Methods for Combining Monthly and Semiannual Data in Regression

Summary

The document addresses a regression problem in which market or company observations are monthly while accounting fundamentals, such as price-to-earnings, book value, and cash flow, arrive semiannually. It points to mixed-frequency econometric methods as ways to use variables observed at different intervals without requiring that every series share the same frequency.

The methods named are bridge equations, mixed-data-sampling models, mixed-frequency vector autoregressions, and mixed-frequency factor models. MIDAS is identified as a well-known approach, but the document does not describe model specifications, timing conventions, or how to avoid using information before it would have been available. It offers no empirical comparison or worked example, and directs readers to a survey for further detail.

Key ideas

• Regression can combine variables observed at monthly and semiannual frequencies using mixed-frequency methods.
• MIDAS models are presented as a prominent approach to mixed-frequency data.
• Other options include bridge equations, mixed-frequency VARs, and mixed-frequency factor models.
• The document lists methods but does not explain estimation, data timing, or implementation choices.

Tags

# Regression extensions


# Regression extensions












I'm trying to find extensions for my regression and obviously would like to use PE, BV and CFO. But I've got monthly data, while all company's fundamentals are semi-annually... Can I deal with it somehow?

## Answer by tagoma (score 1)

https://quant.stackexchange.com/a/7684

It exists several techniques to deal with mixed-frequency data. I believe MIxed DAta Sampling is the best-known.

Eg:

- bridge equation,

- MIxed DAta Sampling (MIDAS) models

- Mixed frequency VARs

- Mixed frequency factor models

- ...

Here is a good document on this topic: A survey of econometric methods for mixed- frequency data

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.