Sunday, December 4, 2011

The circular logic behind Ricardian Equivalence

Let’s examine a common argument that conservative economists make when discussing the possibility of government stimulus and the logic of borrowing when the economy is weak to pay back when the economy is strong.  The argument is a conjecture known as Ricardian Equivalence; here is one telling set in the fabled land of Freedonia where the government is deciding how to pay the bill for an expensive dinner:

Assume you can either impose $1,000 in taxes now to pay for that state dinner, or you can issue $1,000 of government debt, payable in one year for, let's say, 10% interest. If you go with the taxes, Freedonians have $1,000 less to spend today. That's straightforward enough.

If, on the other hand, you go with the debt, then Freedonians, being a savvy bunch who've seen this debt financing in action before, realize that in one year, it will be time for you to pay back the people who buy the government debt. You will owe those people $1,000 plus $100 in interest, for a total of $1,100. Freedonians know that money must come from somewhere, so they expect that in one year, their taxes will go up by $1,100. In order to be ready for that one year from now, they put $1,000 into saving today, earning 10% interest, so that they will have the $1,100 they will need. This is $1,000 dollars today that they cannot spend today or save for reasons other than paying future taxes, so the outcome is that Freedonians have $1,000 less to spend today, just like they do if you raise taxes today.

Now we can note the absurd assumptions about human behavior -- does someone living paycheck to paycheck skip a rent payment to have money set aside for a hypothetical future tax hike? Nevertheless, I'll accept that much for the sake of argument; there's an even more serious problem here.

The problem I want to point out is a fundamental logical flaw. The key question to ask is, what exactly are the Freedonians in the above example saving for? Their future taxes, right? Wrong. They are saving for a future change in their after-tax income. To appreciate the subtle difference, let’s imagine that you are a businiess owner and the government is considering enacting some policy that will cost you $1000 in taxes at some point in the future but will also bring you $1000 more in business at the same time. Do you have any more reason to save if the government chooses to do this than if they do not? Of course not.

But if we assume no overall change in incomes due to the decision to tax or borrow, then the future change in after-tax income is the same as the future change in taxes, right? But therein lies the problem -- the whole point of the notion of Ricardian Equivalence is to argue that it makes no difference to the economy whether government spending is funded by taxation or borrowing, but as we see the argument logically depends on this as an assumption. It is a circular argument, as such it sheds no light on the question of when to tax or borrow and whether it makes any difference.

So this argument is unsound, but putting this argument aside, does it make any difference whether the government taxes or borrows?  I see at least three reasons to suspect that it does make a difference:

1.  Just consider for a moment the different forms a tax hike could take: it could fall on the rich or the poor, on businesses, gasoline, or cigarettes, on sales, income, property, value added, or capital gains, etc.  If it makes any difference which of these taxes is chosen, then what does it even mean to say that borrowing has the same effect as taxation?

2.  We know that borrowing can in some situations be worse.  One funny thing about Ricardian equivalence, if taken to its logical conclusion, is that there’s no reason why we couldn’t just do away with those annoying taxes for a decade or two and just borrow -- after all, if the effect of borrowing and taxation is the same, why should we go through the hassle of filling out those tax forms?  But that would lead to default, while sufficiently high taxes would not.

3.  If the government chooses to borrow, the money comes from people who were looking to save money rather than spend it, so the immediate effect on private spending is likely to be quite small.

UPDATE:  Paul Krugman had a good post taking on the absurd assumptions about human behavior present in the standard Ricardian Equivalence argument.  This prompted responses from Stephen Williamson and John Cochrane.  Now before I get to their responses, it seems to me that there are two parts to the concept of Ricardian Equivalence:  1) the prediction: that the timing of taxes does not matter; taxing and deficit spending are the same  2) the mechanism: we expect this prediction because people will expect higher future taxes and will take those into account when deciding how much to spend today. Now if you consider Ricardian Equivalence to be the combination of these two claims, then it is clear that Paul Krugman was laying out a problem in part 2, and it is also clear that John and Stephen have not laid a finger on him in their responses.  Instead, they seek to rescue part 1 by proposing an entirely different mechanism.  So if you throw out the standard mechanism, is it still correct to call it Ricardian Equivalence?  Unfortunately, David Ricardo could not be reached for comment.

So what is the mechanism behind, oh, let's call it Williamson-Cochrane equivalence? That when someone such as the government borrows, someone else must lend, thus there is no more money to be spent as a result.  Now I have to say this is a much more plausible mechanism, and I'm quite frankly baffled that the more absurd explanation not only exists but is standard when people who want to make the case against deficit financed stimulus could make a much better argument like this one.  But this one doesn't quite work either, for reasons I get into here.  The basic problem with this is that not everyone would be equally quick to spend a dollar, especially those who are looking to save.  So if the government were to sell bonds and use the money to give a temporary tax cut, one thing recipients of the tax cut could do is buy bonds, as is assumed under Ricardian Equivalence.  But that isn't their only choice and it isn't what most of them would do -- they could instead make a rent payment and their landlord could then use the money to buy bonds.  Or they could make a rent payment, and their landlord could pay the janitor, who could buy a TV, and the electronics company could send the money to shareholders who could use it to buy bonds.  Yes, the borrowed money has to come from somewhere, but where it comes from and where it goes to can have a big effect on how much economic activity that money is involved in as it circulates around the economy.


  1. Just found your blog via interfluidity. I certainly agree that Ricardian Equivalence is rubbish.

    Can I respond to your three points to offer what I hope are some constructive remarks?

    1. I does of course make a difference from the point of view of the private sector whether a government taxes or borrows (uses new money) to finance their activities. Taxes are designed to fall on different people and entities in particular ways based on various principles. The effect of capturing resources via an inflation tax (which is what effectively happens when a government borrows money to spend) is distributively quite different I would suspect. But the two distributions do interact for a net outcome from the budget position.

    Aside from these distributive consequences, from the point of view of the government it doesn't really matter. They get the resources to use either way at the time they spend.

    2. Depending on your theory, government can print it's money and do away with taxes altogether. But if you subscribe to MMT or functional finance perspectives, it is the existence of taxation in the money unit of account that provides the impetus for it's adoption in the private sector. If you didn't need government-sanctioned money to pay taxes, why would you use it at all?

    3. No. Like any other borrowing this is new money. It is not a transfer from 'savers' to government. It is a transfer from everyone to government in resource terms via the inflation pressures (which may be differential).

    You might like my recent post in debts etc.

  2. 1. Agreed

    2. I don't think the government can just print and do away with taxes under any theory of money, except hypothetically maybe in cases of extreme deflationary pressure beyond any cases seen in the real world so far.

    3. Not no. The decision on whether to call it new money or not is an arbitrary accounting choice, and neither choice is the right one and either can be used as the basis for sound economic reasoning but some problems may be clearer one way or the other.

    To see how they lead to the same results, let's consider the effect of government borrowing in a slack economy:

    Bonds not money: This is the way I described it above. Government spends money, but little reduction in private spending because the people buying bonds most likely have low marginal propensity consume; otherwise why were they looking to buy bonds?

    Bonds are money: Government spends money, but little reduction in private spending because no one in the private sector has less money -- same prediction.

    Now consider government borrowing in a full employment economy:

    Bonds not money: Government spends money, but it needs to find buyers for the bonds. Either the Fed will take care of this by printing money and leading to inflation (leading to general crowding out of private spending), or interest rates will have to go up to encourage private sector investors to buy government bonds instead of investing in the private sector, leading to crowding out of private investment.

    Bonds are money: Government borrowing leads to more money in circulation, which is inflationary, unless the fed responds to the inflation by contracting the money supply and raising interest rates. Same prediction.

    The implications are equivalent.