Has Wage Growth Fueled Inflation in the UK?
No, and the BOE can safely cut interest rates
This is a joint post with Ivan Petrella;
Related Blogs:
FT Alphaville on the Catch-22 effect August 1st 2024;
At the Core of UK Inflation, July 9th 2024;
Bank of England still in a Catch-22 mode, April 17th 2024;
The figure above emphasizes the key conclusion of our analysis: supercore wage pressure shocks have had minimal impact on supercore service inflation during the post-pandemic recovery.
From a policy perspective, this suggests that the Bank of England should continue its easing cycle in the upcoming meeting.
Stepping back:
We aim to evaluate the relationship between supercore service inflation and supercore wages in the UK economy. Our baseline model employs a Vector Autoregression (VAR) that includes the Global Supply Chain Pressure Index (GSCPI) as a proxy for global supply chain shocks, along with our measures of supercore wage pressure and supercore service inflation (see At the Core of UK Inflation for details on how these measures were constructed).
Key takeaways:
Our econometric analysis suggests that, in general, supercore wage shocks are a persistent source of supercore service inflation.
The historical decomposition shows that in the post-pandemic period, wage pressures are not the main contributor to the persistence of supercore service inflation.
Global Supply Chain Pressure Index (GSCPI) shocks and supercore service inflation shocks have been the primary drivers of inflation in recent periods.
From a policy angle, this analysis provides context for the Bank of England’s justification for further monetary easing.
Background
Assessing the sources of inflation persistence is a key policy question. In the recent Jackson Hole speech, Governor Bailey states:
The job of monetary policy – which it has been doing– is to squeeze the persistent element of inflation out of the system in a way that is consistent with returning inflation to its target on a timely and sustained basis. When we think about inflation persistence, I would distinguish between extrinsic persistence (the duration of external shocks) and intrinsic persistence in the sense of capturing the impact of the echo effects of those external shocks owing to domestic responses (second-round effects).
This is relevant in terms of designing the appropriate policy response. Indeed, Bailey says:
Is the decline of persistence now almost baked in as the shocks to headline inflation unwind, or will it also require a negative output gap to open up, or are we experiencing a more permanent change to price, wage and margin setting which would require monetary policy to remain tighter for longer? This framework is now prominent in our thinking on the MPC.
In the context of the UK, the debate has centered around the persistence of supercore service inflation (see the May 2024 Monetary Policy Report) and the extent to which nominal wage growth has been a driving component of its persistence.
To conduct our analysis, in a previous post we constructed a measure of supercore service inflation and supercore wage pressure.
More specifically, our supercore service inflation refers to a subset of the core inflation measure that focuses on the prices of services excluding housing. Our supercore service inflation removes housing-related costs from the service component of the Consumer Price Index (CPI)). In our supercore measure, only actual rents are excluded, while the other housing components are still included (see At the Core of UK Inflation).
By focusing on non-housing services, supercore inflation aims to capture price changes in areas such as healthcare, education, and personal services, which can be more indicative of the underlying inflation dynamics driven by wage growth. One reason to exclude actual rents is that their adjustment is less directly associated with the labor market and potentially more affected by changes in interest rates and mortgage rates (see the blog post).
In a parallel fashion, we also constructed a supercore wage pressure index meant to capture nominal wage developments associated with the sectors that are included in the supercore service inflation measure.
Analysis
To assess the extent to which supercore wages have been accounting for the persistence of supercore service inflation we conduct a simple econometric analysis using as the baseline specification a three-equation VAR ordering the Global Supply Chain Pressure Index (GSCPI), Supercore Wages and Supercore Inflation. Let’s briefly review its construction and interpretation.
In the VAR model, using a Cholesky decomposition, the ordering of the variables in the system reflects the underlying assumption about how shocks propagate contemporaneously across these variables. We select the following ordering of the variables: GSCPI, supercore wage pressure, and supercore service inflation.
The Cholesky ordering imposes a recursive structure, meaning that the first variable in the list is assumed to be contemporaneously exogenous to the others, while subsequent variables can be influenced by those ordered before them.
The Cholesky decomposition that we have chosen is consistent with an interpretation where global supply chain conditions are more exogenous and drive changes in domestic wages and inflation in the short term, while supercore service inflation is viewed as a downstream effect of both supply pressures and wages, without contemporaneously feeding back into them. With this ordering, we are also implicitly assuming that service prices adjust to shocks faster than wages.
This ordering is consistent with an economic narrative where global supply factors are seen as key drivers of inflationary dynamics, with wages responding to both supply and price conditions, and inflation being the ultimate result of pressures from both the supply and wage sides.
Let us briefly review in more detail the interpretation of this specific ordering:
Global Supply Pressure Index (First in the order)
A shock to the GSCPI (such as a global supply chain disruption) can contemporaneously affect both supercore wage pressure and supercore inflation in the same period. However, the GSCPI itself is neither contemporaneously affected by shocks to supercore wage pressure nor supercore service inflation.
One way to interpret this assumption is to suggest that any changes in the GSCPI are considered independent of short-term fluctuations in wages or inflation during the same period. This implies that GSCPI shocks are treated as exogenous factors, unaffected by domestic economic conditions.
Supercore Wage Pressure (Second in the order)
Supercore wage pressure is affected contemporaneously by a shock to the global supply pressure index, but supercore wage pressure does not contemporaneously respond to supercore service inflation. However, a supercore wage pressure shock (such as an unexpected or sudden increase in wage growth indicating a deviation from the typical wage trend, driven by factors such as tight labor markets, increased demand for services, or other sector-specific pressures) influence supercore service inflation within the same period.
Following our identifying restrictions, wage dynamics (supercore wage pressure) are assumed to respond to global supply chain conditions. However, wages do not respond directly to inflation within the same period. On the contrary, wages may influence inflation—consistent with wage increases leading to higher production costs that would be passed through prices.
Supercore Service Inflation (Third in the order)
Supercore inflation is affected contemporaneously by shocks to both the global supply pressure and supercore wage pressure indexes, but it does not affect either of these variables contemporaneously (since it is last in the order).
Supercore inflation is seen as a result of both global supply and domestic wage pressures. A supply shock (like a global shortage of goods) can directly affect inflation, and if wages increase due to domestic labor market tightness, that pressure can also pass through to prices. However, inflation itself does not feed back into wages or supply conditions within the same period, though it may have a lagged effect over time.
The other aspect that characterizes our empirical specification is that we allow for dynamic interactions with 5-period lags.
This specification implies that the interaction between variables also plays out over time. For example, while supercore inflation does not contemporaneously affect supercore wage pressure or the global supply pressure index, it could influence them with a lag.
Summary of Shocks Propagation Mechanism
A shock to the global supply pressure index immediately affects both supercore wage pressure and supercore inflation in the same period. For example, a sudden increase in global supply chain disruptions increases both wages (through increased production costs or wage demands) and inflation.
A shock to supercore wage pressure can be caused by global supply shocks, and it will affect supercore inflation contemporaneously. However, a wage shock does not immediately change global supply pressures.
A shock to supercore inflation is influenced by both global supply pressures and wage pressures within the same period. However, inflation does not affect either of these variables contemporaneously. In this context, inflation is treated as an outcome of global supply constraints and wage dynamics. Over time, inflationary pressures can feed back into wages or supply conditions, but only with a lag.
Findings and Interpretation
For the most advanced reader, we provide the details of our empirical model in the Appendix at the end of this post.
We estimate the model monthly over the sample from March 2001 to June 2024.
We first examine the impulse response functions of the three shocks (GSCPI, Wage pressure index, and supercore inflation) on supercore service inflation.
As a background, the Impulse Response Functions measure the impact of a one-time unexpected shock (impulse) on one of the variables on the current and future values of all variables in the system.
It helps us understand how shocks to a specific variable propagate through the system over time, affecting both the variable itself and the others in the model.
In the following graphs we plot the response along with the 68% confidence interval (this means that 68% of the time the impulse response of a variable to a shock lies within the 68% confidence band).
The GSCPI shock has a short and relatively smaller impact on UK prices. In contrast, wage shocks tend to accumulate into inflation over time, with a noticeable effect emerging after six months. This effect is also quite persistent, as it remains significant even after 24 lags. Interestingly, wage shocks have a very long lag, almost behaving like a permanent shock, as their impact at 6, 12, and 24 lags are quantitatively similar. This persistence is what makes wage shocks the most concerning for monetary policy authorities among the three shocks considered.
We then move to the historical shock decomposition focusing on the period running from the beginning of the pandemic onward.
As a background, the historical shock decomposition in a VAR is a technique used to assess the contributions of different shocks to the evolution of each variable in the system over time. Here we focus on supercore service inflation and we decompose the contributions of each shock to it.
An interesting finding is that wage pressure index shocks have not significantly contributed to supercore service inflation during the post-pandemic inflationary period. However, historically, there were periods, particularly between 2005 and 2016, where wage shocks were the primary drivers of supercore inflation.
Our findings suggest that the observed wage growth fits a pattern where wages adjust in response to inflation (following an inflation shock) without leading to a harmful wage-price spiral. This is consistent with a bargaining mechanism where wages rise following past cumulative inflation, while inflation expectations remain stable, preventing current wages from factoring in high future inflation.
Additionally, the recent decline in supercore service inflation is largely due to the reduced impact of GSCPI and service inflation shocks, although these factors remain significant and continue to play a role in the gradual adjustment of supercore service inflation.
We emphasize here that most of the variability of service inflation is accounted by the service inflation shock that should be categorized as a “catch-all” shock, encompassing factors such as commodity price shocks but excluding shocks that are directly linked to significant contemporaneous wage increases.
Finally, in light of the current policy discussion, it is important to note that, in the post-pandemic period, wage pressures have not been the primary factor behind the persistence of supercore service inflation.
Conclusions
In interpreting the dynamic of inflation persistence in the UK, Governor Bailey’s Jackson Hole speech outlined three scenarios that correspond to three different policy responses.
“The first of these cases is the more benign – the persistence is essentially self-correcting with the degree of restriction we have in place today easing off over time….
The second case is the intermediate one. Here we would need to maintain restriction for longer and thus open up more of an output gap.
The last case is least benign and would require more restrictive policy than the first two cases. It would suggest that there are structural changes in product and labour markets going on which are causing the supply side of the economy to change as a lasting legacy of the major shocks we have experienced.”
In this blog, we have used a VAR model to examine the key determinants of supercore inflation in the UK, which is typically more sensitive to labor market dynamics, particularly wage inflation.
Our findings suggest that wages have not been the primary driver of supercore inflation.
From a policy perspective, this analysis leans towards supporting Bailey's first scenario and offers context for the Bank of England's argument for additional monetary easing.
Appendix
Formal Representation
More formally we can represent our VAR as
Where:
GSCPI is the global supply chain pressure index;
WP is the supercore wage pressure index;
SCPI is the service supercore inflation.
With
\(c_1, c_2, c_3\)representing the constant terms;
With
\(\alpha_{i,j,i}\)representing the lagged coefficients.,
With
\(\epsilon_1,\epsilon_2, \epsilon_3\)representing the error terms.
Cholesky Decomposition:
The Cholesky decomposition imposes a recursive structure on the contemporaneous relationships among the error terms:
where
are the structural shocks.
The matrix imposes the recursive ordering: shocks to GSCPI (eta(1)) affect all variables contemporaneously, shocks to WP (eta (2)) affect WP and SCPI but not GSCPI, and shocks to SCPI (eta(3)) affect only SCPI contemporaneously.
This system reflects how shocks propagate according to the ordering of the variables under the Cholesky decomposition.