Debt and Taxes I

There might be such a thing as a free lunch.

There will soon be a Democrat in the White House and Republicans will soon rediscover their hatred of deficits (which were no problem when they were cutting taxes on firms and rich individuals). We are going to read a lot of arguments about irresponsibly burdening our children with debt (which ignore the fact that they will also inherit most of the bonds). We will be reminded that sooner or later we will have to pay.

I am not sure if Milton Friedman will be quoted saying “to spend is to tax”. There will be arguments about how deficit spending creates the illusion of wealth (as consumer/investors we forget that we owe the money as well as owning the bonds just as citizens we forget that we are (most of) the creditors as well as the owners of the indebted Federal Government). There will be arguments about how we can pay now or pay later and it will be more costly if we pay later.

All of this is based on the assumption that the Federal Government’s intertemporal budget constraint is binding. Arguments that you can’t get more now without having less later are arguments about a binding budget constraint. The argument that an increase in spending must be financed by increased taxes in the present or in the future of the same present value is the argument that the intertemporal budget constraint is binding (in fact it is a better explanation of the concept than “the intertemporal budget constraint is binding” the two statements are equivalent and “increased spending … increased taxes …” is written in plain English).

It is true that in standard models, the intertemporal budget constraint is binding (this is called the transversality condition just to type more big words). This is a condition for *optimality* — an aspect of the solution to an intertemporal optimization problem. It is not a given or an assumption about the problem agents face. This is embarrassingly simple. In standard models, you can’t get something for nothing, because if you were in a situation in which you could get something for nothing, then you made a mistake not getting it, and it is assumed that you didn’t make a mistake.

When discussing fiscal policy, Friedman et al assumed that it is optimal while criticizing it as suboptimal. This is a plain contradiction and simple error. All of the discussion of fiscal policy which is used to rule out more deficit spending now assumes that policy is optimal which rules out any change by assumption.

This is not a quibble. The condition for a binding intertemporal budget constraint is that r>n that the interest rate the Treasury must pay is greater than the trend growth of GDP. if r<n then debt can be rolled over forever with new bonds sold to pay interest and principle on the old bonds. The debt to GDP ratio shrinks to zero if r<n and debt is rolled over. This is how the USA handled World War II debt. The US did not pay it off by running primary surpluses. The USA rolled the debt over until it was small and then elected Reagan and began the modern era of huge deficits.

The relevant r in the inequality is the interest rate the Treasury pays on it’s debt. This is always much lower than many other interest rates and has historically usually been very very low. It is now exceedingly low, and it has been extremely low since 2008. It was also very low until the 70s and, when corrected for inflation, remained very low until Reagan and Volcker began working with each other.

This is well known to economists. A masterful explanation and empirical demonstration was given as an American Economic Association Presidential address by OJ Blanchard in 2019 . Read it if you doubt my claims (also if you don’t — it is very good).

For a while after 2008, it was assumed that this was temporary and things would return to normal (normal meaning as they were from 1980 to 2000 but never were before 1980 nor have been after 2000). Many economists now agree that this is the new normal (same as the old normal and different from the situation of extremely loose fiscal policy combined with extremely tight monetary policy).

Importantly this is *not* entirely a forecast. The yield on 30 year inflation indexed bonds is currently -0.24% — the US Federal Government can borrow for 30 years paying a negative real interest rate. Pessimists think that the trend of real GDP growth has fallen to 2% a year (from 3% historically). But it is definitely growth not shrinking.

It is true that investors might change their minds and interest rates might shoot up. If the US financed it’s spending with 30 inflation indexed bonds, this would not be a problem for the Treasury for 30 years. Investors who bought the bonds would lose money, but the Treasury would still have to pay the same coupons and face value. There is even a discussion of introducing extremely long duration bonds — 50 year bonds to lock in interest rates longer or even consols (bonds which last forever and pay a constant interest rate — obviously a multiple of the CPI because the old 19th century nominal consols are silly collectors’ items now))..

But basically the main hugely important point is that there is every reason to think that the US Federal Government can get something for nothing, because it has a slack intertemporal budget constraint.

Failing to take advantage of that would be failing to solve an intertemporal optimization problem. It is one of the few (very difficult) ways to fail my Macroeconomics course.

Caveat after the jump

So far I have acted as if the US Federal Government is an agent which wants to consume. However, it is merely an instrument we use to serve our interests and we have different interests. A safe interest rate less than the rate of growth of GDP implies that increasing debt by giving everyone money ($600,$2000,$1400 whatever) is like a windfall — people get something and no one has to pay – ever.

However, increased public debt also has side effects. In particular, unless the economy is in a liquidity trap with interest rates at the zero lower bound, it crowds out private investment. This is not necessarily a cost. A necessary condition for optimal levels private investment is r>n. if r<n, then less capital can be better for everyone.

However, the r in that equation is the return on privately owned capital not the interest rate paid by the Treasury. That return is much higher than the safe interest rate. In fact, Blanchard describes two effects — a windfall which is good if the safe rate rf is less than the growth rate and another effect on welfare which is good only if the risky rate (expected return on privately owned capital) is less than n.

With Barbara Annichiarico, Brad DeLong, and I are working on this. It turns out that the second effect of debt is a transfer from workers to investors. Lower investment implies lower wages and a higher return on capital. This second effect can be cancelled by taxing capital income and subsidizing wages.

The combination of higher debt and a shift from taxing wages to taxing capital income can cause increased welfare (in the model for everyone so a Pareto improvement) so long as the safe interest rate is less than the rate of GDP growth.

The standard of efficiency in public economics is can we achieve a Pareto improvement with this reform and also some taxes and transfers to compensate those who are hurt by the reform.

Also I like imagining Republicans’ reaction to a proposal to increase debt and also tax labor income less and capital income more. It all follows from a very standard Macroeconomic model.