Money, Examined / Publication No. 11
A good advisor can keep a bad moment from becoming a permanent financial decision.
Price the help against the investor you are likely to be, then compare it with the disciplined investor you could become.
The decision, in four lines
01 / Define the job before the price
It is a better decision process when money, fear, taxes, family, and uncertainty collide. Preventing one disastrous sale can matter more than years of small portfolio refinements. Whether that help is delivered, and what it costs, are separate questions.
There is good reason to take investor behavior seriously. Barber and Odean document poor net performance among the most active traders in a historical sample of individual brokerage accounts. Their result challenges the idea that more activity reliably improves outcomes. It does not measure how much an advisor would have added, or establish that every investor trades too much.[1]
The relevant comparison is not always a frictionless, unflappable investor from a textbook. If your realistic alternative is to abandon a sound plan during a drawdown, behavioral support can be economically valuable. But it is also not enough to compare advice only with your worst possible self. Written rules, appropriate risk exposure, automatic contributions, and limited decision frequency are legitimate competing ways to make disciplined DIY more achievable.
A credible promise
Define the allocation before distress. Explain the losses it can plausibly experience. Keep near-term spending separate from long-term risk. Agree on what warrants a change, and who calls whom when markets fall.
A promise that needs evidence
A recurring fee is observable. A prevented decision is counterfactual. Ask for a specific mechanism, a realistic comparison, and limits. The existence of investor mistakes is not evidence that this particular service prevents them at a price worth paying.
Advice may be worth its price. The price still carries a burden of proof.
That burden is a standard for the service, not a judgment about a profession or an individual. An excellent advisor can lose this arithmetic contest against already-disciplined DIY and still provide planning or administrative value a household chooses to buy.
02 / Make the counterfactual visible
Buying after enthusiasm and selling after fear can damage wealth. But not every sale is a mistake: a changed spending need, unsuitable risk exposure, or new constraint can warrant one. This lab isolates a narrow question: what happens when an investor exits a fixed, hypothetical market path and returns on a precommitted schedule?
Start with an advisor who prevents the exit. Then remove the behavior gap, raise the fee, let the advisor fail, or select a continued decline in which cash wins. All are part of the comparison, not inconvenient exceptions.
Lab 01 / A path, not an annual alpha
Same starting dollars. Same contributions. Same market. After a scheduled drawdown, one investor moves money to cash. Does the advisor keep them invested, or just charge for the same mistake?
Advice beats the modeled mistake after its fees. Any behavioral benefit belongs to the specified episode, not every future year.
| Behavioral difference before advice fees | $223,011 |
|---|---|
| Less: terminal fee burden, including lost growth | $163,005 |
| Net advice difference | $60,006 |
| Of the fee burden: actual fee dollars deducted | $97,148 |
| Remaining difference from growth on deducted fees | $65,858 |
| Year | Disciplined DIY | Mistimed DIY | Advice, discipline kept |
|---|---|---|---|
| 0 | $250,000 | $250,000 | $250,000 |
| 1 | $271,193 | $271,193 | $268,522 |
| 2 | $204,565 | $204,565 | $200,595 |
| 3 | $301,685 | $218,866 | $293,045 |
| 4 | $325,979 | $238,192 | $313,695 |
| 5 | $351,731 | $258,676 | $335,367 |
| 6 | $379,028 | $280,390 | $358,110 |
| 7 | $407,963 | $303,407 | $381,977 |
| 8 | $438,634 | $327,805 | $407,025 |
| 9 | $471,146 | $353,666 | $433,311 |
| 10 | $505,608 | $381,079 | $460,897 |
| 11 | $542,137 | $410,137 | $489,847 |
| 12 | $580,859 | $440,939 | $520,229 |
| 13 | $621,904 | $473,588 | $552,113 |
| 14 | $665,411 | $508,197 | $585,574 |
| 15 | $711,529 | $544,882 | $620,689 |
| 16 | $760,414 | $583,768 | $657,540 |
| 17 | $812,232 | $624,988 | $696,214 |
| 18 | $867,159 | $668,680 | $736,801 |
| 19 | $925,382 | $714,994 | $779,394 |
| 20 | $987,098 | $764,087 | $824,093 |
This is not a historical market record. The shock month returns minus the selected drawdown. The next six months compound to the selected rebound, replacing the normal trend. All other months use the annual trend converted to a monthly rate. A 30% loss followed by a 40% gain still leaves the original dollar at 98 cents, before other returns.
The sale happens after the shock-month close. The selected share targets cash for the next N monthly returns; the rest stays invested. Re-entry is at the close of the Nth month, fixed in advance, not chosen by a future recovery. Partial exposures rebalance monthly. Contributions enter at each month's start and follow that month's target weights. If the horizon cuts an episode short, the model simply ends there.
All invested portfolios use the same exposure: one synthetic risky total-return asset. Cash earns the entered yield, not an unstated zero. Returns include reinvested distributions by assumption; taxes, inflation, trading spreads, product expenses, withdrawals, and advisor trading are excluded. This is not an allocation recommendation.
Fees are additional advice costs: returns first, then annual AUM percentage divided by 12, then annual flat dollars divided by 12. The flat charge is capped at the remaining balance and unpaid bills are disclosed. Advice keeps charging in cash and outside the episode. No planning benefit or tax alpha is inserted. The prevention checkbox is a deterministic counterfactual, not an empirically estimated success rate.
The rebound is known to the simulator, not to the investor's re-entry rule. It is still an assumed path chosen by the reader, so the result cannot tell us how likely an advisor is to help. A visually persuasive curve is an accounting example, not an estimate of a treatment effect.
03 / Separate the estimands
The question is not whether mistakes exist. It is whether a particular relationship changes decisions enough to justify its incremental cost.
In the Mullainathan, Noeth, and Schoar audit study, trained auditors presented different investor portfolios to advice providers. Advice often reinforced return chasing and favored higher-fee active funds rather than correcting the bias. The controlled client presentations provide evidence about the advice offered in that setting; they are not a randomized trial of decades of client wealth.[2]
Foerster, Linnainmaa, Melzer, and Previtero find that Canadian advisors have substantial influence on client portfolios but relatively limited customization. Large advisor effects and the costs of the observed portfolios complicate any claim that advice uniformly improves outcomes. This is evidence from a particular distribution system, not a census of contemporary independent planners or hourly fiduciaries.[3]
In The Misguided Beliefs of Financial Advisors, Linnainmaa, Melzer, and Previtero find that advisors' own investing often resembles the expensive, active behavior seen in their clients' accounts. The implication is not simply that incentives are irrelevant or that advisors are insincere. Genuine conviction can coexist with poor investment practice. Removing a commission does not, by itself, establish competence. This paper and the Foerster study draw on overlapping Canadian data, so they are not independent replications.[4]
Selection is a central identification problem. People who hire advisors can differ in wealth, time, sophistication, risk tolerance, and need. The accessible working-paper version of Hackethal, Haliassos, and Jappelli's study illustrates how adjusting for selection changes performance comparisons. Its results do not support treating delegation as an automatic cure. Even with econometric adjustments, an observed advised-versus-unadvised difference is not a universal causal effect of hiring advice.[5]
Advice also has to be accepted and implemented. In judging a service, distinguish access to a recommendation from a process you will actually follow. Neither lab estimates that adherence rate: the reader must explicitly assume whether the advisor prevents the selected mistake.
| Evidence | What it can establish | What it cannot establish on its own |
|---|---|---|
| Account-level observations | Trading, costs, portfolios, and outcomes in a defined sample. | The return you personally gain by hiring a specific advisor. |
| Audit studies | Responses to controlled investor presentations in a specific setting. | The quality of every advisor, or a lifelong coaching effect in a different setting. |
| Industry value frameworks | A useful inventory of potential services and channels. | An independent causal estimate or a guaranteed annual return premium. |
| The labs on this page | The consequences of stated assumptions, with explicit fee accounting. | The probability that a mistake occurs or that advice prevents it. |
Vanguard's Advisor's Alpha framework treats behavioral coaching, planning, and implementation as potential sources of value. It is an industry framework from an asset manager serving advisors. Its widely advertised “3%” framing must not be imported as a guaranteed annual return or added mechanically to a market forecast. The value can depend on the client, the starting portfolio, the services delivered, and infrequent decisions.[6]
Morningstar's Mind the Gap 2025 compares fund returns with investors' dollar-weighted experience. Dollar-weighted returns depend on the size and timing of cash flows; time-weighted returns describe the investment's performance independently of those flows. A difference can reflect performance chasing, but also contributions, withdrawals, the return sequence, and measurement choices. It is neither wholly a panic-selling penalty nor the causal benefit available from an advisor.[7]
Do not add the gaps. A trading shortfall, a dollar-weighted return gap, an industry coaching estimate, and a lab's avoided mistake are not independent benefits that can be summed. They may concern overlapping behavior, different denominators, different populations, and different counterfactuals.
These studies warrant skepticism about universal sales claims, not a conclusion that useful advice is impossible. The strongest case is specific: a suitable plan, a credible process for sticking with it, competent planning, and a price that makes sense against the alternatives. The evidence here does not supply a portable annual “advisor effect.”
04 / Put a number on the hurdle
An AUM fee scales with assets, even when the workload does not. A flat fee fixes a dollar price, but it can be expensive relative to a smaller account. Both reduce capital that would otherwise remain invested. The SEC's fee bulletin is further reading on why recurring charges matter over long horizons and why the whole cost stack needs attention; our ledger discloses its search-indexed, rather than full-page, access.[8]
Instead of assuming an annual behavioral premium, ask what one prevented mistake would have to be worth. Then ask how often a comparable mistake is realistically avoidable. Reducing a return assumption by an arbitrary “behavior tax” forever quietly assumes the conclusion.
For the fee arithmetic without an assumed behavioral benefit, start with The Arithmetic of Fees. The lab below asks what a service must add to overcome that cost, not whether costs stop compounding when they buy advice.
Lab 02 / Price the promise
Isolate the invoice from the market path. A finite mistake removes a selected percentage of the DIY portfolio at a specified date. Advice is assumed to prevent it completely. The question is how large, or how frequent, those savings would have to be.
1 of 1 requested losses fit in this horizon. 4 event dates are available on the schedule. Break-even requires 2 avoided events at the selected size. These are requirements, not forecasts or evidence that the events will occur.
| Behavioral difference before advice fees | $125,730 |
|---|---|
| Less: terminal fee burden, including lost growth | $170,729 |
| Net advice difference | -$44,999 |
| Of the fee burden: actual fee dollars deducted | $101,579 |
| Remaining difference from growth on deducted fees | $69,150 |
Conditional return on the terminal fee burden: -26.36%. This is (avoided terminal loss minus terminal fee burden) divided by terminal fee burden. It is not an annualized investment return. Advice ends at $858,878 versus $1,029,607 for disciplined DIY without mistakes.
| Year | DIY, no preventable loss | DIY, selected finite losses | Advice, all selected losses prevented |
|---|---|---|---|
| 0 | $250,000 | $250,000 | $250,000 |
| 1 | $271,193 | $271,193 | $268,522 |
| 2 | $293,658 | $249,609 | $287,959 |
| 3 | $317,471 | $270,779 | $308,358 |
| 4 | $342,712 | $293,219 | $329,765 |
| 5 | $369,468 | $317,006 | $352,231 |
| 6 | $397,830 | $342,219 | $375,808 |
| 7 | $427,893 | $368,946 | $400,550 |
| 8 | $459,760 | $397,276 | $426,517 |
| 9 | $493,538 | $427,306 | $453,767 |
| 10 | $529,344 | $459,137 | $482,364 |
| 11 | $567,298 | $492,879 | $512,376 |
| 12 | $607,529 | $528,645 | $543,872 |
| 13 | $650,174 | $566,557 | $576,925 |
| 14 | $695,378 | $606,743 | $611,612 |
| 15 | $743,294 | $649,341 | $648,015 |
| 16 | $794,085 | $694,495 | $686,218 |
| 17 | $847,923 | $742,358 | $726,310 |
| 18 | $904,992 | $793,092 | $768,384 |
| 19 | $965,484 | $846,871 | $812,539 |
| 20 | $1,029,607 | $903,877 | $858,878 |
The monthly contribution arrives first, then the common monthly market return, then any specified DIY loss, then advice AUM and flat fees on the advice path. A loss is a permanent, instantaneous reduction of that portfolio's balance on the event date, after that month's return. Future contributions are unaffected. This is an abstract loss-equivalent, not a second model of panic selling.
Single loss needed solves for the percentage removed at the end of the nominated first-loss year that makes DIY terminal wealth equal advice terminal wealth. It uses just one event, regardless of the event-count control. If even 100% at that date cannot cover later fee drag, or the date is outside the horizon, the result is not reachable.
Events needed adds losses at the nominated first year and fixed spacing, stopping at the horizon. The first count that covers the fee burden is reported; repeated losses compound on the remaining DIY balance. Zero events means the fee hurdle is already zero. A saved mistake is never silently repeated.
AUM fees use the quoted nominal annual rate divided by 12; flat fees use annual dollars divided by 12 and do not inflate. The fee burden compares identical portfolios with and without advice charges, including the subsequent return on dollars deducted. No stochastic probabilities, excess market returns, tax savings, planning value, or paid-bill forgiveness are credited. Actual contracts and external payment of fees can differ.
When a service cannot clear the portfolio hurdle, that does not settle its value. It means the remaining case must be made in service terms, rather than relabeling the fee as investment outperformance. Conversely, a favorable modeled result does not demonstrate the service will deliver it.
05 / Buy a service, not a halo
A household can reasonably pay for clarity, coordination, time, and continuity. Those benefits do not have to masquerade as market alpha to be legitimate.
The table below is a purchasing framework, not an empirical estimate of returns. Ask for the scope, deliverable, responsible professional, cadence, and total fee in writing. A sophisticated portfolio is not a substitute for a completed plan.
| Need | A concrete deliverable | A credible alternative or limit |
|---|---|---|
| Behavioral support | A written investment policy, drawdown conversations, and a repeatable response to proposed changes. | Automated investing, an appropriate risk level, a cooling-off rule, and periodic paid reviews. Coaching cannot guarantee compliance. |
| Financial planning | Cash-flow, retirement, insurance, and contingency analysis that leads to assigned actions. | An hourly or project planner. More projections are not necessarily more useful decisions. |
| Tax coordination | A dated checklist shared with the tax professional: realizations, account location, charitable giving, and relevant deadlines. | A qualified tax professional plus an implementation plan. Gross harvested losses are not spendable tax alpha. |
| Administration | Beneficiary review, account consolidation where appropriate, withdrawal coordination, and documented follow-through. | A scoped administrative service or your own checklist. Verify who can legally provide each service. |
| Family continuity | A practical plan for incapacity or death, trusted contacts, and coordination with legal counsel. | A qualified estate attorney and an organized household record. Investment advice is not legal document preparation. |
A tax benefit depends on the household, the alternative strategy, applicable rules, implementation costs, and often future realization. Deferral is not always permanent avoidance. If a benefit is already reflected in after-tax portfolio returns, do not add it again as a separate service credit. Neither lab assigns a tax benefit by default.
Early retirement is one concrete coordination task: Before 59½ makes the conversion-tax bill, waiting periods, and spending bridge visible. Account access requires a funded plan and accurate rules, not a generic promise of “tax alpha.”
Time saved can be valuable even when terminal wealth is lower. Price it as a household preference, not a promised investment return. Compare an ongoing AUM service with annual flat-fee planning, hourly help, a one-time project, or a narrower engagement; these contracts need not cover equivalent work.
Ask how the firm and the individual get paid, whether products or referrals produce compensation, and what costs sit outside the advice quote. AUM pricing can create a tension when a sensible choice moves assets outside the billed account. Commissions can create a product-selection tension. Flat and hourly fees still require scope and quality control. The audit and advisor-belief studies show why both incentives and competence deserve attention.[2][4]
A professional standard matters, but it does not make costs or conflicts disappear. Request Form CRS where applicable and relevant adviser disclosures, ask for plain-language explanations, and verify registration and disciplinary history using the linked regulatory resources. Registration is not a performance endorsement.[9]
06 / Make the relationship falsifiable
If you choose not to delegate, write down a feasible allocation and liquidity plan, automate routine contributions, and specify when you will rebalance or revisit assumptions. Separate a genuine change in circumstances from a headline. Decide in advance when you would seek professional help. These are proposed safeguards, not a guarantee that behavior will improve.
The SWAN Number turns the liquidity part into a household constraint. A reserve that prevents a forced sale and an adviser who prevents a panic sale solve related, but different, problems.
If you choose to delegate, do not outsource the definition of success. An advisor who keeps you from selling low may be worth far more than the invoice in a particular episode. An advisor who adds cost, encourages chasing, or delivers little may be worth less than a simple system you can follow. The difference is the work delivered and the decisions changed.
Pay for useful help. Insist on a clear price. Keep the counterfactual honest.
Reader's ledger / Evidence and its boundaries
Reviewed as of . Claims are summarized, not quoted at length. Peer-reviewed studies, working papers, regulator education, interested industry analysis, and our own scenario arithmetic are different kinds of evidence. No cited paper is used to calibrate a universal annual advice premium.
Brad M. Barber and Terrance Odean (2000). The Journal of Finance 55(2), 773-806. DOI: 10.1111/0022-1082.00226.
Supports: Frequent trading was associated with substantially worse performance after transaction costs in the studied discount-brokerage accounts; gross-return differences were much smaller. See pp. 773-777.
Boundary: Historical directly held common stocks and transaction costs, with 1991-1996 account data. Not random assignment to advice, not a causal estimate of coaching, and not a description of every contemporary low-cost investor.
Sendhil Mullainathan, Markus Noeth, and Antoinette Schoar (2012). NBER Working Paper 17929, March. DOI: 10.3386/w17929.
Supports: Auditors presenting different portfolios encountered advice that often reinforced return chasing and favored higher-fee active funds, including when the initial portfolio was diversified and inexpensive. Abstract and numbered pp. 9-10 describe the setting.
Boundary: 284 Boston-area visits in April-August 2008. An audit of recommendations in initial consultations, not an experiment measuring long-term client wealth or the performance of all contemporary fiduciary relationships.
Stephen Foerster, Juhani T. Linnainmaa, Brian T. Melzer, and Alessandro Previtero (2017). The Journal of Finance 72(4), 1441-1482. DOI: 10.1111/jofi.12514.
Supports: Adviser identity substantially explains portfolio risk and home bias beyond measured investor characteristics in Canadian household data; adviser portfolios predict client portfolios. The published abstract describes limited customization and substantial costs.
Boundary: Historical Canadian distribution setting, not a universal causal hiring effect or a valuation of comprehensive planning. Final abstract and earlier draft differ in their cost figures; this article does not import either number into the labs. Data overlap with source 4 means the two are not independent replications.
Juhani T. Linnainmaa, Brian T. Melzer, and Alessandro Previtero (2021). The Journal of Finance 76(2), 587-621. DOI: 10.1111/jofi.12995.
Supports: Advisers invested their own money in ways resembling client recommendations, including expensive active funds and return chasing; patterns persisted after leaving the industry. This supports mistaken beliefs as an explanation alongside conflicts. See pp. 588-590.
Boundary: Two Canadian mutual-fund dealers, 1999-2013, with a legal and product environment different from many present-day services. Partly overlapping evidence with source 3. Benchmark-relative net alpha is not a raw investment loss or an annual advice benefit.
Andreas Hackethal, Michael Haliassos, and Tullio Jappelli (2012; working-paper version 2009). Journal of Banking & Finance 36(2), 509-524. Text reviewed: CFS Working Paper 2009/04, draft dated March 15, 2009. DOI: 10.1016/j.jbankfin.2011.08.008.
Supports: The accessible draft illustrates why adviser selection matters: investor characteristics differ across advice use, and performance conclusions depend on econometric adjustment. Its adjusted results do not support automatic improvement through delegation. See the draft abstract, PDF p. 4.
Boundary: German brokerage setting and nonrandom selection, with identification assumptions required for the adjustment. The final publisher text was blocked; draft-specific sample sizes and estimates are not represented here as verified final-journal findings.
Francis M. Kinniry Jr., Colleen M. Jaconetti, Michael A. DiJoseph, David J. Walker, and Maria C. Quinn (July 2022). Vanguard, Advisor's Alpha Perspectives.
Supports: Identifies potential implementation, behavioral, planning, and nonfinancial value. Explicitly says improvement should not be expected annually and is client-dependent. See pp. 2-4 and 17.
Boundary: An asset manager serving advisers, not an independent randomized advice study. Portfolios using specified best practices are compared with portfolios not using them. The framing is not a guaranteed 3% annual premium, and it does not show that every fee is justified.
Jeffrey Ptak (August 13, 2025). Morningstar, Portfolio and Planning Research.
Supports: The report explicitly notes that regular paycheck investing and rebalancing can create a return gap, alongside less disciplined trading. Its investor return measure depends on cash-flow size and timing. See pp. 1-3 and 20-21.
Boundary: Not a causal advice experiment, not a measure of the average individual investor, and not wholly panic selling. The 2025 edition covers the decade ending December 31, 2024 and changes its aggregate comparator methodology. No 2026 edition is asserted. Primary-site access was challenged; the report was read from the linked mirror.
U.S. Securities and Exchange Commission, Investor.gov (July 23, 2025). Investor education bulletin.
Supports: Explains that ongoing fees reduce capital available to earn returns and encourages investors to compare the full fee stack. The article uses this as further reading; the labs use their own disclosed monthly arithmetic, not the bulletin examples.
Boundary: The official page and update date were checked through search-indexed primary-page information. Direct retrieval returned HTTP 403, so the current full page was not reviewed. Educational guidance is not an estimate of any adviser treatment effect.
U.S. Securities and Exchange Commission, Investor.gov (August 6, 2020). Investor education bulletin; see also Investor.gov/CRS.
Supports: Describes relationship summaries covering services, fees, conflicts, standards of conduct, and reportable disciplinary history, and directs readers to research their investment professionals.
Boundary: Official indexed content and publication date were checked, but direct page retrieval returned HTTP 403. This is a due-diligence resource, not a complete statement of current legal duties. Registration and disclosure are not guarantees of performance or suitability.
The two labs are original, deterministic educational models. They use no market feed, personal-account data, or external calculator service. Assumptions stay in this browser session and are not sent to analytics. Rounded displays can differ slightly from the unrounded accounting. Controls intentionally admit cases in which advice underperforms.
The first lab changes exposure on a fixed synthetic return path. The second prices finite portfolio losses on a smooth assumed path. Neither is a forecast, optimization recommendation, or estimate of how a named advisor would behave. Their detailed accounting is disclosed inside each lab and in this module's methodology README.