为什么老龄化会导致通货紧缩和长期停滞.pdf

Why aging induces deflation andsecular stagnationR.Anton Braun(Federal Reserve Bank of Atlanta)Daisuke Ikeda(Bank of Japan)August 1,20231/26Deflation,secular stagnation,and govt policySummers(2020,IMF):In the post-crisis period,many countries had1Deflation:low and in some cases negative inflation rates2Low real interest rates3Below trend GDP growth Secular stagnation despite expansionary monetary and fiscal policyOur questions1What was causing deflation and secular stagnation?2Are deflationary and stagnationary forces over?Our answer:1.Aging,and 2.No2/26Why aging?and What we doAging:a higher fraction of old individuals out of labor force Greater stock of saving;greater demand for liquid assets A decrease in the price levelNominal debtP=Demand for liquid assets=New Keynesian narrative:natural rate real ratePropose a quantitative OLG model with rich nominal-real interactionsFeed projected Japanese age distribution into the modelOur model generates deflation and secular stagnation3/26Why Japan?Rapid aging from 1985 to 2020Old age dependency ratioNote:The ratio of the 65+population to the 20-64 populationSource:World Bank4/26Aging is projected to accelerateOld age dependency ratioNote:The ratio of the 65+population to the 20-64 populationSource:World Bank5/26Why Japan?Deflation and secular stagnationNote:Model is Braun,Ikeda,and Joines(2008)Between 1990 and 2020,secular declinesand persistently low levels of1Real interest rate2Per capita GDP3Inflation rate6/26More data facts:government policy reactionsNote:Shadow rate is Ueno(2017).1Monetary policy:large,persistent decline in the nominal interest rateand unconventional monetary policy2Fiscal policy:large,persistent increase in the debtGDP ratio7/26Road map1Quantitative OLG modelOverviewDemand theory of the price level(Hagedorn,2021)Model performance in the short run(Braun and Ikeda,2021)2Main results:quantitative effects of agingPartial equilibriumGeneral equilibriumAnatomy of aging-driven secular stagnationRobustness analysis8/26Quantitative OLG model:firms and govtThe model is based on Braun and Ikeda(2021)Firms:Intermediate goods firms and final good firmsCapital and labor inputNominal price rigidity(Rotemberg,1996)Government:monetary and fiscal authoritiesTaylor rule for the nominal interest rateNominal debt;taxes;public pensions9/26Quantitative OLG model:householdsHouseholds:overlapping generations aged 21-120Representative cohortsMortality riskDeath event known at beginning of final period of lifeNo accidental bequestsAsset demandIlliquid assets(capital;equity)liquid assets(private ious,gov.bonds)Convex costs of adjustment on illiquid assetsNatural borrowing constraintLabor supplyAge profile of efficiency units of work is humpshapedWorking households join a labor unionHours worked identical for all workers10/26Household consumption-saving problemAge j household observes death event zj 0,1 and chooses consumptioncj,liquid assets dj,and illiquid assets ajto maximizeUj(aj1,dj1,zj)=maxcj,aj,dj?j(cj/j)11 1+1h1+1j+zj(1 j+1)Uj+1(aj,dj,0)+j+1Uj+1(aj,dj,1)?,subject to(1+c)cj+aj+(aj,aj1,zj)+djRaaj1+Rdj1+(1 w)wjhj+bj+Mandatory retirement:j=0 for j Jr11/26Demand Theory of the Price LevelHagedorn(2021);Hu,et al.(2021)Govt sets the amount of nominal debt Dnevery period=FTPL where govt sets real surplusesDebt market clearing condition:supply=demandDnP=JXj=1Xzj0,1dj(zj)NjGiven Dn,demographics NjJj=1 demand price P12/26Model performance in steady stateSteady-state age profiles of net wort and assetsNet WorthLiquid assetsIlliquid assetsAgeModelDataModelDataModelDataUnder 300.010.65-0.63-0.080.640.7330390.881.60-0.85-0.581.732.1840492.852.580.19-0.312.652.9050595.544.522.230.763.313.7660697.276.293.631.703.644.6070+4.166.010.941.773.224.25Note:Relative to income of households aged 5059Source:Braun and Ikeda(2021)Net worth is humpshapedLeverage:Young borrow liquid assets to purchase illiquid assetsOld have large positive holdings of liquid and illiquid assets13/26Model performance in the short runImpulse responses to a 1%tightening in MP in the impact yearNotes:Cumulative responses in the impact year,HFI identification based on Kubota andShintani(2021).SVAR identification based on Ikeda et al.(2020).Vertical lines are 90%confidence intervals.The source is Braun and Ikeda(2021).14/26Long-run simulation(this paper)One period=1 year;starting from 2014Population by age for years 20142060 from IPSS1Year 2014 ageasset distribution2survival probabilities 201420603birth rates 20142060.Other conditioning assumptionsNominal per capita government debt fixed in all periodsGovernment budget constraint closed by adjusting lumpsum taxCentral bank follows monetary policy rulelog(Rt/R)=0.35 log(Rt1/R)+(1 0.35)2 log(t)15/26Partial equilibrium:asset demand glutChanges in population distribution due to1Aging of babyboomers:initial distribution2Longer life expectancy:higher survival rates3Lower fertility rates:birth rate of households aged 21Year of maximum increase in assets(percentage change from 2014)Demographic ScenarioLiquid assetsYearIlliquid assetsYearAging of Babyboomers19.8320382.322029Longer life expectancy0.6320450.072044Lower fertility rates24.1220656.182067Baseline27.120435.24205316/26General equilibrium:interest rates,inflation and output17/26Why does output decline?(Raw)hours per worker flatHours in efficiency units exhibits steady decline due to agingCapital deepening(real interest rate declines)18/26Anatomy of aging-driven secular stagnationDemographic transition:aggregate labor input h declines in futureFuture real rates are lower:Rk k1h1Increases in demand for liquid and illiquid assetsk RkDemand for liquid assets is particular strong and price level fallsDntPt=demand for liquid assetsMonetary policy:P R.Reaction of MP spreads out theresponse of the price level over time and depresses real interest rates.Asset substitution channel:R demand for k Rk19/26Contribution of nominal rigiditiesCostly price adjustmentFlexible pricesNominal rigidities are not important for secular stagnation20/26Does the reaction of monetary policy matter at all?Suppose instead nominal interest rate is fixed tYes!Severe deflation;real rate initially increases;output increases21/26Unpleasant monetarist arithmeticDeflation followed by inflation(Goodhart and Pradhan,2017)22/26Robustness:impose the effective lower boundSevere deflation;output puzzle;higher debtoutput ratioUMP has been effective(see e.g.Ikeda et al.,2020).23/26Robustness:an increase in nominal govt debtAccommodate higher liquid asset demandDebt-output ratio 1.6 2.1 in 2040.Deflation floor is 1 percent.24/26Robustness:starting yearDemographic transition is starts from 1983 and ELB is imposed.25/26Concluding remarksDemographic transition can induce“secular stagnation”MP rule has large macro effects during the demographic transitionWhy MP matters is asset substitution,not NK channels,not FTPLOur results are consistent with the following narrative:Savings glut;safe asset shortage;weak investment MP and FP have accommodated deflationary pressure induced by aging UMP has been reasonably successful in addressing the ELB An increase in debt-output ratios26/26Thank you!daisuke.ikedaboj.or.jp26/26