities’ old rule suspended in 2015).
    - New rule (authorities’ proposed conservative US$40 pb benchmark).
    - Staff’s proposed rule (futures-based benchmark; surplus target of 1 percent of GDP).
  - General counterfactual finding:
    - Consistent fiscal rules would have made fiscal policy more countercyclical, increased savings, and lowered debt-to-GDP ratios.
    - Russia would have more assets than liabilities under all fiscal rules: 16 percent on average, compared to actual liabilities of 4 percent at end-2014.
  - Comparative projected non-resource primary deficits (series for 2010–2016):
    - Old rule: 7.4, 6.6, 7.3, 7.5, 8.8, 12.3, 11.4
    - New rule (benchmark@40): 3.8, 3.5, 3.7, 3.6, 4.0, 5.6, 5.3
    - Proposed rule: 7.1, 6.8, 6.3, 5.7, 6.1, 7.3, 5.0
    - Baseline: 11.2, 8.4, 9.4, 9.3, 9.5, 9.0, 8.9
  - Under the old rule counterfactual: least savings, highest spending during high oil prices; Reserve Fund would have run down to 3 percent of GDP; gross debt ratcheting to 15 percent of GDP by end-2016.
  - Staff’s proposed rule aims for faster adjustment and more savings via a surplus target.
- Forward simulations (2018–2022 baseline projected non-resource primary deficits):
  - Old rule: 5.6 5.5 5.8 5.6 5.5 5.4 5.4
  - New rule (benchmark@40): 4.7 4.7 4.6 4.5 4.5 4.4 4.4
  - Proposed rule: 4.2 4.1 4.0 4.0 4.1 4.1 4.1
  - Baseline: 6.5 5.6 4.8 4.8 4.9 4.6 4.6
- Oil price scenarios (2018–2022):
  - Baseline: 55.4 55.2 55.5 56.1 59.3
  - Persistent Low Oil: 30.0 38.7 40.3 42.8 46.8
  - Temporary Low Oil: 30.0 42.9 48.5 56.6 57.6
  - Persistent High Oil: 80.0 71.8 71.8 70.6 72.9
  - Temporary High Oil: 80.0 67.5 62.3 55.5 60.8
- Key conclusion on rules:
  - Staff’s proposed rule is preferred given considered shocks: allows faster adjustment to persistent low prices while avoiding excessive pro-cyclicality when shocks are temporary; new rule accumulates largest buffers under persistently high prices but risks rigidity if benchmark is mis-specified.

### J. Annex I — FSGM for Russia (model features relevant to simulations)
- Model highlights:
  - Annual, multi-economy, forward-looking FSGM; economies structurally identical except for commodities.
  - For Russia: oil dominates (most exports; oil production affects potential output one-for-one).
  - Households: overlapping-generations plus liquidity-constrained types.
  - Investment: Tobin’s Q; firms are net borrowers.
  - Trade: reduced-form driven by competitiveness and demand; competitiveness improves one-for-one with domestic prices.
  - Commodities: three commodities (oil, metals, food); demand and supply price-inelastic in short run.
  - Inflation and monetary policy: Phillips curves with REER weight; inflation-forecast-based monetary rule.
- Oil benchmarking observations:
  - 50-year average (1966–2015) oil price in real terms (US$ 2015) = US$42.6.
  - Authorities’ proposed benchmark: 50-year average in real terms (freeze relative price at end-2015).
  - Using long MAs reduces pro-cyclicality but can leave observed prices far from benchmark for long periods.
  - 50-year MA through 2021 (using futures) ≈ US$46/barrel in 2015 terms vs US$42.6 in 2015 terms for 1966–2015.
  - Relative price of oil to US consumption basket rose ~2.3 percent per year (1972–2015); futures as of July 27, 2016 imply ~1.3 percent annual increase through 2021.

### K. Fiscal federalism, regional transfers, and subnational finances (key findings)
- Legal framework:
  - Russia’s legal framework scores higher de jure on co-determination of federal policies, fiscal rules, and fiscal constitution stability than the average of advanced and EM economies, but de facto variability exists (rules changed/suspended).
  - Fiscal constitution classified as integrated/centralized.
- Regional revenues and transfers:
  - Regional revenues = own revenues + federal transfers; federal taxes (PIT and CIT) are largest source of regional fiscal revenue (~70 percent of own revenues).
  - Equalization grants constitute about 50 percent of federal transfers.
  - Federal Medical Insurance Fund transfers to Territorial Medical Insurance Funds represented 1.7 percent of GDP in 2016.
- Transfers’ impacts:
  - Transfers reduced cross-regional spending dispersion and supported increases in health and education real per capita spending.
  - Regions receiving higher average transfers experienced higher investment-to-GRP ratios and higher physical capital growth; very high investment ratios in some regions suggest initially low capital stocks.
  - Empirical results: transfers have not positively impacted regional fiscal sustainability; transfers increased the size of the public sector, with negative indirect impacts on GRP growth outweighing direct positive impacts.
  - Quantitative illustrative effect: a one-standard deviation difference in transfers (~17 percent of regional GRP) associated with a negative cumulative bilateral difference in real per capita GRP growth (2005–15) of around 1.2 percentage points, increase in public sector share of GRP of ~1.5 percentage points, and own revenue-to-expenditure ratio change around -0.1 percentage point.
- Persistence and policy implications:
  - Regional dependence on transfers likely to decrease only slowly; complete elimination of regional dispersion is unlikely.
  - Policy options: include sustainability measures in grant allocation formulas; increase horizontal transfers in the margin to counteract concentration; streamline and increase transparency of transfers (reduce number of subsidies, allocate by program appendices).
  - Strengthen regional tax bases (e.g., expand personal property taxes—currently 0.4 percent of consolidated own revenues) to improve accountability; 28 regions started transition to market-value-based property taxation in 2016; Moscow projects a five-fold increase in property tax collections by 2020 (collection increased 55 percent in 2016).
- Subnational fiscal rules and limits:
  - Regional deficit cannot exceed 15 percent of own revenues (excluding grants).
  - Debt not allowed to exceed own annual revenues (excluding grants); debt service should not exceed 15 percent of total expenditures (excluding subventions).
  - Borrowing constraints differ for regions receiving equalization transfers; new foreign borrowing limited by credit-rating and arrears conditions.

### L. Time-varying Phillips Curve for Russia (summary and policy relevance)
- Motivation and findings:
  - Unemployment muted since 2013; inflation volatile; transition to Inflation Targeting (IT) in late 2014 motivates time-varying Phillips curve analysis.
  - Hybrid NKPC with time-varying coefficients estimated on quarterly data 2000Q1–2016Q3.
  - Main findings:
    - The core inflation Phillips curve is "alive"; slope expected to increase with recovery.
    - Impact of cyclical unemployment on core inflation changes over time: tends to increase in normal times and decrease after crises.
    - Weight on inflation expectations in PC has increased recently, reflecting improved anchoring with IT adoption.
    - Model-implied unemployment slack has been coming down fast post-GFC.
- Specification and estimation:
  - Model includes unemployment gap, long-run inflation expectations (5-year forward WEO-based), lagged core CPI inflation, lagged relative import price inflation, and lagged REER; time-varying coefficients follow random walks.
  - Estimation via Constrained Extended Kalman Filter with sign and parameter constraints.
- Constant-coefficient regression fit (summary statistics):
  - R-square: 0.87; Adjusted R-Squared: 0.86; Root Mean Squared Error: 1.43.
  - Estimated coefficients (constant model example):
    - Lagged unemployment: -0.3686 (SE 0.195, p 0.0635)
    - Inflation Expectation: -0.1258 (SE 0.1324, p 0.3456)
    - Lagged inflation: 0.9586 (SE 0.0605, p 0)
    - Import inflation: -0.0255 (SE 0.0145, p 0.0831)
    - REER: -0.0079 (SE 0.0095, p 0.41)
- Time-varying results and policy implications:
  - Weight on long-run expected inflation increased since 2012; still below advanced-economy levels (e.g., US ~0.7).
  - Slope steeper in Russia than average EMs—monetary policy should monitor slack carefully.
  - Import-price inflation weight rose until sanctions onset; from 2015 onward the effect of import prices on inflation diminished.
  - Policy relevance: time-varying NKPC supports anchoring expectations via credible IT and monitoring of slack as recovery proceeds.

*Source: cr17198 (IMF staff excerpts).*

### 1. Russia’s exports composition and competitiveness __________________________________ 4

### 1. Russia’s exports composition and competitiveness

### A. Introduction and summary
- Lower commodity prices can trigger real depreciation and potentially unwind Dutch disease, but macro-level recovery of Russian tradable sectors has not materialized.
- The paper explores two sets of factors explaining the muted non-commodity export response:
  - (i) disruptiveness of sudden terms-of-trade driven devaluations; and
  - (ii) external demand and access to external markets.
- Key observations:
  - The most recent oil price collapse was primarily driven by supply factors, so global demand remained reasonably strong over the last two years, yet Russia’s main trading partners (CIS) performed relatively poorly—markets where Russia has a high concentration of non-commodity exports.
  - Structural transformation results are mixed: long-term trends in diversification have stalled and have not been accompanied by a move into more sophisticated products.
- Policy recommendations (Section F):
  - Efforts to insulate the non-commodity sector from oil price volatility;
  - Structural reforms to support reallocation of resources across sectors;
  - Initiatives to improve penetration of global markets;
  - Measures to ensure that the financial system can support reallocation of resources even during periods of stress.

### B. Export elasticities during terms-of-trade driven depreciations
- Methodology:
  - Elasticities of manufacturing exports with respect to the REER and trading partner growth estimated using a standard panel regression (first differences of log manufactured exports), data averaged over three year periods (interpreted as medium-term elasticities).
  - REER is lagged one period to alleviate endogeneity concerns.
  - Export Commodity Price Index (ECPI) from Gruss (2014) used to interact with ∆ Ln REER to estimate elasticities separately when ECPI is rising versus falling.
- Empirical findings:
  - Baseline results for a set of 61 advanced and emerging markets: REER elasticity is negative and external demand elasticity is positive, both significant at the 1 percent level.
  - When ECPI is falling (emerging from a period of falling commodity export prices), the elasticity of manufacturing exports w.r.t. REER is close to zero.
  - For commodity-exporting countries (commodity exports represent over 20 percent of all export or over 10 percent of GDP):
    - The elasticity of manufacturing exports w.r.t. REER is nearly twice as large when commodity prices are on the rise than when they are on the decline.
    - The elasticity during commodity price declines is not statistically significant from zero.
  - Results for extreme swings: column 3 focuses on periods when annualized swings in ECPI exceed 2.5 percent (corresponds to the 90th percentile in the three-year change in ECPI); differences between upswings and downswings are more dramatic in commodity exporters.
- Mechanisms and literature explanations for asymmetric elasticity:
  - Krugman (1987): learning-by-doing externalities—during commodity booms marginal tradable industries are driven out, foreign competitors gain an advantage, and a subsequent real depreciation may not restore lost capacity.
  - Krugman (1989): hysteresis and sunk costs—uncertainty about future exchange rates delays incurring fixed exporting costs; depreciation may be viewed as temporary overshooting, delaying re-entry.
  - Commodity–banking channel: deep linkages between commodity and non-commodity sectors can stress the financial sector during a commodity-price driven depreciation, restricting lending when the tradable sector needs investment.

### C. Banking stress and non-performing loans (NPLs)
- Stylized evidence from the recent oil shock:
  - Russian banks experienced an increase in NPLs of a scale uncommon among the 20 largest EMs, though a few countries with equally large oil-to-exports ratios showed similar deterioration in asset quality.
  - Figure evidence: NPL evolution (p.p. change since 2014 Q1) highlighted Russia with notable deterioration relative to peers.
- Consequences:
  - Lending becomes restricted and expensive just as non-commodity tradable industries need to invest to take advantage of depreciation.
  - Established manufacturers may finance expansions internally, but tight credit can prevent new entry into the sector.
- Supporting empirical studies:
  - Kinda et al. (2016): negative shocks to commodity prices tend to weaken financial sectors of commodity-exporting emerging and developing markets; larger shocks have more pronounced impacts (higher non-performing loans, higher bank costs, reduced bank profits, liquidity, and provisions).
  - Agarwal et al. (2016): at the bank level for 46 commodity-dependent LICs, falling commodity prices reduce bank lending by domestic banks via deteriorated bank capitalization.
- Data note:
  - Some bias may exist in the NPL data: Russia and India were involved in large-scale banking cleanups, so increases may also reflect more stringent supervision; in some oil-rich countries, deterioration may be obscured by loan restructurings (evergreening).

*Prepared by Alexander Culiuc. Source: Russia—Selected Issues (chapter excerpt).*

### 15.      Changes in commodity prices don’t happen in a vacuum. An important driver is global

### 15.      Changes in commodity prices don’t happen in a vacuum. An important driver is global

### Global demand and commodity prices
- Trading partner growth is positively and statistically significantly correlated with the export commodity price index for a group of 29 commodity-exporting countries (Figure 4).
- The recent drop in oil prices has been linked primarily to supply factors; depressed demand did not appear to prevent countries from taking advantage of improved price competitiveness.
- Focusing on 2014–16, there is little correlation between trading partner growth and commodity prices (the 2014–16 regression line is virtually flat in Figure 4) — trading partner growth stayed roughly at the level of long-time averages.

### Russia’s divergence during the 2014–16 oil price decline
- Russia’s trading partners performed significantly worse than those of the average commodity exporter: Russia’s 2014–2016 data points lie some 1–1½ percentage point below the corresponding regression line (Figure 4).
- Russia’s trading partners performed roughly in line with what the longer-term correlation would suggest (i.e., Russia’s low trading partner growth is not an outlier relative to the long-term relation).
- Russia is at the bottom of the distribution of trading partner growth among major EMs (Figure 5, left panel), along with a few European EMs, despite Russia being much less dependent on the EU’s slowly-recovering economy.
- Russia’s manufacturing exports performed relatively poorly in 2013–15 (Figure 5, right panel).

### Common pattern for EM commodity exporters
- All EM commodity exporters are well below the regression line, supporting the result that non-commodity tradable industries of commodity exporters generally face an uphill battle during periods of commodity price collapses, even when controlling for trading partner growth.

### Russia’s export geography and consequences
- CIS markets account for just over ½ percent of global GDP, but they absorbed 12 percent of Russia exports in 2013 and some 28 percent of manufacturing exports.
- Growth in CIS countries is strongly correlated with Russia’s growth because those countries are either commodity exporters or dependent on exports to or remittances from Russia.
- Russia’s manufactured exports registered particularly large drops on CIS markets (left chart in figure 6).
- Excluding exports to Ukraine does not change the overall result: manufactured exports to other CIS members decreased by over 30 percent, more than twice the drop registered on non-CIS markets.
- Russia’s growth is much more correlated with that of its export partners than its trade openness would suggest (right chart in figure 6).

### Market access and regional agreements
- Russia has no free trade agreements beyond its neighbors; Russia currently has regional trade agreements (RTAs) only with other EAEU members and Serbia.
- The Eurasian Economic Union (EAEU) includes Armenia, Belarus, Kazakhstan, Kyrgyz Republic and the Russian Federation; it provides for free movement of goods, services, capital and labor, as well as coordinated, agreed or common policy in different areas. The bilateral agreement with Serbia applies to select goods only.
- The 1994 CIS free trade agreement was never ratified by Russia; Russia terminated the 2011 CIS Free Trade Agreement in 2016.
- Historically, according to the WTO database, none of the large EMs had RTAs with countries representing more than 1 percent of global GDP. Since the mid-nineties, participation in RTAs has grown dramatically for most large EMs, but Russia is a rare exception with virtually no preferential access to major markets and no meaningful change over nearly three decades (Figure 7).
- By the metric of RTA partners’ share in world GDP, Russia is last among the twenty top economies considered (figure 7, right chart).
- Ongoing efforts to establish bilateral agreements (e.g., with India and Vietnam) and plans for other RTAs included in announced development strategies could support Russian exporters and multinationals contemplating including Russia in global value chains.

### Trade restrictiveness faced by exporters
- Using MA-OTRI (Market Access Overall Trade Restrictiveness Index) from Kee et al. (2009), Russia faces relatively low tariff and overall trade restrictions on its external markets (Figure 8).
- Caveat: MA-OTRI may suffer from selection bias because it is affected by the markets to which countries export and the basket of goods exported, which are endogenous and partly driven by trade barriers (e.g., low estimated barriers for Russian manufacturers may reflect high shares of exports to EUEA).

### Structural transformation: margins of export growth
- Export growth decomposition (Zahler (2007) methodology) separates contributions of intensive margin (more of old products to old markets) and extensive margin (new products and/or new destinations).
- Over the medium term, Russia’s manufacturing exports grew in a balanced way: new products contributed to 14 percent of total exports growth, placing Russia in the top quintile of analyzed countries.
- Despite market access constraints, Russia managed to grow exports by introducing products to new markets.
- During the 2013–2015 period associated with falling oil prices, commodity price-driven ToT shocks were not conducive to structural change: neither Russia nor other commodity exporters managed to introduce new products, and few made significant gains on external markets.

### Export sophistication (EXPY)
- EXPY (Hausmann et al. (2007)) tracks export sophistication. Russia’s EXPY decreased significantly between 2001 and 2015, along with most other commodity exporters (Figure 10).
- EXPY weights are dollar-based, so ToT shocks mechanically change EXPY even if volumes remain constant (commodities typically have low PRODYs). To control for this, EXPY was recomputed for manufactured exports only.
- EXPY for manufactured exports shows the overall picture unchanged: Russia, along with most commodity exporters, registered a negative evolution in the sophistication of the manufactured exports basket (Figure 11).
- Interpretation: structural transformation in commodity-exporting countries is difficult (in line with Krugman (1987)).

### Outlook and policy implications
- Structural reforms facilitate resource reallocation after a negative REER shock, but when the REER shock is driven by unfavorable commodity prices, several factors blunt competitiveness effects:
  - The stress from worsened terms of trade (ToT) impedes reallocation to the non-commodity tradable sector; uncertainty and banking sector weaknesses are two channels.
  - Episodes of ToT-driven depreciation usually coincide with reduced trading partner growth, lowering demand for non-commodity exports.
  - Low trading partner growth reduces incentives to introduce new products, limiting opportunities for large structural transformation.
- The non-commodity tradable sector suffers from an overvalued exchange rate during commodity booms, but busts are not conducive to a rapid reversal.

Policy recommendations for Russia (to compensate for structural handicaps):
- Attenuate the effects of commodity price swings on the non-commodity sector through highly counter-cyclical fiscal policy — the new mechanism introduced by the Ministry of Finance is a welcome step in this direction.
- Ensure product and labor market regulations are conducive to reallocating resources in response to price signals (2016 Article IV Selected Issues Paper). This may need to be accompanied by strengthening the social safety net.
- Strengthen regional and multilateral trade relations to allow greater penetration of foreign markets by Russian entities and to facilitate Russia’s integration into global value chains.
- Ensure the financial system is healthy enough to shift credit to new sectors even during periods of external stress.

*Source: IMF staff analysis in the Russian Federation country report chapter reproduced from the supplied PDF content.*

### Annex II. Export Growth Decomposition: Methodology and

### Annex II. Export Growth Decomposition: Methodology and Extended Results

### Methodology
- The decomposition analyzes export growth along the intensive and extensive margins:
  - Intensive margin: growth due to exporting “more of the same” (surviving product-destination combinations).
  - Extensive margin: growth due to new products and/or new destinations (product and market dimensions).
- The full product-destination (PD) space is conceptualized as a matrix with some 200 columns (countries) and some 5000 rows (products in the Harmonized System).
- The exercise follows Zahler (2007) methodology.

### Stylized example (conceptual illustration)
- Initial situation (2001): country filled only 7 cells of the PD matrix by exporting 4 products (P1–P4) to 3 countries (A–C). Export values within cells sum to 26.
- Outcome (2015): exports doubled to 52 in value.
- Contributions to export growth between 2001 and 2015:
  - Surviving PDs (old goods to old destinations): 42 percent of growth.
  - Filling cells within the old potential PD space (new PD, old space): 12 percent of growth.
  - Old products to new destinations (quadrant II): 19 percent of growth.
  - New products to old destinations (quadrant III): 31 percent of growth.
  - New products to new destinations (quadrant IV): 4 percent of growth.
  - Death of old PD combinations (extinct PDs): negative contribution of 8 percent.

### Interpretation of results
- The sum of surviving PDs and extinct PDs represents the intensive margin of export growth.
- The remaining categories (new PD in old space; new P old D; new D old P; new P and D) form the extensive margin, with product and destination sub-dimensions.
- The relative importance of these five margins indicates the degree of experimentation by exporters and the economy’s ability to capture new business and engage in structural transformation.
- The decomposition can only be performed between two points in time:
  - Shorter intervals: larger contribution from surviving PDs (intensive margin), since few products/markets are added each year.
  - Longer intervals: extensive growth plays a more prominent role because it incorporates all new PDs added in intervening years and their interim growth.

### Detailed results and applications
- Figure 2 presents full decompositions for Russia, focusing on different periods (for example, pre-GFC boom vs. post-GFC) and different bases of analysis (all exports vs. manufacturing exports only).
- Panel charts in Figure 3 compare Russia to other large EMs and a small sample of advanced economies across several time spans and for both all exports and manufacturing exports.
- The decomposition categories consistently reported in the figures are:
  - Surviving PD
  - New PD, old space
  - New P, old D
  - Old P, new D
  - New P, new D
  - Dead PD
- Across country comparisons and period splits (examples shown in the figures include 2001/15, 2001/08, 2008/15, 2001/03, 2013/15, and 2008/15 for manufacturing), the decomposition illustrates heterogeneity in export growth drivers (intensive vs. various extensive margins) across economies and time periods.

### Practical implications
- Decomposition results can:
  - Diagnose whether export growth is driven mainly by intensive-margin expansion or by extensive-margin experimentation (new products/destinations).
  - Inform policy on supporting diversification (if extensive margin contributions are low) or on improving competitiveness/scale (if intensive margin dominates).
  - Help interpret the speed and nature of structural transformation in exporters by quantifying contributions of new products and new markets.

*Source: cr17198 - Annex II. Export Growth Decomposition: Methodology and Extended Results.*

### Box 1. Russia’s Previous Fiscal Frameworks

### Box 1. Russia’s Previous Fiscal Frameworks

### History and design of oil-related funds (2004–2008)
- Russia established an Oil Stabilization Fund (OSF) in 2004 to save windfall oil revenues—export duties and the mineral extraction tax—and shield the budget from oil price fluctuations.
- Oil revenues above a cut-off price: US$20 pb in 2004–05; US$27pb in 2006–07 would be accumulated in the OSF.
- OSF balances above US$20 billion would be freely usable.
- Despite heavy use, the OSF reached US$157 billion at end-2007.
- Lack of a fiscal rule meant no targets for fiscal balance or limits on new borrowings, and the OSF did not prevent pro-cyclical fiscal policy.
- Non-oil and gas federal deficit evolution:
  - 2.9 percent of GDP in 2002
  - 5.1 percent of GDP in 2007
  - 6.5 percent of GDP in 2008
- As part of 2008 reforms the OSF was abolished and two funds created:
  - Reserve Fund (initial balance of US$25 billion) to smooth public spending against oil price fluctuations.
  - National Welfare Fund (initial balance of US$32 billion) to finance long-term pension liabilities.
- Windfall saving rule: oil revenue windfalls saved in the RF until it reached 7 percent of GDP; 50 percent of the excess then accrues to the NWF and the remainder finances infrastructure and other priority projects.
- Balances after a decade of high oil prices:
  - NWF at US$73 billion
  - Reserve Fund declined from US$125 billion in early 2008 to US$16 billion as end-2016

### 2008 fiscal rule and suspension
- A formal fiscal rule introduced in 2008 targeted a long-term non-oil fiscal deficit of 4.7 percent of GDP to be achieved by 2011, beginning at a deficit of 6.6 percent of GDP in 2008.
- The rule was consistent with a POIM approach, keeping government spending constant in real terms and supporting intergenerational equity and fiscal sustainability.
- The rule was suspended to allow a fiscal stimulus during the global financial crisis and abolished in 2012.

### 2013 expenditure rule and its weaknesses (abandoned 2015)
- The 2013 redesign replaced the budget balance rule with an expenditure rule setting a ceiling on federal expenditures equal to:
  - oil revenues measured at a benchmark oil price, plus non-oil revenues, plus a net borrowing limit of 1 percent of GDP.
- Benchmark determination: minimum of (i) a backward-looking moving average of up to ten years of Urals oil price and (ii) a three-year backward-looking average to protect the budget from sustained falls in oil prices.
- The rule failed to adjust benchmark oil prices rapidly enough, producing unwarrantedly large non-oil fiscal deficits.
- Example: the 3-year moving average escape clause yielded a benchmark price of about US$85pb versus an actual oil price of US$42pb in 2016.

### Modifications proposed to strengthen the fiscal rule (paragraphs 10–14)
- Adjust oil-price benchmark to perceived long-term oil price changes:
  - Fixed US$40 pb in the oil rule formula may lack credibility if oil prices are persistently higher.
  - Tradeoff: smoothing expenditures versus adjusting to oil price changes; faster adjustment desirable if oil shock is permanent/persistent.
  - Proposal: include futures oil prices in the benchmark calculation to allow faster adjustment, with caveat that futures are strongly correlated with observed prices and may overweight current prices, possibly increasing expenditure volatility and pro-cyclicality.
- Target a surplus informed by long-term fiscal considerations:
  - Recommendation to target a surplus rather than a balance to consider inter-generational equity.
  - With primary balance of zero, non-oil primary deficits are each year around 1 percentage point higher than long-term benchmarks consistent with inter-generational equity.
  - Saving more through the fiscal target may be more credible than assuming an artificially low oil price to accumulate reserves.
  - Caution: choosing a primary balance target vs overall balance is questionable because incorrect assumptions on interest rates or growth may set debt on an unsustainable path.
- Add an additional target on spending to avoid pro-cyclicality:
  - Include a rule that primary expenditures do not grow by more than the estimated long-term growth rate in real terms to address residual pro-cyclicality of a non-cyclically adjusted primary balance rule.
  - Fixed-oil price benchmark removes main source of pro-cyclicality (volatile oil prices) but is inferior to a full structural balance rule that excludes the cyclical component of output beyond oil prices.
  - If demand shocks are unrelated to the oil price gap, the authorities’ primary balance rule would be procyclical.
- Strengthen escape clauses to allow adjustment to persistently low oil prices:
  - Escape clauses should cap withdrawals from the reserve fund; whenever triggered, expenditure should adjust down.
  - Escape clauses should be complemented with borrowing constraints to internalize permanent drops in oil prices into the budget process.
- Consider targeting a structural non-oil balance once more data are available:
  - Structural primary non-oil balance (adjusting for economic and commodity cycles) as a share of potential non-oil GDP would reduce pro-cyclicality embedded in the proposed rule.
  - Complications: estimating cyclical adjustments is uncertain and often revised ex-poste; Russia should compile non-oil GDP data to calculate a non-oil structural fiscal balance.
  - Profits from oil and gas producing companies should be excluded from non-oil revenues and included in oil revenues.
  - Final objective: establish a rule ensuring a constant flow of oil revenues to the budget, as in Norway.

### Box 2 — Determining a benchmark oil price
- Forecasting oil prices is increasingly difficult due to high volatility and unclear time series properties, complicating separation of permanent vs temporary price components.
- Using futures prices helps budgets adjust to new oil price levels but may not anchor long-term benchmarks because futures are strongly correlated with observed prices and overweight current prices if current price is off its long-term equilibrium.
- The proposed US$40 pb (a 50-year average) is an improvement compared to prior benchmarks:
  - Large oil price changes tend to be persistent.
  - Using short moving averages (3, 5, 10 years) as proxies for long-term prices may cause persistent over- or undervaluation of the real effective exchange rate because fiscal expenditures would be tied to an oil price above or below the long-term price.
  - Shorter averaging periods make the benchmark more volatile—undesirable for macroeconomic stability.
- Governance proposal: an independent committee of experts could set the benchmark oil price for the budget, review the oil price formula periodically or after significant price moves, and help mitigate political economy pressures from the Ministry of Finance.

### Simulations of fiscal rules (Section C) — design and key assumptions
- Rules evaluated:
  - The authorities’ old rule (suspended in 2015): oil-price benchmark = minimum of (i) backward-looking moving average up to ten years of Urals; and (ii) three-year backward-looking average. Federal expenditures capped at sum of projected non-oil revenues, oil revenues at benchmark price (in US$) converted to ruble, and net financing of 1 percent of GDP.
  - Staff’s proposed rule: modifies the old rule by (i) using futures prices to establish a benchmark oil price (5-year average past and 5 years ahead futures prices) to allow faster fiscal adjustment; and (ii) increasing the target to a surplus of 1 percent of GDP (instead of a deficit of 1 percent of GDP) to generate more savings.
- Methodology:
  - IMF Flexible System of Global Models (FSGM) used to simulate fiscal and macroeconomic outcomes under alternative rules and oil price shocks.
  - Counterfactual simulation period: 2010–16, capturing high oil prices and the large negative 2014/15 oil shock.
  - Second set of simulations assesses rules under temporary and persistent positive and negative oil supply shocks.
- Key modelling assumptions:
  - Calculate non-resource primary deficits that would have prevailed under each rule; fiscal adjustment measured by change in the non-oil primary deficit.
  - Adjustment assumed to be met by equal cuts to government transfers and government consumption.
  - Financing of deviations from baseline: first from reserve fund holdings, residual by domestic borrowing.
  - Consolidation assumed achieved via cuts to transfers and government consumption (rather than investment), limiting negative growth impact.
  - Monetary policy is supportive: lower policy rates lead to a more depreciated exchange rate, higher exports and GDP growth, offsetting short-term negative impacts of consolidation.
  - Small negative impact on potential output through capital accumulation as real investment drops in the short run.

### Counterfactual simulation results and comparative assessment
- General counterfactual finding:
  - Consistent implementation of fiscal rules would have made fiscal policy more countercyclical: lower non-resource primary deficits when oil prices were high, enabling a looser stance after 2014.
  - Savings would be higher, with lower overall deficit and debt-to-GDP ratios.
  - Russia would have more assets than liabilities under all fiscal rules: 16 percent on average, compared to actual liabilities of 4 percent at end-2014.
  - Early consolidations, combined with supportive monetary policy, produce a more depreciated exchange rate and higher growth, offsetting consolidation effects; growth and inflation remain close to actual under all fiscal rule scenarios.
- Comparative rule assessment:
  - The old rule produced the lowest savings and highest spending during high oil prices:
    - Old rule effectively used an oil benchmark of US$79 (almost equal to spot) plus non-oil revenues, yielding the least competitive economy.
    - Higher spending increased inflation, prompting higher short-term policy rates and upward pressure on the exchange rate.
    - Under persistent oil price drops the old rule would have led to massive fiscal stimulus, depleting buffers—Reserve Fund run down to 3 percent of GDP—and gross debt ratcheting up to 15 percent of GDP by end-2016, leaving the economy no better off than actual outcomes despite higher initial buffers.
  - Staff’s proposed rule (futures-based benchmark and surplus target) aims to generate more savings and allow faster adjustment to oil-price developments.
- Projected non-resource primary deficits (table reproduced as series for 2010–2016)
  - 1. Old rule: 7.4, 6.6, 7.3, 7.5, 8.8, 12.3, 11.4
  - 2. New rule (benchmark@40): 3.8, 3.5, 3.7, 3.6, 4.0, 5.6, 5.3
  - 3. Proposed rule: 7.1, 6.8, 6.3, 5.7, 6.1, 7.3, 5.0
  - Baseline: 11.2, 8.4, 9.4, 9.3, 9.5, 9.0, 8.9
  - (Label: IMF staff estimates)
- Additional simulation outcomes:
  - The old rule would have resulted in the least favorable combinations of savings, competitiveness, and debt dynamics versus alternatives.

*International Monetary Fund — Box 1. Russia’s Previous Fiscal Frameworks*

### 20.      The authorities’ proposed new rule would have built more buffers, but staff’s rule is

### 20.      The authorities’ proposed new rule would have built more buffers, but staff’s rule is 

### Tradeoff between countercyclicality and buffer accumulation
- The simulation illustrates the tradeoff between countercyclicality and building buffers when designing fiscal rules.
- The new rule saves the most in the reserve fund through a conservative US$40 pb benchmark oil price (average actual oil prices are US$83 pb) and compared to an average benchmark of US$76 pb under the proposed rule.
- The proposed rule also results in substantial savings but rather through a more stringent fiscal target.
- The proposed rule allows a more countercyclical response to the large negative shock to oil prices.
- Output losses (in growth and levels) across the rules are similar and the differences in the debt trajectory derive from the extent of adjustment rather than growth differentials across rules.
- A caveat of the exercise is that we assume no reaction in Russia’s country risk premium.

### Simulation design: Oil price shocks and baseline
- Scenarios assume positive and negative shocks to oil prices (temporary and persistent) are a result of changes in the supply of non-Russian oil producers.
- Simulations are calculated as deviations from the baseline projections for Russia included in the 2017 staff report.
- Under the baseline scenario, spending is frozen per the authorities’ medium-term budget plan and revenues are calculated at baseline oil prices.
- Implementation of the fiscal rules starts in 2018.
- As in the counterfactual we calculate three fiscal rule adjustment scenarios on projected non-resource primary deficits under the baseline oil prices and four shocks to oil prices.
- Projected non-resource primary deficits (Baseline)
  - 1. Old rule: 5.6 5.5 5.8 5.6 5.5 5.4 5.4
  - 2. New rule (benchmark@40): 4.7 4.7 4.6 4.5 4.5 4.4 4.4
  - 3. Proposed rule: 4.2 4.1 4.0 4.0 4.1 4.1 4.1
  - Baseline: 6.5 5.6 4.8 4.8 4.9 4.6 4.6
- Oil Price Scenarios (2018–2022)
  - Baseline: 55.4 55.2 55.5 56.1 59.3
  - Persistent Low Oil: 30.0 38.7 40.3 42.8 46.8
  - Temporary Low Oil: 30.0 42.9 48.5 56.6 57.6
  - Persistent High Oil: 80.0 71.8 71.8 70.6 72.9
  - Temporary High Oil: 80.0 67.5 62.3 55.5 60.8

### Key simulation findings and comparisons across rules
- Staff’s proposed rule is preferred given the shocks that are considered.
- Persistently high oil prices:
  - The new rule results in the highest savings—net debt falls to zero, compared to 8 percentage points under the proposed rule.
  - Should high oil prices prove to be temporary, the proposed rule begins to have higher overall savings because the proposed rule targets a surplus, rather than a primary balance as in the new rule.
  - When oil prices are high, the impact on the broader economy is similar under the new and proposed rules.
  - Compared to the baseline, neither rule has a large negative impact on growth and both rules result in better outcomes on potential GDP.
  - Inflation is contained and policy rates are low resulting in a more depreciated exchange rate, increased competitiveness and an improved current account.
- Persistently low oil prices:
  - The persistent low price oil scenario illustrates the dangers of the authorities’ new rule of getting the benchmark price wrong.
  - Although the shock is persistent, the new rule doesn’t adjust, spending at a benchmark price of US$40 despite permanently lower prices.
  - This results in a rapid depletion of the RF and increasing debt dynamics with debt ratios that are around 2 percentage points higher every year throughout the projection horizon.
  - The proposed rule adjusts quickest to the new reality of low oil prices, with net debt decreasing to 8 percent.
  - However, should the negative shock be transitory, the proposed rule is tight and forces adjustment, when ex-post it was not necessary.
  - The impact of a tighter fiscal policy under the proposed rule to a temporary shock does not result in a significantly lower growth path compared to the new rule as looser monetary policy and the accompanied depreciated exchange rate offset the drag from a tighter fiscal policy.

### Quantitative and graphical outcomes (high-level)
- Evolution of Net Debt under shock scenarios shown in Figure 4 (In Percent of GDP) for:
  - Persistently High Oil Prices (2017–2022)
  - Temporarily Low Oil Prices (2017–2022)
  - Persistently Low Oil Prices (2017–2022)
  - Temporarily Low Oil Prices (2017–2022)
- Multiple panels and charts compare Old Rule, Authorities' Proposed New Rule, and Staff Proposed Rule for:
  - Real GDP (% Difference), GDP Growth (%pt difference), Potential Output (% difference)
  - Output Gap (%pt difference), REER (% difference, +=appreciation), Current Account (%pt GDP difference)
  - Real Exports (% difference), Real Imports (% difference)
  - Core CPI Inflation (%pt difference), Headline CPI Inflation (%pt difference), Policy Interest Rate (%pt difference)
  - Government Debt (%pt of GDP difference), Government Deficit (%pt GDP difference)
  - Private Saving, Private Consumption, Private Investment, Government Consumption, Government Investment (as applicable across scenarios)

_Italic: Source: IMF staff estimates._

### Annex I. FSGM for Russia

### Annex I. FSGM for Russia

### Model overview
- The simulations are calibrated using the IMF’s Flexible System of Global Models (FSGM), an annual, multi-economy, forward-looking model combining micro-founded and reduced-form formulations.
- Each economy in the model is structurally identical (except for commodities) but with different steady-state ratios and behavioral parameters; countries are distinguished by unique parameterizations.
- Russia’s parameters are strongly determined by the fact that its economy is dominated by oil.

### Households and firms
- Consumption:
  - Micro-founded via overlapping-generations households that can save and smooth consumption, and liquidity-constrained households that must consume all current income every period.
- Investment:
  - Firms’ investment is determined by a Tobin’s Q model.
  - Firms are net borrowers.
- Risk premia:
  - Rise when the output gap is negative (periods of excess capacity) and fall when the output gap is positive (booms), capturing the effect of falling/rising real debt burdens.

### Trade and commodities
- Trade equations are reduced-form and driven by a competitiveness indicator and domestic or foreign demand.
- Competitiveness:
  - Improves one-for-one with domestic prices—there is no local-market pricing.
  - For Russia, competitiveness changes play a small role because most exports are oil and gas.
- Exports:
  - For Russia, most (90 percent) exports are oil and gas, so exports of oil respond largely to Russian production decisions.
- Commodities in the model:
  - The model includes three commodities—oil, metals, and food—allowing distinction between headline and core CPI inflation.
  - Demand for commodities is driven by world demand and is relatively price inelastic in the short run.
  - Supply of commodities is price inelastic in the short run.
  - Global real commodity prices are determined by a global output gap (only a short-run effect), the overall level of global demand, and global production of the commodity in question.
  - Commodities can moderate business cycle fluctuations via price adjustments when aggregate demand is in excess or shortfall.

### Potential output
- Potential output is endogenous and modeled by a Cobb-Douglas production function with exogenous total factor productivity (TFP), and endogenous capital and labor.
- For Russia, potential output moves one-for-one with the long-run average production of oil (but not cyclical swings in oil production).

### Inflation and monetary policy
- CPI and wage inflation:
  - Modeled by reduced-form Phillips curves including lags and leads of inflation and the output gap.
  - Consumer price inflation has a weight on the real effective exchange rate and second-round effects from food and oil prices.
  - As energy prices in Russia do not respond to global oil price developments, there is no feed-through from oil price changes to CPI inflation.
- Monetary policy:
  - Governed by an interest rate reaction function.
  - The rule is inflation-forecast-based, working to achieve a long-run inflation target through a risk-adjusted uncovered interest rate parity.

### Role of oil in calibration and fiscal implications (annex observations)
- Oil’s roles in the model:
  - Oil price fluctuations affect government revenues.
  - Oil prices have little effect on household wealth because households have no direct ownership stake in the oil sector.
  - Oil prices have little effect on households’ and firms’ decisions, as oil prices are held fixed domestically.
- Oil price benchmarking analysis (historical period and benchmarks examined):
  - Historical oil prices analyzed for 1923–2016 (in US$ nominal terms) and expressed in 2015 US$ terms using the US CPI.
  - Benchmarks considered: moving averages (in 2015 US$ terms) of lengths 3, 10, 20, 25, 30, 40 and 50-year MAs.
  - Analysis extended using future oil prices (as of July 27, 2016) through 2021 and assuming US CPI inflation gradually converges to 2 percent per year (from the current 1 percent) by end 2019.
- Key quantitative observations about potential benchmarks and persistence:
  - The 50-year average (1966–2015) oil price in real terms (for the US imported oil basket) is US$42.6 in 2015 terms.
  - The authorities’ proposed benchmark: the average of the last 50 years (in 2015 terms) and adjusted yearly by the variation of the US CPI (i.e., assuming a constant relative price between oil and the US consumption basket going forward).
  - Using short-length MAs (e.g., a 3-year MA) results in more “realistic” oil prices but can induce strong pro-cyclicality in periods of sustained increases or decreases in oil prices.
    - Definition of “far”: observed oil prices were either higher (or lower) than the benchmark by +/- 25 percent.
    - During 1972–2015:
      - The longest period in which observed oil prices were “far” higher than the 3-year MA was 1.5 years.
      - The longest period in which observed oil prices were “far” lower than the 3-year MA was 2 years.
      - The average period in which observed oil prices were either far above or below the 3-year MA was 0.2 years.
  - Using long-length MAs (e.g., a 50-year MA) reduces pro-cyclicality of government spending but can leave observed oil prices “far” from the benchmark for long periods.
    - During 1972–2015:
      - Observed oil prices were “far” above or below the 50-year MA during, on average, 2.5 years.
      - The longest period in which observed oil prices was higher than the 50-year MA was about 12 years.
      - The longest period in which observed oil prices were far lower than the 50-year MA was 1.5 years.
    - The asymmetry arises because the 50-year moving average includes a long spell (1966–1972) of low and stable real oil prices (about US$13/barrel in 2015 terms).
  - Adopting a benchmark based on the 50-year MA (instead of freezing the relative price of oil at end-2015) would result in an increase in the oil price benchmark in the coming years as older low-price periods are replaced by more recent higher-price periods; for example, the 50-year moving average oil price through 2021 (using oil price futures through 2021) would be around US$46/barrel in 2015 terms, compared with US$42.6/barrel in 2015 terms in the rule proposed by the authorities.
  - The relative price of oil with respect to the US consumption basket during 1972–2015 increased by about 2.3 percent per year.
  - Oil price futures (as of July 27, 2016) imply an average annual increase in such relative price of about 1.3 percent per year through 2021.
  - Therefore, the authorities’ proposal to “freeze” the oil price at the 50-year moving average in real terms at end-2015 also results in a “saving” bias.
- Policy implications highlighted:
  - Short-length MA benchmarks can produce pro-cyclical fiscal outcomes in sustained price trends but typically avoid extended periods where observed prices are far from the benchmark.
  - Long-length MA benchmarks reduce pro-cyclicality but can produce long episodes where observed prices exceed the benchmark, requiring political restraint to accumulate net assets when revenues exceed the benchmark.
  - Intermediate-length MAs (e.g., 10- or 15-year) may overlook long persistent low or high price regimes and thus can entail relatively long periods when observed oil prices are far above or below the benchmark, potentially leading to debt increases that could exceed market willingness.

*International Monetary Fund (excerpts from Annex I and Annex II in the source PDF).*

### 9.      Russia’s legal framework obtains higher marks than the average of advanced and

### 9.      Russia’s legal framework obtains higher marks than the average of advanced and emerging market economies in co-determination of federal policies, fiscal rules, and the stability of its fiscal constitution

### Legal framework: de jure strengths versus de facto variability
- Russia’s legal framework obtains higher marks than the average of advanced and emerging market economies in:
  - co-determination of federal policies,
  - fiscal rules, and
  - the stability of its fiscal constitution.
- De jure versus de facto considerations affect the assessment:
  - Russia’s budget code has included some form of a fiscal rule since 2008, but its parameters have changed, and its implementation has been suspended a few times.
  - The operational framework establishing the relation between federal and regional governments (including on tax sharing and transfers) has experienced numerous modifications.

### Fiscal constitution classification
- Russia’s legal framework is consistent with an integrated fiscal constitution (as opposed to a decentralized one).
- Characteristic contrasts (as summarized from OECD (2016)):
  - Decentralized fiscal constitutions (e.g., Canada and the United States): more SNG autonomy and responsibility, low co-determination, relatively weak numerical budget rules and frameworks.
  - Centralized/integrated fiscal frameworks: lower SNG autonomy and responsibility and, at least de jure, strong fiscal rules and frameworks.

### Fiscal Federalism at Work: achievements and challenges — empirical approach
- Empirical analysis:
  - Panel data of 79 regions across economic activity, labor, fiscal, financial, and structural variables.
  - Data spans the period 2000–16, with some variables shorter (e.g., regional fiscal data for 2005–16, GRPs for 2000–15, GRPs’ composition for 2004–15).
  - Analysis based on a cross-sectional bilateral dataset of regional differences, producing more than 3000 observations.

### Some stylized facts on regional finances and responsibilities
- Composition of regional revenues:
  - Regional revenues = own revenues + federal transfers.
  - Federal taxes (notably personal and corporate income tax) are the largest source of regional fiscal revenue, representing on average about 70 percent of own revenues.
  - Tax sharing (primary distribution) is performed directly in regions where taxes are collected at predetermined rates; sharing arrangements and rates governed by the Budget Code and, for corporate income tax, by the Tax Code. Rates tend to be adjusted frequently.
- Cross-regional revenue disparities:
  - Significant cross-regional differences in own revenues in real per capita terms.
  - Real per capita fiscal revenues are:
    - positively associated with the share of the private sector in regional GRP,
    - positively associated with the share of mining in GRP,
    - negatively associated with the share of agriculture.
  - Regions with lower real per capita GRP have lower real per capita own revenues.
- Transfers and their composition:
  - Intragovernmental transfers aim to level horizontal fiscal inequality and include:
    - non-earmarked and non-matching transfers (dotatsii, with equalization grants most important),
    - subsidies (earmarked matching transfers),
    - subventions (earmarked non-matching transfers),
    - other transfers.
  - The Federal Medical Insurance Fund makes transfers to Territorial Medical Insurance Funds, which represented 1.7 percent of GDP in 2016.
  - Equalization grants constitute about 50 percent of federal government transfers.
- Regional spending responsibilities (2016 shares of general government expenditure):
  - housing and utilities: 95 percent by regional spending,
  - education and cultural activities: 80 percent by regional spending,
  - health (including territorial extra-budgetary medical funds): around 85 percent by regional spending.

### Federal transfers and public goods supply disparities — empirical findings
- Transfers reduced cross-regional spending dispersion:
  - Federal transfers lifted real per capita fiscal spending in lower GRP per capita regions and reduced cross-regional spending dispersion.
  - Grants mainly achieved reductions in real per capita spending disparities; subsidies and subventions in real per capita terms have been broadly allocated to regions with higher per-capita income.
- Education and health:
  - Higher average transfers in 2005–16 (in real per capita terms) have been positively associated with larger increases in real per capita annual spending in health and education.
  - This helped regions with initially lower real per capita GRP partially close the gap in real per capita spending in health and education.
- Human capital:
  - Regions with lower initial real per capita income and weaker educational attainment experienced faster increases in years of education of the average worker (regional labor data for 2002–15).
  - Human capital measures (using Hall and Jones (1999) methodology) increased at relatively higher rates in regions that received higher average transfers (in regional GRP terms) during the last decade; result partially driven by cross-regional differences in labor supply.
- Physical capital and investment:
  - Regions receiving larger federal transfers (in GRP terms) generally experienced higher investment-to-GRP ratios and higher growth rates of physical capital.
  - Very high investment ratios reported in some regions (in some cases to the order of 50 percent of GRP), suggesting initially very low capital stocks in poorer regions.
- Transfers and cross-regional growth correlation:
  - Several models relating correlation of cross-regional growth of real per capita GRP with correlation of cross-regional growth of real per capita federal transfers and other variables show:
    - Aggregate transfers do not have a strong or robust association with bilateral cross-regional growth correlation.
    - By transfer type: correlation in the growth of grants is not associated with correlation of growth rates; subsidies and subventions are positively associated.
  - Caution due to possible endogeneity; different transfer types have different impacts on cross-regional GRP growth correlation.
  - Policy implication: whether positive association between cross-regional GRP growth correlations and those of subsidies and subventions is desirable depends on whether federal fiscal policy amplifies or lessens overall economic cycles; smoothing aggregate cycles and strengthening cross-regional growth correlations would have positive spillovers for monetary policy.

### Federal transfers and the sustainability of regional budgets — empirical findings
- System of equations estimated to assess direct and indirect effects of federal transfers on fiscal sustainability, allowing interactions among:
  - own regional revenues-to-expenditures ratio (proxy for fiscal sustainability),
  - per capita GRP growth,
  - GRP structure,
  - federal transfers.
- Main empirical conclusions:
  - Federal transfers have not positively impacted regional fiscal sustainability.
  - Transfers appear to have changed GRP structure by increasing the size of the public sector.
  - While transfers have a direct positive impact on GRP growth (through stronger accumulation of production factors), they also have a negative indirect impact via a larger public sector; the negative indirect impact more than offsets the direct positive impact.
- Quantitative illustrative example:
  - A one-standard deviation difference in the level of federal transfers (about 17 percent of regional GRP) is associated with:
    - a negative cumulative bilateral difference in real per capita GRP growth (over 2005–15) of around 1.2 percentage points,
    - an increase in the bilateral share of public sector in GRP of around 1.5 percentage points,
    - an (own) revenue-to-expenditure ratio that stays around unchanged (indeed, a decrease in such ratio of about 0.1 percentage point).
  - Given the positive association between own revenue-to-expenditure ratio and GRP growth, federal transfers have not resulted in an improvement in regional fiscal sustainability.
  - These results are particularly relevant for around 1/3 of Russia’s regions (28 out of 79 in the sample), which receive federal transfers that are higher than the average by between 1 and 3 standard deviations.
- Persistence of fiscal dependence:
  - Regions receiving larger federal transfers have not been able to close the gap between expenditures and own revenues.
  - Economic growth based on expansion of government services did not improve own revenue-to-GRP ratios, which (in levels) are positively correlated with the size of the private sector.
  - Financial dependence on federal transfers has remained broadly unchanged for many regions; for many, own revenues are barely sufficient to finance health and education spending.
- Transfers and private-sector-led growth:
  - During the period analyzed, federal transfers were insufficient to jumpstart self-sustaining, private-sector-led growth in regions receiving relatively more transfers.
  - Transfers flowed to regions with lower initial real per capita GRP and a relatively larger footprint of the state (measured as number per capita of regional budget and non-budgetary entities, including state unitary enterprises and joint-stock companies).
- Total factor productivity (TFP):
  - Neutral TFP levels recovered using a production function approach (regional capital stocks via perpetual inventory method; effective human capital corrected for labor utilization).
  - TFP levels for 2000–15 recovered using regional human and physical capital, assuming identical Cobb-Douglas production functions for all regions.
  - Analysis suggests cross-regional TFP growth differentials are negatively associated with cross-regional differences in average transfers; distance in productivity levels between low and high-income regions increased in the last decade.
  - Additional analysis (Pedroni and Yao (2006) approach, not shown) suggests no convergence in real per capita income across Russian regions during 1998–2015, supporting the conclusion that federal transfers have not helped speed up regional convergence.

### Demographic concentration and spatial implications
- Geographic population concentration increased in the last 15 years:
  - Population of the city of Moscow increased by more than 30 percent since 2000.
  - Population of Saint Petersburg increased by 10 percent since 2000.
  - Total population broadly constant, implying other less densely populated regions experienced population decreases of 15–20 percent.
- Implications:
  - Concentration brings advantages for recipient regions and cities (economies of scale, firm localization, improved job matching) and drawbacks for regions losing population.
  - Increased concentration raises costs of per capita federal transfers and contributes to geographically unbalanced development in a continental-sized country.
  - Current fiscal federalism institutions appear not to take into consideration advantages/disadvantages related to increased concentration or unintended effects they may create.

### Conclusions and issues for discussion — policy implications and recommendations
- Role of federal government and need for coordination:
  - Russia’s fiscal federalism assigns a strong role to the federal government; increased policy coordination with regions could be beneficial.
  - System evolved from disorderly decentralization in the 1990s into a more centralized system in the last 15 years.
  - Regions play an essential role in human and physical capital formation but have less autonomy and lower control of fiscal policy than in other federal countries.
  - Increased coordination between federal and regional governments to tackle complexity and address cross-regional infrastructure and human capital bottlenecks could yield a more integrated national market with positive spillovers for inter-regional and international trade and investment.
  - Ongoing work to measure regional business climate differences to strengthen institutions should be pursued and deepened, avoiding stigma but promoting jurisdiction competition.
  - Regional convergence can yield a growth dividend and more balanced geographical development.
- Federal macroeconomic and tax policy recommendations:
  - Appropriate federal macroeconomic and tax policies can contribute to development of regional tax bases and support regional sustainability.
  - Adoption of a fiscal rule along realistic parameters should promote a more stable and more aligned-with-fundamentals real exchange rate with positive spillovers for lower per-capita income regions where agriculture (a tradable sector) represents a larger share of GRP.
  - Current plans to rebalance domestic taxes to tax labor less strongly should support decreases in informality (likely more prevalent in low-per-capita income regions, as attested by weaker tax bases).
  - From a macroeconomic perspective, adopting a fiscal rule should eliminate the role fiscal policy has played in transmitting terms of trade shocks.
  - Under this backdrop, the role of transfers in supporting correlation in regional growth should have positive spillovers for monetary policy.
- Strengthening regional tax bases and accountability:
  - Strengthening regional tax bases could improve regional sustainability and accountability.
  - Option: expand use of personal property taxes (OECD, 2016).
    - Personal property taxes currently represent only 0.4 percent of the consolidated own revenues of regions.
    - In 2016, 28 regions started a transition to market value-based instead of accounting value-based taxation of property.
    - Example: the city of Moscow is projecting a five-fold increase in property tax collections by 2020 (with tax collection increasing by 55 percent in 2016).
  - Stronger regional tax bases should balance somewhat the strong de jure role of the federal government and increase accountability of regional governments.

*IMF — Russian Federation chapter (content unit: 9).*

### 32.      Federal transfers have been effective in supporting factor accumulation in lower per

### cr17198 - 32.      Federal transfers have been effective in supporting factor accumulation in lower per

### Effectiveness of federal transfers
- Federal transfers have been effective in supporting factor accumulation in lower per capita income regions and increasing growth correlation, but less effective in supporting self-sustaining GRP growth and productivity increases.
- Given relatively rigid tax sharing arrangements, federal transfers constitute one of the main levers through which federal policy operates at the regional level.
- Transfers have expanded government services but have not been as effective in expanding productive activities.
- Large cross sectional differences in own fiscal revenues (in per capita and GRP terms) have persisted, as well as the associated dependence on federal transfers.
- Federal transfers have flowed more strongly to regions where the footprint of the state is larger.

### Most likely scenario and strategic implications
- The most likely scenario going forward is one in which regional dependence on transfers decreases only slowly.
- From a regional perspective, equalization grants will likely keep their leading role. Sudden decreases or reallocations could create disruptions especially in the most financially dependent regions.
- The complete elimination of regional dispersion is unlikely.
- Enhanced strategic direction could help increase federal transfers’ growth effectiveness.
- Open-ended transfers may have had the unintended effect of weakening regional incentives to enlarge their tax bases, further supporting a pattern of dependence.
- Policy consideration: include in the formulas defining grant allocation, gradually and in the margin, a measure of sustainability together with the current objective of equalization.
  - Transition periods and reasonable time frames to achieve sustainability would be essential.
- Macroeconomic consideration: the expected persistence of current volumes of transfers will add up to the existing earmarking of federal revenues that also includes transfers to EBFs, which may complicate addressing intertemporal equity considerations in the use of oil revenues.

### Use of horizontal transfers and tax redistribution options
- There may be scope to increase the use of horizontal transfers in the margin to counteract economic and population concentration.
- The large cross-regional dispersion of per capita own fiscal revenues may have contributed to economic and population concentration, creating negative spillovers for regions with population outflows.
- Horizontal transfers, in the margin, may support improved use of human and physical capital in lower-per-capita income regions.
- Policy option: gradually improve the primary distribution of corporate income tax (CIT), and make more permanent the ongoing redistribution (by the federal government) of 1 percentage point of CIT to finance equalization grants.

### Streamlining, simplification, and transparency of transfers
- There may be room to streamline, simplify and increase the transparency of transfers:
  - Streamline the number of transfers (especially subsidies), in particular for agriculture development, housing and utilities and education, and allocate them in appendices to the federal budget law.
  - Allocate subsidies one-to-one to government programs (or subprograms), instead of to a multiplicity of them.
  - Transform and further consolidate “other transfers” into subsidies.
  - Regulate budget loans, which are increasingly used because of their concessional interest rates.
- These steps should result in a simpler, more transparent, and easy to administer system.
- Ongoing work towards streamlining the Budget Code should be pursued and finalized.
  - The Budget Code was approved by the Federal Assembly in 1998 and has since been amended by 120 federal laws.
- Note on recent legal adjustments: shares (of certain revenues) are suspended for 2017-2020 by law 409-FZ of 30 November 2016.

### Empirical evidence and analysis (summarized)
- IMF staff calculations and official data underpin graphical and econometric analysis linking transfers to:
  - Per capita GRP (Figures show associations between Log real pc Transfers, Log real pc Grants, Log real pc Subsidies, Log real pc Subventions and Log real pc GRP).
  - Accumulation of factors of production (associations between Differences in Transfers as Percent of GRP and Real PC Education Spending, Real PC Health Spending, Human Capital Index, Physical Capital growth differentials).
  - Public sector expansions and TFP increases (differences in bilateral change of share Public Sector (2004-15 pp GRP) versus differences in transfers; bilateral annual productivity change versus differences in transfers).
- Regression and simultaneous-equations results are reported:
  - Table 1: definitions of variables used (e.g., Real per capita growth correlation; Federal transfers-to-GRP ratio; Footprint of state = Ln of number of per capita budgetary and non-budgetary state institutions).
  - Table 2 and Table 3: regressions for bilateral regional per capita GDP growth correlations (estimated coefficients and model specifications; significance markers *, **, *** for 1, 5, and 10 percent levels).
  - Table 4 and Table 5: federal transfers in a simultaneous equations system and estimation results across methods (SUR, 2SLS, 3SLS, FIML, GMM) with reported coefficient estimates and significance markers.
- Appendix I catalogs revenue sources and sharing arrangements and describes devolved spending responsibilities:
  - Table A1 lists tax and non-tax revenue sharing arrangements (examples include VAT share: 100; PIT share: 30/85/15; CIT sharing and note that for 2017-20 the federal government will receive an additional 1 pp to be redistributed via equalization grants).
  - Specifics on special regimes and shares (e.g., MET (Oil and Gas): Formula-based depending on oil price; MET (Other subsoil resources): Ad valorem and specific 40/60; Excise on gasoline and motor oil shares are subject to suspension for 2017-2020).
  - Notes: Regions are authorized to adjust their portion of the CIT rate down, but no more than to 13.5 percent (12.5 percent in 2017-20). For 2017-20, the federal government will receive an additional 1 pp to be redistributed via equalization grants. This may result in a financing gap for some regions.

*Source: IMF staff (cr17198).*

### 42.6 percent in 2018, and 39.8 percent in 2019.  The remaining portion will go to the regional budgets.

### cr17198 - 42.6 percent in 2018, and 39.8 percent in 2019.  The remaining portion will go to the regional budgets.

### Spending responsibilities and jurisdiction by level of government
- General allocation of responsibilities:
  - Federal (exclusive): authority on federal property, regulation of social and economic development, federal energy systems, national defense and security, international relations, law enforcement; meteorology and statistics.
  - Joint federal-regional jurisdiction: public safety and law enforcement; administrative, labor, family, housing, land, subsoil, forest, water relations; environmental protection; emergencies and natural disasters; education, science, culture, sports; public health, social security. Responsibilities usually divided by jurisdictional attribution or relevance; sometimes shared between levels.
  - Regional (exclusive): all government responsibilities beyond federal and joint jurisdictions as stipulated in regional constitutions and legislation.
  - Local governments' jurisdiction: urban, rural settlements; electricity, heating, water, gas, fuel supply; roads; municipal housing; public transport; emergencies, fire safety; public amenities, eateries, retail trade; culture (local cultural heritage, folk art and crafts); physical culture, sports, public entertainment, recreation; archives; cemeteries; local resorts; public safety, rescue operations; waste management; support to agriculture and SMEs; terrorism/extremism prevention; education (less vocational + vacations); public health.

- Delegated federal responsibilities supported by federal subventions (selected items):
  - National Census and Agricultural Census
  - Prevention of homelessness
  - Housing for disabled, veterans, retired servicemen, etc.
  - Subsidization of housing and utility payments for veterans, disabled, radiation-exposed, etc.
  - Payouts to radiation-exposed
  - Unemployment benefits; maternity and childcare benefits
  - Monthly compensation payouts to various categories (e.g. exposed to radiation, blood donors)
  - Water and forest relations: management (partial) of federal water facilities and forests; animal world, hunting, fishing (partial)
  - Protection and oversight of cultural heritage; education: oversight, licensing, accreditation (all partial)
  - Public health: licensing; procurement of drugs, mandatory medical insurance

- Delegated federal responsibilities unsupported by federal subventions (selected items):
  - Audit of construction plans and engineering surveys; environmental audit
  - Land relations: provision of plots of land for construction, demolition of real estate, easement
  - R&D management

- Selected areas (examples of functional allocation):
  - Education: Universities; Vocational, primary and secondary schools
  - Employment: Unemployment benefits (delegated) and employment facilitation
  - Social security: Social support to war veterans, radiation victims (some responsibilities delegated); social support to senior citizens, disabled, orphans, labor veterans, low income households; payment of medical insurance contributions on behalf of non-workers
  - Industry support: For instance, Aviation; Support to agriculture (beyond that from federal programs) and to SMEs (since 2015)
  - Waste management: Radioactive waste; Solid waste

- Legal basis and lists of responsibilities:
  - Responsibilities of regional governments in areas of joint jurisdiction are stipulated in legislation/regulations: 114 responsibilities listed in the framework law (184 FZ of 1999); 61 responsibilities prescribed in various specific laws; 20 responsibilities arising from Presidential decrees; 162 responsibilities according to GoR decrees. Regional governments implement 55 federal government programs and federal special-purpose programs (financed with own funds and subsidies).

### Limits imposed by the Federal Government on Regional Budgets
- Monitoring, reporting and transparency standards and sanctions for rule violations (including adjustments in the size of transfers, excluding subventions).
- Budget balance requirements:
  - The deficit of regions cannot exceed 15 percent of their own revenues (excluding grants).
  - Rules are stricter if federal grants exceed 40 percent of the consolidated region budget revenues (excluding subventions).

- Tax limits:
  - Sub-federal governments can set tax rates and reliefs for regional and local taxes.
  - For the CIT, regions can set rates for the regional part of the tax within limits set by the Tax Code but not reliefs.
  - Excise taxes on gasoline and alcohol are shared annually between regions and federal government.
  - The Tax Code does not allow regions to legislate on PIT, fees and charges, rates and reliefs, which constitute the remaining 40 percent of their revenues.

- Expenditure limits:
  - Regions with a share of federal grants exceeding 10 percent of consolidated region budget revenues (excluding subventions), cannot assume and execute expenditures assigned to regional governments by Constitution and federal laws; and cannot exceed federal norms for budgetary sector wages and regional government activity financing.
  - Similar restrictions exist for municipalities getting equalization grants from regions.

- Borrowing constraints:
  - Domestic borrowing is not directly restricted.
  - New foreign borrowing (for deficit financing or refinancing) is allowed only for regions that do not receive federal equalization transfers, do not have debt arrears, and have proper credit ratings from at least two international agencies.
  - Regions receiving federal equalization transfers can borrow externally to refinance existing external debt, if no debt arrears and credit rating requirements are satisfied.
  - Total yearly borrowing of regions and municipalities is bound up by deficit financing and debt amortization.

- Debt levels and service:
  - Debt is not allowed to exceed own annual revenues (excluding grants).
  - Rules are stricter if federal grants share exceed 40 percent of consolidated region budget revenues (excluding subventions).
  - Debt service (interest payments) should not exceed 15 percent of total expenditures (excluding subventions).
  - Escape clauses: budget credit financing, privatization, use of regional precautionary saving funds.
  - Debt ceilings are currently allowed to be exceeded for an amount equal to federal budget credits.

### PUTTING THE CURVE BACK IN RUSSIA’S PHILIPS CURVE: A TIME-VARYING APPROACH — Introduction and motivation
- Key observations:
  - Unemployment has stayed muted since 2013, while inflation has been volatile.
  - Transition to Inflation Targeting (IT) regime by Central Bank of Russia (CBR) in late 2014 increases importance of understanding inflation–slack relationship.

- Empirical strategy:
  - Estimate a hybrid New Keynesian Phillips curve (NKPC) for Russia’s core inflation with time-varying coefficients.
  - Compare results to simpler bivariate estimations to show how bivariate relationships can be misleading.

- Main findings (summary):
  - The core inflation Phillips curve in Russia is "alive" and the slope is expected to increase with the recovery.
  - Impact of cyclical unemployment on core inflation changes over time: tends to increase during normal-times and to decrease in aftermath of crisis-times.
  - Weight on inflation expectation in the PC model has increased recently, thanks to the introduction of an IT regime—implying CBR is on track in anchoring long-run inflation expectations and gaining credibility.
  - Model-implied unemployment slack has been coming down fast post-GFC, due to flexible labor market conditions.
  - Overall fitted inflation shows the PC is a good model in explaining inflation dynamics.

- Caution on bivariate analysis:
  - Bivariate relationship between inflation and slack can reverse sign during crisis episodes (crisis time episodes used: 1998Q3–2000Q4, 2008Q4–2009Q4, 2015Q1–2016Q4).
  - Reversal highlights open economy implications and the role of REER and import prices.
  - Dispersion in estimated slopes across measures and episodes suggests multivariate specification is superior.

### Model specification, estimation and key equations
- Sample: quarterly data from 2000Q1 to 2016Q3.
- Variables highlighted:
  - ߨ௠,௧ is inflation in the relative price of imports (import price deflator relative to GDP deflator).
  - ߨ௘௧ is long-run inflation expectation (defined as a 5-year forward-looking forecast of inflation based on the WEO vintage database (1993–2016)).
  - Model accounts for expected and lagged core inflation, unemployment rate, lagged relative import price inflation and lagged REER.

- Estimated (constant-coefficient) regression example (as presented):
  - ߨ௧ = ା0.37ݑ௧ିଵ ା0.13ߨ௧௘ ା0.96ߨ௧ିଵ ା0.03ߨ௠,௧ିଵ ݎ݁݁ݎ ା0.007௧ିଵ ߝ൅௧
  - Interpretation from this regression:
    - Presence of a Phillips Curve with strong hysteresis (lagged inflation) and mild impact of import price inflation.
    - Coefficients of REER and inflation expectation are not significant over whole sample — indicating time variation due to policy regime changes (IT adoption and free-floating exchange rate in November 2014).

- Time-varying parameters hybrid NKPC estimated (model equations reproduced):
  - ߨ௧ߠ = ଵ௧ (ݑ௧ିଵݑെ௧ିଵ∗)ߠ + ଶ௧ ߨݐ݁ + (ߠ1େଶ௧)ߨ௧ିଵߠ + ୳௧ ߨ௠,௧ିଵ + ସ௧ݎݔ௧ିଵߝ൅௧  (Equation (1))
  - ݑ௧ݑെ௧∗ߩ = (ݑ௧ିଵݑെ௧ିଵ∗)ߟ + ௧௧  (Equation (2))
  - ݑ௧∗ݑܾ = ௧ିଵ∗ + (ܾ1େ)ݑ௧ିଵߥ൅௧  (Equation (3))
  - Time-varying coefficients follow a random walk: ߠ௧ߠ = ௧ିଵ൳ + ௧ (Equation (4))

- Model features and estimation notes:
  - Equation (1) is hybrid NK Phillips Curve in terms of unemployment gap (ݑ௧ݑେ௧∗), long-run inflation expectations ߨ௧௘, lagged core CPI inflation ߨ௧ିଵ, relative import price inflation ߨ௠, and REER ݎݔ.
  - Slope coefficients in equation (1) are time-varying (random walk).
  - Natural rate of unemployment is estimated within the model (unobserved).
  - Equation (2) gives law of motion of the unemployment gap with persistence ߩ.
  - Equation (3) determines dynamics of the natural rate of unemployment; coefficient ܾ determines hysteresis level in the natural rate.
  - The three shocks (ߝ௧, ߟ, ߥ௧) are assumed normal and i.i.d; model is Gaussian but not linear in latent states.
  - Time-variation motivated by institutional changes: adoption of IT in 2014 with headline inflation target of 4 percent by end of 2017, and the shift to a free-floating exchange rate in November 2014.

### Empirical implications and policy relevance
- Evidence supports using a multivariate, time-varying Phillips curve to:
  - Capture changing roles of unemployment slack, inflation expectations, REER, and import prices.
  - Inform monetary policy in the transition to and consolidation of Inflation Targeting.
- Specific empirical implications:
  - Increasing weight on inflation expectations suggests improved anchoring—relevant for communication and credibility-building by the CBR.
  - Time-varying slope implies that the sensitivity of inflation to slack will strengthen in recovery periods and weaken in crisis periods—important for countercyclical policy calibration.
  - Import-price and exchange-rate channels matter episodically; monitoring REER and import-price inflation remains important for inflation outlook.

*Source: IMF staff calculations and text from the document provided.*

### 7.      Our identification assumptions imply some constraints on the coefficients. More

### 7.      Our identification assumptions imply some constraints on the coefficients. More

### Identification, estimation method, and data
- Identification assumptions:
  - The slope on the unemployment gap is negative.
  - The weight of inflation expectations must be between 0 and 1.
  - The slope of the exchange rate is restricted to be negative.
- Estimation method:
  - A standard Kalman Filter cannot be used because of the constraints; the model is estimated using a Constrained Extended Kalman Filter.
  - Constrained Extended Kalman Filter: nonlinear version of the standard Kalman Filter. The Jacobian of equation (1) is required to update the covariance. Constraints are checked at each step and, if violated, a minimization procedure is performed to obtain a constrained estimate of the states.
- Calibrated parameters:
  - Hysteresis parameter, ܾ: higher values produce smoother estimates of the natural rate; smaller values make the natural rate follow the observed unemployment rate more closely.
  - Variance of the natural rate of unemployment: calibrated via the signal to noise ratio between the variance of the unemployment gap and the natural rate.
- Prior variance of latent time-varying slopes — three calibration options examined:
  1. By using user-defined priors.
  2. By using a calibrated prior variance from a rolling or full sample estimation of the model with constant slopes, using Constrained Maximum Likelihood.
  3. Calibrate the variance from the Constrained Maximum Likelihood, either from a rolling or full sample regression and do not estimate it (methodology in Matheson and Stavrev (2013)).
- Sign restrictions implementation:
  - Sign restrictions are imposed by specifying matrix D and vector d such that D x ≤ d; draws with wrong sign are discarded.
  - Example constraints provided for negative first slope, positive third coefficient, and a second coefficient restricted between 0 and 1.

### Data and transformations
- Data frequency and coverage:
  - Quarterly frequency, covers 2000Q1 to 2016Q3.
  - Seasonally adjusted.
- Long-run inflation expectations:
  - A 5-year forward-looking forecast based on the WEO vintage database (1993–2016).
  - Construction: for each year, use WEO Spring vintage (WEO April or WEO May) as the Q1 and Q2 expectation, and Fall vintage (WEO September or October) as the Q3 and Q4 expectation. Example: 2021 projection in 2016 WEO April vintage taken as 5-year expectation in 2016Q1 and 2016Q2; 2021 projection in 2016 WEO October vintage taken as 2016Q3 and 2016Q4. Y/Y growth rate calculated based on the index.
  - Reason for using YoY transformation: consistency with the measure of inflation expectation which is yoy.
- Series and transformations (Data 2000–2016, quarterly):
  - HICPXEF — Core Consumer Price Index, SA (Dec.2000=100): YoY Growth Rate
  - UR — Unemployment Rate, SA: Level
  - IMPXdef — Imports Deflator/GDP Deflator (SA, 2011=100): YoY Growth Rate
  - REER — Real Broad Effective Exchange Rate Index, CPI Based (2010=100), SA: Annualized log difference
  - LTEXP — Long-term Inflation expectation: YoY Growth Rate

### Estimated coefficients and model fit (constant-time-coefficient model summary)
- Sample and fit statistics:
  - R-square: 0.87
  - Adjusted R-Squared: 0.86
  - Root Mean Squared Error: 1.43
  - Estimated error variance of the model: 2 (MSE around 2.05)
  - F-statistic vs. constant model: 83, p-value = 1.19e-25
- Estimated coefficients (Estimate, SE, tStat, pValue):
  - Lagged unemployment: -0.3686, 0.195, -1.8903, 0.0635
  - Inflation Expectation: -0.1258, 0.1324, -0.9507, 0.3456
  - Lagged inflation: 0.9586, 0.0605, 15.855, 0
  - Import inflation: -0.0255, 0.0145, -1.7626, 0.0831
  - REER: -0.0079, 0.0095, -0.8297, 0.41
- Interpretation of fit:
  - R-squared of 87 percent shows a good fit, but does not guarantee unbiased coefficients or predictions.
  - Residual plot mostly random, except outliers in the last quarter of 2014 and 2015.
  - MSE ≈ 2.05 implies approximately 95 percent of observations should fall within plus/minus 2*standard error of the regression from the regression line (approximation of a 95 percent prediction interval).
  - P-values for inflation expectation and REER > 0.1, so not significant at the 10 percent significance level in the constant-coefficient regression.

### Time-varying Phillips Curve results and dynamics
- General findings:
  - The slope of core inflation Phillips curve is not flat and is steeper in normal business cycle times.
  - The weight on long-run expected inflation (as opposed to the coefficient on lagged inflation) has increased since 2012, attributed to the CBR’s effort in anchoring inflation expectations.
  - The coefficient on inflation expectations is lower than those typically found in Advanced Economies with well-established IT regimes (example: for the US it hovers around 0.7). This is attributed to the transitioning nature of the IT regime in Russia.
  - The Phillips curve implied slack (time-varying NAIRU) is smoother than unemployment by construction.
  - Russia’s estimated NAIRU has been declining since 2000 on average, although it rises during crisis episodes.
  - The wage structure includes a flexible and a fixed component; the flexible component is cyclical and adjusts during the business cycle.
- Time variation patterns:
  - The effect of the unemployment gap on inflation increases before crisis episodes and declines post-crisis with some lag.
  - The slope of the Phillips Curve flattened in the aftermath of the GFC, started to recover in 2012, and recovered until the beginning of the recent crisis in 2015. This suggests potential slope-sharpening as recovery proceeds.
  - The importance of import price inflation increased over time up to the onset of the sanctions.
  - From 2015 onward, sanctions on some foods and compression of consumption and expenditure switching impacted imports of goods, reducing the inflation elasticity to import prices.
  - The impact of REER on core inflation was mostly steady and small until the 2014 move to a floating exchange rate regime, which lowered the absolute value of this coefficient.
- Model fit over time:
  - The Phillips curve fits the inflation data reasonably well, with predicted values matching data except for episodes with particularly large inflation spikes.
  - From 2012 onward the model-implied inflation slightly lags actual inflation, but overall the model can be used to forecast inflation.

### Cross-country comparisons (Russia, EMs, and AEs)
- Weight on inflation expectations:
  - EMs have lower weight on inflation expectations than AEs, reflecting less mature IT frameworks in EMs.
  - After establishing the IT regime in Russia, the weight of inflation expectations in the Phillips curve has risen and is aligned with the average of EMs.
- Slope of the Phillips Curve:
  - The slope of the Phillips Curve is steeper in Russia than on average in EMs. The average reaction to cyclical unemployment in Russia is higher than in other EMs, implying monetary policy should be more watchful of slack.
- Relative import price inflation:
  - Historically, Russia’s weight on relative import price inflation is in line with other EMs.
  - The weight of relative import price on inflation in AEs is larger than EMs, due to higher import penetration in AEs, though scales are small.
  - In Russia, this weight declined after the imposition of sanctions in 2014, implying import price inflation has a smaller impact on domestic inflation since then.

### Conclusions and policy implications
- Summary conclusions:
  - A time-varying parameters hybrid NK Phillips curve estimated with Russian data (2000–2016) indicates the Phillips curve is alive.
  - The slope is expected to increase as the recovery gets underway, emphasizing the role of slack in monetary policy decisions.
  - The weight on inflation expectations has increased recently, reflecting progress in establishing an IT regime in Russia.
  - The PC-implied NAIRU shows slack has been coming down fast post-GFC, reflecting a flexible labor market in Russia.
  - The fitted values demonstrate that the Phillips curve is overall a good model for explaining inflation dynamics and can be used for forecasting.
- Policy implications:
  - Monetary policy should pay attention to developments in slack because the slope of the Phillips curve can strengthen during recoveries.
  - Continued efforts to anchor inflation expectations support the effectiveness of monetary policy via higher weight on long-run expected inflation in the Phillips curve.

*Sources: Haver Analytics; and IMF staff calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17198.pdf_