Separating Global and Local Yield Movements with PCA

Summary

The document considers how to separate a shared global component of bond yields from local influences. It suggests applying principal component analysis to yields at several curve maturities, from shorter tenors to the long end, as a way to identify common movements empirically. The response also proposes that global growth conditions can help explain why yields across regions tend to move together.

Demographics, including population growth, and productivity are cited as possible drivers of longer-run yield levels and movements. The discussion notes a practical data constraint: evaluating population-growth relationships over very long horizons requires historical series that may not be available for many countries. It also flags cross-country inflation differences as a complication. The material does not specify a PCA setup, sample design, or method for distinguishing expected from unexpected yield changes, credit effects, and other local factors; its economic explanations are suggestions rather than demonstrated causal findings.

Key ideas

• Principal component analysis across multiple yield-curve maturities can summarize shared yield movements.
• Global growth is offered as one reason yields across regions may be correlated.
• Demographics and productivity are proposed as potential drivers of interest rates and yields.
• Long historical demographic data may be unavailable for many countries.
• Inflation differences and local influences complicate comparisons of yields across regions.

Tags

# decomposition of yields into global and local components


# decomposition of yields into global and local components












It is reasonable to assume that global yields move in tandem to a certain extent, driven by a global and a local component. Are there any ways to separate the two, beyond the obvious (regress the local yield changes onto an average change across all yields)? Any pitfalls, like expected/unexpected changes, credit risk, inflation etc?

## Answer by demully (score 1)

https://quant.stackexchange.com/a/63290

Yes, there IS a global yield common component here. PCA on 2s, 5s, 10s and the long-end of yield curves would objectively describe this for you.

I agree with the earlier answer that population growth is an important factor here; but there are others. Demographics is one half of this; the other is productivity, ie the output per worker. Faster growing (for either reason) economies will tend to have higher interest rates/yields. All of this before we start to think about different inflation rates in different currencies :-)

Short answer is that yields globally ARE regionally correlated, because global growth is an important factor in all of them regionally.

## Answer by Sergei Rodionov (score 0)

https://quant.stackexchange.com/a/63272

I subscribe to a theory, originally postulated by Kapitsa, Jr. in Модель роста населения Земли и экономического развития человечества, that the main predictor for the global yield component would be the population growth rate. The challenge with incorporating such variables in regression analysis is that one needs to have access to time series which span 100+ years. This data is unavailable for many countries.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.