RObustness/Sensititvity analysis

Good afternoon,

I would like to know if robustness check is mandatory in macro analysis. I have a standard growth (deterministic) model. I’ve estimated most of the substitution elasticities, and I have pined down the other parameters on SS values + FOCs.

Will a reviewer be picky if I do not perform sensitivity analysis on my parameters ? I know that this is very common in econometrics.

Thank you,

Will a reviewer be picky if I do not perform sensitivity analysis on my parameters?

I believe nobody on this forum can answer this question. That said, I think usually people put some sensitivity analysis in their appendix to show that their parameters are not crucial to their main results, or something like that. A good reviewer, even being picky, should ask you to provide what she/he wants in the second round; not every reviewer is good, for sure.

1 Like

I tend to agree with @wupeifan. You should provide some robustness analysis to convince your readers that your results are not an artifact of the particular one choice you made. However, you should not overdo it. After all, a referee’s job typically is to always provide comments in order to signal the editor that he/she is smart. If you preempt all sensible robustness checks that could be asked for, you will most likely get referee requests that are very hard to fulfill. So you are walking a fine line.

1 Like

After all, a referee’s job typically is to always provide comments in order to signal the editor that he/she is smart.

I thought a referee’s job is to provide comments to help the authors with their paper, or to help the editors with their decisions, (and more broadly, to help the economic science going forward).
In reality, unfortunately, you can see a LOT of reviewers showing their wit…

is a naive view. Rule 1 of economics is: incentives matter. What is the incentive of an unpaid and anonymous referee? If you constructively help authors, they will never be able to attribute your contribution to you. That is not to say that there are no intrinsically motivated referees. But most of the time, it’s a signaling game between referee and editor. Your job as a referee is to help the editor to make the decision. But what’s in it for you? It’s an opportunity to show that you are smart and know what you are talking about. You do this in the hope that the next time you submit a paper to the journal and now the editor is the gatekeeper for your own paper, he has a high opinion of you.

Have you ever heard about someone not getting comments because a referee thought the paper was good? No, you always have to provide comments. It happens even to the very best papers, telling you it’s not just about improving the paper.

You may find the following papers interesting.
Chambers, J. M./Herzberg, Agnes M.H. (1968): A Note on the Game of Refereeing, Journal of the Royal Statistical Society. Series C (Applied Statistics), 17(3), pp. 260-263

Gans, Joshua/Shepherd, George B. (1994): How Are the Mighty Fallen: Rejected Classic Articles by Leading Economists, Journal of Economic Perspectives, 8(1), pp. 165-179

1 Like

Thank you for the precious advice @jpfeifer

So what you are saying is that I should submit my paper without sensitivity analysis (although I’ve done it), and wait for the referee to ask me for it ?

Many thanks. I’m just starting in this “publishing” game :slight_smile:

I would always provide some robustness analysis as a signal, but would not overdo it. A personal opinion (seemingly shared by many people): I find most robustness analyses offered by authors worthless. You can summarize most robustness sections as follows: “Of course our results are robust to the variations we offer here. If they were not robust, we would never show them here.”

1 Like

That’s why I think it is very sad.