Bond bubbles, MMT, and the limits to fiscal policy

Some reflections on this discussion between Brad DeLong and Noah Smith.

Most people who believe in a bond bubble hold inconsistent beliefs. A 10-year Treasury yield of 2% is obviously a bubble, right? After all, the economy is already growing at 2-2.5% real and can’t be that far away from full employment.

Ok. So how much do you expect interest rates to rise in the next five years? Most people say 3%, maybe 4%. Let’s stick with 4% – and let’s assume that’s the peak. That could be wrong, but it is not unreasonable.

So we now live in a world where the peak in interest rates is 4% and the trough is zero, so the average is around 2%. Well, the average fed funds rate is roughly what the 10-year treasury should yield. After all, you can hold a deposit for 10-years and earn the Fed Funds rate, or you can buy a 10-year and earn the yield-to-maturity. That’s why most people who believe in a bond bubble hold inconsistent beliefs: they think bonds are an accident waiting to happen, but they don’t believe that short rates will average more than the interest rates implied by bonds.

‘What about the term premium?’, I hear you clamouring. If you don’t know what it is and are currently silent, it’s the idea that the term structure of interest rates compensates for the uncertainty in future rates by paying a premium. I’m not convinced there should there be one. In fact, long-dated bonds are arguably an insurance policy: if the stock market crashes, or recession strikes (particularly when inflation is low) longer duration bonds will likely deliver higher positive returns. They’re diversifiers. Insurance policies don’t usually pay premia.

Of course, there are other considerations. The risk properties of bonds may be attractive due to their negative correlation with risk assets when inflation is very low, but when yields are also very low the objective distribution of returns is skewed to the downside. In English: the returns from bonds are low if we are all reasonably correct about the path of interest rates, but if we’re all completely wrong … the economy booms, the Fed Funds rate goes to 5% – and, thanks to a now-sound financial system – it stays there for a decade … bonds are toast. But hey, if the economy booms, who needs insurance?

So bonds may be a lousy investment, or a reasonable insurance policy – but it’s far from obvious that there’s a bubble.

The real limits to fiscal policy

If the bond market is not a bubble, it’s hard to worry about the government debt or deficits. After all, at these levels of real interest rates, across the yield curve, everyone looks solvent. But does this mean historically high debt and deficits can continue without limit.^[1] If not, where is the limit?

A related subject has been raised by MMTers. I’m not an expert in MMT, by any stretch. And I’m not a big believer in the “football team” approach to thought. That’s why we have sport. But more importantly, most of what they say about fiscal policy seems broadly correct. The constraints on fiscal policy are determined by two factors: 1) can you print your own money, and 2) is unemployment already so low that fiscal stimulus is inflationary.

I think 1) and 2) do determine the limits to fiscal policy – I also think they are logical conclusions from the work of voices as disparate as Keynes, Friedman, Sargent & Wallace, and Stephanie Kelton.

The eurocrisis was not caused by “too much” sovereign debt, but the fact that the ECB wasn’t doing QE (i.e. the issuers of debt were not backed by a printing press). That was reflected clearly in asset price behaviour at the time^[2] (non-Eurozone bonds rallied when Italian spreads were ballooning), and has gradually become consensus among policy-economists (starting with Paul de Grauwe).

But if you can print money and buy government debt, what’s the limit on government borrowing? The limit has to be inflation. One of the things I dislike about the unthinking obsession with “expectations” in today’s monetary policy discourse is the suggestion that “inflation expectations” can suddenly – out of nowhere – go haywire. This a bit like the idea of bond panic. One day – for some unknown reason – the bond market is going to think that the government won’t raise future taxes, and panic.

Now temporary bouts of severe market volatility (to use a more technical expression for “panic”) are not just possible – they are likely (because of myopia, correlation and leverage) – but a sustained problem is what actually matters.

But if the central bank can do QE, there cannot be sustained bond panic in the absence of a genuine inflation problem. Why? Because, as the BoJ is showing, faced with no inflation risk, the central bank can buy all the bonds.

All that matters then is what causes a sustained rise in inflation. The idea that the population wakes up one day and decides that because the national debt has gone through the Reinhart and Rogoff limit, or because a check from the Fed has arrived in the post, there is going to be a wild outbreak of inflation, is unconvincing.^[3] By-and-large, we live in a highly competitive de-regulated world where prices are set independently of what we think policy-makers are up to.

Could that change? Of course it could. If wages start to rise significantly and policy-makers do nothing about it. If the Fed comes out and says ‘we know wages are rising, we expect prices to start rising too, and we’re going to leave rates at zero’. And the government joins in and says, ‘yes wages are rising, prices are rising and we’re going to increase the deficit’. Sure, that would cause a bond panic. It’s also totally unrealistic.

So fiscal policy really isn’t a problem. The constraints on fiscal policy are actually benign: when you approach full employment, let the deficit shrink.

Postscript: a short comment on Benanke

Ben Bernanke addresses the subject of the term premium in bond markets in his latest blog. He essentially points out that it has fallen over the longer term, mainly due to falling inflation risk premia. But in his subsequent analysis he omits the most important factors: co-variance and – prospectively – the zero bound.

If global interest rates are impacting US yields further out the yield curve, it is not really a term premium. I think we have to analyse the global term premium. Also, we have to pay attention to the bond/equity correlation – covariance should impact the risk premium. The shift from a negative to a positive bond/equity correlation was a significant reason why “tapering” was so pronounced. More so when you include myopia & leverage. More fundamentally, there should be two main drivers behind the trend decline in the term premium. Absent inflation risks, bonds are recession insurance – why should an insurance policy “pay” a premium? However, as yields approach the zero bound the objective distribution of returns necessarily is skewed towards loss – term premia should rise (i.e. Japan).

^[1] Actually, this isn’t really true. Public sector debt in all the “QE countries” is much lower than reported due to a basic accounting error: 1) the central bank should be consolidated into the government’s balance sheet, and 2) base money is not a liability. Therefore you can therefore cut the UK national debt by roughly 25% to get to the correct number. See here and here for discussion.

^[2] If you’re in any doubt, read this.

^[3] As I have argued here in response to Tony Yates.