A careless version of this argument would compare long-duration preservation with cash held inertly for forty years.
No modern worker preserves that way, and an argument built against that comparison would deserve to lose.
Compensation arrives as a bank deposit, and deposits may pay interest. The worker can hold savings accounts, term deposits, money-market instruments, treasury bills, short-duration government bonds, and inflation-indexed government securities. The defender of the current arrangement can reasonably say that money itself need not preserve purchasing power, because interest-bearing nominal claims compensate the saver for deferred consumption and expected inflation.
That objection deserves a full answer, and the answer requires distinguishing the instruments rather than treating money as one thing.
Walking Down the Spectrum
Base money, meaning currency and central-bank reserves, pays no interest. It is the pure case of the decay arithmetic, and it is also the case nobody actually relies on.
Bank deposits are claims on a bank whose safety rests on a combination of the bank's solvency, deposit insurance up to limits, and, behind the insurance, a sovereign or public guarantee. The second paper's property-and-promise logic would note the layering. They may pay interest at rates that vary with policy and competition, and transaction deposits have often paid below short-term market rates.
Short-duration sovereign claims, meaning treasury bills and their equivalents, are the closest thing to a riskless nominal asset. Their real return is the empirical crux of this whole question.
Here is what the longest available evidence says. Across twenty-one countries since 1900, bills delivered a real return of roughly one percent per year or less on average. The United States sits near the top of that range and several countries sit below zero. And there are extended periods, the 1940s and the 1970s among them, in which real bill returns were materially negative for a decade or more.
Read that twice, because both halves matter. On average, across a very long run, the safest instrument roughly held its ground. For stretches long enough to span a large fraction of a working life, it did not.
The Tax Layer, Handled Carefully
Taxation enters next, and it has to be handled carefully, because a great deal of long-duration retirement saving happens inside tax-advantaged structures. An argument that depended on the taxation of interest would not reach that saving, and I would rather disarm that objection myself than have it used against the argument later.
The second paper's framework supplies the distinction. The monetary property of the underlying claim is one thing. The tax wrapper in which the worker holds it is a different layer of the stack.
Where nominal interest is taxed without adjustment for inflation, as is common in taxable accounts, the after-tax real return on the safest claims has frequently been at or below zero, and the preservation burden rises.
Where the claim is held tax-sheltered, the burden falls back to the pre-tax real return, which is approximately flat on average with long negative stretches.
So preservation-motivated investment does not depend on taxation to exist. Taxation raises the burden and sheltering lowers it, but the underlying claim's real return sets the level, and that return has been near zero on average and materially negative for extended periods regardless of wrapper.
The Exception That Has to Be Named
Inflation-indexed government securities are the exception, and the second paper named them just as prominently.
They solve the general basket problem by construction. Where a worker can hold them in sufficient quantity, tax-sheltered, the preservation burden falls toward the indexed real yield.
Which has itself been negative for extended periods over the last two decades.
I want to be fair to this instrument rather than dismissive, because any account of monetary hazard that omits it is manufacturing alarm. A worker whose floor tranche is funded with indexed sovereign claims has answered the basket question, the price-level asymmetry question, and most of the regime history the second paper described, with a single instrument.
Why does it not close the problem? For reasons that are specific and sizeable rather than atmospheric. Indexed bonds are nearly absent from the default menus through which most workers' deferred pay actually travels, which is an architecture failure rather than an instrument failure. Their real yields have at times been negative for years. In taxable accounts the inflation adjustment is itself taxed. Their capacity is finite against the scale of aggregate retirement saving. And they remain claims on the sovereign balance sheet.
That first reason is the one that belongs to this profession. The instrument that most completely answers the question is the one your plan menu probably does not contain.
The Proposition, Stated Conditionally
Here is the empirical answer to the objection.
The safest broadly accessible nominal savings instruments have not reliably preserved purchasing power across inflation regimes and horizons. They have come close, on average, over the very long run. They have failed by wide margins for periods long enough to span a large fraction of a working life.
That is what I mean when I say the worker is paid in a claim that must either earn interest or be transformed. Interest on the safest claims has, on the record, been sufficient to approximately stand still on average and insufficient for long stretches. The worker who wants more than approximately, or who happens to live through one of those stretches, has to transform.
That formulation is less arresting than "advised not to hold" and considerably more defensible, and the argument does not weaken for it.
So the proposition has to be stated conditionally, because an unconditional version is falsified instantly by any period of positive real bill yields or by the availability of indexed claims.
Monetary decay creates a preservation burden that must be addressed through interest, indexation, transformation into other instruments, or acceptance of risk. The burden is zero only when the safest available claim reliably delivers a non-negative real return after whatever tax applies. When that condition fails, as the long-run record shows it has for extended periods in many economies, the worker has to transform the holding, accept additional risk, accept shortfall, or some combination of the three.
The argument does not need the condition to hold always. It needs the condition to exist and to matter, and the record establishes both.
A Reduction I Accept
There is a reduction of all this that I think should be accepted rather than resisted, and accepting it is what locates the contribution properly.
In monetary-economics terms, part of what I am calling the preservation burden is simply the price embodied in the real return on the safest accessible claim. A price set by the supply of saving and the demand for investment, rather than by any monetary pathology.
That is right, and the compensation contribution is not to rename the real rate.
It is to locate its incidence inside the transfer of deferred compensation, where it falls on an identifiable worker, through an architecture an employer designed, and where no instrument currently reports it.
Economists have studied the real rate for a century. Nobody has put it on a total rewards statement, because nobody thought it belonged to compensation.
Next week: the question that decides how much of this is really about the money. If the unit had zero expected inflation and its safest forms delivered a reliable zero real return, would workers still need preservation architecture?
The answer is plainly yes, and answering it honestly cuts this argument down to its actual size.