New WRDS Data Release Accounting for Noisy TRACE Prices:
October 2023 Release: Download the new TRACE WRDS-based bond panel data including many price-based bond characteristics (Yield, Credit Spread, Bond Market Cap, Price, Illiquidity, Bond Return, Duration and Convexity) that have been adjusted for measurement error/noise. We follow the cleaning advice of Andreani, Palhares and Richardson (2023) to properly align the TRACE-based WRDS data to that of the industry standard Bank of America Merrill Lynch quote data. The updated factors can be downloaded here. Download the README file for the MMN-corrected data here.
You can download the factors constructed with the industry-standard Bank of America Merrill Lynch quote data here.
To purge bond short-term reversal signals of measurement error/noise, we follow Dickerson, Robotti and Rossetti (2023) To adjust for measurement error in other bond characteristics (credit spreads, yields etc.), we follow an adjusted procedure from Bartram, Grinblatt and Nozawa (2023), and set the “month-end” price-based investment signal to be the value a single business day before the transaction price used to compute a monthly return.
Inspired by Dick-Nielsen, Feldhütter, Heje Pedersen and Stolborg (2023). We also provide a correction for the truncation of bond returns at the +100% level by setting these truncated values to the return from the Bank of America Merrill Lynch data and if missing, the return computed by us using the Enhanced TRACE data. We fix ~99% of the truncation errors using this method. You can download the spreadsheet for the bonds with truncated returns and the associated corrected returns here.
To see the impact that measurement error/noise has on out-of-sample price-based anomalies, see this simple script, which downloads the WRDS measurement-error-adjusted bond panel and forms decile portfolios for credit spread, bond yield, bond return (short-term reversal) and bond market capitalization with and without the effects of measurement-error.
New Paper:
Return-Based Anomalies in Corporate Bonds: Are They There?
Quite simply, they are not. Corporate bond anomalies related to return reversals and momentum have been shown to generate large average returns to investors, even after adjusting for potential sources of risk. We revisit these findings and provide conclusive evidence that these anomalies, once properly constructed, generate average returns and alphas that are close to zero or negative. Code to construct market microstructure noise (MMN) adjusted TRACE bond returns can be accessed here.
Noise in TRACE bond prices is extremely important. After using the measurement error correction procedure in our paper, the reduction in the noisy short-term reversal (REV) premium is to the order of 90%. The measurement-error-adjusted factor (REV*) earns a premium of close to 0%, compared to the noisy REV factor which earns > 0.70% per month. Noise affects all metrics related to bond price (i.e., credit spread, yields and so on), implying the majority of papers which use noisy TRACE price-based metrics are prone to bias.
Priced risk in corporate bonds
This website provides the correctly constructed Bai, Bali & Wen (2019, BBW) four factors, and reproducible code and data used in “Priced risk in corporate bonds“, by Dickerson, Mueller & Robotti (2023). If you use the factors, code or data, please cite our paper:
title={Priced risk in corporate bonds},
author={Dickerson, Alexander and Mueller, Philippe and Robotti, Cesare},
journal={Journal of Financial Economics, Forthcoming},
year={2023}
A key finding in our paper, is that after correcting serious and consequential errors in the original authors’ factors, the bias adjusted sample squared Sharpe ratio of the bond market portfolio (MKTB) is, in fact, greater than holding all four of the bond factors. Economically, and from an investment perspective, it is optimal to simply hold the market portfolio.
The original BBW 4-factors with the lead and lag errors and ex post winsorization can be downloaded here.