## PROFITABILITY AND DETERMINANTS OF FX INTERVENTION IN BRAZIL ## Source details **Canonical URL:** [PROFITABILITY AND DETERMINANTS OF FX INTERVENTION IN BRAZIL](https://www.imf.org/-/media/files/publications/cr/2019/1braea2019002.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2019/1braea2019002.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2019/1braea2019002.pdf.json) --- ### A. Introduction - Objective: examine profitability of FX swaps issued by the Central Bank of Brazil (BCB) between May 2013 and February 2019 to shed light on the rationale for FX intervention. - Conceptual framework: - If FX swaps are used to smooth temporary excessive fluctuations (deviations from uncovered interest parity, UIP), the central bank should be profitable in expectation (Friedman (1953) argument). - Making money is a by-product (not a goal) of leaning against temporary UIP deviations (Gabaix and Maggiori (2015) theoretical support). - BCB operational context: - Brazil has FX reserves of about 375 billion USD. - Price discovery mainly occurs in the derivative market; BCB intervenes in the FX derivative market by offering swaps that provide insurance against a depreciation of the real. - Rationale for analyzing swaps: - Swaps have explicit maturities allowing precise computation of realized and expected financial returns. - Expected returns are computed using market projections for the exchange rate and the Selic at the time of the swap sale. - Data and approach: - Analysis covers each FX swap issued between May 31, 2013 and February 28, 2019. - Expected returns use the Market Expectations System survey forecasts (end-of-month forecasts for Selic and BRL/USD over 18 months, interpolated to daily when needed). ### B. Average Profitability of FX Swaps - Use and evolution of FX swaps: - Outstanding stock increased to almost 120 billion USD between 2013 and 2015. - After declining to about 20 billion USD, swaps increased again in 2018 to almost 70 billion USD and remained stable thereafter. - Swaps are sold through auctions with maturities varying from a few days to two years (average maturity about 7 months). - Average value of FX swaps sold in a given auction is 200 million USD; some auctions reached a few billion USD. Auction size can be negative (reducing outstanding swaps). - Mechanics of a swap auction (BCB long BRL vs USD): - BCB pledges repayment of 50,000 USD at maturity converted in BRL at the spot exchange rate of the previous day. - Market participants bid a discounted value relative to the 50,000 USD face value; the discount rate is the bid cupom cambial. - Market participants pay the BCB the Selic rate between settlement and maturity. - Realized net revenues: - Cumulative realized losses between May 2013 and September 2018 were about 7.3 billion USD. - Losses peaked at 25.6 billion USD in April 2015. - Realized losses driven by unexpected exchange rate depreciations (not by unforeseen Selic shocks). - Expected returns (at time of intervention): - Cumulative expected gains reached 19.8 billion USD by the end of February 2019. - Average profitability of FX swaps was about 5 percent in annual terms. - Average expected excess return of the BRL over the USD was about 3.5 percent. - Interpretation: swaps were profitable in expectation, consistent with BCB intervening to lean against temporary deviations from UIP (issuing swaps when exchange rate perceived as undervalued). - Decomposition of gap between realized and expected revenues: - Using realized exchange rates with expected Selic tracks realized revenues closely → shocks to Selic do not explain the gap. - Using expected exchange rates with realized Selic replicates expected revenues → unexpected exchange rate movements explain realized losses (notably sharp depreciations in 2015 and 2018). - Robustness to delayed survey updates: - Average expected returns using forecasts on the day of action: 5.6 percent. - Average expected returns using forecasts reported one week later: 3.3 percent. - Average returns remain positive even using forecasts reported four weeks after auction dates. - Conclusion: positive expected profitability is not driven solely by delays in survey updates. ### C. Determinants of FX Intervention — Regression analysis and evidence - Purpose and method: - Test whether FX intervention by the BCB was targeted to temporary excessive movements of the exchange rate by examining whether the extent of swap issuance varied over time with the expected profitability of swaps. - Measure: expected return on individual swaps computed by dividing expected net revenues by the swap face value of 50,000 USD. - Annualized expected returns reported: range between -61 to 107 percent, with an average level of 5 percent. - Primary regression: regress the face value of swaps (in millions of USD) issued in a given auction on their expected return and other financial variables. - Key regression estimates: - Coefficient on "Expected return on swap" in Column (1): 4.82*** (robust standard error 1.2). - Coefficient examples in restricted samples: 17.43*** (2.2); 15.82*** (2.4); 11.58*** (2.3); 11.98*** (2.3); 10.72*** (2.3); 11.45*** (2.4); 11.73*** (2.3). - R-squared and sample sizes: - Column (1) R-squared: 0.02. - Column (2) (excluding pre-announced and rollover swaps) R-squared: 0.19. - Later columns R-squared: 0.23, 0.46, 0.49, 0.47, 0.46, 0.46. - Number of swap auctions by column: 2,898; 731; 712; 712; 712; 712; 712; 706. - Other significant determinants (examples): - Exchange rate (ER) depreciation over last ten days: coefficient example 13.63*** (standard error 4.6). - Cupom cambial - US rate: coefficient example -119.06*** (20.7) and -59.43*** (17.1) in different specifications. - Two-year sovereign spread: coefficient example 48.99*** (14.6) in a specification including sovereign spreads. - Magnitudes (one-standard-deviation shocks): - Expected return shock affects issuance by 179 million USD. - Exchange rate shock affects issuance by 46 million USD. - Cupom cambial shock affects issuance by 70 million USD. - Robustness, sample adjustments, and controls: - Excluding pre-announced and rollover auctions materially improves fit (R-squared increases to 0.19). - Column (4): adding sales during the prior ten days to control serial correlation; significance of swap returns and cupom cambial unchanged. - Column (5): winsorize 1 percent of data with respect to auction size and swap returns; results broadly unchanged. - Columns (6)–(8): adding other financial variables; swap issuances positively correlated with sovereign spreads; no association between 2-year CDS on sovereign bonds and domestic stock market performance. - Statistical notation: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. - Interpretation: - The BCB increased swap sales when the degree of undervaluation was more severe (higher expected profitability of swaps). - Additional shaping factors: movements in the cupom cambial, sovereign spreads, and significant autocorrelation in swap issuance. ### D. Policy implications and operational recommendations - Monitoring and decision rule: - The expected profitability of FX swaps can be computed in real time because it depends on expectations at the time of issuance rather than on future realizations of exchange rates and interest rates over the life of the swap. - Policy recommendation: if the central bank wants to lean against UIP deviations, it could decide on the scale of intervention by openly monitoring the expected returns of swaps. - Balance-sheet concerns: - If FX intervention is profitable in expectation, concerns about central bank balance-sheet exposure may be alleviated. - While unexpected depreciation generates losses on FX swaps, well-intended FX intervention tends to be profitable in expectation, suggesting the argument that swaps are financially sustainable only if matched by large FX reserves might be overstated. ### E. Government bond portfolio, TSA, and liquidity management - Government bond portfolio properties: - The BCB’s government bond portfolio earns interest like any other government paper; by law, the Treasury can only roll over its principal value adjusted by a consumer price index. - IMF (GFSM 2014) treats all government debt held by the central bank as a liability of the government; under the BCB definition, bonds issued to the BCB are not included as debt. - Direct issuance of bonds to the BCB by the Treasury immediately impacts the IMF definition of public debt; the BCB definition is affected only when the FS portfolio is used in OMOs. - TSA dynamics and drivers: - TSA balances increased from 10 to 19 percent of GDP over the past decade, despite an average annual fiscal deficit of 4 percent of GDP. - Key factors: BCB profits from valuation gains (credited semi-annually under the old framework), TSA accrued interest, debt management buffer, and earmarking of TSA sub-accounts. - TSA stood at R$ 1.3 trillion (19 percent of GDP) as of December 2018. - About half of the balance (9.5 percent of GDP) can be used to repay debt. - Repos and reserve requirements: - Stock of reverse repos increased from 7 to 17 percent of GDP over the last decade. - Reverse repos are the only relevant OMO instrument used by the BCB. - Reserve requirements are partially remunerated, have a permanent effect on liquidity, and significant changes are rare. - Drivers of liquidity: - Structural liquidity in Brazil increased from 1.7 percent of GDP to 17 percent of GDP from 2005 to 2018. - FX swaps: a negative net result on FX swaps leads to an increase in liquidity because it represents an outflow from the BCB to the financial system, requiring sterilization. - From 2006 to 2012, international reserves increased by US$310 billion and boosted liquidity, prompting sterilization via reverse repos. - From 2012 onwards, exchange rate impact on liquidity was smaller and mainly driven by FX swaps results. ### F. Central bank profit accounting, fiscal interactions, and legacy of the old framework - Old framework accounting: - Financial flows between BCB and Treasury were separated into two items assessed semi-annually: “FX equalization” and operations not related to FX operations. - Losses were compensated through debt issuances; gains transferred to the TSA. - Magnitude of transfers: - From 2008 to mid 2018, transfers from BCB to Treasury amounted to 14.3 percent of 2018 GDP, with 72 percent corresponding to FX equalization. - Treasury transferred bonds equivalent to 9.7 percent of 2018 GDP to the BCB, primarily to cover FX equalization losses (97 percent). - Impact on fiscal operations and debt management: - Since 2014, refinancing of debt through new debt issuance represented, on average, just 80 percent of maturing debt (principal and interest). - The government accumulated a financial surplus from fiscal operations instead of paying down debt; this helped limit growth of structural liquidity but increased the cash buffer. - Entangled policies: - Exchange rate policies (NIR accumulation, FX swaps) affect the fiscal deficit and public debt trajectory. - High interest rates perpetuate liquidity, triggering more OMOs, which in turn increase public debt. - Investors increased exposure to short-term liquidity management instruments as an alternative to short-term treasury bills. ### G. The 2019 reform: redesign of Treasury–BCB relationship - Reform objectives: - (i) change profit and loss distribution rules, - (ii) enhance monetary policy execution, - (iii) improve public debt management; better align incentives, secure Treasury backing of BCB losses, and allow significant reduction of costly FS portfolio. - Key changes: - Unrealized FX gains will no longer be credited to the TSA; profits from FX holdings will flow into an accounting reserve in the BCB balance sheet to cover future losses. - If the reserve is insufficient in a given year, the remainder will be covered by BCB's net worth until this reaches 1.5 percent of its assets; if still insufficient, the Treasury will automatically issue securities to the BCB. - When FS portfolio reaches a lower bound of 4 percent of total BCB bond portfolio, the Treasury will automatically issue new securities so FS portfolio reaches 5 percent of the total BCB bond portfolio. - Treasury has a legal obligation to cover any remaining BCB losses if reserve account buffers are insufficient. - Expected effects: - Makes the Golden Rule more effective by reducing transfers that indirectly covered primary expenses with BCB FX results. - Stops the ratchet effect of FX volatility on public debt and paves the way to reduce the FS portfolio to the minimum necessary. - Enables cancellation of a portion of the FS with BCB’s net worth, which rose from 0.04 percent of GDP in end-2007 to 1.9 percent of GDP in end-2017. - Remaining challenges: - The reform reduces financial flows between the two institutions, but does not address the high stock of TSA balances, NIR, and OMO. - Over time, reforms are expected to reduce the relative size of the BCB’s balance sheet to GDP; addressing the stock problems could be subject to further research. - The size of the government’s cash buffer (TSA) should be reviewed given its high opportunity cost. ### H. Annex I — Optimal Cash Buffer for the Government: Benefit and Cost framework - Benefit definition (principal ideas): - Benefit arises from using a cash cushion to avoid market issuance when financial conditions are abnormally tight, assuming the rise in interest rates is temporary. - Key symbols: ܫܥܣ∆ is the change in the average cost of issuance during the period of high volatility; ܯ is the average maturity of debt issuance; ݀ is debt coming due during the period of high volatility. - Consolidated public-sector benefit: - The consolidated benefit accounts for the CB sterilizing increased market liquidity via OMOs at approximate cost of the policy rate (ܴܲ). - Cost considerations: - Opportunity cost for central government holding buffer b is the Average Cost of its debt (AC); CB should remunerate this buffer according to the Market Rate (MR). - Consolidated underlying cost can be negative if AC is lower than the policy rate. - Net benefit maximization and outcomes: - Net benefit NB of holding a buffer is defined and analyzed as a linear maximization in b. - Case 1: Spread (AC - PR) positive → objective function negatively sloped; optimum cushion may be b = d or b = 0 depending on benefit vs cost factor. - Case 2: Spread (AC - PR) negative → objective function positively sloped; best constrained response is to hold a buffer equal to d. - Practical note: - Framework builds on simplifying assumptions (full refinancing of maturing debt during volatility; immediate TSA recomposition) but provides a useful approximation for preliminary analysis of precautionary cash cushions. *Source: IMF staff note prepared by Damiano Sandri (with research assistance from Daniel Cunha), covering FX swaps issued by the Central Bank of Brazil between May 2013 and February 2019.* ### References _______________________________________________________________________________ 12 ### PROFITABILITY AND DETERMINANTS OF FX INTERVENTION IN BRAZIL ### A. Introduction - Objective: examine profitability of FX swaps issued by the Central Bank of Brazil (BCB) between May 2013 and February 2019 to shed light on the rationale for FX intervention. - Conceptual framework: - If FX swaps are used to smooth temporary excessive fluctuations (deviations from uncovered interest parity, UIP), the central bank should be profitable in expectation (Friedman (1953) argument). - Making money is a by-product (not a goal) of leaning against temporary UIP deviations (Gabaix and Maggiori (2015) theoretical support). - BCB operational context: - Brazil has FX reserves of about 375 billion USD. - Price discovery mainly occurs in the derivative market; BCB intervenes in the FX derivative market by offering swaps that provide insurance against a depreciation of the real. - Rationale for analyzing swaps: - Swaps have explicit maturities allowing precise computation of realized and expected financial returns. - Expected returns are computed using market projections for the exchange rate and the Selic at the time of the swap sale. - Data and approach: - Analysis covers each FX swap issued between May 31, 2013 and February 28, 2019. - Expected returns use the Market Expectations System survey forecasts (end-of-month forecasts for Selic and BRL/USD over 18 months, interpolated to daily when needed). ### B. Average Profitability of FX Swaps - Use and evolution of FX swaps: - Outstanding stock increased to almost 120 billion USD between 2013 and 2015. - After declining to about 20 billion USD, swaps increased again in 2018 to almost 70 billion USD and remained stable thereafter. - Swaps are sold through auctions with maturities varying from a few days to two years (average maturity about 7 months). - Average value of FX swaps sold in a given auction is 200 million USD; some auctions reached a few billion USD. Auction size can be negative (reducing outstanding swaps). - Mechanics of a swap auction (BCB long BRL vs USD): - BCB pledges repayment of 50,000 USD at maturity converted in BRL at the spot exchange rate of the previous day. - Market participants bid a discounted value relative to the 50,000 USD face value; the discount rate is the bid cupom cambial. - Market participants pay the BCB the Selic rate between settlement and maturity. - Realized net revenues formula (as presented): - Net revenues = 50,000 * [expression using exchange rates, cumulative Selic over working days, and the bid cupom cambial] (Equation (1) in the source). - Net revenues inversely related to depreciation of the real and positively related to the differential between the Selic rate and the bid cupom cambial. - Empirical findings on realized returns: - Cumulative realized losses between May 2013 and September 2018 were about 7.3 billion USD. - Losses peaked at 25.6 billion USD in April 2015. - Realized losses driven by unexpected exchange rate depreciations (not by unforeseen Selic shocks). - Empirical findings on expected returns (at time of intervention): - Cumulative expected gains reached 19.8 billion USD by the end of February 2019. - Average profitability of FX swaps was about 5 percent in annual terms. - Average expected excess return of the BRL over the USD was about 3.5 percent. - Interpretation: swaps were profitable in expectation, consistent with BCB intervening to lean against temporary deviations from UIP (issuing swaps when exchange rate perceived as undervalued). - Decomposition of gap between realized and expected revenues: - Using realized exchange rates with expected Selic tracks realized revenues closely → shocks to Selic do not explain the gap. - Using expected exchange rates with realized Selic replicates expected revenues → unexpected exchange rate movements explain realized losses (notably sharp depreciations in 2015 and 2018). - Robustness to delayed survey updates: - Concern: Market Expectations System forecasts may be updated infrequently; delayed updates could bias expected profitability measures. - Results using delayed forecasts: - Average expected returns using forecasts on the day of action: 5.6 percent. - Average expected returns using forecasts reported one week later: 3.3 percent. - Average returns remain positive even using forecasts reported four weeks after auction dates. - Conclusion: positive expected profitability is not driven solely by delays in survey updates. ### C. Determinants of FX Intervention (introductory points from source) - The analysis also investigates whether the scale of intervention varied with swap profitability: - A positive correlation between number/size of swaps sold and their expected profitability would suggest the BCB intervened more when the exchange rate was perceived as more undervalued. - Other potential determinants to be examined include: - Movements in the exchange rate. - Movements in the cupom cambial (onshore USD interest rate). *Source: IMF staff note prepared by Damiano Sandri (with research assistance from Daniel Cunha), covering FX swaps issued by the Central Bank of Brazil between May 2013 and February 2019.* ### 18. To further understand the rationale for FX intervention by the BCB, we perform a ### 1braea2019002 - 18. To further understand the rationale for FX intervention by the BCB, we perform a ### Regression analysis: purpose and method - Objective: test whether FX intervention by the BCB was targeted to temporary excessive movements of the exchange rate by examining whether the extent of swap issuance varied over time with the expected profitability of swaps. - Measure: expected return on individual swaps computed by dividing expected net revenues by the swap face value of 50,000 USD. - Annualized expected returns reported: range between -61 to 107 percent, with an average level of 5 percent. - Primary regression: regress the face value of swaps (in millions of USD) issued in a given auction on their expected return and other financial variables. ### Key empirical findings - Positive correlation between swap issuance and expected return: - Coefficient on "Expected return on swap" in Column (1): 4.82*** (robust standard error 1.2). - Coefficient increases in restricted samples shown (examples): 17.43*** (2.2), 15.82*** (2.4), 11.58*** (2.3), 11.98*** (2.3), 10.72*** (2.3), 11.45*** (2.4), 11.73*** (2.3). - R-squared values: - Column (1) R-squared: 0.02. - Column (2) (excluding pre-announced and rollover swaps) R-squared: 0.19. - Later columns R-squared reported: 0.23, 0.46, 0.49, 0.47, 0.46, 0.46. - Sample sizes (Number of swap auctions): 2,898; 731; 712; 712; 712; 712; 712; 706 (by column). - Other significant relations: - Exchange rate (ER) depreciation over last ten days: coefficient example 13.63*** (standard error 4.6) in one specification. - Cupom cambial - US rate: coefficient example -119.06*** (20.7) in one specification and -59.43*** (17.1) in another. - Two-year sovereign spread: coefficient example 48.99*** (14.6) in a specification including sovereign spreads. - Magnitudes: a one-standard-deviation shock to - expected return affects issuance by 179 million USD, - exchange rate affects issuance by 46 million USD, - cupom cambial affects issuance by 70 million USD. ### Robustness, sample adjustments, and controls - Pre-announced and rollover auctions: - Excluding auctions that were pre-announced or used to rollover maturing swaps materially improves fit (R-squared increases to 0.19). - Example of pre-announcement: on August 22, 2013 the BCB announced “swap auctions will take place every Monday, Tuesday, Wednesday and Thursday, when 500 USD million will be offered for each day” at least until December 2013. - Definition: “rollover” swaps are those for which on settlement date the stock of outstanding swaps differs less than 0.5 billion USD from the previous day. - Controls added and adjustments: - Column (4): controls for serial correlation by adding sales during the prior ten days (lags positive, absorb effects of exchange rate movements; significance of swap returns and cupom cambial unchanged). - Column (5): winsorize 1 percent of data with respect to auction size and swap returns; results broadly unchanged. - Columns (6)–(8): adding other financial variables; swap issuances positively correlated with sovereign spreads (difference between two-year sovereign yields and the Selic); no association between 2-year CDS on sovereign bonds and domestic stock market performance. - Statistical notation: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. ### Interpretation and conclusions - Primary conclusion: the BCB used FX intervention to lean against temporary excessive movements of the exchange rate. - On average, FX swaps were profitable in expectation, implying issuance when the exchange rate was temporarily undervalued and vice versa. - Regression evidence: the BCB increased swap sales when the degree of undervaluation was more severe (higher expected profitability of swaps). - Additional shaping factors: movements in the cupom cambial, sovereign spreads, and significant autocorrelation in swap issuance. ### Policy implications and recommendations - Monitoring: - The expected profitability of FX swaps can be computed in real time because it depends on expectations at the time of issuance rather than on future realizations of exchange rates and interest rates over the life of the swap. - Policy recommendation: if the central bank wants to lean against UIP deviations, it could decide on the scale of intervention by openly monitoring the expected returns of swaps. - Balance-sheet concerns: - If FX intervention is profitable in expectation, concerns about central bank balance-sheet exposure may be alleviated. - While unexpected depreciation generates losses on FX swaps, well-intended FX intervention tends to be profitable in expectation, suggesting the argument that swaps are financially sustainable only if matched by large FX reserves might be overstated. ### Contextual notes on related balance-sheet and liquidity-management issues (selected excerpts) - FX reserves and valuation: - FX reserves increased from around USD 50 billion until 2005 to USD 374 billion by 2018; at market value FX reserves stood at 20 percent of GDP end-2018. - Under the old framework (until July 2019) valuation gains were credited semi-annually to the Treasury Single Account (TSA); valuation losses were compensated by a bond issuance from the Treasury to the BCB, producing a ratchet effect on the BCB’s balance sheet and TSA balances. - Balance-sheet composition and evolution: - The BCB’s balance sheet nearly doubled over the past decade; expansion driven by FX securities, government bonds, TSA balances, and repos. - Government bonds held by the central bank rose from 13 to 26 percent of GDP in a decade. - Three channels for BCB holdings of treasury securities: compensation for financial losses (including FX valuation losses), automatic rollover of maturing bonds (principal + CPI), and maintenance of a minimum Free Securities portfolio for monetary policy operations. *International Monetary Fund — Brazil chapter excerpts* ### 11. The government bond portfolio has the following important properties: ### 11. The government bond portfolio has the following important properties: ### Government bond portfolio: key properties - Interest on government bonds: - The BCB’s government bond portfolio earns interest like any other government paper. - By law, the Treasury can only roll over its principal value adjusted by a consumer price index. This leaves a portion—the real interest rate earned on government bonds—that is paid to the BCB in cash. - Free Securities (FS) and debt perimeter: - The FS portfolio underpins the difference between the IMF’s and the BCB’s definition of public debt. - IMF (GFSM 2014) treats all government debt held by the central bank as a liability of the government. - Under the common Brazil (BCB) definition, bonds issued to the BCB are not included as debt, but the stock of BCB-held government bonds in the hands of the public through OMOs is. - Consequence: direct issuance of bonds to the BCB by the Treasury immediately impacts the IMF definition of public debt; the BCB definition is affected only when the FS portfolio is used in OMOs. ### Liabilities and the Treasury Single Account (TSA) - TSA dynamics drivers: - Fiscal revenues and expenditures, financial revenues, and public debt operations drive TSA dynamics. - The balance on the TSA increased from 10 to 19 percent of GDP over the past decade, despite an overall expansionary fiscal policy stance and an average annual fiscal deficit of 4 percent of GDP. - Key factors behind TSA increase: - BCB profits from valuation gains: - Under the old framework, FX valuation gains from currency depreciation credited the TSA semi-annually. When the currency appreciated, there was no drawdown in the TSA because the Treasury capitalizes the BCB through public bonds. - TSA accrued interest: - TSA balances earn interest equivalent to the average cost of public bonds held by the BCB, close to the average cost of domestic public debt; this corresponds roughly to the real interest payment that the government pays the BCB for its government debt holdings. - Debt management buffer: - The National Treasury earmarks several financial revenues (interest on TSA balances and debt repayments from loans with subnational governments and BNDES) to build and maintain a cash cushion to redeem government debt. - Earmarking of TSA subaccounts: - Many TSA sub-accounts are legally earmarked for specific purposes; reallocating TSA balances is complex and often requires parliamentary approval. ### Repos (compromissadas) and reserve requirements - Repos: - Reverse repos are the only relevant OMO instrument used by the BCB. - Stock of reverse repos increased from 7 to 17 percent of GDP over the last decade. - Reverse repos: BCB (borrower) sells a government bond (collateral) to another agent (lender) with a repurchase commitment—effectively a collateralized loan that reduces liquidity. - Reserve requirements: - Bank deposits are traditional reserve requirements; like repos, they have a contractionary impact on liquidity. - An increase in reserve requirements triggers an outflow from banks to the BCB, reducing liquidity. - These compulsory reserve requirements are partially remunerated, have a permanent effect on liquidity, and significant changes are rare. ### Drivers of liquidity (accounting and economic perspectives) - Structural change: - Structural liquidity in Brazil increased from 1.7 percent of GDP to 17 percent of GDP from 2005 to 2018. - Accounting drivers (BCB, 2018): - FX purchase: purchase of FX for BRL increases liquidity, which is immediately sterilized by reverse repos. - Interest on OMOs: repos initially reduce the monetary base (liquidity reduction) and increase repo operations (liability); on repo closing day liquidity increases via repayment and interest, typically prompting rollover into a new repo with a higher nominal value due to interest. - Debt operations: net debt issuance by the Treasury to the market reduces liquidity as TSA balances increase. - Fiscal flows (non-debt): tax collection, expenditures, and financial revenues (including BNDES loan repayments) affect liquidity; expansionary fiscal policy increases liquidity, triggering OMOs. - FX swaps: involve payments between the BCB and the market and impact liquidity; FX swaps are drawn based on the BCB’s short position on USD and coupon cambial with a corresponding long position in SELIC. A negative net result on FX swaps leads to an increase in liquidity because it represents an outflow from the BCB to the financial system, requiring sterilization. - Reserve requirements: changes affect the monetary base and liquidity (requirements increased in 2010 to curb credit growth, later partially reversed in 2012). - Economic drivers summarized: - Fiscal policy: - From 2007 to 2014, fiscal policy had on average an expansionary impact on liquidity as primary surpluses and other fiscal inflows were offset by average debt rollover ratios below 100 percent. - From 2015 onwards, fiscal policy impact on liquidity turned contractionary because liquidity-enhancing primary deficits were offset by repayments of loans from BNDES and average debt rollover ratio above 100 percent. - For much of the last decade, the liquidity impact of fiscal policy often had the opposite direction to fiscal policy. - Paradox: a debt redemption is immediately offset by an increase in reverse repos, leaving overall debt level unaffected but changing composition from debt held directly by the public to debt held via compromissadas. - Monetary policy: - Positive variations in base money cause, ceteris paribus, a contraction of the same magnitude in liquidity as banks exchange reverse repos for some component of the monetary base. - Example: sharp increase in reserve requirements in 2010 caused a sizeable contraction in liquidity. - Repos’ interest payments are key drivers of structural liquidity and are linked to the policy rate. - Exchange rate policy: - From 2006 to 2012, international reserves increased by US$310 billion and boosted liquidity, prompting sterilization via reverse repos. - From 2012 onwards, exchange rate impact on liquidity was smaller and mainly driven by FX swaps results. ### Institutional framework: benefits and unforeseen consequences - Institutional setup: - Macroeconomic tripod: floating exchange rate, inflation targeting, and fiscal primary surpluses. - Institutional safeguards: TSA moved to BCB in the 1980s; central bank barred from financing the Treasury under the 1988 Constitution; Fiscal Responsibility Law (2001) prohibited public financial institutions from financing their controlling entities. - Evolution and accumulated balance-sheet positions: - Institutional setup remained the same until mid-2019, with very large balance sheet positions accumulating through interactions of institutional rules and evolving domestic/international conditions. - The document assesses potential unforeseen consequences on monetary policy execution, public debt trajectory/profile, and fiscal operations. ### Monetary policy execution, liquidity, and interest rates - Implementation: - BCB maintains the market interest rate (SELIC over) close to the monetary policy rate (SELIC target), targeting a difference of 10 bps. - To achieve convergence BCB conducts reverse repos to control liquidity surplus. - Liquidity level directly relates to SELIC over pricing; SELIC over affects the overnight inter-banking interest rate (DI), which benchmarks several fixed income products. - Compromissadas and SELIC: - BCB uses different repo maturities: 1 day (overnight), 45 days (between COPOM meetings), 3 months, and 6 months. - Overnight operations directly affect SELIC pricing; longer maturities matter because their volume is set by BCB—if not rolled over, market players lend via overnight repos, putting downward pressure on SELIC over. - Objective: minimize difference between SELIC over and SELIC target; deviations lead to social loss from deflation or inflation. - Empirical observation: - High structural liquidity does not seem to have had an adverse effect on monetary policy execution; the SELIC over rate has not been affected by the increasing structural liquidity. - BCB has effectively controlled the price (SELIC spread) of daily reverse repos as the spread remained constant through periods of large OMO accumulation. ### Central bank profit accounting and FX volatility - Old framework accounting: - Financial flows between BCB and Treasury were separated by law into two items; losses in one could not be offset by gains in the other. - First item: “FX equalization” (equalização cambial) — net profit/loss of international reserves holdings and FX swaps, including exchange rate effect on international reserves, gains/losses from FX swaps in the domestic market, and the opportunity cost of carrying reserves. - Second item: operations not related to FX operations (gains/losses on government securities portfolio, monetary base, and repo operations). - Gains/losses in each item assessed semi-annually; losses compensated through debt issuances, gains transferred to the TSA. - Magnitude of transfers and transfers-to-BCB: - From 2008 to mid 2018, transfers from BCB to Treasury amounted to 14.3 percent of 2018 GDP, with 72 percent corresponding to FX equalization. - In the same period, the Treasury transferred bonds equivalent to 9.7 percent of 2018 GDP to the BCB, primarily to cover FX equalization losses (97 percent). ### Impact on fiscal operations - Interaction with debt refinancing: - BCB’s large balance sheet positions affected fiscal operations and fiscal transparency (2017 Fiscal Transparency Evaluation recommended strengthening accounting/statistical treatment). - Since 2014, refinancing of debt (both held by the market and the BCB) through new debt issuance represented, on average, just 80 percent of maturing debt (principal and interest). - The remainder was financed by various sources, including extrabudgetary financial revenue, which included the BCB’s valuation gains. - The government accumulated a financial surplus from fiscal operations instead of paying down debt; this helped limit growth of structural liquidity (because net treasury debt repayments would have increased liquidity and required compensating OMO) but also increased the cash buffer. *Source: 1braea2019002 - 11. The government bond portfolio has the following important properties (PDF chapter/section).* ### 25. Interest revenue on TSA balances has played an increasingly important role. TSA ### 1braea2019002 - 25. Interest revenue on TSA balances has played an increasingly important role. TSA ### Interest revenue on TSA balances and fiscal operations - TSA balances, including those originating from valuation gains, generate sizeable interest revenue that is not earmarked and has become a relevant factor in fiscal operations. - Until 2014, the treasury used TSA interest revenue to repay debt, helping the government to comply with the golden rule. - From 2015 onwards, interest revenue from high TSA balances is also being used to finance primary expenditures. ### Golden rule, BCB profits, and fiscal space - Constitutional golden rule: government cannot use more proceeds from debt operations to finance the budget than capital expenditures in any given year; capital expenditures include investments, financial investments, and public debt amortization. - BCB profits and valuation gains credited to the TSA have increased the space provided by the golden rule by reducing rollover needs. - Disregarding the contribution of BCB profits to the government’s financial revenue, the golden rule would have become a binding constraint, at least since 2015. ### Impact of BCB foreign exchange (FX) operations on debt measures and the golden rule - Differences between BCB and IMF debt concepts arise from the BCB’s FS portfolio being affected by valuation losses of international reserves. - As long as the FS portfolio is large enough to back issuance of reverse repos to mop up monetization caused by reserve purchases or FX swap losses, the IMF gross debt measure will not be affected by these operations. - Net debt is affected by net changes in non-financial public sector assets, making it less sensitive to valuation and FX swaps operations. - The new framework corrects distortions of international reserves valuation gains/losses on public debt and their side effect on the golden rule. - Table 1 summary (operation impacts preserved as in source): - Valuation gains used to reduce BCB-held or market debt: BCB Gross Debt unchanged; IMF Gross Debt decrease; Net Debt decrease; Golden Rule Fiscal Space increase. (Old framework) - Valuation losses compensated by bonds issued to BCB: BCB Gross Debt unchanged; IMF Gross Debt increase; Net Debt increase; Golden Rule Fiscal Space unchanged. (Old framework) - Under the new framework, valuation gains/losses transferred/compensated to reserve account leave BCB Gross Debt, IMF Gross Debt, Net Debt, and Golden Rule Fiscal Space unchanged. - Acquisition of international reserves: BCB Gross Debt increase; IMF Gross Debt unchanged; Net Debt unchanged; Golden Rule Fiscal Space unchanged. - Sale of international reserves: BCB Gross Debt decrease; IMF Gross Debt unchanged; Net Debt unchanged; Golden Rule Fiscal Space unchanged. - FX swaps gains: BCB Gross Debt decrease; IMF Gross Debt unchanged; Net Debt decrease; Golden Rule Fiscal Space unchanged. - FX swaps losses: BCB Gross Debt increase; IMF Gross Debt unchanged; Net Debt increase; Golden Rule Fiscal Space unchanged. - (Assumes FS portfolio > 0 in all scenarios.) ### High TSA balances, precautionary cash buffer, and opportunity cost - Debt rollover risks motivate holding a cash cushion to: - mitigate the need to tap markets when issuance costs rise, and - temporarily enable payment of debt service in extreme cases of excessively high rates. - Opportunity cost: government cash buffers have an opportunity cost reflected in the cost of debt; debt falling due during high volatility helps determine an upper-bound for the optimal precautionary cash buffer. - Brazil specifics: - TSA stood at R$ 1.3 trillion (19 percent of GDP) as of December 2018. - About half of the balance (9.5 percent of GDP) can be used to repay debt (National Treasury Secretariat, 2019b). - Estimates indicate the TSA’s debt cushion could finance market-held debt coming due in the next 12 months and beyond, well above the 3 months level established as a lower bound by the National Treasury. - Using the non-debt-cushion part of the TSA to service debt is possible but requires legal and accounting changes. - Other considerations: reducing TSA via debt redemption could trigger compensating liquidity operations, shifting debt composition toward short-term debt; risk aversion of debt managers and market perceptions also influence optimal TSA size. ### Compromissadas, OMOs, and implications for debt sustainability - Compromissadas (reverse repos) offer investment opportunities and can be regarded as sovereign debt from an investor perspective. - Open market operations (OMOs) effectively transform longer-dated debt instruments into short-term collateral debt operations, shortening the effective debt maturity profile significantly. - Including OMOs in debt sustainability analysis is important because: - longer dated bonds issued to the BCB are used as collateral in overnight OMOs, - the OMO investor base differs from traditional treasury bond investors, - extreme risk tail events could lead investors to withdraw from OMOs, affecting liquidity, interest rates, and sovereign refinancing decisions. - Brazil: average debt maturity falls materially when reverse repos are included (chart in source). ### Institutional incentives, international experience, and liquidity management - Under the old framework, incentives of the Central Bank (BCB) and the Treasury were not fully aligned: - Bond issuances to the BCB are costly for the Treasury (interest payments) and do not generate cashflow. - The BCB had incentive to hold a large FS portfolio to ensure firepower to manage liquidity. - Equilibrium: Treasury finances debt service of FS portfolio, increasing its rollover needs; BCB monetary policy subject to non-legally enforceable Treasury action. - Aligning incentives could improve equilibrium: e.g., setting BCB holdings of government debt to match necessary reverse repos would eliminate FS portfolio, lower costs, and increase efficiency. - International evidence (selected structural liquidity and features): - Brazil structural liquidity 15.8 percent of GDP (reverse repos; CB uses central government bonds for liquidity management: yes; CB can draw on exogenous resources to cover losses: Yes). - Other countries show diversity in instruments and whether CB uses central government bonds or can draw on exogenous resources; many central banks accommodate losses with buffers and/or future profits. - Optimal choice of instruments depends on institutional setup, market development, and monetary policy transmission; use of treasury bills may be first best in certain contexts. - Commitment of the fiscal authority to back central bank losses is critical in a structural liquidity surplus environment because OMOs to drain structural liquidity are costly and could jeopardize central bank solvency over time. ### The 2019 reform: redesign of Treasury–BCB relationship - Reform objectives: (i) change profit and loss distribution rules, (ii) enhance monetary policy execution, (iii) improve public debt management; better align incentives, secure Treasury backing of BCB losses, and allow significant reduction of costly FS portfolio. - Key changes: - Unrealized FX gains will no longer be credited to the TSA; profits from FX holdings will flow into an accounting reserve in the BCB balance sheet to cover future losses. - If the reserve is insufficient in a given year, the remainder will be covered by BCB's net worth until this reaches 1.5 percent of its assets; if still insufficient, the Treasury will automatically issue securities to the BCB. - When FS portfolio reaches a lower bound of 4 percent of total BCB bond portfolio, the Treasury will automatically issue new securities so FS portfolio reaches 5 percent of the total BCB bond portfolio. - Treasury has a legal obligation to cover any remaining BCB losses if reserve account buffers are insufficient, strengthening monetary policy execution and BCB solvency. - Expected effects: - Makes the Golden Rule more effective by reducing transfers that indirectly covered primary expenses with BCB FX results. - Stops the ratchet effect of FX volatility on public debt and paves the way to reduce the FS portfolio to the minimum necessary. - Enables cancellation of a portion of the FS with BCB’s net worth, which rose from 0.04 percent of GDP in end-2007 to 1.9 percent of GDP in end-2017. ### Conclusion: drivers and legacy of the old framework - Structural liquidity surplus growth mainly explained by four factors: - (i) the effect of FX reserve accumulation, - (ii) an expansionary fiscal policy (in particular from 2015 onwards), - (iii) debt rollover rate less than 100 percent, - (iv) compounding effect of high interest payments on the existing stock of reverse repos. - Under the old framework: - The TSAs balance was regularly boosted with unrealized FX valuations, reducing Treasury debt rollover needs and helping compliance with the Golden Rule. - BCB profits and interest on TSA balances were used to refinance debt and finance primary expenditure. *Source: IMF staff analysis in the provided chapter content.* ### 46. The setup has entangled monetary, exchange rate, and fiscal policies. Exchange rate ### 1braea2019002 - 46. The setup has entangled monetary, exchange rate, and fiscal policies. Exchange rate ### Entangled policy setup and market implications - Exchange rate policies (NIR accumulation, FX swaps) affect the fiscal deficit and public debt trajectory. - High interest rates perpetuate liquidity, triggering more OMOs, which in turn increase public debt. - As the stock of OMOs has increased and debt rollover rates have fallen, investors have increased their exposure to short-term liquidity management instruments as an alternative to short term treasury bills. ### 2019 reform, TSA balances, and balance-sheet imbalances - The 2019 reform will reduce financial flows between the two institutions, but does not address the high stock of TSA balances, NIR, and OMO. - Holding unrealized FX valuation gains within the BCB’s balance sheet reduces flows between the Treasury and the BCB. - Over time, the reforms are expected to reduce the relative size of the BCB’s balance sheet to GDP; addressing the stock problems could be subject to further research. - The size of the government’s cash buffer (TSA) should be reviewed: - Brazil’s TSA stands out in international comparison due to the institutional framework and earmarking of TSA sub-accounts. - Holding a large cash buffer has a high opportunity cost. - Within a new operational framework, a portion of the TSA could be used to repay the free securities FS portfolio, improving public debt dynamics and closing the gap between the BCB’s and IMF’s debt perimeter. ### Treasury role and fiscal policy to curb structural liquidity surplus - The Treasury could help control structural liquidity by: - Speeding up fiscal consolidation. - Envisaging higher rollover rates of its securities. - Blommestein and Turner (2012) recommend moving public debt managers from a micro portfolio mandate toward one incorporating macroeconomic considerations to better coordinate public debt operations with monetary policy. - Consideration could be given to mandating that Treasury operations have at least a neutral impact on liquidity in normal times. ### Central bank independence and liquidity management toolkit - Full central bank independence and expansion of the liquidity management toolkit are key milestones. - The new institutional setup ensures the BCB’s solvency and provides a legal basis for the Treasury to issue bonds to back monetary policy execution. - It remains desirable to grant the central bank full de jure independence and to expand its toolkit to include additional instruments to manage liquidity. ### Annex I — Optimal Cash Buffer for the Government: Benefit and Cost framework - Benefit definition (equation (1) principal idea): - Benefit arises from using a cash cushion to avoid market issuance when financial conditions are abnormally tight, assuming the rise in interest rates is temporary. - Key symbols: ܫܥܣ∆ is the change in the average cost of issuance during the period of high volatility; ܯ is the average maturity of debt issuance; ݀ is debt coming due during the period of high volatility. - Economic scenario: Treasury ceases refinancing volume d during volatility, avoiding additional issuance cost ܫܥܣ∆ over a period equivalent to average issuance maturity. - Consolidated public-sector benefit (equation (2) principal idea): - The consolidated benefit accounts for the CB sterilizing increased market liquidity via OMOs at approximate cost of the policy rate (ܴܲ). - Additional symbols: ′ܯ is the duration of the period of high volatility; ܴܲ is the policy rate. - Cost considerations: - Opportunity cost for central government holding buffer b is the Average Cost of its debt (AC); CB should remunerate this buffer according to the Market Rate (MR). - Treasury cost representation (equation (3)) uses AC, MR, and b. - CB effective cost representation (equation (4)) uses MR, ܴܯ, ܴܲ, and b, recognizing the CB’s implicit return from not having to wipe out liquidity at cost PR. - Consolidated underlying cost (equation (5)) combines Treasury and CB costs; the cost can vary and can be negative if AC is lower than the policy rate. - Net benefit maximization (equation (6)): - Net benefit NB of holding a buffer is defined and can be analyzed as a linear maximization in control variable b, with other parameters exogenous. - The expression highlights a benefit factor and a cost factor and is sufficient to study the sign (inclination) of signals to estimate the optimal buffer. ### Optimal buffer outcomes — three cases from the maximization - Case 1: Spread (AC - PR) is positive - Objective function negatively sloped; initial response: optimum cushion equals zero. - Local analysis (neighborhood where ,ܾ≅݀) yields two sub-outcomes: - If benefit factor > cost factor → optimum buffer equals the volume that would be refinanced during the period of volatility (b = d). Interpretation: insurance fair value equals expected loss. - If benefit factor < cost factor → optimum buffer equals zero; b does not act as fair insurance. - Case 2: Spread (AC - PR) is negative - Objective function positively sloped; without constraints b would go to infinity. - Recognizing temporary nature (AC typically lags PR) and risk aversion of debt manager, impose constraint ݀൑ܾ; best response is to hold a buffer equal to d. - Practical note: equation (2) and the framework build on simplifying assumptions (full refinancing of maturing debt during volatility; immediate TSA recomposition) but serve as a good approximation for preliminary analysis of precautionary cash cushions. *Source: Excerpt from IMF content unit 1braea2019002 (Annex I and related sections).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2019/1braea2019002.pdf_