## 1. Electronic Invoicing Adoption and VAT Revenues

## Source details

**Canonical URL:** [1. Electronic Invoicing Adoption and VAT Revenues](https://www.imf.org/-/media/files/publications/cr/2020/english/1perea2020002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2020/english/1perea2020002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2020/english/1perea2020002.pdf.json)

---

### Introduction and overall context
- Peru’s gains in revenue mobilization were almost nil in 1999–2017 despite strong macroeconomic performance and fiscal reforms.
- Public debt declined from around 50 percent of GDP in 1999 to less than 26 percent of GDP at end-2018.
- Ambitious revenue mobilization target: 1 percent of GDP over 2019–20.
- Large spending needs (including infrastructure gaps) and an uncertain external outlook pose risks to meeting fiscal needs.

### Tax policy structure and revenue trends
- Taxes shifted away from tariffs toward income and sales:
  - Taxes on imports declined from 1.7 percent of GDP in 1999 to 0.2 percent of GDP in 2018 (page 16).
- IGV (VAT) showed lower net growth over 1999–2017; IR (corporate and personal income tax) contributed a larger share over time but is more volatile than IGV.
- VAT rate change:
  - VAT rate lowered from 19 to 18 percent in 2011 — likely weakened VAT collection gains relative to peers.
- VAT performance relative to peers:
  - Over 1999–2017, Peru’s growth in VAT revenues was well below the Latin American average of 1.9 percentage points of GDP.
- VAT gap and c-efficiency:
  - VAT gap reduced from 50 percent in 2003 to below 35 percent in 2018.
  - Peru outperforms several peer countries with higher income levels in terms of VAT c-efficiency.
  - The VAT policy gap for Peru is lower than in European countries.

### VAT gap dynamics and risks
- The VAT gap is countercyclical and co-moves with the commodity cycle:
  - Largest reduction during the commodity boom of the mid 2000s.
  - Weakening commodity prices have led to reversion to moderately higher VAT gaps.
- Implication: a weaker external outlook could further exacerbate low VAT collection.
- Electronic invoicing is a policy lever to improve compliance, but complementary tax administration measures are critical.

### Electronic invoicing (e-invoicing) — empirical findings (Box 1)
- Reform timeline and scope:
  - In 2014 Peru mandated a switch from paper invoices to electronic invoices for VAT filing.
  - From October 2014, successive waves of firms were mandated to shift, starting with the largest firms and gradually extending to medium and small firms.
- Mechanism:
  - E-invoicing reduces cost and complexity of aggregating and cross-checking transactions data in real time, enhancing detection of misreporting and lowering compliance costs.
- Empirical results (IMF and SUNAT administrative data study):
  - Small firms (defined as firms with sales under 1700 UIT—Unidad Impositiva Tributaria––equivalent to S/4,200 in 2019) reported higher sales, purchases, and taxable value added one year after adoption.
  - These small firms showed over 5 percent higher VAT payments in the first year after adoption.
  - No such results were found among larger taxpayers, which constitute over 80 percent of VAT collections in Peru.
  - Sectoral effects larger in retail, services, and construction.
- Firms with outstanding VAT credits (representing 40 percent of the total):
  - No increase in VAT liabilities or VAT payments observed after e-invoicing introduction.
  - Larger increases in reported taxable purchases than sales among these firms increased their stock of VAT credits.
  - Indicates need for effective control of VAT credits to realize e-invoicing benefits.
- Interpretation and limitations:
  - Results present a lower bound of potential benefits; e-invoicing roll-out is ongoing and more firms being inducted could create larger aggregate effects.
  - SUNAT made no significant change to its risk management strategy over the period examined; complementary audit strategies using e-invoicing data could yield larger reductions in noncompliance.

### Tax expenditures and excise reforms
- Consumption tax expenditures of 1.6 percent of GDP (higher than many regional countries).
- Two largest VAT tax expenditures in nominal terms in 2016:
  - Agriculture sector: 6.4 percent of total VAT collections and 2.4 percent of overall tax revenues.
  - Amazonia region: 5 percent of total VAT collections.
- Reforms:
  - 2018 elimination of VAT exemptions in the Amazonia region and tax refunds for infrastructure development in Loreto, in exchange for annual transfers to regional governments — expected to improve revenue collections and efficiency of transfers.
- Excise tax changes (May 2018):
  - ISC on fuel, vehicles, alcoholic beverages, other beverages and tobacco were all raised.
  - Anticipated annual impact on revenue: 0.3 percent of GDP.
  - As of September 2019, excise revenues increased by 30 percent year-on-year, relative to 1.8 percent in 2017–18, contributing 0.1 percent of additional revenue.
  - Underperformance in 2019 due to avoidance by alcoholic beverage sectors; authorities expect full performance in 2020 after legal actions.

### Corporate tax policy and small business regimes (Box 2)
- Corporate tax rate changes:
  - Maximum corporate tax rate reduced in 2015 to 28 percent.
  - Rate increased to 29.5 percent in 2017, but impact on collections has been limited partly due to tax stability agreements.
- Corporate tax revenue outcomes:
  - Corporate tax revenues declined from an average of 4 percent of GDP in 2013–14 to an average of 3 percent of GDP in 2015-2017.
  - Reduction intended to be balanced by increase in tax on dividends, which was not approved by Congress.
- Multiple tax regimes and their effects:
  - Until 2017, businesses could register under two simplified schemes for micro and mid-sized businesses and one general scheme.
  - 2017 reform introduced RMT (Regimen MYPE Tributario) with a bracket of lower marginal tax rates for smaller sales, intended to encourage transition to the General Regime (GR).
  - The proliferation of multiple regimes has contributed to deterioration of corporate tax revenues by creating opportunities for misreporting and evasion.

Box 2 — design, effects, and recommendations
- Regimes before 2017:
  - NRUS (Nuevo Regimen Unico Simplificado): for microbusinesses; single fee based on sales and purchases; ineligible to provide invoices for VAT refunds; firms in the General Regime can deduct up to 6 percent of their purchase with payment slips instead of invoices.
  - RER (Regimen Especial del Impuesto a la Renta): taxes firms on gross sales and permits them to generate purchase invoices for larger firms in the general regime.
- 2017 addition:
  - RMT (Regimen MYPE Tributario): intermediate regime to ease transition from NRUS into the general regime.
    - sales under 15 UIT being taxed at 10 percent
    - sales above 15 UIT taxed at 29.5 percent (same as all sales of firms in the general regime)
  - UIT – Unidad Impositiva Tributaria – equivalent to S/4,200 in 2019.
- Observed effects after RMT introduction:
  - Large outflow of firms from the General Regime (GR) into the RMT, resulting in an overall loss in corporate tax revenues owing to the lower marginal rates in RMT.
  - Minimal effect on the number of firms under NRUS, suggesting RMT progressivity failed to convince firms to switch from RER or NRUS.
  - Introduction of RMT increased the number of thresholds in the tax system susceptible to arbitration by firms through misreporting.
- Identified problems:
  - Low rate at which sales are taxed in the RER creates opportunities to generate false invoices for firms in the general regime.
  - RMT increased threshold complexity and facilitated misreporting and tax arbitrage.
  - Peru’s burden of taxes is not cited as a major reason for not formalizing, implying simplification rather than lower marginal rates may be more relevant.
- Policy recommendations:
  - Rationalize and reduce the number of regimes for small businesses to prevent leakage from the general regime and reduce arbitrage opportunities.
  - IMF technical advice recommends eliminating the RMT and RER regimes and replacing them with a new simplified regime that:
    - Has the same marginal tax rate as the General Regime to prevent arbitrage at thresholds.
    - Uses simplified accounting (for example, using cash flow as the tax base) to lower cost of compliance and encourage registration of sales and purchase invoices.
    - Would increase incentives to formalize workers because firms could deduct labor costs from cash flow.
  - Expected fiscal impact from rationalization:
    - Revenue gains of around 0.14 percent of GDP from consolidating the RER and RMT into a single simplified regime with the same marginal rate as the General Regime.

Complementary tax-administration findings (from same source)
- Historically, lowest observed VAT gap was 29 percent in 2014.
- Under plausible assumptions, VAT collections could increase by 0.17 percent of nominal GDP in 2019 from a reduction of the VAT gap to 29 percent.
- SUNAT estimated the VAT gap narrowed by 3.8 percent to 32.9 percent in 2018.
- Other numeric findings:
  - In 2017, 78.2 percent of taxpayers declared income subject to zero marginal tax and an additional 14 percent were in the lowest marginal tax bracket, resulting in an effective tax rate of 5.6 percent across all taxpayers.
  - Property valuations used to establish tax obligations are estimated up to 200–300 percent below market valuations.
  - The number of control orders increased by 43 percent in 2018 (reaching 84,316 cases) and the amount of tax revenues subject to control under these cases reached 2,408 million soles, with a growth of 31 percent.
  - SUNAT’s tax recovery from control procedures increased by 75 percent between 2017 and 2018.

### Policy implications and priorities
- Short-term and medium-term priorities:
  - Strengthen tax administration to leverage e-invoicing data (risk management, audits, control of VAT credits).
  - Reform tax regimes for small businesses to reduce opportunities for misreporting and evasion.
  - Continue reforms that reduce inefficient tax expenditures and broaden the tax base.
  - Further reform of excises and enhancement of personal income tax coverage would be necessary in the medium term to meet spending needs.
- Risks and considerations:
  - Historical strong co-movement between tax revenues and commodity prices implies external shocks can sharply affect revenue mobilization.
  - Complementary tax administration strategies are necessary for e-invoicing to yield substantial aggregate VAT gains, especially given that large taxpayers constitute over 80 percent of VAT collections.

*Source: 1perea2020002 - 1. Electronic Invoicing Adoption and VAT Revenues*

### 1. Electronic Invoicing Adoption and VAT Revenues ____________________________________ 7

### 1. Electronic Invoicing Adoption and VAT Revenues

### Introduction and overall context
- Peru’s gains in revenue mobilization were almost nil in 1999–2017 despite strong macroeconomic performance and fiscal reforms.
- Public debt declined from around 50 percent of GDP in 1999 to less than 26 percent of GDP at end-2018.
- Ambitious revenue mobilization target: 1 percent of GDP over 2019–20.
- Large spending needs (including infrastructure gaps) and an uncertain external outlook pose risks to meeting fiscal needs.

### Tax policy structure and revenue trends
- Taxes shifted away from tariffs toward income and sales:
  - Taxes on imports declined from 1.7 percent of GDP in 1999 to 0.2 percent of GDP in 2018 (page 16).
- IGV (VAT) showed lower net growth over 1999–2017; IR (corporate and personal income tax) contributed a larger share over time but is more volatile than IGV.
- VAT rate change:
  - VAT rate lowered from 19 to 18 percent in 2011 — likely weakened VAT collection gains relative to peers.
- VAT performance relative to peers:
  - Over 1999–2017, Peru’s growth in VAT revenues was well below the Latin American average of 1.9 percentage points of GDP.
- VAT gap and c-efficiency:
  - VAT gap reduced from 50 percent in 2003 to below 35 percent in 2018.
  - Peru outperforms several peer countries with higher income levels in terms of VAT c-efficiency.
  - The VAT policy gap for Peru is lower than in European countries.

### VAT gap dynamics and risks
- The VAT gap is countercyclical and co-moves with the commodity cycle:
  - Largest reduction during the commodity boom of the mid 2000s.
  - Weakening commodity prices have led to reversion to moderately higher VAT gaps.
- Implication: a weaker external outlook could further exacerbate low VAT collection.
- Electronic invoicing is a policy lever to improve compliance, but complementary tax administration measures are critical.

### Electronic invoicing (e-invoicing) — empirical findings (Box 1)
- Reform timeline and scope:
  - In 2014 Peru mandated a switch from paper invoices to electronic invoices for VAT filing.
  - From October 2014, successive waves of firms were mandated to shift, starting with the largest firms and gradually extending to medium and small firms.
- Mechanism:
  - E-invoicing reduces cost and complexity of aggregating and cross-checking transactions data in real time, enhancing detection of misreporting and lowering compliance costs.
- Empirical results (IMF and SUNAT administrative data study):
  - Small firms (defined as firms with sales under 1700 UIT—Unidad Impositiva Tributaria––equivalent to S/4,200 in 2019) reported higher sales, purchases, and taxable value added one year after adoption.
  - These small firms showed over 5 percent higher VAT payments in the first year after adoption.
  - No such results were found among larger taxpayers, which constitute over 80 percent of VAT collections in Peru.
  - Sectoral effects larger in retail, services, and construction.
- Firms with outstanding VAT credits (representing 40 percent of the total):
  - No increase in VAT liabilities or VAT payments observed after e-invoicing introduction.
  - Larger increases in reported taxable purchases than sales among these firms increased their stock of VAT credits.
  - Indicates need for effective control of VAT credits to realize e-invoicing benefits.
- Interpretation and limitations:
  - Results present a lower bound of potential benefits; e-invoicing roll-out is ongoing and more firms being inducted could create larger aggregate effects.
  - SUNAT made no significant change to its risk management strategy over the period examined; complementary audit strategies using e-invoicing data could yield larger reductions in noncompliance.

### Tax expenditures and excise reforms
- Consumption tax expenditures were relatively high:
  - Consumption tax expenditures of 1.6 percent of GDP (higher than many regional countries).
  - Two largest VAT tax expenditures in nominal terms in 2016:
    - Agriculture sector: 6.4 percent of total VAT collections and 2.4 percent of overall tax revenues.
    - Amazonia region: 5 percent of total VAT collections.
- Reforms:
  - 2018 elimination of VAT exemptions in the Amazonia region and tax refunds for infrastructure development in Loreto, in exchange for annual transfers to regional governments — expected to improve revenue collections and efficiency of transfers.
- Excise tax changes (May 2018):
  - ISC on fuel, vehicles, alcoholic beverages, other beverages and tobacco were all raised.
  - Anticipated annual impact on revenue: 0.3 percent of GDP.
  - As of September 2019, excise revenues increased by 30 percent year-on-year, relative to 1.8 percent in 2017–18, contributing 0.1 percent of additional revenue.
  - Underperformance in 2019 due to avoidance by alcoholic beverage sectors; authorities expect full performance in 2020 after legal actions.

### Corporate tax policy and small business regimes
- Corporate tax rate changes:
  - Maximum corporate tax rate reduced in 2015 to 28 percent.
  - Rate increased to 29.5 percent in 2017, but impact on collections has been limited partly due to tax stability agreements.
- Corporate tax revenue outcomes:
  - Corporate tax revenues declined from an average of 4 percent of GDP in 2013–14 to an average of 3 percent of GDP in 2015-2017.
  - Reduction intended to be balanced by increase in tax on dividends, which was not approved by Congress.
- Multiple tax regimes and their effects:
  - Until 2017, businesses could register under two simplified schemes for micro and mid-sized businesses and one general scheme.
  - 2017 reform introduced RMT (Regimen MYPE Tributario) with a bracket of lower marginal tax rates for smaller sales, intended to encourage transition to the General Regime (GR).
  - The proliferation of multiple regimes has contributed to deterioration of corporate tax revenues by creating opportunities for misreporting and evasion.

### Policy implications and priorities
- Short-term and medium-term priorities:
  - Strengthen tax administration to leverage e-invoicing data (risk management, audits, control of VAT credits).
  - Reform tax regimes for small businesses to reduce opportunities for misreporting and evasion.
  - Continue reforms that reduce inefficient tax expenditures and broaden the tax base.
  - Further reform of excises and enhancement of personal income tax coverage would be necessary in the medium term to meet spending needs.
- Risks and considerations:
  - Historical strong co-movement between tax revenues and commodity prices implies external shocks can sharply affect revenue mobilization.
  - Complementary tax administration strategies are necessary for e-invoicing to yield substantial aggregate VAT gains, especially given that large taxpayers constitute over 80 percent of VAT collections.

*Source: 1perea2020002 - 1. Electronic Invoicing Adoption and VAT Revenues*

### Box 2. Tax Regimes for Small Businesses in Peru

### Box 2. Tax Regimes for Small Businesses in Peru

### Design objectives and trade-offs
- Special regimes for taxation of small business must be designed to increase formalization while discouraging tax evasion and avoidance.
- Taxation policy (marginal tax rates, average tax rates and the complexity of the taxation system) is identified as a key factor discouraging formalization; there is a direct relationship between ease of paying taxes and the size of the informal sector.
- Design must balance reducing tax compliance costs (for example, reducing accounting requirements) with limiting use of special regimes for tax evasion and tax avoidance in the larger economy.

### Description of regimes and recent reform
- Until 2017, small businesses could choose between:
  - NRUS (Nuevo Regimen Unico Simplificado): for microbusinesses; allows firms to pay corporate tax and VAT as a single fee based on sales and purchases. Firms filing under NRUS are ineligible to provide invoices to their buyers for VAT refunds; they can provide payment slips, and firms in the General Regime can deduct up to 6 percent of their purchase with payment slips instead of invoices.
  - RER (Regimen Especial del Impuesto a la Renta): taxes firms on gross sales and permits them to generate purchase invoices for larger firms in the general regime.
- In 2017 the government added a third regime:
  - RMT (Regimen MYPE Tributario): created as an intermediate regime to ease transition from NRUS into the general regime. For eligible firms under a sales threshold, the corporate tax rate was made progressive:
    - sales under 15 UIT being taxed at 10 percent
    - sales above 15 UIT taxed at 29.5 percent (same as all sales of firms in the general regime)
  - UIT – Unidad Impositiva Tributaria – equivalent to S/4,200 in 2019.
- Effects observed after the RMT introduction:
  - Large outflow of firms from the General Regime (GR) into the RMT, resulting in an overall loss in corporate tax revenues owing to the lower marginal rates in RMT.
  - Minimal effect on the number of firms under NRUS, suggesting RMT progressivity failed to convince firms to switch from RER or NRUS.
  - Introduction of RMT increased the number of thresholds in the tax system susceptible to arbitration by firms through misreporting.

### Identified problems and empirical concerns
- The low rate at which sales are taxed in the RER creates opportunities for firms in the RER to generate false invoices for firms in the general regime, enabling tax evasion.
- The fourth regime (RMT) increased threshold complexity and facilitated misreporting and tax arbitrage.
- Peru’s burden of taxes is not cited as a major reason for not formalizing, implying simplification rather than lower marginal rates may be more relevant.

### Policy recommendations and expected impacts
- Rationalize and reduce the number of regimes for small businesses to prevent leakage from the general regime and reduce arbitrage opportunities.
- IMF technical advice recommends eliminating the RMT and RER regimes and replacing them with a new simplified regime that:
  - Has the same marginal tax rate as the General Regime to prevent arbitrage at thresholds.
  - Uses simplified accounting (for example, using cash flow as the tax base) to lower cost of compliance and encourage registration of sales and purchase invoices.
  - Would increase incentives to formalize workers because firms could deduct labor costs from cash flow.
- Expected fiscal impact from rationalization:
  - Revenue gains of around 0.14 percent of GDP from consolidating the RER and RMT into a single simplified regime with the same marginal rate as the General Regime.

### Complementary tax-administration and revenue considerations (contextual links in source)
- Improvements in tax controls, electronic invoicing, and stronger risk profiling (SUNAT reforms) are highlighted as necessary complements to regime rationalization to reduce compliance gaps and increase VAT revenues.
- VAT-related findings and potential gains:
  - Historically, lowest observed VAT gap was 29 percent in 2014.
  - Under plausible assumptions, VAT collections could increase by 0.17 percent of nominal GDP in 2019 from a reduction of the VAT gap to 29 percent.
  - SUNAT estimated the VAT gap narrowed by 3.8 percent to 32.9 percent in 2018.
- Other numeric findings in related tax-administration context (from same source):
  - In 2017, 78.2 percent of taxpayers declared income subject to zero marginal tax and an additional 14 percent were in the lowest marginal tax bracket, resulting in an effective tax rate of 5.6 percent across all taxpayers.
  - Property valuations used to establish tax obligations are estimated up to 200–300 percent below market valuations.
  - The number of control orders increased by 43 percent in 2018 (reaching 84,316 cases) and the amount of tax revenues subject to control under these cases reached 2,408 million soles, with a growth of 31 percent.
  - SUNAT’s tax recovery from control procedures increased by 75 percent between 2017 and 2018.

*Source: Box 2. Tax Regimes for Small Businesses in Peru (from the provided IMF country report content).*

### Introduction

### Introduction

### Background and monetary framework
- Two decades of severe macroeconomic imbalances and hyperinflation until the early 1990s entrenched low confidence in the national currency and widespread dollarization.
- Monetary regulation enhancements and instruments introduced: certificates of deposit (1994) and repos (1997).
- Reserve requirements (RRs) on domestic deposits were reduced substantially in 1991–93 to foster an effective interbank market.
- The BCRP (Banco Central de Reserva del Perú) is constitutionally mandated to maintain monetary stability and adopted an inflation targeting framework in 2002, becoming the first monetary authority to implement it under a dual monetary system.
  - The target range for inflation was initially set at 1-3 percent.
  - In 2002, the target was set to 2.5 per cent with a tolerance margin of +/– 1 per cent.
  - From 2007 onwards, the target was lowered to 2 percent with the same tolerance margin.
- The reference interest rate has been the main monetary policy instrument; foreign-exchange intervention—sterilized to preserve the inflation objective—remained a key policy tool because of exchange rate effects on dollar liabilities.

### Financial system reform and macroprudential policy
- The crisis at the end of the 1990s led to bank failures and a collapse in credit; the SBS (Superintendencia de Banca, Seguros y AFP) strengthened banking supervision and the regulatory framework.
- New macroprudential tools were introduced to enhance resilience, dampen the credit cycle, prevent liquidity shortages, and mitigate risks from dollarization.
- Both the BCRP and the SBS operate macroprudential policy tools:
  - BCRP’s instruments target mainly liquidity risk, both in domestic and foreign currency, and excessive credit growth.
  - SBS’s instruments focus primarily on strengthening the financial system’s loss absorption capacity.
- Reserve requirements (RRs) and capital flow measures (CFMs) have a special status in the policy taxonomy; in the paper RRs are classified under MPPs following Alam et al. (2018).

### Organization of the paper
- Section B: concept of the policy triangle characterizing Peru’s multi-instrument framework since 2002.
- Section C: events leading to the Global Financial Crisis (GFC), 2002–07.
- Section D: the GFC and aftermath, 2008–11.
- Section E: episodes of global risk aversion after the GFC, 2012–19.
- Section F: macroeconomic performance over 2002–19.
- Section G: conclusions.
- Figures 1–5 in the Annex present performance of key macroeconomic variables; Annex I details policy trade-offs with the role of financial stability.

### A stylized representation: the policy triangle
- The policy triangle represents combinations of three policy objectives: price stability, financial stability, and exchange rate stability (the latter instrumental to the first two).
- The three main policy instruments (vertices) are:
  - Policy interest rate (MP)
  - Exchange rate intervention (FXI)
  - Macro-prudential policies (MPPs)
- Movements inside the triangle indicate revealed preferences of policymakers in response to shocks:
  - Movement toward a vertex: increased use of that instrument over time.
  - Movement toward a side: changing trade-offs between objectives in response to shocks.

### Broad evolution (2002–19) summarized
- 2002–03: large capital inflows and favorable terms of trade; FXI used intensely to accumulate reserves and reduce exchange rate pressures.
- Up to the GFC: continued movement toward more FXI usage with initially little use of MPPs (points 1 and 2).
- 2008–11 (GFC and aftermath): terms of trade worsened and capital inflows stalled; many new MPPs and RRs were used together with foreign currency sales (point 3).
- 2012–16: with declining terms of trade and volatile capital flows, MPPs were aggressively deployed and more exchange rate flexibility allowed.
- Last years covered: continued decline in terms of trade and volatile capital flows; policy rate has been used more intensely and FXI less so, while relying on a large number of MPP tools.

### The commodity price boom (2002–07)
- External environment: large metal price increases created appreciation pressures.
- Foreign exchange reserves:
  - Between January 2002 and December 2007, foreign exchange (FX) reserves increased by more than US$18 billion (12 percent of GDP), nearly all sterilized.
- Dollarization remained high at end-2005:
  - 55 percent of broad money denominated in foreign currency.
  - 70 percent of credit to the private sector denominated in foreign currency.
- Monetary and prudential policy actions:
  - After record low interest rates in early 2002, the BCRP raised its reference rate by 150 basis points from September in response to regional uncertainties, FX pressures, and rising inflationary pressures.
  - Banks were required to assess exchange-rate-related credit risks; mortgage lending in foreign currency to unhedged borrowers was scrutinized.
  - Reserve requirements, both in domestic and foreign currency, were increased.
  - Loan loss provision requirements were augmented to include dynamic provisioning and sectoral provisions (housing loans in particular).
- Outcomes:
  - Inflation reduced to about 2 percent on average.
  - Output gap was negative at 0.5 percent (reflecting disinflation effort).
  - Movement from point 1 to point 2 in the policy triangle (more MP use alongside reserve accumulation and early prudential tightening).

### The Global Financial Crisis (2008–11)
- External shock: commodity prices declined and capital flows became very volatile.
- BCRP policy response included a broad range of unconventional instruments:
  - Currency swaps, repo operations, repurchase of central bank certificates of deposit.
  - Sold nearly US$7 billion (20 percent of international reserves).
- Policy rate response:
  - Starting in February 2009 the BCRP began a series of aggressive rate cuts, with the reference rate reduced overall by 525 bps to 1.25 percent.
  - Later, as recovery proceeded in mid-2010, the policy rate was increased by a total of 300 bps to 4.25 percent by May 2011.
- Macroprudential and reserve requirement measures:
  - RRs were tightened considerably to curb credit growth, which stayed at about 20 percent in 2010–11, with foreign currency credit growth surpassing domestic currency credit growth.
  - Marginal requirement for domestic currency deposits increased from 15 to 25 percent.
  - Required ratio for non-residents increased from 15 to 120 percent by October 2008.
  - RRs for foreign currency deposits and short-term debt increased from 30 percent to 49 percent by September 2008.
  - RRs on long-term external liabilities were reintroduced at 9 percent in July 2008.
  - First phase of comprehensive reform of capital requirements announced by SBS in July 2008; in 2010 new asset-based instruments were created.
  - First CFM introduced in February 2010 as a RR on foreign credit lines and other external obligations.
- Interaction and outcomes:
  - At times opposite movements of RRs, MPPs, and policy rate changes could have weakened monetary policy transmission.
  - Large use of FXI and macroprudential policies may have helped preserve financial stability.
  - Inflation expectations temporarily increased and the inflation target was missed twice.
  - Peru performed better than many peers on inflation; growth picked up quickly.
- Movement in the policy triangle: from point 2 to point 3, reflecting heavier use of FXI and MPPs during the crisis.

### Global risk aversion events (2012–19)
- Triggering events: China’s slowdown in 2012, U.S. fiscal policy uncertainty and the taper tantrum in 2013, Renminbi depreciation in 2016, trade and geopolitical tensions in 2017–19.
- 2012 developments:
  - Net capital inflows in the first three quarters of 2012 were over $16 billion (about 8 percent of GDP), only about half of which was foreign direct investment.
  - Nonresident holdings of domestically-issued debt continued to increase, reaching over half of the total (about 3.5 percent of GDP) in 2012.
  - The BCRP kept the policy rate unchanged at 4.25 percent until the third quarter of 2013 and placed large amounts of central bank securities to sterilize FX purchases.
  - Gross international reserves reached US$64 billion (32 percent of GDP).
  - Despite large intervention, the sol appreciated about 4 percent (yoy) against the US dollar and by almost 9 percent (yoy) in real effective terms during January–November 2012.
- Macroprudential steps and capital reforms:
  - RRs were further tightened through November 2013.
  - Second phase of capital reforms commenced: allowing more investment abroad of private pension funds; prepayment of external debt; larger accumulation of FX to repay debt; significant increase in the Fiscal Stabilization Fund.
  - A countercyclical capital buffer as part of the second phase of raising capital requirements was introduced in 2012 through October 2014; the noncyclical capital buffer included concentration risk buffers, a risk-propensity buffer, and a very small systemic risk buffer for the four largest banks.
- 2013 policy easing and subsequent cuts:
  - External conditions deteriorated rapidly in 2013, prompting a preemptive easing of monetary policy, FX sales, and new MPPs to support monetary policy.
  - The BCRP cut the policy rate by 0.25 bps to 4 percent in November 2013, citing slower domestic and global growth and declining inflationary expectations.
  - The reference rate was further reduced in steps to 3.25 percent through January (text ends).

### Selected calculated averages and indicators (as presented)
- For the periods 2002q1–2007q4, 2008q1–2011q4, 2012q1–2019q1, selected averages and volatilities included:
  - Inflation: 2.71, 11.92, 3.41, 2.97 (period averages listed in table format in the source).
  - Output gap: -0.08, -0.46, 0.05, 0.17.
  - Inflation volatility: 1.08, 1.14, 1.79, 0.65.
  - Output volatility: 1.90, 1.67, 3.41, 1.25.
  - Real credit growth, y-y, average for the period: 9.10, 5.03, 16.53, 8.36.
  - REER y-y change, average for the period: 0.07, -1.67, 1.66, 0.64.
  - REER volatility: 3.00, 2.50, 2.69, 3.59.
  - FXIs, net million USD: 43375.30, 20306.10, 15461.90, 7607.30.
  - FXI, in percent of GDP: 30.29, 27.36, 10.99, 3.73.
  - MPP measures (share of total): 1, 2, 2, 9, 59 (as provided in the source table).
  - NIR, average for the period, million USD: 41926.56, 15247.33, 39313.00, 63242.75.
  - NIR, in percent of GDP: 29.28, 20.55, 27.94, 31.03.
  - Notes in source: volatility measured as 8 quarters rolling standard deviation; T stands for tightening measures, L for loosening measures; MP stance measured by change in policy rate for the entire period by end points (T tightening, L loosening, N neutral).

*Prepared by Manuk Ghazanchyan, Joannes Mongardini, and Alexei Goumilevski (Introduction section).*

### 2015. RRs on domestic currency deposits were also cut by

### 2015. RRs on domestic currency deposits were also cut by

### Reserve Requirements, Macroprudential Policies (MPPs), and FX Intervention (FXI)
- RRs on domestic currency deposits were cut by a cumulative 23 bps to 7 percent between August 2013 and May 2015 and the legal minimum requirement rate was further reduced from 9 percent to 6.5 percent by year-end.
- The BCRP raised marginal RRs on FX deposits from 60 to 70 percent and introduced additional RRs on short positions in FX derivatives and others conditional on FX loan balances.
- Two additional RRs for financial institutions considered as CFMs were introduced in February and March 2015.
- Other MPPs introduced or tightened:
  - Increasing capital buffers on FX loans to reduce FX-related credit risk.
  - Additional capital buffers on consumption loans.
  - Higher risk weights on mortgages with high loan-to-value (LTV) ratios.
  - New liquidity coverage ratios (LCRs) introduced in 2014; in January 2015, LCRs in both local and foreign currency were increased from 90 to 100 percent.

### Policy Actions and Evolution (2015–2019)
- 2016:
  - The BCRP raised the policy rate from September 2015 by a full percentage point to 4.25 percent.
  - The sol depreciated by 14 percent with respect to the U.S. dollar by the end of 2015 over a year.
  - The SBS continued to tighten the countercyclical capital buffers on past lending.
- 2017–2019:
  - The BCRP gradually reduced the policy rate from 4.25 percent in May 2017 to 2.75 percent in April 2018, and later to 2.25 percent after two 25 bps cuts in August and November 2019.
  - The BCRP reduced reserve requirements to 5 percent for domestic currency deposits and 35 percent for FX deposits in mid-2017 and mid-2018, respectively, while introducing new limits on banks’ FX transactions.
  - March 2018: measures allowing foreign currency loans to increase if local currency credit to enterprises rises more than proportionally, subject to thresholds.
  - By September 2018: operational limit on private pension funds’ investment in foreign securities increased from 46 to 50 percent of their portfolio.
  - March 2019: the BCRP relaxed limits on FX operations with forwards and swaps.
  - Use of FXIs limited in size amounting to less than 3 percent of GDP in 2018-2019 with balances on swap agreements reduced to zero and broadly symmetric.

### Outcomes: De-dollarization, Inflation, and Financial Stability
- Dollarization trends:
  - Latest data shows that credit and deposit dollarization in Peru have declined to about 27 and 35 percent from more than 65 and 55 percent, respectively in the early 00s.
- Monetary policy and instrument mix:
  - As dollarization declined, the use of FXI was reduced in favor of a more active use of the policy rate, while the MPP toolkit was further expanded.
  - Augmented Taylor rule estimations suggest the policy rate in Peru was more expansionary and less volatile than predicted by the rules when including terms for the REER, real credit, and the US policy rate.
  - Significant coefficients for FXI and a variable for financial stability proxied by the domestic lending-deposit interest rate spread suggest that, without these policies, the reference rate would have had to be higher on average and more volatile to attain the inflation target.
- Inflation and macroeconomic performance:
  - Inflation performance was good relative to other EMEs although target misses were more frequent than in Chile and Brazil.
  - With a strong and resilient financial system, Peru was one of the best EME performers in terms of output growth, inflation, and inflation volatility; output volatility, however, was higher than in most EMEs.
- FXI effectiveness:
  - Model simulations (using the FPAS model) show FXI can mitigate output responses to real and financial shocks when balance sheet effects from dollarization are large.
  - With dollarization declining, the positive impact of FXI correspondingly diminishes.

### Evidence on MPPs and Credit Cycle
- Character of MPPs:
  - Despite episodic reductions in RRs, Peru’s MPPs have been mostly tightening; Alam and others (2018) classify Peru’s MPPs as tightening, a large part capital-related, followed by provisioning and dollarization measures.
  - Most MPPs adopted have never been removed, complicating evaluation of their effectiveness.
- Empirical findings:
  - Fang et al., (2018) find no effects of higher capital requirements on credit in Peru.
  - Regressions estimating MPPs’ impact on output and output volatility show:
    - No impact of MPPs on real GDP growth in the short run.
    - MPPs increase output volatility (at the 10-percent significance level only), an effect that can be reversed with deeper financial development and more capital account openness.
    - In the longer run (using 2- and 5-year averages), the impact of MPPs on growth and its volatility dissipates.
- Overall assessment:
  - MPPs seem to have been successful at dampening the credit cycle, especially in 2012-19, but the evidence on their impact is inconclusive.

### Conclusions and Policy Lessons
- Since adopting inflation targeting in 2002, Peru’s multi-instrument policy framework gradually moved away from reliance on FXI toward increased use of the policy rate and an expanding macroprudential toolkit.
- Key drivers of this evolution:
  - Success in gaining credibility and stabilizing inflation.
  - Limiting the impact of large capital flows.
  - Reducing dollarization, which improved the effectiveness of the interest rate transmission channel.
- Observed trade-offs and roles:
  - FXI may have contributed to stabilizing output fluctuations, but benefits decline with de-dollarization.
  - FXI and MPPs may have helped monetary policy achieve objectives.
  - MPPs appear to have played a role in dampening credit cycles but may have contributed to output fluctuations.
- Research needs:
  - Evidence is suggestive but lacks clear counterfactuals; further analysis is necessary to fully understand interactions among instruments and to design improved combinations of instruments and policies.

*Source: International Monetary Fund (selected excerpts from the Peru country chapter).*

### Annex I. Policy Goals and Trade-Offs

### Annex I. Policy Goals and Trade-Offs

### Methodological strategy
- Focus: frequencies and the significance of co-movements in macroeconomic variables during the adoption of inflation targeting, 2012–19.
- Expectations used to interpret comovements:
  - Output gap moves in the same direction as inflation except at times of supply shocks.
  - Exchange rate: depreciation associated with higher inflation; appreciation with lower inflation.
  - Financial stability / credit: credit expansions precede high inflation (policy rate hikes); when inflation is considerably below target, credit conditions are more likely contractionary.
  - Capital flows: inflows associated with higher inflation; outflows with lower inflation, other things being equal.
- Construction of indicators:
  - Standard deviation measured over 2002–Q2 2008 (pre-Lehman set) for variables to define booms/recessions and depreciation/appreciation quarters.
  - Discrete inflation status variable: -2 (below target lower bound), -1 (below target but within range), 1 (above target but within range), 2 (above target upper bound).
  - Capital flow measure: change in capital flows in GDP relative to pre-crisis mean ± one pre-crisis standard deviation.
  - Dummy variables for status changes used to test significance of correlations with inflation status and inflation gap.

### Key empirical findings on trade-offs and comovements
- General statement:
  - Stabilizing financial and foreign exchange market conditions may have conflicted with price and output stability in Peru; broader financial stability concerns at times can outweigh price stability concerns when monetary policy actions may not be enough.
- Co-movement overview:
  - While real credit and capital flows were generally correlated with inflation developments in Peru, exchange rate movements were not correlated with inflation in most periods owing to interventions.
  - Real credit was off sync in the immediate aftermath of the adoption of the IT framework but broadly in line with inflation in all other sub-periods.
  - Except for 2008–2011, capital flows were largely in line with movements of inflation though not with the exchange rate.
  - Where large negative output gaps were present (the pre-crisis period) and associated with higher inflation (perhaps due to supply shocks), the BCRP responded with policy hikes.
  - The closer link of capital flows with inflation in Peru suggests inflation was more correlated with financial rather than real shocks.
  - Deviations from orthodox inflation targeting (relative to Chile) were largely modest and transitory, differing primarily on policy rate setting and real exchange rate outcomes; policy rate decisions in Peru have been affected by reserve requirement settings, particularly in response to capital flow shocks.

### Regression coefficient estimates (selected)
- Notes:
  - Table shows coefficient estimates from a simple regression of the inflation gap (inflation minus its target value) separately from also the status index of inflation relative to target.
  - ***, **, and * indicate significance at the 1, 5, and 10 percent levels, respectively.
- Coefficients reported (as in source table):
  - Output gap:
    - 0.41***
    - 0.53***
    - 0.19
    - 0.31*
    - 1.04***
    - 0.94***
    - 0.08
    - 0.05
  - Change in ER:
    - 0.9*
    - 0.55
    - 0.08
    - 0.06
    - -0.19
    - -0.46
    - 2.62***
    - 3.69***
  - Change in Real credit:
    - 2.35***
    - 2.99***
    - 0.75
    - 1.41
    - 1.74
    - 2.18**
    - 2.56***
    - 2.66***
  - Share of net capital flows in GDP:
    - 1.29***
    - 1.24***
    - 1.33**
    - 1.40**
    - 0.31
    - 0.32
    - 2.09**
    - 2.69**

### Frequency-consistency findings (selected percentages and periods)
- Exchange rate movements consistent with inflation dynamics:
  - Between 2002-07: 57 percent of cases.
  - Between 2008-2011: 40 percent of cases.
  - Between 2012-2019: 65 percent of exchange rate movements were consistent with inflation dynamics.
- Capital flows consistent with inflation dynamics:
  - Between 2002-07: more than 80 percent of the time.
  - Between 2008-2011: more than 80 percent of the time.
  - Between 2012-2019: 80 percent of the time.
- Real credit consistent with inflation dynamics:
  - Between 2002-07: slightly less than half of the time.
  - Between 2008-2011: 80 percent of the time.
  - Between 2012-2019: 93 percent of the time.

### Comparative benchmarking and policy instruments
- Benchmark: Peru’s multi-instrument inflation targeting framework compared with Chile’s orthodox inflation targeting regime.
- Main dimensions of difference:
  - Policy rate setting: policy rate decisions in Peru have been affected by reserve requirement settings.
  - Real exchange rate outcomes: policy rate decisions, foreign exchange intervention, and reserve requirements seem to explain differences in real exchange rate outcomes.
- Interpretation: Deviations from orthodoxy were modest and transitory; multi-instrument use reflected responses to capital flow shocks and financial stability considerations.

*Source: IMF Staff calculations.*

### Introduction

### Introduction

### Multi-instrument policy framework in Peru
- Peru uses a multi-instrument policy framework to manage macroeconomic fluctuations, comprising:
  - a traditional policy rate (monetary policy rate),
  - foreign exchange intervention (FXI),
  - reserve requirements (RRs).
- FXI is used to manage exchange rate fluctuations to mitigate balance-sheet risks associated with financial dollarization.
- RRs are used to reduce financial dollarization and to partially offset the impact of changes in external credit conditions.

### Analytical objectives and benchmarking strategy
- Key research questions:
  - To what extent does the use of FXI and RRs affect the adjustment of the monetary policy rate?
  - To what extent does Peru’s framework generate policy-induced deviations of the real exchange rate (RER)?
- Benchmark: Chile, a regional peer that uses an orthodox inflation targeting framework with a flexible exchange rate and reliance almost exclusively on the monetary policy rate.
- Rationale for benchmarking Chile:
  - Both countries are inflation targetters and share important similarities (export structure dominated by mining and agricultural products).
  - Important policy-framework differences (Chile: flexible exchange rate and policy-rate adjustments; Peru: multi-instrument usage).

### Economic developments (January 2002–March 2019)
- Terms of trade:
  - Very similar trends and cycles for Peru and Chile.
  - Almost identical volatility (both in standard deviation and coefficient of variation).
- Net capital inflows:
  - Monthly average: Peru = 0.2 months of imports, Chile ≈ nearly zero.
  - Similar volatility (standard deviation and range); coefficient of variation not appropriate for Chile given near-zero mean.
- Real GDP growth and inflation:
  - Real GDP growth (average): Peru = 5.3 percent, Chile = 3.9 percent.
  - Inflation (average): Peru = 2.7 percent, Chile = 3.2 percent.
  - Volatility (coefficient of variation) lower for Peru for both series.
  - Wider inflation range for Chile driven by months around the global financial crisis (GFC).
- Exchange rate developments:
  - RER volatility (coefficient of variation): Chile = 5.6, Peru = 4.1.
  - RER range: Chile ≈ 30 points, Peru = 17.
  - Nominal exchange rate volatility (coeff. of variation): Chile = 13.5, Peru = 9.1.
  - Nominal exchange rate range: Chile = 46, Peru = 31.
  - Deviation corrections between nominal and real series tended to occur through adjustments in Chile’s nominal exchange rate; inflexion points largely coincide.

### Policy instruments: comparative behavior
- Monetary policy rates:
  - Cycles and averages similar in nominal and real terms.
  - Average real rate about 40 basis points higher in Peru.
  - Volatility higher in Chile, suggesting more active use of the policy rate in Chile.
- FXI and RRs:
  - Volatility of FXI and RRs significantly higher in Peru.
  - Chile barely used these instruments: FXI in just three small episodes; RRs were not changed at all.

### Other policy and structural differences potentially relevant
- Fiscal FX transactions:
  - Government ownership of mining/hydrocarbon resources higher in Chile than in Peru (in Peru resources mainly private).
  - Chile’s government accumulated a large stock of external assets during the commodity boom, alleviating pressures on the FX market.
  - Some FX transactions by Peru’s Fiscal Stabilization Fund and tax stability arrangements with mining companies may have mitigated FX market pressures (potential bias toward non-significance in analysis).
- Restrictions to capital flows:
  - Chile eliminated some 1990s capital controls but appears to have remained more restrictive than Peru (per Fernandez, Klein, Rebucci, Schindler, and Uribe (2016) index).
  - May explain lower and less volatile capital flows in Chile.
- Financial sector structure:
  - Banks are relatively more dominant in Peru’s financial sector than in Chile’s, which may make RRs more effective in Peru.

### Estimation strategy and data
- Empirical approach:
  - Analyze deviations (gaps) of Peruvian variables vis-à-vis Chilean counterparts.
  - Start from country-specific model: Yi,t = αi + β Xi,t + γ Zi,t + vt + ui,t  (equation (1)).
  - Difference specification for country pairs: Yi,t − Yj,t = (αi − αj) + β(Xi,t − Xj,t) + γ(Zi,t − Zj,t) + (ui,t − uj,t)  (equation (2)).
  - Advantage: differencing helps control for unobservable common shocks (vt).
- Data characteristics:
  - All variables measured monthly.
  - Sample period: January 2003–March 2019; most regression data start in September 2003 (first month policy rate for Peru available).
  - Real policy rate = nominal policy rate − 12-month inflation expectations.
  - FXI variable deflated by imports of goods and services (unit: central bank intervention in months of imports) to account for size differences and facilitate coefficient interpretation.
- Terminology:
  - “Gap” denotes differences between Peru’s economic variables and Chilean counterparts (policy rate gap, real exchange rate gap, inflation gap, terms-of-trade gap, FXI gap, etc.).

### Analysis and regression results — Policy question 1 (effect of FXI and RRs on monetary policy rate)
- Behavior of the real interest rate gap:
  - Stationary series with an average of around 0.4 percentage points (reflecting that the average real policy rate was higher in Peru by that amount).
  - Very large gaps around the GFC in both directions, but short-lived.
- Regression results (Table 1 summary):
  - Dependent variable: real interest rate gap.
  - Explanatory variables included: inflation gap and policy instrument gaps.
  - Key findings:
    - Inflation gap effect: a one percentage point increase in the inflation gap translates into a 0.11 percentage points increase in the real policy rate gap.
    - Reserve requirements (RRs) appear to act as a substitute for policy rate adjustments (negative coefficient).
    - Quantitative statement on RRs is incomplete in the source text (sentence ends mid-thought): "A 10-percentage point increase in reserve requirements translates into a reduction of" — source text cuts off at this point.

*Source: IMF staff text from "Introduction" (1perea2020002).*

### 0.2 percentage points in the real policy rate.

### 1perea2020002 - 0.2 percentage points in the real policy rate.

### Key findings on monetary policy instruments and outcomes
- Reserve requirements (RRs) played an important role in Peru’s response to capital inflows and influenced the intensity of policy rate use; a 10 percentage points increase in reserve requirements leads to the real interest rate being 0.2 percentage points lower than its orthodox benchmark.
- FX intervention (FXI) does not appear to influence interest rate setting: the coefficient of FXI in the real policy rate regressions is not statistically significant in the main specifications.
- Deviations from a more orthodox framework were modest and transitory; Peru and Chile performed similarly on output growth and inflation and their volatility, with Peru slightly lower volatility for both series.
- For the real exchange rate, the real policy rate and FXI are the most relevant instruments in explaining deviations, with RRs possibly playing a role depending on capital flows.

### Regression results — Determinants of the Real Monetary Policy Rate Gap (Table 1)
- Dependent Variable: Real Monetary Policy Rate Gap. Robust Least Squares, period Dec. 2002-Mar. 2019 (most regressions start Sep. 2003).
- Coefficients and standard errors (in parentheses):
  - Inflation Gap: 0.1046*** (0.0335); 0.1043*** (0.0336); 0.1119*** (0.0324); 0.1401*** (0.0306); 0.1138*** (0.0316)
  - FXI Gap: 0.0958 (0.1417); -0.0373 (0.1387); -0.1989 (0.1331); 0.0060 (0.1360)
  - RR Gap: -0.0216*** (0.0055); -0.0103 (0.0066)
  - RR Gap IV: -0.0475*** (0.0076) — instrumenting RR with lagged capital flows gap (three, six, nine month lags)
  - Capital Flow Gap: -0.1549*** (0.0507)
- Goodness of fit and sample:
  - R-square: 0.0481; 0.0503; 0.1153; 0.2013; 0.1523
  - Observations: 195; 187; 187; 187; 187
- Interpretation:
  - Instrumented RR remains a significant explanatory variable for the real policy rate gap and the coefficient increases when instrumented.
  - Including the capital flows gap as a control causes the RR gap to lose significance (likely reflecting endogeneity and high correlation between RR and capital flows gap; correlation reported as 0.58).

### Relationship between RRs, capital flows, and endogeneity
- Reserve requirements are highly correlated with the capital flows gap, and Peru appears to have used RRs to tighten monetary conditions amid capital inflows to avoid triggering additional inflows via policy rate increases.
- To address endogeneity of RRs, RRs are instrumented with lagged values of the capital flows gap; the instrumented RR remains significant with larger coefficient magnitude (see RR Gap IV: -0.0475*** (0.0076)).
- Including the capital flows gap as a regressor renders the RR gap insignificant (Table 1, fifth column), consistent with endogeneity between RR and capital flows gap and their high correlation (0.58).

### Regression results — Determinants of the Real Exchange Rate Gap (Table 2)
- Dependent Variable: RER Gap. Robust Least Squares, period Dec. 2002-Mar. 2019 (most regressions start Sep. 2003).
- Coefficients and standard errors (in parentheses) across selected specifications:
  - TOT Gap: 0.0862 (0.0726); 0.0642 (0.0768); 0.0502 (0.0723); 0.0353 (0.0692); 0.0374 (0.0681)
  - Real Pol. Rate Gap: 2.6304*** (0.7165); 2.7407*** (0.6750); 3.5684*** (0.6609); 4.0637*** (0.6625)
  - FXI Gap: -6.0190*** (1.2809); -5.1560*** (1.2430); -5.4552*** (1.2280)
  - RR Gap: 0.1805*** (0.0513); 0.0843 (0.0597)
  - Capital Flow Gap: 1.3589*** (0.4680)
- R-square and observations:
  - R-square: 0.0069; 0.0670; 0.1605; 0.2053; 0.2378
  - Observations: 195; 187; 187; 187; 187
- Quantitative interpretations (from equation (4) discussion):
  - A one percentage point increase in the policy rate gap leads to an almost 3½ percentage points appreciation of the real exchange rate (equation (4) estimate).
  - Foreign exchange sales equivalent to one month of imports of goods and services (equivalent to almost two percent of GDP) lead to an appreciation of the real exchange rate in the order of over 5 percentage points (based on equation (4) coefficient).
  - A 10 percent increase in reserve requirements leads to a 1.8 percent appreciation of the real exchange rate (equation (4) implication), though higher RRs also lower the policy rate and thus partly offset this effect.

### Credit growth gap and other linkages
- Correlations:
  - Real credit growth gap is highly correlated with the GDP growth gap: correlation of 0.55.
  - Real credit growth gap has weak and negative correlation with the inflation gap: correlation of -0.26.
- Links with policy instruments are relatively weak and period-specific:
  - Pre-GFC: increased RRs when the credit gap was highly positive; RRs subsequently reduced when credit decelerated.
  - Other RR tightening episodes do not consistently align with credit gap developments; RRs appear more closely tied to capital flows.
  - Pre-GFC large positive credit gap occurred during a negative policy rate gap (suggesting easier monetary conditions contributed).
- Overall, credit gaps appear highly correlated with output gaps but only weakly correlated with inflation gaps and policy instrument use.

### Conclusions and implications
- Peru’s multi-instrument inflation targeting framework generated modest and transitory deviations from a more orthodox framework, primarily in:
  - Policy rate setting: RRs appear the most relevant instrument affecting the real policy rate, subject to capital flow developments.
  - Real exchange rate outcomes: real policy rate and FXI are the primary instruments; RRs may also contribute depending on capital flows.
- Magnitudes and examples:
  - 10 percentage points increase in RRs → real interest rate 0.2 percentage points lower than orthodox benchmark.
  - Average monthly FX sale episode ≈ 0.25 months of imports → translates into an appreciation of the real exchange rate in the order of 1.3 percent.
- Research needs:
  - Further work to better understand endogeneity linkages among outcomes and instruments (Appendix II provides initial work).
  - Expand analysis to other countries and explore implications for economic vulnerabilities and welfare.
  - Complement with country-specific studies to relax the identical-coefficients assumption used in the empirical regressions.

### Data and measurement notes
- All variables are measured monthly. Sample period: January 2003-March 2019.
- Gaps are calculated as differences between Peru and Chile variables. For REER and TOT gaps natural logs are taken before differencing; capital flows gap uses USD billions difference.
- Key data sources: Haver, IMF staff estimates, Central Bank of Peru, BCRP, IMF INS database.
- Notes on series construction:
  - Real policy rate: nominal policy rate minus 12-month inflation expectations.
  - RRs for Peru: marginal reserve requirements; charts and calculations use simple average of reserve requirements for domestic and foreign currency.
  - FX Intervention: divided by (12-month rolling) imports of goods and services for normalization in charts.

*Source: IMF staff estimates and analysis from the provided chapter.*

### Annex II. Addressing Endogeneity through VAR Analysis

### Annex II. Addressing Endogeneity through VAR Analysis

### Approach and scope
- Treat all variables as endogenous and estimate the model through a VAR to address endogeneity problems.
- Analysis is restricted to assessing whether results from the main text are present in the VAR estimation; a more comprehensive VAR analysis (including impulse response functions) is left for future work.

### Key VAR estimation findings
- Real policy rate equation (first column):
  - The inflation gap is a significant explanatory variable and has the same sign as in the regressions presented in Table 1.
  - The RR gap also has the same sign as in Table 1, but the significance in this equation "falls slightly short of the standard significance levels."
- Real exchange rate gap equation (fifth column):
  - The FXI gap is highly significant and has the same sign as in Table 2.
  - The real policy rate exhibits similar results as those in Table 2, but the significance level is borderline.
  - The terms of trade gap, which was not significant in Table 2, is significant in this VAR estimation.

### Selected statistics and model diagnostics reported in Table A1
- R-square values by equation (columns 1–8): 0.8252, 0.3937, 0.9664, 0.9048, 0.8929, 0.4538, 0.9615, 0.5736
- Observations: 182 for each equation
- Estimation sample period: Dec. 2002-Mar. 2019. Note: most regressions start in September 2003, when Peru's policy rate becomes available.
- Table reporting conventions: Standard errors in parentheses, t-statistics in brackets.
- Coefficient examples from the VAR (lagged regressors, shown with associated standard errors and t-statistics in the table):
  - IRATEGAP(-1) coefficients across columns include 0.8071 (t [ 18.9788]), 0.0252 (t [ 0.64328]), -0.0770 (t [-0.33610]), 0.0279 (t [ 0.43040]), 0.4680 (t [ 1.60895]), -0.2436 (t [-0.96134]), 0.0508 (t [ 0.28755]), -0.1885 (t [-1.79469]).
  - FXIGAP(-1) coefficients include -0.0982 (t [-1.37301]), 0.5718 (t [ 8.67847]), 1.7498 (t [ 4.54466]), -0.0804 (t [-0.73781]), -1.7758 (t [-3.63046]), 0.5563 (t [ 1.30576]), 0.3903 (t [ 1.31262]), 0.3085 (t [ 1.74685]).
  - RRGAP(-1) coefficients include -0.0050 (t [-1.48227]), -0.0034 (t [-1.09408]), 0.9884 (t [ 54.3697]), -0.0032 (t [-0.62432]), 0.0124 (t [ 0.53716]), -0.0075 (t [-0.37344]), 0.0017 (t [ 0.12240]), 0.0305 (t [ 3.65425]).
  - INFGAP(-1) coefficients include 0.0454 (t [ 2.39025]), 0.0059 (t [ 0.33837]), -0.0243 (t [-0.23766]), 0.9373 (t [ 32.3371]), 0.0469 (t [ 0.36047]), -0.2573 (t [-2.27152]), -0.0491 (t [-0.62163]), -0.0282 (t [-0.60082]).
  - REERGAP(-1) coefficients include 0.0103 (t [ 2.12204]), -0.0023 (t [-0.51527]), 0.0208 (t [ 0.79617]), 0.0082 (t [ 1.10515]), 0.8706 (t [ 26.2172]), 0.0215 (t [ 0.74393]), -0.0148 (t [-0.73514]), 0.0253 (t [ 2.11139]).
  - GDPGGAP(-1) coefficients include 0.0146 (t [ 1.22509]), 0.0032 (t [ 0.29195]), -0.1201 (t [-1.86647]), -0.0089 (t [-0.48851]), 0.0776 (t [ 0.94997]), 0.3763 (t [ 5.28572]), 0.1430 (t [ 2.87686]), -0.0466 (t [-1.57861]).
  - CREDGGAP(-1) coefficients include 0.0051 (t [ 0.98624]), -0.0012 (t [-0.25960]), 0.0727 (t [ 2.60284]), 0.0094 (t [ 1.19185]), -0.0632 (t [-1.78111]), 0.1125 (t [ 3.63986]), 0.9367 (t [ 43.4409]), 0.0221 (t [ 1.72219]).
  - CAPFLWGAP(-1) coefficients include -0.0358 (t [-1.37421]), -0.0163 (t [-0.68133]), 0.0050 (t [ 0.03552]), 0.0632 (t [ 1.59224]), 0.1858 (t [ 1.04418]), 0.0421 (t [ 0.27144]), -0.0966 (t [-0.89273]), 0.5196 (t [ 8.08604]).
  - Constant (C) coefficients include 0.2795 (t [ 3.00666]), 0.1023 (t [ 1.19450]), 0.2103 (t [ 0.42007]), 0.0471 (t [ 0.33224]), -1.3417 (t [-2.10990]), 0.8080 (t [ 1.45871]), -0.1596 (t [-0.41283]), -0.0547 (t [-0.23841]).
  - TOTGAP coefficients include -0.0061 (t [-1.37469]), 0.0008 (t [ 0.19608]), -0.0374 (t [-1.56861]), 0.0024 (t [ 0.35015]), 0.0652 (t [ 2.15258]), -0.0126 (t [-0.47701]), 0.0017 (t [ 0.09214]), -0.0048 (t [-0.43742]).

### Estimation details and notes
- Source of calculations: Author's calculations.
- Notes from table: "Standard errors in parentheses, t-statistics in brackets."
- Estimations are for the period Dec. 2002-Mar. 2019. Most regressions start in September 2003, when Peru's policy rate becomes available.

*Source: Annex II. Addressing Endogeneity through VAR Analysis (Table A1) — Author's calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2020/english/1perea2020002.pdf_