Crowding Out

The mechanism

Start with a single market for loanable funds. Households, firms, and foreigners supply savings; investors and the government demand them. Equilibrium real interest rate r* clears the market.

Now the Treasury runs a deficit and issues bonds. Government demand for funds shifts the demand curve right; r* climbs to r**. At the new rate, private investment falls. The reduction in private investment is the crowding-out quantum. Whether it's small or large depends on:

• Slope of the saving supply curve. Steep (people don't save much more when rates rise) → most adjustment falls on investment → strong crowding out. Flat → savings rise to absorb the deficit → weak crowding out.
• Slope of the investment demand curve. Steep (investment is rate-insensitive) → little crowding out. Flat → small rate increases knock out a lot of projects.
• Capital mobility. An open economy can borrow from foreigners, blunting the rate increase. The U.S., the world's biggest borrower, also has the deepest external savings pool.

Worked example: a Treasury auction in a closed economy

Suppose private investment demand is I = 100 - 5r (in % terms), private saving supply is S_p = 40 + 5r, and the public deficit D = 20. With no deficit, equilibrium gives I = S_p → 100 - 5r = 40 + 5r → r = 6%, I = 70.

Now the Treasury auctions $20 of new bonds. Total demand for funds is I + D = (100 - 5r) + 20 = 120 - 5r. Re-solving: 120 - 5r = 40 + 5r → r = 8%. Private investment falls to I = 100 - 5(8) = 60. Of the $20 deficit, $10 came from extra savings (private saving rose from 70 to 80) and $10 came from displaced private investment.

Crowding-out coefficient: 0.5. This matches the central tendency of empirical estimates — around half-displacement is the typical case in tight economies.

Ricardian equivalence: the deficit doesn't matter

Robert Barro (Journal of Political Economy, 1974) revived a hint in Ricardo: rational households see deficit-financed spending as future taxation. If the government borrows $20 today, every household knows the present value of future taxes has risen by $20. Permanent-income hypothesis (PIH) households reduce consumption to save the difference. Private saving rises by exactly the amount of the deficit. The supply curve shifts right by the same amount the demand shifted, leaving r and I unchanged.

If Ricardian equivalence holds, deficits are neutral — fiscal policy has no real effect. Empirical tests reject the strict version: households are liquidity-constrained, finite-lived, and don't optimize across generations as cleanly as Barro's model assumes. Bernheim (1987) finds partial Ricardian behavior — perhaps 20-50% of deficits offset by saving — leaving room for both crowding out and Ricardian effects to coexist.

Financial vs real (resource) crowding out

Two distinct channels often get conflated.

1. Financial crowding out. Works through the credit market. Government bond issuance pushes up yields; private borrowing becomes more expensive. This is the IS-LM textbook channel.
2. Real (resource) crowding out. Works through factor markets at full employment. If the government hires construction workers for infrastructure, private firms must pay more for the same workers. The displacement happens in labor and materials, not credit.

Financial crowding out can be neutralized by central-bank purchases (QE absorbs the bond supply). Real crowding out cannot be neutralized — if every worker is employed, more government spending means less private spending in real terms regardless of how it's financed.

When crowding out is and isn't strong

Conditions	Crowding-out strength	Why
Full employment, closed economy	Strong (toward 1.0)	Fixed savings pool; deficit must displace something
Recession, slack capacity	Weak or negative (crowding-IN)	Idle resources mean both can rise; multiplier > 1
Zero lower bound (e.g. 2009-15)	Near zero	Central bank pins short rates; long rates respond little
Open economy, mobile capital	Weakened	Foreign savings absorb part of the issuance
QE active	Near zero (financially)	CB buys the bonds with reserves
Liquidity-constrained households	Stronger	Less Ricardian offset; deficits feed through

This state-dependence is why the same fiscal policy debate keeps recurring. In 2009 the case for stimulus rested on slack and ZLB; the crowding-out objection was weak. In 2021-22, with output near potential and rates rising, the same case looked weaker.

What the data say

• CBO long-run analyses. The U.S. Congressional Budget Office uses a crowding-out coefficient of roughly 0.33 (each dollar of long-run debt reduces capital stock by ~33 cents) — derived from the Solow framework with empirical capital-output and saving-rate parameters.
• Cross-country evidence. IMF's World Economic Outlook (Oct 2010 chapter on fiscal multipliers) finds multipliers of ~1.5 in slack economies and ~0.5 in expansions — consistent with crowding-in then crowding-out as activity rises.
• WW2 financing. The U.S. ran deficits of 25% of GDP from 1942-1945. Private investment did fall — but mostly because of price controls, rationing, and the Fed's pegged-yield curve, not market crowding out per se.
• Reagan-era deficits. 1981-86 deficits coincided with high real rates (10-year TIPS-equivalent ~7%) and a strong dollar — Engen and Hubbard (2004) estimate ~3 bps per percentage point of debt/GDP, modest but nonzero.

Variants of the idea

• International crowding out. In an open economy, deficits raise rates, attract foreign capital, appreciate the currency, and crowd out exports rather than domestic investment — the "twin deficits" link.
• Crowding-out of private R&D by public R&D. Wallsten (2000) found federal SBIR grants partly displaced firm-funded R&D dollar-for-dollar; later studies find more crowding-in. Direction varies by program.
• Crowding-out of charitable giving by government welfare. Andreoni's literature on warm-glow giving estimates partial substitution: a $1 increase in government welfare reduces private giving by ~$0.20-0.40.
• Crowding-out in housing. Below-market public housing can displace private rental supply when developers stop building units the public sector now provides.

Counterarguments

• "Crowding out is irrelevant when the economy has slack." The Keynesian rejoinder: in a recession, idle resources mean fiscal expansion adds output, and may even raise investment by lifting expected demand (the accelerator effect).
• "With QE, financial crowding out is moot." True for the period when the central bank is actively buying. Once QE unwinds (QT), the bond supply hits private balance sheets and the channel reactivates.
• "In an open economy, the U.S. faces an infinitely elastic foreign saving curve." Approximately true at small scale, but the deeper the deficits the more the curve slopes up — and the appreciation crowds out tradables instead of investment.
• "Public investment crowds in private." If the spending builds productive capacity (highways, broadband, R&D), the marginal product of private capital rises, attracting more investment than was displaced.

Common pitfalls

• Comparing nominal to real rates. Crowding out works through real rates. A nominal yield rise that just offsets higher inflation is not crowding out.
• Ignoring the central bank's reaction function. Whether the Fed accommodates or leans against the deficit changes everything — the same fiscal package can be 0% or 50% crowded out depending on monetary stance.
• Treating short and long horizons identically. Short-run crowding out can be negligible (multipliers above 1) while long-run crowding out is substantial (slower growth via lower capital stock). Both can be true.
• Forgetting the composition of spending. A deficit funding consumption transfers crowds out investment more directly than a deficit funding productive infrastructure that itself raises capital returns.