The debate between Larry Summers, Brad DeLong and Paul Krugman over models and confidence is summarised excellently by Martin Sandbu. On one point, Krugman is absolutely right: there was never any reason to be concerned about a panic in the money-printing world’s sovereign bond markets after 2008.
The evidence of financial markets discriminating between bonds in the Eurozone and everywhere else is unambiguous. And Krugman correctly argues that this empirical fact is supported by ‘theory’ (or models), and the ‘gut instincts’ of those thinking otherwise are unhelpful and poorly thought-through.
The first point to establish is that the ability to print money is key. If Greece faces deflation it also runs the risk of government default. If the UK faces the risk of deflation, the Bank of England will buy government bonds. [Note: although there are precedents where money-printing governments have defaulted on domestic currency debt (i.e. Russia) – NONE have done so when the threat was deflation].
It follows logically that pricing Greek bonds is all about credit risk – and there can be self-fulfilling runs, caused also by austerity. If Greek bond yields rise because the market worries about default, default risk rises because financing the government is mathematically more onerous.
But how do markets price government bonds in the rest of the developed world? In the money printing world, the relevant model is the no arbitrage model of asset pricing, which in this instance means bonds prices will predominantly be determined by interest rate expectations.
So we can accurately define what a bond market ‘panic’ in the money-printing world looks like – a large sustained decline in bond prices can only occur if the market suddenly anticipates a sustained increase in short-term interest rates.
Importantly – and this is key – government bond markets in the money-printing world are self-correcting – the opposite of bonds in the credit risk world, where panics are self-fulfilling. Why? Because if bond yields in the US rose because of a some panic about debt levels, this itself would cause the economy to weaken and result in a fall in interest rate expectations (and/or more QE), which supports bond prices.
In summary, when the risk is recession and deflation, there cannot be a panic in government bond markets in money printing economies – and a fall in bond prices would self-correct. However, if a government can’t print money and there is the threat of recession and deflation, sovereign bond panics are self-fulfilling. That is what theory – and all the evidence – overwhelmingly suggests. It is also why the ECB caused the Eurocrisis.
Ok, so this is precisely why a ‘bond panic’ in the money-printing world was always a fiction, reflecting a poor understanding of asset pricing. If the economic threat is deflation it makes no sense for markets to anticipate a sustained rise in short-term interest rates. If rate expectations sit materially below bond yields – regardless of the level of government debt – bond prices will rise.
So Krugman is right, and was right. And market behaviour and correlation repeatedly confirmed this view – not just in singling out Eurozone bonds, but also for example, when JGBs rallied around Fukushima, despite a huge expected increase in Japan’s budget deficit.
It is also the case that this is precisely what the most compelling, and basic, models of asset pricing would predict.
We don’t need to get distracted by the term premium – which should almost certainly be close to zero if there is a material risk of deflation. There are certainly situations where term premia should rise with government debt levels, but none were relevant in the last decade. Moreover, during the Eurocrisis the global stock of “safe” assets was shrinking – where “safe” is best thought of as negatively correlated with risk assets, such as equities.