## wpiea2021264-print-pdf ## Source details **Canonical URL:** [wpiea2021264-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf.json) --- ### I. Scope, definitions, and methodology - Scope: - Analyze vulnerabilities in commercial real estate (CRE) markets across 23 advanced economies and 7 emerging market economies by assessing misalignments between observed CRE prices and fair-value implied by economic fundamentals. - CRE definition: - Property owned for the primary purpose of benefitting from investment returns (includes the multifamily segment), distinct from owner-occupied and noninvestment leased real estate. - Core methodology: - Model fair value as a function of expected income of the commercial property and the return of holding the property (present-value relationship of Campbell and Shiller (1989)). - Expected return specification includes required risk compensation for exposure to the overall property stock market; preferred specification augments risk model with macro factors: output gap, CPI-based inflation, and broad money-to-output ratio. - Extended model for a subset of economies includes vacancy rates for scenario analysis of sustained CRE demand shocks. - Objectives: - Quantify CRE price misalignment (observed price versus fundamentals-implied fair value). - Assess how misalignment forecasts downside risks to GDP growth and financial stability. - Evaluate effectiveness of CRE-related macroprudential policy measures in reducing downside risks to CRE prices. ### II. Key empirical findings and market facts - Market size and exposures: - As of end-2019, the commercial real estate sector had total assets of about 20 percent of GDP, on average, across the sample, up from 17 percent a decade ago. - CRE sector size reached as high as 50 percent or more of GDP in Singapore, Sweden, and Switzerland. - In the United States and some European economies (Estonia and Poland), direct lending related to commercial real estate constituted more than 50 percent of total bank lending to nonfinancial corporations in 2019. - Pre-pandemic dynamics and returns: - Median CRE price across economies steadily increased in the run-up to the COVID-19 pandemic; in Sweden and the United States, real CRE prices almost doubled between 2009 and 2019. - Nominal annual capital appreciation for office buildings and multifamily dwellings averaged about 3 percent globally in the pre-pandemic period. - Pandemic-period disruptions (2020): - Global real estate transaction volume fell by 39 percent in 2020, to its lowest level since 2012. - Same-store net operating income (NOI) changes (first half of 2020, six-month same-store): - Retail assets: declined by 21.4 percent. - Hotel assets: declined by 40 percent. - Industrial property NOI: grew 1.4 percent. - Office sector NOI growth: remained flat for the six months to June 2020. - Residential property: recorded flat net income growth for the six months to June 2020. - Delinquencies and servicing: - Special servicing rates in September 2020: lodging 26.04, retail 18.32; overall reading reached a post-GFC crisis high of 10.48 percent. - Vacancy rates: - U.S. office vacancy: about 9 percent at end-2019, increased to 17 percent at end-2020 (Cushman and Wakefield, 2021). - Model associations and valuation dynamics: - CRE prices associated with market risk premiums, financing price and nonprice terms, and unconventional monetary policy (broad money-to-output). - During the peak of COVID-19, model indicates signs of overvaluation as actual prices did not fall as much as model-implied fair values. ### III. Scenario analysis: permanent increase in vacancy rates - Shock considered: - Permanent increase in the vacancy rate used to proxy a sustained decline in CRE-specific demand (e.g., persistent structural shifts such as e-commerce and teleworking). - Calibration(s) reported in source: - Shock size calibrated so that vacancy rate would gradually increase on average by 5 percentage points in the next 10 years. - Alternative wording: simulation assumes a 5 percent sustained increase in vacancy rates. - Key quantitative result: - A permanent increase in the vacancy rate of 5 percentage points would result in a median drop in fair values of about 15 percent after five years (median across the subset with vacancy data: Australia, Canada, Denmark, Spain, Italy, Portugal, Sweden, the United Kingdom, and the United States). - Uncertainty: - Interquartile ranges are substantial in impulse-response plots; outcomes contain notable cross-country heterogeneity. ### IV. Macroeconomic and financial-stability effects of CRE price misalignment - Misalignment as predictor of tail risks: - A 50-basis-point increase in the misalignment measure (corresponding to one standard deviation) could raise downside risks to GDP growth (cumulatively) by: - 1.4 percentage points over four quarters. - 2.5 percentage points over 12 quarters. - CRE-price downside risk (CaR) findings: - A one-standard-deviation higher price misalignment (corresponding to 10 basis points in advanced economies under the CaR misalignment proxy) is associated with: - a (cumulative) 2.5 percentage point increase in downside risks to CRE prices in advanced economies over four quarters. - a 1.1 percentage point increase in downside risks to CRE prices in emerging market economies over four quarters. - The association is persistent and reaches 10 percentage points in advanced economies over twelve quarters (cumulatively) under some specifications. - Amplification channels: - Effects of CRE price misalignments on future downside risks to growth are more pronounced in economies with: - higher financial leverage (credit-to-GDP gap); example: high-leverage scenario implies a 50-basis-points larger GDP decline over four quarters (cumulatively) for the same level of misalignment. - surging CRE cross-border capital inflows; cross-border amplification becomes statistically and economically significant after about six quarters. - Underlying borrower and lender balance-sheet weaknesses (leverage, maturity mismatch) can create feedback loops between credit growth and asset prices. - Asymmetry: - Positive misalignment (overvaluation) matters more for financial fragility; interaction terms with an indicator for positive misalignment are negative and significant. ### V. Macroprudential policies: effects on downside risks to CRE prices and misalignment - Policy instruments analyzed: - CRE-specific borrower-based tools: loan-to-value (LTV) limits, debt-service-coverage (DSTI/DSTI-like) ratio limits, total debt servicing ratio (TDSR) examples. - CRE-specific capital-based tools: higher risk weights, sectoral capital buffers, minimum risk-weight floors. - Capital flow management (CFM) measures: overall inflow restrictions and real-estate specific inflow restrictions. - Identification and coding: - CRE-specific measures coded quarterly as -1 (loosening), 0 (no change), or 1 (tightening) and purged of credit-to-GDP variation to address endogeneity. - Main quantitative results (policy effects on CRE prices and misalignment): - Aggregate headline: - A targeted tightening measure reduces downside risks to CRE prices by 2.5 percentage points over 8 quarters on average (reported in source summary). - Panel/quantile estimates: - Baseline result: a tightening of targeted CRE measures reduces downside risks to CRE price growth by 0.3 percentage points per quarter in the near term (over 8 quarters). - Example: a macroprudential tightening targeted to CRE vulnerabilities two years before the global financial crisis would have reduced on average the decline in CRE prices from about 11 percent to 8.5 percent. - Heterogeneous timing and persistence: - Borrower-based measures: larger near-term effect; less significant twelve quarters ahead in examples. - Capital-based measures: larger and more persistent effects after eight quarters; twelve quarters ahead per-quarter impact on the lower tail of CRE price growth reported as 0.35 percentage points in some estimates. - CRE-price-misalignment reduction: - Panel quantile estimates of the effect of CRE-specific macroprudential measure on future CRE price misalignment change report coefficients (5th percentile) that become large and significant at medium horizons. Selected coefficients (coefficients with standard errors shown in source): - h=6: -1.063** (0.490) - h=7: -2.781** (1.313) - h=8: -3.264** (1.266) - h=12: -4.116* (2.301) - h=14: -5.045** (2.546) - h=15: -4.967** (2.497) - h=16: -4.406* (2.286) - Capital flow management measures: - Overall capital inflow restrictions (CFM Overall Inflow Restriction Shock) associated with reductions in downside CRE price risks; reported positive coefficients on the 5th percentile of CRE price growth (exact coefficients preserved in source) include significant positive values across h=1..h=16 (e.g., h=3: 0.244*** (0.085); h=5: 0.303*** (0.069); h=16: 0.174** (0.074)). - Real-estate specific inflow restrictions (advanced economies sample) also associated with significant reductions in downside CRE price risks (e.g., h=1: 0.183*** (0.059); h=5: 0.195*** (0.049); h=16: 0.143** (0.067)). - Timing and sequencing: - Impact of CRE-related policy measures is more long-lasting when measures are introduced during the early build-up phase of CRE price misalignment. - Tools aimed at increasing buffers should be deployed when risks are still building up; once the market has entered a downswing, tightening may be too late to prevent large price drops. - Transmission and policy choice guidance: - Borrower-based measures limit credit to new borrowers and curb amplification via defaults and credit growth—suitable for near-term control. - Capital-based measures enhance lender loss absorbency and resilience—suitable for longer-term durability but take longer to materialize. - Residential-targeted borrower-based measures that constrain multifamily credit can reduce amplification between residential and commercial segments; unreported results indicate such tightenings may reduce downside risks to CRE prices by about 2 percentage points (cumulative) in medium and long terms. - Circumvention and limits: - Macroprudential measures apply primarily to domestic banks and can be circumvented via foreign borrowing or nonbank channels; capital flow management and ownership restrictions have been used in some cases to limit foreign investor activity. ### VI. Policy implications, monitoring, and recommended actions - Monitoring: - Use present-value based frameworks that incorporate market risk premiums, financing terms, and measures of unconventional monetary policy to monitor CRE valuations relative to fundamentals. - Segment-specific monitoring recommended: retail and hotels faced the largest pandemic shocks; industrial and multifamily segments can behave differently and warrant segment-specific data and surveillance. - Risk management: - Recognize large CRE price misalignments increase tail risk for GDP growth; effects amplified by financial leverage and cross-border CRE inflows. - Prepare for valuation lags that can transmit further downward pressure on CRE values for some segments and economies. - Macroprudential strategy: - Use targeted CRE-related macroprudential measures (LTV limits, debt-service-coverage limits, CRE-specific risk-weights) to reduce downside risks to CRE prices. - Earlier intervention during the build-up phase of misalignment yields more durable risk reduction. - Combine borrower-based near-term tools with capital-based buffer increases for longer-term resilience. - Consider capital flow management or ownership restrictions where foreign inflows materially amplify CRE risks. - Supervisory and structural considerations: - Enhance supervisory attention to the CRE sector given its large size, heavy reliance on debt funding, and interconnectedness with the real economy. - Account for uncertainty in estimating corporate credit losses, potential capital shortfalls, levels of corporate indebtedness, and the role of non-bank financial institutions when calibrating tools. ### VII. Key descriptive statistics and robustness highlights (selected exact figures) - Table 1 descriptive statistics (selected): - CRE price growth: Mean 0.53; SD 1.85; Min -20.78; Max 8.37; Observations 1,836 - CRE price misalignment: Mean 0.00; SD 0.10; Min -0.57; Max 0.69; Observations 1,757 - Capitalization rate (nominal): Mean 5.77; SD 1.42; Min 2.20; Max 10.11; Observations 1,757 - NOI growth: Mean -0.46; SD 4.07; Min -50.55; Max 76.56; Observations 1,598 - CRE-specific macroprudential policy: Mean 0.00; SD 0.11; Min -0.99; Max 1.01; Observations 1,793 - CFM Overall Inflow Restriction Shock: Mean 0.01; SD 0.21; Min -1.11; Max 1.00; Observations 1,568 - Robustness: - CRE Prices-at-Risk and Growth-at-Risk results robust to percentile choice (5th, 10th, 20th), alternative panel quantile estimation (Machado and Silva (2019)), inclusion of VIX and time-varying effects, and country fixed effects. - Significance and magnitudes consistently identify CRE price misalignment as a key driver of downside tail risks. *Source: wpiea2021264-print-pdf - References ______________________________________________________________ 30 — https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf* ### References ______________________________________________________________ 30 ### wpiea2021264-print-pdf - References ______________________________________________________________ 30 ### I. Introduction — scope, motivation, and methodology - Focus: analyze vulnerabilities in commercial real estate (CRE) markets across 23 advanced economies and 7 emerging market economies by assessing misalignments between observed CRE prices and fair-value implied by economic fundamentals. - CRE definition for the chapter: property owned for the primary purpose of benefitting from investment returns (includes the multifamily segment), distinct from owner-occupied and noninvestment leased real estate. - Methodology: - Model fair value as a function of expected income of the commercial property and the return of holding the property, building on the present-value relationship of Campbell and Shiller (1989). - Expected return specification includes required risk compensation for exposure to the overall property stock market; preferred risk model adds macro factors such as output gap, inflation, and money supply (broad money-to-output ratio). - Extended model for a subset of economies to include vacancy rates for scenario analysis of sustained CRE demand shocks. - Objectives: - Quantify CRE price misalignment (observed price versus fundamentals-implied fair value). - Assess how misalignment forecasts downside risks to GDP growth and financial stability. - Evaluate effectiveness of CRE-related macroprudential policy measures in reducing downside risks to CRE prices. ### II. Key empirical findings and market facts - Market size and exposures: - As of end-2019, the commercial real estate sector had total assets of about 20 percent of GDP, on average, across the sample, up from 17 percent a decade ago. - CRE sector size reached as high as 50 percent or more of GDP in Singapore, Sweden, and Switzerland. - In the United States and some European economies (Estonia and Poland), direct lending related to commercial real estate constituted more than 50 percent of total bank lending to nonfinancial corporations in 2019. - Pre-pandemic dynamics: - Median CRE price across economies steadily increased in the run-up to the COVID-19 pandemic; in Sweden and the United States, real CRE prices almost doubled between 2009 and 2019. - Nominal annual capital appreciation for office buildings and multifamily dwellings averaged about 3 percent globally in the pre-pandemic period. - Pandemic-period disruptions (2020): - Global real estate transaction volume fell by 39 percent in 2020, to its lowest level since 2012. - Same-store net operating income (NOI) changes in the first half of 2020 (biannual, same-store basis): - Retail assets: declined by 21.4 percent. - Hotel assets: declined by 40 percent. - Industrial property NOI: grew 1.4 percent. - Office sector NOI growth: remained flat for the six months to June 2020. - Residential property: recorded flat net income growth for the six months to June 2020. - Aggregate: declines in NOI translated into large drops in valuation in the U.S. and in Europe; impact varied widely across and within economies. - Model associations: - CRE prices are associated with movements in market risk premiums, the price and nonprice terms of financing, and unconventional monetary policy (measured by broad money-to-output ratio). - During the peak of COVID-19, model indicates signs of overvaluation as actual prices did not fall as much as model-implied fair values. ### III. Scenario analysis: permanent increase in vacancy rates - Shock analyzed: a permanent increase in the vacancy rate used to proxy a sustained decline in CRE-specific demand (e.g., due to persistent structural shifts such as e-commerce and teleworking). - Key quantitative result: - A permanent increase in the vacancy rate of 5 percentage points would result in a median drop in fair values of about 15 percent after five years (median across the subset with vacancy data). ### IV. Macroeconomic and financial stability effects of CRE price misalignment - CRE misalignment as a leading indicator of downside risk: - A 50-basis-point increase in the misalignment measure (corresponding to one standard deviation) could raise downside risks to GDP growth (cumulatively) by: - 1.4 percentage points over four quarters. - 2.5 percentage points over 12 quarters. - Amplification by financial vulnerabilities: - The effect of CRE price misalignments on future downside risks to growth is more pronounced in economies with higher financial leverage or with surging CRE cross-border capital inflows. - Underlying borrower and lender balance-sheet weaknesses (leverage, maturity mismatch) can create feedback loops between credit growth and asset prices. ### V. Macroprudential policies: effects on downside risks to CRE prices - Modeling approach: a CRE prices-at-risk model including changes in CRE-related macroprudential policy measures (loan-to-value limits, debt-service-coverage ratio limits, and risk-weights). - Key quantitative result: - A targeted tightening measure reduces downside risks to CRE prices by 2.5 percentage points over 8 quarters on average. - Timing effect: - The impact of CRE-related policy measures is more long-lasting when measures are introduced during the early build-up phase of CRE price misalignment. ### VI. Implications and takeaways for policymakers - Monitor CRE valuations relative to fundamentals using present-value based frameworks that incorporate market risk premiums, financing terms, and measures of unconventional monetary policy. - Recognize that large CRE price misalignments increase the economy’s tail risk for GDP growth, with effects amplified by financial leverage and cross-border CRE inflows. - Use targeted CRE-related macroprudential measures (LTV limits, debt-service-coverage restrictions, risk-weight adjustments) to reduce downside risks to CRE prices; earlier intervention during the build-up phase of misalignment yields more durable risk reduction. - Prepare for sectoral heterogeneity: retail and hotels faced the largest pandemic shocks; industrial and multifamily segments can behave differently and require segment-specific monitoring. *Source: wpiea2021264-print-pdf - References ______________________________________________________________ 30 — https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf* ### 1. Real Commercial Real Estate Prices in Selected Economies ### 1. Real Commercial Real Estate Prices in Selected Economies ### Market developments during the COVID-19 crisis - Commercial real estate (CRE) prices and returns: - Panel 1 shows CRE prices for economies in the core sample (Index, 2009=100). - Panel 2 shows the global total return decomposition for CRE segments over 2016–19. Capital growth measures the change in property valuations, net of any capital expenditure and receipts, relative to the capital employed. Income return measures the net income receivable in relation to the capital employed. - Transaction volumes and price changes: - Figure 2, panel 1 reports change in CRE transaction volumes (Percent, 2020:Q2 and 2020:Q4, year-over-year). - Figure 2, panel 2 reports CRE price growth rate (Percent versus 3-year average). - Figure 2, panel 3 reports global net operating income growth rate (Percent, six-month growth rate). - Figure 2, panel 4 reports commercial real estate prices (Percent, 2020:Q2 and latest, year-over-year). Latest data available are for January 2021 in Europe and February 2021 in the United States. Prices in US dollars are used to construct panels 3 and 4. - Sectoral impacts and delinquencies: - Lower revenues translated into reduced debt-servicing capacity and expectations of higher delinquency rates on CRE loans; delinquencies on commercial mortgaged-backed securities (CMBS) surged. - While overall delinquency rates for the sector are comparable to those during the global financial crisis (GFC), delinquencies in the retail and hotel sectors reached an all-time high in the second quarter of 2020. - Special servicing rates during the COVID crisis hit new all-time highs in September: lodging 26.04, retail 18.32, raising the overall reading to a post-GFC crisis high of 10.48 percent. - Vacancy rates: - Global office vacancy rates continued to rise, most notably in the United States: the vacancy rate of office space in the United States was about 9 percent at the end of 2019 but increased to 17 percent at the end of 2020 (Cushman and Wakefield, 2021). - Vacancy rates also increased in Australia and Asia Pacific markets such as Singapore; in Europe vacancy remained low in Paris and major German cities such as Berlin and Munich but increased in London. ### Comparison with previous CRE market downturns - Cyclicality: - Short-term cross-correlation between changes in real CRE prices and real GDP growth is strongly positive across economies (Figure 3, panel 1). - Historical price corrections and magnitudes: - United States CRE price declines: by 3 percent after the dot-com bubble burst in 2001 and by 31 percent during the global financial crisis (GFC). - Ireland’s CRE prices experienced a sharp nominal decline during the global financial crisis of about 70 percent and never fully recovered. - Timing and lags: - Post-GFC, real estate asset performance lagged wider equity markets: equities lost about a third of their value in the second half of 2008 with a rebound in 2009, while private real estate distress became much more severe in 2009 and played out over subsequent years. - During the GFC CMBS delinquencies and special servicing rates did not reach their peak of 10.34 percent and 13.36 percent until mid-2012. - Multifamily and lodging peaked at delinquency rates of 16.93 percent (special servicing 20.07 percent) and 19.4 percent (special servicing 25.59 percent), respectively, between the second half of 2010 and first half 2011. Overall delinquencies for each of the major five property sectors were elevated in the double-digits until early 2013. - Distinguishing features of the COVID-19 downturn: - The COVID-19-induced downturn is exogenous (public health origin) compared with prior endogenous systemic crises. - Heterogeneous impact on NOI across segments, with contact-intensive segments most affected. - Policy action was swifter and more sizable than during the GFC, cushioning the shock; investors had generally better solvency positions pre-COVID than during the early stages of the GFC. - Growth of real CRE assets was more limited in the runup to the pandemic: even in the top quartile of markets ranked by growth in stock, only 2 percent of stock was expected to be added each year compared with an average of 6 percent per year recorded ahead of previous downturns. - Nonetheless, valuation lags imply possible further downward pressure on CRE values going forward for some segments and economies. ### CRE valuation framework and variables - Conceptual foundation: - Following Campbell and Shiller (1989), CRE price (deflated by CPI) is expressed in terms of current and expected growth of net operating income (NOI), and current and expected total returns on CRE holdings. Equation (1) in the source sets out log Price-to-NOI as a function of expected future NOI growth and expected total returns. - Total returns are expressed as the sum of the 3-month short-term real interest rate and the spread of total returns over the real 3-month rate. - Structural estimation: - The model is embedded in a general equilibrium framework expressed as a structural vector autoregression (SVAR) as in equation (2). The vector of fundamentals y_t includes: - NOI growth - CRE market risk premium (spread between capitalization rate and government bond yields) - output gap - CPI-based year-on-year inflation - non-financial corporate credit-to-output - 3-month short-term interest rate - broad money-to-output - capital flow-to-output - All variables except output gap and non-financial credit-to-output gap are demeaned. The output gap is estimated by HP-filtering GDP with a smoothing parameter of 1600. Credit-to-GDP corresponds to the Private Nonfinancial Credit to GDP ratio directly obtained from the Bank for International Settlements. - Definition of misalignment: - CRE price misalignment m_t is defined as the component of the detrended log Price-to-NOI ratio that cannot be explained by economic fundamentals (equations (3)–(5)). The residual m_t represents the market’s over(under) valuation of CRE price beyond fundamentals. - From equation (5) the contribution of different fundamentals to the fair value of CRE prices is computed as the predicted difference in CRE prices with and without the observed shock to that variable. ### Empirical findings on misalignment and drivers - Cross-country misalignment (2001–19 data availability for 11 economies): - Most economies did not enter the pandemic crisis with large price misalignments. - On average, the deviation of CRE prices from fair values before the pandemic is at around negative 2 percent—in contrast to the 8 percent overvaluation before the global financial crisis (Figure 5, panel 2). - CRE price misalignments generally increased in 2020 despite a decline in CRE prices, with the median value across economies reaching about 3.6 percent. - The largest deviations of price-to-NOI ratio from fundamentals were observed in the United Kingdom and the United States exceeding 10 percent of the price level implied by the model. - Decomposition for the United States (Panel 3, Figure 5): - Predating the GFC, a significant portion of the increase in valuations was driven by a rise in financial leverage and an increase in NOI; risk premia contributed only marginally to long-term price deviations during the run-up to the GFC. - During the GFC, risk premia and NOI were negatively affected; CRE valuations fell by more than implied decline based on fundamentals, offsetting pre-GFC overvaluation. Prices recovered since 2010 to an extent rationalized by expansion of unconventional monetary policy. - During the COVID-19 crisis, overvaluation increased because fundamentals deteriorated by more than CRE prices. The decline in aggregate demand and NOI was not fully offset by easing of monetary policy, leaving CRE prices overvalued despite their decline. - Segmental note: - Magnitude of CRE price misalignments in late 2020 varied across market segments. In general, the extent of misalignment is smaller in the office sector compared to retail, though in some economies large overvaluations have emerged in both segments. - Data limitations: - Lack of data availability for some variables precluded reliable estimation of fair values for other CRE segments such as the hotel and industrial segments. ### Implications and next steps - Potential for further price adjustments: - Given lags in valuation and the differential impact across segments, further downward pressure on CRE values is possible for some segments and economies. - Structural shifts in demand: - The analysis acknowledges potential structural changes in CRE demand going forward (e.g., shift toward e-commerce and teleworking induced by COVID-19). These structural shifts are difficult to forecast accurately; the source proceeds to examine their effects via scenario analysis (introduced in the following section of the source). *Source: wpiea2021264-print-pdf - 1. Real Commercial Real Estate Prices in Selected Economies* ### 3. United States: Decomposition of Estimated CRE Valuations ### 3. United States: Decomposition of Estimated CRE Valuations ### Fair-value model extension and vacancy-shock scenario - The fair-value model is extended to include vacancy rates; a permanent CRE-specific demand shock is introduced as a permanent increase in the vacancy rate (equivalently a CRE-specific demand shock included in the structural shock vector 풖풖풕 in Equation (3)). - Impulse response (ImS) is defined as the deviation of simulated price paths under the shock (푅푅_vvP_vvPv_v,t = 1) from the one forecasted without shocks (푅푅_vvP_vvPv_v,t = 0), as in equation (6). - Using the companion form 풚풚�_t = M0 + M 풚풚�_{t−1} + 풖풖�_t and equation (4), price-to-NOI and NOI are simulated by forward iteration; the model is linear so shock-size rescaling is feasible. - Two distinct calibrations/wordings appear in the source: - The shock size is calibrated so that the vacancy rate would gradually increase on average by 5 percentage points in the next 10 years. - The simulation assumes a 5 percent sustained increase in vacancy rates. - Intuition: if commercial spaces remain unoccupied due to a structural change in preferences, CRE cash flows and prices decline. ### Simulation results: CRE-price response to a permanent vacancy shock - Panel 4 (Figure 5) reports median and interquartile range of CRE prices in response to the permanent CRE-specific demand shock. - Key quantitative result: the median drop in fair values following a permanent increase in the vacancy rate by 5 percentage points would be about 15 percent after five years. - Uncertainty is substantial: interquartile ranges are reported alongside medians in the impulse-response plots. - The restricted sample for vacancy-data-based estimation includes: Australia, Canada, Denmark, Spain, Italy, Portugal, Sweden, the United Kingdom, and the United States. - Footnote/context notes: - Since shocks to vacancy rates are exogenous in the model, the shift in demand due to structural change in preferences is assumed to be unexpected. - If actual prices do not adjust downward accordingly (e.g., due to valuation uncertainty), prices may become overvalued, increasing the risk of a sharp correction later. - A 5-percentage point decline in the vacancy rate is equivalent to that experienced by the United States during the global financial crisis; the scenario abstracts from potential repurposing across CRE sectors. ### CRE price misalignment and downside risks to CRE price growth (CaR) - Approach: construct commercial-real-estate-prices-at-risk (CaR) as the conditional quantile (here the 5th percentile) of future CRE price growth using panel quantile local projections (two-step Canay (2011) procedure; equation (7) and (8)). - Dependent variable: Δ_h Y_P,t+h = average log change in CRE prices h periods ahead for economy i; explanatory vector X_P,t includes lagged CRE growth, CRE price misalignment, GDP growth, credit-to-GDP growth, capital flow-to-GDP, and an index of financial conditions and monetary aggregates. - Misalignment measure for CaR analysis (full sample limitations): calculated as the sign-inverted deviation of capitalization rates from the historical trend (instead of the fair-value-model misalignment used in earlier subsections). - Main findings on misalignment effects: - A one-standard-deviation higher price misalignment (corresponding to 10 basis points) is associated with: - a (cumulative) 2.5 percentage point increase in downside risks to CRE prices in advanced economies over four quarters. - a 1.1 percentage point increase in downside risks to CRE prices in emerging market economies over four quarters. - The association is persistent and reaches 10 percentage points in advanced economies over twelve quarters (cumulatively). - Time-varying and percentile robustness: - Results are robust to controlling for time-varying effects and to alternative percentiles; alternative panel quantile methodology (Machado and Silva (2019)) yields consistent results. - Historical CaR dynamics: - 1-year projection of CaR deteriorated during the GFC, recording an average quarterly CRE price decline of 12 percent on an annualized basis at the end of 2007. - Median CaR synchronized across major economies during the COVID-19 pandemic, showing a common decline. - Relative importance: - Standardized coefficients place CRE price misalignment among the largest drivers of downside CRE-price risk across key factors, especially after the fourth quarter in the forecasting horizon. ### CRE price misalignment, amplification channels, and downside risks to GDP growth - Conceptual channels: - Price misalignment raises likelihood of future price correction. - Balance-sheet vulnerability (leverage of borrowers and lenders) can amplify a CRE price downturn via credit-quality deterioration, bank losses, weakened capital, and credit supply contraction. - Cross-border capital flows and foreign investor behavior can amplify boom-bust cycles and synchronization across domestic and global CRE markets. - Empirical specification for GDP downside risks (equation (9)): - Δ_h Y_P,t,τ denotes average percentage change in real GDP growth h periods ahead at quantile τ (5th percentile); regressors include CRE misalignment (lagged), financial-conditions index (purged of CRE price variations), credit-to-GDP gap, lagged GDP growth, and house price growth controls. - For broader sample inclusion, CRE misalignment is proxied by the deviation of capitalization rates from historical trend. - Main quantitative results on GDP downside risk amplification: - A one-standard-deviation increase in the misalignment measure (corresponding to a negative deviation of the capitalization rate from its long-term trend by 10 basis points) raises downside risks to GDP growth by: - 1.4 percentage points in the short term (cumulatively over 4 quarters). - 2.5 percentage points in the medium term (cumulatively over 12 quarters). - Estimated coefficients are negative and statistically significant across the forecasting horizon and robust to inclusion of credit-to-GDP gap measures. - Subsample analysis: - The model is re-estimated separately for advanced and emerging market economies to account for structural differences; full regression results are reported in tables referenced in the source and coefficients are plotted (Figure 7). *Source: IMF staff calculations and figures from the content unit "3. United States: Decomposition of Estimated CRE Valuations" (wpiea2021264-print-pdf).* ### 3. Emerging Markets: Impact of CRE Price Misalignment on ### 3. Emerging Markets: Impact of CRE Price Misalignment on ### Impact of CRE price misalignment on downside risk to GDP growth - A one standard deviation increase in CRE price misalignment is defined as: - a negative deviation of the capitalization rate from its long-term trend by 10 basis points in advanced economies, and - 0.08 percent in emerging markets. - Estimated impacts: - Advanced economies: a one-standard-deviation increase in CRE price misalignment is associated with an increase in downside risk to GDP growth of 0.5 percentage points in the short term and 0.25 percentage points in the medium term. - Emerging market economies: the impact is about 0.2 percentage points in the short term. - Interpretation: - The effect is significant in both advanced and emerging market economies, though smaller and statistically weaker for emerging markets, possibly due to smaller CRE markets and lower credit-to-GDP gaps relative to advanced economies. ### Amplification by other financial vulnerabilities (interaction analysis) - Extended specification (equation (10)) interacts CRE price misalignment with: - the credit-to-GDP gap (to capture financial leverage), and - the cross-border CRE-capital-flows-to-GDP gap (to capture intensity of cross-border investments). - Leverage amplification (results summarized from Table 5 and Figure 8, panel 1): - Economies with a higher credit-to-GDP gap are more likely to experience a severe downturn across the forecasting horizon. - An economy with a higher level of leverage would incur a 50-basis-points larger GDP decline over four quarters (cumulatively) for the same level of CRE price misalignment at time t should large downside risks materialize. - Cross-border investment amplification (Table 6 and Figure 8, panel 2): - Economies with higher CRE cross-border investments have a larger effect of CRE price misalignments on downside risk to GDP growth. - Difference between “Average” and “Higher” cross-border investment trajectories is not significant in the short-term but becomes statistically and economically significant after six quarters. - Example: eight quarters ahead, the per-quarter price decline due to a one-standard-deviation increase in CRE price misalignment increases from 0.3 to 0.4 percentage points in economies more reliant on CRE cross-border flows. ### Interpretation of amplification channels - Possible mechanism: periods of rapid credit growth are typically accompanied by easing of bank lending standards; increased tolerance for risk leads to lending to less credit-worthy businesses, producing higher credit losses in a downturn, especially with a large asset price correction. - Cross-border amplification may also be linked to institutional investors (e.g., pension funds, insurance companies) whose share in cross-border flows increased since 2010 and who may be prone to flights to safety in large global shocks. ### Robustness and sign of misalignment - Robustness checks: - Controlled for time-varying effects (Annex Table A3.4). - Compared coefficients across choice of percentile (Annex Table A3.5). - Alternative panel quantile estimation methodology (Machado and Silva, 2019) reported in Annex Table A3.6. - Results are robust and consistent with the baseline specification. - Sign of misalignment: - Interaction of misalignment with an indicator for positive misalignment (overvaluation) shows the interaction coefficient is negative and significant, while the single-term coefficient is no longer significant (Annex Table A3.7). - Interpretation: positive misalignment (overvaluation) matters more for financial fragility. ### Macroprudential policy effects on CRE prices (overview) - Focus: targeted (“CRE-specific”) macroprudential measures that apply specifically to the CRE sector, including: - borrower-based policies (CRE-specific loan-to-value (LTV) and debt service-to-income (DSTI) ratios), and - capital-based policies (higher risk weights and sectoral capital buffers for CRE exposures). - Policy data sources: IMF iMaPP database, BIS, ESRB policy databases. - CRE-specific measure coding: - categorical variable equal to -1 (loosening), 0 (no change), or 1 (tightening) per quarter. - policy measure is purged of variation in credit-to-GDP to address endogeneity. ### Estimation approach for policy effects - Quantile regression model used to assess impact of policy measures on the 5th percentile of future CRE price changes (equation (11)). - Controls include change in macroprudential policy stance (ΔMP), monetary policy shock, lagged real GDP growth, changes in credit-to-GDP ratio, capital inflows to GDP ratio, the VIX index, and CRE price misalignment. - Interaction specification (equation (12)) includes ΔMP × CRE price misalignment to identify whether policy effectiveness depends on contemporaneous overvaluation. ### Quantitative findings on macroprudential effectiveness - Baseline result (Table 7 and Figure 9, panel 1): - A tightening of targeted CRE measures reduces downside risks to CRE price growth by 0.3 percentage points per quarter in the near term (over 8 quarters). - Economic implication: a macroprudential tightening targeted to CRE vulnerabilities two years before the global financial crisis would have reduced on average the decline in CRE prices from about 11 percent to 8.5 percent. - Heterogeneous effects by measure type (Table 8 and discussion): - Borrower-based measures: - Have a larger effect in the near-term. - Not significant twelve quarters ahead in the example provided. - Capital-based measures: - Have a larger effect after eight quarters. - Twelve quarters ahead, the per-quarter impact of capital-based measures on the lower tail of CRE price growth is equal to 0.35 percentage points. - Capital-based policies are more long-lasting but take more time to materialize. - Interaction with level of overvaluation (Table 9 and Figure 9, panel 2): - Macroprudential policies have the largest effect on curbing downside risk to CRE price growth during high overvaluation periods (CRE price misalignment one standard deviation higher than historical average). - The effect diminishes in the long-term. - A tightening of CRE-specific measures in periods of low overvaluation (historical average) can lower CRE price declines after three years by about (text truncated in source). ### Policy interpretation and recommended targeting - Conceptual channels: - Borrower-based measures (LTV, DSTI) limit credit to new borrowers, reduce default risk, and curb excessive credit growth. - Capital-based measures increase lenders' loss absorbency and enhance banks’ resilience to CRE loan defaults. - Policy choice guidance: - Appropriate selection of macroprudential tools should be contingent on the source and intensity of identified vulnerabilities: borrower-based tools for near-term credit growth control; capital-based tools for longer-term bank resilience. *Italicized source attribution: IMF staff calculations and figures from the provided chapter content.* ### 0.2 percentage points (per quarter) more than in periods of high overvaluation. ### wpiea2021264-print-pdf - 0.2 percentage points (per quarter) more than in periods of high overvaluation. ### Timing and effectiveness of buffer-increasing tools - Tools aimed at increasing buffers should be deployed when risks are still building up; when the market has already entered a downswing, it may be too late to tighten to prevent large price drops. - Empirical magnitude noted: "0.2 percentage points (per quarter) more than in periods of high overvaluation." - Sequencing and calibration of policy tools should consider: - (i) the potential capital shortfall and extra capital needed to maintain investors’ confidence during stress periods; - (ii) the uncertainty surrounding the estimation of corporate credit losses; - (iii) the level of corporate indebtedness; and - (iv) risks of policy leakage and the role of non-bank financial institutions. ### Transmission channels: borrower-based measures and residential linkages - Borrower-based measures targeting residential real estate can: - directly limit a borrower’s access to credit for multifamily housing (such as apartments); - dampen amplification effects from the interaction between residential and commercial real estate prices that threaten financial stability. - Unreported results using a similar identification strategy indicate borrower-based measures that include residential-targeted measures tend to have a significant impact, with a tightening reducing downside risks to CRE prices by about 2 percentage points (cumulative) in the medium and long terms. ### Macroprudential policies and CRE price misalignment - Macroprudential policies affect downside risks to CRE prices also by limiting CRE price misalignment (Annex A4.2). - Possible mechanism: tightening of borrowing limits and higher risk weights, if credible and large enough, can lead CRE investors to revise down expectations of future CRE prices and thereby reduce speculative incentives that play a key role in bubble dynamics. ### COVID-19 impact, CRE vulnerabilities, and macro-financial links - The commercial real estate sector was severely affected by the COVID-19 crisis, with transaction volumes and prices falling globally in 2020, especially in segments such as retail, hotels, and offices. - The sector’s characteristics that warrant enhanced supervisory attention: - large size of the sector; - heavy reliance on debt funding; - strong interconnectedness with the real economy, making it a source and amplifier of adverse macro-financial shocks. - Using a novel approach to assess CRE market fair values, the gap between observed prices and model-implied fair value increased in 2020 following a large drop in aggregate demand and net operating income. - Some drivers of the CRE shock were conjunctural; others indicate underlying structural changes, particularly in the retail segment where demand for traditional brick-and-mortar retail had been eroding before the pandemic as consumers shifted increasingly toward e-commerce. ### Macro-financial consequences of CRE price misalignments - CRE price misalignments increase the probability of a large CRE price correction and affect macro-financial outcomes. - An increase in commercial real estate price misalignment is associated with an increase in downside risk to GDP growth in the near- and medium term. - The impact is smaller and statistically weaker for emerging market economies relative to advanced economies. - The effect of CRE price misalignments on future GDP growth is amplified in the presence of underlying financial vulnerabilities, such as: - firms’ financial leverage; - cross-border capital inflows. ### Policy implications and recommended tools - Targeted macroprudential policy tools can help address vulnerabilities and price misalignments as the CRE market recovers and easy financial conditions persist. Examples highlighted: - limits on the LTV and DSTI ratios; - CRE-specific risk-weights. - Macroprudential measures are generally applicable to domestic banks but could be circumvented if CRE debt funding occurs by borrowing directly from abroad or through nonbank financial institutions. - In some cases, borrowing from abroad has been limited through capital flow management measures restricting investments by nonresidents (for example, ownership restrictions on nonresidents, or higher stamp duties for nonresidents on purchases of real estate). The analysis of such measures' effects on CRE prices is reported in Annex A4.3. *Source: wpiea2021264-print-pdf* ### References ### wpiea2021264-print-pdf - References ### Literature and empirical background - Citations focus on: commercial real estate (CRE) valuation, CRE and financial stability, macroprudential policy impacts, Growth-at-Risk and Prices-at-Risk methodologies, quantile regression and SVAR inference, cross-border capital flows, and data on pandemic policies. - Representative entries (authors and topics as listed): Adrian et al. (Vulnerable growth; term structure of growth-at-risk), Aizenman and Jinjarak (Current Account Patterns and National Real Estate Markets), Alam et al. (Digging deeper—effects of macroprudential policies), Barajas et al. (Loose Financial Conditions, Rising Leverage, and Risks to Macro-Financial Stability), Brandao-Marques et al. (riskiness of credit allocation), Baumeister and Hamilton (SVAR inference), Brueckner et al. (Work-from-Home spatial hedonic), Campbell et al. (rent–price ratio variance decomposition), Canay (Quantile Regression for Panel Data), Chernozhukov (Extreme Value Inference for Quantile Regression), Chaney et al. (Collateral Channel and corporate investment), Cushman and Wakefield reports, Davis et al. (price of residential land), Deghi et al. (Predicting Downside Risks to House Prices and Macro-Financial Stability), Duca and Ling (commercial CRE boom and bust), ESRB reports (2015, 2018), Fendoglu (CRE and US banking sector), Fernandez et al. (Capital Control Measures dataset), Fratzscher (Capital Flows), Hale et al. (Oxford COVID-19 Government Response Tracker), Iacoviello and Navarro (Foreign Effects of Higher U.S. Interest Rates), Nareit (Estimating the Size of the CRE Market), Olszewski (CRE market and macroprudential policy), Panagopoulos and Vlamis (Real estate IT and Basel II), Shibut and Singer (Loss Given Default for Commercial Loans), Van Nieuwerburgh (Why are REITs currently so expensive?). ### Table 1 — Descriptive statistics (exact sample measures) - CRE price growth: Mean 0.53; SD 1.85; Min -20.78; Max 8.37; Observations 1,836 - House prices growth: Mean 0.58; SD 1.97; Min -14.11; Max 14.08; Observations 1,836 - Credit growth: Mean 0.43; SD 1.85; Min -9.41; Max 28.77; Observations 1,836 - GDP growth: Mean 0.58; SD 1.15; Min -6.48; Max 20.89; Observations 1,836 - Financial conditions index: Mean 0.05; SD 0.72; Min -1.83; Max 4.06; Observations 1,836 - Credit-to-GDP gap: Mean 1.55; SD 17.92; Min -99.70; Max 87.20; Observations 1,836 - Capital Flow-to-GDP: Mean 0.02; SD 6.74; Min -79.36; Max 32.13; Observations 1,836 - VIX: Mean 19.24; SD 7.63; Min 10.12; Max 51.72; Observations 1,836 - Capitalization rate (nominal): Mean 5.77; SD 1.42; Min 2.20; Max 10.11; Observations 1,757 - CRE price misalignment: Mean 0.00; SD 0.10; Min -0.57; Max 0.69; Observations 1,757 - Total return: Mean 1.97; SD 1.78; Min -17.70; Max 9.54; Observations 1,757 - 3m interest rate: Mean 2.60; SD 2.40; Min -0.81; Max 12.00; Observations 1,699 - NOI growth: Mean -0.46; SD 4.07; Min -50.55; Max 76.56; Observations 1,598 - Monetary policy shock: Mean 0.00; SD 0.29; Min -1.52; Max 1.35; Observations 1,793 - CRE-specific macroprudential policy: Mean 0.00; SD 0.11; Min -0.99; Max 1.01; Observations 1,793 - Borrower-based macroprudential policy: Mean 0.01; SD 1.15; Min -4.60; Max 6.67; Observations 1,793 - CFM Overall Inflow Restriction Shock: Mean 0.01; SD 0.21; Min -1.11; Max 1.00; Observations 1,568 - CFM Real-estate Inflow Restriction Shock: Mean 0.00; SD 0.17; Min -1.04; Max 1.07; Observations 1,568 ### Tables 2–4 — CRE Prices-at-Risk and Growth-at-Risk baseline results (selected coefficient patterns and statistical significance preserved) - Table 2 (CRE Prices-at-Risk, dependent variable: 5th percentile of average CRE price growth distribution; full core economies sample): - GDP growth: coefficients across h=1..16 include 0.175*, 0.174**, 0.104, 0.178, 0.177, 0.086, 0.072, 0.024, 0.042, 0.045, 0.017, -0.003, -0.006, -0.016, -0.046, -0.039 (standard errors shown in parentheses). - FCI conditions index: initial negative and significant coefficients (e.g., -0.244**, -0.146**, -0.140**) diminishing across horizons. - Capital Flow-to-GDP: consistently negative and often highly significant (e.g., -0.338*, -0.473***, -0.449***, -0.534***, ... , -0.281***). - Credit-to-GDP growth: consistently negative and often highly significant (e.g., -0.439***, -0.438***, ... , -0.258***). - CRE price misalignment: strongly negative and highly significant across horizons (e.g., -0.346***, -0.487***, -0.572***, ... , -0.647***). - Observations decline from 1,845 at h=1 to 1,406 at h=16. - Table 3 (Growth-at-Risk with CRE Price Misalignment, dependent variable: 5th percentile of average GDP growth distribution; full core economies sample): - CRE price Misalignment: strongly negative and highly significant across horizons (e.g., -0.423*** at h=1 to -0.187*** at h=16). - Financial conditions index: negative and significant at short horizons (e.g., -0.399***, -0.242***), turning to positive and significant at some medium horizons (e.g., 0.097** at h=9). - Credit-to-GDP gap: consistently negative and highly significant across horizons (e.g., -0.491*** to -0.257***). - House prices growth: positive and significant across horizons (e.g., 0.266**, 0.321***, ... , 0.126***). - Observations from 1,792 at h=1 to 1,424 at h=16. - Table 4 (Growth-at-Risk split: Advanced Economies and Emerging Market Economies): - Advanced Economies: CRE price Misalignment coefficients strongly negative and highly significant across horizons (e.g., -0.450*** to -0.202***). Observations 1,415 to 1,139 across horizons. - Emerging Market Economies: CRE price Misalignment coefficients weaker and less consistently significant (e.g., -0.290* at h=1; many subsequent coefficients are smaller and not significant). Observations 377 to 285 across horizons. - Credit-to-GDP gap: negative and often stronger in emerging markets at several horizons (e.g., -0.821** at h=1), and consistently negative in advanced economies (e.g., -0.504*** at h=1). ### Tables 5–6 — Amplification effects (CRE price misalignment on Growth-at-Risk through leverage and cross-border flows) - Table 5 (Amplification through Credit-to-GDP Gap): - Scenario with Average Leverage (effect 훽̂τh): h=1..16 coefficients: -0.404***, -0.458***, -0.416***, -0.384***, -0.312***, -0.326***, -0.279***, -0.233***, -0.210***, -0.158***, -0.139***, -0.132**, -0.121***, -0.119***, -0.132***, -0.132*** (standard errors in parentheses). - Scenario with High Leverage (effect 훽̂τh + λ̂τh): h=1..16 coefficients: -0.582***, -0.597***, -0.551***, -0.475***, -0.391***, -0.404***, -0.365***, -0.301***, -0.28***, -0.24***, -0.229***, -0.221***, -0.206***, -0.185***, -0.182***, -0.173***. - Observations: 1,792 down to 1,424 across horizons. - Table 6 (Amplification through Cross-Border Capital Flows-to-GDP Gap): - Scenario with Average Cross-Border Investments (훽̂τh): coefficients include 0.029, -0.069, 0.206, 0.241***, 0.211**, 0.201***, 0.095**, 0.088**, 0.098***, 0.016, 0.034, 0.060**, 0.025, 0.005, 0.013, 0.023. - Scenario with High Cross-Border Investments (훽̂τh + λ̂τh): coefficients strongly negative and often highly significant across horizons (e.g., -0.452**, -0.545***, -0.325***, ... , -0.275***). - Observations: 1,792 down to 1,424 across horizons. ### Tables 7–9 — Impact and channels of CRE-specific macroprudential policies on CRE Prices-at-Risk - Table 7 (Impact of CRE-Specific Macroprudential Policy Measures on CRE Prices-at-Risk; extended CRE Prices-at-Risk specification): - CRE-specific macroprudential measure: coefficients across h=1..16 include 0.086, 0.156***, 0.165**, 0.202***, 0.244***, 0.282***, 0.307***, 0.326***, 0.201**, 0.125, -0.025, -0.066, -0.073, -0.024, 0.006, 0.037 (standard errors shown). - CRE price misalignment: strongly negative and highly significant across horizons (e.g., -0.923***, -1.204***, -1.315***, ... , -0.912***). - Other controls of note: Capital Inflow-to-GDP (-1) negative and often highly significant (e.g., -1.074***, -1.414***, ... , -0.279***); VIX (-1) negative and significant at some horizons (e.g., -0.744**, -0.775***). - Observations: 693 down to 532 across horizons. - Table 8 (Separate effects of CRE-specific borrower-based and capital-based measures): - CRE-specific borrower-based measures: some positive and significant coefficients at medium horizons (e.g., 0.222* at h=5, 0.282** at h=7, 0.332*** at h=8), with coefficient pattern reported across h=1..16. - CRE-specific capital-based measures: positive and increasingly significant at medium to longer horizons (e.g., 0.321** at h=9, 0.379*** at h=11, 0.354*** at h=14, 0.288*** at h=15, 0.244*** at h=16). - CRE price misalignment: strongly negative and highly significant across horizons (e.g., -0.882*** to -0.899***). - Observations: 693 down to 532 across horizons. - Table 9 (Effect of CRE-Specific Macroprudential Measures through CRE Price Misalignment scenarios): - Scenario with Low CRE price Misalignment (φ̂τh): coefficients h=1..16: 0.088, 0.239***, 0.245***, 0.258***, 0.260***, 0.284***, 0.307***, 0.315***, 0.245***, 0.229***, 0.185***, 0.160*, 0.154**, 0.149**, 0.139***, 0.120***. - Scenario with High CRE-Price Misalignment (φ̂τh + σ̂τh): coefficients h=1..16 include 0.328, 0.483**, 0.471**, 0.439***, 0.358***, 0.275***, 0.314***, 0.289***, 0.114, 0.0531, -0.00631, -0.0512, -0.0531, -0.0227, 0.0114, 0.0407 (standard errors in parentheses). - Observations: 693 down to 532 across horizons. ### Methodological notes and estimation details (as reported) - Dependent variables: - CRE Prices-at-Risk specifications: dependent variable corresponds to the 5th percentile of the average CRE price growth distribution over the forecasting horizon h. - Growth-at-Risk specifications: dependent variable corresponds to the 5th percentile of the average GDP growth distribution over the forecasting horizon h. - Covariates are standardized so that magnitudes of coefficients indicate relative importance of variables. - Standard errors are bootstrapped and shown in parentheses. - Significance notation: *** p<0.01, ** p<0.05, * p<0.1. - CRE-specific measures coded as categorical variable taking values -1 (loosening), 0 (no change), or 1 (tightening); measures are purged of credit-to-GDP ratio to address potential endogeneity. - CRE-specific borrower-based measures include CRE-specific LTV and DSTI ratios; capital-based measures include higher risk weights and sectoral capital buffers for CRE exposures. - Monetary policy shock measured as predicted residual from regressing policy rate on contemporaneous and lagged variables and a quadratic time trend (as in Iacoviello and Navarro (2019)). - Sub-sample analyses: results reported for full sample of core economies, advanced economies, emerging market economies, and sub-samples where at least one CRE-specific measure was announced. *Italic: Authors’ calculations as presented in the References and Tables section of the source PDF.* ### Annex 1. Data Description and Sources ### Annex 1. Data Description and Sources ### Variable definitions and data sources - Bank Stock Returns: Refinitiv Datastream's bank sector return index — Refinitiv Datastream - Broad Money: Broad money, seasonally adjusted — IMF, World Economic Outlook - Break Even Inflation: 10-years break even inflation rate — Bloomberg - Capital Flow Management Measures: Measures that are designed to limit capital flows — Fernandez and others (2017); and IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER) - Capital Inflows: Sum of portfolio investment and foreign direct investment — IMF, Balance of Payments database - Consumer Price Index: Consumer price index, percent — IMF, World Economic Outlook - CRE and CMBS Delinquency Rates: The percentage of CRE and CMBS loans within a financial institution's loan portfolio whose payments are delinquent — Trepp - CRE Capitalization rate: Net Operating Income per CRE value — MSCI Real Estate - CRE Investments: Investments in commercial real estate — Real Capital Analytics - CRE Prices: Asset value index — MSCI Real Estate - CRE Vacancy Rates: Total market rental value in vacant units / total market rental value — MSCI Real Estate - Credit-to-GDP Ratio: Private-sector credit in percent of GDP — Bank for International Settlements - Credit-to-GDP Gap: Deviation of Credit-to-GDP Ratio from the trend. — Bank for International Settlements - Financial Condition Index: For methodology and variables included in the FCI, refer to Annex 3.2 of the October 2017 Global Financial Stability Report. Positive values of the FCI indicate tighter-than-average financial conditions. — IMF staff estimates - Gross Domestic Product: Gross domestic product (GDP) — IMF, World Economic Outlook - Global Liquidity Indicator: The sum of bank loans to non-banks and debt securities issuance by non-banks — Bank for International Settlements - Long-Term Nominal Interest Rate: 10-years government bond yield (Please confirm. Just used the shared dataset). — IMF, World Economic Outlook - Long-Term Real Interest Rate: 10-years real interest rate index — Bloomberg - Macroprudential Measures: CRE-specific measures: Measures that are designed to limit the build-up of vulnerabilities in CRE sector — IMF, The integrated Macroprudential Policy (iMaPP) database, BIS Macroprudential Database, ESRB Macroprudential Measures Database, ESRB (2015, 2018, 2019) - Macroprudential Measures: Capital Requirements: Capital requirements for banks, which include risk weights, systemic risk buffers, and minimum capital requirements. Countercyclical capital buffers and capital conservation buffers are captured in their sheets respectively and thus not included here. — IMF, The integrated Macroprudential Policy (iMaPP) database - Macroprudential Measures: Limits on the Debt-Service-to-Income Ratio: Limits to the debt-service-to-income ratio and the loan-to-income ratio, which restrict the size of debt services or debt relative to income. They include those targeted at housing loans, consumer loans, and commercial real estate loans. — IMF, The integrated Macroprudential Policy (iMaPP) database - Macroprudential Measures: Limits on the Loan-to-Value Ratio: Limits to the loan-to-value ratios, including those mostly targeted at housing loans, but also includes those targeted at automobile loans, and commercial real estate loans. — IMF, The integrated Macroprudential Policy (iMaPP) database - Net Operating Income: Total net operating income for the period as an absolute amount — MSCI Real Estate - Net Operating Income Yield: Total net operating income for the period in percent of CRE value — MSCI Real Estate - Short-Term Nominal Interest Rate: Short-term deposit rate — IMF, World Economic Outlook - Policy Rate: Monetary policy rate and shadow rate by Leo Krippner(2013, 2015) — IMF, World Economic Outlook, Bank for International Settlements, Leo Krippner(2013, 2015) - VIX: CBOE Volatility Index — Refinitiv Datastream ### Data usage notes - Financial Condition Index methodology: see Annex 3.2 of October 2017 Global Financial Stability Report (positive values indicate tighter-than-average conditions). - Vacancy rate is transformed via inverse logit in the SVAR (see identification below). - Long-term nominal rate note: 10-years government bond yield used from shared dataset (request confirmation in source). --- ### Annex 2. Commercial Real Estate Prices and Fundamentals ### Identification of shocks in SVAR and CRE prices - Model extension: vacancy rate is added to capture CRE market–specific demand shocks distinct from aggregate demand. - Model representation (as in the source): - The vector y_t includes CRE prices and drivers plus the inverse logit-transformed vacancy rate (V S P S R p y R S R e_t is the inverse logit-transformation of the vacancy rate). - Once dynamics of fundamentals are expressed in terms of shocks, CRE price dynamics are expressed as the sum of those shocks and misalignments, allowing simulation of additional shocks' impacts on CRE prices. - Contemporaneous parameter identification: requires sign restrictions following Baumeister and Hamilton (2015, 2018). ### Sign identification and priors - Sign restrictions are imposed on contemporaneous relations following Baumeister and Hamilton (2015, 2018). - Prior distributions: - Contemporaneous matrix A: truncated t-distribution to satisfy sign restrictions. - Magnitude of shocks (standard deviation of u_t): inverse Gamma distribution. - Lagged structural coefficients B*(L): assumed normal and set by Minnesota priors as outlined by Baumeister and Hamilton (2018). - Table A2.1 sign assumptions (shocks × variables). The source presents sign indicators (plus/minus) by shock and variable; key entries (as in the source) include: - For variable labeled N R R S R R N S S S (first row): signs include + + − and other combinations as in the source table. - For variable labeled I R I I S R P I R: signs include − + − - For variable 3M R S R e: + (in demand columns) - For various ratio variables (e.g., C P e S P R / N R R S R R R): mixed + and − signs per the source table. - Vacancy rate row: shows − + in supply/demand columns and additional +/− across shocks. - Implementation: prior of contemporaneous matrix set to satisfy these sign restrictions via truncated t-distribution; other priors as specified above. (Note: the source provides the full sign table as Table A2.1. The above bullets summarize the structure and indicate that mixed sign restrictions are applied across variables and shocks as in Table A2.1.) ### Sample - Country-by-country analysis period: 2001:Q2 to 2019:Q4 - Countries analyzed (unless stated otherwise): Australia, Canada, Denmark, Germany, Italy, Portugal, South Africa, Spain, Sweden, United Kingdom and United States. ### Historical decomposition - Any variable in the SVAR can be written as the sum of past shocks (equation (14) in the source): - y_t = sum_{k=0}^{∞} M_k u_{t−k}^* - Logarithm of price decomposed into fundamental shocks as in equation (A2.2) in the source (source provides the full algebraic decomposition). - This historical decomposition permits attributing movements in CRE prices to sequences of identified fundamental shocks and residual misalignments. ### Impulse response functions and figures - Example: Figures A2.1–A2.2 present impulse response functions for the United States: - Figure A2.1: Impulse Response Functions of CRE Price to shocks in CRE valuation drivers. Shaded areas correspond to 68 percent confidence interval. Labels: convmp = conventional monetary policy; rp = risk premium; cftoy = capital flow-to-output; noig = net operating income growth; ump = unconventional monetary policy; vcrate = vacancy rate. - Figure A2.2: Impulse Response Functions of Risk Premia to shocks in CRE valuation drivers. Shaded areas correspond to 68 percent confidence interval. Labels as above. - The full set of impulse response functions is available upon request (per source). *Annex 1. Data Description and Sources; Annex 2. Commercial Real Estate Prices and Fundamentals — as provided in the source PDF.* ### Annex 3. Robustness analysis ### Annex 3. Robustness analysis ### CRE Prices-at-Risk: controlling for aggregate time-varying effects (Table A3.1) - Model: CRE Prices-at-Risk specification described in equation (7) augmented with VIX; full sample of core economies; dependent variable = 5th percentile of the average CRE price growth distribution over forecasting horizon h. All covariates standardized. Standard errors bootstrapped (shown in parentheses). *** p<0.01, ** p<0.05, * p<0.1. - CRE price misalignment coefficients (h=1 to h=16): - -0.330***, -0.443***, -0.494***, -0.629***, -0.539***, -0.577***, -0.636***, -0.599***, -0.700***, -0.674***, -0.641***, -0.637***, -0.638***, -0.601***, -0.629***, -0.651*** - Corresponding standard errors (h=1 to h=16): (0.073) (0.093) (0.129) (0.157) (0.173) (0.109) (0.110) (0.097) (0.091) (0.081) (0.109) (0.116) (0.114) (0.110) (0.098) (0.053) - Other notable coefficient patterns (selected): - Capital Flow-to-GDP: -0.363**, -0.481***, -0.460**, -0.681***, -0.653***, -0.644***, -0.624***, -0.608***, -0.505***, -0.505***, -0.429***, -0.371***, -0.377***, -0.418***, -0.338***, -0.287*** - Credit-to-GDP growth: -0.322***, -0.342***, -0.333**, -0.184, -0.186*, -0.246***, -0.277**, -0.293**, -0.327***, -0.219*, -0.247***, -0.251**, -0.228**, -0.227***, -0.234***, -0.255*** - VIX: -0.621***, -0.668***, -0.691***, -0.683***, -0.690***, -0.678***, -0.599***, -0.489***, -0.349***, -0.303*, -0.149, -0.052, -0.003, 0.025, 0.003, -0.010 - Fixed effects and controls: Country FE = YES; CRE price growth lag = YES. - Observations by horizon (h=1 to h=16): 1,845; 1,818; 1,790; 1,762; 1,734; 1,706; 1,676; 1,646; 1,616; 1,586; 1,556; 1,526; 1,496; 1,466; 1,436; 1,406. ### CRE Prices-at-Risk: estimated at different percentiles (Table A3.2) - Purpose: test robustness against choice of percentile by estimating CRE Prices-at-Risk at 5th (baseline), 10th, and 20th percentiles. Dependent variable = percentile of average CRE price growth distribution over horizon h. All covariates standardized. Bootstrapped standard errors shown. - 5th percentile (baseline) CRE price misalignment coefficients (h=1 to h=16): - -0.346***, -0.487***, -0.572***, -0.578***, -0.593***, -0.523***, -0.561***, -0.577***, -0.639***, -0.685***, -0.658***, -0.648***, -0.638***, -0.604***, -0.635***, -0.647*** - SEs: (0.068) (0.099) (0.115) (0.121) (0.130) (0.107) (0.143) (0.059) (0.114) (0.090) (0.104) (0.098) (0.078) (0.099) (0.106) (0.045) - 10th percentile coefficients (h=1 to h=16): - -0.308***, -0.323***, -0.485***, -0.529***, -0.545***, -0.564***, -0.576***, -0.588***, -0.624***, -0.689***, -0.674***, -0.684***, -0.709***, -0.687***, -0.689***, -0.688*** - SEs: (0.085) (0.079) (0.125) (0.121) (0.093) (0.120) (0.102) (0.099) (0.089) (0.106) (0.097) (0.078) (0.075) (0.076) (0.104) (0.074) - 20th percentile coefficients (h=1 to h=16): - -0.178***, -0.233***, -0.298***, -0.356***, -0.475***, -0.509***, -0.572***, -0.610***, -0.627***, -0.668***, -0.710***, -0.728***, -0.738***, -0.743***, -0.777***, -0.775*** - SEs: (0.061) (0.050) (0.060) (0.097) (0.105) (0.097) (0.111) (0.105) (0.095) (0.052) (0.061) (0.066) (0.069) (0.052) (0.048) (0.046) - Fixed effects and controls: Country FE = YES; CRE prices growth lag = YES; Macro Controls = YES. ### CRE Prices-at-Risk: alternative quantile estimation methodology (Table A3.3) - Comparison of estimation methodologies for 5th percentile dependent variable: Canay 2-step (baseline) vs Machado and Silva (2019). - Canay 2-step (baseline) coefficients (h=1 to h=16): - -0.346***, -0.487***, -0.572***, -0.578***, -0.593***, -0.523***, -0.561***, -0.577***, -0.639***, -0.685***, -0.658***, -0.648***, -0.638***, -0.604***, -0.635***, -0.647*** - SEs: (0.068) (0.099) (0.115) (0.121) (0.130) (0.107) (0.143) (0.059) (0.114) (0.090) (0.104) (0.098) (0.078) (0.099) (0.106) (0.045) - Machado and Silva (2019) coefficients (h=1 to h=16): - -0.531***, -0.669*, -0.821**, -0.884***, -0.943*, -0.945***, -0.939***, -0.944***, -0.905**, -0.881***, -0.866***, -0.836***, -0.814***, -0.814**, -0.802**, -0.786*** - SEs: (0.179) (0.342) (0.330) (0.322) (0.542) (0.282) (0.243) (0.295) (0.411) (0.187) (0.163) (0.162) (0.149) (0.406) (0.337) (0.277) - Fixed effects and controls: Country FE = YES; CRE prices growth lag = YES; Macro Controls = YES. ### Growth-at-Risk with CRE price controlling for time effects (Table A3.4) - Model: Growth-at-Risk specification described in equation (9) augmented with VIX; full sample of core economies; dependent variable = 5th percentile of the average GDP growth distribution over horizon h. All covariates standardized. Bootstrapped SEs shown. *** p<0.01, ** p<0.05, * p<0.1. - CRE price misalignment coefficients (h=1 to h=16): - -0.326**, -0.349***, -0.351***, -0.311***, -0.279***, -0.276***, -0.225***, -0.207***, -0.199***, -0.177***, -0.154***, -0.163***, -0.163***, -0.168***, -0.167***, -0.181*** - SEs: (0.137) (0.072) (0.044) (0.040) (0.029) (0.024) (0.035) (0.037) (0.029) (0.027) (0.025) (0.037) (0.029) (0.026) (0.025) (0.020) - Other key variables (selected coefficients, h=1 to h=16 where relevant): - House prices growth: 0.245***, 0.343***, 0.354***, 0.338***, 0.309***, 0.268***, 0.284***, 0.218***, 0.233***, 0.214***, 0.174***, 0.139***, 0.155***, 0.138***, 0.121***, 0.126*** - Credit-to-GDP gap: -0.583***, -0.466***, -0.370***, -0.371***, -0.333***, -0.330***, -0.268***, -0.326***, -0.347***, -0.301***, -0.291***, -0.292***, -0.289***, -0.283***, -0.258***, -0.230*** - VIX: -0.890***, -0.154, 0.113, 0.027, 0.001, -0.246***, -0.066, 0.013, -0.005, 0.073*, 0.103**, 0.113***, 0.096**, 0.091**, 0.117**, 0.085** - Fixed effects and controls: Country FE = YES; GDP growth lag = YES. - Observations by horizon (h=1 to h=16): 1,792; 1,792; 1,785; 1,757; 1,730; 1,703; 1,676; 1,648; 1,620; 1,592; 1,564; 1,536; 1,508; 1,480; 1,452; 1,424. ### Growth-at-Risk: CRE price misalignment estimated at different percentiles (Table A3.5) - Purpose: robustness to percentile choice (5th baseline, 10th, 20th) for dependent variable = percentile of average GDP growth distribution; covariates standardized; bootstrapped SEs. - 5th percentile (baseline) CRE price misalignment coefficients (h=1 to h=16): - -0.423***, -0.372***, -0.331***, -0.311***, -0.271***, -0.280***, -0.230***, -0.212***, -0.197***, -0.180***, -0.148***, -0.163***, -0.167***, -0.171***, -0.178***, -0.187*** - SEs: (0.069) (0.058) (0.041) (0.040) (0.028) (0.028) (0.033) (0.030) (0.039) (0.029) (0.040) (0.033) (0.027) (0.023) (0.029) (0.025) - 10th percentile coefficients (h=1 to h=16): - -0.292***, -0.250***, -0.249***, -0.275***, -0.239***, -0.241***, -0.220***, -0.194***, -0.192***, -0.159***, -0.155***, -0.149***, -0.135***, -0.139***, -0.141***, -0.154*** - SEs: (0.053) (0.056) (0.037) (0.041) (0.037) (0.032) (0.025) (0.031) (0.029) (0.036) (0.033) (0.036) (0.033) (0.048) (0.044) (0.045) - 20th percentile coefficients (h=1 to h=16): - -0.153***, -0.133***, -0.182***, -0.171***, -0.179***, -0.176***, -0.185***, -0.190***, -0.189***, -0.169***, -0.158***, -0.155***, -0.148***, -0.143***, -0.139***, -0.146*** - SEs: (0.047) (0.029) (0.030) (0.024) (0.024) (0.025) (0.027) (0.025) (0.019) (0.018) (0.016) (0.017) (0.016) (0.019) (0.019) (0.022) - Fixed effects and controls: Country FE = YES; GDP growth lag = YES; Macro Controls = YES. - Observations by horizon (h=1 to h=16): 1,792; 1,792; 1,785; 1,757; 1,730; 1,703; 1,676; 1,648; 1,620; 1,592; 1,564; 1,536; 1,508; 1,480; 1,452; 1,424. ### Growth-at-Risk: alternative estimation methodology (Table A3.6) - Comparison: Canay 2-step estimation (baseline) vs Machado and Silva for CRE price misalignment effect on 5th percentile of average GDP growth distribution. - Canay 2-step (baseline) coefficients (h=1 to h=16): - -0.423***, -0.372***, -0.331***, -0.311***, -0.271***, -0.280***, -0.230***, -0.212***, -0.197***, -0.180***, -0.148***, -0.163***, -0.167***, -0.171***, -0.178***, -0.187*** - SEs: (0.069) (0.058) (0.041) (0.040) (0.028) (0.028) (0.033) (0.030) (0.039) (0.029) (0.040) (0.033) (0.027) (0.023) (0.029) (0.025) - Machado and Silva coefficients (h=1 to h=16): - -0.343**, -0.367**, -0.345***, -0.303***, -0.256***, -0.231***, -0.211***, -0.180***, -0.173***, -0.155**, -0.140**, -0.137**, -0.135**, -0.135***, -0.139***, -0.150*** - SEs: (0.174) (0.155) (0.132) (0.091) (0.083) (0.069) (0.068) (0.064) (0.063) (0.063) (0.057) (0.055) (0.053) (0.049) (0.047) (0.045) - Fixed effects and controls: Country FE = YES; GDP growth lag = YES; Macro Controls = YES. - Observations by horizon (h=1 to h=16): 1,792; 1,792; 1,785; 1,757; 1,730; 1,703; 1,676; 1,648; 1,620; 1,592; 1,564; 1,536; 1,508; 1,480; 1,452; 1,424. ### Growth-at-Risk: asymmetric CRE price misalignment effects (Table A3.7) - Model: Growth-at-Risk specification with interaction between CRE price misalignment and indicator equal to 1 when misalignment is positive; dependent variable = 5th percentile of average GDP growth distribution over horizon h. All covariates standardized; bootstrapped SEs shown. - CRE price misalignment (main effect) coefficients (h=1 to h=16): - -0.033, -0.066, -0.051, -0.011, -0.017, 0.066, 0.049, 0.095*, 0.103***, 0.118***, 0.096**, 0.107*, 0.065, 0.031, -0.002, -0.030 - SEs: (0.150) (0.067) (0.097) (0.066) (0.063) (0.063) (0.045) (0.056) (0.036) (0.036) (0.044) (0.058) (0.050) (0.035) (0.053) (0.036) - CRE price misalignment × Positive CRE price misalignment (interaction) coefficients (h=1 to h=16): - -0.734**, -0.687**, -0.282**, -0.398***, -0.343**, -0.350***, -0.332***, -0.359***, -0.336***, -0.338***, -0.295***, -0.273***, -0.249***, -0.207***, -0.176***, -0.154*** - SEs: (0.368) (0.290) (0.131) (0.120) (0.141) (0.072) (0.080) (0.062) (0.047) (0.041) (0.035) (0.047) (0.043) (0.035) (0.055) (0.036) - Positive CRE price misalignment (separate row) coefficients (h=1 to h=16): - 0.045, -0.102, -0.329**, -0.238**, -0.187**, -0.218***, -0.147**, -0.143**, -0.104**, -0.106*, -0.082, -0.088*, -0.091*, -0.084**, -0.049, -0.053 - SEs: (0.146) (0.148) (0.143) (0.094) (0.092) (0.068) (0.066) (0.057) (0.052) (0.056) (0.052) (0.053) (0.047) (0.039) (0.031) (0.033) - Other controls (selected): Financial conditions index: -0.395***, -0.276***, -0.186*, -0.161***, -0.147***, -0.158***, -0.056, -0.026, 0.027, 0.032, 0.048, 0.039, 0.052*, 0.033, 0.055*, 0.045 - Fixed effects and controls: Country FE = YES; GDP growth lag = YES. - Observations by horizon (h=1 to h=16): 1,792; 1,792; 1,785; 1,757; 1,730; 1,703; 1,676; 1,648; 1,620; 1,592; 1,564; 1,536; 1,508; 1,480; 1,452; 1,424. *Annex 3. Robustness analysis — tables and notes as presented in the source PDF.* ### Annex 4. Macroprudential Policies and Downside Risks to CRE Prices ### Annex 4. Macroprudential Policies and Downside Risks to CRE Prices ### CRE-Specific Macroprudential Policies (Table A4.1) - Selected country measures (Date and description preserved exactly as in source): - Denmark Jun. 2003: 60% LTV limit on recreational dwellings, office properties and retailing properties, industrial properties and craftsman's properties, collective energy-supply plants. - Hong Kong SAR Feb. 2013: 10 pp lower LTV limit on mortgage loans for all commercial and industrial properties - Hong Kong SAR May 2017: Lower the applicable DSR limit by 10 percentage points for mortgage to borrowers whose income is mainly derived from outside of Hong Kong SAR - Indonesia Jun. 2012: LTV limit of 70% on 2nd loan for an office/shop house; 60% for 3rd or more loans for an office/shop house - Indonesia Jun. 2015: Lifting LTV ratio for property (including office houses) loans - Indonesia Aug. 2016: Lifting LTV limit on office houses based on banks’ internal policy (first loan), 85% (second loan), 80% (third loan or more) - Indonesia Jun. 2018: Lifting regulatory limits on the first mortgage on home stores/home offices - Ireland Jan. 2007: Minimum risk weight on commercial property lending increased from 50% to 100% - Ireland Jan. 2014: Minimum risk weight applied to commercial property lending was increased to 100% from 50% - Norway Sep. 2014: Risk weight of 100% on CRE lending for banks using the standardized approach - Poland Jan. 2005: 100% risk weight on non-residential property - Poland Jun. 2014: 75% or 80% LTV limit on CRE loans if the part above 75% is insured or collateralized with funds on bank account, government or NBP securities - Poland Dec. 2017: For banks using the Standardized Approach to determine capital requirement: 100% risk weight on exposures secured by commercial immovable property located in Poland - Singapore Jan. 2013: Seller's stamp duties for industrial properties - Singapore Jun. 2013: Total Debt Servicing Ratio (TDSR) to the loan applied for both residential and nonresidential property (e.g., industrial and commercial property), and covers property both in- and outside of Singapore - Spain May. 2008: Stringent capital requirements on commercial real estate and residential real estate exposures - Sweden Jan. 2014: Risk-weight floor framework for commercial mortgages at 100% for exposures calculated according to the standardized approach for credit risk - United Kingdom Jan. 2014: Stricter criterion requirement for firms to determine whether the annual average loss rates for lending secured by mortgage on commercial real estate in the UK did not exceed 0.5% over a representative period - United Kingdom Oct. 2014: Stricter criteria for the eligibility of the 50% risk weight (RW) exposures fully and completely secured by mortgages on commercial real estate located in non-EEA country entered into force - United States Dec.2006: Guidance to banks with high CRE risk concentrations to tighten managerial controls - United States Jan. 2015: 150% risk weight on HVCRE exposure held by a banking organization - United States Dec.2016: Implementation of risk retention rule. The risk retention rules require that at least one sponsor of a securitization (or its majority owned affiliate) retain a 5 percent interest in the credit risk of the securitized assets. ### Effect of CRE-Specific Macroprudential Measures on CRE Price Misalignment (A4.2) - Model specification: Panel quantile model estimating Δ_h Y_{P,R,τ} (change in CRE price misalignment from t to t+h) as a function of CRE-specific macroprudential measures (MPP), macroprudential policy indicators (MP), lagged CRE price misalignment change, and other controls (equation (A4.2.1) in source). CRE-specific measures take values {-1, 0, 1} for loosening, no change, and tightening, and are purged of credit-to-GDP ratio. - Key qualitative finding: - "The findings suggest that macroprudential policies have an important role in curtailing CRE prices misalignment." - Estimated coefficient on CRE-Specific Macroprudential Measure from Table A4.2.1 (dependent variable: 5th percentile of future CRE price misalignment change distribution). Coefficients and standard errors preserved exactly: - h=1: -0.023 (0.138) - h=2: -0.032 (0.188) - h=3: -0.526 (0.436) - h=4: -0.487 (0.334) - h=5: -0.324 (0.229) - h=6: -1.063** (0.490) - h=7: -2.781** (1.313) - h=8: -3.264** (1.266) - h=9: -2.846* (1.603) - h=10: -2.918* (1.905) - h=11: -3.301 (2.184) - h=12: -4.116* (2.301) - h=13: -4.679* (2.539) - h=14: -5.045** (2.546) - h=15: -4.967** (2.497) - h=16: -4.406* (2.286) - Model controls and estimation details (preserved from source note): - Country FE: YES for all horizons. - CRE price misalignment change lag: YES for all horizons. - Observations by horizon: 691, 683, 675, 664, 653, 642, 631, 620, 609, 598, 587, 576, 565, 554, 543, 532. - All covariates are standardized. Standard errors are bootstrapped and shown in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1. ### Effect of Capital Flow Management Measures on Downside Risks to CRE Prices (A4.3) - Framework: Δ_h Y_{P,R,τ} regressed on change in capital flow management measures (Δ CFM) and controls (equation (A4.3.1) in source). The capital flow management indices are two-year rolling sums of individual measures (+1=tightening; 0=no change; -1=loosening), purged of capital-flow-to-GDP. - Qualitative finding: - "The results show that such measures are also associated with a reduction of downside risks in CRE prices." - Impact of Overall Capital Inflow Restrictions (Table 4.3.1). Dependent variable: 5th percentile of average CRE prices growth distribution. CFM Overall Inflow Restriction Shock coefficients and standard errors preserved exactly: - h=1: 0.172* (0.104) - h=2: 0.194* (0.104) - h=3: 0.244*** (0.085) - h=4: 0.249*** (0.078) - h=5: 0.303*** (0.069) - h=6: 0.287*** (0.082) - h=7: 0.218* (0.115) - h=8: 0.231*** (0.084) - h=9: 0.259*** (0.094) - h=10: 0.246*** (0.064) - h=11: 0.272*** (0.070) - h=12: 0.229*** (0.067) - h=13: 0.188** (0.083) - h=14: 0.177** (0.089) - h=15: 0.169*** (0.058) - h=16: 0.174** (0.074) - Model controls and estimation details for Table 4.3.1 (preserved): - Key controls include GDP growth (-1), Financial conditions index (-1), Capital Inflow to GDP (-1), Change in Credit-to-GDP (-1), VIX (-1), Monetary policy shock. - Country FE: YES for all horizons. - CRE prices growth lag: YES for all horizons. - Observations by horizon: 1,664 (h=1..h=8), 1,642 (h=9), 1,620 (h=10), 1,591 (h=11), 1,561 (h=12), 1,531 (h=13), 1,501 (h=14), 1,471 (h=15), 1,441 (h=16). - All covariates are standardized. Standard errors are bootstrapped. Significance: *** p<0.01, ** p<0.05, * p<0.1. - Impact of Real-Estate Specific Inflow Restrictions (advanced economies sample) (Table 4.3.2). CFM real estate inflow restriction shock coefficients and standard errors preserved exactly: - h=1: 0.183*** (0.059) - h=2: 0.166*** (0.037) - h=3: 0.163 (0.105) - h=4: 0.184** (0.081) - h=5: 0.195*** (0.049) - h=6: 0.164*** (0.057) - h=7: 0.115 (0.089) - h=8: 0.122** (0.061) - h=9: 0.122 (0.075) - h=10: 0.126** (0.050) - h=11: 0.141** (0.071) - h=12: 0.147*** (0.050) - h=13: 0.110** (0.048) - h=14: 0.081* (0.045) - h=15: 0.128** (0.055) - h=16: 0.143** (0.067) - Model controls and estimation details for Table 4.3.2 (preserved): - Key controls include GDP growth (-1), Financial conditions index (-1), Capital Inflow to GDP (-1), Change in Credit-to-GDP (-1), VIX (-1), MP Shock. - Country FE: YES for all horizons. - CRE prices growth lag: YES for all horizons. - Observations by horizon: 1,346 (h=1..h=8), 1,331 (h=9), 1,316 (h=10), 1,294 (h=11), 1,271 (h=12), 1,248 (h=13), 1,225 (h=14), 1,202 (h=15), 1,179 (h=16). - All covariates are standardized. Standard errors are bootstrapped. Significance: *** p<0.01, ** p<0.05, * p<0.1. *Source: wpiea2021264-print-pdf - Annex 4. Macroprudential Policies and Downside Risks to CRE Prices* --- _Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021264-print-pdf.pdf_