Free consultations, gifts to referral sources, above-market pay, the sample that costs you nothing to give. Every one of these rests on a claim that people repay generosity. The claim is well supported. What follows is about the second question.
Key Takeaway
From a field experiment of roughly 10,000 solicitation letters: "Compared to the no gift condition, the relative frequency of donations increased by 17 percent if a small gift was included and by 75 percent for a large gift."[1] From the labour market: field evidence found "small effort increases (25%-70%) that waned within hours", against laboratory effects of "e.g., 300%"[2].
Our Grades For These Claims
Applying the scheme from the first article in this series.
Grade A for the charitable field result. A randomised field experiment with roughly 10,000 subjects, published in Econometrica, abstract obtained verbatim from two independent sources.
Grade B for the labour market decay finding, which reaches us through a working paper's detailed description of the original rather than the original itself.
Grade B that immaterial gifts compound while monetary ones do not, from a paper whose abstract we obtained.
Ungraded for whether any of it pays, because no source we found reports the cost of the gifts, and our own arithmetic on that question uses invented figures throughout.
Our position: this is among the best-evidenced findings in the series and the commercial question is almost entirely unanswered, which is an unusual combination worth stating.
A Note On Method
Everything here is verified to August 2026.
We obtained the 2007 field experiment's abstract verbatim from two independent sources[1][3]. We did not obtain the paper, and full text access is recorded as restricted to subscribers[3].
We did not obtain the 2006 labour market paper. Its results reach us through a Yale working paper describing them in detail[2] and through a second working paper's summary[4], which disagree on one number.
We obtained two further abstracts in full[5][6] and a passage from a third[7].
No source we found reports what the gifts cost, which is the number the commercial question turns on.
All arithmetic is ours and uses invented response rates and donation sizes.
This article discusses research on reciprocity. It is not marketing, compensation or human resources advice.
Ten Thousand Letters
The design, which is why this article grades well.
Falk published Gift Exchange in the Field in Econometrica, 75(5), 1501–1511, in September 2007[3].
Its method: "In collaboration with a charitable organization, we sent roughly 10,000 solicitation letters to potential donors. One-third of the letters contained no gift, one-third contained a small gift, and one-third contained a large gift. Treatment assignment was random."[1]
Three observations, ours.
Ten thousand subjects, randomised, in a real setting with real money. After forty-eight articles frequently reporting samples of nine or twenty, this is worth pausing on.
The design has three arms rather than two, which means it recovers a dose-response relationship rather than a single contrast.
And the paper says why it exists: it "extends the current body of research on gift exchange, which is almost exclusively confined to laboratory studies."[3] That sentence describes the gap this series keeps encountering.
Seventeen And Seventy-Five
The result.
"The results confirm the economic importance of gift exchange. Compared to the no gift condition, the relative frequency of donations increased by 17 percent if a small gift was included and by 75 percent for a large gift."[1]
Three observations, ours.
The effect is large. A 75 percent increase in donation frequency is not a marginal lift, and it came from a randomised comparison rather than an observational one.
The effect is graded. Small gift, small response; large gift, large response. That monotonic pattern is itself evidence, because a spurious result has no reason to order itself correctly.
And note the phrase relative frequency of donations. This is about how many people gave, not how much. The abstract does not report amounts, and we have not seen them.
Why This One Is Different
Placing it against the rest of the series. Ours.
Three points.
Most articles here have concerned laboratory findings of contested generality. This is a field experiment on real donors making real decisions with their own money.
It also has no replication controversy we could locate, and the paper circulated for three years under an earlier title before publication[5], which is a long exposure to review.
And the honest consequence is that this article cannot be about whether the effect is real. It is, on the evidence we found. So the article has to ask something else.
The Labour Market Version
The other famous field test, and the one that complicates things.
Gneezy and List published Putting Behavioral Economics to Work: Testing for Gift Exchange in Labor Markets Using Field Experiments in Econometrica, 74(5), 1365–1384, in September 2006[4].
We did not obtain this paper. What follows is a working paper's description of it, and we flag that at every point.
That description credits the design: it is "the most persuasive despite its small sample because they address the two major confounds to gift exchange: threat of dismissal, by advertising tasks as a one-time job, and selection of higher quality workers, by contracting at approximately the going market wage."[2]
Two observations, ours.
Those are the two confounds that make workplace generosity hard to study. If people work harder for a generous employer, it may be because they fear dismissal, or because generous employers attract better applicants. The design removes both.
And despite its small sample is the describer's own qualifier, which we report rather than smooth over.
And It Waned Within Hours
The finding.
The same description: "In a first experiment, with a data-entry task, they find that a 67% wage raise increased the number of records entered by 25%, but it disappeared after three hours. In the second experiment, with a fundraising task, a 100% wage raise increased funds raised by 70% only in the first three hours."[2]
Another source states the general finding: in a labour market setting, "reciprocal behavior triggered by one monetary gift is temporary."[6]
Three observations, ours.
The effects were real and immediate. A quarter more output, or seventy percent more funds raised, is not nothing.
They were also over within a shift. Which means a manager observing the first two hours would draw a conclusion that the fourth hour would contradict.
And that is a measurement trap with general application. Any evaluation window shorter than the decay time will overstate the effect, and nobody designing a trial knows the decay time in advance.
Three Hundred Percent Versus Seventy
The gap between settings, which is the sharpest thing in our sources.
The same working paper: "Laboratory experiments suggest that gift exchange is a powerful incentive mechanism, generating not only involuntary unemployment but also large effort increases in one-shot interactions (e.g., 300%). Persuasive field evidence, however, has found more modest effects: small effort increases (25%-70%) that waned within hours."[2]
It adds that for smaller wage increases, another study found "effort responses that were not statistically significant", and that "despite gift exchange's importance, the reasons why it is curbed in the field are still unknown."[2]
Three observations, ours.
A laboratory effect of 300 percent against a field effect of 25 to 70 percent that then vanishes is roughly an order of magnitude, and it is in the direction this series has come to expect.
The candid admission is worth recording: the reasons why it is curbed in the field are still unknown. A well-established effect with an unexplained gap between settings.
And note that the charitable result does not show this pattern. It is a field study with a large effect, so the lab-field gap is not universal, and whatever explains it is not simply that field settings are noisy.
A Timing Discrepancy
A sourcing note, reported because our two descriptions do not agree.
One source says the data-entry effect "disappeared after three hours"[2]. Another describes the same study as showing "a higher wage for books catalogued increased productivity during the first 1.5 hours of work, but had no significant effect thereafter."[4]
Two observations, ours.
Three hours or one and a half. The descriptions may refer to different experiments within the same paper, or to the same one described loosely, and we could not resolve it because we did not obtain the paper.
It does not change the conclusion, since both figures are short. But it is the eighteenth bibliographic inconsistency this series has recorded, and here it sits in a load-bearing number rather than a page range.
Does It Pay
The question no source we found answers. Ours, and the arithmetic below uses invented figures throughout.
Two observations before the numbers.
An effect being real is a different question from an effect being worth its cost, and almost every treatment of reciprocity in business writing stops at the first.
No source we obtained reports what the gifts cost, in either the charitable or the labour study. That absence is the reason this section is arithmetic rather than reporting.
The Wage Case Does Not
Our own calculation on the reported figures.
A 67 percent wage rise producing a 25 percent output rise, lasting three hours.
Over the three hours it worked: output per wage dollar is about 25 percent worse than baseline.
Over a full eight-hour shift: about 35 percent worse.
Two observations.
It never pays on these numbers, and it does not pay even in the window where the effect was strongest, because 67 percent more wage cannot be recovered by 25 percent more output.
Which is not a criticism of the study. The authors were testing whether reciprocity exists in labour markets, not whether it is a profitable management technique, and it is the business literature that has quietly converted the first result into the second claim.
The Gift Case Might
The other direction, asked as a break-even question. Our arithmetic; the response rate and donation size are invented and appear in no source.
Assume 10,000 letters, a five percent baseline response, and an average gift of fifty dollars.
The small gift lifting response by 17 percent could cost up to about 42 cents per letter before it stops paying.
The large gift lifting it by 75 percent could cost up to about $1.88 per letter.
Three observations.
The break-even framing is more useful than a return estimate, because it turns an unknown into a threshold you can check against a real quote.
The large gift is not merely more effective; it has more than four times the budget to work within, which is a different and more useful statement than saying it worked better.
And every input above is invented. The paper reports relative frequency changes and we could not obtain the underlying rates. This shows how to ask the question, not its answer.
Compliments Compound, Money Does Not
The finding we would most want a professional firm to notice.
A field study of five gifts across five consecutive transactions reports: "We find that both immaterial gifts (compliments) and material gifts (tips), given in advance, induce positive reciprocity, i.e., salespersons provide more product weight. While monetary gifts trigger a larger level effect, only the effect of immaterial gifts increases significantly over repeated interactions."[6]
Three observations, ours.
Money produced the bigger single hit. Compliments produced the only effect that grew.
Which sits directly beside the decay finding. One monetary gift is described as temporary; repeated immaterial gifts are reported as increasing. Two studies, consistent shape.
And it is free, which is worth saying plainly. The intervention with the compounding effect in this literature costs nothing.
It Showed Up In Quality
A result that changes what to measure.
A hybrid field-laboratory experiment reports: "We find that although the total number of surveys entered did not vary in terms of wages, high wage workers made fewer errors and entered more surveys after controlling for errors."[7]
And its economic conclusion: "for low costs associated with errors, offering a low wage maximises profits; but for higher costs, paying a high 'gift exchange' wage maximises profits."[7]
Three observations, ours.
The effect was invisible in the headline measure. Volume did not move. Errors did.
Which means a firm measuring output alone would have concluded the gift did nothing, and been wrong.
And the conclusion is properly conditional: whether generosity pays depends on what an error costs you. For work where mistakes are cheap, it does not. For work where they are expensive, it may. That is a considerably more useful statement than a general principle, and for a professional services firm it points one way.
It Depends Who You Give To
A moderator with an unusually specific claim.
A field experiment reports: "Results indicate a strong complementarity between the initial wage-gift and the agent's ability to 'repay the gift'. We control for differences in ability and reciprocal inclination and show that gift-exchange is more effective with more reciprocal agents." The authors say their results "help to reconcile the conflicting evidence on the efficacy of gift-exchange outside the lab."[5]
Two observations, ours.
Two separate conditions are named: the recipient must be able to repay, and inclined to. A gift to someone who cannot increase their output has nowhere to go.
And if reciprocal inclination varies between people, then an average effect across a workforce understates it for some and overstates it for others, which is the heterogeneity problem this series has met in nearly every literature it has covered.
What This Means For A Firm
The application. Ours, untested.
Four points.
Free work generates reciprocity and probably does not pay for itself directly. On the wage arithmetic above, a gift priced at 67 percent of the fee buying 25 percent more of anything is a loss, and a free consultation is a gift priced at 100 percent.
The case for it is the second interaction, not the first. The only compounding effect in this literature was the immaterial, repeated one, and a single monetary gift is described as temporary.
Measure quality, not volume. In one study the entire effect appeared in error rates and was absent from output counts.
And give to people who can act on it. The complementarity finding says the gift works where the recipient has both the ability and the inclination to repay, which is an argument for targeting rather than for generosity in general.
The Line We Would Draw
Ours, in the manner of the twenty-eighth article.
Three positions.
A gift given because you value the relationship is not a technique, and nothing here bears on it. Most professional generosity is of this kind.
A gift engineered to trigger obligation is a different act, and the test is whether you would be comfortable saying why you sent it. A firm that would not put its reasoning in the accompanying note has answered the question.
And the honest observation is that the literature's most effective repeated instrument is a compliment, which is difficult to weaponise and cheap to mean sincerely.
What To Do
Accept that the effect is real. A randomised field experiment of roughly 10,000 letters found donation frequency up 75 percent with a large gift, and this is among the best-evidenced findings in this series.
Ask separately whether it pays. No source we found reports the cost of any gift, and on the labour figures a 67 percent wage rise buying 25 percent more output never covers itself.
Work out the break-even, not the return. It converts an unknown into a threshold you can check against an actual quote.
Expect monetary effects to fade fast. Field evidence puts the decay in hours, and any evaluation window shorter than that will overstate what you bought.
Use repetition and use it immaterially. The only effect reported as increasing over repeated interactions was the immaterial one, and it costs nothing.
Measure error rates. One study found the entire effect there and none of it in volume.
Target people who can repay. The complementarity finding says the gift needs somewhere to go.
Do not import laboratory magnitudes. Three hundred percent in a lab against 25 to 70 in the field, and the reason for the gap is described as still unknown.
The Limits Of This Analysis
Several caveats matter. This article discusses research on reciprocity and is not marketing, compensation or human resources advice. Everything is verified to August 2026. We obtained the 2007 field experiment's abstract verbatim from two independent sources but did not obtain the paper, whose full text is recorded as restricted, and we report none of its statistics, amounts donated or gift costs. We did not obtain the 2006 labour market paper at all; its results reach us through two working papers' descriptions, which disagree on whether the effect lasted three hours or one and a half, a discrepancy we report and could not resolve. No source we found reports what any gift cost, which is the figure the commercial question turns on. All arithmetic is ours; the wage calculation uses the reported percentages but assumes a constant baseline output rate, and the break-even calculation uses a response rate and donation size that are entirely invented and appear in no source, so those figures demonstrate how to ask the question rather than answering it. The applications to professional firms and the section on where the line falls are our own reasoning, untested.
Frequently Asked Questions
Is reciprocity real?
Does that mean gifts pay?
How long does the effect last?
What works better than money?
Why do lab and field results differ so much?
What should I measure?
References
- Publisher record for Falk, Armin, Gift Exchange in the Field, Econometrica, volume 75, issue 5, pages 1501–1511, reproducing the abstract, on the study reporting evidence from a field experiment conducted to investigate the relevance of gift exchange in a natural setting; on roughly 10,000 solicitation letters having been sent to potential donors in collaboration with a charitable organization, with one-third containing no gift, one-third a small gift and one-third a large gift, and treatment assignment random; and on the results confirming the economic importance of gift exchange, with the relative frequency of donations increasing by 17 percent for a small gift and 75 percent for a large gift compared to the no gift condition. Note: the publisher's record. We obtained the abstract and did not obtain the paper, its statistics, the amounts donated, or the cost of either gift. onlinelibrary.wiley.com
- University working paper on revisiting gift exchange, on the gift exchange hypothesis first proposed by Akerlof (1982) postulating that workers reciprocate above-market wages with above-minimal effort and having garnered mixed empirical support; on laboratory experiments suggesting gift exchange is a powerful incentive mechanism generating large effort increases in one-shot interactions, for example 300 percent; on persuasive field evidence having found more modest effects, being small effort increases of 25 to 70 percent that waned within hours, citing Gneezy and List (2006), and effort responses that were not statistically significant for smaller wage increases, citing Kube, Maréchal and Puppe (2012); on the reasons why gift exchange is curbed in the field still being unknown; on Gneezy and List (2006) being the most persuasive test despite its small sample because it addresses the threat of dismissal by advertising tasks as a one-time job and selection of higher quality workers by contracting at approximately the going market wage; and on a first experiment with a data-entry task finding a 67 percent wage raise increased records entered by 25 percent but disappeared after three hours, and a second with a fundraising task finding a 100 percent wage raise increased funds raised by 70 percent only in the first three hours. Note: a working paper describing another paper's results. Our principal source for the labour market findings, which we did not obtain directly. isps.yale.edu
- Bibliographic database record for Armin Falk (2007), Gift Exchange in the Field, Econometrica, Econometric Society, volume 75, issue 5, pages 1501–1511, September, reproducing the abstract identically and adding that the study extends the current body of research on gift exchange, which is almost exclusively confined to laboratory studies; and recording that access to the full text is restricted to subscribers. Note: a bibliographic database record; a second independent reproduction of the abstract, and the source for the paper's own statement about the laboratory-dominated literature. Full text access restricted. ideas.repec.org
- Institute discussion paper on the multi-dimensional effects of reciprocity on worker effort, on the authors having found a higher wage led to directionally greater productivity during Day 1 but not Day 2, described as similar to Gneezy and List's (2006) evidence that a higher wage for books catalogued increased productivity during the first 1.5 hours of work but had no significant effect thereafter; and confirming the citation as Gneezy, U., and List, J. (2006), Econometrica, 74(5), 1365–1384. Note: a discussion paper's summary. This source gives the effect duration as 1.5 hours where reference 2 gives three hours; the discrepancy is reported in the body and we could not resolve it without the original paper. docs.iza.org
- Publisher record for a field experiment varying both the presence of a gift-exchange wage and the effect of the worker's effort on the manager's payoff, reproducing the abstract, on results indicating a strong complementarity between the initial wage-gift and the agent's ability to repay the gift; on the authors controlling for differences in ability and reciprocal inclination and showing that gift-exchange is more effective with more reciprocal agents; on a principal-agent model with reciprocal subjects motivating the findings; and on the results helping to reconcile the conflicting evidence on the efficacy of gift-exchange outside the lab. Note: a publisher record; we obtained the abstract only and not the paper or its magnitudes. link.springer.com
- Full-text repository copy of a field study on immaterial and monetary gifts in economic transactions, published in Experimental Economics, on Gneezy and List (2006) and Ockenfels and colleagues (2015) showing the importance of investigating how reciprocity develops over time; on the former finding that in a labor market setting reciprocal behavior triggered by one monetary gift is temporary while the latter report higher work performance for the same total wage when wage is increased in two steps rather than once; on the authors providing five immaterial or five monetary gifts in five consecutive transactions over the same number of working days; and on the finding that both immaterial gifts, being compliments, and material gifts, being tips, given in advance induce positive reciprocity, with monetary gifts triggering a larger level effect while only the effect of immaterial gifts increases significantly over repeated interactions. Note: a full-text repository copy; we obtained these passages and not the study's full results or magnitudes. pmc.ncbi.nlm.nih.gov
- Bibliographic database record for Kim, Min-Taec, and Slonim, Robert, The multi-dimensional effects of reciprocity on worker effort: evidence from a hybrid field-laboratory labour market experiment, reproducing the abstract, on the authors examining the gift exchange hypothesis on both the quantity and quality of work; on participants having been recruited to enter survey data for a well-known charitable organisation and paid either a high or low wage; on the finding that although the total number of surveys entered did not vary in terms of wages, high wage workers made fewer errors and entered more surveys after controlling for errors; and on the finding that for low costs associated with errors a low wage maximises profits, while for higher costs paying a high gift exchange wage maximises profits. Note: a bibliographic database record; we obtained the abstract only. ideas.repec.org
This article discusses research on reciprocity and is not marketing, compensation or human resources advice. Neither of the two principal papers was obtained; one abstract was verified across two independent sources and the other reaches this article only through working papers describing it, which disagree on the duration of the effect. No source obtained reports the cost of any gift. All arithmetic is the authors' own and uses invented response rates and donation sizes, demonstrating how to frame the question rather than answering it.