The Impact of Gifts and Shared Experiences on an Investor-Manager Relationship

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General Assessment:
This paper presents an experimental investigation into the impact of gifts and shared experiences on investor-manager relationships, contributing to the literature on social preferences and affective ties in economic decision-making. The study is well-designed and addresses an important gap by comparing the effects of monetary transfers (gifts) versus shared project outcomes on investor behavior. The findings are intriguing, particularly the strong effect of transfers on switching rates, which aligns with prior work on gift exchange. However, there are several areas where the paper could be improved to strengthen its theoretical and methodological rigor.

Major Comments:

Theoretical Framework and Hypotheses:

The paper relies heavily on the Affective Ties Model (ATM) to explain the observed gift effect. While this is a reasonable approach, the authors should more explicitly contrast ATM with alternative explanations, such as reciprocity norms or inequity aversion, especially given the mixed evidence in the literature. A deeper discussion of why ATM is uniquely suited to explain the results would strengthen the theoretical contribution.

Hypothesis 2 posits an ordered effect (Control < History < Transfer), but the results only partially support this (Transfer shows the strongest effect, but History does not differ significantly from Control). The authors should revisit the framing of this hypothesis or provide a more nuanced interpretation of the findings.

Experimental Design:

The distinction between the Historyand Transfer treatments is clear, but the rationale for including the Control treatment could be better articulated. Specifically, how does the Control treatment help isolate the social element, given that it lacks a manager entirely? A more detailed justification would clarify its role in the design.

The paper mentions that the Controltreatment is identical to History except for the absence of a manager. However, this could introduce confounding factors, as the absence of a manager might alter the decision context beyond just removing the social element. The authors should address this potential limitation.

Results and Interpretation:

The finding that shared project outcomes (History) do not significantly affect switching rates is interesting but somewhat under-discussed. The authors suggest this aligns with ATM, but could there be alternative explanations, such as participants attributing outcomes to luck rather than the manager’s actions? This should be explored further.

The decision-time results are intriguing but not fully explained. Why are decision times longer in social treatments? Is this due to increased cognitive load, moral deliberation, or other factors? The authors should speculate on possible mechanisms.

 

Minor Comments:

The abstract could be more concise. The current version is detailed but somewhat lengthy; condensing it would improve readability.

The figures (e.g., Figure 5) are informative but could benefit from clearer labels or captions to guide the reader.

The discussion of Bayesian updating (Section 2.2) is technically sound but might be overly detailed for some readers. A brief summary or intuitive explanation could make this section more accessible.

Author Response

Response to R1

We are grateful for the Reviewer’s appreciation of our study (“well-designed”, “addressing an important gap”) and the many useful major and minor comments aimed at areas where the paper could be improved. Our responses (in roman) follow each comment (in italics).

Major comments

Theoretical framework and hypotheses

The paper relies heavily on the Affective Ties Model (ATM) to explain the observed gift effect. While this is a reasonable approach, the authors should more explicitly contrast ATM with alternative explanations, such as reciprocity norms or inequity aversion, especially given the mixed evidence in the literature. A deeper discussion of why ATM is uniquely suited to explain the results would strengthen the theoretical contribution.

 

A deeper discussion is now provided in the Concluding Discussion and Appendix C.

Hypothesis 2 posits an ordered effect (Control < History < Transfer), but the results only partially support this (Transfer shows the strongest effect, but History does not differ significantly from Control). The authors should revisit the framing of this hypothesis or provide a more nuanced interpretation of the findings.

 

This comment is not completely clear to us. Hypothesis 2 is our pre-formulated main analysis point of the paper. Changing the framing noticeably would therefore conflict with intellectual honesty. In the Discussion section we outline several possible reasons why the effect in the History treatment is not like we expected.

We nuanced Result 2 a bit, for greater clarity.

 

Experimental design

Rationale for including the Control treatment could be better articulated. Specifically, how does the Control treatment help isolate the social element, given that it lacks a manager entirely?

Absence of a manager might alter the decision context beyond just removing the social element. The authors should address this potential limitation.

 

The rationale is that only this way one can attribute an effect in History to the fact that the (random) choice of a project is delegated to a manager instead of being made by the investor him- or herself. It is not clear to us how one would control for this in another way. As outlined in the paper, we attempted to exclude as many possible alternative explanations as possible.

Results and interpretation

History finding: The authors suggest this aligns with ATM, but could there be alternative explanations, such as participants attributing outcomes to luck rather than the manager’s actions? This should be explored further.

Of course, one cannot exclude that there are other explanations. Our focus, as argued in the Introduction, concerns Gurdal et al. (2013) because it provides a model and evidence of ‘unjustified blame’, which is potentially relevant in our case in view of the randomness of choices. We also find and discuss some supportive (but weak) evidence in our experiment.

Decision-time results are intriguing but not fully explained; authors should speculate on possible mechanisms.

Our speculation on a possible mechanism – related to our model (subsection 2.4) – is presented in the Concluding Discussion: “Although it is not clear at this stage what exactly the reason is, a plausible driving factor concerns the extended utility of an investor in that situation, due to an additional norm and/or interaction partner’s payoff related utility component.”

 

Minor comments

Abstract could be more concise.

Done

 Clarity of labels/captions of Figures.

Done

 Discussion of Bayesian updating.

We’ve decided not to change it, because it is a standard calculation for optimal decision making, and it makes apparent that it is indeed a challenging task (see the last paragraph of Section 2.2). Moreover, our other Reviewer has no comment in this respect.

 

Additional Technical Changes

• We changed a lot of the formatting of tables and figures. Essentially, a lot of detail was lost in the process of porting the layout from our original manuscript to the journal layout, which we corrected.
• Some minor improvements to the consistency of data presentation (For example, a missing significance marker in table 2, spec. 5 was added.)
• We recently made the analysis code publicly available, as per journal requirements, and discovered a small inconsistency in the test that was reported to be used for Result 1. We had stated that we used a clustered chi2 test, but that was not the case. We switched this to a logit regression, with qualitatively similar and largely identical results. This leads to minor changes in footnotes 15 and 16 and the text just prior to Result 1.
• We found a missing number in the copy of the instructions: the digital version of the instructions in the live experiment said the exchange rate was “1 Euro per 35 ECU”, but in the version in the document it only said “1 Euro per ECUO”. This was corrected.
• Multiple minor language adaptions because of further read-throughs.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This is an interesting paper and I enjoy reading it. The central message seems to be that bringing an initial incident of good luck (a good outcome from the first project in History) does not invite reciprocation, while an initial gift (a transfer decision in Transfer) does. The first finding probably has more novelty than the second, which has a rich literature behind it as discussed by the authors. The results suggest that shared positive experience does not necessarily maintain a relationship, but direct gift giving can. The decision time analysis adds some interest too.

I think the authors could try to make Sections 1 and 2 clearer. As alluded above, it seems to me that the research is essentially about whether there could be reciprocation toward: (1) an unintentional good outcome (stroke of luck) brought about by the other party, and (2) a voluntary act of gift giving from the other party. The Introduction could be more clearly structured along these two research objectives, with hypothesis development structured in a similar way. At present, the exposition in both sections seems a bit mixed up and sometimes convoluted. For example, there could be separate hypotheses regarding these two reciprocation possibilities instead of just one (Hypothesis 1).

The authors point out, somewhat incidentally, on p.12 (the paragraph under Hypothesis 1) that gambler’s fallacy could be a confounding factor. I think illusion of control could also be considered. It would be good to include further discussion on these factors where appropriate. I cannot follow the rationale behind the last two sentences in the same paragraph (lines 429-431); where does “as discussed above” in the last sentence refer to?

The design regarding the alternative project is an important feature and could be described clearly in Section 2.1.1 rather than later in the theoretical analysis in Section 2.2. Also, while I appreciate that “For design efficiency, we condition the alternative project returns that investors are offered on the success or failure of the original project.” (p.10, line 332), in each case of success/failure of the original project, there were still eight outcome pairs to choose from (Tables A1 and A2). Was the choice among the eight in each case random? Could the authors clarify in the paper?

Maybe there could be more analysis on the data of the History and Transfer conditions that is conditioned on all four of the “Experienced Histories” scenarios in Table 1. That is, in the History condition, whether the first project outcome was high or low combined with whether the second project outcome (“prior result”) was high or low; in the Transfer condition, whether a transfer happened combined with whether the prior result was high or low. The current analysis sometimes seems to address them only one at a time (e.g., Figures 5(c) and 5(d)) and does not consider interaction effects, which could be interesting.

In the theoretical development, π_H and π_L are used to denote lotteries with high and low expected payoffs, respectively. But these notations are also often used to denote high and low possible payoffs within the same lottery. To avoid confusion, the authors could consider using a different set of notations for the lotteries, such as simply H and L for the lotteries with high and low expected payoffs.

There are occasional typos. For example, line 245 on p.7 has the word “managers” that should be “manager”; the first term in the right-hand-side of (6b) should have an expectation operator E. It might be better to replace the term “product” in Hypothesis 1 (and a few other places in the paper) to “project” in keeping with the terminology in the rest of the paper. Overall, it could be useful to proofread the paper once more.

Author Response

Response to R2

We are grateful for the Reviewer’s appreciation of our study (“interesting paper”, “enjoyed reading it”) and the useful comments for improvement. Our responses (in roman) follow each comment (in italics).

Comments and suggestions for authors

• The authors could try to make Sections 1 and 2 clearer: (……) it seems to me that the research is essentially about whether there could be reciprocation toward: (1) an unintentional good outcome (stroke of luck) brought about by the other party, and (2) a voluntary act of gift giving from the other party. The Introduction could be more clearly structured along these two research objectives, with hypothesis development structured in a similar way. (….) For example, there could be separate hypotheses regarding these two reciprocation possibilities instead of just one (Hypothesis 1)

The essence of our research is twofold; it concerns not only this “reciprocation” issue, but also a novel relevant context (dynamic investor-manager relationship). As argued in the Introduction, we view both novelties as important. Although we appreciate the suggestion, we have therefore kept our presentation as it is.

• Further discussion of factors like gambler’s fallacy (p. 12), and rationale behind the last two sentences in the same paragraph (lines 429-431); where does “as discussed above” in the last sentence refer to?

We changed the sentence before Hypothesis 2 and added brief discussions on the gambler’s fallacy and the illusion of control in our Discussion section. We also made the sentence before Hypothesis 2 more clear.

• The design regarding the alternative project is an important feature and could be described clearly in Section 2.1.1 rather than later in the theoretical analysis in Section 2.2

 

This was adapted in the paper.

 

• In each case of success/failure of the original project, there were still eight outcome pairs to choose from (Tables A1 and A2). Was the choice among the eight in each case random?

 

This has been clarified in the paper. The order was randomized, but each investor faced each dilemma exactly once.

 

• Maybe there could be more analysis on the data of the History and Transfer conditions that is conditioned on all four of the “Experienced Histories” scenarios in Table 1.

 

The interaction between these two variables is indeed an interesting question. However, we think that complicating the analysis even further would severely harm the readability of the paper. Additionally, testing this formally means asking a lot the number of observations that we have available in our data. Some exploratory regressions proved very tricky to interpret. In any case, what we are looking at is informative enough for our purposes.

 

• In the theoretical development,πH and πL are used to denote lotteries with high and low expected payoffs, respectively. But these notations are also often used to denote high and low possible payoffs within the same lottery. To avoid confusion, the authors could consider using a different set of notations for the lotteries, such as simply H and L for the lotteries with high and low expected payoffs.

 

We followed the suggestion and now use H and L for the lotteries.

 

• Overall, it could be useful to proofread the paper once more.

We went through another couple of revisions and improved on language and clarity in several aspects. Additionally, we corrected some potentially confusing elements that were introduced as part of the conversion of the paper to the journal layout.

 Additional Technical Changes

• We changed a lot of the formatting of tables and figures. Essentially, a lot of detail was lost in the process of porting the layout from our original manuscript to the journal layout, which we corrected.
• Some minor improvements to the consistency of data presentation (For example, a missing significance marker in table 2, spec. 5 was added.)
• We recently made the analysis code publicly available, as per journal requirements, and discovered a small inconsistency in the test that was reported to be used for Result 1. We had stated that we used a clustered chi2 test, but that was not the case. We switched this to a logit regression, with qualitatively similar and largely identical results. This leads to minor changes in footnotes 15 and 16 and the text just prior to Result 1.
• We found a missing number in the copy of the instructions: the digital version of the instructions in the live experiment said the exchange rate was “1 Euro per 35 ECU”, but in the version in the document it only said “1 Euro per ECUO”. This was corrected.
• Multiple minor language adaptions because of further read-throughs.

Author Response File: Author Response.pdf