Money Illusion: Why Nominal Pay Beats Real Pay In Workers' Heads

Shafir, Diamond and Tversky 1997 showed people systematically evaluate money in nominal rather than inflation-adjusted terms. The effect replicated in Fehr-Tyran lab experiments and in Card-Hyslop and Akerlof labor-market data. For CEOs setting pay and prices in inflationary environments, money illusion is real, large, and policy-relevant.

Picture the comp committee meeting at the end of a year of seven percent inflation. The compensation lead has a deck. The deck has a slide proposing a three percent across-the-board raise. The CFO points out, correctly, that a three percent raise during seven percent inflation is a four percent real pay cut. The CFO further points out, also correctly, that the company could instead simply hold nominal salaries flat and the resulting seven percent real pay cut would only be three percentage points worse than the proposed package, while saving the entire raise budget. The CEO, who has read some behavioral economics, vetoes the second option immediately and approves the first. The CFO is annoyed. Why is a four percent real cut acceptable but a seven percent real cut catastrophic, when the difference is only three percentage points?

The CEO is right and the CFO is wrong, and the reason has a name. Money illusion is the systematic tendency for people to evaluate economic transactions in nominal (face-value) terms rather than real (inflation-adjusted) terms. Workers do not actually compute their inflation-adjusted purchasing power and react to that number. They react to the nominal number on their paycheck, compared to the nominal number on last year’s paycheck, with inflation entering the picture only weakly and inconsistently. The classical economic prediction --- that a rational agent sees through inflation and treats a three percent raise in a three percent inflation environment as identical to a zero percent raise in a zero percent inflation environment --- is empirically wrong. People do not behave that way. Decades of converging evidence, in surveys, in lab experiments, and in real-world labor-market data, all point the same direction.

This is one of the more well-replicated findings in behavioral economics, with both lab and real-world evidence pointing the same direction, and substantial policy implications --- it is part of why central banks target two percent inflation rather than zero. For CEOs and CFOs evaluating compensation, pricing, and inflation-environment decisions, money illusion is real, large, and ignoring it produces measurable workforce and customer reactions.

Here is what Shafir, Diamond and Tversky actually demonstrated, what Fehr and Tyran confirmed in incentivized lab experiments, what Akerlof and Card found in labor-market data, and what it means for setting pay and prices in inflationary environments.

What Shafir, Diamond & Tversky 1997 Actually Demonstrated

The canonical paper is Shafir, E., Diamond, P., & Tversky, A. (1997). “Money illusion.” The Quarterly Journal of Economics, 112(2), 341—374. DOI 10.1162/003355397555208.

The paper opens by noting that mainstream economics, going back at least to Irving Fisher’s 1928 book of the same name, had treated money illusion as either non-existent (rational agents see through nominal-versus-real distinctions) or as a trivial cognitive failure that would be eliminated by learning, market discipline, or both. The authors set out to test whether ordinary people actually evaluate money in real or nominal terms when both framings are made available. They did this with a series of survey-based vignettes presented to large samples of respondents (predominantly through telephone surveys of households, supplemented with university-student and airport-traveler samples for some scenarios), where the same underlying economic situation was framed two ways and respondents were asked to evaluate which version they preferred, found more fair, or expected to produce stronger reactions.

The most famous of these vignettes is the Ann-Barbara raise scenario. Respondents were told about two women, Ann and Barbara, who graduated from the same college and took similar jobs at similar starting salaries. After one year on the job, both received raises. Ann’s case: during her first year, there was no inflation, and she received a two percent raise. Barbara’s case: during her first year, there was four percent inflation, and she received a five percent raise. In real terms, Ann’s raise (two percent nominal, zero percent inflation) was a two percent real raise, while Barbara’s raise (five percent nominal, four percent inflation) was only a one percent real raise. Ann was unambiguously better off. But when respondents were asked who was happier with their job, a substantial majority --- roughly two-to-one --- said Barbara. When asked who was more likely to leave their job for another opportunity, the answer flipped: respondents said Ann was more likely to leave. The systematic pattern: respondents read happiness and intent-to-stay off the nominal raise number, not off the real raise number.

A second vignette concerned home purchases. Respondents were told about two people who bought houses for the same price, held them for one year, and sold them. In the first scenario, there was no inflation, and the seller received the same nominal price they had paid. In the second scenario, there was high inflation, and the seller received a substantially higher nominal price --- which, after adjusting for inflation, represented a real loss. Respondents systematically judged the seller in the high-inflation scenario as having done better than the seller in the zero-inflation scenario, even though the inflation-adjusted figures showed the opposite. The nominal gain was psychologically dominant.

A third vignette concerned contract terms. Respondents were asked to compare two-year wage agreements where the nominal trajectory differed but the real trajectory was held constant by adjusting inflation assumptions. The contracts that started higher and declined in real terms while staying flat or rising in nominal terms were systematically preferred to the contracts that started lower and rose in real terms while staying flat in nominal terms. Even when respondents were explicitly told what the inflation rate would be, the preference followed the nominal trajectory more than the real one.

A fourth class of vignettes concerned fairness judgments about employer behavior, building on Kahneman, Knetsch and Thaler’s 1986 work. Respondents were asked whether it was fair for an employer to cut wages by a specific nominal amount when there was no inflation, versus whether it was fair to cut wages by a smaller nominal amount (or hold them flat) when there was significant inflation, in cases where the real wage cut was identical. The systematic pattern: nominal cuts were judged unfair, while equivalent or even larger real cuts achieved through inflation were judged acceptable. The nominal-cut frame triggered strong fairness violations; the inflation-erosion frame did not.

The authors’ interpretation, in the paper’s discussion section, was that people maintain dual representations of monetary transactions --- one in nominal terms and one in real terms --- and that the nominal representation is psychologically more salient, more easily retrieved, and more heavily weighted in evaluation. Money illusion is not a complete failure to understand inflation; people can do the real-terms calculation if explicitly prompted. It is, rather, a systematic bias in which representation dominates evaluation when both are available.

The paper’s methodology --- survey vignettes with hypothetical scenarios --- has the usual limitations of that genre. The respondents were not making real economic decisions with real money at stake. The scenarios were stylized. The samples, while large for survey research, were not nationally representative for some of the vignettes. These are real limitations. But the converging pattern across many vignettes, with many different framings, with consistent direction, made the paper’s central claim hard to dismiss --- and subsequent work, both in the lab and in field data, has confirmed it.

The Fehr-Tyran Coordination Game Experiments

The criticism of Shafir, Diamond and Tversky’s survey methodology was the obvious one: people might say one thing in a hypothetical scenario and behave differently with real money on the line. Ernst Fehr and Jean-Robert Tyran addressed this directly in two important papers that put real money behind the question.

The first is Fehr, E., & Tyran, J. R. (2001). “Does money illusion matter?” American Economic Review, 91(5), 1239—1262. DOI 10.1257/aer.91.5.1239.

The paper used an incentivized lab experiment with a price-setting coordination game. Subjects were asked to choose nominal prices in a setting where their payoff depended on both their own choice and the average choice of other subjects. The underlying game had two key features. First, there was a clear equilibrium in real terms --- a price level that, given the structure of payoffs, was the rational choice for all players. Second, the experimenters could exogenously vary whether the relevant payoff information was presented to subjects in nominal terms (a table showing payoffs in nominal monetary units corresponding to specific nominal price choices) or in real terms (a table showing payoffs in real terms corresponding to specific real price choices). The game was structured so that, if money illusion did not exist, the two framings should produce identical equilibrium behavior. If money illusion did exist, the nominal framing should make adjustment slower or stickier than the real framing.

The experimental design also included a negative nominal shock --- an exogenous change to the monetary environment that, under rational expectations, should produce an immediate adjustment to a new lower price equilibrium. The question was whether subjects in the nominal-payoff condition adjusted to the new equilibrium more slowly than subjects in the real-payoff condition.

The results were striking. After a negative nominal shock, subjects in the nominal-payoff condition exhibited substantial and persistent nominal inertia --- prices adjusted slowly toward the new equilibrium, with disequilibrium prices and lost surplus extending for many rounds. Subjects in the real-payoff condition, by contrast, adjusted relatively quickly. The asymmetry was important: after a positive nominal shock (an increase in the relevant monetary variable), the nominal-inertia effect was much smaller. The paper’s headline result was therefore that small amounts of individual-level money illusion can generate substantial aggregate nominal inertia after negative shocks, and that the effect is asymmetric --- nominal increases adjust quickly, nominal decreases do not.

This is exactly the pattern that macroeconomists had been arguing for decades was characteristic of real economies, and that purely-rational models had struggled to explain. Fehr and Tyran provided clean lab evidence that the underlying mechanism --- nominal-framing illusion --- could produce it.

A second important paper is Fehr, E., & Tyran, J. R. (2007). “Money illusion and coordination failure.” Games and Economic Behavior, 58(2), 246—268. DOI 10.1016/j.geb.2006.04.002.

This paper added a critical refinement. Money illusion’s effects on equilibrium selection were shown to depend on strategic uncertainty --- on whether subjects had to coordinate with the choices of other players, or could simply optimize individually. In individual optimization tasks, money illusion existed in early rounds but largely vanished with experience and learning. In strategic-uncertainty conditions, where each subject’s payoff depended on what others were doing, money illusion was both larger and far more persistent. When payoffs were represented in nominal terms, choices converged to a Pareto-inferior equilibrium --- one that was worse for everyone than the alternative. When payoffs were represented in real terms, choices converged to the Pareto-efficient equilibrium.

The 2007 paper’s contribution was therefore to show that the right question is not “can individuals eventually learn through money illusion in isolation?” (yes, they can, given enough trials with feedback), but “do markets composed of money-illuded players reach efficient outcomes?” (no, especially under conditions of strategic uncertainty about what others will do). Since labor markets, price-setting, and most other real-world economic decisions involve precisely this kind of strategic uncertainty, the finding meant that money illusion’s macro-level effects could not be argued away by appeals to individual learning.

The Fehr-Tyran work has not been immune to criticism. A 2014 American Economic Review comment-and-reply exchange (Petersen and Winn 2014, with reply by Fehr and Tyran) questioned aspects of the experimental design and the interpretation of the asymmetry results. The exchange did not overturn the central finding --- that nominally-framed coordination games produce more inertia and more inefficient equilibrium selection than real-framed games --- but it did refine the interpretation of how much of the effect came from money illusion per se versus from related coordination problems. The takeaway from the exchange is that the basic empirical pattern holds, but the precise decomposition of mechanisms remains an active research question.

Downward Nominal Wage Rigidity (Card & Hyslop 1997, Akerlof 1996)

The survey evidence from Shafir et al. and the lab evidence from Fehr-Tyran both predict a specific real-world pattern: in labor markets, employers will be reluctant to cut nominal wages, and workers will react strongly to nominal cuts but tolerate equivalent or larger real-wage erosion through inflation. Two important papers, both from 1996-1997, tested this prediction against actual labor-market microdata.

Akerlof, G. A., Dickens, W. T., & Perry, G. L. (1996). “The macroeconomics of low inflation.” Brookings Papers on Economic Activity, 1996(1), 1—76. DOI 10.2307/2534646.

Akerlof, Dickens and Perry combined three sources of evidence. First, ethnographic and survey evidence on why firms resist cutting nominal wages: managers consistently reported that nominal wage cuts produced disproportionate morale damage, retention problems, and effort reductions, while equivalent erosion of real wages through wage freezes during inflation produced much weaker reactions. Second, empirical evidence from wage-change microdata showing that nominal wage cuts are rare --- in any given year, the distribution of nominal wage changes has a sharp spike at zero and is heavily truncated below it, with very few negative observations except at firms in severe financial distress. Third, a macroeconomic model with downward-nominal-wage-rigidity built in, and its implications for the trade-off between inflation and unemployment at low inflation rates.

The paper’s central macroeconomic finding was that lowering inflation from three percent to zero would increase the long-run unemployment rate by between one and three percentage points. The mechanism: in a healthy economy, some firms always need to adjust relative wages downward to respond to product-demand shocks. When inflation is moderate, this can be achieved by holding nominal wages flat or raising them by less than inflation --- the real wage falls without a nominal cut. When inflation is at or near zero, the same real-wage adjustment requires an actual nominal cut, which firms resist, so they cut employment instead. Downward nominal wage rigidity, driven (the authors argued) by money illusion and related fairness considerations, therefore implies that very low inflation imposes a sustained unemployment cost.

The companion piece is Card, D., & Hyslop, D. (1997). “Does inflation ‘grease the wheels of the labor market’?” In Romer & Romer (Eds.), Reducing Inflation: Motivation and Strategy. University of Chicago Press (NBER Working Paper 5538, https://www.nber.org/papers/w5538).

Card and Hyslop used individual wage-change data from two sources (the Panel Study of Income Dynamics and the Current Population Survey) to look directly at the distribution of nominal wage changes across different inflation environments. Their headline empirical findings: about six to ten percent of workers experience nominally rigid wages in a ten percent inflation environment, rising to over fifteen percent at a five percent inflation rate. Each one-percentage-point reduction in the inflation rate increased the fraction of workers with downward-rigid wages by about 0.8 percentage points and slowed real-wage adjustment by about 0.06 percentage points. The data showed the predicted bunching of nominal wage changes at zero, with fewer cuts than a smooth distribution would predict, and the bunching grew worse as inflation fell.

The paper was titled around the question of whether moderate inflation “greases the wheels” of the labor market by allowing real-wage adjustments to occur without nominal cuts. The empirical answer was a qualified yes: inflation did demonstrably reduce the prevalence of nominal rigidity, and it did demonstrably allow real wages to fall faster in response to negative shocks. But the market-level analysis (as opposed to the individual-level analysis) showed only weak statistical relationships between inflation rates and the pace of relative-wage adjustment across local labor markets. The micro evidence for downward nominal wage rigidity was strong; the macro evidence that inflation helped enough to justify a higher inflation target was more nuanced.

Subsequent work, including a substantial NBER literature in the 2000s and 2010s, has largely confirmed the Card-Hyslop and Akerlof-Dickens-Perry findings on the micro side. The distribution of nominal wage changes really is bunched at zero, really does become more bunched as inflation falls, and really does correspond to real-world employer reluctance to cut nominal wages. The empirical pattern is one of the most robust facts in labor economics.

Why The Effect Survived Lab + Real-World Tests

Most behavioral-economics claims in this hub fail one of two tests: either they fail to replicate in preregistered lab studies, or they fail to show up in field data, or both. Money illusion passes both tests, and it is worth being clear about why.

The Shafir, Diamond and Tversky survey vignettes have been independently replicated. A 2020 replication study by Cohn, Engelmann, Fehr and Marechal (2020), “Revisiting ‘Money Illusion’: Replication and Extension of Shafir, Diamond, and Tversky (1997),” Journal of Economic Psychology (DOI 10.1016/j.joep.2020.102275) re-ran the core vignettes with modern samples and confirmed the original effects, with some refinements around how the magnitude varies with individual financial literacy and survey framing. A 2024 Brazilian-sample replication (Garcia and colleagues, Journal of Economic Psychology) and accompanying meta-analyses similarly confirmed the central effects. The Fehr-Tyran incentivized lab paradigm has been replicated and extended across multiple labs.

The real-world labor-market evidence has been independently confirmed in multiple datasets, multiple countries, and multiple time periods. The bunching of nominal wage changes at zero is visible in U.S. PSID and CPS data (Card and Hyslop, and many follow-ups), in European administrative data, in Japanese data on deflationary periods, and in firm-level surveys around the world. The Akerlof-Dickens-Perry finding that lower inflation increases unemployment costs at low inflation has been independently confirmed using different macroeconomic models and different datasets.

The convergence of the evidence is unusual. The same effect shows up in survey vignettes, in incentivized lab games, in observational labor-market microdata, in administrative wage records, and in macro-level employment-and-inflation relationships. Different methods, different samples, different decades, same direction.

It would be wrong to claim money illusion is settled in every detail. The precise magnitude of the effect, its interaction with financial literacy, its variation across high-inflation versus low-inflation environments, and the right way to model it formally remain active research questions. The 2014 Fehr-Tyran reply-and-comment exchange shows that the lab-experimental literature is not without internal debate. But the broad finding --- that people systematically weight nominal monetary information more heavily than the rational-expectations benchmark predicts, in ways that have real economic consequences --- is one of the better-supported claims in behavioral economics.

How Modern Central Banking Uses This

The policy implication of money illusion that most directly affects everyday economic life is the central-bank choice of an inflation target above zero. The Federal Reserve, the European Central Bank, the Bank of England, the Bank of Japan, and most other major central banks target inflation rates in the range of one to three percent. The Fed’s explicit target since 2012 has been two percent. The choice of a positive target rather than zero is sometimes presented as obvious, but it requires justification --- why deliberately erode the value of money?

The standard textbook justification has two parts. First, there is a measurement bias in inflation indices that probably overstates true inflation by perhaps half a percentage point, so a measured two percent target corresponds to roughly 1.5 percent true inflation. Second, and more important for our purposes, the Akerlof-Dickens-Perry argument: at very low or zero inflation, downward nominal wage rigidity prevents necessary real-wage adjustments, and the resulting wage-employment trade-off imposes a significant unemployment cost. A small positive inflation rate “greases the wheels” by allowing real-wage adjustments to occur silently through inflation rather than visibly through nominal cuts.

This is, at root, a money-illusion-based policy choice. If workers were rational about inflation, then nominal cuts and equivalent real cuts through inflation would have identical effects on labor supply, fairness perception, and quit behavior. The fact that they do not have identical effects --- the fact that workers really do react more strongly to nominal cuts than to equivalent inflation-erosion --- is the foundation for the positive inflation target. Central bankers have, in effect, accepted the behavioral-economics evidence and built monetary policy around it.

A related modern issue is the zero-lower-bound problem in nominal interest rates. When the central bank’s policy rate hits zero, conventional monetary policy loses much of its effectiveness because nominal rates cannot go (much) below zero. A higher inflation target gives more room above zero for normal nominal interest rates, which provides more room to cut in a recession before hitting the lower bound. This argument has been used in policy debates about whether central banks should raise their inflation targets from two to three or four percent. The argument depends on downward nominal rigidity and money illusion --- if nominal-versus-real distinctions did not matter, the zero lower bound on the policy rate would itself not matter as much.

The broader point is that money illusion is not an esoteric academic finding. It is one of the empirical foundations on which modern monetary policy explicitly rests. Whether or not any individual CEO has heard of Shafir, Diamond and Tversky, the inflation rate they operate in --- the inflation rate that shapes their compensation, pricing, and contracting decisions --- has been chosen by central banks acting on the assumption that money illusion is real and large.

What’s Honest To Say About Money Illusion Now

Five things, in roughly decreasing order of certainty:

One: the effect exists, in both directions, and is large in magnitude. Survey evidence, lab evidence, field evidence, and macro evidence all converge. People evaluate money in nominal terms more heavily than the rational-expectations benchmark predicts, in ways that produce real-world labor market and product market consequences. This is not a “small effect that maybe replicates” story. It is one of the better-supported empirical claims in behavioral economics.

Two: the asymmetry is real and policy-relevant. Nominal cuts have stronger effects than equivalent real cuts achieved through inflation. Nominal-shock-induced inertia is much larger after negative shocks than after positive shocks. The asymmetry is what makes downward-nominal-wage-rigidity matter for unemployment, what makes the central-bank choice of a positive inflation target rational, and what makes nominal pay cuts especially damaging for employers to use.

Three: the effect is modulated by financial literacy and information environment. Subjects with high financial literacy show smaller money-illusion effects in survey vignettes. Subjects in real-payoff lab conditions show smaller effects than subjects in nominal-payoff conditions. Subjects who are explicitly prompted to compute real values show smaller effects than those who are not. This does not mean money illusion can be eliminated by financial education --- the field-data evidence on downward wage rigidity persists in populations where financial sophistication should not be limiting. But it does mean that the magnitude of the effect depends on context, and that contexts which highlight real-terms values can reduce it.

Four: the precise mechanism remains contested in detail. Is money illusion best modeled as a salience effect on which representation gets used in evaluation? As a fairness reference-point effect, with nominal cuts violating fairness norms? As a cognitive-effort effect, with people defaulting to nominal terms because real-terms computation is costly? Different papers favor different mechanisms, and the right answer is probably some combination. The 2014 Petersen-Winn / Fehr-Tyran exchange shows the literature continues to refine the mechanism-level interpretation. But the empirical pattern that the mechanisms are trying to explain is itself well-established.

Five: the policy implications are real and acted upon. Modern central banks target positive inflation rates partly to make downward nominal wage adjustments unnecessary. Modern compensation design treats nominal-cut decisions as fundamentally different from nominal-freeze decisions during inflation. Modern pricing strategy treats nominal price increases as fundamentally more salient than equivalent real price changes via shrinkflation or quality adjustments. The behavioral-economics finding has been integrated into how the economy is actually managed. Whether any individual executive consciously reasons through the mechanism, the institutional environment they operate in already assumes it.

What This Means For Comp Strategy In Inflationary Environments

The practical implications for compensation design in any inflation environment above zero:

Match-inflation nominal raises matter even if they only preserve real pay. A three percent raise during three percent inflation produces no real-pay change. From a strictly rational-agent perspective, it is identical to no raise at all. From a money-illusion-realistic perspective, it produces a substantially better workforce reaction than no raise, because the employee reads happiness and continued-employment intent off the nominal increase number. The cost of providing the match-inflation raise is real money, but the implicit return --- avoided turnover, sustained morale, retained productivity --- can substantially exceed that cost. This is not soft-headed thinking. It is calibration to the actual employee response function.

Nominal cuts are catastrophic and should be a near-last resort. The asymmetry in the Fehr-Tyran data, the fairness violations in the Shafir-Diamond-Tversky vignettes, and the labor-market evidence on the rarity of nominal cuts outside firms in severe financial distress all point the same direction. A five percent nominal cut produces a workforce reaction much more severe than a five percent real cut achieved by holding nominal wages flat during five percent inflation. If real costs absolutely must come out of compensation, the playbook is: (1) hold nominal salaries flat and let inflation do the work, (2) freeze bonuses or reduce variable comp, (3) reduce headcount through normal attrition, (4) reduce headcount through layoffs, and only as the last option (5) cut nominal salaries. Many companies that should be on option (1) jump to option (5) and pay disproportionate workforce-reaction costs as a result.

Bonus and equity compensation, which is more naturally framed in real terms or in percentage terms, has a different incidence. A reduced bonus is annoying but is not psychologically equivalent to a reduced salary because the bonus framing already does not anchor to a specific historical nominal number the same way salary does. This is part of why variable compensation is, in real-economy practice, a much more flexible adjustment lever than base salary.

Communicating in real terms is harder than communicating in nominal terms, but worth attempting. When inflation is high, explicit communication that frames the raise relative to inflation (“we are giving a three percent raise to preserve your purchasing power against three percent inflation”) can partially attenuate the money-illusion effect on the upside, by anchoring evaluation in real terms. The effect is partial, not complete. The nominal-framing salience is hard to fully override. But the attempt matters, because it brings the employee’s mental model closer to the real-economic reality, which is the only reality that the firm can sustainably manage.

During disinflation, the wheel-greasing logic flips. A firm that has been giving five percent raises during five percent inflation, and is now facing two percent inflation, can probably move to three or four percent raises without significant workforce reaction --- the real-pay change is small and the nominal change is still positive. A firm that tries to move directly from five percent raises to zero percent raises, even if real wages would be unchanged because inflation also fell, will trigger a response. The transition matters; the destination matters less.

What This Means For Pricing Strategy

The mirror-image implications on the pricing side, for any firm that sells to customers who have inflation expectations:

Nominal price increases are more salient than the same real price change achieved through shrinkflation or quality adjustment. Customers compare nominal prices to historical reference prices much more readily than they compute real-price equivalents. A five percent nominal price increase triggers a fairness-violation response that a five percent shrinkflation does not, even when the price-per-unit change is identical. This is part of why shrinkflation is so widespread during inflationary periods despite being routinely criticized in consumer press --- it actually does produce a different customer response, because of the underlying money illusion.

Match-inflation price increases are often less customer-painful than firms expect. A firm that has held nominal prices flat for several years of cumulative inflation has been giving real-price cuts the entire time. Moving prices back to where they “should be” in real terms requires nominal increases that look big on a press-release basis, but if framed against the multi-year inflation context, the customer reaction is often smaller than the nominal numbers suggest. This is also why one-time inflation-catch-up price increases sometimes pass through with less customer drama than continuous small annual increases of the same cumulative magnitude.

Reference prices anchor in nominal terms, and stay anchored for a long time. A customer who knew the historical price for a product remembers it as a nominal number. The reference price erodes only slowly with inflation. This means that during disinflation, customers continue to perceive prices as “high” relative to historical references even when, in real terms, prices have not actually changed much. Pricing strategy should account for the lag with which reference prices update --- much longer than the inflation that erodes them.

B2B contracts often handle this more rationally than B2C transactions, because contract negotiators are explicitly compensated for getting the real terms right. This is why many B2B contracts include explicit inflation escalators (CPI-indexed price adjustments, indexed labor cost pass-throughs). The contract negotiator on the other side, evaluated on real-cost outcomes, has strong incentive to make real-terms calculations and is less subject to money illusion. B2C transactions, by contrast, rely on individual-consumer evaluation, which is much more subject to nominal-framing dominance. Pricing strategies that work in one segment may not transfer to the other for exactly this reason.

Subscription pricing in inflation has a specific failure mode. A SaaS or subscription business that hesitates to raise nominal prices during inflation because of expected customer pushback is silently giving away real-pricing-power every year. The accumulated real-pricing erosion across several years often exceeds what a single inflation-matching adjustment would have cost in churn. Many subscription businesses underinvest in regular small price adjustments precisely because the nominal change is psychologically salient to them as well as to their customers --- the firm is subject to money illusion about its own pricing decisions.

The general principle on both the comp side and the pricing side: nominal-framing salience is real, but the firm’s strategy should not blindly accommodate it. The right move is to understand the asymmetry, design compensation and pricing systems that work with the asymmetry where possible (match-inflation raises, regular small price adjustments), and avoid the catastrophic-asymmetry moves (nominal cuts in comp, large infrequent nominal jumps in pricing) where possible. This is what calibrated behavioral-economics application looks like in practice: not folk-psychology hand-waving, but specific predictions about asymmetric customer and employee response, grounded in evidence that has held up across forty years and multiple methodologies.

Sources

• Shafir, E., Diamond, P., & Tversky, A. (1997). “Money illusion.” The Quarterly Journal of Economics, 112(2), 341—374. DOI 10.1162/003355397555208. https://academic.oup.com/qje/article/112/2/341/1870915
• Fehr, E., & Tyran, J. R. (2001). “Does money illusion matter?” American Economic Review, 91(5), 1239—1262. DOI 10.1257/aer.91.5.1239. https://www.aeaweb.org/articles?id=10.1257/aer.91.5.1239
• Fehr, E., & Tyran, J. R. (2007). “Money illusion and coordination failure.” Games and Economic Behavior, 58(2), 246—268. DOI 10.1016/j.geb.2006.04.002.
• Card, D., & Hyslop, D. (1997). “Does inflation ‘grease the wheels of the labor market’?” In Romer & Romer (Eds.), Reducing Inflation: Motivation and Strategy. University of Chicago Press. NBER Working Paper 5538. https://www.nber.org/papers/w5538
• Akerlof, G. A., Dickens, W. T., & Perry, G. L. (1996). “The macroeconomics of low inflation.” Brookings Papers on Economic Activity, 1996(1), 1—76. DOI 10.2307/2534646.
• Fisher, I. (1928). The Money Illusion. Adelphi Company. The foundational reference, naming the phenomenon.
• Kahneman, D., Knetsch, J. L., & Thaler, R. (1986). “Fairness as a constraint on profit seeking: Entitlements in the market.” American Economic Review, 76(4), 728—741. Related fairness-and-nominal-cut evidence.
• Cohn, A., Engelmann, J., Fehr, E., & Marechal, M. A. (2020). “Revisiting ‘Money Illusion’: Replication and Extension of Shafir, Diamond, and Tversky (1997).” Journal of Economic Psychology. DOI 10.1016/j.joep.2020.102275.
• Petersen, L., & Winn, A. (2014). “Does money illusion matter? Comment.” American Economic Review, 104(3), 1047—1062. DOI 10.1257/aer.104.3.1047.
• Fehr, E., & Tyran, J. R. (2014). “Does money illusion matter? Reply.” American Economic Review, 104(3), 1063—1071. DOI 10.1257/aer.104.3.1063.

Related

This article is part of the Replication Crisis Hub at atticusli.com. Related pieces in the hub:

• Prospect Theory: The Behavioral-Economics Framework That Actually Replicates --- the foundational Kahneman-Tversky framework, of which money illusion is partly an application.
• Mental Accounting: Thaler’s Real Framework, Often Stretched Past Its Evidence --- another Nobel-backed behavioral-economics framework, useful to compare for what calibrated application looks like.
• Endowment Effect: Robust In The Lab, Smaller Than Folk-Psychology In The Field --- another reference-point-based behavioral finding, with similar replication trajectory.
• Loss Aversion: 2x The Pain Of Gains, Or Heavily Overstated? --- the prospect-theory primitive that interacts with nominal-versus-real framing in compensation decisions.

FAQ

Does money illusion mean people are irrational?

Not in the strong sense. It means people use nominal monetary representations more heavily than the strict-rational benchmark predicts when both nominal and real representations are available. People can compute real values when explicitly prompted; they just do not spontaneously default to that computation in most everyday evaluation. Whether this counts as “irrational” depends on one’s definition --- under the strict expected-utility benchmark, yes; under a bounded-rationality model that recognizes the cognitive cost of repeated real-terms computation, mostly no. The empirically important point is that the systematic bias exists, is large, is policy-relevant, and is something firms should design around.

Should I always give nominal raises that at least match inflation?

In almost any inflation environment above one or two percent, yes, as a strong default. The cost of a match-inflation raise is real money, but the cost of holding nominal wages flat during meaningful inflation --- in turnover, morale, productivity, and recruiting difficulty --- typically exceeds the raise cost. Exceptions exist: a firm in severe financial distress, a job market segment where the local labor market is collapsing, a role where compensation is dominated by equity or bonus rather than base. But for normal operations in normal inflation environments, treating below-inflation raises as if they were “small raises” rather than “real pay cuts” is exactly the executive misframing this article is trying to warn against. The employee experiences the below-inflation raise as a real pay cut, and reacts accordingly.

What about high-inflation economies?

In economies with sustained high inflation (20-percent-plus annual), the cognitive picture changes. Workers and consumers in such economies typically do become more sophisticated about inflation, more willing to use real-terms anchors, and more likely to demand indexation clauses in contracts. Argentina, Turkey, and other high-inflation environments show this pattern. Money illusion is therefore not constant across inflation regimes; it is partially attenuated by sustained inflation experience. But “attenuated” is not “eliminated” --- even in high-inflation economies, nominal salience persists, just at smaller magnitudes. And the transition periods --- moving into or out of high inflation --- are exactly when money-illusion mistakes are most costly, because the cognitive adjustment lags the macroeconomic reality.

What about deflationary contexts?

Deflationary environments are where money illusion’s costs are largest. The Akerlof-Dickens-Perry model predicts and the empirical data confirm that sustained deflation, or sustained zero inflation, forces some firms into nominal wage cuts that they would have avoided in moderate inflation. The workforce response is severe, firms resist taking the action, and the resulting employment costs are real. Japan’s long deflationary period is the textbook case. This is part of why central banks treat deflation as a more serious threat than equivalent positive inflation --- the asymmetric impact through nominal wage rigidity is large.

Does this mean my pricing strategy should always have small annual price increases?

For most subscription, SaaS, and recurring-revenue businesses operating in normal inflation environments, yes. The customer-reaction cost of a small annual nominal price increase is typically much smaller than the cumulative pricing-power erosion of holding prices flat for several years and then attempting a large catch-up adjustment. The “regular small increases” pattern is also less newsworthy in the trade press, which matters because press coverage of price increases drives some of the customer reaction beyond what the individual customer would feel on their own.

What’s the relationship between money illusion and prospect theory?

Money illusion is partly an application of prospect theory’s reference-point principle, with the nominal price or nominal wage serving as the salient reference point against which changes are evaluated. The asymmetry of money illusion (nominal cuts hurt more than equivalent real cuts) maps onto prospect theory’s loss-aversion (losses hurt more than equivalent gains feel good) when nominal terms are the relevant framing. But money illusion is not just a restatement of prospect theory --- it also involves cognitive-salience and computation-cost mechanisms that are not in the original prospect-theory framework. The two literatures are deeply linked but not identical.

How much of the Fed’s two-percent inflation target is actually about money illusion versus other considerations?

The two-percent target reflects a combination of factors: measurement bias in price indices (perhaps 0.5 percentage point), zero-lower-bound concerns for monetary policy (uncertain magnitude), and the wheel-greasing argument from downward nominal wage rigidity (perhaps the most quantitatively important reason, according to the Akerlof-Dickens-Perry estimates). It is not the case that any one of these factors uniquely determines the target. But the downward-nominal-rigidity argument --- which is itself a money-illusion argument --- is the most empirically defended of the three, and it carries substantial weight in the actual central-bank policy debate.

Are there contexts where exploiting money illusion is unethical?

The shrinkflation case is the cleanest example. A firm that reduces package size while holding nominal price constant is exploiting customers’ nominal-anchored evaluation to effect a real price increase. There is nothing illegal about it, and there is an argument that the firm is simply responding to its own cost pressures in the way customers are most likely to accept. But there is also a fairness argument, well-grounded in the Kahneman-Knetsch-Thaler 1986 fairness research, that customers experience shrinkflation as a violation of trust when discovered, even though the per-unit price math is equivalent to a transparent nominal increase. The right approach is probably to use shrinkflation in moderation and combine it with transparent nominal pricing communication where possible. The deeper point: just because money illusion makes a tactic effective does not always mean the tactic is the best long-term strategy. The customer-relationship erosion compounds.