The Not Unreasonable Podcast

Uwe Dulleck on Credence Goods

David Wright

Uwe Dulleck is a leading theorist on Credence Goods, a class of economic good defined by the fact that consumers can't verify the quality of what they've bought. Think car mechanics or medical procedures. Or insurance!
Uwe's expertise doesn't end there of course and we talk about why people pay taxes, cybersecurity and some interesting ways Uwe has innovated on connecting academic work to real hard tech innovation. 
I had never heard to credence goods until a few months ago and it astonished me that there was a whole corner of economic thought devoted to consumers figuring out who to trust to get the right deal for themselves. It's an incredible and, in a world of ever-increasing specialization, increasingly important concept!

Show Notes:
https://notunreasonable.com/2021/11/26/uwe-dulleck-on-credence-goods/

Twitter: @davecwright
Surprise, It's Insurance mailing list
Linkedin
Social Science of Insurance Essays

David Wright:

My guest today is Uwe Dulleck, professor of economics at Queensland University of Technology in Brisbane, Australia. The subject of our talk, today is Uwe's work on credence goods. But we won't top or start there. Actually, we, welcome to the show.

Uwe Dulleck:

Thanks, David. Nice to be here.

David Wright:

First question, why do people pay taxes?

Uwe Dulleck:

So I did this work with, together with a colleague of mine who actually had very early in, like where I did my credence goods work, he started to look at this question, why do people pay taxes and his work was saying that there's something like text morale. So it's not just that we pay taxes because we are worried that otherwise the taxman will catch us. Because we haven't paid taxes. And then there's all the shame and maybe the the type of penalties that you have to pay. But more or less, you find, and we find actually similar patterns in our work on credence goods, that that people are driven very often to do the right thing. And there is a willingness to pay taxes in people in particular, if you if you can make the case for what these taxes are used.

David Wright:

And so tell me about later the specific mechanism you identify there, which is this feeling of guilt, right? Or, like it's very interesting study, because your heart rate monitoring device you put on them right, to test how they felt about pretending to pay taxes, not tells you

Uwe Dulleck:

No. So we what we are we're interested in is more or less, what's the level of stress? Yeah, that is caused when you when you consider so to say whether you should pay taxes or not. And what we found is that there are people for whom if you tell them, well, this is the text you have to pay. And we can see the behavior that they show us where they are just honest in their texts, and they're not particularly stressed about it. And we actually could find a little bit more that if you are very, like in this experiments, like simulated situation, in this situation, those people that are very dishonest. They also are not very much stressed. For them, it's maybe it's just a game. And they just don't really think about it, maybe they don't have this type of morale. And for that reason, it doesn't cause any stress to withhold the tax. And then we had the people actually that were a little bit like that paid some taxes, but not all, that showed the highest levels of stress, which is more or less showing that it's it's it varies between people. But also it causes more stress. If you're in this in between range.

David Wright:

That's that's like, it's not very cynical. Right? Whenever we're reading, you're reading about the credence goods in a second, but I read your papers on that, and a few other papers where you analyze the game theory of things, right? You have to assume the maximum cynical person because you're trying to design the boundaries of what decisions people can make. But then, what I think we find, in these instances, and in many other instances that you study, is that it just doesn't apply. Like for most people, they doesn't matter though, they don't need that stuff. And I'm wondering like, how do you as an economist analyze that kind of behavior in any kind of rigorous way.

Uwe Dulleck:

Like, for me, it's like, do feel it's a little bit a minority position. Currently, as a behavioral economist, I do actually feel this, this theory gives you a very good benchmark about whether purely selfish behavior can actually get you to good outcomes or not, like in the credence goods example, in eneral, like the work that we o there is shows that if verybody behaves like it's lmost the perfect, Adam Smith ike worlds, if everybody seeks is own benefit, we end up in a ood situation. And if we have he right institutions, right, o if we know that this is the ase, then I think we can really se a lot of standard economic nstruments. But we also show hat, again, in the credenc goods example, you see that qui e nicely. There are plenty f situations where that's not t e case. And we just talked abo t the resistance to notches a d smack. And if we would go in o policy land. I think if you lo k at current discussions, it ge s even even stronger. So pushi g people in certain directio s doesn't work either. And a l t of the other work that we' e done, done, particularly t e behavior work shows that the e are enough drivers for go d behavior in in, in us in most f us. And we can build on tha. And we should actually be mu h more open and maybe much mor. We shouldn't we could do mo e research about how we can ex m to get the right people into t e right spots to make decision. That would be how I how I s e that. So do I think economis s are cynical in their approach r is the homely canonica s particularly cynical model, I think only in the way, it s sometimes what I would ca l first year economics hord s toward their very selfish ty e of problem or approach. And y u could go further and actual y take into account that peop e care about others. Like the wo k on text honesty is an examp e where one way to look at t would be saying that peop e actually have preferences ha e like derive utility feel bett r if they are honest. And th t drives their behavior. And e should just should make t e right assumptions that peop e are not just selfish and on y caring about their own benefit. We do care about others as wel

David Wright:

The our governments, so credence goods, hich talk a little bit what hat is. So this is when you buy omething without knowing hether without being able to erify the quality, that of the ood itself, or that you even eceived it after you've paid or it. Right? And I think, and allow you to redefine it in our words, but I think of overnment services as a kind of reated, it's good for you think bout this, and ensure I'm going o give a social insurance cheme or any kind of collective ction where, you know, you're ou're you're putting money in nd somebody else benefits, ight somewhere else. That's ike you paying taxes, for ocial welfare for social ecurity for healthcare, perhaps or roads, may not your roads rom somebody else's roads, it eems to me that government ervices are kind of credence ood, true or false.

Uwe Dulleck:

Like to some extent, I do feel they are, I just want to be careful about how like we need to go a little bit more to detail that there are two aspects of credence goods that are interesting. And you cannot just one very interesting one that was particularly important in the in the what you would call the Agricultural Economics type of literature, where we talked about characteristics of goods that are not that you're not necessarily are able to observe. In general, like the way we talk about credence goods is saying, it's not just the quality of the good that you won't be able that you may not be able to observe. But it's also you're not sure about what you really need for in a specific situation. So like the example that I would like to work with, because I think it's the easiest, but we can talk about plenty of other examples, and I'm pretty sure we will touch up on them is this example of seeing a car mechanic. So if you have this feeling that your car makes this funny noise or something is not really right, you go and see a mechanic. And we all know this type of weird feeling, it's pretty hard to describe what's wrong. But we also don't know what is wrong, but we are pretty good to to know when the car is fixed afterwards the noises gun or it has whatever all the traction that it had before all the power, whatever you want to whatever you complain about with your car, when it when it is in this problem. What we don't know is whether the it was just simple repair or whether really inexpensive part or whatever needed to be replaced. So that's the type of situation we talk about, that I find maybe most interesting to think about in the credence goods example, that the viewer is somebody who buys something doesn't even know what he or she needs. So sometimes low quality will be fine. It does everything that you need, right, and maybe just a screw that needs to be replaced in your car. And it's a five cent thing and it's a five minute job. But if the car mechanic says Well, I needed to replace the whatever the muffler in your car, you wouldn't be able to tell whether the screw would have done the job or not.

David Wright:

So this so that's useful that you made that clarification because my next question is, if not knowing whether or not you got the thing in the first place was is an important feature of credence good. What was fascinating to me, is in one of your papers, you actually come across the result, where the verifiability of the product doesn't actually become an important feature and whether or not the the market outcome is good for the customer. Right? So verifiability in this kind of sense, doesn't matter. Now to me, this is like real head scratcher. So if verifiability is part of the definition of ingredients, good How could it not matter? economically speaking in the outcomes of trading ingredients, goods,

Uwe Dulleck:

yeah I think the interesting bit is like in that, in that paper, when we talk about verifiability, we talk about what you could call the verifiability of the input. So you are not able to verify whether you got a high quality treatment, good service or low quality, treatment good or service. Right. So that's what you can't verify. But you can verify whether it was successful that you were treated or not. So this is, like we talked back to, if we go back to the mechanic example, if the mechanic just tightens screw, or replaces a couple of small parts in your car, you won't be able to observe that really, right. And the more we just

David Wright:

catch up versus truly fix the problems, that would change. But if you if you put

Uwe Dulleck:

in a new muffler, most likely you could check if you would want to check, right maybe a little bit inconvenient. Like I think in that paper or or the table discussions that we had around that we talked sometimes about, you can ask a mechanic to put the replaced parts into the boot of your car. Right? And that way, you have some sort of evidence, okay, you must have replaced that because there is something in my car there. And there's another one there, right? He could still achieve, but it's much, much harder, in principle we can think about that people are able to, to verify whether they received the service. But that's different from being able to verify that the thing works afterwards. Okay, right. So in principle, once you take your car from, from the repair shop, you could have somebody else look at the car and saying yes, the problem is fixed. Right. And that's the type of verifiability of outcome that we talk about when we discuss liability. So I think the reside you're talking about is that if we have this problem, that you can't just observe the quality, we need to have a way to get the expert to guarantee that he provides the service, or, more or less gives you an outcome that that that you wanted to achieve.

David Wright:

So here we're talking about what the big result of a paper and list your six paper was the kind of the parsimonious model of credence goods, right. So what are the things that have to exist? In order for the credence good market to function? You mentioned liability. That means if somebody under treats you, you're you needed, you know, you'd heard surgery and they gave you, I don't know, aspirin, right? you sue them? Right? You attack them, because they violated the law, because they've you know, they've had a fiduciary duty that they breached. And then what is interesting about the verifiability point is that you don't actually need to verify whether or not and this is the point where you know, what you were expecting to see if I got this right. Is I in the absence of verifiability, they will always just they will always give me or they will they will sometimes overcharged me was that what it was there was there was some there's some mechanism there that I was missing, right?

Uwe Dulleck:

Yeah, no, no. If I if I have liability, then what will happen is that that you like in this article in this model, right. So it simplifies a lot of aspects that you pay more or less a fixed price or the you get a service contract. Now, I recall the in the US you have the system of health, is it called health maintenance? HMO? HMOs. Yes. Yeah.

David Wright:

Organizations,

Uwe Dulleck:

I think, yeah, yeah, exactly. Yeah. Not here not not to have made it into membership yet. But we this idea, you pay a fixed fee, and they ensure that keep you healthy, so to say, right, so in that sense, you do pay sometimes a high price when you get only a low quality treatment. And you could interpret that as as as overcharging. Like we don't explain it. We don't call that overcharging when we discuss overcharging it's more this idea that I give you something. And claim it is something in which the two have different prices. But but that's more or less the finding that we have there. If you have liability, you can actually end up in a situation where you pay a fixed price and the expert makes sure whatever he delivers works.

David Wright:

And you know, and when it comes to verifiability, though, you mentioned that it doesn't matter. The the the surprising result there was that you were expecting the lack of verifiability to have negative consequences. Right. And it didn't. Yeah, in the presence of liability.

Uwe Dulleck:

Exactly. Like I don't know whether I expected that not to happen. I think we We like I would like to move on a little bit in my work and start moving to the experimental work on that. The interesting bit is more that is the argument in this in this theory that you need either liability or verifiability. For the credenc goods market to work, th credence goods market work wi h this very homely economic ho t type of agents on both side. And the argument is slight y different. And that's what I find interesting, and I thi k shows us this power of a why w y it is important to move a litt e bit more move away from th s model of perfectly ration l behavior. In the case f liability. The simple insight s if we have liability, we c n actually put sort of, say t e expert into the driver's sea, right. Because just say, Wel, you know, you write a servi e contract, and the expert mak s sure that your problems g t fixed. But that he more or le s decides, and their econom c incentives for him to use t e most cost effective way to mo e or less fix your car or sol e your, your credence good problem. If we hav verifiability of input, so thi idea that you can observe as customer, whether you get th small or the large type o treatment, the high or the lo quality type of treatment, yo rely on customers actuall thinking about what are th incentives for the experts t behave? Right, you have a incentive to make sure that th expert doesn't make more profit out of selling you to hig quality, compared to the righ quality or to low qualit compared to the to the rid quality, and, and competition more or less than forces expert to do choose prices that kee them in, give them the righ incentives to to deliver what i best for the customer. So it, i relies a lot more on this typ of verifiability of inpu reliance relies a lot more on on on people understanding thi type of strategic interactio that's going on. It's much mor game theoretic. So to say in th way it's argued. And like, i you're okay, and I take th bridge to make the move to wor on that, then what we find i the lab study all th experiments and and I found tha really interesting at the tim where the theory says, i doesn't matter whether you hav liability or verifiability. I the experiment, we fin liability works almos perfectly, not perfectly Nothing is perfect when human behave but pretty close, lik the type of prediction work while verifiability seems t fail dramatically and seems t fail with respect to th behavior of the experts whe they don't react correctly t this type of incentives. And th same way the customers don' predict the behavior of experts Correct

David Wright:

And so the in the net result that you uncovered there, what was you know, I tried to get what just figure out whether they're at the paper, right, but the, what you discovered was the reason why didn't matter was that the providers were actually doing the right thing anyway. Right, or at least that there was a distribution of providers, and you could pick one that was going to do the right thing.

Uwe Dulleck:

Yeah, like the the pain point, or the very important thing. And I think that that's already the next step in the work. So what we found, even if you, even if the theory would predict, people should completely whatever exploited customers or the market shouldn't work at all, we still find that the market still works reasonably well, in the sense of you get about efficiency of about 60%. And that's due to the fact that I would say maybe about half of people just do the right thing. Could be a little bit more, we could talk about the like, like how we get to this 60% efficiency. So there are enough people out there that do the right thing, independent of the institutions that we have. So the baseline so to say compared to the theory, that should have said nothing should happen, is much better, because there are enough good people out there that do the right thing and depend on incentives. And it's more about what happens to the rest. Like do we get the institution so to say, to get as close to the 100% efficiency that you could have, and I think we ended up in the paper with around 90 95% which is To some extent, due to the fact that people just make mistakes. But we only got that out of this liability setting. So it did work to some extent, I think it shows that maybe customers in this situations in this in this simulated situations, the experiments that we did, were much more willing to interact when liability was there. But it also forced then efficient behavior of the experts to a much larger extent.

David Wright:

So one of the things that another result from that, I think it was that same paper, the paper on the experiment that I found really interesting was that reputation didn't matter. Now, let me pause and tell you a little bit about my perspective. Because I work in the insurance industry. Right. And, and not well, a little bit in the healthcare insurance industry, but for the most part in the property casualty insurance industry. And I came across your work, because I first heard about credence oods, like, two months ago. And was just like, what, there's his whole corner of economics hat I didn't even know existed. nd it really baffled me. And I hink a lot about trust, right? nd the return to the insurance ndustry has, is it mean? It's I hink of it, it's like a double ided credence good where the nsurance purchaser buys an nsurance policy from a company, hey don't know if it's going to ork. The insurance companies ell something to insure it, and hey don't know if that person s going to screw them. Like, re you going to file a raudulent claim? And so nobody nows what's going on? Right? here's not enough information bout what will happen in the uture. Like it's a really omplex forecasting exercise, ight? Both sides. And so it's iven me this framework for hinking very carefully about ow, how the economics of the nsurance industry through this egion's goods lens, very nteresting. And we're gonna get nto broadening the I think the efinition, recruiting is good, ecause you do that a lot in our letter in your literature, oo. But I say all that at this oint to say that the number one hing when you talk to insurance gents that matters, and I would ay this, for most salespeople s reputation. And in your xperiment paper, you make the oint that reputation actually oesn't matter if you've got hese other institutional actors in place. So maybe you ould explain that result a ittle bit.

Uwe Dulleck:

Like I find, for me, the easiest way to explain explain the reside is to think about what liability actually is. And I think in situations where you can't have hard liability, we just talked about the medical context, right? Yes, in principle, if you need some sort of more severe intervention with your heart, and not just aspirin, the doctor should be culpable. And you should be held able to hold them responsible. But it's pretty difficult to do that, right? It's a lot of money in in this type of legal cases. And I think there's a lot of people that would be avoided, would not be held liable and can avoid this type of responsibility. So for me, if you have liability reputation doesn't matter that much. Because reputation is in my eyes, just a replacement for liability. And I think the result that we have there, like I would need to go now into the details of the result is that reputation only has an effective liability does not apply. And if you don't have anything else, then actually it can be shown to be effective. And there are good reasons. Interestingly, you could even make an economical standard economic based argument that if you have a little bit of reputation, you allow different types of behavior for for most of the game. So that's probably the first big insight like in very, like, if you talk about the insurance example, it's very hard to hear hold an insurance agent liable for not providing you the right level of insurance, when you asked him for that break. It's your decision. Right. And therefore, given that you don't have liability in the market reputation is the best replacement of that. I see. And in the experiment, we we can't really see that that much because we only have, for example, a limited limited number of rounds that we can play in an experiment. While the reputation is almost in finite if you think about your industry, because it's not just maybe you have only one interaction with with or only few interactions with a certain client, but the other peoples that know about that. And you could really see there's an infinite number of interactions that matter and reputation there can replace liability.

David Wright:

That's pretty interesting. So you're saying that the presence of reputation in the actual insurance market is a kind of critique of the liability regime that's in place there, because if the liability constraints were stronger, you wouldn't need such a strong reputational element in them. marketplace.

Uwe Dulleck:

Yes. And I think that's what I would say we do see in the experiments as well, that once you have a good liability reputation doesn't matter that much.

David Wright:

I did notice in your way you structured the experiments where you simply prohibited somebody from under treating, right, and you just can't do it. Whereas in the real world, you know, there's this messy kind of process whereby you can get away with it for a little while. And so you're not going to just straight you can't strictly prohibit kind of much of anything, really, because there's always a mechanism of enforcement and your game, you can artificially make that absolute. Right. Yeah. You know, I'd actually kind of different interpretation of the of you're trying to reconcile reputation. Because another another result, Another interesting feature of the of the credence good market, is that you make the point that in order for a market to function, you have to have what you call equal markups, right? So you have to have a low quality good and an high quality good have to have the same amount of profit available for the provider. And the reason why is that because the consumers know this, and they know that they know that they understand incentives, right? And they know that if you're incented, to give me the wrong kind of advice, and I don't know, you're the expert, you know what I should do, then if your incentives are wrong, that means that you're going to, you're going to give me the wrong advice, right. So to me reputation becomes almost like a guidepost for like, I felt like it was something similar to the idea of, of grading, trust to transact. Right, you know, because you don't always listen to insurance, you don't always observe the markups that people are providing, right. So in this case, the reputation winds up as another kind of feature to make me feel comfortable to work with you. And then that was consistent with your with your findings. And the experiments were actually reputation to increase the number of transactions which were occurring. Now what you think but,

Uwe Dulleck:

you know, like, I think we are not too far away from that. Right. So what you point out is, in a lot of industries, we, you can't really observe what exactly are the markups? For that reason? We, it it's much harder to apply this this type of rationale. But what reputation really does is do did I have good experiences in the past with you, as my agent, so to say, or not? So did I find out maybe after I ran into a couple of friends, and I found out the deal that you found me isn't really the best? That would be one way to think Well, you didn't really justify the trust that I put in you into you. But what it does is it forces you because you have now if you're talking about a reputation, you have a reputation to lose, right, that's the argument here. And that gives me is what I would call is the substitute for this type of liability that I want to have. So you're completely right, if we can't have verifiability, which is I think the example you set up, we do need to do something around liability, we can't have it in a formal way, then reputation is for me the way to go. And for me, that explains a lot of why reputation matters so much in many industries.

David Wright:

Another Another feature of your experiments, which was, which I sort of was deciding to decide whether or not it bothered me was that they were they were bilateral, right? And so you didn't introduce the idea of a social reputation. So it looked to me like your reputation was I could go back to you as a provider, yes or no. As opposed to I can talk to the guy next to me and say Is he is a crook or not,

Uwe Dulleck:

I would say we have a little bit more in there. But in the sense of you do see the reputation of several people. So I think we looked at the record that you know, your own person, but we also looked at the case that you can see how other customers have feared with what was their experience with interacting with it with a different expert, like the big challenge in the lab, and that's I think, why we moved later to the field is yes, like, there we actually could look at social reputation because you just share with everybody but it's only a small group. So, how realistic that is. That is the challenge I think in the in the lab experiments you try to identify like you tried to separate one specific aspect and whether we chose the right ones or not is always an interesting question.

David Wright:

Well, I got over my problem anyway, but I think that it did introduce the the idea of the social mechanisms right. So reputation is a social mechanism. As is this as this other idea, which is yet another paper which I loved. Which actually this one is my favorite, because the the idea of different distribution In channels along with price discrimination, so that to me is a very powerful result, at least one that's perfectly descriptive of the way the world works, as I understand it, or as we know it insurance, because what you see is you see an insurance industry a very fragmented distribution mechanism, right. And if you take the interpretation and interpretation of your result there of saying, if I have to have equal markups for everybody, that means I have to treat everybody the same. But everybody's not the same. Right? The heterogeneity or the homogeneity criterion, right? That's violated, right? So every is different, but I have to treat everybody the same. Otherwise, they're all going to think I'm lying, or they're gonna, they're not going to trust my ability to discriminate their heterogeneity. And they're going to think that I'm out to get them. So now what I have to do is I have to supply them through different channels, each of which serves a homogeneous group of customers. Right? And that's, that's insurance, like insurance is amazing variety of distribution channels. And it provides a wonderful example, wonderful explanation for that. Because you need to have you need to laser out a group of people and actually have them treated the same, but not in the way that we typically interpret that an insurance is, is the regulator makes you do it. And you know, there's some there's a way you analyze the data, there's artifacts of the data analysis, we have to group people in order to minimize the variance of the result notice, but actually, it's a it's an efficient distribution outcome, because now the people in that segment trust you, because you know, you're treating them all the same. That's amazing thing. And you did you did it for computer parts, or something that's very derive that result. That was what was really cool.

Uwe Dulleck:

Yeah, like, the interesting thing is, at that time, we really had this big transition in whatever personal computing. So So, at the time, I still remember talking to computer scientists, you talked about right sizing computers with respect to the type of memory that they have. That's not a discussion today. Because it's just became so cheap. And that's exactly what I find was the powerful insight. And then I don't know whether we are able to get that into the insurance industry as well and see a similar transition there. But maybe we we can discuss that a bit more. For me, what was the powerful insight there is once this additional quality of service becomes very cheap, right? The relatively expensive Expert service, the ability to diagnose what is needed, becomes less and less important. Right. So in the computer market, I found it interesting that we had highly specialized, but more low quality computing in the computing in the early times, very much, I would say focused on, I guess, we could talk about personal computers, at one end on whatever very specific gaming application. So I don't know. I don't know how old you are, but the C 64, or the Sinclair computers, and

David Wright:

Z for Commodore, I know that

Uwe Dulleck:

the very old type, the very early days that computers somehow started to appear in our households, were definitely more low quality. And then on the other end, you had whatever the mainframe, big business computing, that was very expensive, right, and you had some sort of experts that provided this high quality end of the market. And you could buy the Commodores at your corner store, or maybe not the corner store, but you know, some sort of mass on some sort of mass market, not really with a lot of advice, right. And for me, that made sense, because in this low end market, there wasn't much much in or not enough value to actually try to save on memory or for specific, specific aspects. While at the high end market, you really, it made a big difference to get actually the type of setup that fits your needs. Nowadays, I think for for most applications, even for small business, you can buy some sort of standardized set up and it does enough, especially when we talk about the cloud, you just need to be someone connected to the cloud. And therefore right sizing of this equipment is not that important anymore, maybe right sizing of the service is important, but not all not not the equipment. So that's what I found quite interesting and powerful to see that, you know, what are the areas in the economy where this expert advice that is some more costly, actually is merited by and I would really say in the way that you have an expert that can make sure that low quality does the job that you need to be done. So you don't need to invest in in very high quality to ensure yourself that always functions the expert can say for what you want to do. The simple stuff does it right, like the good car mechanic that says well your car makes a funny noise. But I don't need to replace the whole muffler. I can just fix the one screw. And then it's all good, right? The stuff that you have is still good enough. So that's the the benefit that we get from expert services. I don't know Without we'd get that too into insurance. But

David Wright:

well maybe come back to it. But here's like a general observation that I read in one of your papers which said that up until that point, which was in the 2000s, and misses, maybe changes that the vast majority of the data and analysis that had been done on ingredients could have been done in two markets, auto mechanics and healthcare. Right. And I wonder maybe, if you, if you what you think about how limiting that scope might have been for earlier researchers, and it maybe it's changed now?

Uwe Dulleck:

I'd say it's a good question. Like, I find it interesting, when I look at, like, I still review a lot of papers that do very standard credence goods analysis, I do feel that economists are not particularly good to have a bigger picture. So very often they work on on very specific examples. And there are few people that are that that look a little bit further. Maybe it's the incentives in our market. Maybe it's the problem that we're talking about micro economics, so people are used to analyze very specific situations.

David Wright:

And then what do they miss? What do you think they miss, that you understand? Who's analyzed? I forget what market system the hukou system in China, you've analyzed data security, right? I mean, so you've ranged more widely? What do you kind of see that they don't?

Uwe Dulleck:

Well, like I, I, for me, the best example there's, like, look at the academic environment, and a lot of leading research comes out of big economics departments. Right. So if this is very different to the setup, here at Qt, so he had Qt, it's a very small economics department. And that made me and I found that, like, when I first accepted, the job heard was just economically attractive was a nice place to live. So so that was what what attracted me here. And I had two colleagues that I knew and and valued. And I thought, you know, why not? Why not do that, but it's a good offer, time to move, do something different. But what I have found and started to really enjoy is that you start talking with very different people. So I think, like, we have a local competitor here, which has a typical American sized big econ department and what happens is people just talk to other economists. And and then you you, you sit in your own silo and, or ghetto and just talk with other people with a very narrow mindset. What I enjoy being at a smaller University, where you have a couple of sales of excellence is that you start with talking to whatever information security specialist in that example. And and we started talking about how does credence good theory matter for cyber security information security systems? And you start talking about that?

David Wright:

Or what was the answer to that question,

Uwe Dulleck:

when we started to look at more or less using ideas from credence goods theory and the game theory that we have there, to improve mechanisms, like in that is, was, for example, the idea in the healthcare context to, to say that by giving people budgets of access to data, you could actually give people that are usually have a very hard time to access, patient data, more access, and more ability to choose what type of data they need on a certain patient. Then always involving whatever the the head doctor that will otherwise need to clear access. And the argument was more or less that people will know that they can only access on so much data. And if they access data wrongly, at a time when they shouldn't, they will run out of more or less budget at the end of the month. And people would then be able to identify that they actually use data where they shouldn't have had access to data. So I I found that quite interesting to use these mechanisms and transfer them into like, take them further from from just abstract ideas to think about not only observing what happens on markets, but actually engaging in you could call it market designs.

David Wright:

But what is the link to credence because there because I mean, the one that comes to mind is that of reputation, right where you know, you I know whether or not I should access this data, and the data provider gives it to me. And that's it, trust me for a minute. But if I abuse that, then I lose the ability to transact is that I got that right.

Uwe Dulleck:

The the the the example that we discussed at that time, and where this came from was, like, in a hospital setting, you have different patients, patient data is protected by privacy. But we do know that let's say the nurse that is looking after you, for some for administering, for example, certain medications, it would help if she gets access to some of your medical record. In most cases, the nurse itself can do that. And it just gets the order so to say, or if it would want to access the record, you will need whatever the head to the chief, head of the department to clear her from seeing that, that seems to be a very simplified version of this type of setup. And the argument that we made there is saying, playing game theory, applying the credence goods argument, the nurse and service in the best position to judge how much access of data she needs. Like whatever the head of department would be able to know that as well, whether she should get that or not, and we have this checks. But it's costly that he actually or she needs to check on the nurse and what is needed there. Instead of saying that we could say what do you have a certain capacity to access data. And we know let's say in a normal work week, you need five times high private access, high privacy access, and the rest of the time normally access. If we document that we could say, well, if you don't go over your five times higher privacy access, you'll get a automatic waiver. And they will make the system much more effective. And it is this thing of of legacy, as you pointed out, people that need a certain type of data are the best judges about how much data they need in this situation. The moment we have this, a lot of checks, not really checks and balances really mainly checks that say we put it very high up in the chain of responsibility to clear people's access. That's very expensive, if you think about, like this head of department needing to grant access all the time. And the question is whether it actually then happens because becomes then more or less automatic. Right. So the checking may not even happen as we formally haven't.

David Wright:

So to meet To me, that's like a that's a pretty interesting expansion of the of the tools. Right. You know, I think a little more prosaically, credence goods are appearing verywhere. I mean, look at look t it online, ordering of nything at all, I mean, you uying something you don't know hat you're going to get, and ometimes it ain't good. You ave reputation systems, right? gain, you have you have free eturns, liability. Right? So it eems to me like I remained affled at how little known the redence good line of thinking s, given that it's massive loss f importance in our society. nd I think I have a story too, or kind of like what's going on here. I think that the ncreasing specialization of our conomy requires narrower or arrower fields of expertise of reater and greater quality. And o we were gonna wind up having o go into all sorts of more pecifically, trained experts or kind of almost anything, ook at the medical profession. mean, you don't really see eneral practitioners so much nymore. It's all these deeper nd deeper experts. And I feel ike a lot of professions are oing in that direction. So to e, they all become more and ore credence, like, is that ight? Like,

Uwe Dulleck:

I would argue, it's very, like I would make very similar arguments there. Just one thing that I want to take on and it makes sense to think about that, like, early in the theoretical work on creden e goods, we talk about this difference between what I would call standard goods like let's say your milk or your fuel you buy, right so I sometimes say the example when I talk to my students or the goods You are a you're happy that your mother buys for you and or your your, your parents or whoever you are IP to delegate right standardized goods, no problem. Then you have search goods, maybe clothing that we buy, because you want to look at it before you buy it, but you can look at it. And that's really changes with with the internet. The next thing that economists talked about his experience goods and I just wanted to just quickly talk about that because I think the internet is a good example of an experience good. You more or less need to make the you need to trust The expert, not the expert, the seller, in the sense of the quality that is advertised on the internet is actually the quality that is delivered. So reputation they are matters, right? Same like in the literature, and it's not a new concept we like if you buy a bottle of wine, you just see the bottle. Even if you buy it in a shop, as long as you can try it there, you have to more or less make this first trusting decision to say, Well, I trust the label, or I trust the seller, that it won't be complete. Like, unbearable vinegar.

David Wright:

Yep. swear if you'd like but yes, and,

Uwe Dulleck:

and, and then you like it's more or less repeat business that really makes this work. Right. So you buy a certain brand, you like it, or you don't like it, if you don't like it, you just don't buy it again. And you somehow hope that disappears. That's the the experience. Good example. And we started out this discussion and talked about how agricultural economists think about creating schools, and they have this type of perspective. That's, it's a lot about, I can't observe the quality. The important thing is it's also what do I need? Right. So if we think about buying clothes in a shop, there are different quality levels, let's say you look for some sort of function where to go on a on a hike in summer, and you would have the expert being able to tell you what is needed for the type of trip you do take. And that's much harder to provide, I guess, over the Internet, but it can be done right. So so that's for me that the interesting thing where we have this differences between the goods that it's not just the quality where you need to have trust, because the channels have changed, and have become maybe more anonymous or reputation becomes also harder, or more difficult. But it's also this what is needed to identify, like, to what extent can you actually identify what is the need of a customer, where I think that's important. But I do feel that like the word gets more and more complex technologies become more and more complex, and makes it much harder for all of us to actually choose what is really the option that matter that matches our needs. So I think we do have more and more specialization that is needed that where we rely on advice. Yeah,

David Wright:

I think that So to me, that specialization and intermediation kind of go hand in hand, right. So like, if you have where once you had sort of one or two generalists, now you have a whole plethora of specialists, you need a guide, right? You need somebody to say, here's all the specialists hold my hand, I'm gonna tell you where to go. Or I'm gonna tell you which ones you need. And then, you know, we get the efficiency of extraordinary expertise among specialists, but we have this, we have this need for that, to interpret that for the rest of us. And now we need that we need advice. And now we have, you know, in other regions good pops up. You know, I'm long credence goods, who've been in y intellectual forecasting of he future. But on that note, I m intrigued by one more, you now, we'll finish up on this opic. But one more strand of our work, which is very recent, ery recent, which is on valuating the intersection etween academic literature and atent research. And so to me, ike, I think of the credence ood angle here as being, what o we get for all these rofessors that we are hiring. nd there's like an interesting echanism of output, which is atents into the economy. And hat can generate real outcomes or us in terms of economic rowth, which is nice. I'm ntrigued by what you think of hat, whether this becomes a, a ind of evaluation of the output f academics in the sense of ike, evaluating. They're the nes who are on your list of cademics who publish papers hat wind up in patent eferences are more productive n some sense than those who do ot. Is that to bold?

Uwe Dulleck:

Like my interest or our interest, there was slightly different at first, to make the connection between the discussion that we just had in the patent literature, or academic output in general. I think it's interesting to see that this type of literature actually requires experts to make it accessible. Right, so for me, the interesting thing is when I started to have these discussions with with a co author when we are co authors that are on this paper to look at the connection between the academic and the patent literature and Was that for a long time, IP was actually not very well organized literature so that we had the the type of progress that you have more or less now, Google's four patents, search mechanisms that can actually, and that's what I think a lot of the type of church search algorithms and programs that we have now have done for us, they were able to actually organize a very unstructured literature. So the way I think about IP is it was even worse, much worse than the academic literature because it wasn't very structured. Like I always, if I talk to colleagues, and I say, you know, the way to think about the Payton the knowledge, the paitent, knowledge before the arrival of women modern library search techniques, it's it's all of this was like, like really, like a big archive that wasn't particularly well organized, more or less by time. And you couldn't really navigate the knowledge well, and that was used to some extent by, by large corporations, right. So if you think about the pain and knowledge, which gives you a right to exclude others, you actually don't want to know. Or you don't necessarily want a competitor to know what you can exclude the competitor from, because that way you actually reduce competition. So this uncertainty for some players in the market made a lot of sense. It also led to giving a certain type of experts a lot of power. Namely, if you can navigate this type of Yep, not very well organized. information, then you have a lot of power, right? Because you can tell people what's happening there, but a big investment not very efficient. So so the type of mapping a mapping exercise actually came about, how can we bring the right experts together with the people that want to apply this type of knowledge? Right, so that's how I thought about it. So the the mapping and trying to understand where, you know, what are the the the the academic researchers, what are the industry, the people that are able to turn knowledge, ideas into products, services, commercially valuable outcomes? How can we bring them together? That's much more the way how I see that. So So connecting the right expert to the right problem. In many ways, I'm like I can see and I think it's important for from a society perspective to actually have a better idea about what our what our contributions to the advancement of knowledge, and I agree that part of the work is really there to say, well, we should be cautious like we talked about, you know, what is the academic environment that brings you to interesting ideas. And we have the problem of the ivory towers and and not just one ivory tower, several ivory towers that write their own game. And I'm definitely as we discussed before, I'm I'm very much into getting people more out of their own little corner and interacting more, because I found myself on a much more exciting, I also believe it's much more useful from a former from from a social perspective. But as the credence good person, I am really reluctant to push harder for using this as a as a way to evaluate the academics because I find a problem. The problem is that the type of advice they provide may be different for that reason. If they're too much driven by the incentives,

David Wright:

indeed, so I was intrigued, just but as a kind of, like poking around in there, and when you're on there for your cybersecurity work, which is super interesting. Dude, you know, I would say, as an academic, you're certainly, you know, most well known for creating this good literature. But then here you are this tiny little kind of strange detour you made. And now in the world of patents, that's who you are. You're the guy who helped design this interesting, kind of like gating mechanism for people to get more data. And yeah, I was looking for all sorts of economists on there. And I found, you know, a phase black of correlation completely with the, you know, the ranking of citations of economists and patents, right. So that's kind of interesting. What are you really doing out there? And then, you know, if you go to venture into more liberal arts, like I put in a couple of philosophers and stuff and there's stopping, right, they didn't they weren't cited any patents. Daniel economists and no 25 or so which that feels about, right, in terms of be real economists. So I mean, I was intrigued by the idea of this sort of like, I don't know, completely distinct and independent evaluation and quality of academics, academics, output. And and so like, in some ways, I imagine a lot of academics would not like it, because they wouldn't feel as threatening to their expertise, perhaps, I'm not sure. But it's been the reaction to that.

Uwe Dulleck:

Like, I it's an interesting question, like the economists don't really care, because and I think one of the reasons is that, that maybe in general, with the social sciences, a lot of the work we do is producing for the commons or for the public, public goods, right. It's very abstract, I think, very often very useful knowledge. But it's not knowledge that I feel is particularly well protected by, by whatever the IP type of setup. Like there's a big discussion in the patent literature about the business methods, Payton's. And I personally think it doesn't make much sense to protect business methods by by patents, because there's not like it's very often ideas, they change relatively quickly. The IP system is not particularly fast to react to this, this type of thing. So what you actually have is when when the most of the benefits of a certain business method have been developed, the system hasn't yet come into place. And then you protect something that is more or less legacy protection doesn't make much sense for me. And for that reason, I think a lot of the economists don't care about it. I do feel and that was the the push that we tried to do with this work is that there's a lot of data, there would be interesting to study from an economics perspective that we currently not, that is not studied, at much depth. There are pockets of research in this field, but it's Yeah, I feel it's an under under researched area. And it came out of work where I had discussions with a colleague of mine that was building this this patent database, the lens. about, you know, what, what is the economy about that? How do people engage with this abstract knowledge? To what extent are experts in this field patent lawyers really helping what happens there To what extent do a hindering to actually get innovation out to reach its full benefit?

David Wright:

So last question, what are you working on next? What's the next frontier credence good for you?

Uwe Dulleck:

Like, for me, we currently do a series of papers where he tried to find out, we talked about the role of experts to try to find out who actually makes the decision when you have this type of credence good situation. Right? So I think in the medical context, it's most obvious. If you have a certain health problem, do you actually feel that you at the end of the day, decide about the treatment always at the doctor? And in the early papers that we have a couple of these papers out? Now? What we find is that is not clear. It's definitely no agreement there that both patient and doctor believe that the patient makes the decision, which would be some are what we assumed from an institutional setup and from a legal setup. And I find that fascinating and interesting. So who actually is in control? And there are lots of implications that we'll have from from a legal perspective as well. But it becomes difficult, right? It's like, if you look at the economics literature, it's a little bit this type of discussion about formal and real authority. So who can make actually a decision? And to what extent do feet like to feed people rarely feel they just received advice, but then make the decision or they're delegated the decision in the credence goods context, that's, find a very important uestion. And it really applies hrough all these different pplications of credence goods. W rk.

David Wright:

My guest today is Uwe Dulleck. Thank you very much.

Uwe Dulleck:

Thanks, David.