Personal Finance Gurus vs. Economists: When Simple Rules Beat Textbook Models

In 2022, Yale finance professor James Choi did something nobody had bothered to do systematically: he read the 50 most popular personal finance books and checked their advice against what economic models prescribe. The results were published in the Journal of Economic Perspectives, and the summary table is brutal reading if you like the books. Save 10 to 15% of your income no matter what? The models say your savings rate should swing with your income. Keep your wealth intact in retirement? The models say spend it down. Divide your money into labeled buckets? A dollar is a dollar.¹

The paper is usually passed around as proof that the gurus are frauds. That reading does not survive contact with Choi’s own conclusion, which is worth quoting in full because almost nobody does:

“Popular financial advice can deviate from normative economic theory because of fallacies. But popular financial advice has two strengths relative to economic theory. First, the recommended action is often easily computable by ordinary individuals; there is no need to solve a complex dynamic programming problem. Second, the advice takes into account difficulties individuals have in executing a financial plan due to, say, limited motivation or emotional reactions to circumstances. Therefore, popular advice may be more practically useful to the ordinary individual.”

Choi (2022), p. 186.

Some of these disagreements represent a genuine fallacy in the popular advice. Some represent something real that the models leave out. And some are artifacts of comparing a person’s Tuesday-morning decision against a frictionless model. We read the paper, checked its claims against the underlying research, and sorted all of it into those three buckets. One of the disagreements, it turns out, does not survive a look at the paper Choi cites for it.

What Choi did

The method matters for interpreting the result. Choi took the 50 most popular personal finance books as ranked by Goodreads in May 2019. Three contained no advice on his chosen topics, leaving 47. The list runs from The Richest Man in Babylon (1926) and Benjamin Graham to Bogle, Bernstein, Malkiel, Ramsey, Orman, Kiyosaki, JL Collins, and Ramit Sethi. He then compared what they recommend against benchmark results from academic finance on saving, portfolio equity share, dividends, factor tilts, international diversification, active management, debt paydown, and mortgages.¹

These books are not fringe. Choi notes that Rich Dad Poor Dad had sold 32 million copies, that Dave Ramsey’s Total Money Makeover had sold 1.5 million, and that Ramsey’s radio show reported 18 million weekly listeners. Felix Chopra’s working paper, using the staggered expansion of the show’s radio affiliates from 2004 to 2019, estimates that households in areas gaining access cut retail spending by 1.3%, which the paper bounds at a decrease of at least 5.4% for households exposed to the show. Whatever you think of the advice, it moves money.²

One limitation to hold onto from the start: this is a survey of what books recommend, compared against what models prescribe. It is not a study of whether people who follow guru advice end up wealthier. No part of the paper measures outcomes, and that gap is exactly where the most interesting arguments live.

The scorecard, as the paper prints it

Here is Choi’s Table 1, condensed. Read the last two rows carefully, because they are the ones that get dropped when this table is reposted.

One row agrees outright. On active management, both columns say the same eight words. Twenty-four of the books recommend indexing and only seven recommend active management, one of them written by Peter Lynch, whose position is at least understandable. The lifecycle equity row agrees on its most-quoted element: both columns say the allocation should be hump-shaped with age. The rest of that row differs, with the books putting near-term money entirely in cash while the model conditions the equity share on marginal utility. Choi is explicit that “economic models also recommend equity allocations that are hump-shaped with age, but for somewhat different reasons than those found in popular books.” When this table circulates as a list of things gurus get wrong, the hump-shape agreement is routinely miscounted as a disagreement.¹

Where the economics should win

Start with the disagreements where the popular side is making an actual mistake, and a behavioral justification would just be an excuse for bad arithmetic.

Dividends treated as income the company hands you

Nine of the books reject dividend irrelevance in some form, and not one recommends avoiding dividends for tax reasons. Choi quotes Peter Lynch arguing that “the presence of the dividend can keep the stock price from falling as far,” then checks it: from July 1927 through June 2022, a value-weighted portfolio of all non-dividend-paying stocks had more positively skewed monthly returns than dividend payers in the bottom 30% or middle 40% of the positive-yield distribution, which is inconsistent with the cushion story, though a skewness comparison is an indirect test of downside protection.¹ A dividend moves value out of the share price rather than adding to it, and in a taxable account it accelerates the tax bill, which is the subject of our guide on dividends.

Picking active managers

This one is not a disagreement between the camps, so much as a disagreement between the good books and the bad ones. Choi cites Kenneth French’s estimate that the average actively managed US equity fund trails the average passive fund by 0.67% per year. The evidence since has not been kind to the seven books recommending active management: in S&P’s year-end 2025 SPIVA scorecard, 78.78% of actively managed large-cap US equity funds underperformed the S&P 500 during 2025, and the failure rate climbs with the horizon, reaching 88.96% over five years and 92.89% over twenty.³ See our indexing guide for the full record.

The stated reasons for extreme home bias

Twenty-six books address international investing; only two say to skip it entirely; and among those naming a number, the average recommendation is 27% of equities, ranging from 12.5% to 50%. Choi works through the usual justifications and finds most of them weak: US multinationals’ foreign revenue is not the same as owning foreign securities (the subject of a guide of ours), rising cross-market correlations have not eliminated the diversification benefit, currency risk can be hedged cheaply in major currencies, and a widely expected strong economy is already in the price. That last point is the one worth tattooing somewhere: expected growth is not expected return.¹

One number here needs care, because it will otherwise look like it contradicts your brokerage statement. Choi’s benchmark is that non-US stocks were 59% of global market capitalization in 2021, which is correct on the basis he cites: SIFMA’s research quarterly put US equities at 41.1% of a $122 trillion global market.⁴ That measure counts all listed domestic market capitalization. The index inside a total-world fund does not. FTSE Global All Cap, which VT tracks, adjusts for free float and investability, and had ex-US at 38.0% as of July 31, 2026 (40.0% at the end of 2021).⁵ Closely held and state-owned stakes are far more common outside the US, so float adjustment cuts the ex-US share hard. Both numbers are right; they answer different questions. Market weight for an investor is the float-adjusted one, and 27% is still an underweight against it, just a much smaller one than the paper’s framing implies.

Ignoring the interest rate when repaying debt

The arithmetic is not in dispute: a dollar sent to a 27% balance beats a dollar sent to a 6% balance. Choi cites evidence that households frequently do not do this, and it is not simple laziness. Gathergood, Mahoney, Stewart, and Weber found that people repaying multiple credit cards follow a “balance-matching” heuristic, allocating repayment in proportion to each card’s share of total balances rather than to the rate. That pattern explains more than half the predictable variation in repayments.⁶ Whatever the motivational case for a snowball, and there is one below, balance matching is just a costly habit.

Preserving principal as a goal in itself

Of the 15 books giving retirement spending advice, five explicitly set the withdrawal rate at or below an expected real return, making capital preservation the objective. The lifecycle model says savings exist to be converted into consumption. Dying with your retirement portfolio intact in real terms is a fine outcome if you wanted a bequest or held it back against longevity, medical, or long-term care risk, and a sign of underspending if it served none of those. The outlier in the sample is Ramsey’s 8% withdrawal rate, built on an assumed 12% nominal return minus 4% inflation, which we take apart in a dedicated guide. Seven books say 4%, which is the more common answer, and our withdrawal-rate guide covers why even that is a starting point rather than a rule.

What the gurus caught that the models leave out

Now the other direction. In these cases the popular advice looks inefficient on a spreadsheet and is defensible anyway, because it is solving a problem the benchmark model assumes away.

Automation and paying yourself first

The phrase “pay yourself first” appears in 16 of the books and dates to Clason in 1926. It reframes saving as the thing you consume rather than the thing you give up, and it works by removing the decision: money leaves before you see it. Choi points out the discipline argument is “almost always missing from economic models of optimal saving,” which is an unusually direct concession.¹ Thaler and Benartzi’s Save More Tomorrow, which commits future raises to saving in advance, reported participant savings rates rising from 3.5% to 13.6% over 40 months at one midsize manufacturer. Treat that as a strong idea rather than a settled fact: participants self-selected into the program, and the US Department of Labor rates the causal evidence as low.⁷

Choi’s own later work sharpens that caveat. Across nine 401(k) plans, he and coauthors estimate that once job turnover, early withdrawals, and opt-outs are counted, automatic enrollment raises steady-state saving by about 0.6% of income and default auto-escalation by about 0.3%.⁸ Automation still clears Choi’s computability and motivation bars; it cannot substitute for choosing an adequate rate and revisiting it.

Mental accounting

Seventeen books tell you to split money into labeled buckets, and standard theory says a dollar is a dollar. Both are right about different things. Mental accounting causes real errors, most visibly when someone protects a labeled savings balance while carrying a credit card at 25%. But Choi makes the defense himself, and it is the most interesting paragraph in the paper: economic models generally assume utility is the same every period and that goods are infinitely divisible, when real spending is lumpy and unusually valuable at specific moments, like a wedding, a move, or a first year of college. Funding a labeled account ahead of a known large expense has the useful effect of lowering consumption during the period when consumption is worth less, and it lets you check whether you are on track.¹ Choi also cites Karlan and coauthors, who found in field experiments at three banks that reminder messages raised commitment-savings attainment, with messages mentioning savings goals and financial incentives particularly effective.⁹

The synthesis we would offer: keep buckets for behavior, and use one consolidated balance sheet for decisions. Name the accounts if it helps you fund them. But when you are deciding whether to pay down debt, rebalance, or fund a 529, look through the labels and evaluate the household as a whole, which is the argument in our household portfolio guide.

Emergency liquidity

Twenty-eight of the books say everyone should build an emergency buffer, with recommendations ranging from $1,000 to two years of income. Choi does not put a liquidity row in Table 1, but he does discuss the “wealthy hand-to-mouth” literature showing that roughly 20% of US households hold substantial illiquid assets and almost no liquid ones, a pattern that can be rationalized if illiquid assets earn high enough risk-adjusted returns.¹ The books reject that trade, and so, in his other work, does Choi: he is a coauthor on research proposing automatic enrollment into employer-sponsored rainy-day accounts, motivated partly by the finding that for every $1 flowing into 401(k)-type accounts, 30 to 40 cents leaks out before retirement.¹⁰ The books are directionally right and wrong on the specific number, which is why we size it by risk exposure rather than by a flat multiple in our emergency fund guide.

The debt snowball

The book sample splits evenly: ten recommend prioritizing the highest-interest debt, ten recommend something else, nine of those being the snowball, and two saying to attack whichever debt bothers you most. Ramsey is unusually candid about why, and Choi quotes him: “if you were doing math, you wouldn’t have credit card debt, would you? This is about behavior modification.”¹

The research splits the same way, and the two halves are often conflated. Amar, Ariely, Ayal, Cryder, and Rick documented what they called debt account aversion: in an incentive-compatible lab game, people paid off small balances first at real cost to themselves.¹¹ That is the cost side. On the benefit side, Kettle, Trudel, Blanchard, and Häubl, using a field study of indebted consumers plus three experiments, found that concentrating repayment into one account raises motivation to become debt free, and that the effect is strongest when the repayments go to the smallest account, because people read progress from the largest proportional balance reduction in any one account.¹² Neither of these is cited in Choi’s paper, and together they say the snowball is a real behavioral device with a real price tag.

Our position, which matches our debt payoff guide: default to the avalanche, and reach for the snowball when a credible adherence benefit is worth the rate gap. The snowball does not beat the avalanche mathematically. It beats an avalanche you abandon. And the size of the rate gap decides how much that insurance costs: choosing a 5.5% debt ahead of a 6% debt is nearly free, while choosing a 3% loan ahead of a 29% card is not.

Holding cash while carrying expensive debt

Fourteen books endorse co-holding, and the most commonly cited reason (seven books) is that keeping a buffer prevents you from running the balance back up. Choi notes that only one of the fourteen mentions any justification from the academic literature. The behavior is common: Gross and Souleles found that a third of households paying high credit card rates simultaneously held at least one month of income in low-interest liquid assets, and in May 2022 the average card rate was 16.65% against 0.06% on savings.¹

The most useful recent evidence suggests this is a choice rather than an oversight. In a 2026 study of Commonwealth Bank of Australia customers, co-holding affected 23% of credit card users, and the median co-holder was paying about AUD$245 in interest while earning about AUD$15 on liquid assets. A randomized trial across 125,328 customers sent app notifications spelling out that cost, and repayment moved by about AUD$20, roughly 1.4% on balances near AUD$6,500, which is neither statistically nor economically meaningful. That rules out the simplest reading, that co-holders have not noticed the cost, and is consistent with the buffer being a deliberate purchase, though the experiment cannot say what households believe they are buying.¹³ That does not make the cost vanish. Our reading: keep enough liquidity that a small shock does not immediately recreate the debt, send everything else at the expensive balance, and rebuild the buffer after.

Where the benchmark column is a single model

The third bucket is the one that changes how you should read the whole table. Choi flags in the paper itself that Table 1 involves “some oversimplification.” In several rows, the “benchmark academic advice” is the output of one stylized model rather than a position modern household finance would defend.

The mortgage row does not match the paper it cites

Table 1 says economists would choose an adjustable-rate mortgage “unless interest rates are low.” The citation is Campbell and Cocco. We read the 2003 paper, and it does not say that.

What it says is that ARM attractiveness depends on household characteristics. From its abstract: “While an ARM is generally an attractive form of mortgage, a household with a large mortgage, risky labor income, high risk aversion, a high cost of default, and a low probability of moving is less likely to prefer an ARM.” That is a five-part test about the borrower, not a rule about the rate environment.¹⁴

On the rate level specifically, the paper runs the opposite direction: “the ARM is even more advantageous if the interest rate is initially low, since in this case the ARM has a lower cost in the early years when borrowing constraints are most severe. The FRM is more attractive if the interest rate is initially high.” And Campbell and Cocco explicitly reject rate timing as a basis for the choice, quoting a popular finance book advising readers to lock in when rates are low and responding that “movements in long rates are extremely difficult to forecast” and that it “seems overambitious for the average homeowner to try to predict movements in long-term interest rates.”¹⁴

So the row compresses a conditional household result into a rate-timing rule whose direction the source paper argues against. Choi’s spoken version elsewhere is closer to the research, adding that someone stretching their budget should take the fixed rate, which maps onto the paper’s large-mortgage condition. The practical upshot for a reader: a fixed-rate mortgage buys payment certainty plus a valuable refinancing option, and whether that is worth its price depends on your income stability and how large the payment is relative to your budget. Comparing the fixed and adjustable rates actually on offer belongs in that decision; forecasting where rates go next is the part Campbell and Cocco argue against. Our mortgage guide covers what the fixed rate does for you on the liability side.

Full annuitization is a 1965 result with five assumptions

The row saying to fully annuitize traces to Yaari (1965). That result requires expected-utility preferences, intertemporally separable utility, mortality as the only uncertainty, no bequest motive, and actuarially fair annuities. Davidoff, Brown, and Diamond generalized it in 2005 and found that with complete markets the case for some annuitization survives dropping most of those assumptions, but that under incomplete markets, “positive annuitization remains optimal widely, but complete annuitization does not.”¹⁵

Pricing matters too. Poterba and Solomon, using US retail annuity prices as of May 2024, estimate that an immediate annuity bought at 65 delivers an expected discounted payout of about 87 cents per premium dollar for a buyer with population mortality, versus close to a dollar for a buyer whose mortality resembles the typical annuity purchaser.¹⁶ Add that most retirees already hold a large inflation-indexed annuity called Social Security, and “fully annuitize” is a theoretical benchmark rather than planning advice. Partial flooring is the live question, which is what our income floor guide works through.

Factor tilts are an open debate presented as a verdict

“May or may not be attractive” is an honest summary of an unsettled literature, and it is not a rebuttal of the 26 books making equity-style recommendations, among them eight recommending a value tilt and fourteen recommending small stocks. Choi’s sharper observation is that the books mostly present tilts as free improvements: only three mention that value stocks might be riskier, and he checks one specific claim, that growth outperforms in recessions, against the data. From July 1926 to June 2022, the value-minus-growth factor averaged 0.37% per month during US recessions plus the first year of recovery, against 0.35% otherwise.¹ The tilt may be defensible; the story told to justify it often is not. See our factor guide.

The concept the books never mention

The most original finding in the paper gets almost no attention. Only five of the books suggest that diminishing marginal utility should drive your portfolio, and none mentions the covariance of returns with marginal utility, the engine of consumption-based asset pricing. Choi offers a conjecture: these models may fail empirically because people simply do not make portfolio decisions with such covariances in mind. Equilibrium prices do not require anyone to compute that covariance consciously, so read this as an intriguing lead rather than a settled diagnosis.¹ That is a paper about popular books quietly raising a question about why a large academic literature underperforms.

The savings-rate fight, in numbers

The headline disagreement deserves its own treatment because it is the one most likely to be misquoted at you. Forty-five of the books give savings advice: 32 stress starting immediately, 31 invoke compound interest, and 21 recommend a positive rate that does not vary with age, with 10 to 15% covering most of them. Only one of those 21 adjusts for how much you have already saved. Only four books account for Social Security at all, despite it replacing 64% of final earnings for the median new beneficiary aged 64 to 66 in 2005.¹

The theory objection is real. Lifecycle models target smooth consumption, and the savings rate is just whatever is left over. Since income is hump-shaped over a career, the implied savings rate is low or negative early, high in midlife, and negative in retirement. Choi also notes the models make the opportunity cost of current consumption a key driver, which popular authors replace with an appeal to compound interest. Compound interest tells you what a dollar saved becomes. It cannot tell you whether that dollar was worth more to you at 25 than at 55.

The calculator below runs both plans against the same income path. The two plans spend identical lifetime resources in present value, so the only difference is timing. Then flip the borrowing switch, which is where the theoretical prescription meets the world.

On the default settings, the textbook plan asks a 25-year-old earning $60,000 to spend about $98,500, which requires borrowing roughly $38,000 a year against income they have not earned yet. No lender offers that product. Turn borrowing off and the smoothed plan does something more troubling: it prescribes saving nothing at all for about 15 years, then saving over a third of income from 40 onward. That is a plan that requires a person to flip from zero to supersaver on a specific birthday, which is the exact objection David Chilton raises in one of the books Choi quotes: “most people aren’t going to be able to transition from setting aside nothing to being supersavers at the flip of a switch.”

Both sides have a point here, and the frictions favor the rule. A flat savings rate works as a stable feedback rule under conditions the benchmark model leaves out: you cannot forecast your income precisely, cannot borrow freely against it, and cannot rely on your future self to triple a savings rate on schedule. The improvement available is not abandoning the rule; it is escalating it with raises and recalibrating it against what you have already accumulated, which is the one adjustment 20 of the 21 books forgot.

Three things the paper is repeatedly said to prove

“Economists say not to save in your 20s.” The models say the optimal savings rate is low or negative when income is low relative to lifetime income, under assumptions that include being able to borrow freely and knowing your future earnings. Choi also discusses buffer-stock models in which uncertain labor income makes several months of liquid assets rational, and he coauthors research arguing for automatic emergency savings. The prescription is about the shape of the path, not a license to spend your twenties.

“Economists say to fully annuitize.” One model from 1965 says so under five assumptions. The modern generalization concludes that some annuitization is widely optimal and complete annuitization is not, and market prices are not actuarially fair.

“Economists say to get an ARM.” The cited paper says ARM preference depends on five household characteristics, and it argues against choosing based on a forecast of where rates are headed. The current level of rates does enter its analysis, in the opposite direction from the condition the table attaches: a low initial rate makes the ARM more attractive.

What the paper does not establish

This is a good paper being asked to carry more than it weighs. Five limits worth stating plainly:

• Goodreads rank is not exposure. A snapshot of one website’s rankings in May 2019 is a reasonable sampling device, not a measure of how much each recommendation reaches households.
• Each book counts once, so prolific authors count several times. Ramsey and Orman each contribute three books to the sample, Bach three, Bernstein and Bogle two apiece. Those observations are not independent, so a prolific author’s view is counted several times.
• The sample spans almost a century. A 1926 book and a 2019 book faced different tax codes, retirement systems, product menus, and trading costs.
• The benchmark column is a choice. It was not produced by surveying economists. Choi selects benchmark models and says so, flagging “some oversimplification” in the table itself. The mortgage row shows how much that can matter.
• Nothing here measures welfare. The paper compares prescriptions. It cannot tell you whether a person who follows the snowball ends up better or worse off once you account for whether they stick with it, which is the question that decides the advice.

Choi’s own next move is telling. His paper closes by suggesting that building normative models with computability and limited motivation built in “may be a fruitful direction for future research,” and in 2025 he coauthored exactly that: an approximate solution to lifecycle portfolio choice that runs in a spreadsheet and costs an average of 0.06% of welfare relative to the optimal solution across 5,103 parameter sets.¹⁷ The author of the critique agrees the usability problem is real enough to work on.

A framework for using both

The division of labor that falls out of all this is simple enough to keep: economics is good at telling you where you should end up, and behavioral rules decide whether you get there.

What we recommend

A rule worth following has to clear three bars. The economics has to point the right direction, because a behavioral hack built on a fallacy is just a fallacy with better marketing. It has to survive imperfect forecasts, since personal finance runs on estimates you cannot verify. And you have to be able to execute it for decades, because a theoretically optimal plan you abandon in year three is worth less than a decent plan you keep.

Applied to Choi’s topics, that produces: automate saving but recalibrate the rate against what you have already accumulated and what Social Security will provide; hold liquidity sized to your own shock exposure rather than a flat multiple; use labeled buckets to execute and a consolidated balance sheet to decide; index; skip the dividend tilt; treat float-adjusted global market weight as the default starting point and any US overweight as a deliberate bet; repay the highest rate first, accepting the snowball’s premium only when it is small and adherence is the binding constraint; treat the mortgage as a risk-management decision priced against your budget rather than a moral question; and plan to spend your retirement savings, using partial guaranteed income if the fear of running out would otherwise distort your spending.

The deeper lesson is about what kind of problem this is. Personal finance is a control problem run by an imperfect controller under uncertainty. Academic models describe the ideal trajectory given the model. Popular finance designs crude feedback rules that stay stable when the inputs are noisy and the operator is tired. You want both, and you should be suspicious of anyone selling you only one.

Key takeaways

• Choi’s own conclusion is not a takedown. He credits popular advice with two strengths: it is computable without solving a dynamic programming problem, and it accounts for limited motivation.
• One Table 1 row agrees outright, and a second agrees on the hump shape. Both columns say index rather than pick active managers, and both say the equity share should be hump-shaped with age, for somewhat different reasons.
• Some guru advice is a genuine fallacy. Dividends as free income, active manager selection, and ignoring interest rates when repaying debt are all cases where the arithmetic decides it.
• Some guru advice solves a problem the models skip. Automation, emergency liquidity, goal-labeled accounts before lumpy expenses, and the snowball as a commitment device all have defenses Choi partly makes himself.
• One benchmark row conflicts with its own source. Table 1 says to choose an ARM unless rates are low; Campbell and Cocco condition on five household characteristics and argue that a low initial rate makes an ARM more attractive.
• The saving disagreement is about shape. The smoothed plan needs borrowing no lender offers, or else prescribes saving nothing for roughly 15 years and then over a third of income. A flat rate is a stable rule rather than the optimum.
• The paper compares prescriptions, not outcomes. It never tests whether readers of these books end up wealthier, which is the question that would settle most of the argument.

How Summitward helps

The single adjustment that would improve the books’ advice most, recalibrating your savings rate against what you have already accumulated, requires knowing what you have accumulated. Summitward tracks net worth over time across accounts and tax buckets, projects it forward, and shows your progress against financial independence targets, which turns a fixed percentage into a number you can revisit with evidence.

Frequently asked questions

Is Dave Ramsey wrong?

On the parts that are arithmetic, sometimes yes: an 8% withdrawal rate built on 12% assumed nominal returns is not defensible, and the debt snowball costs more interest than the avalanche. On the parts that are about behavior, he is making a claim Choi partly endorses, and he says so openly: the snowball is “about behavior modification.” Chopra’s research also suggests the show measurably reduces listener spending, which is the outcome the advice is aiming at.

Do economists really say you should not save in your 20s?

No. Benchmark lifecycle models imply a low or negative savings rate when current income sits far below lifetime income. Those models assume you can borrow against future earnings and know what they will be. Remove those two assumptions and the same models produce buffer-stock saving, which is several months of liquid assets. Choi himself coauthors research promoting automatic emergency savings accounts.

Should I switch from the snowball to the avalanche?

If you are currently executing a snowball and on track, the switching decision depends on how far apart your rates are. Across a spread like 3% to 29%, ordering by rate is worth real money. Across debts all clustered near the same rate, the interest difference is small and the motivational benefit documented by Kettle and coauthors may be worth more. Run your actual balances through the calculator in our debt payoff guide before deciding.

Does the paper say I should get an adjustable-rate mortgage?

The table says so, with a rate-level condition attached. The research it cites conditions instead on the size of the mortgage relative to your income, the riskiness of your income, your risk aversion, your cost of default, and how likely you are to move. If your payment is large relative to your budget or your income is volatile, that research points toward the fixed rate.

Did Choi test whether guru readers end up richer?

No, and this is the most important thing to know about the paper. It compares what books recommend against what models prescribe. It contains no outcome data on readers, so it cannot settle whether a simple rule that people follow beats an optimal plan that they do not.

Which books came out best?

Choi does not rank the books, and the paper is organized by topic rather than by author. The books recommending indexing, warning against credit card debt, and pushing automatic saving line up with both columns of the table. Our own tiered reading list is in finance books for DIY investors.

Related guides

• Finance Books for DIY Investors: which of these books are worth your time, tiered by what they are good for
• Consumption Smoothing for Professional Students: the savings-rate argument applied to the readers it fits best, people with steep expected income growth
• Debt Payoff Strategies: avalanche and snowball run head to head on real balances, with a calculator
• Emergency Fund Sizing: replacing the flat three-to-six-month rule with your own shock exposure
• Lifecycle Finance: the Modigliani and Yaari framework the benchmark column is drawn from
• Safe Withdrawal Rate: where 4% came from and why it is a starting point
• The 8% Withdrawal Rate: the specific claim in Choi’s sample that fails on its own return assumption
• The Case for Global Diversification: why home bias costs you, and what market weight means
• Just Invest in Index Funds: the row where both columns agree, and the SPIVA record behind it
• You Have One Household Portfolio: using buckets for behavior and one balance sheet for decisions

Author disclosure

Summitward sells a personal finance product, which makes this a paper about our own industry. We have tried to be at least as hard on the advice we agree with as on the advice we do not. Nothing here is personalized financial advice.

Sources

1. Choi, J.J. (2022). “Popular Personal Financial Advice versus the Professors”. Journal of Economic Perspectives, 36(4), 167-192. The source for the sample construction, Table 1, every book count in this guide, the dividend skewness and value-factor recession checks, and the conclusion quoted at the top.
2. Chopra, F. (2023). “Media Persuasion and Consumption: Evidence from the Dave Ramsey Show.” Working paper, November 2023 version, still unpublished as of 2026. Source for the 1.3% intent-to-treat spending effect and the bounded 5.4% estimate for exposed households.
3. S&P Dow Jones Indices (March 2026). SPIVA U.S. Scorecard, Year-End 2025, Report 1a, data as of December 31, 2025. Source for the 78.78% one-year and 88.96% five-year large-cap underperformance figures.
4. SIFMA (2021). SIFMA Research Quarterly, 3Q21. Source for US equities at 41.1% of the $122 trillion global market, the basis of Choi’s 59% ex-US figure.
5. FTSE Russell. FTSE Global All Cap Index factsheet. Source for the float-adjusted ex-US weight of 38.0% as of July 31, 2026 and 40.0% as of December 31, 2021.
6. Gathergood, J., Mahoney, N., Stewart, N. & Weber, J. (2019). “How Do Individuals Repay Their Debt? The Balance-Matching Heuristic”. American Economic Review, 109(3), 844-875. Cited by Choi for the finding that repayment follows balance shares rather than interest rates.
7. Thaler, R.H. & Benartzi, S. (2004). “Save More Tomorrow: Using Behavioral Economics to Increase Employee Saving”. Journal of Political Economy, 112(S1), S164-S187. The 3.5% to 13.6% result is from the first implementation at one anonymous midsize manufacturer with self-selected participants; the US Department of Labor rates the causal evidence low.
8. Choi, J.J., Laibson, D., Cammarota, J., Lombardo, R. & Beshears, J. (2024). “Smaller than We Thought? The Effect of Automatic Savings Policies”. NBER Working Paper 32828, revised November 2024. Source for the nine-plan steady-state estimates: about 0.6% of income from automatic enrollment and 0.3% from default auto-escalation, after job turnover, withdrawals, and opt-outs.
9. Karlan, D., McConnell, M., Mullainathan, S. & Zinman, J. (2016). “Getting to the Top of Mind: How Reminders Increase Saving”. Management Science, 62(12), 3393-3411. Field experiments at three banks; cited by Choi for salience raising savings motivation.
10. Beshears, J., Choi, J.J., Iwry, J.M., John, D.C., Laibson, D. & Madrian, B.C. (2020). “Building Emergency Savings Through Employer-Sponsored Rainy-Day Savings Accounts”. Tax Policy and the Economy, 34, 43-90. Source for the 30 to 40 cent retirement-account leakage figure and Choi’s own support for emergency liquidity.
11. Amar, M., Ariely, D., Ayal, S., Cryder, C.E. & Rick, S.I. (2011). “Winning the Battle but Losing the War: The Psychology of Debt Management”. Journal of Marketing Research, 48, S38-S50. Source for debt account aversion in an incentive-compatible lab setting. Not cited by Choi.
12. Kettle, K.L., Trudel, R., Blanchard, S.J. & Häubl, G. (2016). “Repayment Concentration and Consumer Motivation to Get Out of Debt”. Journal of Consumer Research, 43(3), 460-477. A field study plus three experiments; source for concentrated repayment raising motivation, most strongly on the smallest account. Not cited by Choi.
13. Batista, R.M., Mao, E., Sussman, A.B., Mahoney, N. & Min, J. (2026). “Disclosing the costs of co-holding liquid assets and high-interest debt has limited impact on behavior”. PNAS Nexus, July 28, 2026. Commonwealth Bank of Australia data; source for the 23% co-holding rate, the median AUD$230 annual cost, and the null result from a randomized disclosure trial across 125,328 customers.
14. Campbell, J.Y. & Cocco, J.F. (2003). “Household Risk Management and Optimal Mortgage Choice”. Quarterly Journal of Economics, 118(4), 1449-1494. Source for the five household conditions, for the finding that a low initial rate makes an ARM more attractive, and for the paper’s rejection of rate-timing advice.
15. Davidoff, T., Brown, J.R. & Diamond, P.A. (2005). “Annuities and Individual Welfare”. American Economic Review, 95(5), 1573-1590. Source for the enumeration of Yaari’s assumptions and for the finding that positive annuitization remains widely optimal while complete annuitization does not under incomplete markets. Yaari, M.E. (1965), Review of Economic Studies 32(2), 137-150, is the original result.
16. Poterba, J.M. & Solomon, A. (revised October 2025). “New Evidence on the Money’s Worth of Immediate and Deferred Individual Annuities”. NBER Working Paper 28557, using US retail annuity prices as of May 2024. Source for the roughly 87 cents per premium dollar at population mortality.
17. Choi, J.J., Liu, C. & Liu, P. (2025). “Practical Finance: An Approximate Solution to Lifecycle Portfolio Choice”. NBER Working Paper 34166. Source for the spreadsheet-computable approximation costing an average 0.06% of welfare across 5,103 parameter sets.