## Appendix I — _wp1051

## Source details

**Canonical URL:** [Appendix I — _wp1051](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1051.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1051.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1051.pdf.json)

---

### Introduction and purpose
- Use yield-curve modeling strategies to study macro-finance linkages and cross-border spillovers in the Czech Republic, Hungary, and Poland (Central and Eastern European — CEE — countries).
- Methodological framing:
  - Employ a dynamic Nelson-Siegel (NS) representation of the yield curve with a different estimation strategy and open-economy variables.
  - Use generalized impulse responses of NS parameters (level, slope, curvature) re-aggregated to analyze shifts in the full yield curve after macroeconomic and financial shocks.
  - Characterize spillovers from CEE yield curves and the euro area with impulse responses and variance decompositions to examine global versus domestic drivers.
- Data coverage:
  - Sample period: 2000:01 through May, 2008 (data cut-off end-May, 2008).
  - Zero coupon yields for maturities: 3, 6, 9, 12, 24, 36, 48, 60, 84, 96, and 120 months from Bloomberg.
  - Bloomberg euro-area data: composite of sovereign yields of Spain, Austria, Germany, France, Ireland, Finland, and the Netherlands.

### Model and identification
- Nelson-Siegel representation:
  - y_t(τ) = β1t + β2t [ (1−e^(−τ/λ)) / (τ/λ) ] + β3t [ ( (1−e^(−τ/λ)) / (τ/λ) − e^(−τ/λ) ) ]
  - Interpretation: β1t = level; β2t = slope; β3t = curvature. λ determines exponential decay speed.
  - Slope approximated by y(3) − y(120) ≈ β2t.
- VAR/GVAR and generalized impulse responses:
  - For each country estimate VAR with variables x = [i*, β1, β2, β3, IP, REER, π, ip] and lag polynomial L = 3.
  - Variables (except β'i and interest rates) expressed as 12-month year-on-year changes.
  - Shocks μ ~ N(0, Ω); GIR analysis uses v = σi,i1/2 standardization.
  - Innovation: reconstruct movements in full yield curve from impulse responses of β'i rather than reporting β'i alone.
- Spillover decomposition:
  - Regional principal component for each NS factor across [β_i,CZE, β_i,HUN, β_i,POL].
  - Global principal component across [β_i,DEU, β_i,US, β_i,JPN].
  - GVAR comprising nine CEE NS factors plus euro-area NS factors for generalized impulse responses and variance decompositions.

### Stylized facts and country backgrounds (2000–08)
- Cross-country stylized observations:
  - U.S. example cited: positively sloped yield curve between 5.5 percent at the short end and 7 percent at the long.
  - Euro average upward-sloping yield curve between 3 and 4½ percent.
  - CEE countries observed negatively sloped yield curves for many years (contrast with U.S./euro).
- Country-specific median/regime notes (subsamples January 2000–April 2004 and May 2004–May 2008):
  - Euro median curve: first half upward sloping; second subsample flatter and anchored at long end at 3½–4 percent.
  - Czech Republic median: upward sloping for whole sample; policy rate cuts in 2003 and 2005 contributed to relatively low and slightly upward-sloping curve.
  - Hungary median: downward sloping for whole sample; inverted during most of sample.
  - Poland median: flat for whole sample; steep negatively sloped curve early, un-inverted in 2003, then flat or positively sloped except last months of sample.
- Appendix I country monetary developments (selected exact figures and dates):
  - Czech Republic:
    - CNB repo policy rate cut to just 2 percent in 2002 and 2003.
    - Cut to 1 3/4 percent in April 2005.
    - Spread of around 100 bps between long and short rates (as of end-May, 2008).
  - Hungary:
    - Policy rates reduced from 18 percent in 1998 to 8 percent in 2002.
    - Cuts to 7 percent in first half 2003; later increase by 600 bps to 12.5 percent between June and November 2003.
    - Flattened to around 7 percent in 2005 and 2006; inverted again with tightening in 2007–2008.
  - Poland:
    - Policy rates reduced by around 1400 bps between early 2001 and mid-2003; flattened at around 5.5 percent.
    - Cuts of 250 bps to 4 percent between March 2005 and March 2006.
    - Policy rates increased seven times since April 2007 to 5 ¾ percent.
    - Yield curve showed a small hump in the short-end reflecting expectations of near-term tightening.

### Estimated NS factors — fit, diagnostics, and descriptive statistics
- Estimation approach:
  - Grid search for λ0 to seven decimal places minimizing sum of squared residuals.
  - OLS per month using 11 maturities: 3, 6, 12, 24, 36, 48, 60, 84, 96, 108, 120 months.
  - Time series for β1t, β2t, β3t recovered for 101 months (January, 2000 through May, 2008).
- Fit and pricing errors (selected residuals in basis points; mean; standard deviation):
  - Czech Republic (3 to 120 months): 3: -1.2; 6.9 — 6: 0.3; 3.5 — 12: 0.4; 9.7 — 24: 1.8; 7.0 — 36: 1.0; 5.4 — 48: -0.5; 6.0 — 60: -2.2; 8.2 — 84: -2.6; 8.5 — 96: 0.9; 4.3 — 108: 1.6; 4.3 — 120: 0.5; 6.0.
  - Hungary (3 to 120 months): 3: 1.3; 9.0 — 6: -0.8; 7.1 — 12: -1.5; 9.7 — 24: 0.1; 5.8 — 36: 2.3; 6.1 — 48: -1.2; 7.2 — 60: -1.4; 8.2 — 84: 1.0; 6.0 — 96: 1.7; 4.3 — 108: 0.2; 4.8 — 120: -1.9; 6.0.
  - Poland (3 to 120 months): 3: 3.5; 11.2 — 6: -0.2; 7.8 — 12: -3.8; 16.2 — 24: -5.1; 11.9 — 36: 0.7; 6.5 — 48: 5.4; 9.9 — 60: -0.2; 14.9 — 84: 5.9; 11.2 — 96: 2.8; 6.5 — 108: -1.1; 6.2 — 120: -7.9; 12.8.
- Authors’ fit comments:
  - NS approximation provides a reasonably good fit.
  - Czech Republic mean residuals range from about -2½ basis points to less than 2 basis points across maturities.
  - Poland shows larger pricing errors with overpricing about 8 basis points at 120 months and underpricing close to 6 basis points at 84 months.
  - Most large model errors occurred prior to EU accession (before 2004); Poland’s largest residuals concentrated before 2002.
  - Volatility of pricing errors more pronounced in Poland.
- Descriptive statistics for βi (sample 2000–08; exact table entries preserved):
  - β1 — Mean / Std. Dev. / Median / Maximum / Minimum
    - CZE: Mean 6.08 / Std. Dev. 1.34 / Median 6.35 / Maximum 9.28 / Minimum 3.81
    - HUN: Mean 6.53 / Std. Dev. 0.53 / Median 6.53 / Maximum 7.71 / Minimum 4.99
    - POL: Mean 6.36 / Std. Dev. 1.11 / Median 6.02 / Maximum 9.51 / Minimum 4.73
    - EUR: Mean 4.92 / Std. Dev. 0.70 / Median 5.18 / Maximum 6.06 / Minimum 3.50
  - β2 — Mean / Std. Dev. / Median / Maximum / Minimum
    - CZE: Mean -2.97 / Std. Dev. 1.08 / Median -2.82 / Maximum -1.07 / Minimum -5.14
    - HUN: Mean 2.65 / Std. Dev. 1.88 / Median 2.68 / Maximum 6.62 / Minimum -0.66
    - POL: Mean 2.14 / Std. Dev. 4.82 / Median 0.06 / Maximum 13.40 / Minimum -2.13
    - EUR: Mean -1.75 / Std. Dev. 1.13 / Median -1.88 / Maximum -0.07 / Minimum -3.59
  - β3 — Mean / Std. Dev. / Median / Maximum / Minimum
    - CZE: Mean -1.28 / Std. Dev. 2.21 / Median -1.12 / Maximum 4.38 / Minimum -6.04
    - HUN: Mean 1.83 / Std. Dev. 1.86 / Median 1.58 / Maximum 7.49 / Minimum -1.74
    - POL: Mean 1.98 / Std. Dev. 2.98 / Median 1.36 / Maximum 13.40 / Minimum -2.09
    - EUR: Mean -1.68 / Std. Dev. 1.68 / Median -1.95 / Maximum 1.49 / Minimum -4.98
- Correlations between βi and yield-curve metrics:
  - β1 / y(120):
    - The Czech Republic 0.83
    - Hungary 0.98
    - Poland 0.99
  - β2 / slope and β3 / curvature entries presented in source tables (preserved as provided).

### Impulse responses — summary of key empirical findings
- GIRs for macro shocks on yield curves:
  - Shocks to euro area policy rate:
    - Tend to produce a relatively large hump-shaped upward shift in the yield curve at the three-month horizon.
    - Upward shifts persist at least until after a year, except in Hungary.
  - Shocks to economic activity: negligible effects on the yield curve.
  - Shocks to inflation:
    - Systematically shift up yield curves at the three-month horizon.
    - Czech Republic: parallel upward shift.
    - Hungary and Poland: tilting of the curve.
    - Effects die out after one year except in Hungary where tilt persists longer.
  - Shocks to REER (unexpected strengthening): small but persistent downward shift in entire yield curve, more pronounced in Hungary.
  - Shocks to domestic policy rates:
    - Rapid transmission to short end (liquidity effect).
    - Czech Republic and Hungary: curves shift up in a humped fashion and become inverted only after three months.
    - Poland: entire curve shifts upward in a parallel fashion and effect persists longer.
- GIRs for yield-factor shocks on yield curves:
  - β1 (level) shocks:
    - Upward shift combined with short-run positive tilt.
    - Shift vanishes within a year in Hungary; more persistent in Poland and Czech Republic.
  - β2 (slope) shocks:
    - Produce a "pure" negative tilt (inversion) where long end dips below baseline.
    - In Hungary, short-term rate shocks link to exchange rate moves and do not much affect long-term inflation expectations.
  - β3 (curvature) shocks:
    - Generate hump-shaped impact that vanishes quickly (arbitraged away).
- GIRs for yield shocks on macro variables:
  - Responses of macro variables to β1, β2, β3 shocks are small and measured imprecisely; REER responses estimated with somewhat more precision.
  - β1 shocks:
    - Associated with increase in actual inflation and domestic policy rates in Poland and to a lesser extent Czech Republic; reverse in Hungary.
    - REER: appreciates quickly in Czech Republic; depreciates in Poland and Hungary.
  - β2 shocks:
    - Associated with domestic interest rate hikes and a decrease in economic activity in short run in all CEE countries.
    - REER tends to appreciate where short end increases (Hungary and Poland) and depreciate where short-end effect is smaller (Czech Republic).
  - β3 shocks:
    - Associated with rise in policy rates in all CEE countries; no clear pattern in other macro variables.

### Spillovers, principal components, and variance decompositions
- Principal component findings:
  - Roughly two-thirds of variance in β1 and at least half of variance in β2 and β3 for the CEE countries can be associated with a regional component.
  - Global level factor explains 75 percent of level factor variance in G-3 in the referenced literature.
- Correlations with euro-area/global components:
  - Level factor (β1): Czech Republic and Poland correlations with euro-area/global component are quite high and statistically significant; Hungary’s correlation is substantially lower though it increased after EU accession.
  - Slope (β2) and curvature (β3): Czech Republic correlations with euro/global factors increased since EU accession and are larger than in Hungary and Poland; Hungary and Poland correlations turned negative since joining the EU.
- GVAR / GIR spillover patterns (selected exact findings):
  - Euro-area shocks:
    - Increase in β1_euro: gradually transmitted to CEE countries; steepening on impact in Czech Republic and Poland; effect dies out after three months in euro area but lasts longer in CEE countries.
    - β2_euro flattening/tilting: similar effects in CEE countries; more persistent in later subsample.
    - β3_euro shocks: small effects on CEE yield curves.
  - CEE country shocks:
    - Shock to β1_CZE: tilts Czech curve and affects others; effect on Czech long end vanishes by six months but persists longer in Poland.
    - Shock to β2_CZE: inverts Czech curve and shifts down Poland and Hungary (second subsample).
    - Shock to β1_HUN: shifts up yield curves in Poland and (less) in Czech Republic.
    - Shock to β2_HUN: inverts Hungary curve; small responses in Czech and Poland (shift down after EU accession).
    - Shock to β1_POL and β2_POL: transmitted to other countries especially in first three months — stronger spillovers from Poland to Czech Republic and Hungary.
- Variance decomposition highlights (preserved statements):
  - Global/regional factors rarely account for more than 40 percent of variance of yield curve factors in these CEE countries.
  - For Czech Republic and Poland, euro-area yield curve shocks explain between 20 percent and 30 percent of variance of level and slope of their own curves.
  - For Hungary, regional shocks are more important than euro-area-originated shocks, though euro-area importance increased after EU accession.
  - After accession (2004–08), percentage of variance explained by regional factors increases markedly in Czech Republic and Poland; euro-area factor importance grows in Hungary.

### Event study — Poland’s euro adoption announcement (September 10, 2008)
- Event outline:
  - Announcement: Prime Minister Tusk announced goal of adopting the euro in 2011 on Wednesday September 10, 2008; clarifications sought Commission approval and euro notes in circulation January 1, 2012.
  - Market reaction (September 9–12, 2008): Poland’s yield curve steepened on September 10 consistent with expectations of tighter monetary policy and lower long-run inflation; Czech Republic and Hungary yield curves shifted down (Czech with one-day delay).
- Two-stage analysis:
  1. Extract NS β shocks implied by yield curve moves three days after event using OLS regression decomposition of yield changes onto country impact GIRs (h = 1, 2, 3 days; based on 11 maturities).
  2. Use estimated shocks in VAR/macro-financial models to infer macro implications and spillovers.
- Estimated NS shocks for Poland (selected reported estimates and scenario used in models):
  - Poland-only regression: small (statistically insignificant) reduction in β1; 40 to 50 bps increase in β2; decline of almost 200 bps in β3; these shocks explained about 80 percent of Poland yield curve movement following announcement.
  - Panel regression: qualitatively similar but smaller coefficient estimates (around 20 basis points) and explains far less of movement across three countries.
  - Shock scenario fed into models (Table 8, basis points):
    - μβ1 = -10
    - μβ2 = 50
    - μβ3 = -150
- Macro implications for Poland under scenario:
  - Real exchange rate appreciates over next ten months; activity contracts initially; both reverse over following year.
  - Small increase in inflation.
  - Policy interest rates do not increase much — market short-term rate rise makes policy changes unnecessary in model.
  - Caveat: Lehman Brothers bankruptcy one week later constitutes an additional shock altering predictions.
- Spillovers from Poland announcement:
  - Transmitted to Czech Republic and Hungary via NS β linkages.
  - Hungary experienced similar REER appreciation and inflation; activity improves marginally in Hungary (contrast with Poland).
  - Spillovers to Czech Republic much smaller.

### Conclusions — synthesized findings
- Macro-financial linkages in CEE countries show parallels with mature markets:
  - Increases in policy rates and inflation shift up whole curve.
  - Policy rate shocks lead to eventual inversion of the curve; inflation shocks have limited persistence in CEE.
- Directionality and macro effects:
  - Macro variables have stronger influence on yield curves than vice versa.
  - Yield curve inversion has a negative impact on economic activity in CEE, though weaker than in mature markets.
  - Hungary: short-term interest rate increases associated with exchange rate appreciation and smaller declines in long-term inflation expectations.
- Integration and spillovers:
  - Substantial co-movement between euro-area and CEE yield curves, especially post-EU accession (post-2004).
  - Euro-area shocks explain 20–30 percent of Czech Republic and Poland yield curve variation; smaller share for Hungary that increased after 2004.
  - CEE region became more integrated with euro area and U.S., particularly in convergence of level of yield curves; post-2004 convergence reflects policy-framework convergence in Hungary.
  - Shocks to yield curve in one country can produce measurable spillovers to neighbors (illustrated by Poland’s euro adoption announcement).

*Source: _wp1051 — Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries (authors' calculations; sample 2000:01–2008:05; event-table A2 uses September 10–12, 2009 data).*

### Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries ......45

### Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries ......45

### Appendices (contents and page references)
- Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries ......45
- Appendix II. The Generalized Vector Autoregression ............................................................46
- Appendix III. Estimating the Nelson Siegel Beta ....................................................................47
  - A. Estimating the Nelson-Siegel Representation...........................................47
- Appendix IV. Impulse Responses for the Nelson Siegel Beta.................................................54
- Appendix V. Estimating Shock for Historical Episodes ..........................................................57

### Appendix Tables
- Table A1. Descriptive Statistics of the Yield Curve Residuals ...............................................48
- Table A2. Shock’s Underlying Poland’s Euro Adoption Annoucement .................................57

### Appendix Figures
- Figure F1. CEEC’s: Maximum and Minimum Yield Curve Residuals, 2000–2008............49
- Figure F2. Euro Area and CECs: Level Factor (β1t) and Long-Term Yields.....................50
- Figure F3. Euro Area and CECs: Slope Factor (β2t) and Slope of the Yield Curve............51
- Figure F4. Correlation between Curvature Factor (β3t) and Curvature of Yield Curve........51
- Figure F5. CEEC3: Impulse Response Functions of the Yield Factors, 2000–08..............53
- Figure F6. Response of Nelson Siegel Beta Factors to Shocks to β1t 2000–08.................54
- Figure F7. Response of Nelson Siegel Beta Factors to Shocks to β2t 2000–08.................55
- Figure F8. Response of Nelson Siegel Beta Factors to Shocks to β3t 2000–08.................56

*Source: _wp1051 - Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries ......45*

### References .............................................................................................................

### _wp1051 - References

### Introduction
- Purpose: Use yield-curve modeling strategies to study macro-finance linkages and cross-border spillovers in emerging markets: the Czech Republic, Hungary, and Poland (Central and Eastern European — CEE — countries).
- Methodological framing:
  - Employ a dynamic Nelson-Siegel (NS) representation of the yield curve with a different estimation strategy and open-economy variables.
  - Use generalized impulse responses of NS parameters (level, slope, curvature) re-aggregated to analyze shifts in the full yield curve after macroeconomic and financial shocks.
  - Characterize spillovers from CEE yield curves and the euro area with impulse responses and variance decompositions to examine global versus domestic drivers.

### Background — Literature Review
- Two strands in the macro-finance and yield-curve literature (Diebold, Piazzesi, and Rudebusch, 2005):
  - Affine, no-arbitrage, latent-factor models augmented with macro variables (examples: Piazzesi (2005); Ang and Piazzesi (2003); Rudebusch and Wu (2004); Dewachter and Lyrio (2006)).
  - Nelson-Siegel (1987) representations related to macro variables without no-arbitrage restrictions (examples: Diebold and Li (2006); Diebold, Rudebusch, and Aruoba (2006); Evans and Marshall (2007)).
- Key empirical observations from literature:
  - Up to 85 percent of the forecast variance at short-and medium-term maturities can be explained by macro factors in some studies (Ang and Piazzesi (2003)).
  - Inflation expectations determine the level of the yield curve; cyclical factors and monetary responses determine the slope (Rudebusch and Wu (2004); Dewachter and Lyrio (2006)).
  - Global factors extracted from major-country term structures explain more than half the variance of yields in some studies, affecting level more than slope (Diebold and others (2007b)).
- Rationale for studying CEE:
  - Since the mid-1990s, relatively large and liquid fixed-income markets with strong participation from institutional investors.
  - Yield curves in CEE have shown substantial volatility due to convergence plays and macro management challenges around EU accession in May 2004 and euro adoption prospects.
  - Data cutoff for empirical analysis: end-May 2008. Yield data for study cover the period 2000-08 and are based on zero coupon yields for government bonds with maturities of 3, 6, 9, 12, 24, 36, 48, 60, 84, 96, and 120 months from Bloomberg. Bloomberg euro-area data are a composite of sovereign yields of Spain, Austria, Germany, France, Ireland, Finland, and the Netherlands.

### Background — Stylized Facts on CEE Countries
- Comparative stylized facts:
  - U.S. (over last 20 years cited): positively sloped yield curve between 5.5 percent at the short end and 7 percent at the long (Diebold and Li (2006)).
  - Euro: average upward-sloping yield curve between 3 and 4½ percent (Lemke (2008)).
  - CEE countries: observed negatively sloped yield curves for many years (contrast with U.S./euro).
- Sample medians and regime changes (2000–08; subsamples January 2000-April 2004 and May 2004-May 2008):
  - Euro median curve: upward sloping in first half of sample; second subsample became flatter and anchored at the long end at a lower level of 3½–4 percent.
  - Czech Republic median: upward sloping for whole sample; low inflation and appreciating koruna contributed to relatively low and slightly upward-sloping curve after policy rate cuts in 2003 and 2005.
  - Hungary median: downward sloping for whole sample; yield curve remained inverted during most of sample with brief flat periods in 2003, 2005, and 2007.
  - Poland median: flat for whole sample; steep, negatively sloped curve at beginning of century un-inverted in 2003, then flat or positively sloped except in last few months of sample.
- Stylized distributional facts:
  - Higher volatility in short end relative to long end; asymmetry with relatively fat right tails (25–75 percentiles).

### Modeling the Yield Curve and Spillovers
- Nelson-Siegel representation used:
  - y_t(τ) = β1t + β2t [ (1−e^(−τ/λ)) / (τ/λ) ] + β3t [ ( (1−e^(−τ/λ)) / (τ/λ) − e^(−τ/λ) ) ]
  - Interpretation: β1t = level; β2t = slope; β3t = curvature. λ determines exponential decay speed.
  - In this study slope approximated by y(3) − y(120) ≈ β2t (as in Diebold and Li (2006)).
- Macro-financial linkages:
  - Use generalized vector autoregressive (GVAR) analysis (Koop, Pesaran, Potter (1996)) to capture historical dynamic linkages and produce order-invariant generalized impulse responses.
  - For each country estimate VAR with NS factors (β'i for i=1,2,3), industrial production (IP), inflation (π), domestic policy rate (iP), euro area policy rate (i*), and real effective exchange rate (REER).
  - Variables (except β'i and interest rates) are expressed as 12-month year-on-year changes. Lag polynomial C(L) with L=3.
  - Shocks μ ~ N(0, Ω).
  - Innovation: reconstruct movements in the full yield curve from impulse responses of the β'i rather than reporting impulse responses of β'i alone.
- Spillovers:
  - Two approaches:
    - Principal components driving each NS factor for the three CEE countries and for large bond markets (U.S., Germany, Japan); study simple correlations between regional/global components and country NS factors.
    - Estimate GVAR comprising the three NS factors for each CEE country plus the three NS factors of the euro area to characterize linkages by generalized impulse responses and variance decompositions.

### Empirical Results — Estimated βi,t and Macroeconomic Relationships
- Estimation context:
  - Estimated β'i capture key yield-curve features in the CEE countries and are highly correlated with empirical counterparts of level, slope, and curvature.
  - Correlation of β1t and long rate y(120) exceeds 0.8 in all three countries.
  - Slope β2t correlates strongly with slope measured as y(3) − y(120).
  - Curvature β3t correlates with curvature measured as 2y(24) − y(120) − y(3).
- Table 1 — Correlation between βi,t and yield-curve metrics (as presented):
  - β1 / Long-term Yield Level (y(120)):
    - The Czech Republic 0.83
    - Hungary 0.98
    - Poland 0.99
  - β2 / Slope (y(3) − y(120)):
    - The Czech Republic 0.52 0.75 0.06
    - Hungary 0.58 0.98 0.07
    - Poland 0.88 0.99 0.39
  - β3 / Curvature (2y(24) − y(120) − y(3)):
    - The Czech Republic 0.32 −0.11 0.91
    - Hungary 0.04 n.a. 0.95
    - Poland 0.78 0.72 0.77
  - (Sources: IFS, National authorities and authors' calculations. 1/ Based on consumer survey by the European Commission.)
- Table 2 — Descriptive statistics for estimated Nelson-Siegel factors (sample 2000–08), presented as rows for β1, β2, β3 across CZE HUN POL EUR:
  - β1 — Mean / Std. Dev. / Median / Maximum / Minimum
    - Mean: 6.08 6.53 6.36 4.92
    - Std. Dev.: 1.34 0.53 1.11 0.70
    - Median: 6.35 6.53 6.02 5.18
    - Maximum: 9.28 7.71 9.51 6.06
    - Minimum: 3.81 4.99 4.73 3.50
  - β2 — Mean / Std. Dev. / Median / Maximum / Minimum
    - Mean: −2.97 2.65 2.14 −1.75
    - Std. Dev.: 1.08 1.88 4.82 1.13
    - Median: −2.82 2.68 0.06 −1.88
    - Maximum: −1.07 6.62 13.40 −0.07
    - Minimum: −5.14 −0.66 −2.13 −3.59
  - β3 — Mean / Std. Dev. / Median / Maximum / Minimum
    - Mean: −1.28 1.83 1.98 −1.68
    - Std. Dev.: 2.21 1.86 2.98 1.68
    - Median: −1.12 1.58 1.36 −1.95
    - Maximum: 4.38 7.49 13.40 1.49
    - Minimum: −6.04 −1.74 −2.09 −4.98
  - (Source: Authors' calculations.)
- Observed country-specific descriptive findings:
  - Mean level factor (β1) exceeded the euro area mean by 100 basis points or more for CEE countries; smallest margin for the Czech Republic (also most variable); Hungary’s β1 margin over euro exceeded 150 basis points and its standard deviation was about half that in other CEE countries.
  - Slope β2 differences: On average the yield curve was inverted (short rates > long rates) in Hungary and Poland; Poland’s extreme value of 13.4 percentage points corresponds to November, 2000. Czech Republic experienced a normal-sloping yield curve throughout the sample.
  - Curvature β3: Hungary and Poland have seen positive curvature that imparts a hump; Czech Republic has not.
- Correlations between estimated βi,t and macro variables (Table 3 — selected summarized findings; exact numeric entries presented in table in source):
  - β1 (level):
    - Positively correlated with inflation in Hungary and Poland, and to a lesser extent in the Czech Republic.
    - Correlated with European Commission survey-based 12-month inflation expectations except in Hungary.
  - β2 (slope):
    - Negatively correlated with industrial production and capacity utilization in Hungary and Poland but not in the Czech Republic.
    - Positively correlated with real effective exchange rate (REER) in all three countries, suggesting increases in shorter-end rates attract capital inflows and put upward pressure on nominal and real exchange rates.
  - β3 (curvature):
    - Positively correlated with interest rates in Hungary and Poland; opposite sign in the Czech Republic.

### Empirical Results — Spillovers and Regional/Global Components
- Approach summary:
  - Extract principal components for each NS factor across CEE countries (regional) and across large bond markets (global: Germany, U.S., Japan).
  - Compute simple correlations between these components and each country’s NS factors to gauge co-movement.
  - Use a GVAR comprising the nine NS factors (three per CEE country) plus euro-area NS factors to compute generalized impulse responses and variance decompositions to characterize spillovers.
- Methodological detail (principal component construction):
  - Regional first component for vector [β_i,CZE, β_i,HUN, β_i,POL] computed as R_REG = W_REG′ × B_REG, where W_REG is the normalized eigenvector associated with the largest eigenvalue of the sample variance-covariance matrix Σ_REG (Campbell, Lo, and MacKinlay, 1997).
  - Global first component constructed similarly for [β_i,DEU, β_i,US, β_i,JPN].

### Organization of Paper (structure referenced)
- Section layout as in source:
  - I. Introduction
  - II. Background (Literature Review; Stylized Facts)
  - III. Modeling the Yield Curve and Spillovers (Nelson-Siegel; Macro-Financial Linkages; Spillovers)
  - IV. Empirical Results (Estimated βi,t and macroeconomic variables; Spillovers)
  - V. Case study: Poland’s euro adoption announcement on September 10, 2008 (examination of potential macroeconomic and spillover effects) — empirical framework applied to this event.
  - Final section: Conclusions.

*Source: _wp1051 - References (PDF content provided).*

### 0.58    n.a.

### _wp1051 - 0.58    n.a.

### Poland
- 0.58
- n.a.
- 0.07
- -0.54
- -0.13
- -0.34
- 0.30
- 0.88
- 0.98
- 0.39
- -0.65
- -0.58
- -0.64
- 0.40

### The Czech Republic
- 
- 3
- 0.32
- -0.11
- 0.00
- 0.49
- 0.45
- 0.43
- -0.03
- -0.364
- 0.06
- -0.59

### Hungary
- Hungary

*Source: _wp1051 - 0.58    n.a.*

### 0.04    n.a.

### _wp1051 - 0.04    n.a.

### Macro-Financial Linkages — overview
- Analysis focuses on two-way dynamic relationship between macroeconomic variables and Nelson–Siegel (NS) yield curve factors for three CEE countries (Czech Republic, Hungary, Poland) using generalized impulse responses and VAR/GVAR frameworks for 2000–08 and subsample 2004–08.
- Models used: three-country VAR models with x = [i*, β1, β2, β3, IP, REER, π, ip] and L = 3; GVAR analysis for spillovers; principal components for regional/global decomposition.

### Impulse Responses: Macro Shocks on Yield Curves
- Shocks to euro area policy rate:
  - Tend to produce a relatively large hump-shaped upward shift in the yield curve at the three-month horizon.
  - Upward shifts in whole yield curve persist at least until after a year, with the exception of Hungary.
- Shocks to economic activity:
  - Have negligible effects on the yield curve.
- Shocks to inflation:
  - Systematically shift up yield curves at the three-month horizon.
  - Czech Republic: parallel upward shift.
  - Hungary and Poland: tilting of the curve.
  - Effects die out after one year except in Hungary, where tilt persists longer.
- Shocks to REER (unexpected strengthening):
  - Associated with a small but persistent downward shift in the entire yield curve, more pronounced in Hungary.
- Shocks to domestic policy rates:
  - Rapidly transmitted to shorter end of yield curve (liquidity effect).
  - Czech Republic and Hungary: curves shift up in a humped fashion and become inverted only after three months.
  - Poland: entire yield curve shifts upward in a parallel fashion and effect persists longer.

### Impulse Responses: Yield Factor Shocks on Yield Curves
- Shocks to level factor (β1):
  - Associated with upward shift of curve combined with short-run positive tilt.
  - Shift vanishes within a year in Hungary; more persistent in Poland and, to a lesser extent, Czech Republic.
- Shocks to slope factor (β2):
  - Produce a "pure" negative tilt (inversion) where long end dips below baseline.
  - In Hungary, shocks to short-term rates link to exchange rate moves and do not much affect long-term inflation expectations.
- Shocks to curvature factor (β3):
  - Generate a hump-shaped impact that vanishes quickly (arbitraged away by traders).

### Impulse Responses: Yield Shocks on Macro Variables
- General finding: responses of macro variables to shocks in β1, β2, β3 are small and measured imprecisely; REER responses estimated with somewhat more precision.
- β1 (level) shocks:
  - Associated with increase in actual inflation and domestic policy rates in Poland and, to a lesser extent, Czech Republic.
  - Reverse responses in Hungary.
  - REER: appreciates quickly in Czech Republic; depreciates in Poland and Hungary.
- β2 (slope) shocks:
  - Associated with domestic interest rate hikes (quickly reversed in Czech Republic) and a decrease in economic activity in short run in all CEE countries.
  - REER tends to appreciate where short end increases (Hungary and Poland) and depreciate where short-end effect is smaller (Czech Republic).
- β3 (curvature) shocks:
  - Associated with rise in policy rates in all CEE countries.
  - No clear pattern in other macro variables — curvature changes are largely financial market events.

### Cross-Border Spillovers — regional and global components
- Principal components analysis:
  - Roughly two-thirds of variance in β1 and at least half of variance in β2 and β3 for the CEE countries can be associated with a regional component (panel results).
  - By comparison, global level factor explains 75 percent of level factor variance in G-3; Diebold, Li, and Yue (2007) find over 90 percent of level factor variation associated with first principal component for U.S./Germany/Japan/U.K.
- Correlations between CEE NS factors and euro area/global factors:
  - Level factor (β1): correlations of Czech Republic and Poland with euro area/global component are quite high and statistically significant; Hungary’s correlation is substantially lower though it increased after EU accession.
  - Slope (β2) and curvature (β3): Czech Republic correlations with euro/global factors increased since EU accession and are consistently larger than in Hungary and Poland; correlations for Hungary and Poland turned negative since joining the EU (reflecting disinversion and ECB tightening since 2004).
- VAR-based contemporaneous residual correlations:
  - Shocks to euro-area slope (β2_euro) negatively correlated with euro-area level (β1_euro), suggesting inversion associated with fall in long-run inflation expectations.
  - Flattening of euro curve (positive β2_euro) negatively correlated with level factors in CEE countries → positive spillover as lower long-run inflation expectations in CEE.
  - At CEE country level, slope shocks (β2_i) negatively correlated with level factors (β1_i), and across countries β2_i negatively correlated with β1_j, indicating tightening in one country lowers long-run inflation expectations in others.

### Cross-Border Spillovers — GIR (generalized impulse responses)
- Euro area shocks:
  - Increase in euro-area level (β1_euro): gradually transmitted to CEE countries; steepening on impact in Czech Republic and Poland; effect dies out after three months in euro area but lasts longer in CEE countries.
  - Flattening/tilting of euro-area curve (β2_euro): similar effects in CEE countries; more persistent in later subsample.
  - β3_euro shocks: small effects on CEE yield curves.
- CEE country shocks (selected):
  - Shock to β1_CZE: tilts Czech yield curve and similarly affects others; effect on long end vanishes by six months in Czech Republic but persists longer in Poland (and to lesser extent Hungary).
  - Shock to β2_CZE: inverts Czech curve and shifts down Poland and Hungary (second subsample).
  - Shock to β1_HUN: shifts up yield curves in Poland and (less) in Czech Republic.
  - Shock to β2_HUN: inverts Hungary curve; small responses in Czech and Poland (shift down after EU accession).
  - Shock to β1_POL and β2_POL: transmitted to other countries especially in short run (first three months) — stronger spillovers from Poland to Czech Republic and Hungary.
- Variance decomposition highlights:
  - Global/regional factors rarely account for more than 40 percent of variance of yield curve factors in these CEE countries.
  - For Czech Republic and Poland, euro area yield curve shocks explain between 20 percent and 30 percent of variance of level and slope of their own curves.
  - For Hungary, regional shocks are more important than euro-area-originated shocks, though euro-area importance increased after EU accession.
  - After accession (2004–08), percentage of variance explained by regional factors increases markedly in Czech Republic and Poland; euro-area factor importance grows in Hungary.

### Event Study: Poland’s Euro Adoption Announcement (September 10, 2008)
- Event description:
  - Prime Minister Tusk announced goal of adopting the euro in 2011 on Wednesday September 10, 2008; subsequent clarifications indicated Commission approval sought to adopt euro in 2011 and euro notes in circulation January 1, 2012.
  - Market reaction (September 9–12, 2008): Poland’s yield curve steepened on September 10 consistent with expectations of tighter monetary policy and lower long-run inflation; Czech Republic and Hungary yield curves shifted down (Czech with one-day delay).
- Two-stage analysis approach:
  1. Extract NS β shocks implied by yield curve moves three days after event using OLS regression decomposition of yield changes onto country impact GIRs (h-day changes with h = 1, 2, 3; based on 11 maturities).
  2. Use estimated shocks in VAR/macro-financial models to infer macro implications and spillovers.
- Estimated NS β shocks for Poland (first-stage results):
  - Poland-only regression estimates indicate:
    - Small (statistically insignificant) reduction in β1 (long-run inflation expectations).
    - 40 to 50 bps increase in β2 (short end / monetary tightening).
    - Decline of almost 200 bps in β3 (curvature).
    - These shocks explained about 80 percent of Poland yield curve movement following announcement.
  - Panel (three-country) regression yields qualitatively similar but smaller coefficient estimates (around 20 basis points) and explains far less of movement across three countries.
- Shock scenario fed into macro-financial models (Table 8; numbers in basis points):
  - μβ1 = -10
  - μβ2 = 50
  - μβ3 = -150
- Macro effects inferred for Poland:
  - Real exchange rate appreciates over next ten months; activity contracts initially; both reversed over the following year.
  - Small increase in inflation.
  - Policy interest rates do not increase much — market short-term rate rise makes policy changes unnecessary in model.
  - Note: results are illustrative; Lehman Brothers bankruptcy one week later constitutes an additional shock altering predictions.
- Spillover effects:
  - Poland’s announcement transmitted to Czech Republic and Hungary via NS β linkages.
  - Hungary experienced similar real exchange rate appreciation and inflation; level of activity improves marginally in Hungary (contrast with Poland).
  - Spillovers to Czech Republic are much smaller.

### Conclusions — key findings
- Macro-financial linkages in CEE countries show similarities with mature markets:
  - Increases in policy rates and inflation shift up whole curve.
  - Policy rate shocks lead to eventual inversion of the curve; inflation shocks have limited persistence in CEE.
- Directionality:
  - Macro variables have stronger influence on yield curves than vice versa (consistent with other studies).
  - Negative impact of yield curve inversion on economic activity observed in CEE, albeit weaker than in mature markets.
- Country differences:
  - Hungary’s short-term interest rate increases are associated with exchange rate appreciation and smaller declines in long-term inflation expectations — consistent with prior exchange rate band policy.
- Integration and spillovers:
  - Substantial co-movement between euro-area and CEE yield curves, especially post-EU accession (post-2004).
  - Euro-area shocks explain 20-30 percent of Czech Republic and Poland yield curve variation; smaller share for Hungary that increased after 2004.
  - The CEE region has become more integrated with euro area and U.S., particularly in convergence of level of yield curves; convergence after 2004 reflects policy framework convergence in Hungary.
  - Shocks to yield curve in one country can produce measurable spillovers to neighbors (illustrated by Poland’s euro adoption announcement).

*Source: Authors' calculations and analysis drawn from the provided content (2000–08; 2004–08 subsamples).*

### Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries

### Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries

### Czech Republic: monetary developments and yield curve
- After a protracted recession in 1997-99, growth resumed in 2000.
- Inflation was at around 4 percent.
- The yield curve was slightly upward sloping at the 5 and 6 percent level.
- Strong capital inflows and an appreciating koruna led the CNB to cut (two-week) repo policy rates in 2002 and 2003 to just 2 percent—leaving money market rates sometimes below the ECB policy rate.
- An increase in inflation expectations steepened the curve in 2004; after a 50 bps tightening in the summer of 2004 the CNB reassessed the inflation outlook and cut rates to 1 3/4 percent in April 2005.
- As inflationary pressures brought headline inflation above the 3 percent target, monetary policy was gradually tightened in 2007 and 2008.
- Yield-curve dynamics:
  - Steepening in 2006.
  - Followed by a shift upwards and a mild flattening.
  - Current spread of around 100 bps between long and short rates (as of data cut-off end-May, 2008).

### Hungary: monetary developments and yield curve
- NB Hungary reduced policy rates steadily from 18 percent in 1998 to 8 percent in 2002.
- The forint yield curve shifted downwards significantly but remained inverted—attributed to Hungary’s strong convergence prospects.
- In the first half of 2003, policy rates were cut further to 7 percent to fend off an (appreciation) speculative attack against the strong end of the exchange rate band.
- The yield curve “normalized” briefly, then became inverted again in the second-half of 2003.
- Increasing fiscal and external imbalances forced the NBH to increase policy rates by 600 bps (to 12.5 percent) between June and November 2003.
- The yield curve remained inverted during 2004 but flattened to a much lower level of around 7 percent in 2005 and 2006.
- In response to increasing inflationary pressures, policy rates were increased again in the second half of 2007 and early 2008, leading again to an inverted yield curve.

### Poland: monetary developments and yield curve
- After substantial monetary tightening following the Russian crisis, the MPC reduced policy rates by around 1400 bps between early 2001 and mid-2003—when the yield curve normalized and flattened at around 5.5 percent.
- A new MPC increased rates in 2004 to prevent one-off inflation pressures from becoming entrenched in expectations, then cut rates by 250 bps (to 4 percent) between March 2005 and March 2006.
- Following two years of strong economic growth, policy interest rates were increased seven times since April 2007 to 5 ¾ percent.
- The yield curve currently displays a small hump in the short-end—reflecting expectations of further near-term tightening.

*Data cut-off date: end-May, 2008.*

---

### Appendix II. The Generalized Vector Autoregression (GVAR)

### Model setup and identification
- The GVAR analysis follows Koop, Pesaran, and Potter (1996), postulating a vector autoregressive (VAR) model that assumes shocks μ are distributed multivariate normal, N(0,Ω), and thus Xt is multivariate normal.
- GIR (Generalized Impulse Responses) considers the conditional expectation of Xt given a specific shock to μt.
- The expected (average) effect of a shock to the ith component μit is obtained as the conditional expectation of Xt and, given the properties of the multivariate normal, equals:
  - 11,,(,)(), titi G I R XvC Lv μσ −− = = ×Ω××
- Setting v = σi,i1/2 is standard in GIR analysis as this corresponds to a one standard error shock.
- In that case, the GIR for a shock to the ith variable can be obtained from Cholesky analysis by placing the ith variable first in the ordering; the first column of the Cholesky factoring corresponds to the GIR in that ordering.

---

### Appendix III. Estimating the Nelson-Siegel Betas

### Estimation approach
- Estimation follows Diebold and Li (2006) and Nelson and Siegel (1987).
- For a given λ0 (determined by a grid search to seven decimal places that minimizes the sum of squared residuals for the full sample), time series for β1t, β2t, and β3t are recovered for the 101 months of the sample period (January, 2000 through May, 2008).
- For each month, OLS estimates use data for 11 maturities: 3, 6, 12, 24, 36, 48, 60, 84, 96, 108, and 120 months.

### Fit and pricing errors (Table A1: Descriptive Statistics of the Yield Curve Residuals)
- Table A1 reports mean and standard deviation of residuals (in basis points) for each maturity and country (Czech Republic, Hungary, Poland), sample 2000:01 through 2008:05.
- Czech Republic residuals (mean; standard deviation) by maturity (in months):
  - 3: -1.2; 6.9
  - 6: 0.3; 3.5
  - 12: 0.4; 9.7
  - 24: 1.8; 7.0
  - 36: 1.0; 5.4
  - 48: -0.5; 6.0
  - 60: -2.2; 8.2
  - 84: -2.6; 8.5
  - 96: 0.9; 4.3
  - 108: 1.6; 4.3
  - 120: 0.5; 6.0
- Hungary residuals (mean; standard deviation) by maturity:
  - 3: 1.3; 9.0
  - 6: -0.8; 7.1
  - 12: -1.5; 9.7
  - 24: 0.1; 5.8
  - 36: 2.3; 6.1
  - 48: -1.2; 7.2
  - 60: -1.4; 8.2
  - 84: 1.0; 6.0
  - 96: 1.7; 4.3
  - 108: 0.2; 4.8
  - 120: -1.9; 6.0
- Poland residuals (mean; standard deviation) by maturity:
  - 3: 3.5; 11.2
  - 6: -0.2; 7.8
  - 12: -3.8; 16.2
  - 24: -5.1; 11.9
  - 36: 0.7; 6.5
  - 48: 5.4; 9.9
  - 60: -0.2; 14.9
  - 84: 5.9; 11.2
  - 96: 2.8; 6.5
  - 108: -1.1; 6.2
  - 120: -7.9; 12.8
- Authors’ comment on fit:
  - NS approximation provides a reasonably good fit.
  - Czech Republic mean residuals range from an over pricing of about 2½ basis points (at 84 months) to an under pricing of less than 2 basis points (at 24 months).
  - Hungary shows a similar range; largest errors at 120 and 36 months.
  - Poland shows larger pricing errors: over pricing about 8 basis points (at 120 months) and under pricing close to 6 basis points (at 84 months).
  - Most of the largest model errors occurred prior to EU accession (before 2004); in Poland largest residuals concentrated before 2002.
  - Volatility of pricing errors: more volatile in Poland.

### Factor diagnostics and correlations (figures summarized)
- Figures (F1–F4) illustrate:
  - Maximum and minimum yield curve residuals, 2000–2008 (Czech, Hungary, Poland).
  - Level factor (β1t) and long-term yields (y(120)).
  - Slope factor (β2t) and slope of the yield curve (y(3) - y(120)).
  - Curvature factor (β3t) and curvature of yield curve (2y(24) - y(3) - y(120)).

### Impulse response functions for βi (Appendix III.B)
- Autoregressive models (order 12) were fitted to estimated βit and used to obtain impulse response functions.
- Key findings:
  - Shocks to β1t (level) are roughly about 40 basis points in these countries; persistence lasts longer in Poland where after the initial jump the level remains virtually unchanged for about a quarter.
  - Shocks to β2t (slope) differ in size and persistence:
    - Larger in Hungary—about 70 basis points compared to 40 basis points in Czech Republic and Poland—but the effect dies out quicker.
    - In Poland, following the initial jump the slope remains virtually unchanged for a number of months and even rises further before dissipating.
    - The effect persists the longest in the Czech Republic.
  - Shocks to β3t (curvature) are larger than those to other βit—ranging between 100 and 120 basis points—but their effects dissipate faster.

---

### Appendix IV. Impulse Responses for the Nelson-Siegel Betas

### VAR specification for impulse responses (figures F6–F8)
- Responses of Nelson Siegel beta factors to shocks to β1t, β2t, and β3t are presented (2000–08).
- Source note for figures:
  - Authors' calculations.
  - Based on three country VAR models with x = [i*, β1, β2, β3, IP, REER, π, ip] and L = 3.

---

### Appendix V. Estimating Shock for Historical Episodes

### Table A2: Shock’s underlying Poland’s Euro adoption announcement (September 10–12, 2009)
- Table A2 reports coefficient estimates for μβ1POL, μβ2POL, μβ3POL from regressions explaining one-day (September 10), two-day (September 11), and three-day (September 12) changes in the yield curve. Regressions are shown for a Poland sample and a stacked sample.
- Selected reported statistics (as presented):
  - Standard error of the estimate:
    - Columns reported values: 0.13, 0.12, 0.15, 0.06, 0.09, 0.08, 0.12, 0.06 (for the one-day specification).
    - For two-day specification: 0.13, 0.11, 0.14, 0.06, 0.09, 0.09, 0.13, 0.08.
    - For three-day specification: 0.13, 0.11, 0.14, 0.06, 0.11, 0.10, 0.16, 0.10.
  - R squared and R bar reported in table cells (examples from three-day change):
    - R squared: 0.00, 0.31, -0.16, 0.83, -0.02, 0.06, -1.28, 0.15.
    - R bar: 0.00, 0.31, -0.16, 0.78, -0.02, 0.06, -1.28, 0.09.
  - Number of observations:
    - Poland sample regressions: 11 observations each.
    - Stacked sample regressions: 33 observations each.
- Source: Authors' calculations based on data from September 10 through September 12, 2009.

*Source: _wp1051 - Appendix I. Main Developments and Monetary Policy Decisions in the CEE Countries (authors' calculations and figures), data from 2000:01 through 2008:05; data cut-off end-May, 2008; Table A2 based on data September 10–12, 2009.*

### References

### _wp1051 - References

### IMF reports and surveys
- AREAER, 1997, “Annual Report on Exchange Arrangements and Exchange Restrictions,” (Washington: International Monetary Fund).  
- Bayoumi, Tamim, and Andrew Swiston, 2007, “The Ties that Bind: Measuring International Bond Spillovers Using Inflation-Indexed Bond Yields,” IMF Working Paper 07/128 (Washington: International Monetary Fund).  
- Roldós, Jorge E., 2004, “Emerging Local Bond Markets,” in D. Mathieson and others, Emerging Local Securities and Derivatives Market, World Economic and Financial Surveys, International Monetary Fund, Washington, D.C.  
- Schadler, Susan, 2005, ed. Euro Adoption in Central and Eastern Europe: Opportunities and Challenges, International Monetary Fund, Washington, D.C.  

### Term structure, yield curve, and macro-finance models
- Ang, Andrew, and Monika Piazzesi, 2003, “A No-Arbitrage Vector Autoregression of Term Structure Dynamics with Macroeconomic and Latent Variables,” Journal of Monetary Economics, Vol. 50 (May), pp. 745–87.  
- Diebold, Francis, C. Li, and V. Yue, 2007, “Global Yield Curve Dynamics and Interactions: A Dynamic Nelson-Siegel Approach,” NBER working paper # 13588  
- Diebold, Francis, and C. Li, 2006, “Forecasting the Term Structure of Government Bond Yields,” Journal of Econometrics, Vol. 130 (February), pp. 337–64.  
- Diebold, Francis, Glenn Rudebusch, and Boragan Aruoba, 2006, "The Macroeconomy and the Yield Curve: A Dynamic Latent Factor Approach," Journal of Econometrics, Vol. 131, pp. 309–38 (March-April).  
- Diebold, Francis, Monika Piazzesi, and Glenn D. Rudebusch, 2005, “Modeling Bond Yields in Finance and Macroeconomics,” National Bureau of Economic Research Working Paper No. W11089.  
- Dewachter, Hans, and Marco Lyrio, 2006, “Macro Factors and the Term Structure of Interest Rates, Journal of Money, Credit, and Banking, Vol. 38, (February).  
- Evans, C.L., and D.A. Marshall, 1998, “Monetary Policy and the Term Structure of Nominal Interest Rates: Evidence and Theory,” Carnegie-Rochester Conference Series on Public Policy, Vol. 49 (December), pp. 53–111.  
- Evans, C.L., and D.A. Marshall, 2007, “Economic Determinants of the Nominal Treasury Yield Curve,” Journal of Monetary Economics, Vol. 54, pp. 1986–2003.  
- Hordahl, Peter, and Oreste Tristani, 2007, “Inflation Risk Premia in the Term Structure of Interest Rates,” BIS Working Paper No. 228, (Basel, Switzerland: Monetary and Economic Department, Bank for International Settlements (BIS)).  
- Lemke, Wolfgang, 2008, “An Affine Marco-Finance Term Structure Model for the Euro Area,” The North American Journal of Economics and Finance, Vol. 19, No. 1, pp. 41–69.  
- Litterman, R. and J. Scheinkman, 1991, “Common Factors Affecting Bond Returns,” Journal of Fixed Income, Vol. 1, pp. 51–61.  
- Nelson, Charles R. and A. F. Siegel, 1987, “Parsimonious Modeling of Yield Curves,” The Journal of Business, Vol. 60, No. 4, 473-489.  
- Rudebusch, Glenn D., and T. Wu, 2004, “A Macro-Finance Model of the Term Structure, Monetary Policy, and the Economy,” mimeo, Federal Reserve Bank of San Francisco.  

### Exchange rates, forecasting, and macroeconometrics
- Fama, E., 1984, “Forward and Spot Exchange Rates,” Journal of Monetary Economics, Vol. 14 (November), pp. 319–38.  
- Estrella, A., and F.S. Mishkin, 1997, “The Predictive Power of the Term Structure of Interest Rates in Europe and the United States: Implications for the European Central Bank,” European Economic Review, Vol. 41, No. 7, pp. 1375–1401.  
- Koop, Gary, Hashem Pesaran, and Simon Potter, 1996, “Impulse Response Analysis in Nonlinear Multivariate Models,” Journal of Econometrics, Vol. 74, No. 1, pp. 119–47.  
- Sims, Christopher, 1980, "Macroeconomics and Reality," American Economic Review, Vol. 62, pp. 1-48.  

### Econometric and time-series methods
- Campbell, John, Andrew Lo, and Craig MacKinlay, 1997, The Econometrics of Financial Markets, (New Jersey, Princeton: Princeton University).  
- Dhrymes, Phoebus, 1978, Introductory Econometrics, (New York, New York: Springer-Verlag).  
- Hamilton, J. D., 1994, Time Series Analysis, Princeton University Press, Princeton, New Jersey.  

### Country- and market-specific studies
- Kladívko, Kamil, 2010, “The Czech Treasury Yield Curve from 1999 to the Present,” forthcoming Czech Journal of Economics and Finance.  

*Source: _wp1051 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1051.pdf_