Is there a case for a further co-ordinated global fiscal stimulus? Part 3

(IV). When does a fiscal stimulus boost aggregate demand?

A fiscal stimulus is a key weapon in the policy arsenal used to address an undesirable weakening of aggregate demand.  For the policy to make sense, either an increase in public spending on goods and services (public consumption or investment) or a tax cut (an increase in transfer payments) must raise aggregate demand at a given price level, wage, interest rates, exchange rates and other asset prices.  In the textbook IS-LM model this means that the fiscal measure shifts the IS curve to the right in output – interest rate space – there is no full direct crowding out, Ricardian equivalence or Minsky equivalence.

We may still not get any effect on output and employment, even if the IS curve shifts to the right, either because there could be ‘financial crowding out’ through higher interest rates, lower asset prices or a stronger exchange rate or because there is ‘real crowding out’ through scare real resources on the supply side; real crowding out or ‘factor market crowding out’ occurs through rising real wages and other real factor costs, and through rising inflationary pressures.

But unless the fiscal stimulus shifts the IS curve to the right, it achieves nothing at all – we don’t even have to investigate whether there is financial or real crowding out.

(IVa). Ricardian equivalence

Tax cuts

Even if financial markets were perfect, life-cycle theories of consumption would imply that postponing taxes by government borrowing (that is, cutting (lump-sum) taxes today and raising them by the same amount in present discounted value at some later date) boosts aggregate consumption demand because it redistributes resources from people with longer expected remaining life-spans (the young and the unborn (future generations) to people with shorter expected remaining life-spans (the old and those current alive).  Strictly speaking, this requires that the tax cuts (transfer payments) be labour income tax cuts or lump-sum tax cuts or transfer payments accruing to persons (owners of human wealth – the non-tradable present discounted value of future after-tax labour income), rather than tax cuts on the returns to or on the value of non-human, financial and real assets that are traded and owned by those currently alive.

Life-cycle principles imply that, because people try to smooth consumption over the life-cycle, the old will have a higher marginal propensity to consume out of current income windfalls than the young.  The unborn of course don’t consume at all (at any rate prior to conception).

To negate these life-cycle arguments for an expansionary demand effect from tax cuts, the Ricardian equivalence or debt neutrality school assumes (1) that the government always satisfies its intertemporal budget constraint (there is no default risk on public debt) and (2) that aggregate consumption can be viewed as the consumption of a single, representative infinite-lived consumer.  The awkward fact that people are born, live and die is finessed by assuming that everyone is linked to all past and future generations through an unbroken chain of operative intergenerational bequest motives.

Stating the assumptions required for Ricardian equivalence to hold is to deny its relevance.  Postponing taxes through borrowing, without changing their present discounted value, will boost aggregate demand because it redistributes resources from the young to the old, from the unborn to those currently alive and from permanent-income or life-cycle households to households constrained by liquidity and current disposable income (Keynesian households).  The only exception, discussed below, is when households are highly indebted and extremely risk-averse and cautious.

A key point to note is that these aggregate-demand-boosting redistributions can also be achieved without the need for public deficits.  If we can identify the young and the old, the life-cycle consumers and the Keynesian consumers, and if we have a sufficiently rich arsenal of taxes and transfers, we can do balanced-budget redistributions that will boost aggregate consumption.  In addition, a balanced-budget increase in public spending on real goods and services (exhaustive public spending) will boost demand unless households are ultra-Keynesian, with a marginal propensity to consume out of current disposable income of 1.

Proposition 4. Balanced-budget redistribution between households with different marginal propensities to spend out of current income can boost demand as effectively as deficit financed tax cuts.  Examples include the following:

1. An increase in social security retirement pensions financed fully by higher social security contributions by workers and employers (pensioners have a higher mpc than workers).

2. An increase in student grants financed fully through a levy on financial wealth (students are likely to be liquidity-constrained, unlike owners of financial assets).

3. An increase in short-term unemployment benefit financed by a reduction in long-term unemployment benefit (short-term and temporarily unemployed workers are more likely to be liquidity-constrained.)

The consumer-oriented tax cuts and transfer payment increases recommended in the IMF Staff Position Note, ‘Fiscal Policy for the Crisis’ (Spilimergo et. al. (2008)) overlap mostly with what I recommend here (increased unemployment benefits, increases in earned income tax credits and the expansion of safety nets in countries where such nets are limited (e.g. China)).

Increased public spending on goods and services

Even if there is Ricardian equivalence for tax cuts or increases in transfer payments, a temporary increase in public spending on goods and services (exhaustive public spending) will stimulate demand.  The reason is that a one-year (say) increase in public consumption or investment of $1 bn will, reduce permanent income by much less than $ 1bn – to a reasonable approximation, private consumption would only fall by an amount given by the product of the time preference rate and $ 1 bn – maybe by $ 30 mn or so. In the Ricardian view, a permanent increase in exhaustive public spending would not boost aggregate demand, as it would lower permanent income and thus private consumption by the same amount as the permanent increase in public spending.

If the Keynesian consumption function with its liquidity-constrained consumers describes reality, a balanced-budget increase in public consumption or investment spending (funded with higher taxes or lower transfer payments) would boost aggregate demand.

Proposition 6: A temporary increase in public consumption or investment will always boost public spending, even if the budget is kept balanced. If there are liquidity-constrained (Keynesian) households, even a permanent balanced-budget increase in public spending on goods and services will boost aggregate demand.

(IVb) Minsky equivalence

When households are highly indebted, face an uncertain employment and labour income future and are afflicted by particularly strong risk-aversion and caution (that is, when they are suffering from a complete collapse of consumer animal spirits), increases in current disposable income, including those associated with tax cuts or higher transfer payments may be saved virtually in their entirety.  Under such conditions household consumption is constrained from below by a socially defined ‘subsistence’ level of consumption, which moves only gradually through habituation and the observation of the consumption patterns of peer groups or other reference groups.  Higher disposable income is devoted to reducing household financial vulnerability by paying down debt.  I call this form of fiscal policy ineffectiveness Minsky neutrality.  It can be modelled either as an extreme form of precautionary saving (see Kimball (1990)) or as a strong form of target-wealth saving or buffer stock saving (see Deaton (1991) and Carroll (1992, 1997).  It may well play a role in countries like Iceland, the UK and the US where household gross debt has grown spectacularly during the period of rising house prices and optimistic permanent income perceptions, net financial wealth has taken a major beating through the collapse of house prices, stock prices and land prices and unemployment has risen sharply.

(IVc) More on different types of ‘crowding out’

The conditions for Ricardian equivalence are unrealistic and don’t hold in practice.  The empirical evidence is inconclusive, however, on the (time-varying) magnitudes of the spending and tax multipliers (see e.g. Favero and Giavazzi (2009), Ilzetzki, Mendoza and Vegh (2009), Mountford and Uhlig (2002), Perotti, (2004, 2008), Romer and Bernstein (2009), Romer and Romer (2009) and Spilimbergo (2008)).  One reason for the inconclusive evidence is that none of these studies distinguish between the ‘ceteris paribus’ effect of past, present and anticipated future fiscal policy actions on demand holding constant interest rates, exchange rates, other asset prices, wages and prices (the horizontal shift of the IS curve in the simplest IS-LM/aggregate demand – aggregate supply model) and the total equilibrium effect of these policy changes, allowing for the endogenous responses of asset prices, wages and prices.

It is clear that, even if expansionary fiscal policy can stimulate aggregate demand at given values of current and future prices, wages, interest rates, exchange rates and other asset prices, this does not mean that it will boost demand when the responses of prices, wages, interest rates, exchange rates and other asset prices to the fiscal stimulus are allowed for.  Three of crowding out can be distinguished: financial crowding out, real resource crowding out and direct crowding out.

Financial crowding out

Financial crowding out occurs though the response of interest rates, the exchange rate and other asset prices to past, current and anticipated future fiscal actions.  The textbook examples in the IS-LM framework are interest rate crowding out in a closed economy when the path of the nominal money stock is kept constant, and exchange rate crowding out in an semi-small open economy under a floating exchange rate and a high degree of international capital mobility.

Interest rate crowding out will be full or 100% in a closed economy when the nominal money stock is kept constant and velocity is constant.  Even when the monetary authorities peg the short nominal interest rate, there will be full crowding out of an (unanticipated, immediate, permanent) fiscal expansion under perfect international capital mobility when the exchange rate floats and the world interest rate is given.  A smaller trade surplus (larger trade deficit) undoes the effect of the fiscal stimulus on output through an appreciation of the nominal and real exchange rates.

A large country or region (like the USA or the Eurozone) with a floating exchange rate could use domestic expansionary fiscal policy to raise domestic demand to the extent that its actions raise the world real and nominal interest rate, but even for a large country or region, a significant part of a domestic fiscal stimulus may end up boosting output abroad through larger imports and reduced exports.  At a given world rate of interest, the entire demand stimulus leaks abroad through a larger trade deficit.  When output and employment are demand-constrained, these international demand spillovers are not just pecuniary externalities, but can have first-order welfare effects.  If there are domestic costs to or constraints on expansionary fiscal policy, fiscal expansion will, like any positive externality, be under-supplied.

This discussion has an obvious implication:

Proposition 6: International coordination of cooperatively designed fiscal stimuli is likely to be necessary to allow the internalisation of the effective demand externalities of a fiscal stimulus through the trade balance and the real exchange rate. This case is likely to be strongest when the degree of international capital mobility is high and the exchange rate floats.

Real resource crowding out

Real resource crowding out occurs when, regardless of the degree of financial crowding out, real resource constraints (capital and labour bottlenecks) limit the expansion of output in response to a fiscal impulse.  It will tend to be accompanied by rising prices and wages, often by rising real wages and by rising inflationary pressures.  In an open economy, the domestic supply constraint on final demand can be relaxed through the trade balance.  For the world as a whole this is not possible.  Fiscal policy cannot relax physical supply-side constraints in the short run, unless one believes there are significant effects of changes in labour income tax rates and other non-lump-sum taxes on labour supply.

However, I shall argue in Section (VI) that credit policy may reduce working capital constraints on production and employment, so credit easing policies may relax effective (financial) supply constraints on output, thus permitting a fiscal stimulus to have a stronger expansionary effect.

Direct crowding out

The effect on aggregate demand of an increase in public spending on real goods and services depends not only on the way it is financed and on the marginal propensities to consume of current and future tax payers.  It also depends on whether the real resources consumed or invested by the state are direct substitutes for or complements with private consumption and investment.  Public spending on free (at the point of delivery) public education and healthcare may be a substitute for private spending on education and healthcare.  Public spending on policing is a substitute for private spending on security guards and other means of enhancing personal security and keeping private property safe.  Public infrastructure spending (roads, railroads) may boost private investment in tourism or residential construction.  There is hardly any hard evidence on the presence and importance of such direct crowding out or crowding in.  The issue is not considered in Spilimbergo et. al. (2008).  But it is something that could usefully be taken into account when setting priorities for the detailed composition of public spending programmes.

(V) Sovereign default risk and the expansionary effect of fiscal policy

If a tax cut or an increase in public spending is deficit-financed, and if markets do not believe it to be certain that the government or its successors will raise future taxes (including monetary issuance) or cut future public spending by the same amount in present discounted value terms as the up-front tax cut or public spending increase, perceived default risk will increase and the government’s cost of borrowing will rise.  Government borrowing rates tend to set a floor for private sector borrowing rates.  Although it is possible that the private sector could borrow on better terms than its sovereign, such situations are few and far between.

A sufficiently large increase in the government deficit (or an increase in the effective net public debt through any other mechanism) could therefore increase the default risk premium on the public debt to such an extent that the net effect of the tax, spending and financing decisions on aggregate demand could be negative.  Note that this has nothing to do with Ricardian Equivalence, which assumes that the government never defaults but instead always meets its intertemporal budget constraint.  It is a form of financial crowding out, but not through increases in the risk-free rate, but through increases in the default risk premium.

Proposition 7: Financial crowding out is always and every where a (monetary) policy choice) if the financial crowding out occurs through an increase in risk-free rates.  Monetary accommodation may not be able to neutralise financial crowding out if this occurs through increases in sovereign default risk premia.

So far, in most of the industrial world and in the emerging markets, the increases in sovereign default risk premia have not been of sufficient magnitude to create worries about the effectiveness of expansionary, debt-financed fiscal policy through this default-risk driven financial crowding out mechanisms.  In most countries, including the US, the UK and Germany, the decline in the risk-free rate has was at first larger than the increase in the default risk premium on the sovereign debt and the total cost of government borrowing actually declined.

This situation began to reverse itself at the end of 2008, and both government default risk premia and long-term interest rates on sovereign debt rose for a couple of quarters.  In a number of European countries (Greece, Ireland, Spain, Portugal, and some of the Central and Eastern European countries) sovereign default risk premia rose to the point that a higher degree of financial crowding out of debt-financed fiscal stimuli would have been a certainty, ad in been tried.

In other large countries, including the US, the UK, Japan, Germany and France, we may not yet have reached that position.  The accumulation of public debt and of other hard or contingent exposure to the banking sector, and other financial institutions (AIG, Fannie Mae, Freddie Mac, RBS, Lloyds Group, Commerzbank), and non-financial enterprises deemed too large, too interconnected or too politically connected to fail (GM, Chrysler, Opel, Renault, Airbus), is such, however, that financial crowding out through rising sovereign default risk premia could become a real issue.  In that case, only balanced-budget measures or central bank money-financed support measures would have any expansionary effect.

Deficit-financed fiscal stimuli should be modulated across countries according to the ‘fiscal spare capacity’ in that country.  As discussed in Section (IIIB), this is the difference between the maximum value of the permanent primary surplus that is economically, administratively and politically sustainable and the minimal permanent primary surplus required for government solvency.  We must not be fooled by the contemplation of the very high public debt to GDP ratios found in the US and the UK immediately after WWI and WWII.

The willingness of the public to make great sacrifices, including fiscal sacrifices, in order to pay down a debt incurred in a noble, national cause – a war against an external aggressor – is not present today.  The increases in public debt we have seen in recent years and are likely to see in the next few years, were incurred as a result of a war on ourselves – civil war.   The political constraints on spending cuts and tax increases are much tighter than they were immediately following WWII, when the US had public debt around 120% of GDP and the UK around 220% of GDP.  Don’t take those figures as a guide to what the fiscal authorities will be able to get away with today without spooking the rating agencies and the markets.  The limits to fiscal burden sharing are much tighter in the economically, politically and socially polarised US of today than they were in the 1950s and 1960s.

Maverecon: Willem Buiter

Willem Buiter's blog ran until December 2009. This blog is no longer active but it remains open as an archive.

Professor of European Political Economy, London School of Economics and Political Science; former chief economist of the EBRD, former external member of the MPC; adviser to international organisations, governments, central banks and private financial institutions.

Willem Buiter's website