Here are two employees. One of them took a job in 1990 at fifty thousand dollars a year. The other took what is recognisably the same job this year, also at fifty thousand dollars a year.
Open the payroll record for each. Base salary: 50,000. Base salary: 50,000. Identical fields, identical values. Every system either of them will ever appear in agrees that these two people are paid the same.
Nobody believes that. You did not believe it while you were reading it. And yet there is no field anywhere in the apparatus that disagrees.
The system is not lying
It is worth being precise about what has gone wrong here, because the sloppy version of this observation is the reason serious people stop listening to it.
The payroll system is not making an error. It is answering the question it was built to answer, which is how many currency units this person receives. That is a real question with a correct answer, and the answer is fifty thousand.
The trouble is that almost nobody actually wants to know that. What an employee wants to know is what the money will do. What a compensation professional wants to know, when they are being honest about it, is whether the offer is good. Those are questions about purchasing capacity, and purchasing capacity requires a denominator that the system has no column for.
The nominal fallacy. The treatment of nominal value as economic value. It is not a mistake anyone makes on purpose. It is the default output of every instrument the profession uses.
Put the distinction to any experienced practitioner directly and they will get it right instantly. Real wages are a standard concept. Cost-of-living adjustments exist. Inflation assumptions sit inside every actuarial valuation. Nobody in this field is confused about the difference between nominal and real.
The claim is not that we do not know. The claim is that everything we have built produces the nominal answer automatically, and the real answer only if somebody goes looking for it. What gets measured by default is what gets managed, and what gets managed is the number that does not mean anything.
The arithmetic is not dramatic, which is the problem
The reason this does not get taken seriously is that inflation, in the period most working professionals have experienced, has not been frightening. It has mostly been the target, which is around two percent, and two percent does not feel like an emergency because it is not one.
Run it forward anyway.
At two percent, the purchasing capacity of a fixed sum halves in roughly thirty-five years. At three percent, which is still inside the band most central banks would describe as acceptable, it halves in about twenty-three.
A working life is forty years. A retirement can be thirty more. The horizon over which compensation has to hold its value is longer than the halving period of the target rate of inflation. That is not an accusation of policy failure. Policy is doing what it says it will do. It is an observation about what happens when a system designed for the short run gets asked to carry something for a very long time.
Two percent is not a crisis. It is the target. It is also enough to cut a person's work in half within a career.
What the second employee is actually being offered
Go back to the two records.
The 1990 hire, in a large enough employer, was likely being offered something more than the salary line. There was a reasonable chance of a defined benefit pension attached to the job, and if there was, the employer had made a promise denominated in outcomes rather than in dollars. Somebody else was being paid to worry about the forty-year problem on that employee's behalf.
The employee hired this year at the same nominal number is, in most cases, being offered a savings account with a tax wrapper and a menu of funds. That is a genuinely useful thing and I do not want to be glib about it. But it is a different kind of object. It hands back to the individual a set of decisions the pension used to make institutionally, without the pooling, without the professional stewardship, and without any promise about what arrives.
So the two records are identical in the only field the system has, and different in almost every respect that determines what the work will eventually be worth. The difference is not in the compensation. It is in the architecture around the compensation, and the architecture is not something we currently think of as compensation's business.
The column that does not exist
What would fix this is not complicated, and it is not a technology problem.
Every element of a compensation package could carry three additional attributes. The horizon it is intended to serve, which is to say when the value is supposed to be used. The unit it is denominated in. And whether it survives the employee leaving.
None of those are exotic measurements. All three are knowable today for every instrument any employer offers. Almost no employer records any of them, because the frameworks the profession inherited have no place to put them.
That is the whole observation. Not that our numbers are wrong, but that the fields we have describe the transaction and not the journey, and we have been treating the transaction as though it were the whole of the thing.
Two people earning fifty thousand dollars. One question the system can answer, and one it cannot even be asked.