Many sovereign bonds trade at a premium for being safe assets that retain their value even in deep recessions. This premium, or ‘convenience yield’, allows governments to borrow cheaply. This column argues that sovereign bonds remain particularly valuable in comparison to other assets, such as stocks, only in disinflationary demand recessions. In inflationary supply recessions, their real value erodes due to inflation and rising interest rates. Therefore, the recent resurgence of supply shocks has contributed to the erosion of the ‘convenience yield’ premium in sovereign bonds. This insight strengthens the case for fiscal prudence as well as for committing to restrained fiscal policy during inflationary recessions.
For a long time, many advanced economy governments have issued sovereign bonds at exceptionally low sovereign yields (Tambalotti et al. 2018). They have benefited not only from the low interest-rate environment, but also from investors’ willingness to pay a premium – the so-called ‘convenience yield’ – for very safe and liquid sovereign bonds (Vayanos et al. 2016, Krishnamurthy and Vissing-Jorgensen 2012).
However, this ‘convenience yield’ premium has declined substantially over the last three years, thereby driving up governments’ borrowing costs and reducing their fiscal space. One important reason is that the net supply of government debt available to investors is expected to increase. Sovereign debt issuance is projected to rise while the ECB’s bond holdings are expected to decline, making these bonds more abundant (Schnabel 2025, Bellon and Gnewuch 2024).
In this column, we argue that next to increasing net supply, there is another important reason for the decline in convenience yields. That is, the resurgence of supply shocks reduces the demand for sovereign bonds.
Generally, investors have a strong demand for very safe and liquid sovereign bonds and are willing to pay a premium, because these bonds are, in the words of Brunnermeier et al. (2024), “like a good friend”. That is, they are around (i.e. they are tradeable and even appreciate) when needed the most (i.e. in a recession, when most other asset prices fall). This negative correlation with other asset prices means that sovereign bonds act like an insurance for investors.
Sovereign bonds, however, provide insurance only in demand (disinflationary) recessions. In supply (inflationary) recessions, sovereign bonds are no longer an insurance because their real value erodes just like other assets. The special demand for sovereign bonds, which gives rise to the convenience yield, therefore depends on the expected composition of supply and demand shocks. In an environment where supply shocks are expected to dominate, sovereign bonds are less valuable, because they are less useful as insurance.
Following the COVID-19 crisis, the ensuing supply chain disruptions, and the energy crisis of 2022-2023, the perceived risk of another ‘inflation disaster’ remains elevated (Hilscher et al. 2024, Braggion et al. 2023) and supply shocks remain in focus. By the above logic, this resurgence of supply shocks in investors’ perceptions leads to a lower demand for sovereign bonds and thus provides another reason for the decline in convenience yields.
Sovereign bonds are “good friends” in demand recessions, but not in supply recessions
Figure 1 shows the real value of government bonds and stocks in two crises. In the Great Financial Crisis (left panel) – a disinflationary (demand-type) recession – real bond prices rose, supported by low inflation and falling interest rates, while real stock prices fell substantially. In contrast, during the energy crisis of 2022-2023 (right panel) – an inflationary (supply-type) recession – real bond prices fell, and even more so than real stock prices.
These episodes illustrate that government bonds are a valuable insurance against demand-type recessions, but not against supply-type recessions. In the face of demand shocks, bond prices are negatively correlated with stock prices (‘negative stock market β’), allowing investors to diversify. In the face of supply shocks, however, bond prices are positively correlated with stock prices and thus become unattractive for diversification.
Figure 1 Real asset prices in demand and supply recessions
Notes: Real asset prices are normalized to 100 in September 2007 (left panel, Great Financial Crisis) and in October 2021 (right panel, Energy Crisis). All nominal prices are deflated using euro area HICP (ex. Tobacco). Nominal bond prices are computed based on synthetic 10-year zero-coupon yields.
More supply shocks reduce convenience yields
The relationship between the expected composition of demand and supply shocks hitting an economy and the convenience yield attached to sovereign debt can be captured in a model in the spirit of Arcidiacono et al. (2024), in which investors buy sovereign bonds and a stock portfolio not only as a buy-and-hold investment, but also to potentially resell these assets in a crisis. The value of an asset thus reflects its buy-and-hold return as well as its resale value in a crisis. The convenience yield captures this resale value in a crisis.
The economy is exposed to both demand and supply shocks. In line with Figure 1, the real value of bonds falls less (or even rises) during a demand recession, whereas the real value of stocks falls less during a supply recession. It follows that, as the expected share of supply shocks increases, bonds become less attractive as a hedge against recessions, while stocks become more attractive. Hence, the convenience yield falls, as shown in the left panel of Figure 2.
Figure 2 Model: The convenience yield falls when the expected share of supply shocks rises
Notes: (Left panel) In this model, the convenience yield reflects the resale value of bonds in an average expected crisis. For more details, see Arcidiacono et al. (2024). (Right panel) The expected inflation rate is always 0 because there is no trend inflation and all shocks can be positive or negative with equal probability.
The expected share of supply shocks is not easily observable, but the inflation risk premium (IRP) constitutes a suitable proxy. 1 Conceptually, as shown in the right panel of Figure 2, the IRP is closely linked to the expected share of supply shocks. This is because the IRP reflects the value of a payout in a high inflation situation compared to the value in a low inflation situation. Between 2015 and 2021, the IRP was negative (see Figure 3), suggesting that demand shocks were expected to dominate, and periods of high inflation were associated with a good state of the economy (low value of payout). Since 2021, the IRP is positive, suggesting that the expected prevalence of supply shocks has increased, and periods of high inflation are increasingly associated with a bad state of the economy (high value of payout; see Burban et al. 2024).
Figure 3 Data: Inflation risk premium suggests that supply shocks are expected to be more important
Empirically, we find that an increase in the IRP – proxying for an increase in the expected share of supply shocks – is indeed associated with a decline in the convenience yields of several issuers of relatively safe and liquid euro area sovereign bonds. As shown in Figure 4, an increase of 100 basis points in the IRP is associated with a reduction in the convenience yield ranging between 5 and 7 basis points. To understand the magnitude of this effect, consider the case of German Bunds: the average convenience yield during the sample period (2015 – 2024) is 37 basis points, suggesting that a 100-basis point increase in the IRP – as observed between March 2020 and December 2023 – erodes 20% of this premium. The resurgence of supply shocks is thus an important reason for the decline in convenience yields.
Figure 4 Data: Convenience yields fall when the inflation risk premium rises
Notes: The coefficients depict the effects of a (daily) change in the IRP on the respective convenience yields. Convenience yields are measured as 10-year OIS rate + 10-year CDS rate – 10-year sovereign yield. The IRP is measured as five-year, five-years-ahead inflation linked swap rate – linearly-interpolated expected long-run inflation rate from the ECB’s SPF. Sample: 2015-2024.
Implications for fiscal policy
Our analysis highlights that sovereign debt is less attractive for investors in an environment where supply shocks are expected to dominate over demand shocks. This has at least two implications for fiscal policy.
First, it emphasises the need for fiscal prudence. The recent resurgence of supply shocks has reduced convenience yields on sovereign bonds, thereby driving up governments’ borrowing costs and tightening their fiscal space. However, the composition of shocks affecting the economy – being largely outside the control of the government – may change even further in the future. Tariffs, for example, represent a supply shock (Werning et al. 2025) and more frequent trade policy interventions may therefore further erode the convenience yield. 2
Second, our analysis highlights a novel mechanism that contributes to the trade-off that governments face during inflationary supply recessions. Smets et al. (2024) show that during such crises, the optimal policy features subsidies for firms and poor households while taxing rich households. Our work reiterates that these redistributive policies should be implemented as fiscally neutral as possible. By focusing on redistributive policies and restraining aggregate stimulus, governments not only avoid further fuelling inflation, but also limit the increase in the supply of sovereign debt, thereby propping up its price. In this way, government bonds become a more attractive investment in an environment with frequent supply shocks. In the words of Jiang et al. (2025), with an adequate fiscal policy, governments can to some extent “insure their bondholders” even against supply recessions. In demand recessions, bondholders are naturally insured as evidenced by Figure 1. Hence, committing to restrain aggregate fiscal stimulus in supply recessions makes sovereign bonds ex ante a more attractive investment and thus raises convenience yields.
Source : VOXeu
