## cr18195

## Source details

**Canonical URL:** [cr18195](https://www.imf.org/-/media/files/publications/cr/2018/cr18195.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2018/cr18195.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2018/cr18195.pdf.json)

---

### Major findings on Ireland’s business cycle and potential output
- Assessing the business cycle in Ireland is complicated by the open character of its labor market and a large presence of globally active multinationals.
- Recent evidence points to a strong rebound of potential growth and a currently advanced cyclical position, with a positive and widening output gap.
- All methods (univariate (HP), multivariate filters, and the production function applied to GNI* in constant prices) suggest a positive output gap in 2017.
- The multinational sector is assumed to continuously operate at full capacity and is added to potential GNI* to derive potential GDP (GDP_p = GNI*_p + GDP_MNE).
- Structural models central element: the labor market (wage Phillips curve and Okun’s law in the multivariate filter; potential employment growth in the production function).
- Labor market equilibrium estimated separately by factoring in the importance of migration.
- Indicators of cyclical upswing: upward wage pressures and weak net immigration to date; risk of a boom-bust cycle if momentum continues.

### Labor market features, measurement implications, and NAWRU
- GNI* definition used: GNI* = GDP – (Net Factor Income) – (Factor Income of Re-Domiciled Companies) – (Depreciation on R&D-related IP imports) – (Depreciation on Aircraft Related to Leasing); GNI* may still contain some multinational activities.
- External job mobility is important: labor force fluctuations driven by age group 14–44 (the age group of most net migrants).
- High correlations documented:
  - between changes in the labor force and net migration;
  - between changes in the labor force and employment;
  - between labor force participation rate and net migration.
- Because labor force and employment often move jointly, part of the labor market cycle is not captured by the unemployment rate; the labor force cycle may affect wage dynamics beyond the unemployment rate gap.
- Estimated equilibrium unemployment rate (NAWRU):
  - Close to 9 percent over the last two decades (Figure 4).
  - Estimated NAWRU: 8 percent at end–2016; compared with a non-employment index of 9 percent.
  - NAWRU dynamics: mild decline during early-2000s boom; increase during the crisis.

### Weakened cyclical signals and wage dynamics (Phillips curve findings)
- Labor force cycles have weakened the Phillips curve relationship, suppressing wage signals about the labor market cycle.
- Historical patterns:
  - 1990s: wage growth accelerated only marginally when unemployment fell by 10 percentage points (outward migration effect).
  - 1999–2000: strong wage acceleration amid steady unemployment.
  - Early 2000s: wage growth trend-decelerated while unemployment stayed low.
- Migration affected Okun’s Law: after 2004 EU enlargement, GDP and employment growth strong while unemployment stable.
- Empirical Phillips/Okun relationships (selected regressions and statistics preserved exactly as reported):
  - Labor force (15–44) vs labor force (all) regression: y = 0.5395x + 0.3931, R² = 0.734.
  - Labor Force and Employment, Age 14–44 regression: y = 0.0451x + 0.828, R² = 0.66.
  - Phillips curve scatter: y = −2.0777x + 5.2398, R² = 0.397.
  - Okun’s Law alternative regression (illustrative weakened subperiod): y = −0.3915x + 5.5508, R² = 0.0054.

### Modeling approach for small open labor market and multivariate stochastic-trend model
- Proposal: extend Phillips curve specification with a labor force gap (LF gap = percentage deviation of actual LF from NAWLF) and model wage growth as function of unemployment rate gap (actual unemployment minus NAWRU) and the labor force gap.
- Standard structural Kalman filter inadequate; implemented alternative multivariate Kalman filter.
- Multivariate stochastic-trend model (Watson (1986) extension) decomposes Employment (E) and Labor Force (LF):
  - Et = Et* + Et^c ; LFt = LFt* + LFt^c.
  - Trends follow random walks with drift:
    - Et* = α + Et-1* + εt,  εt ~ iid N(0, σ)
    - LFt* = λ + LFt-1* + ψt,  ψt ~ iid N(0, σ)
  - Cycles co-determined by common factor net migration (NM):
    - Et^c = γ Et-1^c + δ NMt + υt,  υt ~ iid N(0, σ*500)
    - LFt^c = μ LFt-1^c + θ NMt + φt,  φt ~ iid N(0, σ*500)
  - Trend variance σ assumed 500 times smaller than cycle variance (noise-to-signal akin to HP filter).
  - Estimated parameters include α, γ, δ, λ, μ, σ, and θ.
- Net migration treated as co-determining common factor for employment and labor force cycles.

### Empirical outcomes from trend-cycle estimation and NAWRU table highlights
- Trends represent sustainable paths in employment and labor force.
- Cycle dynamics:
  - Cycle unsustainable during 2005–2008; correction in employment undershot potential; recent hiring brought employment close to sustainable trend.
- Selected Table 1 reported coefficients and sample (preserve values exactly):
  - Constant (trend): 49.1*** (0.2)
  - Constant (cycle): 0.001 (0.001)
  - Employment cycle t-1: 0.06*** (0.004)
  - Labor force cycle t-1: 3.1*** (0.01)
  - Net migration t: -31*** (0.01) and 25*** (0.01)
  - σ: -0.35*** (0.001)
  - No. of observations: 84

### Phillips curve estimation and gap testing (selected Table 2 results)
- Extending Phillips curve with the stochastic labor force gap significantly improves specification versus OECD and EC gap measures.
- Selected coefficients and diagnostics (preserve exactly):
  - Constant: 0.011*** (0.003), 0.015*** (0.003), 0.0078*** (0.0027), 0.009* (0.005), 0.0057* (0.003)
  - Wage growth t-1: 0.69*** (0.078), 0.571*** (0.084), 0.696*** (0.078), 0.705*** (0.083), 0.842*** (0.06)
  - Unemployment gap (Stochastic) t: -0.002*** (0.00076), -0.0033*** (0.0008)
  - Unemployment gap (OECD) t: -0.002** (0.0008), -0.0026** (0.0013)
  - Unemployment gap (EC) t: -0.0007 (0.001)
  - Labor force gap (Stochastic) t: -0.0012** (0.0004)
  - Labor force gap (OECD) t: -0.0005 (0.001)
  - R2 adj.: 0.76, 0.78, 0.76, 0.76, 0.73
  - D.W. statistics: 1.94, 1.86, 1.99, 1.99, 2.1
  - No. of observations: 84
- Long-term effects (in percent) reported:
  - Average wage growth rate: 3.55, 3.49, 2.57, 2.57, 3.61
  - Unemployment rate gap elasticity: -0.65, -0.77, -0.66, -0.88, Insignificant
  - Labor force gap elasticity: -0.28, Insignificant

### Potential output methodology and key results
- Potential output decomposition and methods (applied to GNI* in constant prices, quarterly):
  - Univariate HP filter with smoothing parameter 1600.
  - Multivariate Kalman filter relating output gap to wage inflation and unemployment rate gap; extended to include labor force gap:
    - Wage Phillips curve: π_t^w = ρ π_{t−1}^w + α g_t + β y_t + ε_t^π.
    - Okun’s law: u_t = θ u_{t−1} + δ y_t + ε_t^u.
  - Production function (Cobb-Douglas) with Solow residual B_t:
    - Y_t = B_t K_t^{(1−α)} (EDU_t L_t)^α, B_t = A_t CU_t^{(1−α)} (AHW_t)^α.
    - K_t via perpetual inventory: K_t = (1−ρ) K_{t−1} + I_t; depreciation ρ calibrated from Penn World Tables 9.0.
    - I_t modified to filter out IP-related MNE investment; initial capital stock set in 1997 assuming unaffected by IP-related MNE investments.
    - EDU: human capital index (Penn World Tables 9.0); CU approximated by Euro Area average; AHW = average hours worked; α = labor share from Penn World Tables 9.0.
    - Potential output derived using calibrated α, smoothed employment trends, TFP A (HP filter smoothing 1600), and actual capital stock; average hours and capital utilization at long-term values.
- Structural multivariate filter selected coefficients (Table 3) preserved:
  - Wage growth t-1: 0.01*** (0.003)
  - Phillips curve: 0.56*** (0.08)
  - Net migration / other coefficient: -31.9*** (3.11)
  - Okun’s law: 0.18*** (0.05); 0.68*** (0.07)
  - No. of observations: 84
- Findings:
  - Estimated potential GNI* growth recovered strongly in the post-crisis period; all methods point to strong recovery (Figure 8).
  - GDP potential growth remained positive during the crisis due to multinational sector effects.
  - Current potential GNI* growth rates comparable to 1990s levels.
  - Production function decomposition: rebound driven by total factor productivity gains and non-MNE capital accumulation; MNE sector consistently added momentum and smoothed crisis impact.

### Business cycle position and output gap decomposition
- Output gap switched from negative to positive in 2015.
- Domestic economy (GNI*) output gap more volatile than GDP gap because multinationals assumed to operate at closed output gap.
- Decomposition (production function) identifies cyclical employment and stretched capacity utilization (CU) as major drivers of recent positive and widening output gap; average hours worked (AHW) gradually approaching long-term values, reducing drag.
- Conclusion: Irish economy in upswing phase; positive output gap confirmed with confidence as range across three methods has narrowed.

### FDI concentration, corporate tax (CIT) regime, and CIT revenue composition
- Inward FDI concentration:
  - United States UBO of more than two thirds of total inbound FDI stock.
  - Immediate inbound shares: Netherlands 14 percent, Luxembourg 9 percent, United Kingdom 8 percent, Switzerland 7 percent, Bermuda 6 percent.
  - Only about 30 percent of inbound investment flows directly from the U.S. (implying more than half of U.S. investment channeled through a third jurisdiction).
  - Official UBO data identify only the UK share (< 3 percent).
  - Fifteen percent of inbound FDI ultimately owned by Irish investors (round-tripping).
- Outbound FDI major destinations: Luxembourg nearly half, United Kingdom 12 percent, United States 11 percent, Netherlands 7 percent, Bermuda 5 percent.
- Key CIT features:
  - Statutory trading income rate: 12.5 percent; passive income rate: 25 percent.
  - R&D tax credit: 25 percent for qualifying expenditure in EEA; unused credits transferable or refundable over three years or carried forward.
  - Knowledge box (2015): effective 6.25 percent income tax on qualifying IP profits proportional to R&D share, with up to 30 percent uplift.
  - Withholding tax typically 20 percent on outbound dividends/interest/royalties, with numerous exemptions.
  - Extensive tax treaty network: more than 70 countries.
  - Prior to 2015, tax residence based on place of effective management; 2015 amendment made resident if incorporated or effectively managed in Ireland; grandfathering through 2020.
  - No CFC rules prior to ATAD implementation; transfer pricing rules based on OECD guidelines apply only to trading income of larger companies.
- Capital allowances: Equipment 8 years; scientific equipment 1 year (expensing); IP 15 years; non-agricultural structures 25 years.
- Capital gains tax ring-fenced at 33 percent (raised from 20 percent after 2008).
- CIT revenue composition and productivity:
  - Foreign MNEs account for about 80 percent of total CIT revenues.
  - About 70 percent of foreign MNE CIT revenues (or 56 percent of total CIT revenues) comes from companies with U.S. parents.
  - Purely domestic Irish companies pay about 15 percent of total CIT revenues.
  - Ireland’s CIT revenue-to-GDP in 2016: 2.7 percent; European average: 2.6 percent.
  - Ireland’s CIT productivity in 2016: 21 percent; European average: 12 percent.
- Decomposition term values for Ireland (2016):
  - Ratio of CIT revenue to CIT base: 10 percent (passive income at 25 percent accounted for about 6 percent of taxable corporate profits in 2016).
  - Ratio of CIT base to CGOS: 49 percent.
  - Ratio of CGOS to GDP: 53 percent — highest in Europe by a large margin.
  - Resulting ratio of CIT revenue to CGOS ≈ 5 percent (European average 10.7 percent).

### Evidence of profit-shifting and U.S. MNE affiliate statistics (Table 1 preserved)
- U.S. BEA affiliate data for Irish affiliates (Millions of USD, unless specified):
  - Net Income: 6,635
  - Employment (headcount): 131,625
  - Payroll: 124,800
  - Property, Plant and Equipment: 9,304
  - Third-party sales: 73,438
  - Income Tax Amount: 2,356
  - Share of total (percent): 12.3, 0.9, 1.5, 5.2, 2.3, 4.4 (respectively)
- Interpretation: Irish affiliates show unusually high ratios of value added and total assets per employee; Ireland’s share of U.S. MNE profits much higher than shares of productive factors and third-party sales — consistent with inward profit-shifting.

### Estimated impact of U.S. TCJA on Irish FDI and CIT revenues
- Methodology follows Beer, Klemm and Matheson (BKM); key parameters (Table 2 preserved):
  - γ (Semi-elasticity of profits w.r.t statutory tax rate): 1.2
  - εFDI (Semi-elasticity of FDI w.r.t statutory tax rate): 0.5-2.4
  - ωUS weights: IMF CDIS bilateral FDI 20.2 (in text 25.2 percent cited as alternative); ORBIS microdata 44.3
  - κ (Irish capital weight): 60
  - ΔτUS (statutory rate change): 14 percent points
  - ΔτUS (EATR): 11.7 percent points
  - ΔτUS (EMTR): 7.6 percent points
- FDI responds more to EATRs and EMTRs than to statutory rates; U.S. expensing reduced EMTR to zero and EATR fell by 11.7 points for a company earning 20 percent economic profits.
- Two ωUS schemes used:
  - Bilateral FDI (IMF CDIS): 25.2 percent (text reports 20.2 in Table 2 parameters summary; both figures appear in source).
  - MNE microdata (ORBIS): 44.3 percent.
- Adjustments: BMK capital income share set to 60 percent for Ireland (vs BMK assumption of one third); ORBIS data replaced by Ireland Revenue and CSO 2015 data with minor upward adjustments.

### Quantitative estimates of TCJA impacts (preserve exact reported ranges)
- Average reduction in inbound FDI stock: 10 percent.
- Range of estimated FDI stock reductions: minimum 1.4 percent; maximum 30.5 percent.
- Average reduction of CIT revenues: 10 percent, equivalent to approximately 0.25 percent of GDP (or 0.37 percent of GNI*).
- Minimum revenue reduction: 0.007 percent of GDP.
- Maximum revenue reduction: 0.45 percent of GDP.
- Scope and limitations: estimates capture tax-induced capital and income movements from the 14-point statutory cut and expensing; do not capture other TCJA features, location-specific business exigencies, or ultimate beneficial ownership channels.

### TCJA major provisions and interaction with Irish position (selected)
- TCJA changes (selected exact figures):
  - Federal CIT rate cut by 14 points from 35 percent to 21 percent.
  - Expensing of capital investment: full and immediate deductibility (temporary through 2022).
  - Interest deductibility limited to 30 percent of EBITDA (from 2022 to 30 percent of EBIT).
  - 100 percent dividends-received deduction for foreign dividends from companies with ≥ 10 percent share.
  - GILTI effective statutory tax rate: 10.5 = 21*0.5 percent; after 20 percent FTC haircut effective rate 13.125 percent = 0.8*10.5 percent.
  - One-time transition tax on accumulated unrepatriated U.S. MNE foreign earnings (~US$3 trillion as of end 2017): 15.5 percent on earnings held in liquid assets; 8 percent on earnings reinvested in real assets.
- Observations:
  - Irish authorities expect U.S. transition tax to have little effect on Irish CIT revenues.
  - Moody’s (2018): Ireland largest single location for U.S. MNEs’ unrepatriated earnings, roughly US$670 billion or 20 percent of total.
  - Ireland’s 12.5 percent CIT rate appears well-positioned relative to GILTI focal rate 13.125 percent.

### Strategic implications for tax competition and policy options
- GILTI turns into a practical ceiling/floor (~13.125 percent) for effective foreign tax incidence; foreign countries can set CIT rates near that level without deterring U.S. MNE investment.
- Low-tax jurisdictions still have incentives to offer rates below GILTI to attract investment because U.S. MNEs may blend down foreign tax rates.
- Policy options to compensate potential CIT revenue declines (selected exact figures and ratios):
  - Ireland’s total tax revenues-to-GDP: 22.3 percent (OECD European average: 37 percent).
  - Options recommended: broaden VAT base (eliminate domestic zero-rating; tax energy at full rate), broaden recurrent property tax base, streamline PIT to reduce high marginal rates at relatively low incomes to attract skilled immigration.

### BEPS, ATAD, digital taxation, CCCTB, and implications for Ireland
- BEPS implementation and ATAD timelines:
  - BEPS actions centralize “substance” (DEMPE) and expect resolution of hard-to-value intangibles by 2020.
  - Ireland implemented BEPS minimum standards (Actions 5, 6, 13, 14); peer reviews passed or scheduled.
  - Ireland signed the Multilateral Instrument (MLI) in 2017; expects ratification and implementation "within the next two years"; the U.S. has not signed.
  - "Substantially all" BEPS actions expected in force for Ireland "by the end of 2020."
- ATAD requirements and Irish sequencing (selected dates and rules preserved):
  - Interest limitation (Article 4): limit net interest to 30 percent of EBITDA; general implementation by January 1, 2019, with possible deferment until 2024; Ireland to defer until January 1, 2024 citing “equally effective” rules.
  - CFC rules (Articles 7 and 8): adoption by January 1, 2019; Ireland must adopt a CFC regime.
  - Hybrid mismatch rules: revised provisions transposed by end of 2019; reverse hybrid rule deferrable until end of 2021.
  - Exit tax (Article 5): member states must adopt by January 1, 2020; Ireland’s existing exit tax not as broad as ATAD requires and must be amended.
  - GAAR (Article 6): adopt by January 1, 2019; Ireland’s existing GAAR largely compliant.
- Digital taxation and CCCTB risks:
  - OECD digital PE work aiming for consensus "by 2020"; EC draft directive considers "user participation".
  - EC provisional interim 3-percent equalization tax on certain gross revenues proposed.
  - CCCTB formulary apportionment formula: 1/3 * Sales + 1/3 * Assets + 1/3 * (1/2 * Payroll + 1/2 * Employment).
  - CCCTB adoption would require unanimous EU Council vote; Ireland exposed given lower shares of productive factors and third-party sales relative to profits.
- Annex I tax-minimizing structures summarized (Double-Irish Dutch Sandwich; Single Malt; IP on-shoring) and noted grandfathering and rule changes (2015 residence change; 2015–17 IP on-shoring impact on 2015 GDP surge 25.5 percent).

### CRE international synchronization, stylized facts, and transmission to Irish economy
- Drivers of CRE synchronization: synchronized fundamentals (income, interest rates), enhanced financial integration, currency unions, international lenders/investors, and global confidence/risk appetite shifts.
- Irish CRE market stylized facts:
  - Domestic banks’ CRE exposure: declined from nearly 30 percent of total loans in 2009 to about 10 percent in 2017q4.
  - Impaired CRE loans: about a quarter of total CRE loans remain impaired.
  - Foreign investment in CRE (2011–17): about €10 billion total; €5.5 billion by U.S.-based investors; U.K. investors 8 percent; other European investors 23 percent.
  - Irish investor CRE activity recovered since 2013 but averaged €1.6 billion annually in 2013–17, below pre-crisis levels.
  - Dublin office vacancy rate fell to 6 percent at end-2017 from about 24 percent in 2010; take up highest in last 15 years.
  - CBRE projects Dublin office supply increase annual average 220,000 square meters in 2018–20 vs historical average annual take up ~200,000 square meters in 2010–16.
- Two synchronization indicators defined and used (Synch1 level; Synch2 cycle); data: annual MSCI total return series 2000–16 across 27 countries in euro terms.
- Summary statistics (three sub-periods 2000–07, 2008–11, 2012–16) (selected exact table entries preserved):
  - Total Return averages: 10.5 (2000–07), 4.1 (2008–11), 8.8 (2012–16)
  - Synch1 averages: -9.2 (2000–07), -18.2 (2008–11), -11.5 (2012–16)
  - Synch2 averages: 0.2 (2000–07), 1.1 (2008–11), 0.1 (2012–16)
  - Total Return standard deviations: 8.4 (2000–07), 14.2 (2008–11), 7.2 (2012–16)
- Patterns:
  - Ireland’s cyclical synchronization (Synch2) relatively high with U.S., Great Britain, and Asian countries; Ireland’s level synchronization (Synch1) low (high volatility).
  - During crisis (2008–11) Synch1 declined while Synch2 increased: Irish downturn coincided with peers but was more severe.
- Empirical determinants (instrumental two-stage specification):
  - Bilateral linkages raising synchronicity: OUTPUTGAP_SYNCH, FINANCIAL (portfolio investment share), ABS_SPREAD (correlation of long-term yields).
  - GLOBAL factors: higher global liquidity increases synchronization; higher VIX reduces synchronization.
  - Crisis dummy: CRE return synchronicity declined significantly on average during global financial crisis.
  - TRADE bilateral variable not significant for CRE synchronization.
- Selected SYNCH2 Annex Table A1 coefficient highlights (preserve exact signs/levels where reported):
  - OUTPUTGAP_SYNCH: 0.442*** (significant)
  - FINANCIAL: 0.137* (in one specification)
  - CRISIS: 1.309*; 1.506*; 2.716**; 1.089; 1.134 (significant at 10/5 percent in some specs)
  - ABS_GOV YIELD SPREAD: -0.275* (one specification)
  - R2 overall range: 0.258–0.342 across specifications; # obs. range: 320–377; # countries 25–26.
- VAR analysis (2002q1–2017q3) (preserve exact impulse magnitudes reported):
  - VAR variables: ADVANCED_GDP; INT_CRE (international CRE factor); IRL_CRE; IRL_MFDD.
  - Common international CRE shock: one standard deviation = 0.8 percentage points → Irish CRE prices increase up to 3.5 percent y/y by fifth quarter (real terms).
  - One standard deviation shock to Irish CRE prices (~2 percent real) → IRL_MFDD increases 0.7 percent y/y by sixth–seventh quarters.
  - One standard deviation shock to IRL_MFDD (2.6 percent) → Irish CRE prices increase up to 5 percent y/y by fifth quarter.
  - One standard deviation shock to ADVANCED_GDP (0.4 percent) → IRL_MFDD increases on average 0.5 percent y/y in the first year.
  - Correlation matrix (exact entries):
    - ADVANCED_GDP with INT_CRE 0.688; with IRL_CRE 0.732; with IRL_MFDD 0.671.
    - INT_CRE with IRL_CRE 0.832; with IRL_MFDD 0.657.
    - IRL_CRE with IRL_MFDD 0.824.
- Policy implications and monitoring recommendations:
  - International CRE prices can be early-warning indicators for domestic CRE movements.
  - Close monitoring and data gap closure (transactions, leases, financing) recommended.
  - Prudential management of new bank lending to CRE and existing property exposure necessary to enhance bank resiliency.
  - Monitor nonbank financial institution funding and exposures; monitor banks’ risk appetite and business model changes.
  - Given lack of independent monetary policy and externally-financed CRE activity, scope exists to use tax measures (stamp duty, property tax, capital gains tax) to manage demand and dampen sensitivity to external CRE price movements.

*Italic: IMF staff analysis and calculations in cr18195 (IMF Country Report PDF chapter excerpts).*

### References _____________________________________________________________________________ 17

### References _____________________________________________________________________________ 17

### Major findings on Ireland’s business cycle and potential output
- Assessing the business cycle in Ireland is complicated by the open character of its labor market and a large presence of globally active multinationals.
- Results point to a strong recent rebound of potential growth as well as a currently advanced cyclical position, with a positive and widening output gap.
- All methods (univariate (HP) and multivariate filters, and the production function applied to GNI* in constant prices) suggest a positive output gap in 2017.
- The multinational sector has been assumed to continuously operate at full capacity and added to potential GNI* to derive potential GDP.
- The central element in the structural models is the labor market, namely:
  - the wage Phillips curve and Okun’s law (multivariate filter), and
  - potential employment growth (production function).
- Equilibrium in the labor market is estimated separately by factoring in the importance of migration in the Irish labor market.
- The Irish economy is in the midst of a cyclical upswing; signs include upward wage pressures and weak net immigration to date.
- These signs are consistent with a cyclical upswing amid strong estimated potential output growth and point to risks of a boom-bust cycle should growth momentum continue.

### Labor market features and implications for cyclical measurement
- The GNI* is used as a proxy for the mostly-domestic part of the economy; GNI* = GDP – (Net Factor Income) – (Factor Income of Re-Domiciled Companies) – (Depreciation on R&D-related IP imports) – (Depreciation on Aircraft Related to Leasing). GNI* may still contain some multinational activities.
- Labor market conditions in Ireland are affected by both internal and external job mobility.
- External job mobility is particularly important: labor force fluctuations have been driven by the age group of 14–44, the age group of most net migrants.
- High correlations observed:
  - between changes in the labor force and net migration,
  - between changes in the labor force and employment, revealing the jobs-driven nature of net migration,
  - between labor force participation rate and net migration.
- Native Irish account for an important part of labor force swings.
- Because the labor force and employment often move jointly, part of the labor market cycle is not captured by the unemployment rate; the labor force cycle may have additional effects on wage dynamics beyond the unemployment rate gap.

### Weakened cyclical signals and wage dynamics
- There are signs that the labor force cycle has weakened the Phillips curve relationship, suppressing wage signals about the labor market cycle.
- Historical patterns highlighted:
  - In the 1990s, wage growth accelerated only marginally when unemployment plummeted 10 percentage points, also due to outward migration.
  - In 1999–2000, wage growth accelerated strongly amid a rather steady unemployment rate.
  - In the early 2000s, wage growth trend-decelerated while the unemployment rate stayed low and did not change much.
- Migration patterns also affect Okun’s Law: during the early 2000s (especially after 2004 EU enlargement), GDP and employment growth were strong while unemployment was stable.
- As a result, employment and GDP growth can be strong without declining unemployment or clear wage pressures, potentially leading to unsustainable dynamics.

### Modeling approach to account for small open labor market
- Extending the Phillips curve specification with the labor force cycle is proposed to account for Ireland’s small open labor market.
  - Define a labor force gap as the percentage deviation of the actual labor force from the “non-accelerating wage growth labor force growth (NAWLF)”.
  - Wage growth change is modeled as a function of the unemployment rate gap (actual unemployment rate minus NAWRU) and the labor force gap.
- The usual structural Kalman filter estimation of NAWRU is inadequate because it does not ensure consistency between filtered NAWRU and its components (trends in employment and labor force), and because trends enter the signal equation non-linearly (ratio Et* / LFt*), violating standard Kalman filter requirements.
- Therefore, an alternative multivariate Kalman filter model is implemented.

### Multivariate stochastic-trend model for employment and labor force
- The trend-cycle decomposition of employment (E) and the labor force (LF) uses a Watson (1986) specification extended to a multivariate Kalman filter.
- Key model assumptions and structure:
  - Employment and labor force each decomposed into trend and cycle: Et = Et* + Et^c ; LFt = LFt* + LFt^c.
  - Trends follow random walks with drift:
    - Et* = α + Et-1* + εt,  εt ~ iid N(0, σ)
    - LFt* = λ + LFt-1* + ψt,  ψt ~ iid N(0, σ)
  - Cycles follow autoregressive processes co-determined by a common factor — net migration (NM):
    - Et^c = γ Et-1^c + δ NMt + υt,  υt ~ iid N(0, σ*500)
    - LFt^c = μ LFt-1^c + θ NMt + φt,  φt ~ iid N(0, σ*500)
  - The estimated variance (σ) of the trends is assumed to be 500 times smaller than that of the cycles (similar to the noise-to-signal ratio in the Hodrick-Prescott filter).
  - Other estimated parameters: α, γ, δ, λ, μ, σ, and θ.
  - Superscript c denotes cycle; * denotes trends.
- The model treats net migration as the co-determining common factor for cycles in employment and labor force.

### Empirical outcomes from trend-cycle estimation
- Resulting trends represent sustainable paths in employment and the labor force (Table 1 and Figure 3).
- Observed cycle dynamics:
  - The cycle was unsustainable during 2005–2008.
  - The subsequent correction in employment undershot potential employment.
  - Significant recent hiring brought employment back close to the sustainable trend.

### Figures and estimated statistical relationships (selected)
- Labor force (15–44) and labor force (all) plotted as y/y percent change (1999–2017). Regression: y = 0.5395x + 0.3931, R² = 0.734.
- Labor Force and Employment, Age 14–44 (1999–2017, y/y percent change). Regression: y = 0.0451x + 0.828, R² = 0.66.
- Phillips curve scatter (Wage growth vs Unemployment rate): y = −2.0777x + 5.2398, R² = 0.397.
- Alternative regression (1998–2017 vs 2002–2007) shown for Okun’s Law: y = −0.3915x + 5.5508, R² = 0.0054 (illustrating weakened relationship in certain subperiods).

*Source: cr18195 - References (IMF country report excerpts and chapter material, June 11, 2018).*

### 12.      According to estimates, the equilibrium unemployment rate has been close to

### 12.      According to estimates, the equilibrium unemployment rate has been close to

### Equilibrium unemployment (NAWRU) estimates and labor market trends
- The equilibrium unemployment rate has been close to 9 percent over the last two decades (Figure 4).
- The derived stochastic trends of the labor force and employment determine the estimate of NAWRU = 1 − E_t* / LF_t*.
- The estimated NAWRU shows:
  - A mild decline during the boom of early 2000s.
  - An increase during the crisis.
- Interpretation:
  - Hiring less skilled workers during booms reduces structural unemployment.
  - Layoffs during the crisis lead to gradual deterioration of skills of the unemployed and increase the equilibrium level of unemployment.
- Estimated NAWRU: 8 percent at end–2016; this is close to the non-employment index of 9 percent and is considered a plausible estimate.

- Table 1: Trend and Cycle Estimation (selected reported coefficients and sample)
  - Constant (trend): 49.1*** (0.2)
  - Constant (cycle): 0.001 (0.001)
  - Employment cycle t-1: 0.06*** (0.004)
  - Labor force cycle t-1: 3.1*** (0.01)
  - Net migration t: -31*** (0.01) and 25*** (0.01)
  - σ: -0.35*** (0.001)
  - No. of observations: 84
  - Source: IMF staff calculations.

### Testing gaps with Phillips curve and gap measures
- The plausibility of the implied unemployment rate gap and the labor force gap is tested in a Phillips curve estimation (Figure 5 and Table 2).
- Gaps compared: the stochastic gap derived here, OECD gap, and European Commission (EC) gap.
- Findings:
  - Gaps showing higher amplitudes tend to perform better and imply a likely sluggish equilibrium level.
  - The Irish labor market is one of the most flexible in the euro area and would be expected to exhibit relatively large movements of unemployment compared to its equilibrium level.
  - Extending the Phillips curve with the labor force gap significantly improves the Phillips curve specification and allows separating effects of the labor force cycle from the unemployment rate cycle.
  - The labor force gap of the OECD does not improve the specification (Table 2, column IV.), perhaps due to inconsistency between trend labor force and employment in the NAWRU estimation.
  - The unemployment rate gap of the European Commission is statistically and economically insignificant (Table 2, column V.).

- Figure 5 summary (unemployment and labor force gaps, percentage points):
  - Unemployment Rate Gaps plotted for 1999–2017: Stochastic, OECD, European Commission (range: about -6 to 8 percentage points).
  - Labor Force Gaps plotted for 1997–2017: Stochastic, OECD (range: about -15 to 15 percentage points).

### Wage pressures, net migration, and Phillips curve estimation
- Wage pressures have been importantly affected by both the unemployment rate gap and swings in net migration.
- Regression results (Table 2, column II) indicate:
  - Observed wage growth was dwarfed by the large net immigration during the boom of 2005–08 (Figure 6).
  - If the labor force had grown at a sustainable pace (in line with the derived stochastic trend), wage growth would have been higher during the boom, exceeding the in-sample average wage growth and aligning with the positive unemployment rate gap.
  - In recent years, with a broadly closed labor force gap, rising wage pressures reflect the sharply falling unemployment rate gap.

- Table 2. Ireland: Wage Phillips Curve Estimation (selected coefficients and diagnostics)
  - Columns I–V reported; notable coefficients:
    - Constant: 0.011*** (0.003), 0.015*** (0.003), 0.0078*** (0.0027), 0.009* (0.005), 0.0057* (0.003)
    - Wage growth t-1: 0.69*** (0.078), 0.571*** (0.084), 0.696*** (0.078), 0.705*** (0.083), 0.842*** (0.06)
    - Unemployment gap (Stochastic) t: -0.002*** (0.00076), -0.0033*** (0.0008)
    - Unemployment gap (OECD) t: -0.002** (0.0008), -0.0026** (0.0013)
    - Unemployment gap (EC) t: -0.0007 (0.001)
    - Labor force gap (Stochastic) t: -0.0012** (0.0004)
    - Labor force gap (OECD) t: -0.0005 (0.001)
  - Model diagnostics:
    - R2 adj.: 0.76, 0.78, 0.76, 0.76, 0.73
    - D.W. statistics: 1.94, 1.86, 1.99, 1.99, 2.1
    - No. of observations: 84 in all columns
  - Long-term effects (in percent):
    - The average wage growth rate: 3.55, 3.49, 2.57, 2.57, 3.61
    - Unemployment rate gap elasticity: -0.65, -0.77, -0.66, -0.88, Insignificant
    - Labor force gap elasticity: -0.28, Insignificant
  - Note: Wage growth as dependent variable; using quarterly data series; OLS.
  - Source: IMF staff calculations.

- Figure 6 summary (Wage Growth Decomposition, 1999–2017):
  - Series shown: Impact of LF gap (rhs), Wage growth (closed LF gap), Actual wage growth (fitted), Average wage growth.
  - Wage Per Employee (y/y; percentage points) plotted; impact of net migration evident during 2005–08.

### Potential output methodology and results for a globally-integrated economy
- Potential output for Ireland is modeled as sum of domestic and multinational parts:
  - GDP_p = GNI*_p + GDP_MNE.
  - The multinational sector is assumed to be continuously operating at full capacity.
- Methods applied to domestic economy (GNI* in constant prices, quarterly):
  - Univariate filter: Hodrick-Prescott with smoothing parameter 1600.
  - Multivariate filter: adapted IMF Research Department model relating output gap to wage inflation and unemployment rate gap; extended to include labor force gap:
    - Wage Phillips curve: π_t^w = ρ π_{t−1}^w + α g_t + β y_t + ε_t^π.
    - Okun’s law: u_t = θ u_{t−1} + δ y_t + ε_t^u.
    - The unemployment rate gap is estimated separately (stochastic trend) and used as an observed variable.
  - Production function approach: Cobb-Douglas with capital and labor inputs; unexplained part is total factor productivity B_t (Solow residual).
    - Y_t = B_t K_t^{(1−α)} (EDU_t L_t)^α, with B_t = A_t CU_t^{(1−α)} (AHW_t)^α.
    - K_t from perpetual inventory: K_t = (1−ρ) K_{t−1} + I_t; depreciation rate ρ calibrated using historical average from Penn World Tables 9.0.
    - I_t is modified real domestic investment filtering out IP-related MNE investment.
    - Initial capital stock set in 1997, assuming unaffected by IP-related MNE investments.
    - L is number of employed persons (national accounts concept).
    - EDU is human capital index (Penn World Tables 9.0).
    - CU is capacity utilization in manufacturing approximated by Euro Area average.
    - AHW is average hours worked (national accounts).
    - α is labor share (compensation of employees to gross value added, Penn World Tables 9.0).
    - Potential output derived using calibrated labor share, smoothed employment trends (stochastic trend), total factor productivity A (HP filter with smoothing factor 1600), and actual capital stock. Average hours worked and capital utilization assumed at long-term values.

- Table 3: Structural Multivariate Filter (selected reported coefficients)
  - Wage growth t-1: 0.01*** (0.003)
  - Phillips curve: 0.56*** (0.08)
  - Net migration / other coefficient: -31.9*** (3.11)
  - Okun’s law: 0.18*** (0.05); 0.68*** (0.07)
  - No. of observations: 84
  - Source: IMF staff calculations.

- Figure 7 summary (GDP Growth Contributions, 1999–2017):
  - Series shown: GDP of MNEs, GNI*, GDP (y/y percent change).
  - The MNE sector contributes strongly and smooths crisis impact.

### Potential output growth, drivers, and output gap implications
- Estimated potential output growth has recovered to its pre-crisis rate; all methods point to a strong recovery of potential GNI* growth in the post-crisis period (Figure 8).
- GDP potential growth remained positive even during the crisis, thanks to positive effects of the multinational sector.
- Current rates of potential GNI* growth are strong and comparable to those recorded during the 1990s.
- Drivers of the rebound (production function decomposition):
  - The rebound in potential growth has been driven by gains in total factor productivity and non-MNE related capital accumulation.
  - The MNE sector has consistently added momentum to potential GDP growth and helped smooth the impact of the crisis (Figure 9).

- Figure 8 summary (Potential Growth, 1999–2017):
  - Methods shown: PF (production function), HP (Hodrick-Prescott), MVKF (multivariate Kalman/filter).
  - GNI* Potential Growth and GDP Potential Growth series plotted (y/y percent).

### Business cycle position and output gap decomposition
- Despite strong potential growth, the output gap has become positive:
  - The output gap switched from negative to positive in 2015.
  - Output gap for domestic economy (GNI*) is more volatile than for GDP because multinationals are assumed to operate with closed output gap.
- Changes in the output gap are driven by cyclical employment and capacity utilization.
  - Decomposition of the GNI* output gap (production function approach) identifies cyclical employment and stretched capacity utilization as major drivers of the recent positive and widening output gap (Figure 11).
  - Average hours worked are gradually approaching long-term values and reducing the drag on the output gap.
- Conclusion on cycle position:
  - The Irish economy has firmly moved into the upswing phase of a business cycle.
  - The currently positive output gap is confirmed with reasonable confidence as the range of the three applied methods has recently narrowed.

- Figure 10 summary (Output Gap Estimates, 1999–2017):
  - GNI* Gap and GDP Gap average and range plotted (percent of potential).

- Figure 11 summary (Output Gap Decomposition, 1999–2017):
  - Components shown: CU (capacity utilization), TFP, Employment, AHW (average hours worked), and overall Output gap (percentage points).
  - CU and Employment are notable positive contributors to the recent output gap.

*Source: IMF staff calculations and analysis in cr18195 - 12.      According to estimates, the equilibrium unemployment rate has been close to (IMF Country Report PDF).*

### 4. Ireland’s bilateral FDI stocks are highly concentrated (Figure 3).  The largest source

### 4. Ireland’s bilateral FDI stocks are highly concentrated (Figure 3). The largest source

### Inward and outward FDI concentration and composition
- The United States is the ultimate beneficial owner (UBO) of more than two thirds of the total inbound FDI stock.
- Only about 30 percent of inbound investment flows directly from the U.S., indicating that more than half of U.S. investment is channeled through a third jurisdiction.
- Other major immediate sources of inbound FDI: Netherlands (14 percent), Luxembourg (9 percent), United Kingdom (8 percent), Switzerland (7 percent), Bermuda (6 percent).
- Official data on UBO investment identify only the UK share, which is less than 3 percent of total inbound FDI.
- Fifteen percent of inbound FDI is ultimately owned by Irish investors, indicating “round-tripping” of some domestic investment.
- Largest destinations for outbound FDI: Luxembourg (nearly half of total), United Kingdom (12 percent), United States (11 percent), Netherlands (7 percent), Bermuda (5 percent).

### Ireland’s CIT regime — key features that attract FDI
- Statutory corporation tax rate applied to active “trading” income: 12.5 percent.
- A higher 25 percent rate applies to passive income.
- Trading losses: may be carried back one year, offset against passive income on a value basis, or carried forward indefinitely against trading profits.
- Group consolidation: not allowed; companies may transfer losses and certain expenses where there is at least a 75 percent ownership link.
- Capital assets: may be transferred within a group without triggering the 33 percent capital gains tax.
- R&D and IP incentives:
  - 25 percent research and development (R&D) tax credit for qualifying expenditure within the European Economic Area (EEA), including wages, structures, equipment and overhead.
  - Unused R&D tax credits may be transferred to key R&D employees, refunded over three years, or carried forward.
  - A knowledge box (introduced in 2015) applies an effective 6.25 percent income tax rate to profits from qualifying IP assets proportional to the share of the cost of research and development by the Irish company in total development or acquisition costs of the IP. That share may receive a further “uplift” of up to 30 percent.
- Worldwide CIT: earnings of foreign subsidiaries are subject to taxation in Ireland when distributed to the Irish parent; foreign tax credits (FTCs) are granted and excess FTCs can be used to offset tax on other foreign income or carried forward indefinitely.
- Withholding tax (WHT): domestic law generally imposes a 20 percent WHT on outbound dividends, interest and royalties, but numerous exemptions exist (e.g., dividends to EU member states or tax treaty signatories, dividends to publicly listed companies and their subsidiaries; royalties and interest on long-term obligations paid to EU members or tax treaty signatories are usually exempt, provided the income is taxed in the foreign jurisdiction).
- Extensive tax treaty network: more than 70 countries; treaties often reduce or eliminate cross-border WHTs and have facilitated sectors such as global aircraft leasing.
- Tax residence rules: prior to 2015, certain companies were deemed Irish tax resident only if their place of effective management was in Ireland; in 2015 Ireland amended rules to include companies that are either incorporated or effectively managed in Ireland. Structures existing prior to 2015 are grandfathered through 2020.
- Cross-border anti-abuse rules: no controlled foreign corporation (CFC) rules; no thin capitalization rules (though certain related-party interest may be disallowed or recharacterized as dividends); transfer pricing rules based on OECD guidelines apply only to trading income of larger companies. The EU Anti-Tax Avoidance Directive (ATAD) requires introduction of earnings stripping and other rules within the next few years.

### Capital allowances and capital gains
- Depreciation choices: companies may choose between financial-account depreciation and straight-line tax depreciation allowances with terms: Equipment 8 years; scientific equipment 1 year (expensing); intellectual property 15 years; non-agricultural structures 25 years.
- Capital gains: ring-fenced and subject to a tax rate of 33 percent (raised from 20 percent following the 2008 financial crisis).

### Corporation tax revenues — composition and productivity
- Foreign MNEs account for about 80 percent of total CIT revenues.
  - About 70 percent of foreign MNE CIT revenues (or 56 percent of total CIT revenues) comes from companies with U.S. parents.
- Purely domestic Irish companies pay about 15 percent of total CIT revenues.
- Irish-parented MNEs account for a much smaller share (share rose from 3.6 percent to 5.8 percent in 2015).
- Ireland’s ratio of CIT revenue to GDP in 2016: 2.7 percent; European average: 2.6 percent.
- Ireland’s CIT productivity (CIT revenue divided by statutory CIT rate and GDP) in 2016: 21 percent; European average: 12 percent.
- Decomposition (Equation (1)):
  - CIT revenue / (CIT rate * GDP) = (1 / CIT rate) * (CIT revenue / CIT base) * (CIT base / CGOS) * (CGOS / GDP)
  - CGOS = corporate gross operating surplus.
- Key term values for Ireland (2016):
  - Ratio of CIT revenue to CIT base: 10 percent (includes passive income taxed at 25 percent, which accounted for about 6 percent of taxable corporate profits in 2016).
  - Ratio of CIT base to CGOS: 49 percent.
  - Ratio of CGOS to GDP: 53 percent — the highest in Europe by a large margin.
- Combining terms, the ratio of CIT revenue to CGOS is about 5 percent in Ireland, compared to a European average of 10.7 percent.
- Ireland’s high CIT productivity is driven largely by an unusually high share of corporate profits in GDP.
- The share of Irish wages in GDP is the lowest in Europe (Figure 9).

### Evidence suggesting profit-shifting and MNE profiles
- A high share of corporate profits in GDP—particularly for a low-CIT rate economy—may indicate inward profit-shifting by MNEs.
- U.S. MNE affiliate data (U.S. BEA) show Irish affiliates have an unusually high ratio of value added and total assets per employee (Figures 10 and 11).
- Ireland’s share of U.S. MNE profits is much higher than its shares of real productive factors and third-party sales (Table 1).
- Table 1 (Irish affiliates of U.S. MNEs, Millions of USD, unless otherwise specified):
  - Net Income: 6,635
  - Employment (headcount): 131,625
  - Payroll: 124,800
  - Property, Plant and Equipment: 9,304
  - Third-party sales: 73,438
  - Income Tax Amount: 2,356
  - Share of total (percent): 12.3, 0.9, 1.5, 5.2, 2.3, 4.4 (respectively)

### Assessment of the impact of the U.S. tax reform on Irish FDI and CIT revenues
- Methodology: follows Beer, Klemm and Matheson (BKM). Changes in statutory tax rates affect corporate tax revenues through profit-shifting and changes in FDI.
- Percentage change in FDI from change in U.S. statutory tax rate (ΔτUS) = ΔτUS * ω * εFDI, where εFDI is elasticity of FDI w.r.t tax rate and ω is share of total foreign investment involving the U.S.
- Percentage change in CIT revenues from changes in FDI and profit-shifting:
  - [ΔτUS * ω * (εFDI * κ + γ) * τIRL * πit] / CIT_revenue_it
  - κ = share of capital revenue in total revenue; γ = elasticity of corporate profits w.r.t statutory tax rate; τ = corporate tax rate (12.5 percent); π = corporate profits.
- Parameters summarized (Table 2):
  - γ (Semi-elasticity of profits w.r.t statutory tax rate): 1.2 (Beer, de Mooij, and Liu (2018))
  - εFDI (Semi-elasticity of FDI w.r.t statutory tax rate): 0.5-2.4 (Beer, Klemm and Matheson (2018), De Mooij and Ederveen (2008))
  - εFDI (w.r.t EMTR): 1.3-4 (De Mooij and Ederveen (2008))
  - εFDI (w.r.t EATR): 3.2-5.9 (De Mooij and Ederveen (2008))
  - ωUS weight based on bilateral FDI: 20.2 (IMF CDIS database)
  - ωUS weight based on corporate microdata: 44.3 (ORBIS database)
  - κ (Irish capital weight): 60 (OECD database)
  - ΔτUS (Statutory tax rate change): 14 percent points (Beer, Klemm and Matheson (2018))
  - ΔτUS (EATR): 11.7 percent points (Beer, Klemm and Matheson (2018))
  - ΔτUS (EMTR): 7.6 percent points (Beer, Klemm and Matheson (2018))
- FDI responds more to EATRs and EMTRs than to statutory rates. Due to expensing in the U.S. reform:
  - U.S. EMTR reduced to zero (from 7.6 percent prior to the reform).
  - U.S. EATR for a company earning 20 percent economic profits fell by 11.7 points.
- Two weighting schemes for ωUS:
  - Bilateral FDI (IMF CDIS): sum of Ireland’s bilateral inbound and outbound FDI with the U.S. divided by sum of total inbound and outbound FDI = 25.2 percent.
  - MNE microdata (ORBIS): average value for MNEs with an Irish affiliate of one divided by total number of countries with affiliates (less one) = 44.3 percent.
- Adjustments to BMK parameters used for Ireland:
  - BMK assumes average capital income share of one third; Ireland’s capital income share is much higher. Estimates were made using a capital share of 60 percent (Ireland’s average for the past five years).
  - ORBIS MNE profit and FDI stock data used in BMK were replaced by 2015 data from Ireland Revenue and the CSO, resulting in only minor upward adjustments.

*Source: IMF staff synthesis of chapter content.*

### 17.      Estimates of the FDI and

### 17.      Estimates of the FDI and

### Estimates of FDI and corporate tax revenue impacts
- The average reduction in the inbound FDI stock was 10 percent.
- The range (minimum to maximum) of estimated FDI stock reductions: minimum 1.4 percent; maximum 30.5 percent.
- The average reduction of CIT revenues was 10 percent, which as a share of GDP is approximately 0.25 percent (or 0.37 percent of GNI*).
- The minimum revenue reduction was 0.007 percent of GDP and the maximum was 0.45 percent of GDP.
- The range of estimates reflects variation in elasticities and U.S. investment shares applied.
- These estimates reflect tax-induced movements of capital and income only and do not capture location-specific business exigencies (for example, necessity of maintaining an operation in the EU to avail an EU passport).

*For technical note: EMTR measures the “tax wedge” on an investment that just breaks even on an after-tax basis; EATR measures the average tax burden on a discrete investment earning economic profits and is bounded between the statutory rate and the EMTR. Statutory rates are used in many empirical FDI studies.*

### Scope and limitations of the estimates
- Estimates capture only a portion of U.S. corporate tax reform effects: the 14–point cut in the statutory rate and expensing of capital investment (which affect the EMTR and EATR).
- Analysis incorporates removal of prior law’s deadweight loss associated with the cost of maintaining the old “deferral” system, as expressed by the profit-shifting elasticity.
- Estimates do not capture:
  - Other TCJA features and their possible implications (discussed in the following section).
  - Location-specific business exigencies that might induce or preserve investment in Ireland regardless of U.S. tax rate changes.
- The FDI relationships considered are based on immediate FDI relationships, rather than ultimate beneficial ownership.

### The U.S. Tax Reform — major provisions of the 2017 TCJA (as described)
- The standard federal CIT rate was cut by 14 points from 35 percent to 21 percent. This measure is permanent.
- Capital investment expense was made fully and immediately deductible (“expensing”). This measure is temporary through 2022, scheduled to phase down and revert to pre–2018 depreciation unless extended.
- Interest deductibility limited to 30 percent of EBITDA; from 2022 applied to 30 percent of EBIT. Applies to interest paid to related and unrelated parties.
- Dividends received by U.S. MNEs from their CFCs are generally removed from taxable income: 100 percent dividends-received deduction for foreign dividends from companies with at least a 10 percent share. Subpart F rules retained for specified passive and mobile income. Foreign branch earnings taxed at 21 percent.
- New worldwide tax without deferral: global intangible low-tax income (“GILTI”) tax applies to total non-U.S. active income of a U.S. MNE’s CFCs, reduced by excess of 10 percent of QBAI over CFC interest expense. GILTI is calculated as a single pool of all non-U.S. income for the MNE.
- Only 50 percent of GILTI income is included in the U.S. parent’s tax base, making the effective U.S. statutory tax rate on GILTI 10.5 = 21*0.5 percent.
- FTCs allowed against tentative U.S. GILTI tax are subject to:
  - A 20 percent general haircut (raising effective GILTI rate to 13.125 percent = 0.8*10.5 percent).
  - A second “inclusion percentage” depending on the share of total positive foreign income (including the 10 percent return to QBAI excluded from GILTI base).
- FTCs not allowable against GILTI in the current year cannot be carried forward or back, or used against other foreign source income.
- The TCJA repeals the former corporate alternative minimum tax, introduces a “base erosion and anti-abuse tax” (BEAT) to prevent income-stripping.
  - BEAT applies to groups with at least $500 million in average annual domestic gross receipts and deductible payments to foreign affiliates totaling 3 percent (2 percent for banks and securities dealers) or more of total tax deductions.
  - Exclusions: 100 percent deduction for foreign dividends, certain payments, deductions for GILTI and FDII.
  - Amortization or depreciation of property purchased from a foreign affiliate treated as deductible payment to that affiliate.
  - BEAT alternative tax rate is 10 percent–11 percent in the case of financial companies (5 percent or 6 percent for 2018 only; 12.5 percent as of 2026).
  - No foreign tax credits are allowed against BEAT income.
- Export-oriented patent box: foreign-derived intangible income (FDII) provides a 37.5 percent deduction for foreign-source income related to exploitation of U.S. intangibles, less a 10 percent return to domestic tangible assets, yielding an effective tax rate of 13.125 = 21 * 0.625 percent (this rate will increase to 16.4 percent in 2026).
- New anti-hybrid rule: symmetrical treatment of cross-border financial flows; certain outbound payments will not receive deductions in the U.S. if not taxed in the destination country.
- One-time transition tax on accumulated unrepatriated U.S. MNE foreign earnings (~US$3 trillion as of end 2017): 15.5 percent on earnings held in liquid assets; 8 percent on earnings reinvested in real assets. Liquid investments believed to account for roughly half of the offshore earnings stockpile.

### Complexity and uncertainty of the TCJA
- The TCJA is extraordinarily complex and uncertain:
  - Passed along partisan lines; includes provisions that change or sunset.
  - Speed of drafting leaves application and interaction of provisions subject to significant debate.
  - IRS regulation issuance and MNE analysis will take time.
- FDII viewed by some U.S. tax experts as the most unstable provision due to potential WTO treaty conflicts; may be unlikely to induce MNEs to repatriate IP currently offshore solely to take advantage of FDII.
- WTO disputes usually take many years; if FDII overturned, U.S. government may not “claw back” past MNE tax benefits.
- Questions exist whether FDII breaches international rules on harmful tax competition as agreed at the OECD forum on Harmful Tax Competition.

### Implications for Ireland and observations
- Irish fiscal authorities expect the U.S. transition tax to have little effect on Irish CIT revenues.
- Moody’s (2018) reports Ireland is the largest single location for U.S. MNEs’ unrepatriated earnings, with roughly US$670 billion or 20 percent of the total.
- Department of Finance officials state that profits accumulated primarily in Irish registered companies that are not tax resident in Ireland, and invested by those non-Irish resident companies in non-Irish source assets, are not part of the Irish tax base.
- Breakdown of Irish assets into liquid and real investments is unknown; overall shares thought to be roughly equal.
- Ireland’s 12.5 percent CIT rate on trading income appears well-positioned with respect to GILTI (focal GILTI rate 13.125 percent).

### GILTI mechanics, example, and interactions
- Illustration example (as given):
  - Aggregate CFC results: $1,000 positive foreign income from CFC 1; $100 foreign losses from CFC 2; QBAI of CFC 1 of $1000; no CFC interest expense; foreign taxes of $125.
  - GILTI income = $800 = $1000 - $100 - $1000*0.1.
  - Inclusion percentage = 800/1,000 = 80 percent.
  - Tentative U.S. tax liability = 0.21*800/2 = $84.
  - FTC offset = $125*0.8*[800/1000] = $80; net U.S. liability = $4.
  - Total taxes paid = $129 or 14.33 percent of net profits.
- Domestic interest allocation effect (example):
  - If $100 domestic interest expense allocated to foreign income, FTC limited to ($400–$100)/$400=75 percent of tentative U.S. GILTI tax liability.
  - FTC usable = $84*0.75 = $63; $17 of FTC cannot be used.
  - New net GILTI liability = 0.21*($400) – $63 = $21.
- GILTI may create incentive to locate low-yielding tangible assets offshore to reduce GILTI liability when foreign tax rates are very low; however:
  - Inclusion of the 10 percent return on QBAI in denominator of FTC inclusion percentage dampens this incentive when foreign effective tax rates are near the Irish statutory rate or higher.
  - U.S. expensing creates zero-rate EMTR on marginal investments in the U.S., counterbalancing incentives to locate tangible assets offshore.
- Interaction with BEAT:
  - MNEs subject to BEAT may lose full value of FTCs against GILTI income because BEAT does not permit FTCs, risking double taxation on non-U.S. income.
  - BEAT penalizes related-party deductible payments (except COGS and certain zero-markup services), incentivizing firms to minimize related-party payments outside of COGS and to stay below BEAT thresholds.

### Strategic implications for tax competition and corporate location decisions
- GILTI, as a worldwide tax without deferral, should allow foreign countries to set CIT rates near the GILTI rate without deterring foreign investment by U.S. MNEs.
- Historically, when U.S. system allowed indefinite deferral, countries had incentives to lower CIT rates to attract investment; elimination of deferral restores incentive for foreign countries to set tax rates at or just below 13.125 percent.
- Pooling of worldwide income and FTCs under GILTI preserves incentive for U.S. MNEs with investments in high-tax jurisdictions to “blend down” their foreign tax rate by investing in low-tax jurisdictions.
- From a corporate CFO perspective, the GILTI rate functions as both a practical ceiling and floor: foreign tax credits in excess of 13.125 percent of foreign income cannot be used or carried forward.
- Governments may still offer tax rates below the U.S. GILTI rate to attract additional investment.
- While GILTI reduces incentives for corporate “inversions,” the U.S. international tax system remains worldwide; inversions could resurface if GILTI rate is raised or if foreign tax rates continue to decline. The TCJA includes measures designed to limit inversions; their impact is not yet tested.

*Source: IMF staff (chapter text).*

### 30. Irish tax advisors and U.S. MNE representatives reported that they have only just

### 30. Irish tax advisors and U.S. MNE representatives reported that they have only just

### Response to the U.S. Tax Cuts and Jobs Act (TCJA)
- Irish tax advisors and U.S. MNE representatives have "only just begun to respond to TCJA."
- First priorities:
  - Calculate liability under the one-time transition tax.
  - Make necessary adjustments of first-quarter 2018 financial statements.
- No major decisions yet on revision of global structures; future revisions will depend on each MNE’s particular configuration and future growth opportunities.
- Tax advisors reported that international tax system changes due to OECD/G-20 BEPS initiatives are "at least as important to European operations as the U.S. tax reform."

### OECD/G-20 BEPS — overview and Irish implementation
- BEPS project initiated in 2013; objective: identify practices contributing to base erosion and profit-shifting, recommend legal and administrative measures, and encourage adoption at national and regional levels.
- BEPS reforms aim at tightening rules of the existing tax system while retaining "separate accounting and arm’s length pricing."
- The BEPS project consists of 15 “actions,” divided into minimum standards and best practices.
- EU, including Ireland, is in the vanguard of implementation.
- Actions 2, 3, and 4 are embodied in the ATAD.
- The Coffey Report (Coffey, 2017) recommended transposing BEPS commitments into Irish law, with detailed recommendations on transparency (Chapter 5), transfer pricing (Chapter 6), and anti-abuse issues including hybrid mismatch arrangements and controlled foreign corporation (CFC) provisions (Chapter 7).
- Ireland is considering adopting the updated 2017 OECD transfer pricing guidelines and broadening application of transfer pricing.
- Ireland has implemented the four BEPS “minimum standards”:
  - Action 5: exchange of information
  - Action 6: anti-treaty shopping
  - Action 13: country-by-country reporting and transfer pricing documentation
  - Action 14: mutual agreement procedures
  - Peer reviews for all four have been passed or scheduled.
- "Substance" is central: location of income determined by observable commercial undertakings (headcount, managerial responsibilities, tangible investments, commercial risk assumption and management), not pure financial risks divorced from other functions.
- IP-specific functions summarized as DEMPE: development, enhancement, maintenance, protection, and exploitation.
- Work on hard to value intangibles ongoing; anticipated resolution of open transfer pricing issues "by 2020."
- Ireland signed the Multilateral Instrument (MLI, BEPS Action 15) in 2017; expects to fully ratify and implement it "within the next two years." The U.S. has not signed the MLI.
- Ireland has reserved on Article 10; reservations on Article 11 and Article 12 noted; expected adoption of the "principal purpose test" for treaty shopping.
- "Substantially all" BEPS actions expected to be in force in respect of Ireland "by the end of 2020."
- Grandfathered IP conduit entities will either expire or be nullified by ongoing multilateral initiatives, including ATAD.
- Implications for Irish CIT revenues:
  - Unclear but "potentially positive."
  - MNEs may relocate some IP assets to jurisdictions with significant R&D functions; those assets are not currently part of the Irish tax base, so this "in itself should not undermine Irish revenues."
  - If MNEs have invested real assets in Ireland to support profit- or IP-shifting into Ireland, that incentive "may nonetheless diminish."
  - Conversely, MNEs may choose to locate more productive assets in Ireland, particularly related to R&D, to support retaining a larger share of IP in a relatively low-tax jurisdiction — potentially benefiting Irish CIT revenues and overall productivity.

### EU Anti-Tax Avoidance Directives (ATAD I and II)
- ATAD I (2016) and ATAD II (2017) implement several BEPS Actions and are cornerstones of the European Commission’s Anti-Tax Avoidance Project.
- As European Council Directives, ATAD requires each Member State to amend internal law to comply within specified time frames.
- ATAD addresses tax avoidance along five margins (detailed below).

#### Interest Limitation Rules (Article 4 of ATAD I)
- Article 4 requires limiting deductible "exceeding borrowing costs" (net interest expense) to 30 percent of EBITDA.
- ATAD targets disproportionate debt capitalization of one member of a financial accounting consolidated group; application depends on options exercised by member states.
- Simplified cap application options:
  - (i) “exceeding” interest costs paid by one member to another member in a presumptively lower-taxed jurisdiction, or
  - (ii) third-party exceeding interest costs of one member used to finance an equity investment in a low-taxed affiliate.
- Article 4 implementation deadline: in general by January 1, 2019; member states with adequate existing rules may defer until 2024.
- Irish authorities maintain that Ireland’s rules are “equally effective” to Article 4, allowing Ireland to defer until January 1, 2024.
- Given Ireland’s low statutory CIT rates, MNEs are "not likely to have engaged in widespread interest stripping out of Ireland."

#### Controlled Foreign Company (CFC) Rules (Articles 7 and 8 of ATAD I)
- Articles 7 and 8 require adoption of CFC rules (BEPS Action 3) by January 1, 2019.
- ATAD definition of a CFC:
  - (1) Entity: more than 50 percent of voting power or capital is owned, directly or indirectly, by a taxpayer in a Member State (the "controlling shareholder");
  - (2) The entity or PE is not subject to tax in the jurisdiction of the controlling shareholder;
  - (3) Actual CIT paid by the entity or PE is less than 50 percent of the CIT that would have been paid if fully subject to tax under the controlling shareholder’s CIT.
- Member states may choose between two CFC regimes:
  - Regime A: resembles U.S. subpart F; requires inclusion of passive or mobile income (interest, dividends, royalties from IP) unless CFC "carries on a substantive economic activity supported by staff, equipment, assets and premises." Member states may elect not to apply safe harbor to CFCs resident in non-EEA countries. Option to exclude entity if one-third or less of its income falls into the proscribed list.
  - Regime B: subjective purpose test; includes income arising from "non-genuine arrangements... for the essential purpose of obtaining a tax advantage." Income inclusion limited to an arm’s length calculation linked to significant people functions performed by the controlling shareholder.
- Ireland currently does not have a CFC regime and will need to adopt one before 2019.
- Most Irish operating companies likely not captured because CIT rate should not be lower than 6.25 percent (as measured under Irish tax accounting principles).
- Rules will be relevant to pure tax-haven subsidiaries and financing vehicles used to strip income.
- Irish authorities expected to publish a proposal in late summer, aiming to adopt legislation by the end of 2018.

#### Hybrid Mismatch Arrangements
- ATAD I and II provide complementary rules for hybrid mismatch arrangements.
- Article 9 of ATAD I contained two simple hybrid mismatch rules for transactions between member states; ATAD II revised provisions with more complex measures applying to transactions involving at least one EU member state and to "reverse hybrids."
- Revised provisions must be transposed by the end of 2019; reverse hybrid rule can be deferred until the end of 2021.
- Hybrids and reverse hybrids are described as "creatures of aggressive tax planning."
- ATAD anti-hybrid rules, together with U.S. Treasury actions under TCJA, expected to substantially reduce the impact of hybrid schemes globally.
- ATAD anti-hybrid rules are not expected to have a material impact on domestic operations in Ireland, where hybrids are "reportedly uncommon."
- Hybrids feature heavily in U.S. tax structures involving Irish operations, particularly due to reliance on "check the box" planning.

#### Exit Tax (Article 5 of ATAD I)
- Article 5 requires member states to adopt an "exit tax" applying to transfers of assets in or out of the tax base; must take effect by January 1, 2020.
- Exit tax base: fair market value of transferred assets less their tax cost; rate left to member states.
- Ireland has an exit tax in place but it is "not as broad as required by ATAD" and must be amended.
- Design of exit tax important to protect long-term Irish CIT revenues.
- Historical practice: For IP on-shored in 2015–17, Irish CIT allowed capital allowances and interest expense on debt incurred to purchase such IP to offset 100 percent of revenues generated by that IP. For IP on-shored from 2018 onwards, these expenses limited to 80 percent of IP revenue.
- When capital allowance is exhausted, firms will face Irish CIT at the 12.5 percent rate on large amounts of unsheltered income.
- Robust exit taxes will preserve for Ireland the benefit of collecting taxes on this income, or an economic substitute in the form of an exit tax.
- Current rate of the Irish exit tax is the 33 percent capital gains tax rate.
- Corporate taxpayers object to exit tax being levied at a rate "more than double the general CIT rate," raising the issue whether corporate capital gains tax rate should be modified to mirror the CIT rate on trading income.

#### General Anti‑Abuse Rule (GAAR) (Article 6 of ATAD I)
- Article 6 requires member states to adopt a GAAR by January 1, 2019.
- GAAR permits or requires ignoring arrangements "which, having been put into place for the main purpose or one of the main purposes of obtaining a tax advantage that defeats the object or purpose of the applicable tax law, [is] not genuine having regard to all the facts and circumstances."
- Ireland has an existing GAAR considered "largely in compliance" with ATAD standards; Article 6 should have little direct effect on Irish CIT administration.

### Digital Taxation
- Taxing digital companies with substantial presence in a country’s online market but minimal physical presence is an increasingly salient problem.
- OECD and EU efforts aim to reallocate some profits of digital MNEs from the location of IP ownership to markets where services are consumed, user data generated, and/or advertising sold — potentially shrinking the Irish corporate tax base by generating deductible or creditable foreign taxes for digital MNEs reporting substantial profits in Ireland.
- BEPS Actions 1 and 7: OECD project focusing on new criteria for digital permanent establishments (PEs). A recent OECD report (OECD 2018) proposed potential definitions; OECD aims for consensus "by 2020."
- European Commission draft directive emphasizes "user participation" as a component of value added for certain digital service companies (e.g., social networks).
- EC provisional interim measure: proposed EU-wide 3–percent "equalization tax" on gross revenues from services where principal value arises from "user participation" — applies to online advertising sales, user data sales, and digital interaction platforms; tax due in the jurisdiction where the user is located and would apply to companies that exceeded certain revenue levels.
- EC initiatives introduce principles not easily reconciled with traditional arm’s length formulations; "user participation" effectively introduces a new "factor of production" distinct from labor and capital.
- Adoption of the EU digital tax directive would require unanimity and appears unlikely; several countries within and outside the EU have moved forward with unilateral measures.
- Irish tax treatment of foreign digital taxes will depend on their nature:
  - Indirect taxes typically allowed as a deduction.
  - Taxes on profits typically allowed as a credit, provided there is a foreign PE.
  - If digital levies resemble a sales tax, they may conflict with the EU VAT directive.

### Common Consolidated Corporate Tax Base (CCCTB)
- The CCCTB was first proposed in 2011 and rejuvenated in 2016 with two draft directives:
  - One on the Common Corporate Tax Base (defines calculation of taxable income).
  - One on the Common Consolidated Corporate Tax Base (defines consolidation and apportionment among Member States).
- If adopted, the CCCTB would be mandatory for all companies with a minimum level of revenues and would apply to all EU income up to the "water’s edge"; Member States would remain free to determine their own CIT rates.

*Source: cr18195 - 30. Irish tax advisors and U.S. MNE representatives reported that they have only just.*

### 57.      The CCCTB is a type of formulary apportionment (FA), according to which

### 57.      The CCCTB is a type of formulary apportionment (FA), according to which

### CCCTB formula and rationale
- The CCCTB allocates corporate income using the formula:
  - 1/3 * Sales + 1/3 * Assets + 1/3 * (1/2 * Payroll + 1/2 * Employment)
- Advocates argue FA would deliver simplification, transparency and fairness relative to separate accounting and arm’s-length pricing.
- Compared to financial accounts, FA factors such as employees, tangible assets and third-party sales are argued to be harder to manipulate.

### Empirical and institutional claims cited
- The European Commission asserts that 70 percent of corporate profits within the EU are profit-shifted for tax purposes.
- The EC estimates that adoption of a CCCTB with a deduction for both debt and equity financing would lead to increased GDP growth of up to 1.2 percent, while compliance costs would fall substantially.
- As an EU tax measure, the CCCTB would have to be passed by unanimous vote of the EU Council, giving every member state potential veto power.

### Implications for Ireland — risks and exposure
- Ireland’s share of productive factors and third-party sales tends to be lower than its share of profits; as a small state where low effective tax rates have led to significant in-shifting of corporate profits, Ireland stands to lose substantial revenue from a move to formulary apportionment, even if it is limited to EU profits.
- Digital taxation and FA have serious negative implications for Ireland’s CIT revenues:
  - As a relatively small market, Ireland would lose from any reallocation of taxing rights based more on consumer or user location.
  - As a low-tax country that has benefited from profit-shifting, Ireland would lose from a system emphasizing the location of real factors.
- Although changes in EU tax directives require unanimous approval and currently appear unlikely, unilateral digital taxes being introduced by individual countries could reduce Ireland’s tax base, and OECD work on defining a digital PE may have a similar effect over a longer horizon.

### Conclusions and policy implications (sections 59–64)
- U.S. tax reform effects:
  - The U.S. reform has reduced headline rates and increased tax allowances, making the U.S. a more competitive location for future investment; over the medium term this may cause moderate erosion of Ireland’s inbound FDI stock and CIT revenues, more likely from reduced future investment than from repatriation of existing intangible assets.
  - The U.S. reform has an uncertain future due to many sunsets, lack of bipartisan support, and fiscal cost; the future of the patent box, FDII, is particularly uncertain due to possible violation of WTO rules.
- Ireland and GILTI:
  - U.S. MNEs will be taxed immediately at roughly 13.125 percent on their worldwide earnings under GILTI.
  - Foreign countries should be able to impose taxes of up to that level without deterring foreign investment.
  - Ireland’s statutory CIT rate on trading income, at 12.5 percent, thus appears well-positioned.
  - Nonetheless, U.S. MNEs operating in high-tax jurisdictions will want to blend down average tax rates, so low-tax jurisdictions still have incentives to offer rates significantly below the GILTI rate.
- BEPS and ATAD:
  - Most OECD/G-20 BEPS measures, including the ATAD, will be put in place from 2021, and together with expiry of existing Irish IP conduit structures should constrain the ability of MNEs to shift profits across borders.
  - BEPS’s new emphasis on substance—location of real factors of production, particularly human capital—could benefit Ireland, which should capitalize on opportunities to develop human capital and grow its share in global value chains to support future CIT revenues.
- Fiscal and tax policy options to compensate potential CIT revenue declines:
  - Ireland’s ratio of total tax revenues-to-GDP is 22.3 percent, well below the OECD European average of 37 percent.
  - Ireland’s ratio of CIT revenue-to-GDP is roughly average, but CIT constitutes a much larger share of tax revenues than in most European countries.
  - To compensate any decline in CIT revenues, recommended options include:
    - Broadening the VAT base, in particular by eliminating domestic zero-rating and taxing energy at the full rate (noting Ireland’s VAT efficiency is quite low).
    - Broadening the base of the recurrent property tax, which is characterized as both efficient and progressive.
    - Streamlining the personal income tax (PIT) to reduce high marginal rates at relatively low incomes and to make Ireland more attractive to skilled immigration needed to support its future as a European IP hub.

### Annex I — Tax-minimizing structures used by MNEs (selected examples)
- Double-Irish Dutch Sandwich:
  - Prior to 2015, Ireland’s tax residence depended on effective management rather than incorporation; an MNE could establish two Irish subsidiaries (an operating company both incorporated and tax-resident in Ireland, and an IP holding company incorporated in Ireland but tax-resident elsewhere with low or no CIT).
  - Operating company paid royalties to a related Dutch company, which distributed payments to the “Irish” IP holding company; Dutch income tax applied only to a small spread and no Dutch withholding tax applied on royalties in many cases.
  - From a U.S. viewpoint the income appeared to be in Ireland via “check the box” elections.
  - The 2015 change in Ireland’s definition of tax residence precluded these structures; existing structures were grandfathered through the end of 2020.
- Single Malt:
  - Royalties paid from an Irish operating company to an Irish-incorporated but Maltese tax-resident company managed and controlled in Malta.
  - No withholding tax on that payment; Malta does not tax income earned entirely outside Malta despite a 35 percent CIT rate.
  - The company is deemed Maltese-resident for the purposes of the Ireland-Malta tax treaty and thus treated as Maltese-resident for Irish domestic law despite Irish incorporation.
- IP on-shoring:
  - Irish resident company purchases high-valued existing IP from an offshore affiliate using accumulated cash and/or debt; Irish law permits amortization of the purchase price over 15 years or at the rate used in financial accounts.
  - The Irish company can claim capital allowances based on the full purchase price and related interest expense against income from licensing or product sales; from a U.S. perspective the purchase may be a transaction within a single corporation with no U.S. tax consequences.
  - On-shored IP income is ring-fenced so losses cannot offset other trading income.
  - Up to 2015, IP capital allowances and interest on acquisition debt were limited to 80 percent of IP income; as of 2015 this limit was abolished.
  - In accordance with a recommendation of Coffey (2017), the 80 percent restriction was re-introduced in 2018; however, IP on-shored in the 2015–17 period is grandfathered with respect to this rule.
  - This on-shoring strategy contributed to the 25.5 percent surge in Irish real GDP in 2015.

*Source: cr18195 - 57.      The CCCTB is a type of formulary apportionment (FA), according to which*

### 4.      The strong real estate price co-movement across countries points to the existence

### 4.      The strong real estate price co-movement across countries points to the existence

### Drivers and mechanisms of international CRE synchronization
- Underlying common driving factors include synchronization of property market fundamentals such as income and interest rates, and enhanced financial integration resulting in synchronized borrowing conditions.
- Currency unions, shared monetary policy, and strong trade and financial market integration can amplify synchronization.
- Expectations about global economic and financial conditions can synchronize business cycles and CRE returns.
- Presence of international lenders and investors (e.g. equity firms, real estate investment funds (REITs), and institutional investors) that operate across markets may contribute to cross-country synchronization.
- Global confidence effects and changes in investors’ risk aversion can increase co-movement in property price dynamics.

### Policy implications of CRE price synchronization
- Coordinated movements in CRE prices imply that country-specific or global shocks can transmit quickly across countries via direct and indirect channels.
- Rapid transmission could inflict widespread bouts of economic and financial instability.
- Historical evidence: large house price swings were associated with periods of financial instability across many countries, especially when accompanied by pullback of global investors and capital flight (Claessens et al. 2008, 2011).
- Understanding drivers of synchronization is important to enhance resilience of the financial sector and the economy against shocks.

### Chapter objectives
- Examine the extent to which returns on Irish CRE properties co-move with those in peers, and whether this synchronization has changed over time.
- Investigate the role of global factors and Ireland’s bilateral linkages with peers in driving synchronization of Irish CRE returns with other countries.
- Assess the sensitivity of Irish CRE prices to international CRE prices, and the linkage between the CRE market and the broader domestic economy through a VAR analysis.

### CRE investment and financing: stylized facts
- Domestic banks’ CRE exposure declined from nearly 30 percent of total loans in 2009 to about 10 percent of total loans in 2017q4.
- Domestic banks’ deleveraging included transfer of large CRE loans to the National Asset Management Agency (NAMA).
- Impaired loans on bank balance sheets dropped significantly but remained elevated at about a quarter of total CRE loans.
- Foreign investors’ activity increased materially in recent years:
  - Total foreign investment in CRE in 2011–17 amounted to about €10 billion.
  - Of that, €5.5 billion is related to direct investment by U.S.-based investors.
  - Investment by U.K. investors represented 8 percent of foreign investment over this period.
  - Investment by investors from other European countries represented 23 percent of foreign investment over this period.
  - Direct investment from other regions such as Asia, Africa, and Canada remained rather small.
- Investment by Irish investors recovered since 2013 but averaged €1.6 billion annually in 2013–17, well below pre-crisis levels.
- Investor composition:
  - Irish REITs entered the market in 2013, acquiring sizeable portfolios but have been less involved in new investments more recently.
  - Institutional investors (pension and insurance funds from the U.S., the U.K., and Germany) increased their role.
  - Other private investors (investment funds, property companies) were also active.
- Market activity and supply constraints:
  - CRE investment moderated to about €2.5 billion in 2017 from about €3.5–4 billion a year in 2014–16, reflecting shrinking supply of offices.
  - Dublin office vacancy rate fell to a record-low of 6 percent at end-2017 from about 24 percent in 2010.
  - Take up reached the highest level in the last 15 years.
  - CBRE Research projects Dublin office supply to increase by an annual average of 220,000 square meters in 2018–20, exceeding the average annual take up of about 200,000 square meters in 2010–16.
  - Post-Brexit relocation decisions may exacerbate demand pressure in prime Dublin locations.

### Measuring CRE synchronization: methodology and data
- Two synchronization indicators used:
  - Synch1 (level): negative of the absolute difference between Ireland’s and country j’s total returns (TR); Synch1 = −|TR_irl,t − TR_j,t|. By construction, values are non-positive; less negative → higher correlation.
  - Synch2 (cycle): instantaneous quasi-correlation: Synch2_irl,j,t = ((TR_irl,t − TR̄_irl)(TR_j,t − TR̄_j)) / (σ_TR_irl σ_TR_j). Positive values → both countries simultaneously above or below long-term averages → higher synchronization; negative → low synchronization.
- Data: annual MSCI total return series for 2000–16 covering all property types across 27 countries; CRE total return measured in euro terms to ensure comparability.
- Countries in sample include: Australia, Austria, Belgium, Canada, Switzerland, China, Czech Republic, Germany, Denmark, Spain, Finland, France, Great Britain, Hong Kong, Hungary, Ireland, Italy, Japan, Korea, the Netherlands, New Zealand, Poland, Portugal, Singapore, Sweden, Taiwan, and the USA. For some countries, data available only for a shorter period.

### Summary statistics and temporal patterns (Table 1 and Figures)
- Sample divided into three sub-periods: 2000–07 (pre-crisis), 2008–11 (crisis), 2012–16 (recovery).
- Total return volatility across countries:
  - Range examples: 40 percent (Canada) in run up to the crisis; about -40 percent (Great Britain) during the financial crisis.
  - Ireland’s volatility was the highest in the sample due to the boom-bust episode in 2005–11 and strong recovery in recent years.
- Table 1: Summary statistics (2000–07, 2008–11, 2012–16)
  - Averages:
    - Total Return: 10.5 (2000–07), 4.1 (2008–11), 8.8 (2012–16)
    - Synch1: -9.2 (2000–07), -18.2 (2008–11), -11.5 (2012–16)
    - Synch2: 0.2 (2000–07), 1.1 (2008–11), 0.1 (2012–16)
  - Minimums:
    - Total Return: -10.5 (2000–07), -40.6 (2008–11), -10.3 (2012–16)
    - Synch1: -31.3 (2000–07), -54.2 (2008–11), -34.9 (2012–16)
    - Synch2: -1.3 (2000–07), -3.8 (2008–11), -1.2 (2012–16)
  - Maximums:
    - Total Return: 39.8 (2000–07), 33.3 (2008–11), 36.1 (2012–16)
    - Synch1: -0.2 (2000–07), -0.3 (2008–11), -0.3 (2012–16)
    - Synch2: 2.2 (2000–07), 8.4 (2008–11), 1.4 (2012–16)
  - Standard deviations:
    - Total Return: 8.4 (2000–07), 14.2 (2008–11), 7.2 (2012–16)
    - Synch1: 7.5 (2000–07), 13.2 (2008–11), 9.6 (2012–16)
    - Synch2: 0.6 (2000–07), 2.3 (2008–11), 0.5 (2012–16)
- Interpretation:
  - On average, Ireland’s degree of synchronization declined slightly in 2012–16 compared to the pre-crisis period.
  - During the crisis (2008–11), Synch1 declined while Synch2 increased: downturn in Irish CRE returns coincided with peers (cycle synchronization) but was much more severe in Ireland (level divergence).

### Patterns of co-movement across regions and countries
- Ireland’s synchronization is relatively high with the U.S., Great Britain, and Asian countries for both measures compared to other countries in the sample.
- Synch1 shows that, apart from the U.S., degree of synchronization declined in recent years relative to pre-crisis levels.
- Synch2 shows increased synchronization with the U.S., Great Britain, and Asian countries over time.
- Cross-country comparison:
  - Ireland’s cyclical synchronization (Synch2) has been one of the highest in the sample—reflecting high openness and presence of foreign-owned companies.
  - Ireland’s synchronization of returns’ level (Synch1) is low—reflecting high volatility of Irish CRE returns over the sample period.
  - Being part of a currency union (euro area) may contribute to higher cyclical co-movement.

### Determinants of synchronization: empirical specification
- Estimated specification:
  - Synch_irl,j,t = α_irl,j + β BILATERAL_irl,j,t + γ GLOBAL_t + ε_t
  - Synch_irl,j,t: co-movement (Synch1 or Synch2).
  - BILATERAL variables include:
    - FINANCIAL: country j's portfolio investment in Ireland as a share of Ireland’s GDP (proxy for financial linkages).
    - TRADE: bilateral trade (imports plus exports) as a share of Ireland’s GDP.
    - ABS_SPREAD: government long-term yield differentials in absolute value.
    - OUTPUTGAP_SYNCH: synchronization of business cycles measured by co-movement of output gaps.
  - GLOBAL_t includes:
    - VIX (risk aversion).
    - Global liquidity gap: deviation of bank credit to the nonfinancial private sector (as a share of global GDP) from its trend.
  - Dummy variables: Crisis (global financial crisis 2008–09), EA (Euro area members).
  - Estimation: two-stage least squares (instrumental variables) with country and time fixed effects to address endogeneity.
  - Random effects reported for Synch1 as Hausman tests validate consistency.

### Key estimation results and findings
- Global factors:
  - Higher global liquidity (associated with low interest rates and search for yield) tends to increase degree of synchronization.
  - Higher risk aversion (VIX) tends to reduce synchronization.
  - During the global financial crisis, synchronicity of CRE returns declined significantly on average.
- Bilateral linkages:
  - Higher business cycle synchronization (OUTPUTGAP_SYNCH) contributes to higher CRE return synchronicity.
  - Higher financial interconnectedness (FINANCIAL: portfolio investment in Ireland as a share of GDP) contributes to higher CRE return synchronicity.
  - Higher correlation of long-term yields (ABS_SPREAD) contributes to higher synchronization—possibly capturing cost of capital effects.
  - Ireland exhibits higher synchronicity with countries within the monetary union (EA), reflecting absence of exchange rate risk for euro-area investors.
  - Bilateral trade (TRADE) was not found to have a significant impact on synchronicity of CRE returns.

_Italic: International Monetary Fund staff analysis (chapter content)._

### 20.      The estimation results for the cyclical synchronization measure (Synch2), which are

### 20.      The estimation results for the cyclical synchronization measure (Synch2), which are

### Estimation results for synchronization (Synch2)
- The Annex (Table A1) shows lower explanatory power for the cyclical synchronization measure (Synch2) compared with SYNCH1, but confirms impacts of bilateral linkages such as business cycle synchronization, correlation of long-term yields, and financial links.
- Selected coefficient signs and significance from Table A1 (Instrumental variable estimation, two-stage least square (GLS2LS), Random Effects 1995–2016; Dependent Variable: SYNCH2):
  - VIX: 0.021; 0.041; 0.019; 0.056; 0.056 (not significant at conventional levels in reported columns).
  - GLOBAL LIQUIDITY GAP: 0.119; 0.052; -0.253; 0.095; 0.097 (not significant at conventional levels in reported columns).
  - CRISIS: 1.309*; 1.506*; 2.716**; 1.089; 1.134 (significant at 10 percent in some specs, 5 percent in one spec).
  - EURO AREA: 0.208*; -0.008; 0.184; 1.095* (some specifications show significance at 10 percent).
  - ABS_GOV YIELD SPREAD: -0.275* (significant at 10 percent in one specification).
  - OUTPUTGAP_SYNCH: 0.442*** (significant at 1 percent in one specification).
  - FINANCIAL: 0.137* (significant at 10 percent in one specification).
  - Constant terms: 0.384; 0.269; 0.447; -0.947; -1.349.
- Goodness-of-fit and sample details from Table A1:
  - R2 (overall): 0.341; 0.342; 0.258; 0.280; 0.321.
  - # obs.: 377; 363; 365; 325; 320.
  - Hausman test p-value (H0: RE H1: FE): 0.177; 0.378; 0.990; 0.999; 0.999.
  - # countries: 26; 26; 26; 25; 25.
- Notes on significance notation: *** indicates significance level of 1 percent; ** indicates significant level of 5 percent; * indicates significance level of 10 percent.

### VAR analysis: sensitivity of Irish CRE prices to international CRE prices
- Model and data:
  - Reduced-form Vector Auto Regression (VAR) including four variables: GDP growth in advanced economies (ADVANCED_GDP), real growth of CRE prices in advanced economies with strong financial linkages to Ireland (INT_CRE), real growth of Ireland’s CRE prices (IRL_CRE), and Ireland’s modified final domestic demand (IRL_MFDD).
  - PCA-derived common international CRE factor based on quarterly CRE capital growth for GBR, USA, and Euro area for 2002Q1–2017Q3.
  - VAR estimated over 2002q1–2017q3 using two lags of endogenous variables; Cholesky decomposition ordering: ADVANCED_GDP first, INT_CRE second, IRL_MFDD third, IRL_CRE last.
  - Unit root tests reject the null hypothesis of unit roots for the endogenous variables.
- Principal Component and correlations:
  - The common factor allocates higher weights to CRE capital growth in the U.S. and GBR compared to the EA.
  - Correlation between the common international CRE factor and Irish CRE capital growth: 0.87.
  - The common factor exhibits lower volatility than IRL_CRE, reflecting the Irish boom-bust period.
- Key impulse response results (exact reported magnitudes):
  - A shock of one standard deviation to international CRE prices, equivalent to 0.8 percentage points, results in an increase of up to 3.5 percent y/y in Irish CRE prices by the fifth quarter (in real terms).
  - A shock of one standard deviation to Irish CRE prices, equivalent to about 2 percent in real terms, leads to an increase of 0.7 percent y/y in Ireland’s modified domestic demand by the sixth and seventh quarters.
  - A shock of one standard deviation to modified domestic demand, equivalent to 2.6 percent, results in an increase in Irish CRE prices of up to 5 percent y/y in the fifth quarter.
  - A shock of one standard deviation to advanced economies’ GDP growth, equivalent to 0.4 percent, results in an average increase of 0.5 percent y/y in Ireland’s modified domestic demand in the first year.
- Additional VAR diagnostics and relationships:
  - Ireland’s CRE prices and domestic demand growth exhibit significantly higher volatility than advanced economies’ counterparts.
  - The correlation matrix (Table A3) indicates strong correlations across ADVANCED_GDP, INT_CRE, IRL_CRE, and IRL_MFDD.

### Summary statistics and correlations (Tables A2 and A3)
- Table A2: Summary statistics (ADVANCED_GDP; INT_CRE; IRL_CRE; IRL_MFDD)
  - Mean: 2.153; 0.000; 0.529; 3.949
  - Median: 2.485; 0.227; 3.390; 5.010
  - Maximum: 4.830; 5.484; 27.263; 13.650
  - Minimum: -4.730; -6.243; -42.943; -11.380
  - Std. Dev.: 1.574; 2.421; 16.513; 5.454
- Table A3: Correlations (exact matrix entries)
  - ADVANCED_GDP correlations: with INT_CRE 0.688; with IRL_CRE 0.732; with IRL_MFDD 0.671.
  - INT_CRE correlations: with ADVANCED_GDP 0.688; with IRL_CRE 0.832; with IRL_MFDD 0.657.
  - IRL_CRE correlations: with ADVANCED_GDP 0.732; with INT_CRE 0.832; with IRL_MFDD 0.824.
  - IRL_MFDD correlations: with ADVANCED_GDP 0.671; with INT_CRE 0.657; with IRL_CRE 0.824.

### Key takeaways and policy implications
- Market developments and exposures:
  - Irish CRE prices and rents recovered sharply after the 2008–11 adjustment, supported by strong investment from nonbank Irish institutions (including REITs and institutional funds) and foreign investors largely from the U.S. and Europe.
  - Domestic banking system exposure to the CRE market remains about 10 percent of total loans, and new lending to the sector is picking up despite prior deleveraging.
- Synchronization and transmission:
  - Both synchronization measures indicate a high degree of co-movement of CRE return cycles between Ireland and peers, with Ireland exhibiting much higher volatility.
  - Synchronization of Irish CRE returns is higher with the U.S. and the U.K. than with other countries; synchronization is affected by both global and bilateral factors.
  - VAR results show international CRE prices are a significant channel affecting Irish CRE prices and that CRE–domestic demand feedbacks are strong.
- Monitoring, data, and macroprudential/other policy options:
  - Movements in international CRE prices can serve as early warning indicators for domestic CRE market movements.
  - Close monitoring of the CRE market and closing data gaps (transactions, leases, financing) is important for financial stability.
  - Prudential management of new bank lending to the CRE sector (including to REITs) and existing property exposure is necessary to enhance bank resiliency and prepare for potential provisioning needs.
  - Monitor increases in banks’ risk appetite and possible changes in banks’ business models.
  - Monitor funding from nonbank financial institutions and their exposures to the banking system.
  - In the absence of independent monetary policy and given largely externally-financed CRE activity, there is scope to continue using tax measures (stamp duty, property tax, capital gains tax) to manage demand and dampen sensitivity of Irish CRE prices to external CRE price movements.

*Source: INTERNATIONAL MONETARY FUND (cr18195).*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2018/cr18195.pdf_