Jeffrey, S. A., Lévesque, M., & Maxwell, A. L. (2016). The non-compensatory relationship between risk and return in business angel investment decision making. Venture Capital (London), 18(3), 189-209. https://doi.org/10.1080/13691066.2016.1172748

• Jeffrey, S. A., Lévesque, M., & Maxwell, A. L. (2016). The non-compensatory relationship between risk and return in business angel investment decision making. Venture Capital (London), 18(3), 189-209. https://doi.org/10.1080/13691066.2016.1172748

Reading this thing right now. Just want to make a note. It seems like what they're saying is, you'd think that angels would make a decision about whether to invest in a compensatory way, that is, by having reward compensate for risk (i.e., it's high risk, but an equally high return), but that's not exactly how they make decisions; they do use a compensatory model, but only after an initial screening. Moreover, in that initial screening, they treat risk and reward separately. I don't know exactly how they're treating reward separately per se, but, once they identify a fatal flaw, which, in this article is mediated by the eight criteria in Maxwell, Jeffrey, and lévesque (2011) (note that it's the same as the authors of the article you're reading now), they're out. It's probably like in grading papers, if you see that the student didn't do something obvious (like the paper is significantly shorter or is in a stupid font), then you're like, "Ok, this person is going to get a B or C." So what you do is scan papers in the stack of papers for obvious flaws and then give them Bs or Cs, then you only read closely the ones in which you can find no obvious errors. Of course, this example also points to how, in  Jeffrey, S. A., Lévesque, M., & Maxwell, A. L. (2016), they're talking about how (a) there's more cognitive effort required later in the process and (b) the process gets more and more subjective, that is, the initial screening for obvious errors is the more objective part of the process.

Ok, it looks like I might have been a little quick to judge. "This article continues where Maxwell, Jeffrey, and Lévesque (2011) left off and analyzes those opportunities not rejected for the presence of a fatal flaw. In our analysis, we showed that BAs use aggregate evaluations of anticipated risk and return and analyze them in a non-compensatory manner. Specifically, that both risk and return must achieve a target level in order to not be rejected. We did not analyze the opportunities that continued beyond this point (i.e. those that were not rejected for insufficient return or excessive risk), but future research should explore the interactions beyond the rejection stage we evaluated." It seems like what Jeffrey, S. A., Lévesque, M., & Maxwell, A. L. (2016) did was, they analyzed pitches after the initial stage (i.e., whether this thing has a fatal flaw, which is what Maxwell, Jeffrey, and Lévesque (2011) did. I'm still confused. 

This makes a little more sense. "Because past work has sufficiently explored the rejection of opportunities for fatal flaws (Maxwell, Jeffrey, and Lévesque 2011), we only examine opportunities that were not rejected for a fatal flaw at an early stage of interaction. for example, rejection reasons offered early in the interaction (fatal flaws) were ‘[w]hile the product has potential, it is still too far away from being commercialized,’ and ‘[t]here is nothing to stop competitors from simply copying this product.’Therefore, opportunities that had one or more fatal flaws (a C) were removed, truncating our data at a grade of B– (score of 4). Of the 602 interactions (presentations of investment opportunities) from the four seasons, the observers coded 436 that contained a fatal flaw in one of the eight venture criteria. This left a dataset of 166 opportunities analyzed for this article." So they just split up the work. In this paper (Maxwell, Jeffrey, and Lévesque 2011), we'll analyze the pitches that were rejected. In this paper (Jeffrey, S. A., Lévesque, M., & Maxwell, A. L. 2016), we'll analyze the pitches that weren't rejected.  

Still confused. Look at the yellow part above. "We examined inter-rater reliability for these two sets of assessments provided by our observers. We first looked at their level of agreement on the reasons for rejection. In our dataset, observers independently evaluated 166 opportunities of which 67 were rejected. Of these 67 rejections, 22 were rejected for insufficient return alone and 18 for excessive risk alone. Agreement between the observers on these 40 items was unanimous. Twenty-five (of 27) investment opportunities were evaluated by both observers as being rejected for both reasons (risk and return), while they disagreed on two of them. One observer believed that these opportunities were rejected for two reasons, but the second observer felt that it was rejected for only one. from calculating Cohen’s kappa, we found that the overall level of agreement was strong with k = .937, p < .001. In order to obtain higher levels of agreement, we asked the two observers to watch the recordings again..." So why were there rejections in the rejections? were there stages?

"this article only examines opportunities that were not rejected for a fatal flaw"--Ok, what if a pitch can be rejected for something other than a fatal flaw? Having a fatal flaw is only one reason to reject something. Ok, so there were 602 interactions. Of those 602, 436 contained a fatal flaw, leaving 166. Of those 166, 67 were rejected for reasons other than a fatal flaw. "Of these 67 rejections, 22 were rejected for insufficient return alone and 18 for excessive risk alone" Thus we get into this idea of decisions being non-compensatory. The angels are evaluating risk and return separately. But that still doesn't really make sense. If you're going to reject something on the grounds that it won't make you enough money, then aren't you rejecting it because the risk outweighs the reward? 

Oooohhh. "In other words, if the risk exceeds a certain level, the return cannot be high enough to compensate for the high risk; conversely, if the return is below a certain level, the risk cannot be small enough to compensate for the low return." That's interesting. So there's a cutoff point at some point. Up to a certain level, risk can compensate for reward or vice versa. But if risk or reward are above or below certain thresholds, then it's a no go. 

I feel like this just gets at uncertainty and trust though. If someone's saying to you, invest in this. The reward is so so, but the risk is super small. That argument doesn't factor in how uncertainty could make the risk higher in the future. So this is kind of like the vesting thing. Or like the thing Ethan Mollick was talking about with founders. Why wouldn't you want to give three founders who all work the same hours the same amount of equity? Because you can't foresee the future. 

https://utexas.box.com/s/nea30p4phankcgif686r03db143few59