核心问题
当私人需求疲弱或冲击高度不均时,财政政策应支持什么、触达谁、又能多精准?
- 政策分析
- 经济史
当宏观问题不只是信贷价格,而是收入水平和收入分配时,财政政策尤其重要。
这使财政政策既有力又复杂。它可以直接推动需求,但每项措施都有分配后果、执行延迟和政治取舍。
政策目标
当私人需求疲弱或冲击高度不均时,财政政策应支持什么、触达谁、又能多精准?
政府支出、税收、转移支付和自动稳定器通过居民收入、企业需求和公共投资发挥作用,而不是通过单一政策利率。
财政政策可以直接触及实体经济,但质量取决于速度、定向程度、行政能力,以及支持是否真正到达约束所在。
工具与手段
政策真正可以调动的杠杆。每种工具都有各自的操作目标、信息需求和政治成本。
政府支出
公共采购和投资直接支撑需求,如果用于基础设施、产能或韧性建设,也能改变供给侧。
税收与转移支付
税收调整、退税和转移支付会改变可支配收入,并可指向边际消费倾向较高的家庭或行业。
自动稳定器
失业保险和累进税制等项目可以在不等待新立法方案的情况下缓冲经济下行。
公共资产负债表选择
赤字融资、债务期限和支持方式会影响财政行动是在稳定需求,还是更快进入长期债务争论。
传导渠道
政策工具影响价格、产出、就业、信贷、对外收支或产能的渠道。
需求与收入
政府支出和转移支付无需经过贷款机构就能进入经济。政府购买本身就是需求。转移支付和减税通过家庭预算发挥作用,其效果取决于谁拿到这笔钱:2020年的现金补助在疫情初期大幅提高了支出,后几轮的效果则小得多,高收入家庭尤其如此。
公共借款与信贷成本
赤字通过出售政府债务来融资。当经济存在闲置产能、央行保持利率不变时,这种借款未必会推高私人借款人的成本。接近充分就业时,它会与私人部门争夺储蓄,部分财政刺激会被更高的利率抵消。
进口与外部平衡
每多出一美元的需求,就有一部分花在进口上,这就是本页乘数公式中的漏出项。因此,财政扩张往往会扩大贸易逆差;根据部门收支恒等式,更大的公共赤字必然对应更多的私人储蓄或更多的对外借款。
时滞
效果究竟何时显现。有些价格会立即调整;数量、合同、招聘和投资通常反应得更晚。
Authorization
Legislation, budget rules, and program design determine whether the tool arrives before the downturn has already changed shape.
Disbursement
Transfers can move quickly; infrastructure, procurement, and tax administration usually move more slowly.
Multiplier window
The demand effect depends on slack, monetary accommodation, household liquidity, import leakage, and whether recipients spend or save.
制度约束
决定一项技术上自洽的政策能否实际使用的法律、政治、资产负债表和操作层面的限制。
Budget process
Fiscal policy works through legislatures, agencies, tax systems, and eligibility rules, not one committee vote.
Debt capacity
Borrowing space is a market and political constraint. It is easier to use when credibility was built before the crisis.
Administrative reach
A policy cannot target what the state cannot identify, verify, or deliver to at scale.
失灵方式
政策可能失灵、来得太晚或作用于错误边际的各种方式。
Mistargeting
Broad support can stabilize demand but still miss the households, sectors, or regions where the constraint is binding.
Arriving after the cycle turns
Slow fiscal measures can become procyclical if they land after private demand has already recovered.
Crowding the supply side
If the economy is supply-constrained, extra demand can raise prices more than output.
取舍
任何有用的干预都会牵动别的东西。
广而快,还是准而慢
向每个家庭发放的款项可以很快送达且不漏一人,但其中很大一部分会落到会把钱存起来的人手中。把援助集中给手头拮据的家庭,每一美元带来的支出更多,但需要资格规则和登记记录,而建立这些需要时间。设计上的取舍是速度与精准之间的权衡。
立法与执行时滞
一揽子财政计划需要起草、经立法机构通过,再由各机构支出,因此可能在它所针对的衰退过去之后才到位。2009年美国刺激计划的设计者预计,新增政府购买的效应会一直累积到第八个季度。自动稳定器绕过了立法时滞:在2012财年,它们在没有任何新表决的情况下,使联邦赤字增加了相当于潜在GDP 2.3%的数额。
现在支持谁,以后谁来付账
每一项财政措施都会按收入、地区或行业选择受益者,因此稳定效应和分配效应无法分开。借债还会把成本在时间上转移。如果利率始终低于增长率,债务可以展期滚动而无需日后加税;否则,账单就由未来的纳税人承担。
需要监测的数据
关闭本页之后,继续追踪那些可能推翻这一诊断的数据序列。
- Budget impulseDeficit indicatorsTrack deficits, cyclically adjusted balances, and primary balances rather than the headline deficit alone.
- Household cash flowDisposable incomeUse income, transfer, saving-rate, and delinquency data to see whether support is reaching the binding constraint.
- Real activity responseGrowth indicatorsCompare spending, employment, real GDP, and sectoral output before assigning the fiscal shock a single multiplier.
历史案例
检验这项政策诊断的真实案例。
2009
ARRA and the fiscal multiplier debate
The post-GFC stimulus made timing, state aid, infrastructure pace, and multiplier measurement central again.
2020-2022
Pandemic policy mix
Fiscal scale, emergency lending, supply disruption, and reopening demand all collided. It is now the benchmark case for how hard it is to separate support, overshoot, and timing.
2022-2023
Energy-price relief in Europe
Governments tried to cushion household energy shocks without fully muting the price signal needed for adjustment.
关键恒等式
本页其余部分所依据的方程。
把恒等式当作检验政策逻辑的工具,而不是指望一个方程就能决定整个政策体制。
Static Keynesian multiplier
The textbook closed-form: \(c\) is the marginal propensity to consume, \(t\) is the average tax rate, \(m\) is the marginal propensity to import. Each leakage shrinks the multiplier.
Debt dynamics
Every term is a ratio to the same period's GDP: \(d_t\) is debt at the end of period \(t\) over period-\(t\) GDP, and \(pd_t\) is the primary deficit over period-\(t\) GDP, positive for a deficit and negative for a surplus. The second form is the usual approximation when rates are small. Compare nominal \(r\) with nominal \(g\), as the simulator below does, or real with real. Setting a real rate against nominal growth subtracts inflation once by mistake. The gap \((r-g)\) decides whether the debt ratio rises or falls on its own, which is why most fiscal arguments turn on it.
Sectoral balances identity
Government deficits, private net saving, and the external balance are linked by accounting. With these signs, one sector's deficit is another sector's surplus.
What we know about the size of the multiplier
The multiplier debate is not whether fiscal stimulus moves output. It is by how much, and under what conditions. The memo that Christina Romer and Jared Bernstein wrote for the incoming administration before the 2009 stimulus [1] assumed a multiplier on government purchases that built up to 1.57 by the eighth quarter and a tax-cut multiplier that reached 0.99, both for the case where the federal funds rate stays put near zero. Others read the evidence as much weaker. Robert Barro and Charles Redlick estimated a multiplier on temporary defense spending of 0.4 to 0.5 within the year and 0.6 to 0.7 over two years [2]. Valerie Ramey, reviewing the literature in 2011, concluded that the multiplier for a temporary, deficit-financed increase in government purchases is probably between 0.8 and 1.5 [3].
The state-dependence literature changed the question. Auerbach and Gorodnichenko estimated the response of output to government purchases in a panel of OECD countries, letting it differ between recessions and expansions [4]. In recessions the multiplier averaged about 2.3 over three years. In expansions the response was much weaker and not statistically different from zero. Their earlier estimates for the United States were about 1.5 to 2.0 in recessions and approximately zero in expansions. The mechanism is intuitive: when there is unused capacity, public demand adds to private demand instead of displacing it. When the economy is at potential, public demand bids resources away from private uses. The same instrument has different consequences depending on the regime.
The pandemic relief of 2020 and 2021 tested both views. Chetty, Friedman, Stepner, and the Opportunity Insights team tracked spending close to real time using data from private companies [5]. They found that cash stimulus payments led to sharp increases in spending early in the pandemic but much smaller responses later, especially for high-income households. Composition mattered: the same dollar of stimulus did more when it reached a household that was short of cash.
Why automatic stabilizers carry so much of the load
Discretionary stimulus gets the headlines. Automatic stabilizers act first, and they act without a vote. When income falls, tax liabilities fall with it, and outlays such as unemployment insurance rise. The Congressional Budget Office estimates that these effects added $386 billion to the federal deficit in fiscal year 2012, an amount equal to 2.3 percent of potential GDP, and that 2012 was the fourth year in a row in which they added 2.0 percent or more [6]. Progressive income taxes, unemployment insurance, food assistance, and Medicaid all work this way.
Cross-country evidence shows how much the design matters. Dolls, Fuest, and Peichl simulated income and unemployment shocks through the tax and transfer systems of the European Union and the United States [7]. Automatic stabilizers absorbed 38 percent of a proportional income shock in the EU against 32 percent in the United States, and 47 percent of an unemployment shock against 34 percent. Within Europe the range is wide: stabilizers in Eastern and Southern Europe are much weaker than in Central and Northern Europe.
The policy implication runs against the conventional wisdom. Some of the most consequential stabilization decisions are not the packages voted during a recession. They are the standing choices about the tax schedule, the generosity of unemployment insurance, and the reach of means-tested transfers. The design can also weaken. Bitler and Hoynes find that after the 1996 welfare reform, cash assistance did not respond in the Great Recession, while food stamps and unemployment insurance did [8]. The protection a household gets in the next recession is being decided now, in the rules of those programs.
The Reinhart-Rogoff debate, fully ventilated
Carmen Reinhart and Kenneth Rogoff's 2010 paper, "Growth in a Time of Debt," argued that real GDP growth fell sharply once gross government debt crossed 90 percent of GDP. The paper was widely cited during the 2010-2013 austerity push. Olli Rehn, the European Commissioner for Economic and Monetary Affairs, cited it in a 2013 letter to euro-area finance ministers. Paul Ryan cited it in his House Budget Committee report. The 90 percent threshold became a political number.
In 2013 Thomas Herndon, Michael Ash, and Robert Pollin of the University of Massachusetts Amherst replicated the Reinhart-Rogoff results and found three problems [9]. A spreadsheet error left five countries out of the average. Some of the available data had been excluded. And the summary statistics were weighted in an unconventional way. Properly calculated, countries with public debt above 90 percent of GDP averaged 2.2 percent real growth, not the minus 0.1 percent originally published. Growth above the line was, in their words, not dramatically different from growth below it, and the relation varied by country and by period.
The right reading is not that fiscal sustainability is irrelevant. It is that there is no debt-to-GDP threshold above which growth collapses. Sustainability depends on the gap between the interest rate and the growth rate, the maturity profile of outstanding debt, and the credibility of the inflation regime that determines whether nominal debts can be repaid in real terms. Olivier Blanchard's 2019 presidential lecture to the American Economic Association put the first of these at the center: when the safe interest rate is expected to stay below the growth rate, a government may be able to roll its debt over without a later increase in taxes, although debt can still carry a welfare cost by reducing capital accumulation [10]. Whether the condition holds is an empirical question, to be checked country by country and year by year.
Sectoral balances: who pays for the deficit
By national-accounts identity, the government's net spending must equal the private sector's net saving plus the foreign sector's net saving. Wynne Godley, writing for the Levy Institute in 1999, used the identity to argue that the U.S. expansion was unsustainable: with fiscal policy and net exports both subtracting from demand, growth required private spending to keep rising relative to income on a record and growing scale [11]. In November 2007 he and his co-authors warned that a significant drop in borrowing was likely in the coming quarters, with severe consequences for growth and unemployment [12]. The recession began the next month.
The identity is not a behavioral claim. It does not say that government deficits cause private saving or the reverse. It says the three balances must sum to zero. The empirical question is which sector adjusts when one of them changes course. If a government cuts its deficit while households and firms are also trying to save more, the external balance is the only one left to move, and if it cannot move enough, income falls until the balances fit.
This is also why household debt matters more than the headline deficit number. Mian, Rao, and Sufi found that the fall in consumption from 2006 to 2009 was unequal, and larger where households were poorer and more indebted: on average spending fell 5 to 7 cents for each dollar of housing wealth lost, and by more in those areas [13]. When households cut spending to repair their balance sheets, another sector has to run a larger deficit or income falls until the balances add up. That is the frame in which the size of the 2009 fiscal response is still argued over.
情景 · 解读政策的方法
在套用政策框架之前,先核对工具、渠道、时滞和失灵方式
工具
在判断政策立场之前,先说清具体使用的是哪种工具。同一个目标可以通过利率、支出、税收、监管、沟通或资产负债表操作来实现。
传导
追踪政策变动如何经由家庭、企业、银行、市场、预期、汇率和公共部门资产负债表传导。
时滞
区分宣布、实施、市场反应、实体经济反应以及数据发布的时间。
失灵方式
说明哪些情况会让政策适得其反、生效太晚、效果外溢到国外、只是转移风险,或无法触及受约束的部门。