Estimating Daily Risk-Free Rates from Monthly or Weekly Yields

Summary

The document considers how to estimate daily risk-free rates when Treasury bill yields are available only monthly or weekly, a data limitation the author says can arise in developing economies. The proposed method converts a period yield into an equivalent daily rate by taking its compound-root over an assumed number of trading days: 20 for a month or 5 for a week. The intended use is to calculate excess returns in international analyses.

The question is whether this interpolation is reasonable or whether it adds noise, and whether ordinary returns might be preferable. The document offers no empirical comparison or conclusion. Its method assumes the observed period yield is a suitable proxy for each day in that period, and the chosen day counts and yield conventions matter. It also mentions interbank rates as an alternative but notes that daily observations may be scarce there too.

Key ideas

• A monthly or weekly yield can be converted into a constant daily compounded rate using a period root.
• The proposed daily-rate estimate is intended for excess-return calculations across international markets.
• The approach assumes a period-end yield represents daily risk-free rates throughout that period.
• Sparse yield data may make daily estimates noisy, and the document does not compare the alternatives empirically.

Tags

# daily risk-free rate proxy


# daily risk-free rate proxy












3-month or 1-month Treasury bill yields are typically used as proxies for the risk-free rate. However, it is often the case that - especially for developing economies - that data is only available at weekly or monthly frequency. This raises the question of whether an end-of-the-month yield on a 3-month Treasury bill for a given month would be a good proxy for daily yields within that month. Basically I would compute the daily yield $r_d$ for all days $d$ belonging to month $m$ from the monthly 3-month yield in that month: $r_m$

$(1+r_m)^{1/20} = 1 + r_d$

In the weekly case we would divide by 5 instead of 20.

The objective is to get an estimate of the risk-free rate to compute excess returns in an international setting.

Other alternatives would include interbank rates but daily data is often scarce as well.

My question is whether this is a viable option to estimate daily risk-free rates or whether this is likely to just introduce noise in the computation of excess returns (maybe simply working with returns would be a superior approach).

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.