How to Choose an ETF — Expense Ratio, Tracking Difference, Liquidity

Expense ratio — the first filter, not the only filter

The expense ratio is the annual management fee charged by the ETF, expressed as a percentage of your invested assets. It is deducted automatically from the fund's net asset value on a daily accrual basis — you never see a bill, and it never shows up as a separate line item in your account. What you do see is slightly lower returns every year compared to the raw index performance.

For broad index ETFs tracking the same benchmark, the expense ratio is the largest single predictor of relative returns. This is not subtle: SPIVA data consistently shows that the single best predictor of whether a fund will outperform its benchmark over 10+ years is its expense ratio. The lower the fee, the higher the probability of outperformance, because the manager simply needs to replicate the index — and lower costs give them more of the index return to pass on to investors.

The compounding of fee drag is what makes this so consequential over long investment horizons. A 0.75% annual drag versus 0.03% feels like a trivial 0.72% difference. On $100,000 invested at 8% gross returns over 30 years: the 0.03% fund grows to approximately $1,006,000. The 0.75% fund grows to approximately $761,000. The $245,000 gap — equivalent to 2.45× the original investment — is the compound cost of choosing the wrong ETF. Warren Buffett noted that an investor who stays in a 1% annual fee fund over a 30-year career would have gifted nearly one-third of their terminal wealth to the fund manager.

Tracking difference — the honest measure of total cost

Expense ratio tells you what the fund charges in management fees. Tracking difference tells you what the fund actually delivers after all costs and income combined. It is calculated as: Index annual return minus ETF annual return. A tracking difference of +0.05% means the ETF underperformed its index by 0.05% after everything. A tracking difference of −0.05% means the ETF outperformedits own index by 0.05%.

Negative tracking differences are real and surprisingly common among Vanguard ETFs. They arise from securities lending: the ETF lends its holdings to short sellers, who pay a borrowing fee. Vanguard returns this income directly to the fund, which can offset part or all of the expense ratio. In some years, lending income exceeds the expense ratio entirely, producing a fund that outperforms its own benchmark index net of costs. This is legally permitted, well-documented, and genuinely beneficial to investors — but only discoverable by looking at tracking difference rather than expense ratio alone.

Tracking error, by contrast, measures the consistency of daily tracking — the standard deviation of daily differences between the ETF and the index. A fund with low tracking error tracks its index smoothly day to day. A fund with high tracking error shows erratic daily deviations even if its annual tracking difference is acceptable. Both metrics matter: use tracking difference to assess annual cost, tracking error to assess daily execution quality.

Replication method and its effect on tracking quality

Physical full replication means the ETF holds every single constituent of its index in exact proportion. This is ideal for liquid indices like the S&P 500 — every stock can be purchased and held without meaningful market impact. Optimised sampling holds a statistically representative subset when the index contains thousands of illiquid small-cap constituents. VTI uses optimised sampling for its micro-cap tail and achieves tracking differences below 0.05% annually — demonstrating that sampling does not necessarily impair tracking quality when executed well.

Bid-ask spread — the hidden cost that depends on your trading frequency

Every time you buy an ETF, you pay the ask price — the price sellers demand. Every time you sell, you receive the bid price — what buyers are willing to pay. The difference between these two prices is the bid-ask spread, and it is an implicit transaction cost paid on every single trade. Unlike the expense ratio, which accrues continuously regardless of trading, the spread is only paid when you actually transact.

For highly liquid ETFs like SPY — the most traded security in the world by dollar volume — the spread is approximately $0.01 on a $500 share price, or 0.002%. This is essentially free. An investor adding $1,000 per month to SPY pays roughly $0.10 per contribution in spread costs over a lifetime of investing. For a niche thematic ETF with $50M AUM and 20,000 daily shares traded, the spread might be $0.50 on a $25 share price — a 2% round-trip cost on every contribution. An investor making twelve contributions per year would pay $2,400 in spread costs annually on a $10,000 annual investment — more than offsetting any expense ratio advantage.

The spread is determined by market maker competition. When dozens of institutional market makers compete to provide liquidity in a popular ETF, they drive spreads to near-zero. In a thinly traded ETF, fewer market makers compete, they hold larger inventory risk per trade, and they demand compensation through a wider spread. This is not manipulation — it is the normal economics of market-making. The solution is simple: prefer ETFs where the spread is consistently below 0.10%.

AUM — why size matters and when it doesn't

Assets Under Management (AUM) is not a quality metric in itself — a good ETF with $100M AUM can track its index as well as one with $100B AUM. But AUM matters for two practical reasons: liquidity and closure risk.

Larger AUM attracts more market makers, which tightens spreads. SPY's $550B AUM and extraordinary daily trading volume make it the most liquid security on earth outside of US Treasuries. Small ETFs with under $50M AUM often have wide spreads because no market maker wants to warehouse significant inventory risk in an instrument few people trade. This creates a self-reinforcing cycle: tighter spreads attract more traders, which attracts more market makers, which tightens spreads further.

Closure risk is the more insidious concern. ETF providers close funds that are unprofitable — typically those with AUM so low that management fees do not cover operating costs. An ETF charging 0.15% on $40M generates only $60,000 per year in revenue. Index licensing, custody, compliance, and regulatory filings alone can cost more than this. When an ETF closes, shareholders are forced to sell and realise any embedded capital gains — creating unexpected tax events at potentially inopportune moments. The safe threshold for core holdings is $500M+ AUM. Below $100M, the closure risk becomes meaningful.

The most-asked ETF question: SPY, VOO, or VTI?

These three ETFs collectively manage over $2 trillion in assets and are the most commonly held by individual investors. All three are excellent. Choosing between them requires understanding not which is better in absolute terms, but which is better for your specific use case.

When to choose SPY

SPY is the correct choice for active traders, options traders, and institutional investors. Its daily volume of approximately 90 million shares creates a spread of $0.01 — orders of magnitude tighter than any competing S&P 500 ETF at normal market hours. More importantly, SPY supports the deepest, most liquid options market of any ETF or individual stock on earth. Every institutional investor that needs to hedge a large equity portfolio, express a directional view on the broad market, or use index options for income generation uses SPY options. If you trade options on indices or hedge your portfolio with puts, SPY is the only practical choice.

When to choose VOO

VOO is the correct choice for buy-and-hold investors who want S&P 500 exposure at the lowest possible cost. Its 0.03% expense ratio is one-third of SPY's 0.0945%. On $200,000 invested for 25 years at 8% returns, that 0.0645% annual difference compounds to approximately $22,000 of additional wealth. VOO's AUM is now approximately $1 trillion — larger than SPY — with tight enough spreads for any non-intraday investor. Vanguard's mutual structure returns all securities lending income to the fund, producing negative tracking difference in multiple years.

When to choose VTI

VTI is the correct choice for investors who want maximum US equity diversification, including exposure to small and mid-cap companies that the S&P 500 excludes. The historical performance difference between VTI and VOO is small — typically less than 0.2% per year in either direction — because the S&P 500's mega-cap constituents drive most of the total market's returns anyway. However, in periods where small and mid-cap companies outperform large caps — as they did from 2000 to 2007 — VTI provides the exposure to capture that outperformance. At the same 0.03% expense ratio and similar Vanguard securities lending structure, there is essentially no cost to choosing VTI over VOO for a long-term investor who wants maximum breadth.