To know what replacing a worker really cost you, add up seven items, not just the two you wrote checks for. The exit paperwork and the job ad are the smaller part. The two biggest items are the days the job sat empty and the months the new hire took to work at full speed, and neither shows up in your books as part of the replacement. In the calculator’s example, a replacement that cost $17,800 comes to $1,483 a month for a year: what you could have paid that person extra and still broken even.
At the end of this article is a one-page replacement record you can fill in for the last person who left, using the payroll, the exit file and a calendar.
The most detailed measurement cited here comes from a hospital, not a farm. Researchers at a Brazilian teaching hospital priced every step of replacing nursing staff, to the cent, over four months. Three items made up 96 percent of the cost: the new hires’ slow months, the days the jobs sat empty, and the drop in output of people about to leave. None of the three goes into the books as part of a replacement. A hospital is not a farm, but in both the work has to be done every day, whether or not someone quit.
On a farm the shares change and the order does not. The hiring and turnover cost calculator comes with an example: an operator leaves partway through the year from a job that costs $4,000 a month in wages, payroll taxes and benefits. The exit paperwork, the ad, the interviews, the hiring paperwork with the medical exam, and the training come to $4,600, all paid with a receipt. The 24 days the job sat empty, covered at $250 a day, cost $6,000. The three months in which the new hire did on average 40 percent of the job cost $7,200. The whole replacement comes to $17,800, and on most farms the $4,600 is the only part anyone adds up.
Why does the most expensive part never get counted?
Because nothing in the farm’s books moves when it happens. While the job sits empty for 24 days, payroll goes down, not up. When the new hire starts at 40 percent of the old output, payroll is the same as the month before. No invoice arrives and no number jumps, so it looks as though nothing was spent.
Kate Bahn and a colleague, writing for the Washington Center for Equitable Growth, a US economic research group, reviewed 31 US case studies. They open with the same point: “many businesses do not know or underestimate the toll that high turnover has on their workforce, their sales, and their bottom lines”.
Zaballa and colleagues, in a published method for analyzing staff turnover, name the three places the money goes. Only the first is a payment: hiring the replacement. The other two are time, the stretch in which the post is vacant and produces nothing, and then the stretch in which the new person is not yet productive. On a farm that time is paid for anyway, in overtime for whoever covers the empty job and in a full wage for a new hire who does less than a full job.
The cost also starts before the notice. Gregorio Billikopf and a colleague studied worker turnover on California dairies, in work published by the University of California. They found that the stretch before a departure usually comes with lower output and more absence, and that the absence is sometimes physical, when the person does not show up, and sometimes mental, when they show up with their mind somewhere else, which they call having quit and stayed. In the hospital study, that drop before leaving was 3.1 percent of the cost, measured over the 30 days before each person left. It is small, but it is the only item you can see happening while the person is still there to talk to.
How did the hospital study price each item?
The hospital study, by Ruiz and two colleagues, appeared in the nursing journal of the University of São Paulo. They followed every step of replacing nursing staff at a teaching hospital for four months and put a price on each one.
They split the cost into seven items, three before the new person is hired and four after. Before the hire come recruiting, the empty jobs and the hiring itself. After it come orientation and training, the new hires’ slow months, the slower last weeks of the people leaving, and the exit itself.
Most of the cost fell in two places. Before the hire, “the pre-hire related costs totaled 32.1% of the expenses generated in turnover, in which the sub-process vacancies (overtime and temporary contract) was the most present, with 29.5%”. After it, “the lower productivity of the newly hired professional was the most expensive, with 63.6%”. Laid out as seven items, from largest to smallest, the cost looks like this.
| Item | Share of the total | In the books as part of the replacement? |
|---|---|---|
| The new hires working below full speed | 63.6% | no, nothing is recorded |
| The days the jobs sat empty | 29.5% | no, it goes in as overtime and temporary contracts |
| The drop in output before someone left | 3.1% | no, nothing is recorded |
| Selection and hiring | 2.1% | yes |
| Orientation and training | 1.1% | yes |
| Recruiting and advertising | 0.5% | yes |
| The exit (only the exit interview was measured) | 0.2% | yes |
The three items that never go into the books as part of a replacement add up to 96 percent. The four that do add up to less than 4 percent.
The empty jobs are the one item of those three that you can find in your books, under another name. The authors split that 29.5 percent in two: 85.1 percent of it went out as overtime for the people who stayed, and 14.9 percent as temporary contracts. Both are on paper, as overtime on the payroll or as a contractor’s invoice, and neither is filed as the cost of replacing someone.
So on a farm the money for the empty job is already in your books; what is missing is the link to the person who left. It went in as March overtime, not as the cost of replacing the tractor driver who left in February. After someone leaves, take the overtime for the month after the exit, set it beside a month when nobody left, and put the difference down as part of that replacement.
Where does the hospital’s 96 percent not apply to a farm?
On the exit side. The authors list what they did not measure. In hiring, the activities around the medical examination were never investigated. On the exit side, they left out the “activities related to registration, documentation, attendance at the approval in the Labor Courts”, the paperwork and legal steps of ending a job in Brazil. The only exit cost they measured was the exit interview. Where leaving costs real money, in severance, a final settlement or legal paperwork, the cash side on a farm is larger than that 0.2 percent.
Even with those costs left out, the authors conclude that the cost per person who left “represented three times the average salary of the nursing staff”. That figure is the whole cost of the four months divided by the number of people who left, not the seven items added up for one person.
The calculator’s seven items are not quite the hospital’s seven. The calculator has a line for the hiring medical exam, which the hospital study did not measure. It has no line for the drop in output before someone leaves, 3.1 percent in the hospital, because almost no farm can date the month that drop began. That makes the calculator’s total a floor, the least the replacement cost. If you do find that drop in your own records, which is step six of the replacement record at the end, write it next to the total instead of adding it in.
So take from the hospital the order of the items and the reason for it, not the percentages. The empty job and the new hire’s slow months weigh the most because a ward has to be staffed at three in the morning whether or not anyone quit, and farm work cannot wait either. In the calculator example, with a farm’s exit paperwork, the cash side rises to about a quarter of the cost, against less than 4 percent in the hospital. The order stays the same: the new hire’s slow months and the empty job are still the two biggest items.
Why does the empty job start costing before the hiring starts?
Most people think of the empty job as the result of hiring slowly. It is the other way round: the job goes empty the day someone leaves, before any hiring has begun. Duda and Žůrková, in a study published by a Czech agricultural university, state the rule that follows: “there is a rule that the longer the period for which the leaving employee is not replaced, the higher the costs associated with and resulting from the employee turnover”. They also price the two ways of covering the job, overtime for the people who stayed or a temporary worker, each with its own calculation.
On a farm the rule weighs more than in an office, because the work cannot wait. The California researchers put it plainly: “Dairy production cannot be downsized temporarily in response to labor shortages as easily as in retail or manufacturing”, so “When a worker quits, a substitute must be found to do the work until the worker is replaced”. Cows get milked twice a day whether or not the milker quit, and a spraying window closes on the crop’s schedule, not yours.
That is also why the same 24 empty days cost different amounts depending on when they fall. In a quiet month they are absorbed by moving people around. Inside a picking window they are not, and what they cost is not the wage you saved but the fruit left on the tree.
What is the learning period, and how do you avoid counting it twice?
The learning period is the months it takes the new hire to do as much as the person who left. Duda and Žůrková point out that lost output shows up in two separate stretches, since “it is necessary to involve costs of lost productivity twice (once because of the temporary substitution and once due to the entering of a new employee)”. Both losses are real. The mistake is counting the same days in both. The dividing line is the day the replacement actually started work: before it, the cost belongs to the empty job; after it, to the learning period.
How big the loss is depends on how you covered the job. In a table, Duda and Žůrková compare three ways of covering it over the first three months: Covered by overtime from people who already know the job, the loss is 25 percent in the first month and zero in the next two; covered by a temporary hand, or by the new replacement, it runs 75, 50 and 25 percent; with the post standing empty, it is 100 percent in all three.
Add those up and covering with people who already know the job loses a quarter of a month of work, while covering with new people loses a month and a half, six times as much. What separates the two is a decision made in the first week after someone leaves, when nobody is thinking about cost yet.
The second trap is easier to fall into and costs more: the hours of whoever does the teaching. In the hospital study, the new hires’ slow months had two parts, and the bigger one was not the new hires. It was the supervisors: 64.4 percent of that line was the time of the person doing the orienting, against 35.6 percent for the work the new hand did not deliver. On a farm that is your best tractor driver standing beside a new hand while the tractor sits parked.
If you already counted those hours as training, do not count them again in the learning period. If you counted them nowhere, the cost is missing the bigger part of that item. A new hire also does not improve in a straight line, and almost no farm measures its own output month by month closely enough to draw one. A rough average, written down and dated, is worth more than a precise figure nobody believes.
What is an hour of lost work worth?
It depends on the price you put on the hour, and that choice changes the total more than any other. You can price the hour at what the worker costs you, or at what the worker brings in. Duda and Žůrková use the second: “The average labour output of the employee (OE) is calculated as the share of the added value created by the company within the given fiscal period (AV) per one employee”. In plain words, that is the value the business creates in a year divided by its number of workers. A farm that makes money gets more out of an hour than it pays for it, so this is the higher of the two prices.
In the calculator the hour is priced at the lower figure, what the job costs per month, so its total is a floor. Duda and Žůrková also count two items that are not in the calculator. One is the loss to everyone else: Duda and Žůrková multiply the output of the whole team by a coefficient of dependence, which measures how many people lean on the leaver’s work, in both stretches, the empty job and the learning period. A milker who quits does not stop only his own milking.
The other is the know-how that leaves with the person, which they price at half a year of the post with charges, plus ten percent of that for each year of service. That is six months of what the job costs, wages plus payroll taxes and benefits, plus a tenth more for each year on the farm. These two extra items are why Duda and Žůrková end up at one and a half times a year’s pay, a figure that comes up in the next section: their calculation has more items in it, not bigger ones.
The know-how item also means that replacing someone with ten years on your farm costs more than replacing someone with ten months, even when both do the same work for the same pay.
How do you compare your total with published figures?
Published figures vary widely, and almost none come from farms. The Washington Center for Equitable Growth pooled 31 case studies and found that “On average, turnover costs represent 39.6 percent of a position’s annual wage”, while “the median cost of turnover represented 23.5 percent of a worker’s annual wage”. The whole range is wide: “The costs of turnover range from 2 percent to almost 150 percent and vary across industries”. Duda and Žůrková sit at the top of that range, with a departure costing one and a half times a year’s pay, or 50 percent “higher than the average annual wage of the leaving employee”.
Where farms fall in that range, there are only clues. The California researchers cite US labor statistics showing that “turnover costs in manufacturing and construction (the industries closest to agriculture) are close to 75% of the highest turnover costs in any industry”. In other words, replacing someone in the industries most like farming costs about three quarters of what it costs in the most expensive one.
Tracey Erickson, a former dairy specialist with South Dakota State University Extension, holds in a text for dairies, citing Moore (2012), that “We can estimate conservatively that hourly employees cost 100 to 150 percent of a position’s salary”. The extension text is not peer-reviewed research and the estimate is secondhand, so treat it as a rough idea of what to expect, not a figure to aim for.
Before comparing, check what each figure was divided by. Those published figures are shares of the annual wage. In the calculator the share is of the annual cost of the job, wage plus payroll taxes and benefits, which is a bigger number, so the same money comes out as a smaller share. The calculator example comes to 37.1 percent of the job’s annual cost ($17,800 against $48,000); divided by the wage alone, it would come out higher.
Then check what went into each figure. Duda and Žůrková say at the start that their work covers only the costs generated by somebody leaving, and does not deal with the cost of recruiting the replacement. So their one and a half times a year’s pay still leaves out the hiring side that the US review includes. The two do not disagree; they count different things. Before you set your total beside any published figure, check what was divided by and what was added up. One of the two almost never matches.
Keep or replace: which decision does this cost help you make?
Divide the replacement cost by twelve and you get a monthly figure. In the calculator example, $17,800 becomes $1,483 a month for a year: that is how much more you could have paid that person every month for twelve months and still come out even against what replacing them cost. Before you add up the cost, you have no such figure, so the amount you are willing to spend to keep someone is zero by default.
Rurivia calculation from the calculator example on this site: $17,800 divided by twelve months.
With the same arithmetic you can see where to spend. Dropping the job ad altogether would save half a percent of a cost like the hospital’s. Cutting the two heavy items means shortening the empty days and the learning period. Each has a concrete fix: a notice period the person actually works, a second person who already knows the job, a handover written down instead of shouted across the yard, and training planned on the calendar instead of improvised on the first wet morning.
None of this requires guessing what people want. The California researchers found that “The reasons why workers leave dairies are mostly the same as they were 30 and 60 years ago”. That is discouraging about people and encouraging about the list of reasons: it is short, it is known, and it has not changed.
In 2009 interviews with 209 San Joaquin Valley dairy workers, covering 222 earlier departures, the reason cited most often was compensation and benefits at 29 percent, followed by the employer’s economic trouble at 14, personal and family reasons at 11, working schedules and time off at 11, housing and transportation at 8, relations with management at 8, and the design of the job itself at 7.
Read that list with the monthly figure in mind. Schedules and time off, housing and transportation, and relations with management add up to 27 percent, nearly as much as pay, and none of the three calls for a raise. They take time on the calendar and conversation.
There is another side to this, and it is uncomfortable right after adding up $17,800: not every departure is a loss. The same researchers point out that a voluntary exit can be the chance to hire someone better qualified without having to fire anybody, and that it sometimes trades an old high wage for an entry-level one. So the monthly figure is not meant for whoever happens to be leaving. It is for the people you would have to replace with someone worse.
Before you use the number, write next to the name of the person who left whether you would hire them again today. If the answer is no, the cost is still real, but the decision is a different one: shorten the empty days and the learning period, and do not pay to hold on.
Then there are people who want to leave and have not, a wish that may or may not turn into a resignation. In the same method, written for organizations in general and not for farms, Zaballa and colleagues rate that hidden wish as more damaging to productivity and to the working atmosphere than the departures that do happen. The hospital counted only the drop in output of people who actually left, and the replacement record also catches only someone already gone. A person still thinking of leaving can stay, and if you would hire them again, that is who the monthly figure is for.
Where to start
Set aside an hour with the payroll, the file from the last exit and a calendar, and work on one replacement only: the last person who left. Use your own papers, not the calculator’s form.
The replacement record is worth more than the total on it, because the total is a floor and with the record you can see where the money went. Two farms with the same cost can have it for opposite reasons: one loses money on a job it never hurried to fill, the other on a learning period nobody shortened. What fixes one does nothing for the other. Comparing a decision with what happened afterwards, as you do with this record, is a habit the rest of farm management runs on too.
To count how often you pay this cost in a year, see the article on how much turnover costs your farm each year, where you multiply the departures in a year by the total on this record. Deciding, before anyone quits, how you will fill each job shortens the empty days; the article on a written hiring process is about writing that plan down. With a first day list, a signed page of what the new hire gets on day one, you shorten the learning period, because the new hire is taught instead of learning by watching. More on managing farm workers is in the people section of the library.