## _wp08140

## Source details

**Canonical URL:** [_wp08140](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08140.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08140.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08140.pdf.json)

---

### I. Introduction — context and motivation
- Thailand is a very open economy; the share of trade in GDP stood at over 140 percent in 2006.
- Exports of goods and services accounted for about 74 percent of GDP.
- Net exports have driven Thai growth over the last two years; contribution of domestic investment to growth was negative for four out of six quarters since the beginning of 2006.
- Imports grew at 7 percent in 2006 and 6.4 percent in 2007 (H1), after average growth of about 14 percent over the previous five years.
- Exports grew at 17.4 percent in 2006 and 18.4 percent in 2007 (H1), after average growth of about 10.5 percent over the previous five years.
- Thailand’s most important export market is the US, accounting for 15 percent of Thai exports in 2006.
- WEO forecasts cited: U.S. growth falls to 1.9 percent in 2007 from 2.9 percent the previous year, and remains at 1.9 percent in 2008; WEO projections became more pessimistic between August and October.
- The referenced U.S. baseline: October WEO lowers the point estimate for U.S. growth in 2008 to 1.9 percent; 1.25 percent falls within the 90 percent confidence band.

### II. Model — structure and equations
- Model type: small New Keynesian macroeconomic model with rational expectations, adapted from Berg, Karam and Laxton (2006); four structural macroeconomic equations.
- Two-country structure: small Thailand economy and the US (rest of the world).
- Thailand equations (summary):
  - Output gap (ygap) depends on real interest rate (RR), real exchange rate (z), U.S. output gap (yusgap), with past and future domestic output gap terms and a disturbance ε.
  - Phillips curve: four-quarter change in the CPI depends on expected and lagged inflation (four-quarter ahead and four-quarter lag), the output gap, the exchange rate gap, and a shock ε.
  - Exchange rate: real IP/UIP links real exchange rate z to domestic and U.S. real interest rates and a risk-premium shock; expectations for exchange rate are allowed (δz not forced equal to 1).
  - Monetary policy: forward-looking Taylor-type rule; nominal short-term interest rate responds to expected future inflation (four-quarter ahead), the output gap, and includes interest-rate smoothing (lagged rate).
- U.S. block: analogous behavioral equations for output gap, Phillips curve, and policy rule but without foreign influences; Thailand assumed too small to affect the U.S.

### III. Estimation methodology and data
- Estimation approach: Bayesian techniques following Schorfheide (2000); use of MCMC (Random Walk Metropolis) to characterize posterior distributions.
- Numerical details:
  - Total iterations: 100,000 replications.
  - Burn-in: first 50,000 iterations discarded.
  - Scaling chosen to attain a 30 percent acceptance rate.
  - Convergence monitored with potential scale reduction factors and trace plots.
- Data used:
  - Thailand: real GDP, headline CPI inflation rate, nominal interest rate, real bilateral exchange rate.
  - US: real GDP, headline CPI inflation rate, nominal interest rate.
  - Sample period: 1999Q1 to 2007Q2 (to avoid the crisis period).
  - Equilibrium values of real GDP, real interest rate, and real exchange rate constructed exogenously using a variant of the Hodrick Prescott filter.
- Priors:
  - Chosen to reflect model dynamics and judgment about the Thai economy (complete priors listed in Table 1 in source).
  - Examples:
    - Output-gap dynamics: expectation of substantial inertia with small forward-looking component; sum of β3 and β4 smaller than β1, with β4 at least as big as β3 in open economies.
    - Phillips curve: δ1 expected significantly below 0.5 in many countries; δ2 chosen at 0.25.
    - Exchange-rate pass-through δ3 tied to openness; δz in UIP set to 0.5.
    - Interest rate responses α2 and α3 set to 1.5 and 0.5 respectively.
    - Interest-rate smoothing α1 chosen as 0.5.
    - For the US, priors assume zero prior for foreign output influence and zero prior for exchange rate in output and Phillips equations.

### IV. Key estimation results and interpretation
- Posterior estimates: many parameters lie in the ballpark of prior assumptions; posterior mean and 5th and 95th percentiles reported in source.
- Notable deviations from priors:
  - Lagged gap coefficient in the output gap equation estimated at 0.21 (prior mean 0.7), indicating limited inertia in the Thai output gap.
  - Sum of β3 and β4 is smaller than β1 but not by much, implying only moderate monetary policy transmission lags.
  - Lagged interest-rate coefficient in the monetary policy rule estimated at 0.84, implying substantial policy inertia in practice.
  - Estimated δz suggests quite adaptive expectations for the exchange rate.
- Selected model parameter posterior summaries (domestic and U.S. posterior mean; 5% mean; 95% mean):
  - Domestic:
    - β1 (Beta): 0.700; 0.059; 0.349
    - β2 (Beta): 0.100; 0.052; 0.144
    - δ1 (Gamma): 0.300; 0.195; 0.472
    - δz (Beta): 0.500; 0.310; 0.526
    - α1 (Beta): 0.500; 0.786; 0.911
    - α2 (Gamma): 1.500; 0.661; 1.931
  - Foreign (U.S.):
    - β1_us (Beta): 0.700; 0.589; 0.925
    - β2_us (Beta): 0.100; 0.053; 0.152
    - δ1_us (Beta): 0.200; 0.111; 0.317
    - α1_us (Beta): 0.500; 0.393; 0.676
    - α2_us (Gamma): 2.000; 1.247; 2.498

### V. Baseline calibration and forecasts
- Staff baseline projection used to solve the model initially.
- Recent growth and projections:
  - 2006 growth: 5 percent.
  - H1 2007 growth slowed to 4.3 percent (year-on-year).
  - Staff estimate for 2007 growth: 4.5 percent.
  - Projected growth: 2008 at 5 percent; 2009 at 5.2 percent (baseline assumes rotation of demand from external to domestic sources).
- Baseline projection for the US used: September U.S. desk forecast; U.S. growth slows from 2.9 percent in 2006 to 2.25 percent in both 2007 and 2008.

### VI. Risk scenarios — U.S. slowdown and Thai impact
- Scenario analysis approach: model used to evaluate risk from a slowdown in U.S. growth, framing policy responses and sensitivity to assumptions.
- Main quantitative scenario described:
  - Shock design: shock the U.S. output gap so that 2008 U.S. GDP falls by 100 basis points from 2.25 percent to (text truncated in source).
- Two monetary policy reaction scenarios for Thailand:
  - Delayed response: no policy reaction for two quarters.
  - Immediate response: model-based contemporaneous movement in policy rate.
- Scenario results — impact on Thai GDP and transmission:
  - Delayed monetary policy response:
    - Thai GDP growth falls by 0.9 percentage points below the baseline in 2008, with output bottoming in Q2 2008 and recovering in 2009.
    - Output begins slowing relative to baseline in Q3 2007 due to forward-looking features, despite the shock occurring at beginning of 2008.
    - Declining output leads to lower inflation; UIP implies real exchange rate appreciation because foreign interest rates fall by more than domestic rates.
  - Immediate monetary policy response:
    - Thai GDP growth falls by 0.6 percentage points below baseline.
    - The accommodative monetary response mutes the decline in output by 0.3 percent.
    - Inflation and exchange rate qualitative responses are similar to the delayed-response case.
  - Trade-share adjusted (lower-bound) scenario:
    - Weighting the estimated U.S. impact coefficient by the U.S. share of Thailand’s exports (about 15 percent) yields a fall in Thai GDP of 0.135 percentage points below baseline.
    - This scenario is equivalent to assuming perfect correlation between the U.S. economy and Thailand’s other trade partners.
- Summary assessment of range:
  - Upper-bound impact on Thai GDP growth: 0.9 percentage points.
  - Lower-bound impact on Thai GDP growth: 0.135 percentage points.
  - Midpoint of the range: 0.52 percentage points.
  - The exact position within the 0.135 to 0.9 percentage point range depends on the correlation between U.S. and Thailand’s other trade partners.
  - A quicker monetary policy response could reduce the impact by 0.3 percentage points.

### VII. Interpretation of uncertainty, limitations, and policy implications
- Drivers of uncertainty and model limitations:
  - Omission of a fiscal policy response that could rebalance growth from external to domestic sources (may amplify negative impacts).
  - Absence of an explicit risk premium in the UIP condition (which could dampen exchange rate appreciation and the fall in output).
  - The two-country framework implicitly assumes the U.S. is Thailand’s only trade partner; omitted-variable bias likely inflates the estimated coefficient on the U.S. output gap if other partners’ cycles are positively correlated with the U.S.
- Confidence intervals and robustness:
  - Bayesian analysis constructs confidence intervals for output responses using the posterior distribution of parameters.
  - Example: first panel in Figure 3 shows the 90 percent lower and upper bound of output response as alpha 1 ranges from its 5 percent level to its 95 percent level, holding other parameters at posterior means.
  - For most estimated parameters, confidence intervals around the immediate policy response scenario are quite tight; only parameters that exhibit wider bands are highlighted.
- Policy implications:
  - The likely impact of a U.S. slowdown on Thailand is nontrivial but small relative to expected movements in domestic demand in 2008 as Thailand recovers from political turbulence and depressed investor confidence.
  - Policymakers can materially mitigate the spillover via timely accommodative monetary policy; a quicker response could reduce the direct impact on GDP by 0.3 percentage points.
  - Accounting for trade-share composition and correlated trade-partner cycles substantially lowers the estimated external impact; using the U.S. export share (about 15 percent) yields a modest lower-bound impact of 0.135 percentage points.

*Source: IMF staff estimates (extracts from _wp08140).*

### 1. Data Used for Estimation.............................................................................................

### _wp08140 - 1. Data Used for Estimation.............................................................................................

### I. Introduction — context and motivation
- Thailand is a very open economy; the share of trade in GDP stood at over 140 percent in 2006.
- Exports of goods and services accounted for about 74 percent of GDP.
- Net exports have driven Thai growth over the last two years; contribution of domestic investment to growth was negative for four out of six quarters since the beginning of 2006.
- Imports grew at 7 percent in 2006 and 6.4 percent in 2007 (H1), after average growth of about 14 percent over the previous five years.
- Exports grew at 17.4 percent in 2006 and 18.4 percent in 2007 (H1), after average growth of about 10.5 percent over the previous five years.
- Thailand’s most important export market is the US, accounting for 15 percent of Thai exports in 2006.
- WEO forecasts cited: U.S. growth falls to 1.9 percent in 2007 from 2.9 percent the previous year, and remains at 1.9 percent in 2008; WEO projections became more pessimistic between August and October (figure referenced).

### II. Model — structure and equations
- Model type: small New Keynesian macroeconomic model with rational expectations, adapted from Berg, Karam and Laxton (2006); four structural macroeconomic equations.
- Two-country structure: small Thailand economy and the US (rest of the world).
- Thailand output gap equation (in the paper's notation):
  - ygap dynamics depend on real interest rate (RR), real exchange rate (z), and U.S. output gap (yusgap), with past and future domestic output gap terms and a disturbance ε.
- Phillips curve specification:
  - Inflation (four-quarter change in the CPI) depends on expected and lagged inflation (four-quarter ahead and four-quarter lag), the output gap, the exchange rate gap, and a shock ε.
- Exchange rate equation:
  - Real IP/UIP condition links the real exchange rate z to domestic and U.S. real interest rates and a risk-premium shock; expectations for exchange rate are allowed (δz not forced equal to 1).
- Monetary policy rule:
  - Forward-looking Taylor-type rule: short-term nominal interest rate responds to expected future inflation (four-quarter ahead), the output gap, and includes interest-rate smoothing (lagged rate).
- U.S. block:
  - Analogous behavioral equations for output gap, Phillips curve, and policy rule but without foreign influences; Thailand is assumed too small to affect the U.S.

### III. Estimation methodology and data
- Estimation approach: Bayesian techniques following Schorfheide (2000); use of MCMC (Random Walk Metropolis) to characterize posterior distributions.
- Numerical details:
  - Total iterations: 100,000 replications.
  - Burn-in: first 50,000 iterations discarded.
  - Scaling chosen to attain a 30 percent acceptance rate.
  - Convergence monitored with potential scale reduction factors and trace plots.
- Data used:
  - Thailand: real GDP, headline CPI inflation rate, nominal interest rate, real bilateral exchange rate.
  - US: real GDP, headline CPI inflation rate, nominal interest rate.
  - Sample period: 1999Q1 to 2007Q2 (to avoid the crisis period).
  - Equilibrium values of real GDP, real interest rate, and real exchange rate constructed exogenously using a variant of the Hodrick Prescott filter.
- Priors:
  - Chosen to reflect model dynamics and judgment about the Thai economy; Table 1 lists complete priors (referenced).
  - Examples of prior choices and rationale:
    - Output-gap dynamics: expectation of substantial inertia with small forward-looking component; sum of β3 and β4 smaller than β1, with β4 at least as big as β3 in open economies.
    - Phillips curve: δ1 (forward-looking inflation) expected significantly below 0.5 in many countries; δ2 chosen at 0.25.
    - Exchange-rate pass-through parameter δ3 tied to openness; δz in UIP set to 0.5 (intermediate between 0 and 1).
    - Interest rate responses α2 and α3 set to 1.5 and 0.5 respectively.
    - Interest-rate smoothing α1 chosen as 0.5.
    - For the US, priors assume zero prior for foreign output influence and zero prior for exchange rate in output and Phillips equations.

### IV. Key estimation results and interpretation
- Posterior estimates (summary):
  - Many estimated parameters lie in the ballpark of prior assumptions; the paper reports posterior mean and 5th and 95th percentiles for each parameter (Table 1 referenced).
  - Notable deviations from priors:
    - Lagged gap coefficient in the output gap equation estimated at 0.21 (prior mean 0.7), indicating limited inertia in the Thai output gap.
    - Sum of β3 and β4 is smaller than β1 but not by much, implying only moderate monetary policy transmission lags.
    - Lagged interest-rate coefficient in the monetary policy rule estimated at 0.84, implying substantial policy inertia in practice.
    - Estimated δz suggests quite adaptive expectations for the exchange rate.
- Baseline model calibration/forecast:
  - Staff baseline projection used to solve model initially.
  - Recent growth and projections quoted:
    - 2006 growth: 5 percent.
    - H1 2007 growth slowed to 4.3 percent (year-on-year).
    - Staff estimate for 2007 growth: 4.5 percent.
    - Projected growth: 2008 at 5 percent; 2009 at 5.2 percent (baseline assumes rotation of demand from external to domestic sources).
  - Baseline projection for the US used: September U.S. desk forecast; U.S. growth slows from 2.9 percent in 2006 to 2.25 percent in both 2007 and 2008.

### V. Risk scenarios — U.S. slowdown and Thai impact
- Scenario analysis approach:
  - The model is used to evaluate risk from a slowdown in U.S. growth, framing policy responses and sensitivity to assumptions.
- Main quantitative scenario described:
  - Shock design: shock the U.S. output gap so that 2008 U.S. GDP falls by 100 basis points from 2.25 percent to (text truncated in source at this point).
- Summary assessment:
  - The analysis estimates that output in Thailand may decline by up to 0.9 percent relative to baseline under the considered U.S. slowdown scenarios (text: "we analyze two main risk scenarios, and estimate that output in Thailand may decline by up to 0.9 percent relative to baseline").

*Source: _wp08140 - 1. Data Used for Estimation (extracts provided).*

### 1.25 percent. This is consistent with the lower end of the revised U.S. forecast of the October

### _wp08140 - 1.25 percent. This is consistent with the lower end of the revised U.S. forecast of the October

### Key premise and modeling approach
- The analysis uses a New Keynesian, two-country, general equilibrium model parameterized for Thailand and the U.S., estimated with Bayesian updating.
- The referenced U.S. baseline: October WEO lowers the point estimate for U.S. growth in 2008 to 1.9 percent; 1.25 percent falls within the 90 percent confidence band.
- Two monetary policy reaction scenarios for Thailand are analyzed: a delayed response (no policy reaction for two quarters) and an immediate (model-based) response that moves the policy rate contemporaneously.

### Scenario results — impact on Thai GDP and transmission
- Delayed monetary policy response:
  - Thai GDP growth falls by 0.9 percentage points below the baseline in 2008, with output bottoming in Q2 2008 and recovering in 2009.
  - Output begins slowing relative to baseline in Q3 2007 due to forward-looking features, despite the shock occurring at beginning of 2008.
  - Declining output leads to lower inflation; UIP implies real exchange rate appreciation because foreign interest rates fall by more than domestic rates.
- Immediate monetary policy response:
  - Thai GDP growth falls by 0.6 percentage points below baseline.
  - The accommodative monetary response mutes the decline in output by 0.3 percent.
  - Inflation and exchange rate qualitative responses are similar to the delayed-response case.
- Trade-share adjusted (lower-bound) scenario:
  - Weighting the estimated U.S. impact coefficient by the U.S. share of Thailand’s exports (about 15 percent) yields a fall in Thai GDP of 0.135 percentage points below baseline.
  - This scenario is equivalent to assuming perfect correlation between the U.S. economy and Thailand’s other trade partners.

### Interpretation of estimated range and drivers of uncertainty
- Estimated bounds for the impact of a 100 basis point U.S. slowdown relative to baseline:
  - Upper-bound impact on Thai GDP growth: 0.9 percentage points.
  - Lower-bound impact on Thai GDP growth: 0.135 percentage points.
  - Midpoint of the range: 0.52 percentage points.
- The exact position within the 0.135 to 0.9 percentage point range depends on the correlation between U.S. and Thailand’s other trade partners.
- A quicker monetary policy response could reduce the impact by 0.3 percentage points.
- Model limitations that may amplify negative impacts:
  - Omission of a fiscal policy response that could rebalance growth from external to domestic sources.
  - Absence of an explicit risk premium in the UIP condition (which could dampen exchange rate appreciation and the fall in output).
  - The two-country framework implicitly assumes the U.S. is Thailand’s only trade partner; omitted-variable bias likely inflates the estimated coefficient on the U.S. output gap if other partners’ cycles are positively correlated with the U.S.

### Confidence intervals and robustness
- Bayesian analysis constructs confidence intervals for output responses using the posterior distribution of parameters.
- Example: the first panel in Figure 3 shows the 90 percent lower and upper bound of output response as alpha 1 (the coefficient measuring the aggressiveness of monetary policy to changes in inflation) ranges from its 5 percent level to its 95 percent level, holding other parameters at posterior means.
- For most estimated parameters, confidence intervals around the immediate policy response scenario are quite tight; only parameters that exhibit wider bands are highlighted.

### Selected model parameter posterior summaries (extract)
- Domestic parameters (Posterior mean; 5% mean; 95% mean):
  - β1 (Beta): 0.700; 0.059; 0.349
  - β2 (Beta): 0.100; 0.052; 0.144
  - δ1 (Gamma): 0.300; 0.195; 0.472
  - δz (Beta): 0.500; 0.310; 0.526
  - α1 (Beta): 0.500; 0.786; 0.911
  - α2 (Gamma): 1.500; 0.661; 1.931
- Foreign (U.S.) parameters (Posterior mean; 5% mean; 95% mean):
  - β1_us (Beta): 0.700; 0.589; 0.925
  - β2_us (Beta): 0.100; 0.053; 0.152
  - δ1_us (Beta): 0.200; 0.111; 0.317
  - α1_us (Beta): 0.500; 0.393; 0.676
  - α2_us (Gamma): 2.000; 1.247; 2.498

### Conclusion — policy implication and overall assessment
- The likely impact of a U.S. slowdown on Thailand is nontrivial but small relative to expected movements in domestic demand in 2008 as Thailand recovers from political turbulence and depressed investor confidence.
- Policymakers can materially mitigate the spillover via timely accommodative monetary policy; a quicker response could reduce the direct impact on GDP by 0.3 percentage points.
- Accounting for trade-share composition and correlated trade-partner cycles substantially lowers the estimated external impact; using the U.S. export share (about 15 percent) yields a modest lower-bound impact of 0.135 percentage points.

*Source: IMF staff estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08140.pdf_