2026-09-05 Weekly Reads: Job Hopping Is Not Just About a Pay Raise

Why I chose this

This week's WeChat reposts have been buzzing about a few connected topics: how young people can fight for rewards that match their contributions, how short the window for career advancement really is, what each choice costs, and whether solo companies and flexible employment are realistic options for most people. A lot of the answers end up landing on personal character—whether you dare to act, whether you can persist, whether you're willing to pay the price.

Alex Clymo, Piotr Denderski, Yusuf Mercan, and Benjamin Schoefer's Cautious Careers is worth reading because it doesn't simply explain away "staying put" as timidity, nor does it frame job-hopping as a single-variable decision that happens whenever the pay raise is big enough. What it asks is: when a new job pays more but carries a higher risk of unemployment, what price do people put on that risk—and how does that price shift when someone lacks cash buffers and insurance?

The most useful part of this paper isn't that it decides for anyone whether to switch jobs. It breaks "the cost of choosing" into observable constraints. Courage exists, sure—but so does the capacity to absorb failure. Two people facing the same opportunity can make different choices, and that isn't necessarily about who's more enlightened. It might come down to how long they can survive after losing a job, whether household spending can shrink, and whether their safety net stays intact.

The paper's core argument

The paper first frames career mobility as a two-dimensional ladder. Standard job-ladder models mostly look at wages—a new job with higher pay counts as moving up. The authors add a second dimension: the probability that a job pushes you into unemployment. In reality, higher-paying jobs are safer on average, but at the same wage level there's still wide variation in unemployment risk, and job switchers move both to safer and riskier positions. So a pay raise and safety are not the same thing.

To measure this trade-off directly, the authors embedded custom questions into NORC's AmeriSpeak nationally representative survey in May–June 2025, with a final analysis sample of 1,008 employed respondents. Respondents first reported their current wage and their subjective probability of unemployment over the next 12 months. Then they faced a set of new jobs identical to their current one except for unemployment risk—at 2%, 5%, 10%, 20%, and 50%—and stated the minimum wage they'd require to accept each.

The preferred estimate for full-time respondents: each 1-percentage-point increase in annual unemployment risk at a new job requires, on average, an additional 1.63% of current wages for the respondent to accept it. The standard error is 0.074, and the coefficient stays directionally stable across checks that drop speedsters, relax completeness and monotonicity requirements, and alter extreme-value handling. This isn't a quote formula for any individual—it's the "price of risk" in the sample-average sense.

The bigger differences come from capacity to absorb shocks. The authors approximate household self-insurance capacity with "how much consumption would fall after job loss": people who expect to cut consumption sharply after unemployment demand roughly 75% more compensation for taking on extra unemployment risk. This suggests that seemingly conservative career choices may not be about failing to see the pay raise—it's that downside risk punches through household cash flow much faster.

The paper also draws supporting evidence from the U.S. Current Population Survey and the Survey of Consumer Expectations. Jobs at similar wages show clear risk dispersion; when people switch jobs, the perceived change in unemployment risk is larger than when they stay; and people who are more willing to take risks in daily life also report higher switching intentions. The authors don't treat these correlations as standalone causal proof—they use them to confirm that the wage-safety trade-off actually exists in real labor markets.

The authors then feed the survey-measured risk price into an equilibrium model with on-the-job search, productivity and unemployment-risk dispersion across jobs, borrowing constraints, and precautionary saving. The model isn't calibrated directly to 1.63, yet it produces an average risk price of 1.53. In the model, complete-markets or risk-neutral counterfactuals raise employer-to-employer mobility by about 12% and labor productivity by 0.19%. This isn't a prediction that real-world reforms would necessarily produce these numbers—it's an estimate, within the authors' model setup, of how much misallocation the lack of insurance causes.

Unemployment insurance in the model is also more complex than "money after job loss." With stronger protection, employed workers are more willing to take high-productivity but less stable jobs, and firms' vacancies are more easily accepted by the employed. Conversely, sharply cutting protection lowers labor productivity in the model by about 1.29%—but unemployment also falls: when people take fewer risks, both job switching and unemployment decline. This is a good reminder that low unemployment, low mobility, and high productivity don't always point in the same direction.

Where to push back

First, the headline 1.63 comes from hypothetical offers, not from respondents' behavior after receiving real job offers. The survey frames new jobs as "identical to the current job except for unemployment risk," and first uses reservation wages for same-risk jobs to strip out switching costs as much as possible. In reality, location, commute, colleagues, career path, probation periods, benefits, and family arrangements can't all stay constant. Stable survey responses prove people understood the trade-off—they don't mean real choices will follow that slope precisely.

Second, respondents have to judge their probability of unemployment over the next 12 months, which is inherently hard. The 2%-vs-5% difference is statistically clear, but individuals may not perceive it that finely; the 50% high-risk scenario is also far from most normal jobs. The authors ran multiple retests and robustness checks, but scale inflation and measurement error in subjective probabilities can't fully disappear.

Third, between the survey facts and the macro-policy conclusions sits a calibrated model. The 12% mobility increase and the 0.19% and 1.29% productivity changes all depend on assumptions about search technology, the job risk distribution, wage bargaining, firm entry, and asset markets. The paper's "excessive caution" is relative to benchmarks of output maximization, approximate risk neutrality, or full insurance—it doesn't mean that, from an overall welfare perspective, every rejection of a risky job is a mistake.

Fourth, both the survey and the calibration target the United States. The authors themselves note that the U.S. already has a labor market with short unemployment spells and high employer-to-employer mobility, and cross-country extension remains future work. Transporting the results directly to China runs into different constraints: probation periods, social insurance continuity, housing and child-related spending, age cutoffs, industry hiring cycles. The mechanism can travel; the coefficients can't.

Fifth, the unemployment insurance experiment deliberately isolates the insurance channel. The temporary benefit increases in recession scenarios don't simultaneously impose full financing feedback through taxes, and the wage-bargaining rules rule out some direct wage effects. So "stronger protection promotes hiring and productivity" should be read as a new possibility emerging within this mechanism and these parameters—not an unconditional policy conclusion.

What this connects to in recent discussions

This paper supplies the second half of the sentence "every choice has a price": whether you can choose also depends on who can afford the price. Talking only about upgrading your mindset tends to hide cash buffers, family responsibilities, and benefit continuity. Talking only about structural constraints erases what individuals can prepare in advance. The more practical approach is to calculate the two separately.

For career decisions, I'd turn "how much of a raise makes it worth leaving" into at least four line items: the income gain at the new job, how much the failure probability increases, how long it would take to restore income after failure, and the non-negotiable monthly expenses during the recovery period. Cash reserves aren't idle money—they're an option that lets you accept high-return, unstable opportunities. Without that buffer, turning down an opportunity can be rational; if you want more options, first lower your cost of surviving failure.

This framework also constrains "networking upward" and fighting for fair rewards. Titles, valuations, or verbal promises aren't portable gains. You should also look at the probability of the position ending, equity vesting conditions, skill transferability, and the search cycle for the next job. Writing the upside as huge while treating the gap period after failure as zero is the most common form of self-deception.

For solo companies, unemployment risk hasn't disappeared—it just changed its name. Client concentration, platform bans, payment delays, health interruptions, and demand swings can all zero out income suddenly. AI can lower the fixed cost of getting tasks done, but it doesn't automatically provide insurance. To judge whether a solo company is viable, look beyond what tools can do: how many months you can survive, what share of revenue comes from a single client, how long income recovery would take, and which obligations can't be paused.

This also explains why "more conservative" isn't necessarily "safer." In a recession, workers in the paper's model trade wages for safety—the share of job switchers accepting pay cuts for stability rises from 7% to 14%. In the short run, risk falls; in the long run, people may stay stuck in low-productivity, low-growth jobs. What's actually worth optimizing isn't blindly taking risks or blindly playing it safe—it's making failure more recoverable, so that one choice doesn't wipe out all the choices that follow.

How to read it

Start with the abstract, introduction, and conclusion. Grab the one-liner: the career ladder has two dimensions—wages and unemployment risk. Then read the survey instrument in Section 2.2, Figure 2, and Table 1 to confirm how 1.63 is estimated from five hypothetical risk levels and reservation wages, and why people with different cash constraints differ.

Don't start the model section from the equations. First look at the three counterfactuals in Sections 5–7: complete markets, sharply cutting unemployment insurance, and the "climb to safety" in recessions. Every time you see a percentage, tag it as a survey fact, a real-data correlation, or a model result—the three can't be swapped.

Finally, stress-test one of your own real opportunities: if the new job ends within three months, how long can your cash last; what happens to social insurance, healthcare, and fixed household expenses; how quickly could you return to similar income; and can the skills, portfolio, and contacts from this failure be reused. The best conclusion after reading isn't "should I jump" or "should I stay"—it's knowing which kind of risk-bearing capacity you need to build first.

Official research page: Clymo et al.: Cautious Careers: Job Mobility Under Incomplete Markets

Authors' public full text: August 2026 paper PDF