The J-curve, and why vintage matters more than fund picking

For the first three years, a fund commitment looks like a loss: the fund has called capital, charged fees and holds young investments at cost, so the reported value sits below what you paid in. This dip is the J-curve, and it says nothing about whether the fund is good. What a private investor can influence in that time is not which fund turns out best. It is whether all the money went into one year. We think spreading commitments across vintages does more for a private LP than fund picking, and that the two are often confused.

What do the first three years of a commitment look like?

The J-curve is the path of a fund's value against the capital paid in: down first, then up as companies are written up and sold. Take a commitment of EUR 500,000 to a venture or buyout fund with a management fee of 2 per cent a year on the commitment. The figures are invented and unremarkable.

In year one the fund calls 20 per cent. EUR 10,000 is the management fee, 2,000 are set-up costs, and the rest is invested and carried at cost. TVPI, the value of what you hold plus what you got back, divided by what you paid in, is 0.88. In year two the fee is charged again on the full commitment although less than half is invested, and one company is written down by 10,000. Good companies are not yet written up, because nothing has happened that would justify a new valuation. Year three brings the first write-up after a follow-on financing round, and TVPI climbs to 0.96. DPI, the cash returned, is zero throughout.

Three things follow. The dip is mostly arithmetic: fees are charged on the whole commitment from the first day, losses show early and gains late. The IRR in these years is negative and jumps with every valuation, so it is not worth reading. And none of this distinguishes a fund that will return three times the money from one that will return 0.9 times.

Cambridge Associates examined more than 2,100 funds raised between 1995 and 2005 and found that a fund needs about six years to settle into its final quartile against peers, and that 80 to 90 per cent of funds pass through at least three different quartiles during their lives (A Framework for Benchmarking Private Investments, 2014). In the early years, the things worth checking are operational: whether calls arrive at the pace the manager announced, what share of the called capital went into fees, and whether the portfolio is being built as described. Our piece on reading a quarterly report covers where to find them.

Why does the vintage matter so much?

The vintage is the year a fund starts investing. Funds of one vintage buy during the same three to five years, at the valuations of those years, and they must sell in the same window some years later. A fund that deployed its capital when prices were high and reached its exit years when buyers were scarce carries that through its whole life, however able the manager. That is why funds are benchmarked against others of the same vintage and not against all funds.

For the investor, this means the commitment year is a risk in its own right, comparable to buying all your shares on a single day. Nobody can know in advance whether a year will be a good one. The distribution drought we described last June hit funds across managers because their exit years coincided.

Why not simply pick the best funds?

Because the information needed arrives too late, and the best-known funds are hard to get into.

Managers raise a new fund every three to four years. When fund IV is offered, fund III is three years old and sits at the bottom of its J-curve, and by the Cambridge figure its final quartile will not be clear for another three. The investor decides on the basis of fund II and a story. Research on persistence makes this concrete. Harris, Jenkinson, Kaplan and Stucke found in a study of US funds that for buyout funds raised after 2000 there is little or no evidence that a manager's past performance, as known at the time of fundraising, predicts the next fund. For venture funds persistence did hold (NBER working paper 28109, 2020). But the venture managers with a record are not short of investors, and a private LP with a ticket of a few hundred thousand euros seldom gets an allocation.

None of this makes diligence pointless. A fund's terms, team and strategy deserve the work, and avoiding a bad fund is worth a great deal. But selection is a judgement made under thin information, and its result is known after a decade. The vintage decision is made under no information at all, which is exactly why it can be handled mechanically.

What does spreading across vintages change?

Compare two investors with EUR 500,000 for private funds. The first commits the whole amount to one fund this year. The second commits EUR 125,000 a year for four years.

The second investor does not escape the J-curve. Each new fund starts its own dip, so the programme as a whole stays below 1.0 for longer than a single fund would. That is the price, and it is worth saying because staggered commitments are sometimes sold as a cure for the J-curve. They are not.

What the second investor gets is four entry periods and four exit windows where the first has one of each. A poor year weighs a quarter. And from about year five or six, distributions from the oldest fund begin to meet the calls of the youngest, so the programme needs less fresh cash than the sum of its commitments suggests. How to plan the cash before that point is the subject of our piece on liquidity for capital calls.

The temptation runs the other way. Commitments are easy to make in years when earlier funds are distributing, markets are up and every manager is fundraising, and they feel reckless after a bad year when little comes back. An investor who follows that feeling ends up with most of the money in the vintages that bought at high prices. A fixed annual amount, set once and kept in bad years too, prevents that.

The trade-off is ticket size. A quarter of the amount per fund may fall below a fund's minimum, and four funds produce four sets of notices and reports. For smaller amounts, a fund of funds or a feeder that itself commits over several years does the staggering inside one vehicle, at the cost of a second layer of fees. That is a fair price for some investors and too much for others.

How do you see the vintages you already hold?

Group your funds by vintage year and add up the commitments per year. A portfolio built deal by deal tends to show two or three crowded years and gaps in between. In Valued the vintage is kept on each fund, and the J-curve and vintage views are drawn from the booked calls and distributions, but a spreadsheet column does the same job. The gaps are where the next commitment belongs, whatever the market feels like that year.