Tax Cuts, Spending Multipliers, and Economic Growth
Click here for a printer-friendly version of this article.
Few policy issues generate more controversy among analysts and policymakers than the economic effects of tax cuts and increases in government spending. These core fiscal policy questions are nearly always at the forefront of the policy debate, but that is especially the case today because of the looming “fiscal cliff” of $600 billion in tax increases and spending cuts set to take effect at the end of the year.
Judging the economic impact of the fiscal cliff depends on core assumptions about the impact that taxes and government spending have on economic aggregates. As is obvious to anyone who’s been watching economic policy debates over the past several decades, these assumptions differ greatly between policymakers and analysts.
The two assumptions of greatest relevance for measuring the economic impact of fiscal policy are the “multiplier” on government spending and the “supply side” or substitution effects associated with marginal tax rate reductions. Elected officials often speak about these issues with a conviction generally not shared by actual experts in the field. This is the case largely because the results cited by policymakers often depend critically on a set of assumptions that may not play out in practice. Moreover, as was clear with the stimulus of 2009 and other laws, macroeconomic analysis can often serve as a rationalization for a previously-decided policy course rather than serving as the substantive basis for a decision.
The notion of a “multiplier” to government spending is essentially the ratio between the increase in government spending relative to the associated increase in gross domestic product (GDP). If the multiplier is 1, then $100 million of incremental government expenditures would be assumed to generate $100 million of incremental GDP. For example, if the government increased public investment by $100 million to build a bridge, this $100 million would directly go into the national income accounts as a $100 million increase in government investment. If the spending led to second order spending effects – perhaps through greater consumption spending by the employees of the construction company that won the contract – the multiplier could conceivably be greater than 1 as the personal income generated by the initial outlay supports ancillary consumption spending.
In 2008, Mark Zandi estimated that the multiplier for food stamps was 1.73. Presumably this is because the spending on food stamps generates additional production of eligible agricultural products and increases retailers’ gross receipts. Some estimate even higher multipliers for direct spending and transfer payments. In May 2012, the Congressional Budget Office (CBO) produced the table below, which provides a summary of the range of multipliers assumed for various forms of government spending.
The most interesting part of the table is that the range around the estimate for each spending category is sufficiently wide to suggest that there’s virtually no significant economic difference between the spending categories. The effect of $1 billion to build a new airport (between 0.5 and 2.5) could have roughly the same impact as $1 billion in checks mailed to 10 million households for not working (between 0.4 and 2.1).
This seems to be an odd result. The basic idea behind the multiplier is that a failure of private markets has left productive resources (like human and physical capital) idle. If the government sends money to people, it has not done anything to increase their productivity. The purchasing power may increase aggregate demand, but at the expense of discouraging employment. As Casey Mulligan explains, the economic effect of spending on roads and bridges is likely to be far different from transfer payments like unemployment insurance because these payments are “contingent on work and production: the people cashing the checks received them by virtue of producing something the government values.” By contrast, unemployment insurance or means-tested transfer payments either pay people not to work or actually reduce benefits if people work too much. This perversity generates costs that are not accounted for in the multiplier framework.
Beyond the impact on labor supply, government spending could actually cause private spending to be lower than it otherwise would be. In the old Keynesian one-period models, this could only occur at full employment (i.e. if unemployment is higher than its natural rate, these people could be put to work at zero cost to private payrolls or production). However, in multi-period models, the cost of today’s spending can impact future levels of taxation. This means that households and business managers could respond to deficit-financed spending by increasing their expected future tax rates. The higher assumed tax rates, in turn, cause them to save more today, hold more conservative portfolios, and otherwise wait until the government makes it clear how the increased spending will be financed. Empirical research finds that the defense build-up during World War II had a 0.8 multiplier, as every $1 of defense spending was offset by a cumulative 20 cents decline in private investment, the non-military parts of government purchases, and net exports (larger trade deficits).
It is important to recognize that the size of the multiplier likely changes dramatically through time depending on circumstances. For example, a country with a relatively low debt load – say 15% debt-to-GDP – could increase deficits for a single year during a slump without creating expectations of future tax increases. This is because the households and business managers could reasonably assume that the debt need not be paid back through higher taxes or less future spending. At low debt levels, the government can essentially roll-over outstanding debt indefinitely and eventually grow its way out of the problem. When a government runs a balanced budget, its debt-to-GDP ratio declines at a rate that’s proportionate to the difference between the nominal GDP growth rate and the effective interest rate on its debt. For this reason, it is reasonable to think that a one-time deficit-financed infrastructure project in a country with little debt and low borrowing costs could generate a multiplier greater than 1.
However, once the debt level exceeds 60% of GDP, incremental additions to the debt stock may require future adjustments in the form of higher taxes or less spending than would have occurred otherwise. This leads to less private sector spending today. More dramatically, if the large deficits are persistent, the fiscal adjustment required is even larger because it must not only “pay back” past additions to the outstanding debt but also balance current revenues and spending. Empirical research finds that once economies exceed a certain level of indebtedness, fiscal multipliers go to zero as additional deficit-financed spending generates offsetting declines in private spending. In these situations, an increase in deficits today reduces private spending by increasing the magnitude of future fiscal adjustment costs. As investor Rob Dugger explains, when deficits reach levels that require radical changes in tax and spending policies “executives and investors take protective actions” that include “reducing domestic investment and hoarding cash.”
The “debt overhang” problem arises when so much future income has been pledged to fund past spending that workers and businesses no longer have any incentive to generate that income. This finding is of equal significance for the tax cut debate as well.
A large cut in taxes today might have very little economic impact because it would be difficult to convince households and business managers that it would be permanent. If expectations of large tax increases are reducing investment, increasing cash hoarding, and pushing investors into more conservative portfolios, this problem is unlikely to be solved by creating larger deficits today that generate even larger expected future tax increases.
A 2006 Treasury paper found that a permanent extension of current (2012) tax rates would result in a 2.3% larger capital stock if those tax reductions were financed by corresponding cuts in spending. With the exception of lower taxes on dividends and capital gains, which increase growth even when deficit-financed, all other tax components of the “fiscal cliff” would actually reduce long-run growth unless they are financed by spending restraint. Higher marginal income tax rates reduce work, effort, savings, and investment by reducing the household or business’ share of the incremental income generated by these activities. This is the same channel through which debt overhang operates, as the implied higher future tax rates generated by large deficits reduce work, savings, and investment. Indeed, because no one is sure who will pay the ultimate costs of the looming fiscal adjustment, persistently large deficits could be worse than higher taxes if households and businesses overestimate their likely share of the future tax burden.
Interestingly, in an economy with a gross debt to GDP ratio of more than 100% and exponential expected increases in retirement and health care costs, the economic benefits of both additional tax cuts or government spending is likely to be marginal at best – and very well could be negative. This is why the key economic policy issue today is reducing the size of government over time through spending reductions that cause households and businesses to reduce the size of expected future tax increases. The best course for policymakers today would be to place the budget on a sustainable path, which would aid business planning by making the fiscal adjustment costs explicit, reduce the risk of a public sector financial crisis, and create fresh confidence in the institutions of government.