Bond Vigilantes and Inflation Do Domestic Bond Markets

“Bond Vigilantes and Inflation: Do Domestic Bond Markets Promote Price Stability? ” Andrew K. Rose Berkeley-Haas, ABFER, CEPR, NBER Mark M Spiegel Federal Reserve Bank of San Francisco 9 th Int’l MIFN Conference, Osaka, Japan, October 29, 2015. Our views are our own and do not necessarily represent those of the Federal Reserve Bank of San Francisco or the Board of Governors of the Federal Reserve

Motivation: Bond Vigilantes “I used to think if there was reincarnation, I wanted to come back as the president or the pope or a . 400 baseball hitter. But now I want to come back as the bond market. You can intimidate everybody. ” – James Carville, Wall Street Journal (February 25, 1993, p. A 1)

Impact of Bond Markets on Inflation • Do (long, local-currency) bond markets help keep inflation low and stable? • Theory: Ambiguous. – Government incentive to inflate away debts – Bond-holders lobby for low inflation • Empirics: Qualified yes – Nominal bond markets in IT countries associated with 3 -4 percentage points lower inflation

Endogeneity issues • Concern that low inflation necessary for local nominal bond markets • Not necessary in data – Just in OECD: Australia, Canada, Denmark, Greece, Ireland, Italy, New Zealand, Spain, UK all inflated >15% with nominal bonds • Still, common cause could cause both inflation fall and bond market creation Long Bond Markets 4

Addressing endogeneity issues • IT regimes – Countries “ready for bonds” – Robustness: also check other regimes • Treatment estimators – Bond market launches timed to coincide with inflation fall • Instrumental variables – Inflation stability [Burger Warnock (2006)] – Political and fiscal IVs • Results all similar to least squares

Theory • 3 agent types – atomistic households – Government – Anti-inflation lobby • Stylized 1 -period model – Government sells debt to households to finance – Households make portfolio decisions – Lobby acts on behalf of nominal bond-holders – Government chooses inflation rate

Households • N atomistic households • Endowed with nominal assets from poorest, to richest • Assets can be held as cash or bonds – Fixed fee to enter bond market – After paying fee, can hold domestic or foreign bonds – Interest parity holds in expectations • Economies of scale in entering bond market – Only “rich” enter securities market

Anti-inflation Lobby • Domestic currency bond purchasers contribute to anti-inflation lobby • Each contributes times exposure • endogenous • Chosen to optimize expected wealth of domesticcurrency bondholders • Interest rate parity condition satisfies • Perfect foresight

Government • Government expenditures equal to G • Issues bonds and services debt at end of period • Finances debt service through taxes and seigniorage • Taxes T/N levied equally across households • Seigniorage satisfies • Government values wealth of median voter and transfers from lobby • where measures relative weight on transfers

Benchmark Case: No domestic bond market • Government issues foreign currency securities • Solution for domestic inflation satisfies – Increased implies lower tax burden – Increased implies higher inflation tax – Increased implies higher seigniorage

Domestic bond market with lobbying • Follow Acemoglu, et al (2008) in assuming antiinflation lobby offers credible contract pair • Lobby makes take-it-or-leave it offer • Constraint is to leave government as well off as it would be with no agreement • Solve for this off-equilibrium outcome,

Equilibrium Solution • Lobby solves problem • Inflation solution • satisfies

Proposition 1 • PROPOSITION 1: Realized inflation is lower with a domestic currency bond market relative to a foreign currency bond market if and only if

Impact on distribution of pre-tax wealth

Proposition 2 • PROPOSITION 2: When the introduction of a domestic currency bond market reduces inflation, it leaves the distribution of end-of-period pre-tax wealth more equal if is sufficiently close to. In particular, a sufficient, but not necessary condition for wealth to equalize is.

Theory summary • Nominal bonds expose politically powerful wealthy to inflation tax – Induces lobbying against tax – Under proper conditions, inflation may fall • If so, distribution of income becomes less skewed • Empirical implications – Launch of nominal bond market may reduce inflation – Take this to data for inflation-targeting countries Long Bond Markets 16

Empirics • Many types of debt: public/private, long/short maturity, nominal/real, LCU/FX … – Hence, new ones added: • Polish 10 -year fixed rate government bonds 1999 • Korea 2000 … • Use panel with fixed effects, within estimation – Can think of as difference-in-difference • Focus on Inflation Targeting Regimes – Examine hard fixes, others monetary regimes Long Bond Markets 17

Econometric Issues • Use variety of estimators as sensitivity checks – Event study – Matching – IV (political and fiscal instruments) • Also examine introduction of other bonds (for negative responses) – Indexed – FX bonds Long Bond Markets 18

Estimation Strategy • Conventional LS panel with FE (Δ in Δ): πit = βBondit + γXit + {δi} + {εt} + ηit • Two measures of inflation: 1) CPI; 2) GDP • Five controls (X): – 1) polity; 2) log real GDP p/c; 3) log population; 4) trade (% GDP); 5) growth Long Bond Markets 19

Handling Potential Simultaneity 1. Restrict attention to IT regimes – Use only countries “ready for bonds” – Robustness: check other regimes 2. Treatment estimators to handle selection – Countries may choose time to create bond markets (expected inflation fall) 3. Instrumental variables – Inflation stability [Burger Warnock (2006)] – Political and fiscal IVs Long Bond Markets 20

Data Issues • Bond Markets – Start with Global Financial Data fixed income database: “recorded electronically for current and historical markets covering 200 countries. GFD provides complete yield curve coverage with data on Interbank Rates, Swap Rates, Treasury-Bill Yields and Long-term Government Bond Yields. The Fixed Income Database enables you to follow changes in yields over different maturities going back several decades using yields at 3 months and 10 years, as well as maturities between and beyond these benchmarks. GFD provides data from both the public sector and the private sector. ” https: //www. globalfinancialdata. com/Databases/Fixed. Income. Database. html. Long Bond Markets 21

Which Bonds? • Government: corporates always later • Maturity: at least one decade (benchmark, outside horizon current monetary policy, responsive to inflationary expectations) • Nominal, Local-Currency – Use indexed, foreign-denominated for robustness Long Bond Markets 22

Coverage • GFD: 819 government bonds from 105 countries – 70 countries with bonds of maturity ≥ 10 years Long Bond Markets 23

Supplementary Sources Used • Bloomberg, Financial Times: 20 countries 10 year government bonds (all in GFD) • BIS: 28 countries with “long-term” domestic bonds (maturity more than one year!) • Investing. com: 59 countries with bonds of maturity ≥ 10 years • Dealogic: 73 countries covered Long Bond Markets 24

Other Series • • Inflation from WDI (also macro controls) Polity 2 from Polity IV project Hard Fix regimes: Ilzetzki, Reinhart and Rogoff IT regimes from Rose (2013) – Basically Mishkin (2004): 1) medium-term numerical inflation target; 2) institutional commitment to price stability as primary monetary goal; 3) information-inclusive instrument strategy; 4) CB transparency; 5) CB accountability Long Bond Markets 25

Summary • Annual data, >200 countries, 1970 -2012 • Main Focus: (32) Inflation Targeters – 4 without bond markets – 8 with long-established bond markets • Australia, Canada, New Zealand, Norway, South Africa, Sweden, Switzerland, and the UK – 13 established bond markets during IT • Armenia, Brazil, Chile, Colombia, Czech Republic, Iceland, Indonesia, Israel, Korea, Mexico, Peru, Romania, and Turkey – (7 created bond markets shortly before IT) • Variation creates identification Long Bond Markets 26

Descriptive Statistics With Bond Market No Bond Market Test for Equality Mean Std Dev Obs CPI 5. 5 32. 1, 108 56. 682. GDP 5. 7 30. 1, 146 63. 638. Inflation Long Bond Markets 2, 967 Mean (t) 2. 5* Std Dev (F) 443** 3, 650 3. 0** 459** 27

More by Monetary Regime Inflation with a bond market Inflation Targeting Hard Fix Neither: Sloppy Center Mean Std Dev Obs CPI 3. 2 277 5. 9 54. 381 6. 6 10. 4 412 GDP 3. 6 2. 9 294 6. 1 50. 383 6. 6 9. 5 421 Inflation without a bond market CPI 7. 2 4. 0 69 6. 6 22. 999 95. 951. 1, 489 GDP 8. 9 5. 7 71 13. 1 169. 1, 229 114. 933. 1, 596 Tests for Equality: presence/absence of bond markets (t- and F-tests for Means/Std. Dev) CPI 11. 4** 3. 4** . 3 . 2 1. 9 8300** GDP 11. 0** 3. 9** . 8 11. ** 2. 4* 9600** Long Bond Markets 28

Histograms and Scatters Long Bond Markets 29

Event Studies Long Bond Markets 30

Benchmark: Effect of Bond Market on Inflation Targeters Hard Fixers Other Monetary Regimes CPI Inflation -2. 9** (1. 0) GDP Inflation -4. 4** (1. 1) 7. 3 (7. 7) 74. (53. ) . 6 (13. 7) 136. (83. ) Coefficients for dummy variable (=1 if bond market exists, =0 otherwise). Robust standard errors (clustered by country). Each cell is the result of a single panel regression of inflation on bond market presence with comprehensive time- and country-specific fixed effects unless otherwise indicated. Control covariates included: a) polity; b) log real GDP per capita; c) log population; d) trade, %GDP; and e) demeaned real GDP growth. Default includes annual data for up to 32 IT countries, 1991 -2012 (up to 116 hard fixers, 1987 -2012; up to 129 others, 19872012). Long Bond Markets 31

Key Result • For Inflation Targeters, inflation is 3 -4% lower when (long, nominal, local-currency) bond market exists – Not true of different monetary regimes • Next: is this robust? Long Bond Markets 32

Sample Sensitivity CPI Inflation Drop pre-1995 Drop post-2006 Drop Poor (real GDP p/c < $10 k) Drop Rich (real GDP p/c > $40 k) Drop Small (population <10 m) Drop Large (population > 100 m) Drop >|2. 5σ| outliers GDP Inflation -2. 6** (1. 0) -4. 7** (1. 2) -5. 4** (1. 0) -2. 9** (1. 0) -2. 8* (1. 0) -1. 8 (1. 2) -2. 8** (. 6) -4. 1** (1. 1) -6. 4** (1. 3) -6. 5** (1. 2) -4. 5** (1. 1) -4. 5** (1. 5) -4. 8** (1. 6) -4. 4** (. 6) Long Bond Markets 33

Estimator Sensitivity Conventional standard errors Random (not fixed) country effects Drop time effects Drop covariates CPI Inflation -2. 9** (. 5) -3. 2** (1. 0) -3. 2** (. 9) -2. 6** (. 9) -2. 8* (1. 1) Long Bond Markets GDP Inflation -4. 4** (. 7) -4. 5** (1. 0) -3. 8** (1. 0) -4. 6** (1. 2) -4. 5** (1. 3) 34

Robustness of Bond Market Measure CPI Inflation 5 -year lag of bond market, -1. 9** not contemporaneous (. 5) 5 -9 year maturity bonds -4. 7** instead of ≥ 10 years (1. 0) And now … two checks expected to fail … Indexed/Adjusted instead of nominal long bonds Bonds denominated in foreign exchange, not LCU -1. 6 (1. 3) 1. 1 (. 6) Long Bond Markets GDP Inflation -3. 4** (. 8) -4. 2* (1. 7) -2. 7 (1. 9) 1. 4 (. 8) 35

Placebo test: Falsified launch dates don’t affect inflation (cluster around 0) Long Bond Markets 36

Treatment Effect Estimation • Match bond-market observations to those without bond markets (IT countries) – Use propensity score, nearest neighbor, other matching techniques – Five control covariates for matching model • Treatment effect of bond market for inflation between 3. 3%-5. 1% – Economically, statistically significant – Close to panel estimates Long Bond Markets 37

Average Treatment Effects (of Long Bond Market on Inflation, ITers) CPI Inflation GDP Inflation Propensity Score Matching (three matches) -3. 6** (1. 2) -3. 6** (. 9) Nearest-Neighbour Matching (three matches) -3. 9** (. 6) -5. 1** (. 8) Regression Adjusted -3. 8** (. 8) -4. 2** (1. 0) Inverse-Probability Weighted -3. 8** (. 6) -4. 4** (. 8) Inverse-Probability Weighted with Regression Adjustment -3. 3** (. 9) -3. 7** (. 9) Augmented Inverse-Probability Weighted -3. 6** (. 8) -4. 1** (1. 0) Long Bond Markets 38

IV Estimation • Potential simultaneity/measurement error for existence of bond market • Begin with two IVs: – Government spending (relative to GDP) – Years since national independence (log) – Motivation: mature, larger governments have more institutional capacity and greater need for bonds Long Bond Markets 39

Instrumental Variables CPI Inflation Instrumental Variables Log Ind, Gov Exp (%GDP) Log Years Indep Historical Inf Volatility Historical Inf Vol, Budget (% GDP) Historical Inf Ave, Budget (% GDP) Least Squares Hausman |t| for β Excludable? (p-value) -8. 0** (3. 0) -5. 8* (2. 4) -5. 2** (1. 8) 1. 7 . 12 1. 7 . 05* 1. 2 . 20 1. 4 . 07 -4. 8* (2. 1) 1. 0 . 08 Obs. Weak IV? (p-value) βIV 353 . 00** 332 . 00** 283 . 00** 353 GDP Inflation -2. 9** (1. 0) Hausman |t| for β Excludable? (p-value) -8. 7** (3. 7) -8. 4** (3. 7) -3. 0 (2. 8) -3. 2 (2. 0) 1. 2 . 23 1. 1 . 24 . 5 . 62 . 7 . 85 -2. 8 (2. 4) . 7 . 95 βIV -4. 4** (1. 1) Long Bond Markets 40

Instrumental Variables • Not weak instrumental variables • Estimates economically and statistically large – Hausman tests not rejected – IVs excludable, except for years indep. • Independence critical; spending not • Inflation history matters • In the large, same results as least squares! Long Bond Markets 41

IV Sensitivity CPI Inflation Instrumental Variables Obs. Weak IV? (p-value) βIV Log Indep, Sec Eff, Legit Log Dem, Gov Exp (%GDP) Log Years Democracy Civil Liberties 329 . 00** 183 . 01** 183 . 00** 353 . 00** Democracy Dummy Least Squares 353 . 01* -4. 1* (1. 8) -4. 4 (4. 1) -6. 5* (2. 6) -7. 9 (5. 2) -2. 9** (1. 0) 353 GDP Inflation Hausman |t| for β Excludable? (p-value) . 8 . 11 . 9 . 61 . 9 . 36 1. 4 . 12 1. 0 . 27 Long Bond Markets βIV -7. 1** (2. 4) -7. 9* (3. 8) -7. 7* (3. 8) -8. 6* (4. 3) -5. 8 (5. 2) -4. 4** (1. 1) Hausman |t| for β Excludable? (p-value) 1. 3 . 14 1. 7 . 14 1. 6 . 06 1. 0 . 3 . 78 42

Summary • Political economy model – Domestic bond market may lower inflation and reduce skewness of wealth distribution • Confirmed in empirics – Existence of a long nominal local bond market lowers inflations ≈3% for Inflation Targeters • Untrue of indexed or foreign-denominated bonds • Untrue of other monetary regimes – Robust to sample/estimator • Matching and IV estimators deliver same message 43

Additional slides Long Bond Markets 44

Robustness to IV Choices • Replace government spending with security effectiveness and legitimacy (Systemic Peace) • Replace years of independence with years of democracy (Acemoglu et al, 2014) – Small sample: only recent democracies • Civil liberties (1 -7) from Freedom House • Dummy for Democracy (FH) • Find: a) all strong IVs; b) all negative IV estimates, 10/14 significant; c) no Hausman rejections; d) excludable Long Bond Markets 45

Bond Markets and Business Cycles (more checks expected to fail) Real GDP detrender Baxter-King Christiano-Fitzgerald Hodrick-Prescott Growth Business Cycle Deviation from Trend. 001 (. 005) -. 002 (. 006). 63 (1. 57) Long Bond Markets Absolute Business Cycle Deviation from Trend -. 005 (. 004) -. 000 (. 007) -. 010 (. 006) -1. 92 (. 99) 46