Graveyard / Fund discounts
Business development companies below NAV
Do the cheapest BDCs relative to their book value outperform?
The idea
BDCs report their net asset value each quarter. When one trades far below it, buy that one and short the basket of BDCs, expecting the discount to close.
How we tested it
| Data | About 45 surviving BDCs. NAV per share from SEC filings, dated at the filing date to avoid look-ahead. Prices from Yahoo. Monthly rebalance into the cheapest quintile by price-to-NAV against an equal-weight BDC basket. |
|---|---|
| Period | December 2022 to October 2026 (47 months). Free data gave NAV for 2022 onward only, so there was no development window. |
| Costs | Charged, with a stress case. Turnover is 24% a month, so cost is not what kills it. |
| Holdout | No development window existed, one month short of the 48 months the spec required. |
| Control | A random-selection twin, which landed at the 53rd percentile. |
Result
| Gross per year | +0.06% |
|---|---|
| Net per year | −1.37% (t −0.41) |
| Net at stress costs | −5.9% a year |
| Hit rate | 47% |
| Average P/NAV, cheapest quintile vs basket | 0.65 vs 0.94 |
Why it failed
Gross return is zero, so costs are not the issue. About 18 dead or merged BDCs could not be priced for free. They were typically cheap, so the cheap quintile is flattered by survivorship and the true result is probably worse.
What would make it worth retesting
NAV per share for 2010-2021 from the SEC's bulk financial statement data sets, then run the frozen rule unchanged on a real 2014-2019 development window and a 2020-onward holdout.
Test id: bdcnav-bdc-v1. Past results of a backtest say nothing certain about the future.