Andrew Ang: Should AI Agents Run Your Asset Allocation? | #653

Andrew Ang · Tau Balance

Federal taxes cut long-run equity returns by about 160-170bps a year, compounding to roughly a third of a portfolio.

Andrew's paper 'Uncle Sam's Cut' estimates the federal tax drag on US equity returns. Over the last thirty years it is about 160-170bps a year. He says that small number compounds to about a third of a taxable investor's portfolio disappearing over thirty years. Dividends are the largest part of the drag, since they are taxed every year and interrupt compounding. Over the full century from 1926 the average drag is about 350bps, and the recent period is at the low end. Taxes on dividends fell to the preferential rate and dividend yields dropped from 4-5% to 1-2%, so the modern era is the anomaly. State and local taxes sit on top, with New York City's top marginal rate at 55.576%.

“It sounds small, but you compound that over thirty years, and that results in 36% of your portfolio that just disappears.”at 0:00

Andrew Ang · Tau Balance

Taxes are going up in some form, in Ang's personal view.

Ang argues that taxes are likely to rise in some form. He first sets out the academic debate on optimal tax rates, noting that estimates range from zero to well above 50% for income, dividend and capital gains rates depending on model assumptions. He notes that Berkeley economists favour high optimal capital and dividend tax rates, and then says that, personally, he thinks taxes are going up in some form. The view rests on his reading of the optimal-tax literature rather than on a specific policy proposal, so it is a personal judgement.

“It all depends on your assumptions. But I think you can definitely say... Well, personally, I think taxes are going up in some form.”at 42:15

Andrew Ang · Tau Balance

Equity factors vary over time and are predictable, so investors should tilt factors using several forecasting models.

Andrew argues that factors such as value change over time, and that the question is how to forecast them, not whether they can be forecast. The 2018-2020 value drawdown was the worst in over a century, and he treats that as proof that factor returns vary. His approach is to use three models: a market similarity model that weights past episodes similar to today, factor momentum, and factor dispersion (the spread in valuations or returns between expensive and cheap stocks, or between winners and losers). He combines these with time and model diversification, and keeps the active tilts small.

“yes, the factors change. Yes, they're predictable... we try the best that we can to forecast it.”at 23:47

Andrew Ang · Tau Balance

Tax uncertainty is priced in stocks; higher tax uncertainty means investors demand compensation.

Asked whether rising taxes would lower stock valuations, Ang says the literature shows taxes are a factor in asset prices. The effect is not simply that taxes subtract cash from investors. Instead, when periods of higher tax uncertainty arise, investors demand compensation for bearing that risk, which pushes valuations down. He adds that estimates of how taxes are priced range from not priced at all to reflecting the marginal rate of an ordinary taxable investor. The marginal investor could be a tax-exempt sovereign wealth fund or a wealthy taxable individual, so the size of the effect varies across studies.

“The literature has shown that taxes are a factor... when there are periods of higher tax uncertainty, right, then people demand to be compensated for that.”at 40:41

Andrew Ang · Tau Balance

AI agents debating allocation calls outperform human committees, since agents tolerate disagreement and it surfaces opportunities.

Andrew's firm is building a swarm of specialised AI agents. Each agent runs an asset class or a portfolio construction method, such as hierarchical risk parity or market-cap weighting. They produce expected returns and write memos, then critique and vote on each other's proposals. He argues human teams struggle to harness disagreement because of politics, HR issues or talent departures, whereas agents 'don't care' about internal conflict. He says the best opportunities are where there is the most disagreement. Agents also improve quickly by finding new skills. He sees humans shifting to oversight of the workflow and exceptions. He says the test is whether it beats human approaches, and he doesn't know yet, and that live performance will decide it. He also expects consolidation at the top and middle of investment teams.

“but you can really do it easily with agents because they're agents. They kind of don't care... I think the best opportunities are the ones where there is a lot of disagreement.”at 6:10

Andrew Ang · Tau Balance

Private credit's pretax yields overstate what investors keep, since the income is taxed as ordinary income.

Private credit yields around 10-12% a year, close to the long-run equity return of about 10%. Andrew argues this comparison is misleading because all of the private credit income is taxed as ordinary income, including payment-in-kind interest. Investors should compare after-tax returns. A 10% private equity return is about 7% in equity terms after tax, and private credit would need roughly a 14-15% yield to match a 10% after-tax equity return. He adds that yield products that engineer income through options or by inducing capital gains are generally taxed as income and are often not very good, so selling appreciated stock is usually better.

“A lot of people are attracted to it because it yields, you know, like, 10, 12% per year... But it's all taxed as income... You can't look at this pretax number. You gotta look at what's the return that you get to keep after taxes.”at 31:15

Andrew Ang · Tau Balance

With debt near WWII levels and interest costs above defense spending, tax increases are likely.

Andrew says US debt to GDP is at the same level as during World War Two, above 100%. About 15% of the federal budget now goes to interest payments on US debt, which is more than defense spending. In that setting he finds it hard to see taxes not going up. He says he would not take the other side of a bet that taxes rise, and that the recent 30 years of low tax rates are the exception rather than the norm.

“debt to GDP ratio is at the same level as it was during World War two, you know, above a 100%, well above a 100%.”at 33:51

Andrew Ang · Tau Balance

The economy is not returning to a 2% inflation world, which challenges bond-heavy portfolios.

Ang argues that the economy is not going back to a 2% inflation regime. He points out that the strategies that performed well over the past thirty years, bonds in particular, benefited from that low-inflation backdrop. He then asks what portfolio should be held once inflation is no longer 2%, calling it a big open question. He thinks fixed income will need extra levers, such as tax-loss harvesting for bonds and structuring bonds to get better payoffs and more favourable tax treatment. The view combines a structural inflation call with the implication that bond-heavy allocations need rethinking.

“We're kind of not going back, I think, to a 2% inflation world... What's the right portfolio to hold when inflation's not going to be 2% anymore?”at 42:42