JoshHeaps.Net/JoshHeaps.Net/Content/Socialism/compliance.md

title, section, order, summary

title	section	order	summary
What the Rules Cost a Small Co-op	The Co-op	32	The model's checks, from audits to the lock-holder's reviews, are scaled to a co-op's size and done once a year in one package, with audits by lot to deter cheating, though even the lighter package costs a small co-op real money.

What the Rules Cost a Small Co-op

Every page of this model adds a check. Someone has to confirm that members were paid the going rate, that the job titles are honest, that the equipment isn't wearing out unreplaced, that nobody is being pushed out before they'd join. Each check makes sense on its own. Together, they could cost a small co-op more than it earns.

That matters for two reasons. Rules a small firm can't afford end up existing only on paper, or they quietly keep co-ops out of small, low-margin work, which is where the model most wants them. This page says what's checked, by whom, how often, and who pays.

What gets checked

Across the model, a locked co-op is asked to do these things every year:

Check	Where it comes from
Year-end accounts, closed and checked by an outside accountant	Capital account payouts, savings cover pricing
Every worker paid at least the going rate for their job, by occupation code	The pay rule, Is the profit share pay?
A sample of timesheets matched to payroll records	What counts as an hour
Direct care job titles checked, where Medicaid's 80% rule applies	When Medicaid decides how much goes to pay
Overtime recalculated once the profit share is set	Overtime and the profit share
Pay ratio and pay bands, by job, gender and race	Who Sets the Pay
Warnings, hearings and dismissals, plus an anonymous workload survey	When a member won't do the work
Hires, and people let go before they could join	No pushing people out
Spending on buildings and equipment next to how much they wore out	Running it down
Big expansion proposals and how members voted	When Growing Means Sharing
A statement to each member of their account	Member Capital Accounts
Contracts with insiders, priced like deals with strangers	The hidden boss
Members' share of all work hours, counting agency and contractor hours	The hidden boss

Most of this is data a co-op already has in its payroll and bookkeeping. The expensive part is having an outside accountant check it.

What outside checking costs

Outside accountants offer three levels of checking. A compilation just puts the books into proper form, with no checking at all. A review asks questions and looks for numbers that don't make sense, without testing the records underneath. An audit tests the records and gives a formal opinion. There's also a fourth kind of job, agreed-upon procedures, where the accountant runs a fixed list of tests and reports what they found, without giving an opinion on the accounts as a whole. It's billed on top of a review, not inside it.

For very small firms, the gap between a review and an audit is large. When the SEC wrote its rules for small companies raising money online, it estimated a review at about $1,500 to $18,000 and an audit at about $2,500 to $30,000. The average audit for companies with under $1 million in both market value and revenue was about $20,000 in 2014 (SEC, 2015). By 2020, commenters and industry sources the SEC cited put a review of a small company at roughly $1,500 to $6,000, and one said at least $6,000 for a company with two years of history (SEC, 2021).

Take a cleaning co-op with six members and $400,000 in sales. A yearly audit at $20,000 could be most of its profit. The model's own rough estimate for the package below is more like this, per year:

Item	Rough cost
Review of the accounts	$3,000 to $6,000
The fixed list of extra tests	$1,500 to $3,000 (the model's guess; no good public data)
Lock-holder's review, every two years	$500 to $1,000 a year (the model's guess)
Staff time	20 to 40 hours

That's roughly $5,000 to $10,000, or 1.25% to 2.5% of sales. In cleaning, where profit might be 5% of sales, that's a quarter to half of the profit. The package is cheaper than yearly audits, but it isn't cheap, and the page doesn't pretend otherwise.

How others do it

Italy has checked every co-op for decades. The law requires a "cooperative revision" at least once every two years, or every year where a special law says so. If the co-op belongs to one of the national federations, the federation does it; if not, the government does. The revision has two jobs: to advise the board on running the co-op and on internal democracy, and to confirm the co-op is real. It checks that the members are real, that they take part, and that the co-op qualifies for its tax breaks, and it checks that worker-members are treated as the co-op's own rules say. A co-op that fails can be struck off the co-op register, put under a government-appointed manager, or dissolved (Legislative Decree 220/2002, articles 2, 4 and 12).

It's paid for by a fee on a published scale. For 2025–26, the smallest co-ops pay €330 for the two years, but only if they have at most 100 members, at most €5,160 in share capital and at most €75,000 in sales. Going over any one of those puts a co-op in a higher band. The top band, for co-ops with over €2 million in sales, is €2,740 (Ministry decree of 12 February 2025, as reported by Fisco7).

Italy is also a warning. Cheap checking by the co-ops' own federations has lived alongside a long-running problem of "false co-ops," firms that use the co-op form to underpay workers. In the first three months of 2016, labor inspectors who went looking for them inspected 934 co-ops and found more than half breaking the rules, with 3,768 workers affected (Italian Ministry of Labour, 2016). Those were targeted inspections, not a random sample, so they don't show how common false co-ops are. They do show that a federation checking its own members isn't enough on its own.

Mondragon's bank asked every co-op for monthly financial reports, and gave each a full audit every four years (Has This Been Tried?).

The rules

Checking is scaled to size

Three sizes. Size is measured by sales and by full-time-equivalent workers, counting hours worked through agencies and contractors, so taking on contractors or part-timers doesn't keep a co-op small. Co-ops that are related, in the sense the pilot already uses (a shared founder, manager, landlord, franchisor or major supplier), are counted together, so splitting a firm in two doesn't either. A co-op moves up a size when it passes either line two years running.

• Small: under $2 million in sales and under 25 full-time equivalents. A yearly review by an outside accountant, plus the fixed list of extra tests (below). No scheduled audit.
• Middle: $2 million to $10 million, or 25 to 100 full-time equivalents, or any co-op owing the bank more than $750,000. A yearly review with the extra tests, and a full audit every other year.
• Large: over $10 million, or over 100 full-time equivalents. A full audit every year.

Every co-op also sends the bank its quarterly payroll tax filings, which it already has, so the bank sees hours and pay four times a year at almost no cost.

What a review can and can't stand in for

Wherever the model says "the co-op's auditor," it means whichever outside accountant does that year's review or audit. But a review isn't an audit, and a few rules need the real thing:

• Capital account payouts can go ahead on reviewed accounts. If a later audit finds profit was overstated, the extra credits are taken back out of members' unpaid credits, the same way losses are shared.
• Savings cover starts only after the pool itself audits the co-op, at the pool's cost. After that, premiums can be priced on reviewed accounts, and the pool can audit any covered co-op at any time (Protecting Members' Savings).
• Spreading the cost of a big project, under the long-run rule, still needs an independent audit. A review doesn't count. A small co-op that wants to use the rule has to pay for one.

One package, done once a year

• The extra tests are the same everywhere. The accountant runs a fixed list published by the bank and the federations before the first loan:
  • every contract with an insider (a founder, manager, member's family, or a company they own), checked against what strangers would pay, the main way a hidden boss gets paid;
  • members' share of all work hours, checked against payroll and against agency and contractor invoices;
  • every worker's pay against the going rate for their occupation code, and, in co-ops under 25 full-time equivalents, every worker's code, with a sample in larger ones;
  • a sample of timesheets against payroll;
  • job titles where the Medicaid rule applies;
  • the overtime recalculation.
• One problem widens the check. If a sample turns up anyone underpaid, the test is widened to every worker. Everyone underpaid, member or not, gets the difference with interest, and the co-op gets a full audit.
• The list can't quietly grow. Adding a test to the package requires publishing what it will cost co-ops, and the whole package's cost is reported every year.
• The data comes from payroll. Pay bands, the pay ratio, hires and departures, and members' statements come out of the standard payroll service the model already uses for overtime (Owners Are Still Workers). The co-op doesn't build them by hand.
• One report. Every yearly report on the list above goes on one standard form, filed once, with the lock-holder, the bank and the members. None of the model's own bodies asks a co-op for the same number twice. Government agencies like Medicaid or the IRS will still ask for their own.
• Small groups are protected from the public, not from the members. Where a published figure would reveal one person's pay or discipline, the public sees only the total. Members still see everything the pay rule gives them, including managers' actual pay. The accountant and the lock-holder see the figures person by person, and a gap in pay or discipline between groups is a flag for the lock-holder's review.
• The federation does the templates. Standard bookkeeping, the report form, the occupation codes and the benchmark for the going rate are supplied centrally, so a small co-op isn't paying an accountant to invent them. Clean books also make the review cheaper.

The accountant isn't the co-op's to pick freely

• From an approved list. The co-op chooses its accountant from a list the bank approves, and changes firms at least every seven years.
• The lock-holder picks the samples, so the co-op can't steer which timesheets or pay records get tested.
• The report goes straight to the lock-holder and the bank, as well as to the members.
• Changing accountants is a warning sign. If an accountant resigns or is replaced, the outgoing one writes to the bank saying why, and the co-op gets a full audit.

Audits by lot, and when something looks wrong

Scaling checks down leaves more room to cheat in small co-ops. Three things close part of that gap:

• One in twenty, by lot, after year-end. Each year, after the books close, the pilot's outside reviewers draw one small co-op in twenty for a full audit, by a method published in advance, the same way they run the support lottery. At least half the draw is purely random. The rest is weighted toward co-ops with warning flags: heavy use of agency or contractor hours, a lot of insider contracts, or high turnover. The auditor is assigned, not chosen. What the audits find is published every year. Once a co-op has repaid the bank, the lock-holders run a narrower draw instead, of insider deals only (Deals with insiders).
• Warning signs trigger an audit. A small co-op gets a full audit if:
  • its accounts are late, or its accountant resigns, is replaced or qualifies its report;
  • the extra tests find a problem;
  • it loses money two years running;
  • equipment spending stays below wear long enough to call in the inspector;
  • payments to insiders grow faster than sales;
  • its payroll tax filings don't match its accounts;
  • members' share of work hours falls toward the minimum, or many hires are let go before they could join;
  • it ends a year below Medicaid's 80% pay ratio, where that rule applies;
  • a labor or anti-discrimination agency makes a finding against it;
  • it falls behind on the bank's loan;
  • or before any sale of a large part of the business under the lock.
• Anyone who works there can complain. Members, non-members and anyone who left in the last two years can complain anonymously, in their own language, to the lock-holder or straight to the bank. They're protected by the same rule that protects members who report problems (When a member won't do the work). An upheld complaint leads to a targeted check of what it's about, not a full audit, so it can't be used just to run up a co-op's costs. Either side can appeal the lock-holder's decision to a different approved lock-holder.

The bank pays for audits by lot and for triggered audits, as part of its cost of lending, counted in its results like any lender's monitoring. When an audit finds fraud, the co-op pays it back.

The lock-holder's review, every two years

• Like Italy's revision, with more workers in it. At least once every two years, the lock-holder checks that the members are real and do most of the work, that the membership rules and the insider-deal rules are followed, and that working conditions and discipline follow the co-op's own rules. It talks privately with a random group of workers, members and non-members, including people who left in the last two years, with an interpreter where needed. Its report goes to the members' meeting and the bank, not just the board.
• Advice is separate and optional. A co-op can buy advice from its lock-holder or federation, but the person who checks a co-op can't be its paid adviser in the same years.
• A fee set from the real cost. Lock-holders charge on a scale set by size, but the scale is approved by the bank from what a real review costs, not set by the lock-holders themselves. A fee too low to pay for a real review would just buy a weak one. Where the federation holds the lock, the fee is part of its dues and isn't charged twice.
• Switching lock-holders is allowed, but not to escape a check. A co-op can move to another approved lock-holder by a members' vote, but not while an audit, an unresolved finding or a vote the lock-holder can veto is open, and not in the 12 months before a review is due. The old lock-holder's file goes to the new one, and the review clock doesn't restart.

What happens when a co-op fails a check

In steps, and each step can be appealed. A finding comes with a deadline to fix it. If it isn't fixed, payouts to board members and managers stop, new savings cover is suspended, and the bank can treat it as a breach of the loan. A co-op that refuses to be reviewed or audited is in breach of the loan and of the lock, and the lock-holder can go to court. None of this gives the lock-holder power to run the co-op.

Who pays at the start

The bank's first loan can include the first three years of checking costs as working capital, so a new co-op doesn't have to find them in its first thin years. The pilot's business-help budget pays only for the shared pieces: the templates, the report form and the payroll service's setup.

The cost is measured, and it can move the rules

Before the first loan, the bank and the federations publish what they expect the whole package to cost a co-op of each size. The pilot then records what it actually costs each co-op, as a share of its sales and of its profit, and publishes it by size and industry, along with what the audits by lot find.

The goal is under 1% of sales for small co-ops. The estimate above already misses it. If the typical small co-op spends more than 1% of sales for two years running, and the audits by lot are finding little, the lightest-value checks are cut back in an order published in advance: the group pay figures to every other year first, then the timesheet sample, then the review itself to every other year, with a compilation in between. If the audits are finding a lot, the checks stay, and the report says what honest checking costs in low-margin work.

What this doesn't solve

• Less checking means more room to cheat. A reviewed small co-op is checked less deeply than an audited one. Audits by lot, warning signs and complaints make cheating risky, not impossible. Italy's false co-ops are the warning.
• It costs real money in the work that can least afford it. At 1% to 2.5% of sales, the package could take a large part of a cleaning or home care co-op's profit. The model has no way to make honest checking free, only cheaper.
• The cost numbers are rough. Two of the four lines in the estimate are guesses. There's no good US data on what small firms spend on outside accounting as a share of sales. The pilot's own numbers will be the first.
• Checkers can be captured. The co-op still pays its accountant, and lock-holders depend on the co-ops they check. The approved list, rotation, samples picked by the lock-holder, draws by outside reviewers and the limits on switching are meant to limit that. Italy has the same tension.
• Fees don't cover staff time. Filling in a form and answering an accountant's questions takes hours in a firm with no office staff. Payroll data and templates cut that down. They don't remove it.