Key Takeaways

  • Under former CIO Peter Holland, Columbia Investment Management Company evaluated allocations through raw factor underwriting: assessing equity beta, liquidity costs, and debt levels rather than checking geographical or sector boxes.
  • When Kim Lew took the CIO helm, she reorganized Columbia around asset class specialization across private and public markets, sparking internal concerns about siloing.
  • Specialization introduced active portfolio construction, pushing allocators to plan explicit portfolio tilts and run targeted research sprints rather than merely vetting inbound pitches.
  • Columbia retained a cross-asset debate culture by requiring the entire investment team to review and debate every fund commitment regardless of asset class.
  • Managing Director Erlina Yeo Terwelp views the industry debate between generalist and specialist endowment models as a false dichotomy that obscures how institutional capital actually operates.

The Fallacy of Asset Class Buckets

Most endowments talk about asset allocation as if it were a filing cabinet. You decide on twenty percent venture capital, fifteen percent buyout, ten percent real estate, and then you send associates out to fill each drawer.

Under former CIO Peter Holland, Columbia Investment Management Company approached capital allocation through a different lens. The team ran a model similar to today's total portfolio approach, stripping away labels to price the underlying risk profile of each prospective commitment.

As Erlina Yeo Terwelp explained: “Bedrock of what we did as a generalist was that every investment opportunity was evaluated in its cleanest forms, which was, what's its equity beta, what's the liquidity, and what's the cost of that, and what's its leverage. Not necessarily, do we want an India VC manager or do we want a X-type of manager? It was very little bucket filling. It was a bottoms-up, first principles of what an investment was.”

Evaluating investments on pure factors protects an endowment from double-counting risk across private markets and public equities. An enterprise software buyout fund and a growth equity fund may sit in different asset classes on an LP committee report. Yet on an equity beta and borrowing basis, they often carry identical exposures. Bottom-up factor analysis stripped away GP branding and made managers compete for the same dollar.

Why Specialization Wins on Portfolio Construction

When Kim Lew took over as Columbia's CIO in 2020, she introduced formal asset class specialization across private and public markets. The shift initially rattled allocators who prized the generalist tradition.

“When Kim came in and made us specialists, I was nervous about the whole transition,” Terwelp noted. “I have learned so much from being a asset class specialist to some extent. We still keep a lot of generalist heritage. We discuss every fund together.”

The unexpected upside was disciplined portfolio construction. Pure generalists often drift into reactive underwriting, spending their days filtering incoming ideas rather than deciding what the portfolio requires.

“The benefit to being a specialist, for me a little bit later in life, was that there was this element of portfolio construction that I had not appreciated,” Terwelp said. “Planning, thinking through your tilts, about researching different areas that you needed to inject into a portfolio, that was something that I had not had a ton of interaction with when I was a generalist.”

Specialization forced the private markets team to set deliberate underwriting themes, identify structural blind spots, and proactively source managers to fill them. Yet by forcing every deal back into a room where public and private allocators debate the merits together, Columbia preserved the friction that keeps specialists from falling in love with their own asset class.

As Terwelp observed: “When you said it was changing from a generalist team to a specialist team, that wasn't how I thought about it. It wasn't how it played out. It's a little bit of false dichotomy in some ways.”

Why It Matters

Institutional capital allocators face higher baseline borrowing costs and compressed private equity distributions, making naive bucket allocation untenable. When LPs combine factor underwriting with asset class specialization, GPs can no longer win commitments purely on past quartile performance within a narrow niche. Private equity sponsors now have to justify their returns against public market equivalents on raw risk factors, while proving they fit an LP's intentional portfolio tilts.