Transmission channels, dynamic mediation, and higher-order impulse responses

summary

The gist

This paper introduces a novel concept of impulse response decomposition to disentangle the dynamic contributions of mediator variables in the transmission of structural shocks.

In short

The study introduces impulse response decomposition to separate how structural shocks affect different variables along a transmission path. By linking Sims' impulse responses with higher-order Granger causality, the authors show that a mediator's impact is determined by its exposure margin and propagation margin. Applying this to oil supply shocks reveals that consumer price movements significantly drive industrial production decline in the first six months.

Key concepts

Impulse Response Decomposition
This method breaks down a total effect of a shock into specific contributions from each variable along its path. It uses recursion between Sims and higher-order impulse responses to identify how much each mediator contributes at every stage of the transmission process.
Exposure Margin
This margin measures how much a specific shock moves a particular mediator variable. It quantifies the initial sensitivity of the mediator to an exogenous change, indicating its direct susceptibility to the structural shock being studied.
Propagation Margin
This is a set of coefficients that carry the movement from one variable (the mediator) forward in time to affect the final outcome. It represents how much a change in the mediator's level influences the subsequent variables and ultimately, the target outcome.

Terminology used across episodes

This episode discusses

The paper

Transmission channels, dynamic mediation, and higher-order impulse responses · Read on arXiv

Jean-Marie Dufour, Endong Wang

McGill University · University of Mannheim

Transcript

Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.

Ruth: We're looking at the paper: "Transmission channels, dynamic mediation, and higher-order impulse responses".

Warren: Here's the gist of the paper. This paper introduces impulse response decomposition to disentangle the dynamic contributions of mediator variables in structural shock transmission, establishing a formal link between Sims and Granger causality Core Methodology The paper develops a novel concept of impulse response decomposition to disentangle the dynamic contributions of mediator variables in the transmission of structural shocks This approach systematically decomposes impulse response functions into distinct components, providing a more precise characterization of macroeconomic transmission mechanisms The decomposition framework formally links Sims and Granger causality, stating that Sims’ impulse responses capture the total effect of an exogenous intervention, while Granger causality corresponds to the mediation effect Theoretical Foundation The authors justify their decomposition by embedding it within the framework of causal mediation analysis, demonstrating that the impulse response decomposition with respect to mediators coincides with the average mediation effect under a parametric VAR model They establish that multi-horizon Granger non-causality is a sufficient condition for zero average mediation effect at that horizon Furthermore, they introduce a novel dynamic mediation index constructed via a cosine-based inner product applied to the weighted components of the impulse response decomposition Decomposition Structure The impulse response can be decomposed into variable-specific contributions at horizon zero, where the contribution of mediator M at horizon zero is defined as θ(M0)h = Φ(h)•M,1θ M,0 This decomposition broadens the conventional interpretation of impulse responses by showing that they can be understood not only as reactions to structural shocks, but also as the cumulative effects of exogenous shifts in economic observables The framework extends to an arbitrary intermediate horizon n, yielding a unified representation of dynamic responses by systematically tracing the evolution of endogenous variables and projecting their sequential effects onto the outcome at horizon h Empirical Findings Applying our framework to studies of the transmission channels of US monetary policy, we find that investor sentiment explains approximately 60% of the peak aggregate output response in three months following a policy shock, while expected default risk contributes negligibly across all horizons The dynamic mediation index for the excess bond premium peaks within the first few months, indicating that intermediary-driven fluctuations in market sentiment constitute a dominant short-term mechanism in the transmission of monetary policy The paper concludes that investor sentiment acts as a significant short-term amplifier, while expected default risk contributes minimally across all horizons This framework offers a unified and flexible econometric tool for investigating causal channels and mechanisms in macroeconomic shock propagation The dynamic mediation index captures the directional and time-varying contribution of mediators to the overall impulse response Proofs

Paper summary: Ruth: So this paper, "Transmission channels, dynamic mediation, and higher-order impulse responses," it introduces a way to break down what an impulse response function is showing us. It claims we can figure out exactly how much of the total effect comes from each variable in the system over time.

Warren: Right, so instead of just seeing one big reaction to a shock, this framework lets you see which parts are driving that reaction at every stage. It links what Sims called impulse responses with Granger causality, which is another way we look at cause and effect in these models.

Iris: The authors justify this decomposition by showing it's equivalent to the average mediation effect under a parametric VAR model, but they also establish a formal link between Sims’ impulse responses capturing the total effect and Granger causality corresponding to the mediation effect > <ref:2509.05284#pg1>

Ruth: Exactly, so what it means is that if you look at Granger noncausality from one of those mediator variables to the outcome, it implies a zero propagation margin for that variable > <ref:2509.05284#pg3>

Warren: That means the contribution of that specific mediator to the total response is exactly equal to what we would see from a controlled mediator path effect > <ref:2509.05284#pg3>

Iris: They provide inference for these contributions and they also offer a test specifically for that propagation margin, which is pretty important because it lets us check if their zero-margin claim holds up when we look at actual data > <ref:2509.05284#pg3>

Ruth: And they apply this decomposition to US oil supply news shocks, and they found that consumer-price movements in the first six months make a large contribution to the subsequent decline in industrial production, while the federal funds rate's contribution is small throughout > Abstract

Warren: It suggests that for things like oil supply news, you can really see how different channels matter at different points in time, whereas other shocks might be more uniformly driven by one thing > Abstract

Iris: That’s interesting because it moves beyond just saying a channel matters to showing *when* and *how much* it matters dynamically in the system > <ref:2509.05284#pg2>

Ruth: It really helps us understand how policy shocks or external news ripple through the whole economy, not just what happens at the final destination > <ref:2509.05284#pg2>

Warren: So this paper is essentially giving us a tool to map out those hidden transmission channels that we usually miss when we only look at the total impulse response > <ref:2509.05284#pg2>

Iris: The dynamic mediation index they construct, using a cosine-based inner product on the weighted components of the decomposition, is a novel way to quantify this evolving role of mediators > <ref:2509.05284#pg1>

Ruth: So what's the big picture here for us? It gives us a structured econometric tool to identify these specific macroeconomic transmission channels, which is much better than just looking at one aggregate reaction > <ref:2509.05284#pg3>

Paper summary: Warren: And it connects the theory of Sims' impulse responses to Granger causality in a way that makes their interpretation more precise about those causal links > <ref:2509.05284#pg1>

Iris: The paper mentions that investor sentiment explains approximately sixty percent of the peak aggregate output response in three months following a policy shock, while expected default risk contributes negligibly across all horizons when applying this framework to US monetary policy studies > Abstract <ref:2509.05284#pg1,that investor sentiment explains approximately 60% of the peak aggregate output response>

Ruth: It shows that investor sentiment acts as a significant short-term amplifier, but expected default risk doesn't really show up as much across all time horizons > Abstract

Warren: So if you’re looking at how things like the federal funds rate affect output, this suggests that market sentiment is the main short-term driver of that response > Abstract

Iris: The paper flags a limitation where it focuses on structural shocks, and they don't explicitly address what happens when people change their minds about those underlying economic assumptions > <ref:2509.05284#pg2>

Ruth: And that’s a key thing for us, because the model is built assuming the shock itself is exogenous, but in reality, people’s expectations are constantly changing > <ref:2509.05284#pg2>

Warren: So while this decomposition gives us great insight into the channels *given* a shock happens, it doesn't fully capture the feedback loop where those mediators themselves are reacting to policy or market changes > <ref:2509.05284#pg3>

Iris: That’s fair, because isolating the response of an outcome variable from each mediator is what they do, and that isolation is what allows them to attribute specific portions of the total impulse response > <ref:2509.05284#pg3>

Ruth: So for a listener just listening to this, it means we can get past seeing one big reaction and start seeing which specific economic actors or variables are doing the heavy lifting at different times > <ref:2509.05284#pg1>

Warren: It gives us a way to test these dynamic roles, not just assume them based on intuition about how markets work > <ref:2509.05284#pg3>

Iris: Ultimately, this framework provides a unified and flexible econometric tool for investigating causal channels in shock propagation, which is useful for setting up those more complex models we need > <ref:2509.05284#pg3>

Ruth: So to wrap up the main point of "Transmission channels, dynamic mediation, and higher-order impulse responses," it’s about moving from just seeing the total effect to precisely mapping out how each variable contributes dynamically along the way > Abstract

Warren: And that's what they do by linking Sims and Granger causality through this decomposition framework > <ref:2509.05284#pg1>

Iris: It gives us a clearer picture of those dynamic causal mechanisms in structural shocks, which is useful for understanding policy transmission more deeply > <ref:2509.05284#pg3>

Conclusion: Ruth: So we’re wrapping up our look at this paper, "Transmission channels, dynamic mediation, and higher-order impulse responses." It basically shows how to untangle what's really happening when a shock hits an economy over time.

Warren: Right, it moves beyond just the total effect of a shock and lets you see which variables are actually doing the work at different points in time.

Iris: The authors are connecting things they used in Sims’ impulse responses with Granger causality, which is a way to measure if one variable causes another over time.

Ruth: So, when they do this decomposition, what's the main point about how we should look at economic transmission?

Warren: The main point is that it lets you pinpoint the exact contribution of each mediator at every step of the chain, which is crucial for understanding policy or news transmission.

Iris: They show that if you see Granger noncausality from a mediator to an outcome, it means there's zero propagation margin for that variable in their framework.

Ruth: That’s a big claim—that zero propagation margin—what does that imply about the causal link?

Warren: It implies that the mediator's contribution to the total response is exactly what you would expect from a controlled path effect, which gives you a clearer causal attribution.

Iris: They provide an index they call the dynamic mediation index, which uses a cosine-based inner product on those decomposed components to quantify this evolving role of mediators.

Ruth: So for someone just listening to the show, what does that actually change about how they see these economic shocks moving through the system?

Warren: It changes it from just seeing one big reaction to seeing which specific economic actors are amplifying or dampening that response at different stages.

Iris: It gives us a more flexible tool to test those causal channels, allowing us to check if our assumptions about how things flow actually hold up in the data.

Ruth: So, when we look at oil supply news or monetary policy shocks, this framework helps us see exactly which components—like consumer prices versus the federal funds rate—are driving the reaction and when.

Warren: Exactly, because it lets you see that for one shock, some variables matter a lot in the short term while others stay small across all horizons.

Iris: It’s a powerful way to map out these hidden mechanisms without just assuming how things work intuitively.

Ruth: This paper sets up a way to get past just seeing the total effect and start mapping out the specific dynamics of economic transmission.

More episodes

← Home