Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. ======================================================================== # Managing Innovation in Firms Authors: Ingrid Haegele, Isabela Manelici, Jose P. Vasquez, Francis Wong Discussant: None Video: https://www.youtube.com/watch?v=7LDwpGoPNFc&t=21297s ## Talk (05:54:57 – 06:34:40) [05:54:57] >> Um well thank you very much for the organizer to the organizers for including our paper in the agenda. This is joint work with my amazing co-authors Ingret Hagul from LMU Isabella from LSC [05:55:10] and Francis also from LMU and this is the first time the team goes into something related to economics of innovation. So we look forward to to hearing your your comments. Um most of the innovation happens in firms at least [05:55:24] threequarters of the R&D expenditure happens in firms not in universities or in government labs etc. [05:55:32] And among those private firms, large firms account for the vast majority of R&D expenditure. For instance, in the United States, multinational corporations, which are basically the [05:55:44] all large firms, account for around 90% of business innovation investment in the United States. There's a large literature studying the effects of this knowledge created by innovation and how [05:55:58] it diffuses. Some papers study how it diffuses across borders or across affiliates within the firm. Other papers study how this innovation knowledge [05:56:12] diffuses to other firms in local economies through knowledge spillovers. [05:56:18] But in general, we know much less about how this innovation is produced and organized within the firm. So this innovation just is not exogenously there. it it has to be created uh the knowledge that could be potentially [05:56:31] diffusing and we know much less about that. What we know is that these large firms are complex organizations and the production of innovation is is going to require a lot of coordination between many people. So in this paper we are [05:56:46] going to use detailed data from one of Europe's largest manufacturing multi multinationals and we're going to study how innovation is organized and produced in large firms. We see this as a first step in an agenda and in particular [05:57:00] today we're going to study what is the effect and the role that managers play in this innovation process. [05:57:07] As we know, innovation is hard. Here all in the room, we're trying to create new stuff. And this is not like producing nuts and bolts. Innovation is highly uncertain. We start a project today. We don't know if in a year or two years [05:57:21] from now we're going to have a draft or maybe a submission, let alone when this uh submission is going to be published etc. It it's also a process that is very knowledge intensive and in particular the type of knowledge is not always easy [05:57:35] to codified. For instance, even if you ask a very successful researcher, they cannot give you like the recipe that you always need to follow from brainstorming to publication at a top five, right? So, it's it's hard to codify a lot of tacit [05:57:49] uh knowledge embedded and it also requires a lot of cooperation. Sometimes you're co-authors, you receive feedback from your other peers, from advisors and in the private sector in particular, uh feedback or opinions of of managers [05:58:02] involved. I'm going to give you an anecdote from Gary Starkweather who's known to be the inventor of the laser printer and he gave a very interesting uh interview at the New York Times where he said one morning I just woke up and [05:58:16] had this idea why don't we just print things directly with the laser and he talked to his manager at a time and this manager thought it was the most stupid idea ever in the world. So he was discouraged from engaging into this [05:58:28] project. It was not only um after years after that he had a new manager and this manager actually was supportive of idea. [05:58:37] The idea was included into a prototype and was in the end uh produced the laser printer which was one of the main revenue uh sources for Xerox. And what I want to say about this is that this is not only an isolated example there are [05:58:52] many others but in general innovation managers have a say in innovation. They can encourage or discourage projects. [05:58:58] They can screen early ideas. They can assign resources to different projects and usually they are also involved into into the technical aspects of it. For instance, in the interview uh he was saying that the reasons why the manager was not supportive of this idea were [05:59:12] related to like the difficulty of of printing with laser etc which are like technical aspects. [05:59:19] So in this paper we want to study how the managers affect innovation and we're going to focus in particular on the role of having hands on in innovation experience in the past. So what I mean by this is like having experience [05:59:33] turning ideas into innovation and and this is correlated but not fully with having credentials such as PhD, seniority at the firm, the rank in the hierarchy, the number of years of [05:59:45] experience, tenure etc. Um the setting that we're going to have is one of Europe's largest manufacturing firms. [05:59:53] Unfortunately, for this talk, I cannot give you the name. So, they're happy to disclose their identity to editors and the journals at the end of the publication process, but in the meantime, we we cannot identify the firm in formal or informal settings. What I [06:00:07] can tell you is that the firm is very large. It's headquartered in Germany and it has over 200,000 workers around the world. From this firm, we're going to have invention disclosures. Uh, I going to tell you more in detail what that is, [06:00:22] but to think about it as research proposals. So, workers need when they have an idea that could potentially lead to a patent, they are obliged to write like a research proposal. And this is what we're calling invention disclosure. [06:00:35] I'm going to tell you more about it later. And we can observe these invention disclosures, which ones were filed for a patent or or granted a patent. and we can observe them until which ones are used for commercialization of any of the products [06:00:49] by the firm. We're going to merge this interesting data with personnel records of the firm where we observe a bunch of uh demographic characteristics, salaries, etc. Uh we're going to have the hierarchy of the firm. So we can see who is the boss of whom, who's managing [06:01:03] which team and we also have access to performance evaluations for workers and managers. In terms of empirical strategy, we're going to rely on manager switch switchers design which is a [06:01:16] design that is getting increasingly more popular in the organization and economics literature and we're going to discuss a lot about it in terms of causality etc. Our findings are that having hands-on inventuring experience, [06:01:30] it seems uniquely predictive of the per of of the performance of the team in terms of innovation output compared to other variables that we will discuss. [06:01:40] If you if a team switches manager from one with zero in invention experience to one in the top desile of invention experience, the number of patents of that team almost doubles. [06:01:53] And this includes patents that are in independent inventions. These are not driven by only co-authorships co-inventions with this with this manager but also independent inventions also increase. And then I'm going to [06:02:07] show some evidence about what we think is the key mechanism that these managers with invention experience they improve the invention capability of their workers. [06:02:18] Okay. So a little bit more about the context. It's a large firm as I said headquarter in Germany. It produces sophisticated high-tech uh intermediate inputs. It's not among the group of superstar firms. It's not like very [06:02:33] weird. We can compare it with with the top 25 multinationals in the world and with the top firms multinationals headquarters in headquartered in Europe and they start similar in terms of like [06:02:46] R&D expenditure patenting uh the workforce revenue that they have uh so it's compatible with other large firms and for for this firm we're going to focus on engineers who are based in [06:03:01] Germany in part because some of the interesting variables that we want to use are only available for the headquarters country and if we want them for other countries we have to go affiliate by affiliate just asking for it but for today so we're going to focus on on engineers in Germany and these [06:03:15] engineers they um are producing most of the innovation at the firm they work on client facing problems where usually clients ask them for stuff and maybe we can add this feature to this product and [06:03:29] then they work on it as part of as part of this work they realize that they need to do something new in order to satisfy this requirement and then this could lead to a patent and and that's how [06:03:42] basically their their work operates. Just to give you a sense of some some quotes. There's a a platform at the firm where workers self describe the tasks that they engage on a daily basis. And basically just some quotes from what the [06:03:56] engineers say is just just basically they analyze customer needs and trans try to transform them into sol solving these pro technical problems and try to transform them into something new. We also have information about what these [06:04:10] managers say and just to say that these are special managers. They are not the type of manager that you could expect managing some uh department of human resources at the firm but actually they [06:04:22] self-describe as their tasks as leading teams of scientists and things like that. So many of them have technical background in engineering. A very large percentage have grad school uh masters [06:04:35] or or PhD and this is not something only special about this firm. We team with EO and they allow us to ask to add a couple of questions in some survey that they ask [06:04:47] to large manufacturers in Germany and actually it's fairly common that managers uh have direct involvement in the innovation process at at large firms. [06:04:59] So okay yes >> in many cases so they in around half of the invention disclosures and patents etc they have the manager as a co-inventor but no there there's also [06:05:13] like the other half the in which the managers are not >> sorry another question of the context are they incentives for so managers get bonuses and promotions if they >> so there there are both career [06:05:26] incentives and monetary incentives around six in some survey that uh we applied there's around 60% of the managers think that inventions are part of their important things for them to be [06:05:41] promoted but there's not only that there are other like basically what they care about is to solve the problems that their clients have and invent if something like that works as an invention then good news and there are [06:05:53] some other monetary compensation but at least in interviews people say that this is not the main reason they care about that actually it's not that big of a deal in terms of the amount of money. [06:06:04] Okay. So yes please >> so we filter by all engineers and [06:06:16] actually they uh 90% of all the patents from the firm come from those engineers. [06:06:23] No, but are there some >> it's just that this is something we're we're uh trying to deal with in [06:06:36] R&D is very decentralized at the firm. So they are all encouraged to do innovation but innovation is not the only thing they have to focus on. So it's not that we can say we can focus on these engineers for which the whole job [06:06:48] description is to do new stuff but it's part of their job but some of them they they do many things >> something [06:07:08] shut up. >> Yeah. Yeah. I don't know in this case we we b we're currently filtering on engineers in general so in that case it will include the adjun faculty uh we have this task descriptions that are [06:07:21] self- description of what they do maybe you can filter on the only on those who describe the having innovation as their only task or most important task okay so as I told you from this firm we have around 43,000 invention disclosures what [06:07:35] is that you can think about this like a research proposal so at the moment one or more workers they have an idea and they think that this idea could eventually lead to a patent they have to [06:07:47] register this idea in a platform of an online digital platform of of the firm and this happens for the main reason is that this codified in the law so there's a law in Germany that requires [06:08:01] researchers and innovators to actually register with the firms these early stage ideas and why do they do that from the point of view of the workers. This is useful because it creates a a paper trail. If that idea eventually becomes a [06:08:16] patent or there has some royalty payments etc. they receive some compensation. Those workers who are in the initial invention disclosure receive compensation. This is also useful from the point of view of the firm because this allows the firm to allocate [06:08:30] resources to prototypes to know what each of the engineers is working on and to assign grants. So and and then we get these 43 research proposals. Some of them become a patent application. [06:08:43] Actually 58% of them. They remain there. They are just released in the sense that someone at the firm, the manager or the patent lawyer realized that this is not worth pursuing. This process between the [06:08:57] invention disclosure and the patent application is around eight months. [06:09:01] Then 30% of the initial 43,000 uh they receive a a patent. So they are the patent is granted and that process from application to patent granted is much longer. It's around like 50 months [06:09:15] and then as very small share they're using commercialization and we know this because we have a dummy that the firm provided to us that includes patents that for which the firm receives some [06:09:28] royalty payments and also patents that are directly used in some of the firm's products even if they don't receive any royalty payment from it. And as you could see only 3% of the initial pool of ideas are eventually commercialized. At the beginning, we thought that was [06:09:42] little. We don't have a benchmark. If someone has one, please share it with us. But uh talking to people, it seems that many of those patents uh that they've submit filed and they are eventually granted. Uh they're that may [06:09:56] be defensive patents. Maybe because the firm might use them at some point. They don't have a clear idea how to use them at the moment, but just because that's enough reason to to pursue the patent. [06:10:04] Yes. not is it all low quality or is that a problem? [06:10:15] >> Unfortunately, we don't I don't have an answer for that. Um I I don't even know how to approach it. It's kind of we don't have the text because there's a lot of secrecy in the what is included in these invention disclosures. Uh but [06:10:29] yeah, what we know is that the firm someone thought it was not a good idea. [06:10:33] Probably quality has a lot a big component. [06:10:37] Okay. So because of the law compliance of disclosing these inventions is very high. So every time someone starts working on something boom invention disclosure there with the research proposal and then something for which we [06:10:49] only have anecdotal evidence but uh is that only substantive contributors are listed. This is something that when we talk to engineers they think this is very natural and actually when we think about our profession that's also kind of [06:11:04] natural. you're you're working on a paper not not because you like you might like very much your colleagues at the office but not because of that you will add them for free into your paper. So the way they think about is we we try to push them a little bit like are you sure? Are you sure you don't add your [06:11:19] friends? Are you sure you don't add your manager just because and they say no no why would I do that? I'm working on this project. I will add the people who help me. Um so I don't have against again hard evidence but anecdotal evidence says that only substantive contributors [06:11:33] are listed in this innovation disclosure. Okay. So very quickly I already told you about the invention data. We merge this with personal records including demographics, position at the firm in the hierarchy. We know who's the manager of which team. Uh we [06:11:47] have some other information about hours, salary, bonus payments. We have the organizational structure and we have performance evaluations for both workers and managers. [06:11:57] Okay. So the objective that we have is that we want to have a measure of hands-on invention experience to see whether managers that have this additional knowledge or experience they they are able to foster the innovation [06:12:12] performance of their teams. The measure that we have as a baseline is the average number of inventions per year. [06:12:18] But we have tried not the average number but the cumulative number of inventions and then by inventions we tried with the disclosures with filings with patents granted with things related to citations [06:12:30] and and and results are very robust. Um what we want to capture as I said at the beginning is to is to have some real experience in the innovation process. [06:12:39] You can think about it in in our fields as having someone who has published a paper versus having no experience in the publication process and this is something that is not fully correlated with all the measures of having as having a PhD the number of years at the [06:12:54] firm etc. And there is very a healthy amount of variation that that would allow allow us to to estimate uh to put into our empirical specification. So we are going to rely on on manager [06:13:07] switchers design. As they said this is something that is getting increasingly more popular in the innovation in the organizational economics literature. [06:13:16] Some characteristics that happen in this firm and it seems that in other large firms as well. First that these switchers are routine. So in our data we see that every year around 5% of the [06:13:30] managers rotate to other teams. The main reason the firm gives is that it's related to career considerations of these managers in the sense that once someone gets into the managerial track, the firm wants these people to have experience in different teams and [06:13:45] departments to get a big picture of what the firm actually does and not to be fully concentrated only in one area. [06:13:52] What we could do is that we will have some uh balancing tests and trying to see whether these movements are targeted because that's a big threat to our identification strategy that actually the firm has some superior knowledge and [06:14:06] we'll assign the good managers to the teams that are going to just explode in productivity. [06:14:12] uh what we can do with our data is we can see that previous characteristics of the team they don't predict what kind of managers they they will receive when when there is a switch and also we do balancing in terms of the workers [06:14:25] workers that experience a changing manager to one high experienced manager versus a low experienced manager they seem to be also similar in terms of characteristics and then we also have some evidence that it's hard to anticipate it's not that [06:14:40] managers always switch after two years. So after three years there's a a very smooth histogram uh on on the time and the number of years that manager stay in a team. Okay. So butical specification [06:14:54] we're going to have an event study design where our main variable of interest is going to be this delta E. [06:15:00] This is going to be the difference in in the invention experience that the new manager has minus the one that the old manager has. So this change in inventuring experience is what will be um we will be exploiting uh to to have [06:15:15] our event study. Yes. >> This one will be workers also switch >> workers also switch >> and so that new man. [06:15:27] >> Yeah. Yeah. So in our baseline sample, we're going to basically restrict the workers to remain as to be in the team four quarters before there's a switch of the manager and to remain at least four [06:15:41] quarters after. Uh there we have some other specifications where we only use incumbent workers uh and then we follow them wherever they go. But uh the results seem to be robust. But yeah, I agree there could be overlap. We're [06:15:54] basically precluding that overlap from the worker side and only having the manager moving around. Okay. Um our main identifying assumption is that absence the switch workers that are exposed to [06:16:07] high to big changes in managerial experience in a positive sense they would have trended similar to to workers who have a a low manager um [06:16:19] switching to their teams. And of course, this is going to be the big monster that we need to fight in this paper. And I'm going to try to provide some evidence and and some robustness checks. And we're we have done many of them. We're [06:16:32] happy to to take more of your suggestions if you have any. Um so the first one is that as you would see there's absence of pre-trends. So at least and what I'm going to show you later. So at least it seems that these are these teams that receive high experienced manager versus low [06:16:47] experience experienced manager were not trending upwards or downwards. Um as I said team characteristics cannot predict the type of manager they're going to receive when whenever there is a switch and worker characteristics they also [06:17:02] seem very balanced between teams that receive a high experience manager versus a low experienced manager. We do do a lot of cuts in the data and add all this fixed effects uh that that we have at hand. And then talking to people they [06:17:15] they were telling us that maybe the senior managers they have a say on and where they go when the firm asks them to switch. They have been in the firm for a long time. They know the firm better. [06:17:25] They know where they are going to be more comfortable. But junior managers usually they know less the firm and also they're more willing to go whatever the firm wants. So uh we also restrict the the sample to managers that are junior [06:17:39] versus senior and the results are very similar to internal switches in the same plant or versus across space or this different divisions. The results are also similar also lateral moves versus [06:17:51] moves that in that come with a promotion and also the results look similar. Okay, but now let me just show you what we find. So in this case the outcome is going to be patented inventions. Yes, [06:18:04] please. I was thinking that you could do something like on the lens paper. You kind of have a ranking of managers and teams pre switch and you can assume like the best manager will take the the empty [06:18:18] position kind of controlling for positions who should move the ranking that >> yeah we don't have that that's systematized in the exact same way but we could see whether good teams based on [06:18:31] preexisting characteristics will be matched with the characteristics of a good manager and it doesn't seem to predict Having a high ranking at the team level versus the doesn't seem to predict whether the manager is going to be high ranked or lower ranked if we [06:18:46] rank them at least in terms of like invention output before the match before the switch. Sorry. [06:18:52] Okay. So we have this event study designed absence of pre-trends. One thing to say is that the way we put inventions here even if the regardless of the outcome into this timeline depends on the date of [06:19:06] disclosure in you know invention disclosure. So if we look at quarter number four these are like the inventions that were disclosed four quarters after the switch regardless of when the patent takes three years or [06:19:20] five years to to become eventually patented. [06:19:24] Um what if to give you an interpretation of the magnitude in this case if you go from no experienced manager to one manager with the top in in the last decile of invention experience the [06:19:37] number of patents roughly doubles at the worker level. We can do that with all inventions. So this includes all the disclosures regardless of whether they were dropped and never filed or they were filed and never patented. If we look at the interpretation, the same [06:19:51] change from no experience to the top decile of experience of the manager will improve the productivity in terms of all inventions of the workers by around 72%. [06:20:02] And we take this as evidence that there's fine there's an increase in quantity but increase in patented inventions was roughly double. So the quality increases a little bit more than quantity. [06:20:13] >> What what do you do with the manager? I'm sorry. So like are the managers still there 16 quarters later or you have another new manager about that? [06:20:20] >> So we have we we this is a discussion we have had and we had those several things in the baseline we only take in the in the data the teams until the quarter the manager leaves in that case they [06:20:34] disappear. If the manager still stays only four quarters then the this team would enter in the data only four quarters and so on so forth. In some of the cases and we also look at in some of the cases we just allow the managers uh [06:20:48] to leave and and then without trying to balance the data or anything then we just continue with with the same timeline even if now in quarter 14 they had a reason and and there is in the baseline this is [06:21:02] what I'm showing you where we only restrict for the quarters in which the they remain with the same manager the moment the manager rotates to another team they the team drops the sample but the results look similar in the other uh [06:21:14] specification as well. So is your sense then of this increasing effect size is that coming condition on the longer survival wondering about selection of stay in the [06:21:31] team when it goes well versus moving on and that's why you see this >> you mean the manager >> so can can you rephrase like selection in which the only the manager who's [06:21:45] you know It is the same manager like the the estimated things for 12 to 16 only have like people who are there for some time they might stay because it's going really well. [06:21:56] >> Yeah, I agree that's a concern and that's why we also do the specification in which >> saying it's a concern but it would be nice earlier while staying for a long time. [06:22:07] >> Yeah. So in this is a little bit attenuated uh if when you we do the comparison that you have in mind. Uh but the results I mean we cannot reject significance significantly different results. [06:22:22] Yes, like you know those kind of car handing in client type uh >> AKM >> where they where they show the by you know for the match like you know is it about a solid is it >> good managers with good teams good you [06:22:36] know like four quarters you want to see those symmetries like >> like getting a bad manager has the same symmetric effect as getting a good manager or >> yes uh we haven't done that specifically if you look at here there's some [06:22:50] symmetry embedded into this specification. What we did is that we could split it into teams that receive a manager that has more experience than the previous one versus team that had in this case is a positive change versus [06:23:04] teams that receive a negative change. And if we do the split of the samples, it seems that this fairly symmetric. So that's why we were comfortable with this. This is not the same car hine inclined type of analysis, but I we can [06:23:18] go a little bit more in that direction. Okay. Um let me see. We can see the number of citations on the left just citations but we know that older papers get cited more because simply they have [06:23:32] been more around for longer. So we can also normalize by uh this citations of other papers in inventions in the same cohort and still seems to go up. Um we can look at all inventions file inventions patenting inventions and [06:23:45] commercialized inventions. As you could see the coefficients are always statistically significant in the post but not in the pre. The magnet seems to be going down. However, when we rescale by the outcome mean in terms of percentage changes, it's actually seems [06:24:00] that as you move downstream in the pipeline uh in the effects are larger. [06:24:06] So effects are larger for commercialized inventions than for all inventions for instance. [06:24:12] >> It's always measured the same way. >> Yes. Yes. Yes. Here we have some other alternative. Oh yes, please. [06:24:21] >> Slide. I mean that's a bit interesting because you would have thought that um you know like the local level treatment of like the marginal uh inventions that they're if it's a really great invention [06:24:34] then all managers should have kind of approved it and said go ahead like have it you know but but it's like the the more liberal managers are even allowing bad invention so bad ideas to kind of go [06:24:48] forward right like >> I'm I'm not sure because I wouldn't say that all managers would approve a good invention because actually a lot of it these are early stage projects and I think a lot of the talent that those managers with experience have it's [06:25:03] actually to screening to screen which of them are going to be a good invention or not. It's like the advisor who receives the students and they pitch the idea and then maybe they some some of them have a good smell test and they would say oh no this is these are the ones that are going to be good for us. This is kind of [06:25:18] part of the mechanism as opposed to say some are so obvious that even the uninformed managers are going to push. [06:25:26] >> Yes. I would change size. [06:25:33] >> So the team size as an outcome. >> Yeah. Yeah. Yeah. [06:25:37] >> Is it the case that like James or >> I I have I have this later on there's a non-significant effect on team size because of the manager switch. So this [06:25:50] is we were reject we were actually thinking of that about a mechanism in which this in good managers bring resources and therefore expand their team. we find no effect on team size, no effect on the networks of their uh [06:26:04] supervises. So they don't start co-authoring with more people or with more connected people. Um we can rule out those things. Okay. So I have to move a little bit faster but here just to tell you that we have the same [06:26:18] regression but as a in terms of the main explanatory variable we could do changes in invention experience which are our baseline versus changes in other things like changing only tenure at the firm. [06:26:30] So receiving a manager that has spent at the firm more years versus fewer years no effect where the manager executives where they have a PhD actually it seems to be a little bit lower. uh uh and [06:26:44] these are and these are controls that we have in our baseline specification. If we look at experience supervising other inventions uh we thought this would be also a big thing but it seems no. So even if those managers have experience [06:26:57] have experience supervising productive workers but they don't have hands-on invention experience themselves it's it seems that it it's not having an effect. [06:27:07] other characteristics like being uh female like the same gender as the worker uh the age etc. It seems that uh our measure of invention experience seems to be the one that driving this this result. So now let's talk about the [06:27:20] mechanisms and because I'm uh I'm short on time let me just go directly to the evidence. On the left we have inventions co-invented with the manager and on the right we we have independent inventions [06:27:35] in the sense that the manager is not part of the co-authors co-inventors and as we could see it's not driven by just having a top superstar researcher manager that basically would add me as a worker into their papers or I would just [06:27:47] trail around even papers without or inventions without the manager also go up. Some interesting pattern is that inventions with the manager they seem to pick up right after the switch whereas the other ones independent they will take a much longer time to pick up and [06:28:02] we want to interpret this as maybe learning takes a little bit more of time. Um one evidence that I like very much and let me explain it to you because it's very important. We use IPC codes in order to create measures of [06:28:16] technological proximity between patents. And this is a standard from Bloom and co-authors. The this let me just do the analogy to our profession. So imagine I'm a trade economist. If I have a new [06:28:30] advisor who is also a trade economist, my research productivity in trade has a significantly positive effect whereas my research productivity micro theory or other fields has no effect. At the same [06:28:43] time, if I'm a trade economist and my new supervisor is a micro theorist, my research productivity in trade zero effect whereas my research productivity in other fields some positive effect. So [06:28:57] what we want to see here like the takeaway is this complementarity in which the effects are driven by the managers that have the same expertise as the work. [06:29:10] Okay. So we have some other types of complimentarity for instance a complimentarity by the experience of the of the engineer it himself. So, and it seems that the effect seems larger when highly experienced engineers in terms of [06:29:25] inventions receive highly experienced managers as well. [06:29:33] >> Four digits. So, >> very >> that's as fine as we could go. We're trying to explo expand this analysis by using text similarity measures and things that are more >> Yes. Yes. [06:29:50] >> Yeah. which for me that's kind of like the I mean this claim is subjective even within the team but for me this is the best evidence of the mechanism that is something related to knowledge uh transmission that is not only effort that you will increase your patented [06:30:04] output in every field regardless of the match between the supervisor and the supervis but it's actually driven by having an advisor that's experiencing the things that you were experienced and and these measures of experience are all [06:30:17] before the switch right it's not like it's not affecting affected by the switch >> is also concern that maybe the firm's doing this strategically to raise the productivities team productivity in a [06:30:30] particular area. They want to bring this guy in now and do this kind of stuff. [06:30:35] >> Not not necessarily because what we have is at at the worker level. Well, we maybe we have to add an extra variable there. But what we have is that the technology specialization of the team doesn't predict which manager you will get. But it when it happens that that [06:30:50] manager is on the on the same technology area then the effects are larger. [06:30:55] Okay, I have like three minutes. Um so we have another mechanism but this is more suggestive. We have 360 degree uh ratings that workers provide to their [06:31:07] managers and we don't have the identities. I mean the the firm promised workers that they won't re reveal that identity. So we average at the team level and it seems that managers that seem to be having high innovation [06:31:21] experience in the past they also create an an environment that for fosters uh innovation. For instance there are some three questions that are very important in this survey. uh whether the manager inspires curiosity, supports [06:31:34] experimentation and resources to develop new ideas and when we have the high experienced manager seem to have a significantly different effect and positive with respect to the other managers. However, if we look at other [06:31:46] characteristics, on the left we have what the workers report about their manager. On the right, we have what the manager reports about himself or herself. And as we could see, this increasing in fostering innovation [06:32:01] environment doesn't come at the detriment of other managerial characteristics. So, and at least in this case, we don't see a strong trade-off. Uh we see that this they tend to be a little bit humble because their own scores are usually lower than than the previous one. and other dimensions [06:32:15] like communication etc. they don't seem to be doing worse. Okay, in the two minutes I have uh yes, some alternative explanations. Worker selection, you can say that the managers bring their own [06:32:29] co- inventors and that's why we see the jump. I would say no because we're focusing on on incumbent workers. So we're following the workers that were in the team before the manager went up. uh and we basically have other uh sources of of of [06:32:43] uh additional evidence but that's the strongest as I said uh before team size not much other outcomes like retention rates uh salary per in the team they don't seem to move uh basically it's not selection it's not we see maybe it's not [06:32:57] resources that's what I want to say um it's not only effort because maybe you can say there's the new manager just put everyone to work if that was the the only thing we would expect regardless of the field there would be an increase in [06:33:11] patenting and an increase in innovation output. But we see this really interesting uh complimentarity between the fields that is not only explained by someone who would just come to cheer some everyone up or or or to ex make [06:33:25] everyone exert more effort. Uh resources and networks as I said at the beginning uh we could see whether these teams become bigger. We can see whether their networks of the co-inventors uh the co becomes like stronger. It's not that [06:33:39] that's not the case. It's not that these managers are connecting their supervises with other productive workers. Uh it it doesn't seem to be the case. Okay. So because I'm out of time, so we're studying this paper, how do managers [06:33:53] affecting innovation in large firms? We have a very special setting of using one of Europe's largest manufacturing multinationals. We have two main findings. The first one is the hands-on invention experience seems to be very important and uniquely predictive of the [06:34:07] innovation output of supervises and more than all the variables like tenure r even rank of the PhD degree because we can see their their CVs etc. [06:34:17] And the key mechanism that we believe is at place are basically two. One that has to do with invention, capability improvement and knowledge transmission in particular fields and second creating a an environment in the team that uh [06:34:32] fosters innovation. Uh thank you very much.