Measuring market-based core inflation expectations
Abstract
- Abstract
We build a novel term structure model for pricing synthetic euro area core inflation-linked
swaps, a hypothetical swap contract indexed to core inflation. - The model provides estimates of market-based expectations for core inflation, as
well as core inflation risk premia, at daily frequency, whereas core inflation expectations from
surveys or macroeconomic projections are typically only available monthly or quarterly. - We
find that core inflation-linked swap rates are generally less volatile than headline inflationlinked swap rates and that market participants expected core inflation to be substantially
more persistent than headline inflation following the 2022 energy price spike. - In this paper, we aim to infer market-based core inflation expectations, which are otherwise
not directly observable because no financial asset directly tied to core inflation exists. - We deem this second assumption reasonable because HICP inflation itself is a linear combination
of core as well as energy and food inflation. - The level of 2 percent and relatively low volatility of
long-term inflation expectations suggests that inflation expectations are firmly anchored at the
ECB?s 2 percent inflation target. - This assumption appears reasonably uncontroversial,
as core inflation is a sub-component of headline inflation, which the observable headline ILS
rates are tied to. - Our estimates of core ILS rates reflect both market participants? genuine core
inflation expectations and a core inflation risk premium, but our model explicitly allows for
this decomposition. - The model-implied estimates of core ILS rates appear reasonable along several dimensions:
(i) like realized core inflation is less volatile than headline inflation, the core ILS rates are less
volatile than headline ILS rates, (ii) core ILS rates comove less with oil prices than headline
ILS rates, (iii) the core inflation expectations, as reflected in core ILS rates, typically evolve
similarly as the core inflation projections by Eurosystem staff, and (iv) consistent with market
commentary at the time, core ILS rates suggest that market participants expected core inflation
to be substantially more persistent than headline inflation following the 2022 energy price spike. - To the best of our knowledge, we are the first to price core ILS rates and decompose them into
market-based expectations for and risks around the core inflation outlook. - Our approach to inferring core ILS
rates from headline ILS rates, realized headline and core inflation as well as survey expectations
for headline and core inflation is also related to Ang et al. - Relative
to their study, we separately measure core inflation expectations and risk premia, we provide
core inflation expectations at a higher-frequency, and we provide evidence on the causal effectsECB Working Paper Series No 2908
6
of monetary policy shocks on core inflation expectations and risk premia.
- Specifically, we decompose the synthetic core ILS rates
into average expected core inflation over the lifetime of the swap contract and a core inflation
risk premium that compensates investors for core inflation risk. - In
our model below, this term is constant over time and relatively small, so we will simply refer
to the core inflation risk premium as the difference between the core ILS rate and the average
expected core inflation over the lifetime of the swap contract. - 3.2
Core ILS rates
To have a joint model for headline and core ILS rates, we need one further assumption on the
dynamics of realized core inflation. - The assumption that core inflation is driven by the same set of factors as headline inflation
should be relatively uncontroversial: since headline inflation is a weighted average of core and
food and energy inflation, it should reflect any factors driving core inflation. - If there are factors
driving food and energy inflation, which do not show up in core inflation, then those factors
should still show up in headline inflation. - In step two, to be able to infer the factor
loadings of core inflation, we would regress realized core inflation onto the estimated latent
factors to identify the additional parameters in equation (12). - Before the fourth
quarter of 2016, the SPF did not ask respondents for their core inflation expectations, so we
are not able to use survey-based information about core inflation before then. - Before
2016, the fitted core inflation series is somewhat above the realized one, potentially reflecting
that the model has limited information about core inflation over this early period due to the
lack of information about core inflation from surveys. - This could have been the
case if one of the factors moved core inflation and energy and food inflation in exactly offsetting
direction, so the overall impact on headline inflation was exactly zero. - During 2021, for example, there were
ECB Working Paper Series No 2908
25
Figure 7: Decomposition of synthetic core ILS rates
2y core ILS5y core ILS
5
45
ILSpremia
exp
4
ILS
premia
exp
3
3
2
2
1
1
0
0
-1
-1
-2
2017 2018 2019 2020 2021 2022 2023-2
2017 2018 2019 2020 2021 2022 202310y core ILS
5y5y core ILS
5
45
ILSpremia
exp
4
ILS
premia
exp
3
3
2
2
1
1
0
0
-1
-1
-2
2017 2018 2019 2020 2021 2022 2023-2
2017 2018 2019 2020 2021 2022 2023Note: Synthetic core ILS rates decomposed into genuine core inflation expectations and core inflation risk
premia. - ECB Working Paper Series No 2908
26
Figure 8: Decomposition of ILS rates
2y ILS5y ILS
5
45
ILSpremia
exp
4
3
3
2
2
1
1
0
0
-1
-1
-2
20062010
2014
2018
2022
-2
2006ILS
2010
10y ILS
2018
2022
5
ILSpremia
exp
4
3
3
2
2
1
1
0
0
-1
-1
-2
20062014
exp
5y5y ILS
5
4premia
2010
2014
2018
2022
-2
2006ILS
2010
premia
2014
2018
exp
2022
Note: ILS rates decomposed into genuine core inflation expectations and core inflation risk premia.
- We find that the headline inflation risk premium
indeed does responds more strongly than the core inflation risk premium. - The key
assumption underlying our approach is that traded headline ILS rates span core inflation, whichECB Working Paper Series No 2908
35
should be reasonably uncontroversial as core inflation is a sub-component of headline inflation.
- We fit the model to euro area headline ILS rates, realized headline and core inflation, and
both headline and core inflation expectations reported in the SPF. - Decomposing our core ILS rates into genuine core inflation expectations and core
inflation risk premia shows that shorter maturities mainly reflect core inflation expectations,
while the core inflation risk premium matters relatively more for longer maturities. - Our results suggest that a monetary policy tightening surprise significantly lowers
near-term core inflation expectations, although less so than it lowers headline inflation expectations.