Hacker News new | comments | ask | show | jobs | submit login
The Natural Rate of Interest Is Zero (2004) [pdf] (cfeps.org)
38 points by jganetsk on Feb 19, 2018 | hide | past | web | favorite | 56 comments

Takeaways: -the federal government is self funding. (my opinion: we should take advantage of this by lowering taxes or spending more)

- taxes create demand for money but don't fund government.

- 'money' doesn't exist till the federal gov spends it into existence

- interest rates are determined exogenously by the fed.

- large deficits tend to drive down , not up, interest rates because of the excess reserves created.

I wish people understood that public debt is not really debt as ordinarily understood. It's really just an accounting artifact and tends to reflect and mirror net private savings.

That is unless your country does not issue it's own sovereign currency. Then it is debt.

Public debt is a promise to fund future obligations and things like pensions would be significantly devalued if it isn't honoured, so in political practice it is debt.

That debts and savings are two sides of the same coin is tautological and doesn't require a state or fiat currency. Your current account with the bank is a debt the bank owes you. All savings are debts.

Right, but public debt is someone's asset as well. To parapharse Stephanie Kelton. The federal deficit is the public's surplus.

Does this last sentence apply to the countries that have the Euro?

That would be correct.

Not entirely. It doesn't apply to the Euro as a whole and that means it doesn't apply to countries that, if they have a big banking crisis would provide a medium-or-higher risk to the Eurozone as a whole.

So it does't apply to France, Germany, Italy, Spain, perhaps Poland.

(2004) and talking about the risk free overnight rate.

"Deficit spending will result in net central bank reserve credits in the aggregate banking system, which will drive the short-term overnight inter-bank lending rate to zero."

Also, we now pay interest on reserve balances which goes against one of the paper's assumptions.

The paper mentions that:

"While government security sales may be used to drain the excess reserves to maintain some positive overnight rate, or the central bank may pay interest on reserve balances, absent such government intervention the base rate of interest is zero. In other words, the natural rate of interest is zero."

That's what I'm saying. The paper was written in a time before interest payments on reserve balances, and it explicitly says that is one of the requirements for the paper to hold. We now pay interest on reserves, but I'm not sure what that entire policy is or how it works.

The paper still holds. It is saying that the natural rate of interest, i.e. the "base case", is zero. It mentions interest on reserves and bond issuance to drain reserves both as ways to "move away from the base case". It claims that, regardless of mechanism, we should justify our actions to move away from the base case. It also claims that the base case is a good policy for a variety of reasons.

I know Mosler and he was a proponent of paying interest on reserves. Doing so avoids the complexity of having to execute Fed interest rate policy using treasuries.

It is also the natural amount of profit in perfect competition!

When you put it that way, I really don't want to live in a perfectly efficient market.

So are the authors lending or just talking?

Warren Mosler was a bond trader, for a while, so yes. And he lent at market rates, not at the natural rate, as one would expect.

I too would be interested in this as I'd like to take out a mortgage.

In Switzerland, the current target interest rate (set by the Swiss National Bank) is -1.25 to -0.25 percent, with current rates around -0.75%.

This means banks need to pay the SNB on their funds deposited there. Many of them pass that on to their customers: Deposits don't earn any interest to speak of, and for large sums (starting from 1M or so), banks commonly charge interest. At some points, the yield on Swiss Gov't bonds was even negative.

If you can offer a safe (and bank-accepted) way to store that money, you stand to make a killing.

On the other hand, even in that environment, mortgages aren't free. Rates range from 0.5 to more than 2%. Part of that is the banks' margin, part of it comes from fixing the interest rate for some (2-10) years.

[1] https://www.snb.ch/en/iabout/stat/statpub/zidea/id/current_i...

I'm surprised the rate would get that negative. Especially when they have a 1000 Franc note (worth around 1000 dollars). At that level you can fit trillions of dollars in a single vault. A respected bank could build a new vault, staff it with multiple guards, charge 0.2% or 0.1% for storage, and make their investment back in under a year. Even with smaller notes. Obviously I'm missing something here, but what? Is it not safe enough?

A that sort of scale, you need government permission to turn the banknotes back into legally recognized bank account balances. So if there's a 1% chance Switzerland says "no, you can't actually just store cash in a vault instead", you're losing money on a risk-adjusted basis.

Needless to say, when rates did go negative, banks did not respect the interest rates they agreed to. Did not respect the contracts they agreed to with customers.

When they were legally forced to do so, the government prevented them from being sued to do it anyway, by destaffing the only judicial office where such procedures could be started.

Also, banko Santander was sold to external investors, with direct involvement from the central bank and the Spanish government, and ... failed to mention these negative-interest-rate obligations. If anyone else ever did anything remotely approaching this they wouldn't just be nailed to the cross, people would be talking about it for decades too.

Oh and despite all this cheating, banks and governments not respecting their own laws (nor even taking the time to change them, or even so little as involve the legislative branch in the process), despite violating every precept of democracy on both the Spanish and European level ... all large Spanish banks are a hair away from bankruptcy.


They're talking about the risk-free rate (no offence intended :-) ).

Being able to use money has value. To get people to forgo the use of their own money, you have to pay them something.

If there were risk free investments (I don't think there are), wouldn't people invest in those directly instead of letting you do it with their money?

Of course, there is always arbitrage.

This paper is talking about intraday or overnight lending between banks... cases where loans have a term of less than 1 day. Yes, "to get people to forgo the use of their own money, you have to pay them something". The length of time the lender forgoes the use of their own money is a factor in the rate the borrower pays. As loan term tends toward zero, interest rate should tend toward zero as well. That's what central banks control, but often choose a non-zero value.

I can see why interest should tend towards zero, but why would the rate?

Intuition: if you were my friend, and I gave you a loan for 5 minutes (maybe because your wallet is in your car and we are paying a restaurant bill), how much interest would I expect you pay it back with? Zero (think about Venmo payments). And the only way to have a zero-interest loan for a non-zero amount of time is setting the rate at zero.

This loan has an extremely short term (5 minutes) and is extremely low risk (it's a small amount of money and we are friends). So all real loans should have a non-zero interest rate, but the rate should asymptotically approach zero as we decrease the term and the risk. If not zero, then what would you choose? What should be the floor on interest rates?

If we think about the interbank lending market, once the aggregate supply of reserves is beyond what is necessary to settle daily payments, the interest rate falls to the floor (the rate the Fed pays on reserves). In this case, why should the Fed choose a non-zero floor? There are many answers the Fed could provide, and they do. But they definitely need to provide that answer, which is what the paper is claiming. And the paper also suggests that zero works pretty well, for a variety of reasons.

I don’t think framing it in terms of friends is useful. I’d make that loan to be kind. I wouldn’t do it as part of a business arrangement, so a stranger would probably be out of luck.

If I were in the business of making five minute loans, I’d charge an astronomical interest rate to cover the fixed transaction costs.

I think you're getting side-tracked into credit risk which is not what the paper is about at all. But your example is useful to understand why credit risk is not as simple as that.

For a start, considered in the context of formal loans, lending money to a friend in that situation is far from a risk-free bet - because my friend might have got it wrong and left his wallet at home, not in his car in which case in 5 minutes time he'll default on payment. Of course I'll almost certainly get my money back eventually (since otherwise I'd either have just said I'd give him the money or declined to either lend or gift it), but now your talking about the risk that I'm deprived of my money for longer than I had anticipated as well as having the the cost of using my time in chasing him for it, so I'd want some interest to compensate for that risk.

In reality, what we do is (to some extent) to agree a social convention that "next time it will probably be the other way around" so that those potential losses even out between us and it makes sense that we reduce our administrative burden by not charging each other for them in the form of interest.

What is true is that as the time a loan is outstanding reduces, ability to judge the repayment risk is likely to improve to some extent in most cases. However even this isn't universal - for example if I lend against you receiving payment on an invoice just issued and where yuou have given your customer 30-day payment terms, I'm likely to have better certainty lending on a 60-day repayment term than a 1-day repayment term.

I don’t think I was discussing risk at all, unless some part of my comment implied a statement about risk that I hadn’t thought of.

> And the only way to have a zero-interest loan for a non-zero amount of time is setting the rate at zero.

This assumption is where things break. You could have an interest rate whose nominal contribution for five minutes that rounds down below 1 mil, the smallest unit of currency. A five-minute interest rate of 0.00098% would result in rounding the interest on $5 to zero. However, taken over the course of 5 minute increments through a year, and you arrive at $14.01 dollars:

$5 * ((1 + 0.0000098)^(12 5-min increments/hour * 24 hours/day * 365 days/year)) = $14.01

Plenty of folks would love a 280% annual return.

> To get people to forgo the use of their own money, you have to pay them something.

Temporarily forgo the use and accept a nonzero risk of never seeing it again. This is one reason why there is a range of interest rates.

There is. It's U.S. Treasury bonds. It's guaranteed profit and risk free

Sovereign debt is only risk-free relative to the issuing government's continued existence.

The historical average duration of empires, republics, dynasties, and monarchies is 349 years. The assumptions we make about financial markets in the developed world seem to completely ignore political factors, as if issuing governments or unions that exist today will always exist - as if history has somehow ended.

When you buy a 30y US Treasury bond at auction, you are getting paid 3.13% to assume that between years 231 and 261, representing 8.6% of the average lifespan of empires, no existentially threatening political upheaval will occur in the United States.

Seeing the political division of the populace in the 2016 election, I wonder if this is a prudent assumption.

I also wonder if this is a prudent assumption. Existential threat is really hard to measure. If we want to know the average lifespan of a modern republic that has existed for about 250 years already, we don't have many data points to compare to. Here are ones that may count:

- San Marino has been a republic for over 1000 years

- The Netherlands was a republic from 1581 until 1806

- Switzerland has been a republic since 1648

- Paraguay, Chile, Argentina, and a dozen other Latin American countries have been republics continuously for over 150 years. Not without their share of upheaval, but even in the case of Venezuela I don't know that you can argue that the country stopped existing in sudden crisis.

And that's all I can find, for "modern" republics >= 150 years old at present or at cessation. The next closest is France, which is also still a republic. We don't have too much data but I would listen to an argument that modern republics seem to be unusually long-lived compared to more traditional government types. With no better information available, you can choose either to trust the examples we have, or to adopt the doomsday prepper mentality. When trust is cheaper, it's not any surprise to me that people trust the government to continue existing.

The US is <50 years from it's last large default (moving off the Gold standard). This is more than most European nations.

And for nonpayment risk you obviously count the time between large defaults. Whether the person/company/country/empire/dynasty/... "dies" or not is irrelevant.

Private companies, mostly because they have no such choice, actually do better than states.

You get paid 3.13% every year. So you are getting paid closer to 94% for making that assumption. Plus you get some of the 94% paid out early, so if you were wrong about there being no terrible upheaval, you still got something out of it (hopefully you didn't reinvest the interest in more Treasuries, or keep it in cash).

So even when you put it like that - the odds don't seem too bad.

It’s not actually risk free. It’s considered risk free by the markets right now. If the US treasury bonds Lost their status as risk-free, all valuation models would collapse and the markets would have to move/rush to some other instrument that could be considered risk-free. For example, black-scholes which values options requires the concept of a risk-free asset.

Also, it’s not guaranteed profits if the interest rate were negative, which happened during the financial crisis. As well, real interest rates, vs inflation, are negative. So it depends on your perspective.

> For example, black-scholes which values options requires the concept of a risk-free asset.

This isn't really true. The risk-free asset doesn't have to exist, we just have to agree on what its value would be. For example if Tesla sold a one-year bond paying 20% interest, and everyone agreed Tesla had a 10% chance of being bankrupt within the year, it's fair to agree the risk-free rate is 10% per year.

It's true that if the US government were at risk of defaulting on Treasuries, the market would be a bit of a mess. But it happens in other countries all the time and the markets eventually continue to function without an asset considered risk free.

I suspected that low interest rates would be around for a while, if not forever. So when I got my mortgage, I got the 7/1 ARM, instead of the standard US 30-year fixed.

The paper is talking about the overnight, risk-free interest rate, not a long-term mortgage, which is just one point on the yield curve. If we think of the yield curve as a 3D graph, where one axis is interest rate, another axis is term, and a third axis is risk, then the paper is claiming that there is a point at (0, 0, 0). But central banks often change the interest rate to something non-zero at (0, 0).

Zero interest means zero economic motivation to lend anyone any money. This is only possible in some ideal world in which there is very little demand loaned money, and those few who borrow always give it back in a timely way, so there is zero risk in lending. And lending is just between friends; you know, can you give me a hundred bucks? I will pay you back in a few weeks (exactly a hundred bucks). And they always do.


Updated. Thanks!

USD is a promise to pay a dollar to the bearer of the receipt. A dollar is an amount of gold, and so is a franc, a pound, a baht, and probably many others. Conversion of a thing into itself is not a conversion. There is no risk free rate, and the behavior of capital constrained by worldwide tax hunt cannot reasonably be called 'natural'. One more thing: banks can only have reserves after covering the totality of their demand deposits, which means that virtually all public banks are insolvent... You say they can sell some assets to cover? Fine! Let's see the selling, and see how many come out alive! I'm waiting!

Where do I trade in this piece of paper in my pocket for gold?

Technically a FRN (Federal Reserve Note) is backed by a US Dollar. These are distinctions which seem odd today, and describe the odd system that we now have.

A "US Dollar" is defined constitutionally as being "gold or silver". However you can no longer redeem a FRN for a USD....only for more FRN's.

It raises several interesting legal questions when pondered, for instance when you pay taxes you are being taxed on USD earnings, although technically you've never been paid this.

There's no constitutional definition of a dollar: gold, silver or otherwise.

The restrictions in article I section 10 bind the states, not the federal government.

Since the suspension of convertibility into gold in the 70s, dollars are essentially a pure fiat currency - whether in FRNs, dollar coins or otherwise.

Yes it bound the states...and its obvious intention was to forbid the promulgation of fiat throughout these "united states". Moreover, the First National Bank of the US was forbidden from both buying US treasuries and issuing debts beyond its capitalization. Clearly two limitations intended by the framers of the constitution to prevent the issuance of a fiat currency.

The dollar was a unit of measure prior to the existence of the United States. Saying the dollar was not defined is like saying an ounce is not defined. A "dollar" was 550 grains in weight.

Its obvious intention was to centralise currency by preventing the states from making their own scrip. The Framers were perfectly capable of writing limitations on the federal government reflecting those on the states if they wanted to - look at the restrictions on bills of attainder and titles of nobility.

And the first Bank was not set up by the Constitution - it was set up in 1791 by Act of Congress after a political fight between Hamilton and his critics.

The Spanish dollar existed at the time of the US's founding, as did various other countries' (differently weighted) dollars; the US dollar wasn't defined until 1792 (when it was, indeed, defined as given weights of either gold or silver - the Spanish dollar was silver only). But the US dollar was defined by Congress and can be redefined by the same. And they have.

So something I've never understood, or had explained to me well, perhaps, is ... the value of gold is not proportional to it's actual utility, so its value also seems equally arbitrary.

So what is the issue people have with "fiat" currency? Is it simply because it is created and managed by the government?

Because you can't print gold.

The fundamental question is: Should we represent the value of real finite things (carrots, oil, land, etc) using an abstraction which is also finite or one that is infinite?

Exponential growth curves can't continue forever. This is the fundamental reason why all fiat currencies eventually fail.

You can't print gold? Are you saying you can't counterfeit "finite things"? Is that the main benefit here?

You could potentially print gold via nuclear fusion / fission reactors. However the cost would be several orders of magnitude higher than few grams of gold produced.

All successful forms of money share the following characteristics: counterfeit resistant, rare, fungible, easily transportable, little to no carrying cost, and high value per unit (another way to say rare). Its no accident that human society settled upon gold as representation of this ideal since it closely fits all these criteria. Its possible that crypto will change this however...

I think you worded your way around my question. You can fake people out without a nuclear reactor. And it looks like we settled on USD (or other workarounds to gold, for ages now), not gold.

In the secondary markets, so long as there is one. Currently a USD sells for a little over 1.5 cents of its face value. If it ever becomes impossible to sell for cash (gold) in the secondary market, it has lost its 'money substitute' quality.

Certainly not in Kansas, anymore, Toto.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact