## wp17128

## Source details

**Canonical URL:** [wp17128](https://www.imf.org/-/media/files/publications/wp/2017/wp17128.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17128.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17128.pdf.json)

---

### I. Introduction — motivation and research questions
- Focus: policy-related uncertainty in Japan as a potentially important, overlapping factor behind chronically weak economic performance.
- Key questions:
  - How has policy uncertainty moved over time in Japan?
  - Which policy areas account for the largest share of policy uncertainty and its movements?
  - What do changes in policy-related uncertainty portend for Japan’s economic performance?
- Data produced:
  - Several newspaper-based policy uncertainty measures for Japan (monthly from 1987 to the present), including overall EPU and indices for monetary policy, fiscal policy, trade policy, and exchange rate policy.

### Measuring the Japan EPU index — methodology
- Source newspapers: Yomiuri, Asahi, Mainichi and Nikkei; raw monthly article counts from January 1987 onwards.
- Article inclusion criteria: must contain at least one term in each of three categories:
  - (E) “economic” or “economy”
  - (P) policy-related terms (examples: “tax,” “government spending”, “regulation,” “central bank”, “upper/lower house”, “Diet”, “Prime Minister”, “government deficit”, “government debt”, “(de)regulation”, “structural reform”)
  - (U) “uncertain” or “uncertainty”
- Construction steps (exact sequence):
  1. Obtain raw monthly EPU article counts by newspaper.
  2. Scale raw counts by the total number of articles in the same newspaper and month to obtain relative EPU frequency counts.
  3. Standardize each newspaper’s relative EPU counts to unit standard deviation from 1987 to 2015.
  4. Seasonally adjust the resulting newspaper-level series.
  5. Average the resulting series across the four papers by month to obtain the overall monthly Japan EPU index.
  6. Multiplicatively normalize the four-paper average EPU series to a mean of 100 from 1987 to 2015.
- Auditing: small-scale audits and investigations informed term set selection and classification choices; appendix details seasonality adjustments and audits.
- Keyword selection (Japanese): “経済”, “景気” for economic; “不透明”, “不安”, “不確実”, “不確定” for uncertainty (narrowed to these four).

### Key descriptive findings on index behavior and composition
- Major peaks/events associated with the overall Japan EPU index:
  - Asian Financial Crisis.
  - Lehman Brothers failure (September 2008).
  - U.S. debt‑ceiling fight in 2011.
  - Brexit referendum.
  - Deferral of the consumption tax hike.
  - Contested national elections and major leadership transitions.
  - Drift down during Junichiro Koizumi’s tenure (2001–2006); gradual decline starting in 2013 coinciding with the launch of Abenomics; resurgence since 2015.
- Composition of EPU-relevant articles (among articles that satisfy E, P and U criteria):
  - 57 percent reference fiscal policy matters.
  - 27 percent reference monetary policy.
  - 8 percent reference trade policy.
  - 3 percent reference exchange rate policy.
- Comparison with Baker et al. (2016) Japan EPU index:
  - Expanded coverage from two to four major newspapers (including Nikkei).
  - Seasonal adjustment implemented here.
  - Expanded and refined term set.
  - This index: more persistent rise during the Asian Financial Crisis and larger spikes around Lehman Brothers failure and the US debt downgrade; Baker et al. index: higher volatility in the late 1980s and early 1990s.
- Co-movement with other indicators:
  - Overall EPU co-varies positively with implied volatilities for Japanese equities, exchange rates and interest rates and with Ito’s (2016) political uncertainty measure.
  - Fiscal and monetary policy uncertainty indices correlate at 0.68.
  - Correlation with Husted, Rogers and Sun (2016b) monetary index: 0.22 at monthly frequency and 0.32 at quarterly frequency (differences driven by international/US papers vs. Japanese papers and different term sets, e.g., "quantitative easing", "negative interest rate", "inflation target").

### Policy-category uncertainty indices — construction and dynamics
- Construction: flag articles that meet E, P and U criteria and also contain one or more policy-category terms (see Table 2 term sets); then follow same scaling, standardization, seasonal adjustment, averaging and normalization steps as for the overall index.
- Observed dynamics:
  - Fiscal policy uncertainty:
    - Responds strongly to contested elections, Twisted Diet episodes, political confusion (e.g., PM Mori), debates over stimulus packages (2002, 2008), and talks on social security and tax bills (2012).
    - Sensitive to major external developments (e.g., US government debt downgrade, European debt crisis).
  - Monetary policy uncertainty:
    - Spikes around introduction of Quantitative Easing (2001) and expansions (2001–02, 2010–11), and introduction of negative interest rates in 2016.
    - Peaks in March 2008 amid concerns about a BOJ Governor vacancy.
  - Correlation patterns: policy-category indices correlate positively with one another while retaining distinct dynamics; implied interest rate volatility correlates more highly with the monetary policy uncertainty index; Ito’s (2016) political uncertainty measure correlates more highly with fiscal policy uncertainty.

### Correlations with other uncertainty measures and financial-market implied volatilities
- Cross-country and temporal correlations:
  - Fiscal and monetary policy uncertainty indices for Japan correlate with U.S. counterparts at about 0.3 for both fiscal and monetary.
  - Correlation of Japan EPU with US and Europe EPU:
    - About 0.3 in the pre-2007 period.
    - 0.6 for the 2007–2016 period.
- Selected correlations (Table 3; samples specified exactly):
  - Equity Market Volatility: 0,50 (Overall); 0,42 (Fiscal); 0,44 (Monetary); -0,01 (Trade); 0,16 (Exchange Rate); sample 1990M6-2017M3
  - Exchange Rate Volatility: 0,53 (Overall); 0,43 (Fiscal); 0,34 (Monetary); 0,04 (Trade); 0,13 (Exchange Rate); sample 2003M1-2017M2
  - Interest Rate Volatility: 0,22 (Overall); 0,06 (Fiscal); 0,27 (Monetary); -0,12 (Trade); -0,04 (Exchange Rate); sample 2007M1-2016M8
  - Political Uncertainty: 0,23 (Overall); 0,30 (Fiscal); 0,07 (Monetary); -0,04 (Trade); 0,19 (Exchange Rate); sample 1987M1-2017M3
  - EPU-Global: 0,50 (Overall); 0,40 (Fiscal); 0,47 (Monetary); 0,69 (Trade); 0,22 (Exchange Rate); sample 1997M1-2017M3
  - EPU-US: 0,49 (Overall); 0,44 (Fiscal); 0,48 (Monetary); 0,36 (Trade); 0,26 (Exchange Rate); sample 1987M1-2017M3
  - EPU-Japan (Old): 0,66 (Overall); 0,61 (Fiscal); 0,52 (Monetary); 0,22 (Trade); 0,20 (Exchange Rate); sample 1988M6-2016M4
  - EPU-Europe: 0,49 (Overall); 0,42 (Fiscal); 0,51 (Monetary); 0,57 (Trade); 0,15 (Exchange Rate); sample 1987M1-2017M3
  - VIX - US: 0,52 (Overall); 0,51 (Fiscal); 0,33 (Monetary); -0,14 (Trade); 0,17 (Exchange Rate); sample 1990M1-2017M3
- Empirical note: Japan EPU correlates above 0.5 with option-implied volatilities for the Japanese stock market and the yen-dollar exchange rate; implied equity volatility fell rapidly after the global financial crisis while EPU remained elevated through 2015–2016.

### VAR specifications, identified EPU shocks and macroeconomic effects
- VAR sample windows:
  - Monthly baseline VAR: 1994M1–2016M12.
  - Quarterly baseline VAR: 1987Q1–2016Q4.
- VAR details:
  - Two lags used in all VAR specifications.
  - Standard Cholesky decomposition to identify shocks, with log Japan EPU ordered first unless noted otherwise.
- Largest monthly EPU shocks (standardized units):
  - 2008M10 — global financial crisis (largest).
  - 2011M8 — U.S. debt downgrade, European debt concerns, PM Kan’s resignation and DPJ leadership election.
  - 1997M11 — policy debates about fiscal consolidation and Asian financial crisis.
  - 2010 — European debt crisis also produced a very large positive EPU shock.
- Baseline monthly VAR impulse-response summary:
  - A unit standard deviation EPU innovation yields a peak estimated fall after about one year of 0.8 percent in industrial production.
  - A 0.2 percent fall in tertiary sector activity following an upward EPU innovation (monthly VAR).
  - Smaller, marginally significant and less persistent response in consumption (synthetic consumption index).
  - Employment response: modest in size but highly persistent.
  - Sectoral investment: larger estimated output responses in sectors producing investment goods.
- Quarterly VAR results:
  - A unit standard deviation upward EPU innovation foreshadows a peak fall in real GDP of about 0.3 percent after one year.
  - Investment response peaks at an estimated 1 percent.
- Historical contribution:
  - EPU shocks account for peak investment and GDP swings of about 6 and 2 percent, respectively.
  - Timing of pronounced EPU-induced movements: 1997–1999, 2001–2002 and 2008–2012.
- Comparison with Baker et al. (2016):
  - Peak effect on GDP: 0.38 percent using this Japan EPU index vs. 0.18 using Baker et al. (2016).
  - Peak effect on investment: 1.1 percent using this Japan EPU index vs. 0.6 using Baker et al. (2016).
- Robustness checks:
  - Results robust to alternative Cholesky orderings, inclusion of option-implied equity price volatility, and inclusion of the Global EPU index from Davis (2016).
  - Including Global EPU: both Japan EPU and Global EPU shocks yield material, statistically significant output and investment responses, implying cross-border spillovers.
  - Discarding data after 2006 yields somewhat smaller responses.
  - Restricting to post-1995 period (BOJ at zero lower bound) yields larger responses.
- Interpretation scenarios for VAR-based evidence:
  - Scenario 1: EPU is an unforeseen policy uncertainty shock causing worsening macro performance via option-value and cost-of-capital channels.
  - Scenario 2: EPU captures bad news about the outlook not fully captured by other VAR variables; EPU amplifies and propagates an external shock.
  - Scenario 3: EPU is a summary statistic for information missing from the VAR and has no causal role; external micro-evidence makes this third interpretation difficult to reconcile with firm-level identification studies.

### Detailed chronology of episodes with heightened EPU (selected)
- October 1987
  - October 19: Global stock market collapse (Black Monday); Nikkei tumbled; BOJ injected liquidity; yen appreciated sharply; Ministry of Finance intervened in FX.
  - October 20: PM Nakasone announced nomination of Noboru Takeshita as next LDP President.
- March 1995
  - Failures of Toyo Kyowa and Anzen credit cooperatives; creation of Tokyo Kyodo Bank with BOJ capital contributions.
- November–December 1997
  - Asian financial crisis intensification; tax break of JPY 2 trillion financed by JGB issuance in November 1997.
- July–August 1998
  - PM Obuchi replaced PM Hashimoto on July 30; LTCB nationalized in October 1998.
- June 2000
  - Lower House elections: ruling parties’ share fell from 66 to 56 percent.
- February–March 2001
  - BOJ cut policy rate by 10 bps from 0.25 to 0.15 percent on February 28.
  - March 19: BOJ introduced QE (switch target to current account balance).
- July 2001
  - PM Koizumi won Upper House election; CEFP produced first annual Honebuto-no-Hoshin report on June 26.
- October 2002
  - Bank restructuring and BOJ eased policy; current account balances increased from JPY 10-15 trillion to 15-20 trillion; long-term JGB purchases increased from JPY 1 trillion to 1.2 trillion per month.
- March 2008
  - Twisted Diet uncertainties; repeated rejections of BOJ Governor nomination; Deputy Governor Masaaki Shirakawa became Acting Governor on March 20 and Governor on April 9.
- September–October 2008
  - Lehman Brothers bankruptcy; Japanese economy shrank by 12.4 and 15.4 percent (saar) in Q4 2008 and Q1 2009, respectively.
  - Nikkei fell below 7,000 from around 12,000; yen appreciated from around 120 in 2007 to 93 per dollar by end-2009.
  - Two supplementary budgets enacted in October 2008 and January 2009; BOJ reduced policy rate twice in October and December 2008.
- February 2009
  - Government estimated Q4 2008 growth at negative 12.4 percent.
- May–June 2010
  - European sovereign debt crisis accelerated; Nikkei lost more than 10 percent in May; yen briefly below 90 per dollar.
  - PM Hatoyama resigned in June.
- August 2011
  - Delay in approval of law allowing deficit‑financing bonds for FY2011 until August; Twisted Diet; European debt crisis intensified; S&P downgraded US sovereign debt in August.
- June 2012
  - Tri-party agreement on tax and social security reform on June 21 included two-step consumption tax increase in April 2014 and October 2015.
- January–February 2016
  - BOJ announced negative interest rates on January 29.
- May–June 2016
  - Brexit referendum on June 23 caused substantial yen appreciation; BOJ in July doubled ETF purchases and expanded US dollar lending program.
  - Mixed reports/speculation up to decision on June 1 about postponing consumption tax hike.
- November 2016–January 2017
  - Policy uncertainty rose ahead of the US election; US withdrew from TPP on January 23, 2017.

### Policy implications and recommendations
- Broad conclusion: high policy uncertainty undermines macroeconomic performance—either as an impulse, as an amplifier/propagator of external impulses, or both.
- Institutional implications:
  - Well-designed policy institutions and rules can limit scope for negative shocks to trigger large jumps in policy uncertainty.
- Specific recommendations for Japan:
  - Credible plans to follow through on trade reforms to promote trade-creating investments.
  - Credible medium-term budget plans to foster confidence about Japan’s fiscal trajectory.
  - Further efforts to improve the BOJ’s communications framework to lessen uncertainty about the direction of monetary policy.
- Additional literature-based implications:
  - Stronger predictability in the policy environment could improve macroeconomic performance and increase the potency of countercyclical stabilization policies.

### Quantitative highlights and recorded magnitudes
- Episodes defined as movements exceeding 1.64 standard deviations from average.
- Proximate shares of EPU articles:
  - Fiscal policy terms: 57 percent.
  - Monetary policy terms: 27 percent.
  - Trade policy terms: 8 percent.
  - Exchange rate policy terms: 3 percent.
- Monetary policy: BOJ policy rate cut of 10 bps from 0.25 to 0.15 percent on February 28, 2001.
- Fiscal policy: Tax break of JPY 2 trillion in November 1997 (financed by JGB issuance).
- BOJ balance and JGB purchase changes in October 2002:
  - Current account balances increased from JPY 10-15 trillion to 15-20 trillion.
  - Long-term JGB purchases increased from JPY 1 trillion to 1.2 trillion per month.
- Japanese GDP declines during global financial crisis:
  - Shrinkage of 12.4 and 15.4 percent (saar) in Q4 2008 and Q1 2009, respectively.
  - Government published Q4 2008 growth estimate at negative 12.4 percent.
- Yen exchange rate movements:
  - Around 120 per dollar in 2007 to 93 per dollar by end-2009.
  - Yen briefly below 90 per dollar in May–June 2010.
  - Yen had been below 80 against the dollar since mid-July 2011.
  - Yen depreciated by around 10 percent in February 2009 (contextual).
- Stock market magnitudes:
  - Nikkei fell to below 7,000 from around 12,000 after Lehman bankruptcy.
  - Nikkei lost more than 10 percent in May 2010.
- VAR effect sizes:
  - Monthly: unit standard deviation EPU innovation → peak fall after ~1 year of 0.8 percent in industrial production.
  - Monthly: 0.2 percent fall in tertiary sector activity.
  - Quarterly: unit standard deviation EPU innovation → peak fall in real GDP of about 0.3 percent after one year; investment response peaks at 1 percent.
  - Historical contribution: EPU shocks account for peak investment and GDP swings of about 6 and 2 percent, respectively.
- Sample windows explicitly used:
  - Monthly VAR sample: 1994M1–2016M12.
  - Quarterly VAR sample: 1987Q1–2016Q4.
  - EPU normalization mean period: 1987–2015 (mean of 100).

*Source: wp17128 (IMF Working Paper).*

### REFERENCES  ______________________________________________________________34

### REFERENCES  ______________________________________________________________34

### Figures
- 1. Japan Economic Policy Uncertainty Index (1987M1–2017M3)  ____________________11
- 2. Fiscal Policy Uncertainty Index (1987M1-2017M3) _____________________________16
- 3. Monetary Policy Uncertainty Index (1987M1-2017M3) __________________________17
- 4. Fiscal Policy Uncertainty Indices for Japan and the United States  __________________19
- 5. Monetary Policy Uncertainty Indices for Japan and the United States  _______________19
- 6. Trade Policy Uncertainty Index (1987M1-2017M3)  _____________________________20
- 7. Exchange Rate Policy Uncertainty Index (1987M1-2017M3) ______________________21
- 8. Japan EPU Index and Option-Implied Volatility of Nikkei Equity Index  _____________23
- 9. Our Fiscal Policy Uncertainty Index Compared to Ito’s Political Uncertainty Index  ____24
- 10. Proximate Sources of Economic Policy Uncertainty ____________________________25
- 11. Impulse Responses to Unit Standard Deviation EPU Innovation, Monthly Data  ______29
- 12. Impulse Responses to Unit Standard Deviation EPU Innovation, Quarterly Data ______30
- 13. Historical Contribution of Japan EPU Shocks to GDP and Investment Fluctuations  ___31
- 14. Robustness to Alternative VAR Specifications with Quarterly Data: GDP Impulse Responses to a Unit Standard Deviation Japan EPU Innovation ______________________32

### Tables
- 1. Term Sets for the Overall Japan EPU Index _____________________________________9
- 2. Term Sets for Policy Category Uncertainty Indices ______________________________13

*Source: wp17128 - REFERENCES  ______________________________________________________________34*

### 3. Correlation of Japan EPU Indices with Other Uncertainty Measures  ________________22

### 3. Correlation of Japan EPU Indices with Other Uncertainty Measures  ________________22

### I. Introduction — motivation and research questions
- Focus: policy-related uncertainty in Japan as a potentially important, overlapping factor behind chronically weak economic performance.
- Key questions addressed:
  - How has policy uncertainty moved over time in Japan?
  - Which policy areas account for the largest share of policy uncertainty and its movements?
  - What do changes in policy-related uncertainty portend for Japan’s economic performance?
- Data produced: several newspaper-based policy uncertainty measures for Japan (monthly from 1987 to the present, with planned monthly updates), including overall EPU and indices for monetary policy, fiscal policy, trade policy, and exchange rate policy.

### Measuring the Japan EPU index — methodology
- Source newspapers: Yomiuri, Asahi, Mainichi and Nikkei; raw monthly article counts from January 1987 onwards.
- Article inclusion criteria: must contain at least one term in each of three categories:
  - (E) “economic” or “economy”
  - (P) policy-related terms (e.g., “tax,” “government spending”, “regulation,” “central bank”, “upper/lower house”, “Diet”, “Prime Minister”, “government deficit”, “government debt”, “(de)regulation”, “structural reform”)
  - (U) “uncertain” or “uncertainty”
- Construction steps (exact sequence):
  1. Obtain raw monthly EPU article counts by newspaper.
  2. Scale raw counts by the total number of articles in the same newspaper and month to obtain relative EPU frequency counts.
  3. Standardize each newspaper’s relative EPU counts to unit standard deviation from 1987 to 2015.
  4. Seasonally adjust the resulting newspaper-level series.
  5. Average the resulting series across the four papers by month to obtain the overall monthly Japan EPU index.
  6. Multiplicatively normalize the four-paper average EPU series to a mean of 100 from 1987 to 2015.
- Auditing: small-scale audits and investigations informed term set selection and classification choices; appendix details seasonality adjustments and audits.

### Key descriptive findings on index behavior and composition
- Major peaks and events associated with the overall Japan EPU index:
  - Peaks during the Asian Financial Crisis, Lehman Brothers failure (September 2008), the U.S. debt-ceiling fight in 2011, the Brexit referendum, and the deferral of the consumption tax hike.
  - Tendency to rise around contested national elections and major leadership transitions.
  - Moderately countercyclical fluctuations.
  - Drift down during Junichiro Koizumi’s tenure (2001–2006) and gradual decline starting in 2013 coinciding with the launch of Abenomics; resurgence in policy uncertainty since 2015 amid multiple domestic and external concerns.
- Composition of EPU-relevant articles (among articles that satisfy E, P and U criteria):
  - 57 percent reference fiscal policy matters.
  - 27 percent reference monetary policy.
  - 8 percent reference trade policy.
  - 3 percent reference exchange rate policy.
- Comparison with Baker et al. (2016) Japan EPU index:
  - The two indices are highly correlated but differ in several ways:
    - This study expands coverage from two to four major newspapers (including Nikkei).
    - Adjustment for seasonality implemented here.
    - Expanded and refined term set.
    - Result: this index shows a more persistent rise during the Asian Financial Crisis and larger spikes around Lehman Brothers failure and the US debt downgrade; Baker et al. index displays higher volatility in the late 1980s and early 1990s.
- Comparison with other measures:
  - The overall EPU index co-varies positively with implied volatilities for Japanese equities, exchange rates and interest rates and with Ito’s (2016) survey-based measure of political uncertainty in Japan.
  - Fiscal and monetary policy uncertainty indices correlate at 0.68.
  - Note: Husted, Rogers and Sun (2016b) construct another newspaper-based monetary policy uncertainty index for Japan; correlation with the present monetary index is 0.22 at monthly frequency and 0.32 at quarterly frequency (differences driven by use of international/US papers vs. Japanese papers and different term sets, including "quantitative easing", "negative interest rate", "inflation target").

### Policy-category uncertainty indices — construction and dynamics
- Construction: flag articles that meet E, P and U criteria and also contain one or more policy-category terms (see Table 2 term sets in source); then follow same scaling, standardization, seasonal adjustment, averaging and normalization steps as for the overall index.
- Observed dynamics:
  - Fiscal policy uncertainty index:
    - Responds strongly to contested elections, Twisted Diet episodes, political confusion (e.g., PM Mori), debates over stimulus packages (2002, 2008), and talks on social security and tax bills (2012).
    - Also sensitive to major external developments (e.g., US government debt downgrade, European debt crisis).
  - Monetary policy uncertainty index:
    - Spikes around the introduction of Quantitative Easing (2001) and its expansions (2001–02, 2010–11), and around the introduction of negative interest rates in 2016.
    - Peaks in March 2008 amid concerns about a BOJ Governor vacancy (Diet approval failure).
  - Correlation patterns: policy-category indices correlate positively with one another while retaining distinct dynamics; implied interest rate volatility correlates more highly with the monetary policy uncertainty index than with the fiscal policy uncertainty index; Ito’s (2016) political uncertainty measure correlates more highly with fiscal policy uncertainty.

### Predictive power and VAR evidence
- VAR specifications indicate:
  - Upward EPU innovations foreshadow deteriorations in Japan’s macroeconomic performance as reflected in impulse response functions for investment, employment and output.
  - Global economic policy uncertainty (Global EPU index of Davis (2016)) has a significant effect on Japan’s economic performance.
  - Upward innovations to this Japan EPU index foreshadow a larger decline in macroeconomic aggregates than innovations to the Japan EPU index in Baker et al. (2016), suggesting improved measurement yields stronger downstream econometric results.
- Caveat: VAR results do not prove causality but show the Japan EPU index contains useful information not captured by other forward-looking indicators.

### Related literature and theoretical channels
- Measurement literature: expands on Baker et al. (2016) by increasing newspaper coverage and refining term sets; other text-based and model-based approaches include Ito (2016), Alexopoulos and Cohen (2015), Azzimonti (2015), Hassan et al. (2016), Hlatshwayo and Saxegaard (2016), Husted, Rogers and Sun (2016a), Fátas and Mihov (2013), Grier and Perry (2000), Vitek (2002), Fernandez-Villaverde et al. (2015).
- Theoretical channels through which uncertainty affects economic outcomes (literature cited):
  - Investment and durable-purchase postponement due to option value of waiting (Bernanke, 1983; McDonald and Siegel, 1986; Dixit and Pindyck, 1994; Bloom et al., 2007; Bloom, 2009; Eberly, 1994).
  - Labor-market search frictions and hiring/firing fixed costs (Schaal, 2015; Valetta and Bengali, 2013); interaction with nominal rigidities (Leduc and Liu, 2016).
  - Higher risk premiums and financial-frictions channels (Arellano et al., 2010; Christiano et al., 2014; Gilchrist et al., 2014).
  - Precautionary savings raising and output reduction under nominal rigidities and policy constraints (Johansen, 2014; Fernandez-Villaverde et al., 2015).
  - Potential positive effect via increased option value of growth (Paddock et al., 1988; Bar-Ilan and Strange, 1996).
- Empirical evidence (selected):
  - Ito (2016): negative effects of policy uncertainty on employment and expenditures for consumer durables.
  - Morikawa (2013, 2016): survey evidence on firms’ perceptions of policy uncertainty and its effects on investment and hiring decisions.
  - Gulen and Ion (2016), Baker et al. (2016): negative effects of EPU on corporate investment; larger effects for firms with greater exposure to policy risks.
  - Constantinescu et al. (2017): high policy uncertainty depresses international trade in goods and services.
- Policy implications emphasized in the literature and summary:
  - Stronger predictability in the policy environment could improve macroeconomic performance.
  - For Japan, potential measures include: a concrete and credible medium-term fiscal plan, clear follow-through on structural reform plans, and a stronger communications framework at the Bank of Japan.
  - A stronger policy framework can increase the potency of countercyclical stabilization policies.

*Source: IMF Working Paper wp17128, Appendix sections and text on Japan EPU index construction, behavior, comparison, and VAR results.*

### 2008. Lehman Brothers failure and stimulus debate. L: August 2009. Lower House election, DPJ takes

### 2008. Lehman Brothers failure and stimulus debate. L: August 2009. Lower House election, DPJ takes office. M: May-June 2010. Greek crisis. PM Hatoyama’s resignation. Inauguration of PM Kan’s cabinet. N: August 2011. U.S. debt downgrade and concerns about European debt crisis. PM Kan’s resignation and DPJ’s leadership election. O: June 2012. Sovereign debt problems in Greece and Spain. Tri-party agreement on tax and social security reform. P: May-June 2016. Consumption tax hike delay and Brexit.

### Major episodic events and annotated index spikes
- K: August 2007 — disarray in global financial markets with the outbreak of U.S. subprime loan crisis. Injection of liquidity by central banks. Debate about policy rate cut in Japan and the US.
- L: March 2008 — concerns over vacancy of BOJ Governor.
- M: October 2008 — Lehman Brothers failure. BOJ cuts policy rate.
- N: August 2010 — debate on QE enhancement amid sharp yen appreciation.
- O: August 2011 — QE enhancement and the BOJ’s unilateral FX intervention.
- P: June 2013 — concerns over FED QE tapering.
- R: January–February 2016 — introduction of negative interest rates.
- S: May–July 2016 — debate on FX intervention amid surge in yen. Brexit.

### Construction and properties of policy-uncertainty indices
- Indices covered: overall EPU, fiscal policy uncertainty, monetary policy uncertainty, trade policy uncertainty, exchange rate policy uncertainty.
- Sample windows used in VAR analyses:
  - Monthly baseline VAR: 1994M1–2016M12.
  - Quarterly baseline VAR: 1987Q1–2016Q4.
- VAR specification details:
  - Two lags used in all VAR specifications.
  - Standard Cholesky decomposition to identify shocks, with log Japan EPU ordered first unless noted otherwise.
- Proximate sources (average shares of EPU articles):
  - Fiscal policy terms: 57 percent.
  - Monetary policy terms: 27 percent.
  - Trade policy terms: 8 percent.
  - Exchange rate policy terms: 3 percent.
- Note on shares: an article may contain terms from multiple categories; shares can sum to more or less than 100 percent of all EPU articles in the month.

### Comparisons with U.S. measures and cross-country correlations
- Fiscal and monetary policy uncertainty indices for Japan correlate with U.S. counterparts at about 0.3 for both fiscal and monetary policy.
- Cross-country correlation patterns:
  - Correlation of Japan EPU index and EPU indices for the US and Europe:
    - About 0.3 in the pre-2007 period.
    - 0.6 for the 2007–2016 period.
- Temporal patterns:
  - U.S. fiscal policy uncertainty was higher and more volatile in the late 1980s and early 1990s; Japan’s fiscal uncertainty was higher during the Asian Financial Crisis and has been elevated in recent years partly due to consumption tax uncertainty.
  - Japan monetary policy uncertainty became highly elevated since the introduction of negative interest rates in late January 2016; the U.S. index remained low except for a Brexit-related spike.

### Relationship with financial-market implied volatilities and other indicators
- Overall Japan EPU index correlations (from Table 3):
  - Equity Market Volatility: 0,50 (Overall); 0,42 (Fiscal); 0,44 (Monetary); -0,01 (Trade); 0,16 (Exchange Rate); sample 1990M6-2017M3
  - Exchange Rate Volatility: 0,53 (Overall); 0,43 (Fiscal); 0,34 (Monetary); 0,04 (Trade); 0,13 (Exchange Rate); sample 2003M1-2017M2
  - Interest Rate Volatility: 0,22 (Overall); 0,06 (Fiscal); 0,27 (Monetary); -0,12 (Trade); -0,04 (Exchange Rate); sample 2007M1-2016M8
  - Political Uncertainty: 0,23 (Overall); 0,30 (Fiscal); 0,07 (Monetary); -0,04 (Trade); 0,19 (Exchange Rate); sample 1987M1-2017M3
  - EPU-Global: 0,50 (Overall); 0,40 (Fiscal); 0,47 (Monetary); 0,69 (Trade); 0,22 (Exchange Rate); sample 1997M1-2017M3
  - EPU-US: 0,49 (Overall); 0,44 (Fiscal); 0,48 (Monetary); 0,36 (Trade); 0,26 (Exchange Rate); sample 1987M1-2017M3
  - EPU-Japan (Old): 0,66 (Overall); 0,61 (Fiscal); 0,52 (Monetary); 0,22 (Trade); 0,20 (Exchange Rate); sample 1988M6-2016M4
  - EPU-Europe: 0,49 (Overall); 0,42 (Fiscal); 0,51 (Monetary); 0,57 (Trade); 0,15 (Exchange Rate); sample 1987M1-2017M3
  - VIX - US: 0,52 (Overall); 0,51 (Fiscal); 0,33 (Monetary); -0,14 (Trade); 0,17 (Exchange Rate); sample 1990M1-2017M3
- Empirical relationship notes:
  - The Japan EPU index correlates above 0.5 with option-implied volatilities for the Japanese stock market and the yen-dollar exchange rate.
  - EPU and implied equity volatility show similar spikes during the global financial crisis, but implied equity volatility fell rapidly afterward while the EPU index stayed elevated (notably through 2015–2016).

### Proximate dynamics and interpretation
- Policy uncertainty tends to spike around contested elections, major political transitions, Twisted Diet outcomes, and leadership transitions at the Bank of Japan.
- Political stability episodes (e.g., Prime Minister Abe’s election end-2012 and Prime Minister Koizumi’s long tenure) coincided with lower or declining policy uncertainty.
- Institutional context matters:
  - Strong automatic fiscal stabilizers reduce the need for discretionary fiscal stimulus and the political conflict/uncertainty that discretionary tools can cause.
  - Heavy reliance on unconventional monetary tools (e.g., QE, forward guidance, negative rates) can raise uncertainty about policy effects and risk eroding political support for central banks.
- Historical episodes:
  - Asian Financial Crisis, downturn of 2001–02, and Global Financial Crisis of 2008–09 all produced spikes in fiscal and monetary policy uncertainty.
  - The build-up of dollar-denominated private and banking-sector debt in some Asian economies increased vulnerability and amplified policy uncertainty during the Asian Financial Crisis; stronger frameworks and larger reserve buffers mitigated uncertainty during the Global Financial Crisis.

### VAR-identified EPU shocks and macroeconomic effects
- Largest monthly EPU shocks (standardized units) from the monthly VAR:
  - 2008M10 — global financial crisis (largest).
  - 2011M8 — U.S. debt downgrade, European debt concerns, PM Kan’s resignation and DPJ leadership election.
  - 1997M11 — policy debates about fiscal consolidation and Asian financial crisis.
  - 2010 European debt crisis also produced a very large positive EPU shock.
- Baseline monthly VAR impulse-response result (Figure 11 summary):
  - A unit standard deviation EPU innovation yields a peak estimated fall after about one year of 0.8 percent in industrial production.
  - Upward EPU innovations foreshadow statistically significant declines in output, employment and consumption expenditures (95 percent confidence intervals shown in original results).

*Italicized source: wp17128 - 2008. Lehman Brothers failure and stimulus debate. L: August 2009. Lower House election, DPJ takes office. M: May-June 2010. Greek crisis. PM Hatoyama’s resignation. Inauguration of PM Kan’s cabinet. N: August 2011. U.S. debt downgrade and concerns about European debt crisis. PM Kan’s resignation and DPJ’s leadership election. O: June 2012. Sovereign debt problems in Greece and Spain. Tri-party agreement on tax and social security reform. P: May-June 2016. Consumption tax hike delay and Brexit (authors' calculations and figures).*

### 0.2 percent in tertiary sector activity. We also find a smaller, marginally significant and less

### wp17128 - 0.2 percent in tertiary sector activity. We also find a smaller, marginally significant and less

### Key empirical findings
- Monthly VAR: a 0.2 percent fall in tertiary sector activity following an upward EPU innovation.
- Monthly VAR: smaller, marginally significant and less persistent response in consumption (synthetic consumption index).
- Employment response: modest in size but highly persistent.
- Sectoral investment signal: breakdown of monthly IP data reveals much larger estimated output responses in sectors that produce investment goods.
- Quarterly VAR: a unit standard deviation upward EPU innovation foreshadows a peak fall in real GDP of about 0.3 percent after one year.
- Quarterly VAR: the investment response peaks at an estimated 1 percent.
- Historical contribution: EPU shocks account for peak investment and GDP swings of about 6 and 2 percent, respectively.
- Timing of pronounced EPU-induced movements: 1997–1999, 2001–2002 and 2008–2012.
- Comparison with Baker et al. (2016) Japan EPU index:
  - Peak effect on GDP is 0.38 percent using our Japan EPU index, while it is 0.18 using the Japan EPU index in Baker et al. (2016).
  - Peak effect on investment is 1.1 percent using our Japan EPU index, while it is 0.6 using the Japan EPU index in Baker et al. (2016).
- Co-movement and drivers:
  - Terms related to fiscal policy appear in about 57 percent of all articles that meet EPU criteria.
  - Terms related to monetary policy appear in about 27 percent.
  - About 65 percent of articles that contain the “economy” and “uncertainty” keywords also contain one of the “policy” terms (ratio ranges from a minimum of 0.45 to a maximum of 0.8).

### VAR robustness and specification results
- Robustness checks: the basic pattern—upward EPU innovations foreshadow future output declines—is robust to:
  - alternative Cholesky orderings;
  - inclusion of option-implied equity price volatility in the VAR system;
  - inclusion of the Global EPU index from Davis (2016).
- Specification changes: these modifications lead to somewhat smaller output responses and somewhat wider confidence intervals.
- Sample splits:
  - Discarding data after 2006 yields somewhat smaller responses.
  - Restricting attention to the post-1995 period (when the BOJ was up against the zero lower bound) yields larger responses.
- Global EPU: when including a Global EPU measure, both Japan EPU and Global EPU shocks yield material, statistically significant output and investment responses, implying cross-border spillovers.

### Interpretations / Scenarios for VAR-based evidence
- Scenario 1: an upward EPU innovation is an unforeseen policy uncertainty shock that causes worsening macroeconomic performance through real options effects, cost-of-capital effects or other mechanisms.
- Scenario 2: an upward EPU innovation captures bad news about the economic outlook not fully captured by other VAR variables; that bad news triggers a rise in EPU that has harmful effects—EPU amplifies and propagates an external causal impulse.
- Scenario 3: EPU acts only as a useful summary statistic for information missing from the VAR system and has no causal role as an impulse or propagation mechanism.
- External micro-evidence: the third interpretation is difficult to reconcile with micro studies using firm-level identification strategies (examples cited in the source).

### Policy implications and recommendations
- Broad conclusion: results favor the view that high policy uncertainty undermines macroeconomic performance—either as an impulse, as an amplifier/propagator of external impulses, or both.
- Institutional implication: past policy decisions and institutions shape the policy uncertainty response to contemporaneous economic shocks; well-designed policy institutions and rules can limit scope for negative shocks to trigger large jumps in policy uncertainty.
- Specific recommendations for Japan:
  - Credible plans to follow through on trade reforms to promote trade-creating investments.
  - Credible medium-term budget plans to foster confidence about Japan’s fiscal trajectory.
  - Further efforts to improve the BOJ’s communications framework to lessen uncertainty about the direction of monetary policy.

### Construction and validation of the Japan EPU index
- Data sources and sample:
  - Main interface: Nikkei Telecom to access archives of four Japanese-language newspapers.
  - Additional sources: Kikuzo II for Asahi Shimbun, MAISAKU for Mainichi Shimbun, Yomidas Rekishikan for Yomiuri Shimbun.
  - Sample start dictated by archive availability: January 1987.
  - Local editions of national papers were excluded to avoid spurious jumps in total article counts.
- Keyword selection:
  - Japanese “economic” terms identified: “経済” and “景気”.
  - Japanese “uncertainty” candidates: “不透明”, “不安”, “微妙”, “不確実”, “不安定”, and “不確定”; narrowed to four: “不透明”, “不安”, “不確実” and “不確定”.
  - Construction checks: an “economic uncertainty” index correlated with the economic cycle and stock market volatility indices as expected.
- Sample note on monthly VAR models: the sample for monthly VAR models is shorter reflecting the more limited sample period for the synthetic consumption index; if the tertiary activity index and synthetic consumption index are excluded and the whole sample (1987M1-201M12) is used, the impact of EPU innovations on employment and industrial production is slightly smaller but qualitatively similar.

*Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17128.pdf*

### 1987. At the second stage, we eliminated terms likely to trigger many false positives such as

### wp17128 - 1987. At the second stage, we eliminated terms likely to trigger many false positives such as

### Methodological note
- At the second stage, terms likely to trigger many false positives such as “interest rate” for monetary policy or “public works” for fiscal policy were eliminated.
- Episodes of heightened Economic Policy Uncertainty (EPU) are defined as movements exceeding 1.64 standard deviations from their average levels.

### Detailed chronology of episodes with heightened EPU (selected, chronological)
- October 1987
  - October 20: PM Nakasone announced nomination of Noboru Takeshita as next LDP President; other candidates included Shintaro Abe and Kiichi Miyazawa.
  - October 19: Global stock market collapse (Black Monday); Nikkei tumbled on opening; BOJ injected liquidity; yen appreciated sharply; Ministry of Finance intervened in FX.
- March 1995
  - Failures of Toyo Kyowa and Anzen credit cooperatives; creation of Tokyo Kyodo Bank with BOJ capital contributions.
  - Rescue plan criticized due to prior fraudulent management; management purged and prosecuted.
  - Yen surge reflecting interest rate differentials despite coordinated FX intervention with Germany and the US.
- November–December 1997
  - Intensification of the Asian financial crisis; Korea experienced currency and banking crisis.
  - PM Hashimoto faced headwinds on fiscal consolidation after bankruptcies; tax break of JPY 2 trillion financed by JGB issuance provided while fiscal consolidation continued.
- July–August 1998
  - PM Obuchi replaced PM Hashimoto on July 30 after ruling party losses; Twisted Diet increased political uncertainty.
  - Finance Revitalization Bill established Financial Reconstruction Commission; LTCB nationalized in October 1998 amid opposition.
  - Russian financial crisis in August raised global uncertainty.
- June 2000
  - Lower House elections reduced ruling parties’ share from 66 to 56 percent; LDP lost majority but government maintained control.
- February–March 2001
  - BOJ cut policy rate by 10 bps from 0.25 to 0.15 percent on February 28.
  - March 19: BOJ introduced QE, switching target from overnight call rate to current account balance.
  - PM Mori’s approval fell below 10 percent; reported imminent resignation; Koizumi won LDP presidency in April.
- July 2001
  - PM Koizumi won Upper House election; administrative reform and debate between fiscal consolidation and fiscal expansion intensified.
  - CEFP produced first annual report, Honebuto-no-Hoshin, on June 26.
- October 2002
  - Conflict over NPL issues; PM Koizumi dismissed Hakuo Yanagisawa and appointed Heizo Takenaka.
  - Takenaka Plan forced stricter asset evaluation, higher bank reserves, compression of deferred tax assets; public injections for insolvent banks (e.g., Resona Bank, Ashikaga Bank).
  - BOJ eased policy: increased current account balances from JPY 10-15 trillion to 15-20 trillion and raised long-term JGB purchases from JPY 1 trillion to 1.2 trillion per month.
- March 2008
  - Twisted Diet uncertainties: failed effort to pass a gasoline tax law before fiscal year end; repeated rejections of BOJ Governor nomination.
  - Deputy Governor Masaaki Shirakawa became Acting Governor on March 20 and Governor on April 9.
- September–October 2008
  - Lehman Brothers bankruptcy in September 2008 precipitated global financial crisis.
  - Japanese economy shrank by 12.4 and 15.4 percent (saar) in Q4 2008 and Q1 2009, respectively.
  - Nikkei fell below 7,000 from around 12,000 pre-bankruptcy; yen appreciated from around 120 in 2007 to 93 per dollar by end-2009.
  - Japanese government enacted two supplementary budgets in October 2008 and January 2009; BOJ reduced policy rate twice in October and December 2008.
- February 2009
  - Government estimated Q4 2008 growth at negative 12.4 percent, the largest decline in about 35 years.
  - Yen depreciated by around 10 percent.
  - US stimulus (American Recovery and Reinvestment Act of 2009) deliberations and delays contributed to uncertainty.
- May–June 2010
  - European sovereign debt crisis accelerated after first plan to rescue Greece in May 2010; Nikkei lost more than 10 percent in May.
  - Yen briefly went below 90 per dollar.
  - PM Hatoyama resigned in June; anticipation of a Twisted Diet after July Upper House elections contributed to EPU rise.
- August 2011
  - Delay in approval of law allowing deficit‑financing bonds for FY2011 until August raised fiscal cliff risks.
  - Twisted Diet caused by loss of Upper House majority in July 2010.
  - European debt crisis intensified; S&P placed US sovereign debt on negative watch in July and downgraded in August; US debt ceiling conflict recently resolved but confidence in US fiscal policy remained weak.
  - Japanese FX intervention responded to resumed appreciation after March 2011 earthquake.
- June 2012
  - Ruling DPJ and LDP accelerated tax and social security reform discussions; agreement on June 21 included two-step consumption tax increase in April 2014 and October 2015.
  - Elevated uncertainty due to conflicts within DPJ on reforms.
- January–February 2016
  - BOJ announced negative interest rates on January 29.
  - Leading up: speculation about BOJ policy amid declining inflation momentum, emerging market risks, and yen appreciation.
  - Negative rate decision surprised markets; Governor Kuroda had earlier ruled out negative rates.
  - Decision lowered the whole yield curve and initially depreciated the currency; effect on exchange rate was short-lived and inflation expectations continued to decline.
  - Concerns about negative rates’ implications for banking sector, limited policy space/effectiveness, and public unease about returns on bank deposits.
- May–June 2016
  - Brexit concerns and eventual June 23 referendum result caused substantial yen appreciation.
  - BOJ in July doubled ETF purchases and expanded US dollar lending program to ensure smooth funding conditions.
  - Rising uncertainty about whether PM Abe would postpone consumption tax hike scheduled for April 2017; mixed reports and speculation up to decision on June 1.
  - Postponement of the tax hike was welcomed by public but raised uncertainty about meeting primary balance goal by 2020.
- November 2016–January 2017
  - Policy uncertainty rose ahead of the US election due to US’s role as major trading partner and implications for TPP.
  - Election of Donald Trump increased uncertainty about US fiscal, trade, and other policies; US withdrew from TPP on January 23, 2017.

### Policy responses and mechanisms observed
- BOJ actions
  - Liquidity injections during market stress (e.g., October 1987).
  - FX intervention by Ministry of Finance during sharp yen appreciation episodes.
  - QE introduced March 19, 2001 (switch target to current account balance).
  - Increased current account balances and JGB purchases during October 2002 measures (from JPY 10-15 trillion to 15-20 trillion; JGB purchases from JPY 1 trillion to 1.2 trillion per month).
  - Two policy rate cuts in October and December 2008; policy rate cut of 10 bps from 0.25 to 0.15 percent on February 28, 2001.
  - Negative interest rate policy announced January 29, 2016.
  - Post-Brexit July 2016: doubled ETF purchases and expanded US dollar lending program.
- Fiscal policy responses
  - Tax break of JPY 2 trillion financed by JGB issuance in November 1997.
  - Two supplementary budgets in October 2008 and January 2009.
  - Agreement for two-step consumption tax increases in April 2014 and October 2015 (reached June 21, 2012).
- Political processes
  - Twisted Diets (no majority in Upper House) repeatedly associated with heightened EPU and delays in fiscal legislation (e.g., 1998, 2008, 2011).
  - Leadership changes and election losses often coincided with spikes in EPU (e.g., 2000, 2001, 2008, 2016).

### Quantitative highlights and recorded magnitudes
- EPU episodes defined as movements > 1.64 standard deviations from average.
- BOJ policy rate cut: 10 bps from 0.25 to 0.15 percent on February 28, 2001.
- Tokyo Kyodo Bank related capital and rescue events involved BOJ capital contributions (no dollar/JPY amounts beyond context mentioned).
- Tax break provided in November 1997: JPY 2 trillion (financed by JGB issuance).
- BOJ balance and JGB purchase changes in October 2002:
  - Current account balances increased from JPY 10-15 trillion to 15-20 trillion.
  - Long-term JGB purchases increased from JPY 1 trillion to 1.2 trillion per month.
- Japanese GDP declines during global financial crisis:
  - Shrinkage of 12.4 and 15.4 percent (saar) in Q4 2008 and Q1 2009, respectively.
  - Government published Q4 2008 growth estimate at negative 12.4 percent.
- Yen exchange rate movements:
  - Around 120 per dollar in 2007 to 93 per dollar by end-2009.
  - Yen briefly below 90 per dollar in May–June 2010.
  - Yen had been below 80 against the dollar since mid-July 2011.
  - Yen depreciated by around 10 percent in February 2009 (contextual).
- Stock market magnitudes:
  - Nikkei fell to below 7,000 from around 12,000 after Lehman bankruptcy.
  - Nikkei lost more than 10 percent in May 2010.
- Political approval and electoral magnitudes:
  - Ruling parties’ share fell from 66 to 56 percent in June 2000 Lower House elections.
  - PM Mori’s approval rate fell below 10 percent prior to his resignation.
- Timeline dates preserved exactly when specified (e.g., October 20; March 19; January 29; June 23; January 23, 2017).

*Source: wp17128 (PDF chapter/section).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17128.pdf_