How much worse can co-investment deals be and still beat the fund?

Adverse selection and the fee saving are both real. The limit where one cancels the other is a number you can compute from the fund's own terms.

Julian R. Sterling · from The Co-Investment Practitioner

A no-fee, no-carry co-investment beats the fund as long as its gross multiple stays above the fund's net multiple. With a fund grossing 2.00× and netting 1.64× after fees and carry, co-investments can be 0.36× worse and still tie, which means at most 24 per cent of them can be 0.50× deals nobody else wanted. The cushion shrinks when the fund does badly and nearly halves on reduced-fee terms.

Worked in full in The Co-Investment Practitioner by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →

The standard worry about co-investment is adverse selection: the manager offers limited partners the deals it would rather not hold in full. The standard reply is that the fee saving compensates. Both are true up to a limit, and the limit is a number you can compute from the fund's own terms before the first deal is accepted.

The assumptions

Illustrative terms. The model charges carry on all fund profit with no hurdle, as a full catch-up does once the hurdle has cleared.
Input	Fund route	Co-investment
LP capital paid in	100	100
Management fees and fund expenses over life	10	0
Capital invested in deals	90	100
Carried interest	20%	0%
Gross multiple on invested capital	2.00×	to be solved

The calculation, step by step

Cushion = fund gross multiple − fund net multiple

Proceeds: 90 × 2.00 = 180
Profit to the fund: 180 − 100 = 80
Carry: 20% × 80 = 16
LP receives 180 − 16 = 164, a net multiple of 1.64×
A fee-free co-investment returns its gross multiple, so it ties at 1.64×. Cushion: 2.00 − 1.64 = 0.36×, or 18 per cent of the fund's gross.

Mix of good and bad deals: if a share s of co-investments return 0.50× and the rest match the fund's 2.00×, the programme ties when (1 − s) × 2.00 + s × 0.50 = 1.64, so s = 0.36 ÷ 1.50 = 24%.

In Excel: =(B2-B3)/(B2-B4) with gross, net and the bad-deal multiple in B2:B4.

Put plainly: in a programme of ten deals, two duds at 0.50× and eight ordinary deals return 1.70×, still ahead of the fund's 1.64×. Three duds return 1.55×, and adverse selection has consumed the whole fee saving and more.

The same test works in reverse once the programme is running. Track the realised co-investment multiple against the fund's net multiple for the same vintage, not against its gross, and count the deals that came in well below the fund's average. When the share of weak deals approaches the tolerable share, the programme is at break-even whatever the headline multiple says, and the conversation with the manager about which deals are being offered should start then, not after the final distribution.

What if the fund does worse, or the terms are not free?

The cushion is not a fixed number. It is the fee and carry drag, and the drag depends on how well the fund performs.

Carry is charged on all profit with no hurdle, a simplification that slightly overstates carry, and so the cushion, at low multiples. At 1.00× there is no profit and the cushion is the fee drag alone.
Fund gross multiple	Fund net multiple	Cushion	As % of gross	Max share of 0.50× deals
1.00×	0.90×	0.10×	10%	20%
1.25×	1.10×	0.15×	12%	20%
1.50×	1.28×	0.22×	15%	22%
2.00×	1.64×	0.36×	18%	24%
2.50×	2.00×	0.50×	20%	25%
3.00×	2.36×	0.64×	21%	26%

The cushion is thinnest exactly when it is needed. In a vintage where the fund grosses 1.00×, carry disappears and co-investments have only 0.10× of room. Adverse selection hurts most in poor vintages, because the deals a manager is keenest to syndicate are more likely to be the weak ones, and that is when the fee saving has shrunk.

Terms matter as much as performance. Many co-investments are no longer free. On reduced terms, 5 of fees per 100 and 10 per cent carry, a co-investment at 2.00× gross nets 1.81× rather than 2.00×. To tie the fund it must gross 1.80×, so the cushion falls from 0.36× to 0.20×, and the share of 0.50× deals it can absorb from 24 to 13 per cent. Deal-level frictions narrow it further: 0.03× of transaction fees and shared broken-deal costs takes the free cushion to 0.33× and the tolerable dud share to 22 per cent.

The common mistake

The usual mistake is to compare the co-investment programme's gross multiple with the fund's gross multiple and call the gap "alpha" or "adverse selection". The right comparison is co-investment net against fund net, because the fund route was never going to deliver its gross. A co-investment book returning 1.80× against a fund grossing 2.00× looks like 0.20× of adverse selection. It is in fact 0.16× ahead of what the same money would have earned through the fund.

The second mistake is quoting the cushion as a fixed percentage of the deal, typically from a single good vintage, and applying it to a programme that will run through bad ones. Size tolerance from the bottom of the table, not the middle.

Takeaway

Adverse-selection limit = fund gross − fund net, on the co-investment's own terms.
On illustrative 2.00× fund economics, free co-investments can be 0.36× worse; at most 24 per cent of them can be 0.50× duds.
The limit falls with fund performance and with any fee or carry on the co-investment.
Measure the programme against fund net, deal by deal, from the first year.

The free workbook for this book, at the companion page, works this limit for the book's own deal. For what the fee saving amounts to across a whole programme, see what a co-investment programme actually saves, or model your own terms with the co-investment fee savings model.

Questions readers ask

What is adverse selection in co-investment?

It is the risk that a manager offers limited partners the deals it is least keen to hold in full, so co-investments perform worse than the fund's average. It is tolerable up to the fee and carry saving: with an illustrative fund grossing 2.00x and netting 1.64x, co-investments can underperform by 0.36x before the programme falls behind the fund.

Should co-investment returns be compared with fund gross or net returns?

With fund net returns, measured on the co-investment's own net terms. The fund route never delivers its gross multiple. A co-investment book at 1.80x against a fund grossing 2.00x looks 0.20x behind, but it is 0.16x ahead of the 1.64x the same money would have earned through the fund.

How do reduced fees on a co-investment change the break-even?

They shrink the cushion sharply. With 5 of fees per 100 and 10 per cent carry instead of none, a co-investment must gross 1.80x to tie a fund netting 1.64x, so the cushion falls from 0.36x to 0.20x and the share of 0.50x deals the programme can absorb falls from 24 to 13 per cent.

Read the whole case

The limit is raised in chapter 3 of The Co-Investment Practitioner, and the companion files compute it for the book's own deal. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.

Get the book on Amazon →Free companion files

Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada

Also on this site

What a co-investment programme actually saves, and what erases it37 basis points on a 15 per cent programme — and a 22.8 per cent total-loss rate wipes it out. At the usual per-deal cap a single failure is 37 per cent of the annual cohort.
What a follow-on reserve is actually worthA 15 per cent reserve is worth 0.019 turns under plain dilution and 0.285 under pay-to-play. The optimum is a property of the document, not the portfolio.
Why is net IRR lower than gross IRR in private equity?A buyout returning 17.05 per cent gross leaves the investor 12.70 per cent net: 1.90 points go to the management fee and 2.44 points to carry.
How many co-investments' fee savings does one write-off erase?At a 0.36x saving per deal, one standard-size zero erases 2.8 deals' savings on capital and 5.6 on expected value. Cheque size decides the rest.
How do you convert MOIC to IRR?IRR = MOIC^(1/years) minus 1 for a single exit: 2.0x is 14.9 per cent over five years and 26.0 over three. Interim distributions break the formula.
How much volume can a price increase lose before EBITDA falls?Break-even volume loss is price rise over contribution margin plus price rise: 9.09 per cent for a 4 per cent increase at a 40 per cent margin.

Reading guide: private equity and private markets → · All 368 articles →

If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.