Freelance or salaried, for someone with no track record yet

A graduate holding a $60,000 offer and thinking about going independent instead usually compares $60,000 with $60,000. That comparison is wrong before it starts. In March 2026 the US Bureau of Labor Statistics measured private-industry compensation at $46.60 per hour worked, of which wages and salaries were $32.60 and benefit costs $14.01 — benefits were 30.1% of what the employer spent. The figure on your offer letter is about seventy cents of every dollar it costs to employ you. The other thirty cents buys a list of things that, as a freelancer, you buy yourself or go without.

That is the first half of the problem, and it is arithmetic. The second half is not. Freelance rates price evidence of prior delivery, and a graduate has none, which means your first-year rate is not the rate the market pays for your discipline. It is the rate the market pays for an unknown. Most advice on this decision treats the ramp as a few months. The evidence suggests it is longer than that, and the part of the market that used to absorb beginners is the part that repriced hardest between 2023 and 2026.

The salary on the letter is about seventy percent of the deal

The BLS series that matters here is Employer Costs for Employee Compensation, which measures what employment costs per hour actually worked rather than per hour paid. For private industry in March 2026 the composition was paid leave at 7.6% of total compensation, supplemental pay 4.1%, insurance 7.8% (health insurance alone 7.3%), retirement and savings 3.4%, and legally required benefits 7.2%.

Not all of that transfers to you. Supplemental pay is overtime and bonuses, which you either replicate by working more or don't. Legally required benefits include the employer's half of Social Security and Medicare, which you do replicate, plus unemployment insurance and workers' compensation, which you don't — and correspondingly you have no unemployment claim when work dries up. Paid leave is the one people double-count. It is easier to handle on the hours side, by working fewer weeks, than as a cash line, so leave it out of the money column and take it out of the calendar instead.

What genuinely converts into cash you must find is health insurance, retirement, and the employer half of payroll tax.

What sixty thousand dollars costs to earn as revenue

Take a $60,000 offer as an illustration. This is a worked example, not observed data; substitute your own figures at every line.

An employer paying $60,000 in wages is spending, at the ECEC ratio, $60,000 × 46.60 ÷ 32.60 = $85,767 in total compensation. Health insurance at 7.3% of that is $6,261. Retirement and savings at 3.4% is $2,916. So $69,177 is the cash you need before touching tax.

Then payroll tax. An employee pays 7.65% and the employer pays another 7.65%. Self-employed, you pay 15.3% on 92.35% of net profit, so the extra half you have absorbed is 7.65% × 92.35% = 7.065% of profit. Required profit P therefore satisfies P × (1 − 0.07065) = $69,177, giving P = $74,438.

Add the cost of being a business at all. Call it $4,800 a year for software, accounting, professional insurance, and hardware you amortize yourself. That is an illustrative figure and it is the one line you have real control over: the freelancer-operations software category clusters at $12–$29 a month for a solo tier, and consolidating three subscriptions into one is a genuine cut rather than a mood improvement — the comparison of what an all-in-one like Bonsai covers against a workspace that also runs the money is the right way to look at that line.

Hold the two ends of that table next to each other. The employer's cost of the salaried version was $85,767 across 1,880 hours worked, or $45.62 per hour — almost exactly the $46.60 national average, which is a useful sanity check on the illustration. The freelance version of the same outcome requires selling at $84. The entire gap is non-billable time and the benefits you now buy at retail.

The billable share is where most of that gap lives, and it is the number graduates guess at most wildly. Fifty percent in year one is not pessimistic. Scoping calls, proposals, revisions you didn't quote for, invoicing, chasing, bookkeeping, and finding the next project are all real hours and none of them are billed. If you want to run your own version, a converter between an hourly rate, a day rate and an annual salary does the direction you need, and the rate calculator that takes desired income, billable hours, weeks off and a tax set-aside does the rest.

Your rate prices a track record, and you don't have one

Everything above assumes you can get $84 an hour. That is the assumption a graduate should interrogate hardest, because freelance pricing is a reputation market and reputation is the input you are missing.

The clearest evidence of how that works comes from a study of Freelancer.com covering more than 1.8 million job posts and 3.8 million users between January 2022 and April 2024, which found that only 6.03% of active users did any generative-AI work at all, and those who did were disproportionately high-reputation accounts. The best-priced new category on the platform was not open to everyone; it was captured by people who already had ratings. That is the general shape of the market, not a quirk of one site.

Reputation is not permanent protection either. A 2023 study of freelancers in occupations heavily exposed to language models found reductions in both employment and earnings, and that top freelancers were disproportionately affected — past performance did not insulate them. The paper's public abstract reports no effect sizes, so no percentage should be attached to it. The useful reading is narrower: a track record gets you considered, it does not guarantee the price holds.

Meanwhile the buyer side is consolidating. Fiverr's second quarter of 2026 showed annual active buyers down 21.9% year over year to 2.7 million while annual spend per buyer rose 15.6% to $368. Fewer buyers, each spending more, is a market that screens harder. Screening harder is bad news specifically for the applicant with nothing to screen.

The rung you were going to start on is the one that automated

There is a version of the freelance-first argument that says: take cheap work, build a portfolio, raise rates. It described a real path. The research on what happened to that path is now reasonably consistent.

An analysis of a large freelance platform found that in the eight months after ChatGPT's release, writing and coding job posts fell 21% relative to manual-intensive work, and image-creation posts fell 17% after image models launched. The posts that survived were more complex, better paid, and more competitively bid. Upwork's own 2026 index, from a survey of 2,400 US skilled workers fielded in March and April 2026 with a 2% margin of error, reports that generative AI and creative production saw contract starts rise 90% year over year while earnings per contract fell 13%. More work, worth less each. That is the price signature of commoditization, and it sits precisely on the entry ramp.

The same sources show where the money went instead. Complex AI-augmented work earned 45% more, and the Freelancer.com study found that software-integration tasks commanded 3.82× higher budgets than content-generation tasks. None of that is entry-level. It is judgment work, and judgment is the thing employment is unusually good at teaching you cheaply, because someone else carries the cost of your mistakes.

This is not an argument that the salaried route is easy. The Federal Reserve Bank of New York's tracking of recent college graduates put their unemployment rate at about 5.7% in the first quarter of 2026, with underemployment at 41.5%. Roughly two in five recent graduates in work are in jobs that don't require the degree. Both doors are heavier than they were. The freelance door is heavier in a specific way that the arithmetic above makes visible.

Outside the US the shape is the same and the numbers are not

The gap between salary and total compensation exists everywhere; its composition changes.

In the UK for 2026/27 an employer pays secondary National Insurance at 15% on earnings above a £5,000 secondary threshold and, under automatic enrollment, a minimum 3% of qualifying earnings between £6,240 and £50,270. On a £30,000 graduate salary that is 15% × £25,000 = £3,750 of employer NI and 3% × £23,760 = £713 of pension, so about £34,463 of employer cost against a £30,000 headline. Smaller proportional gap than the US, mainly because health cover is not sitting inside it.

Leave is the sharper difference. UK statutory entitlement is 5.6 weeks of paid holiday, 28 days for a five-day week, and Article 7 of the EU Working Time Directive requires at least four weeks of paid annual leave in every member state. In the US, the Fair Labor Standards Act does not require payment for time not worked, such as vacations, sick leave or holidays. So a European graduate comparing the two options is giving up a legally guaranteed five or six paid weeks, while an American graduate is giving up whatever the employer chose to offer. The five unpaid weeks in the table above are a policy artifact as much as a lifestyle choice, and in Europe the number to model is higher.

One more European data point worth knowing before you assume everyone your age is doing this: Eurostat counted 2.06 million self-employed people aged 20 to 29 in the EU in 2025, 7.9% of all self-employed aged 20 to 64. Young independents are a small share of independents. That is not a reason not to be one. It is a reason to distrust the impression that the salaried route is the unusual choice.

The threshold at which the answer flips

Run your own version of the table, then apply this rule.

Freelance first if you can name at least two buyers who have already paid you for this specific work (paid, not praised) and the rate they paid, multiplied by the billable hours you can honestly sell in a week, clears the revenue line in your table. In the illustration that meant $79,238 across 940 billable hours. If your realistic number is $45 an hour and 15 sellable hours a week, that is $45 × 15 × 47 = $31,725, which is not a slow start, it is a different income entirely, and knowing that in advance is worth more than optimism.

Take the salary if either input is missing. It is the cheaper way to buy the two things your rate is actually priced on: evidence of delivered work and judgment about which work is worth doing. Then treat the first eighteen months as a paid ramp, keep a small amount of independent work running alongside it so the evidence accumulates in your own name, and revisit the table when your billable share would plausibly start above 50% rather than below it. The complete guide to going freelance organized by money milestones sets out what each of those milestones looks like, and if writing is your discipline specifically, starting as a freelance writer in a market that just repriced is the honest version of that entry route. What independence does to a career over a longer horizon, with the debit column shown as well as the credit one, is in an honest ledger of going freelance.

The one thing to start measuring from day one either way is your billable share, because it is the single largest term in the arithmetic and the one nobody can supply for you. Worklyn's time tracking marks hours billable by default and carries them into invoices, so the share you are currently guessing at becomes a number you can check against the table before you commit to it.