Holy schnikes!! So much meaningless debate over an intensely subjective decision. What's next? A debate as to whether chocolate or vanilla milkshakes are objectively superior (btw, the answer is vanilla)?
If you've got a clear path to 20, legacy is clearly superior--regardless of how long you've been in as the match is just not high enough to compound enough to equal the lifetime earnings of the legacy pension. But, as I tell the people junior to me who seek my opinion, only ~55% of officers make it to 20. There is no clear path to 20 for anybody with less than 10 years in service. Medical conditions incompatible with service happen. RIFs happen. Looming administrative tours that will harm your ability to practice competently happen. Familial circumstances that require you to ETS happen. And to guard against all these possibilities you only have to give up 20% of the pension? That's a great trade for any junior milmed doctor. Plus, as I tell my residents, the BRS guards against the "sunk cost fallacy" that makes an unhappy doc at 12-16 years stay in to 20 when his life would be significantly better if he would just ETS. And for those who opt in and end up making it to 20, it won't even matter. If you're running numbers and debating on this thread about CAGRs, marginal tax rates, and modern portfolio theory you're going to be a mutimillionaire anyway and the small amount of lifetime earnings you miss out on by opting into the BRS won't matter.
That said, a couple statements on this thread require comment.
The people who lost a lot were invested in individual stocks, not in diversified funds.
What? The S&P 500 (which the C fund is indexed to) declined from a value of 1520.77 on 1 Sep 2000 to 827.37 on 27 Sep 2002 (don't even get me started on 2008--my first TSP contribution was Sep 2007--my intern year). That's a 46% decline in a diversified, index fund with low beta (for an equity fund). If you don't consider that a significant loss and understand the sequence of return risk for somebody in the drawdown phase during a bear market like that, I don't know what to tell you.
From 1928 to 2010, the average yearly return in the S&P500 was 11.31% and even from 2001 to 2010(where we sustained both a terrorist attack and a major market crash) the average was 3.54%. The C fund(which mirrors S&P500), since inception in 1988, has performed at 10.16% and that includes the dismal 2001-2010 timeframe.
This a rather academic topic, but significant changes have been made to the methodology of the S&P 500 index in the last decade of which most investors are unaware.
1. Foreign companies that maintained US ADRs (like RDS and UL) were booted from the S&P. Some of these companies (like RDS and UL) are a significant part of the historical gains of the S&P.
Shell and Unilever kicked out of S&P 500 index
2. The S&P changed to a float-adjusted weighting from market cap weighting. Essentially, this means that the S&P holds smaller positions in high growth components with significant insider holdings because these companies have a smaller float. These growth stocks (AMZN, GOOGL, FB) are responsible for a significant component of the S&P's returns; any methodology that decreases their weight within the index is a bad thing.
https://www.virtus.com/assets/files/sa/wm_did_you_know_6623 (1).pdf
BLUF: The S&P's methodology is not the same as it was for the past 100 years. It has changed (IMO) for the worse with future returns unlikely to match historical returns. Relying on historical S&P rates of return to calculate future rates of return will likely result in dangerous overestimation.
As for me, I'm a gambler, but a cautious one. The kind of guy who bets $5 on the pass line but also puts $2 on C&E during a come-out roll at the craps table. I opted into the BRS and had my wife stay legacy.