 So good morning everyone. I welcome you to our course on collective dynamics of firms My name is Frank Schweizer I'm a professor of Systems Design at ETH Zurich and I will teach this course myself I hope that a few more of those students who have subscribed will appear here, so There are actually I think 26 or so that have subscribed to the course If you happen to know why some people are no longer interested or drop out from the course So we would be really interested to know this Some of you may have Realized that I teach two courses during the semester. The other course is on economic networks Is there someone participating in both courses? So for all the others there is some overlap of course in subscribing to these courses Economic dynamics I would say is the most advanced course we teach at the same time It's although discussing formal concepts a lot. This course is Not as ambitious as economic dynamics, it's more hands-on you do something You do a bit of math something, you know you deal with equations. You deal with data. It gives you a more comprehensive View of a real dynamics and What I like about this course is that Things you learn here in terms of how to treat data how to build models That's something you can use in many other areas of scientific discipline So this is not very specific on Companies if you ask me which of the courses I like most and I can say this one of My courses I also teach another one in the fall. It's called systems dynamics and complexity This is a very basic course. It basically covers Dynamics from a systems dynamics point of view that means we deal with non-linear equations and then we look into Control parameters that allow us to reach different scenarios and so on but we never explain how we get the systems dynamics, right? if you think of whole issue from a conceptual point of view, then there must be some entities economic actors that interact and generate the systems dynamics It's nice that we can afterwards describe systems dynamics by a set of non-linear equation, but how do we get that? That's a challenging question and this question is answered in this course and To some extent all the answer in the economic network course So let me start with a few formalities here The course is an elective course in the m-tech master program. It's also listed in other programs like Elective course for physicists. Is there any physicists here the two of you good you Will probably recognize that what we are doing is very much inspired by statistical physics here. That's on purpose. Yeah Okay, it's also listed in mathematics. No one and agriculture I think We have Provided you with a website in Moodle where you find all the material. What do we mean by all the material? We mean these handouts and I managed to inform you in due time that you should hopefully download this the handouts contain all the slides that I show in the lecture and a note field sometimes I write References to literature that we have used in order to create the site Sometimes I give general comments on how to understand the slides or things like this But the notes part is basically for you to take notes. It means I expect that you will have a few important Notes to make during the lecture here. That's the idea Therefore some of these fields are empty so you can fill them up The second important material you find there are the self-study tasks. I will comment on this a bit later in detail Each of the courses we teach has a theoretical upfront teaching part and a self-study part We want you to learn something. We want you to do something yourself and In our particular case the self-study talks is about dealing with data Learning how to get certain features that we describe theoretically in this Course how to get this yourself and to compare the figures that I show here with the figures you produce yourself No matter whether you care about firm dynamics after this course or not This is some skill that you can use in most other fields Let me remind you on this and we also put Literature in the Moodle platform. This is meant as an additional information to you Sometimes you would like to look into the original paper. What did the author say about? These figures in comparison to what I said about it, then this is available to you We encourage you to look into these things Sometimes also interesting for historical perspective, and you see how Herbert Simon has argued about these things like 50 years ago That's also interesting, but it is not part of any further Check or control Then we also give you Parts of code that you can use in particular code for the statistical software are that you can use in order to run your own programs and to educate yourself and last but not least the Moodle platform is also for communication communication among you There is a forum where you can discuss about the self-study talk, but there is also a forum where we communicate to you So who is we? The professor of course, but also the assistant of this course that's Parvillian Marvroviev I'm not sure if some of you have met him before in the lecture Systems dynamics and complexity who attended this Okay, so then you know Parvillian already Parvillian has developed Most of the self-studies for this course. So he is really the skilled person to discuss Why certain things may not have worked out for you and so on There was also a student assistant working for this course Natalia Frey she Responsible for example fixing the errors that we detect every year We also redo the slides every year that means we produce new errors and if you see one and please point us to this Oh, what did I do? I? Press something right The ass was not Intended so I have to be careful here so the exercises are Mandatory in the sense that you first have to submit something to Parvillian that is why we use the Email address that is on front of the handout. So that's an email address. You should write to Regarding everything About the course You can also write me personally Then it's read by my office administrator If you want to write about the course and you write on this others and then we look into it so the groups You should form groups in order to solve these exercises To discuss certain things as well and these groups will be Randomly assigned Parvillian will usually do this He will also take care of if some new people enter the course or some people drop out and we arrange these groups it eases the Effort of doing the exercise and presenting something here. Okay, you can read this in detail Then we have a Test start every month. That is an online test start what you do at home basically an open book test if you like Where we just control whether you were able to follow this course you have Hours enough to do this test you can use all your material in order to answer these questions This will be multiple choice questions It is simply to give you an initial an additional feedback about what you understood from the course and what not and for us, it's a good control to see the same you have to pass two of these Test starts in order to be eligible to take the exam at the very end If you do not participate in the exercises that's the other condition and if you do not Use the test start. We are pretty sure that you are not able to pass the exam These are mandatory conditions for good purpose and then we have a sessions proofing and Exams during the summer break, which is arranged by the recto rate. It's a written exam Close book no additional help material But it covers not only the lectures. It also covers the self study talks. That's very important to understand Are there any questions from your side regarding these formal things? If not, then we just continue with the For the description if there are questions you can either ask them Right away, or you contact me doing the break or before or after the lecture or you write to this email address We have given to you. Yeah, just one here All right So Why do we give so much importance to the self study talks? I already explained this on Tuesdays of the Bologna process requires that 30% of your time is devoted to self study talks It's not the professor who teaches you all the things and you just listen and sleep or whatever you want Read your emails or so You have to do something yourself and therefore we have created the self study talks But in addition to the self study talks, there are other elements of self study One element are the questions that are listed at the very end of each lecture These questions are not the exam and questions the exam and questions Let me repeat it again are much more tougher than the questions at the end of each lecture The questions at the end are meant to help you to repeat The material that I have just presented There is not a one-to-one match between these questions and the exam the exam you really have to do a bit more So then as I said You have the self study talks where you form small groups and Interact and then you go to the exercises and discuss with your fellow colleagues We understand that there are mistakes for example in the exercises but in the in the self studies that you deliver but The exercises are for finding out what went wrong, right? So this is nothing where we assign grades or something it is a platform to discuss issues that Came up when you try to solve this at the end you will have a considerable skill in Handling data for example Okay, I don't read this for you. The first thing is of course that you form the self study groups in order To be able to start with the self study talks. I will explain the first self study talks Later in the lecture. It's a very easy task You could probably do it yourself, but we want you to interact a bit with other people So and then if you have questions you can use the Moodle forum or if you have specific questions and have to Send in your solutions from the self study talks you always use this R2S All right Okay with this I come to the content of the course The first thing I should explain is what this course is not about Collective dynamics of firms sounds a bit like the dynamics of firms Most people drop the issue collective because they don't know what this means Let me start by explaining what the course is not about It is not about individual firms If you participate in the m-tech master program Then you know that there are ample of courses that are just devoted to individual firms to Strategic issues of firm development to management issues to economic issues and so on We are not talking about this in this course. This course is a Very different from those other courses It's not our interest to talk about a particular firm ABB and what ABB should do in the next five years In order to improve its standing in the market or to get a growth rate That's expected by the shareholders or something like that not our topic at all We also do not predict the future of a given company X You learn in other courses how difficult it is. We will predict the future, but not with respect to a particular company So then what do we mean by collective dynamics of firms? We look into what happens if we think about not just one firm, but thousands or millions of firms How many firms do we have on earth? Can you give a rough number? I mean firms we know about There are many firms we don't know about firms who don't pay taxes to the government or file any sort of reports, right? Or there are maybe governments who do not collect all the data from the firms and we cannot know about it But what what's the rough estimate of the number of firms if we know about? Just the order of magnitude what do you think? 1000 Million yes, okay, how many million nine million Anything else no clue we know about about 50 million firms That is the number of firms where we know the name. We know the location. We know There probably the tax they paid to the government. We have a track history We know who's the CEO or the head of this company and so on I'll come to this in a moment and Now let's assume that all these firms have some business with each other. They say are not isolated firms in the market, right? they compete for Selling their products to the same market for example, but they also collaborate in order to Develop new products and have a comparative advantage against other competitors and so on We can It's difficult to observe the strategy of a single firm Right and how they position themselves in the market. Let's think about UBS or something like this You cannot really know about the strategic decisions of that kind of bank, right? Because that's part of their business model and they were not printed in the newspaper But what we see is if all of these firms interact We can see the collective dynamics the dynamics on the systemic level that emerges And we would like to first explore this collective dynamics. What do we see and Then we want to go one step further and want to ask, okay If we know now how the dynamics of all these firms look like are we able to make Assumptions about the dynamics of an individual firm that's a challenge point here and We cannot as I said specifically deduce this Dynamics of an individual firm or the growth strategy, but what we can do is we can provide reasonable assumptions that are compatible with the macroscopic dynamics that we observe and This gives us enough reason to hint on Regulatory issues policies that should be set up in order to encourage firms to Grow or to be established and so on The result of our investigations are as I said not specific strategies the results are Statistical laws of firm growth. I Come to this in a moment. I give you an example these Slides as I recognized yesterday again very much rely on text, right? So I try to give a life and lively performance here in order to give some illustration to these plain facts here Publishing thought it's fine to have just this text because it's introductory. We will have more pictures Okay So what we would like to find here are statistical laws I come to examples later, you know We have divided the material of this course in three parts the first part is about modeling What I think is not the first part in the course That's the first part on the slide the first part in the course is data analysis Maybe we start with number two here because this goes and I know We assume that we have data about these firms For our exercise if we use about 10,000 firms and we learn how to analyze this data statistically For this we would like to Find out about the distribution for example if I give you the size of the firm in 2008 and the size of the firm in 2009 Then you can first plot the distribution of the sizes in 2008 You make a histogram you've been the data, right and then you plot the distribution of the size What do you get by this? Let me use this blackboard I Said an exception so let's assume that's a firm size and then P of X is The empirical distribution actually the histogram that you get from this data You bend this here and then you get a curve like this so What do is the underlying message if you see a distribution like this Let's assume. It's a normal distribution or something. Please. Yeah, let's assume. It's a Gaussian distribution, right? But this is a mu and that's a sigma here. Well, okay What would be the message if you if we were able to find a distribution like that? Yes Right so if we find a distribution like this and we say well interesting The firm seemed to have a typical size Namely Round the mean value here, right and then the economists say well if we think of an evolutionary Process that establishes there must be some kind of optimality behind us What's the optimal size of a firm, right? So that means then we start Discussing what internal processes or External processes in terms of firm firm in direction or whatever Text to be paid with a government result in this optimal size But what I can frankly tell you is We don't find this Because there is no optimal size for firm You already know some of these successful firm are very large and there are some successful firms that are extremely small, right? so Okay, we find Something like This Okay, that's what we find What's the interpretation of that? Okay, let me stop here at this point because we discuss this next week, right, but your first exercise After this would be Getting this distributions and then start thinking of what's the meaning of this If I find a distribution like this, is this can I say anything about this? Is there any regularity behind this if yes, what kind of regularity is it? For example, can I scale the function Against the size and then get some universal distribution as we know from other theories for example in statistical physics Okay, that's the first part and Then we have all these Distributions not just about the size. There are other distributions as well. For example, if you have the Snapshot of two consecutive years, what can you then all the calculate? Obviously Of two consecutive years Yeah, the change of the growth of the firm how the firm the annual growth rate of a firm, right? I mean for this particular firm. It may be a bad year, right? What if I talk about 10 million firms, so then the statistics is probably rather good, right? That means we do not get just this distribution about the size. We also get some about the growth rate and then We look into the distribution What do you guess? It's a symmetric or is it an asymmetric distribution? Any hint Just speculate If it is an asymmetric distribution both will be the economic consequence of these we're talking about growth rates of firms here And you guess it would mean that the growth is more likely than to shrink it, right? if it is Has a bias to what's positive growth rate, right? That would please the economist right because the economist is obsessed with growth right economic growth That's what we want to have So, okay Let's check whether this is true on the level of the firm Right, that's something we can find out and if the biases on the other side the negative side But the message then right Do we have a shrinking economy? Do we have a collapse of the whole economic system? What are the consequences of this? And what are the consequences? If the distribution is symmetric We will discuss this next week, but these are the ideas that are related to this So then after a few lectures we stop and then we see we have a number of what we call stylized facts These are all the statistical laws where we find some regularities You do not expect this regularity if you think on a particular firm But if you think of millions of firms and you see this regularity That's what I sell starting next week and Then we have to ask ourselves, okay But these distributions are made up by individual firms who grow or drink each year What is the dynamic on the individual level of the firm? That is Compatible with this kind of distributions that we see That's our question And once we have identified this dynamic and we can ask about the economic meaning of this We go basically from the top to the bottom And then we end up with Stochastic growth models of firms Do you know what a stochastic model is? Have you got any experience with this and other courses? Yes, yeah, of course Yeah, okay. It is a probabilistic model where Random forces play a considerable role. I give you a very simple idea of a stochastic model so x of T plus 1 That's the time as x of t plus some random shock That is added to this actual value, right? So this can be all the negative then we talk about a shrink Shrinkage of the firm it can be positive than the firm growth, right? So that's a stochastic model because this is a random shock This most simple thing we can think about The physicists know that this relates to Brownian motion or to random walkers and so on so this is a good proxy for economic growth That's something we will discuss We will find out that not a good proxy as you can guess but surprisingly We find that this is a good proxy T plus 1 So we have a multiplicative process Multiplicative means there is a growth rate here. That's multiplied with the actual size and This is a stochastic force Wow You can describe the growth of an economic entity like a firm by using an Assumption that heavily depends on random forces, right? That's a bit of a surprise because you think of firm growth in a more deterministic way So what's the difference between this model and your initial assumption that there is a bit more than just random forces behind it? You are right, of course But this points back to the issue that I addressed before We are not interested in talking about a specific firm here for a specific firm. We would probably have Better assumptions here But if I talk about 10 million firms Then this is a good proxy for the individual growth of a firm because when I aggregate All of these processes then I get exactly the distribution. I have seen empirically You understand I cannot make a prediction about a single firm But I can give you a probability that the firm of that size had the growth rate like that for example the same as with Buying a lot of tickets right you do not know exactly what's the number that they draw next Sunday But you know the probability distribution that this number will be drawn Okay You understand this idea so then we talk a lot about these kind of equations and how they are linked to Economic growth and particular to growth of firms But we also talk about formal concepts here and I start with this today Why are we interested in this what are the kind of conclusions and an economic sense that we can use from this What are the conditions to boost or to enhance? Economic growth that is based on the growth of companies So this slide says more or less the same as I said before it shows you the structure of the course so the first five Lectures are first five lectures are Dealing with this data analysis and the empirical facts so at the end we know what we call stylized facts here What's the stylized fact? It's a statistical regularity. You would not call this a law You would call it maybe a statistical law or the economist says a stylized fact That is what we find About the firm size distribution about the firm growth distribution about the relation to age If you are a very old firm How is the age of the firm affecting the growth rate for example? That's a very interesting question This these are things we find there empirically then we go and model these processes By means of these stochastic equations I have described we go through this step by step you should not fear that you get lost because of You have never dealt with these kind of dynamics. I mean the course is essentially to teach you about and Then gradually we add more economic assumptions to this picture of interacting firms First we look into what do we get if firms are kind of Independent and then we look into what is the effect if the firm start cooperating or if the firm start competing What is the outcome on the dynamic? So that means what we do is different projections of the whole picture if we have further assumptions closer to economy This is the last Overview slide I think so I have divided it into how and why The how relates to the skills that you Obtain during the course so first of all you use you learn to use this state of the art Statistical suffer I come to this in a moment. You also learn how to build up multi-agent models of Interacting entities in our case these are firms But if you are from biology, then you immediately realize that you can use the same assumptions and the same techniques Also in other fields namely in biology for example so and you learn how to Squeeze out from this data on these models some regularities because remember these are Phenomena that rely on lots of fluctuations. How do you squeeze out the regularity behind this? That's what we learn here, but then it's not only a technical Course here. It's also a course that tries to explain Why we see this kind of dynamics? The first and most interesting question is why do we see a kind of universality here? because intuitively we are all convinced that Each firm has a different CEO or managing board and these do their very best to make the strategy of the firm unique Different from the competitor, right? But still on the aggregated level we see a kind of regularities. How does the two things match on one end? the struggle to be different from other firms and on the other end The more or less universality of the dynamics that we observe That's a surprise We will argue why this is and on the other hand, I mean if we see this kind of universality Then we clearly understand that this allows us to to have some prediction also about the future Not on the level of an individual firm as I said, but on the level of the economy for example We also discuss the issue how firms choose different Interactions like cooperation and competition initially you would assume It's good To win against a competitor, right? so You are the Apple CEO and you got the iPhone on the market and because of this you do not like another company called Samsung who comes with another phone and then HTC which comes with another phone and they all look quite similar and have the same functionality, right? Your idea is to kill all your competitors What do you gain by this? You can think a moment, right? Then you rule the market, of course. Yeah, everyone has an iPhone But you also wiped out all these companies that are sources of innovations That's an important thing your competitors not just the competitors also a source of innovation This drives the growth of your own company This is something that you observe that some competitors keep other competitors alive, right? So we will talk about this example of Jill yet versus Wilkinson. So that's a shaving company, right? So Jill yet has about three quarters of the market and Wilkinson and all the other companies have one quarter There were several possibilities in the past to just take over Wilkinson and then to rule the market They didn't do it, right? They kept their competitor alive because they learned that from this competition the market grows But if there was only one company that ruled everything the market will not grow That's an intuitive understanding of how firms interact But how do we put this into the model and how can we see that firms do better by keeping their Competitor alive That's something we will answer here, right? and Then we also look into how the market is driven by innovation and how innovation succeeds in a probabilistic model and so There any question about the outline of the course No, if not, then we start by talking a bit about the motivation behind it if you want to place this course in Somewhere on the map of economic theories, then we belong to industrial dynamics or industrial organizations That's a topic of the course Sorry, I have to be a bit careful here. So what is industrial? Dynamics This is a part of economic Theory think economic thinking where we look into the conditions and The dynamic processes that govern the birth the growth and the death of firms and industries Sounds a bit like biology, right? birth death and growth and in fact as you will see The lifetime pattern of a company is not very different from the lifetime pattern of a human, right? There is birth and death There is growth. There is saturation or maturity. You will all this see this in the dynamics of these firms as well and What we are interested in To find out how can we describe the conditions under which firms are born? Because if you consult for the government to the government So then we would like to give hints under what conditions more firms are born, right? For this we have to understand Empirically and all the theoretically what's the impact of lowering taxes on the birth rate of firms And what's the impact of rising taxes on the death rate of firms? And it's not as simple as you may think about So these are things we discuss in industrial dynamics So in particular we look into these stylized facts about the firms How can we describe this there are other issues like productivity and so on and our aim is here to get an Inside into this regularities if we have this inside when we can make predictions About the future, but we can also control the dynamics to a certain point, right? We know what to do if we see a steep decline in firm number What we call death here, right? And we know what to do from these models and from the data So what are our control parameters to to keep this within limits for example? Okay This is second the next two slides give us Some examples here The first Example why we should know this is related to market power You can assume if you know about the firm sizes Then that's a good proxy about the market power of a firm You can use other proxies as well sales rates for example No, do you think of these browsers here? You can think in terms of download numbers, of course, that's another proxy for market power But it's very clear if you compare different Firms and you do not have a clear whatever relation to the market power You use the size of the firm revenue as a proxy of its market power so If firms have a large Control over the market. This is not always for the best of the customer, right? Because then they can also control the price they can drive the direction of the technological development into the direction that They prefer Take the example of Google or Facebook, right? So That means we should understand how market power is related to the welfare of the consumers And for this we should know how to proxy market power Here there is this example of the different browsers that we come that Competed a couple of years ago. So most of them you will not remember anymore Are we but with the if we know why some of these companies? Have gained in size and in market power and others not then we can for example tells the government What to do in terms of antitrust? Policies is it really necessary to keep this company alive? Is this the best way of? Preventing other companies from ruling the market or are there other ways of Preventing that a single company dominates the mark You see first we need to understand the empirics behind it so This is the second example and with this I stop because I assume that the bell is ringing in the moment The impact on employment policy So firm size is a very abstract notion, right? How do we proxy firm size we proxy it by? number of employees All right, so we measure it by a number of employees because that's a number that we know quite well It's all the report And if we talk about the birth and death and we talk about the creation or the destruction of jobs and If we find that there is a large probability for a large firm to collapse Or to die, right? Then of course this has impact on the job market It leads to recession to all sorts of things So that means we should understand what are the determinants of size in Order to give hints for example to employment policies We can prevent for example unemployment either by keeping large firms alive Influencing their probability to die or there's a way round by Increasing the probabilities that new firms enter the market, right? That's another way and of course a big guy dies, but thousands of new firms Enter the markets that are interested to hire experienced employees, right? That means that these people who are laid off by the big company then find immediately a job In a smaller company that is happy to build on their previous experience, right? And we have to understand all these things We have to understand what is the influence on birth rate on death rate What's the influence on size in order to make this kind of policy advice with this? I stop here when the bell rings the next time we continue So I just said we just continue I'm a bit confused to not hear the bell but This happened on Tuesday as well So As far as I understand it's a correct way is that we have a 10 minutes break from 11 to 11 10 And then we have another 45 minutes to 5 to 12 That is my understanding of the idea So and maybe from next time with or without the bell we make sure that this works Okay, so Let me come to the methodology that we are going to use here This methodology is in line with other quantitative sciences like physics or chemistry That means we have an empirical part there where we identify what we call the stylized facts namely the robust statistical patterns in the data and This is similar to what Kepler did with the astronomic data, right? So he looked into the astronomic data and was able to deduce this to a quite simple law Which is called Kepler's first second and third law But he could not give an explanation why it was in the data Nevertheless, he developed a very interesting theory of what he thought is the origin of the empirical laws He has discovered Right, it's very much related to theology. I recommend you to read it he was able to find the five platonic bodies in these Ellipses and so on the very nice work, right? So but this was his explanation. In fact, what he discovered and would still remains is the empirical law and then a person like Newton was able to derive Hypothesis on how Mars body should behave and this was the underlying principle the law of gravitation which then led To an explanation of what was seen empirically And in a similar way we do the same We identify first the stylized facts and then we make hypothesis of course in this time about the economic behavior and Try to reproduce what we observe in terms of a collective dynamics. That's the idea So with this I come to our empirical source We already estimated the number of firms to 50 million That's the firms we know about I said so how do we know about this? usually from databases There is one database that we also use for the course here. It's called the Amadeus database it contains of about nine million firms from 38 EU countries and You see each of these firms have a specific profile there Which tells us about the shareholders and managers the firms that are owned By this particular firms and so on and then there are also relationships between firms and shareholders and so on We have developed a nice Data browser Which I just run here that shows you what happens if you browse the database What we started from a company you you see all the data that is stored in this database On this side and we started from a company that you probably know co-op and then we looked into how it's co-op related to other companies Right and you and you see the subsidiaries of co-op Listed here in the neighborhood. You know though all these companies You probably never realized as a subsidiary or daughter of co-op, but there are some that you already recognize So these are the first order neighbors Yeah, there's a direct link from co-op and in most cases there are a hundred percent of ownership here, right? So now we can increase the resolution decrease the resolution a bit by going to the second next Neighbors, right and you see other companies appearing here now and the immediate neighborhood of Co-op so we can also zoom out a bit In order to make room for the next iteration bomb And then you see the third nearest neighbors appearing So you see there's a network of firms evolving in this case. We look into the ownerships only Do we see the third nearest neighbors? Already yes, it starts so and then you see well We go a bit. So then we see Mikro appearing What you would assume Mikro is a big competitor of co-op, but there are Economic relationship between co-op and Mikro that you can see here When you go along these lines, there are several companies that are owned both by Mikro and co-op Oh So and maybe let me just look into this Yeah, so then you see Can all the move this a bit? This is just my slow computer, but it's also not When you see here the Mikro appearing with Their daughters and then you see here a number of of companies that they link To co-op for example, so this is about an ownership relation that we can detect from the database But there are other features that we can also look up I'm not sure how well this works, but let me try one thing here Let me go back here to number one and then instead of the investment network Let's look into The member of board network, so it doesn't work. Okay, so then let me It's not better. So then what about manager network? Okay, maybe it takes some time So now we talk about people, right? The CEO of co-op how is the CEO of co-op related to the CEO of Mikro? That's always an interesting question, right? Maybe these are complete independent people with their own decisions. So it could be right So this is now I mean So these are now It's in the manager network Where Positions are listed and then we can do the same thing again We can go along these lines and then you see people Emerging and companies emerging that have a say in this Then we can click on certain People here Let me just zoom out and then we see their network evolving. Okay The relationship is what these managers have to do with the company. So this is listed in the company profile It's listed whether you are the CTO of this company or whether you are in the advisory board of this company and so on And in most cases you are not only in the advisory board of this company You're also in the advisory board of that company And that's the most interesting thing if you think of market regulation if you think of competition, right? It's the same people sit in the board of different companies. So what's the value of competition? What's the value of regulation? These are issues. Okay You see so I can now click on any of these people here the people are listed in green and the Grays are companies that Connect these people right Basically, it's a bipartite network and then we would basically see how different managers Indirectly influencing each other, right? It's an interesting question, but I'm not going to talk more about this. I'm just telling you There is no secret here. Yeah, this is all available in the data We just need to look at it. How do we do it? So another I come to examples later on another database that we also use is the Orbis database that has much more firms in this particular case 90 million firms We all they have other databases available like Bloomberg or Thompson Which basically report on financial data of this company though to put it this way there's a wealth of data available What is mostly difficult is to link data that Belong to different databases to the same company Because mostly even this financial data is not directly linked than to specific other features The data as I said is publicly available But I didn't say that it's for free, right? Most people think it's publicly available. It's yeah for free, right? I didn't say that In fact, it's a very very expensive database We pay a decent amount to have access to this but everyone who wants can have access to this You just need to pay So the data is also not a secret that's data that companies report officially the aim of these Companies is to collect this data and fill this into this databases. That's how they make their business Okay This is Now once we have bought the database a simple query in a my SQL database Where we can get access to all these data that is reported about a specific firm We are not interested in all of these things. Yeah But for example, we are interested in shareholders We are implanted interested in particular the number of employees and the previous year because we use this as a proxy of the Firm source and so on so You are not concerned with this one. We don't tell you how to query this database We assume that we did this for you, right? But you get the data from this database Then already in a cleaned way No, that's the first thing we did it for you We use one example here for the course that is are the TNC's TNC's is an abbreviation of transnational companies Why are we interested in these companies because we think that they play a very important role because they operate in different countries We don't talk about the reason why they operate in different countries But you can have ample of hypothesis. Yeah, why it is attractive to have a Headquarter in Switzerland or Luxembourg or the Netherlands and so on. I'm not arguing about it. That's a matter of fact, right? Well, we look into these companies and from the Orbis database we could identify about thirty four thousand transnational companies I Talk about examples later on the interesting factors. Where are these companies located? They are located in a hundred and sixteen different countries. That's a very interesting thing so and Even more interesting these companies are owned by a Small set of shareholders you see on average are just two shareholders per company by Transnational company. I'm talking about big companies That's not a large number and these shareholders interestingly sit in a 192 countries If you talk about all the industrial countries, of course, you end up maybe the range of 50 60 Who are these shareholders if they are distributed in a hundred ninety two different countries That's an interesting question, right? so The other interesting story is that these companies either of them owns on average ten other companies So you see already a network appearing here, you know so and these Participated countries are only in about 40 countries So the transnational company as you can even see from this extremely simple number is Nothing, but an organization that concentrates Economic power on a very few countries. These are the industrialized countries while probably paying Texas in A large number of other countries where they may have not a real business in so and this Collects the money from people in a hundred ninety two countries, right? You see can you see the flow of money? Can you see this? Yeah? Okay, that's a very impressive thing. So and then you also see that The top 500 of these Translational companies are responsible for the 50% of the board trade 500 companies do 50% of the board trade and 90% of the foreign direct investment of the whole world, right? That means if you look into a subset of One percent of these They 1% of the transnational companies is responsible for 90% of the foreign direct investment So I hope that these numbers if you play a bit of imagination Already ring a bell of how the economy is struck. It's extremely skewed That means though the power The economic power is in the hand of the very few companies Yeah, and also the financial power because the foreign direct investment is ruled by this company Nevertheless these companies are not very large So we looked into two extreme cases here general motors is the largest one was 280,000 employees But there are also small ones with 22 employees in this data set This is empirics and now then we have to think of what is the story behind this What do we learn about economics when we see these numbers? We will do more on this in the course But we look now a bit into the structure of these companies so we took one example here and Looked into the classification of their business because we would like to know what is the business that these transnational companies do Right so and there is a classification for this. It's called the nace classification in the notes I have explained a bit more about the nace classification That is used in the European community It's a four digit number to classify your kind of business. There is another set of Classification the ISIC classification international standard industrial classification that's mostly used in the US I Gave a few hints here that means each company in this database is classified according its main business So we took one example here Fiat So you probably know about Fiat, right? What do you think is a main business of fear? Hmm cars other ideas Okay, so Fiat is classified it classifies itself By this number so we looked it up here code 7415 okay, so 74 first of all it belongs to section key K. I wrote it here in the notes, huh? And K section cable deals real estate Renting and business activities so and 74 specifically means other business activities and 74 one this are the three first digit relates to business and management consultancy holdings So different from you what you think about Fiat Fiat is a management company, right? it runs on consulting business on Renting and other business activity. Okay. That's a very interesting thing, right? Cars are do not exist in this kind of classification, right? And why is this because Fiat owns other major Plants and branches that produce for example cars, but all the other things right so we need to understand this structure Okay, this is one example, yeah, so when you go through this here, then you see if you look into what's the business of all these transnational companies, then you see the only a Small minority is really doing business in the primary sector So that means the primary sector is where you transform natural resources into some sort of primary products no, so and only 10,000 are in Manufacturing that means they use the primary product to build a car or something like this, you know, but about the same size is in the service sector and about half of this is Manufacturing sizes in the financial sector Which is not related to any sort of industrial production so and the most of them are as feared in Real estate renting and business activities Okay, that's a very interesting thing if you think about the economy So we learned this from the data here is another plot where we try to sketches all the with the respect to the countries Here are the transnational companies the transnational companies have subsidiaries Oh, it's all the quality of participated companies that our company they own either jointly or Exclusively right so that's on this level and at the same time they are owned by these shareholders, right and of course the shareholders can also own these Participated companies themselves to some extent, but remember the numbers a hundred ninety two countries a hundred eleven a hundred sixteen countries Forty countries what you see what's the meaning of a transnational company and also the purpose of a transnational company Very much concentrates the economic power in a few and the financial power in a few countries So which are these countries so you see that the lion share is Here in the Netherlands for example, Germany great Britain the US Why do you think the Netherlands is so important here? Last year there was someone from the Netherlands who could explain this precisely, okay? What's so special about the Netherlands any Hint any clue why are these all the companies all in the Netherlands any argument? What do you guess? How many Regulation yes, and in particular taxes, right? So These companies move to those countries where they find the best Economic conditions for their development, right? and remember These companies themselves do not need to do some business, right? Each of these companies owns on average ten other companies and these companies do the business Understand what I mean So, okay, then we can argue a bit what's the role of these different countries here So we I'm not going to do this, but there are interesting stories behind this This is simply an empirical analysis, right? I don't start a discussion on policy on regulations on these things here, so But this is what we can simply get from the data and there is also this very nice paper I think I yes, I mentioned it in the notes the Network of global corporate control which made it then to the news in several countries Occupy Wall Street people though they have it in their hand So I thought on TV and they said okay There is the scientific proof of what these people do to us, right? So here this with scientists. They found it. Yeah, how we are used to Be exploited, okay Anyway, let me Come to more concrete things afterwards first thing is I hope with this simple Example that I have to convince you that industrial dynamics and the structure that we have just talked about is a very interesting topic per se It's all though. It's an empirical Problem that means you can learn from these things from data once you have Generated hypothesis of what you would like to ask Specifically what we would like to find out is about birth and growth processes, but all about decline and growth and death processes of different Parts of the industry and of firms in general. What are the conditions here? I tried to convince you that there is a direct impact on Specific policy issues think of unemployment think of market regulation and so We are able to look into data that we are not just sitting here and speculating and writing down mathematical equations and do proofs on these all these Things that are usually done in economics. We are also able to do Real empirical work here Our aim in the first round is to understand are there any structures behind that that are robust Yeah, things we observe in different data sets Stylized facts or statistical regularities and if we identify these that's of course the first thing we do Then we have to ask How can we reproduce the stylized fact? What's the underlying dynamics in the interaction of firms that gets us to these stylized fact? That's a question So and we have to do two steps now The first step is we need to extract the data. We are interested in from these databases. I Can tell you that my graduate students, Stefania Vitale and James blood father spent a year each To get this data out of the database Yeah, I'm just telling you that this is nothing then it's not the same as pressing a button Yeah, and then you get something. Okay, you have to check this data for consistency You have to clean the data and then after six months you find out that By analyzing the data you find some very important companies you have never heard of in your life, right? So and then you are extremely proud because you have Detected the hidden note right that controls all of the economy just by your analysis But then it turns out that it was not the hidden note that you detected you detected a buck in the database that a Person employed by this company made on a particular day. It was a bad day for this person, right? So he mixed up a few numbers I'm simply by this mistake. So some complete unknown person Winded up the hierarchy and became number two in the world, right? Okay This is then a bad day also for the PhD student of course because then first of all we have to complain to the vendor and secondly We have to start from scratch, right? With the analysis this happened in this particular case So we are not querying the database. We assume someone did it for you But what we do now is we talk about Extracting the regularities and then we have to talk about the software that we are using here That is our who has experience with our you already in what context, please With the ten words. Okay. Good. Yeah. Okay. Who else has Yeah, you of well no one, okay Why did we use our The first sentence says it already are is these that state-of-the-art statistical software package it's vastly accepted in all kind of communities not only in the industry, but also in Science even in social science where you think okay people are not so much in Heavy statistics or modeling or so there is a huge community just developing our and there is there are own Conferences on finance and econometrics. So which entirely focus on our Just for your information I Have put in the self-study a short article from the New York Times Which I encourage you to read in order to find a few more arguments why this is the state-of-the-art I'm not going to comment on this The most important thing for us is it's an open-source project And the second one a to heart series is heavily involved in developing this Determined words does a lot on these financial statistics packages But there are all the other people here in the Institute of statistics there's also course by And mr. Mechler and professor Buhlman on Computational statistics and they do it the same way we do it here So therefore what I have chosen here is you want so it's a canonical way of doing it So We will not use more many of the tools that became available in this In this community only a few but it allows you after you did all the Exercises with are it allows you to fully use this package. Yeah, okay It's a second important program. We are going to use is GNU emacs so RS is statistics suffer But this needs to be Run and edited in an editor And we have chosen the GNU emacs editor. What did he term what shows all the GNU emacs or what did he choose? You were free to choose and what did you recommend? Our studio, yeah We also recommend our studio if you like to have a graphical user interface and so on so that's written up in the Self-study talks one probably recommends you or Tells you how to download our student. That's a graphical interface Why do I still keep? This slide about GNU emacs Our studio is specifically targeted at using our in a graphical user interface No, emacs is targeted at doing everything That's a difference if you learn GNU emacs and you have a tool you can do everything with you can write your diplomats easy as you can Have a file browser you can have all sorts of things, right? That means by going the way we recommend you learn more than just the R studio No, of course you lack the opportunity to click on certain things No, but maybe you do you get to the point where you find it useful to not click on certain things? No, you learn a number of shortcuts and then basically that's it and this gives a boost to your Productivities that you do not realize initially in the Initial phase you usually think so why do I need to learn about this? Program right because it's not these graphical user interface stuff But in the second one you recognize that you are much much faster in handling this thing Our student assistant all have to work with a new emacs because we generate these slides in tech What do you think how many tech files do we have for such a course? We have an idea Just say number Individual tech files For such a course Hey, I do Now let's say 300. Okay This is a 300 individual tech files on different topics that are then sorted and master files that gives us like Now let's assume so I want Natalia to change the sentence About rich Richard Stelman who gave a talk I think last year here at it here at Zurich So what was completely crowded as Marcelo told me? So that's the inventor of this free software foundation and So let's assume she needs to find this in the 300 different files. I mean, how is she doing it? I mean technically, right? Or we have made a misspelling something resources always was to as of these kind of things Okay, it's a consistent misspelling. How do we fix this in 300 different tech file? That's something that happens, right? Oh I write my diploma or Martha season then I have a 50 different references that I want to include How do I do this technically? These are all things that are already implemented in GNU emacs I want to deal with this and you recognize. Okay, that's not just useful for using our it's useful for everything That's why I'm talking to you about by the way as far as I understand Mr. Mechler and Mr. Breelman also do it this way. Yeah, okay I Encourage you to do the step right I do not remember that people ever regret this But I remember a couple of people who thought they know it better and never tried it. Okay, so I meet these people quite often So Okay These are two open source software projects Both are and emacs. What's why are we using open source software? There's also a principal reason for this After what's the time? There's all the principal reason for this Open source software has the advantage of being developed by a vast community Well, that means neither the intellectual property rights are owned by a company nor is this company Controlling the development of this software Instead what we have here is a much more flexible Dynamics in developing this software if there is a need on the user side. This is much more Frequently implemented in the software because there is a strong Interaction between the users on the one hand and the developers on the other hand to implement these things That means from our side. There are lots of advantages. We also contribute ourselves I mean my chair Contributes to different parts of open software so Projects not just not related to our but they're related to other things Well, some of we have developed the style files used by ATR for the official letters of these presentations and things like this So we do a lot of things in this direction. We have also developed a software to visualize Small-sized networks Called cuttlefish you find different hints on our website on this, okay Why don't Companies use so much of open source source of many companies do and you have seen before that I run eclipse for example, you know to show you the network browser That's eclipses an open source software that is hosted by IBM, right? It's an open source software, but IBM Considerably contributes to the development of the and provides the infrastructure because there are about like 30,000 people Yeah at 30,000 notes in the software that need to be developed, right? okay, but most companies refuse to use open source software because There is no one to blame if someone if something does not work, right? And if you are the CEO of a company and then you say well, we can talk about the prize But if something goes wrong, then I want to have someone to blame for this I want to pick up the phone. I'm gonna have someone on the other end Yeah, who promises me to have this solution in half an hour or an hour or day And if not then he's out of business This is the that's a mental thing Yeah, so if you are you have to see it from the side of risk management, of course you pay Yeah, but this gives you a more smooth way of managing the risk associated with using certain software and This cannot be guaranteed with open source software. Nevertheless, there are many companies that use open source software Or provide services for open source software even that they do not develop it They provide the service for this and say, okay, we are the one to blame. Yeah, we fix the things with you Okay, yeah, so this is something that you should understand But what I would also like you to understand is there is no Disadvantage, I mean in terms of performance or productivity if you go and use open source software You will see that this provides the same quality as something you pay for Okay This is something you can read at home So there is the OSS analogy of the Cathedral and the Bazaar So we have used this here from other other Talks about open source software Just to give you more insight into the mental framework of developing software and most Pre-pro, pro-prataire software is Following this cathedral like software development Whereas the Bazaar like software development is a metaphor all the four open source software You can read it at home. Yeah, if you want So with this I come to your first self-study talks you go to the website you download the self-study talks And the best thing you can do is you read it loudly and do everything as described by Parvelyn and Natalia No, it should be complete smooth and error free Okay You have to learn how to get started with R with e-max So and ESS means e-max big statistics. That's this particular Package for e-max that helps you to to run our code and then you have to This is now related to 10 after 11 so we still have five minutes because we started 15 after 11 Okay, and then you have to do one little thing you have to write the R Code for plotting a normal distribution have to include the picture Into the deliverable and census picture to Parvelyn. Okay, so Estimated time. I don't know how much time you need to install R and e-max. I don't know that No, so what once you have installed this the estimated time for this is About 60 seconds You've got it. So the estimated time after you install it is 60 seconds No, so Order to convince you I type this now. So that's exactly what you have to do then so I mean I have of course installed R and e-max now So I Do exactly what's written here on this side. So I first run So e-max is already running as you see here. So I start e-max R mode. I just type this here. Okay All right, so and then I start our So it asks me where is my starting directory? Would be nice if you use always the same starting directory because then you have an our history of all the Things you already process in R. Yeah, so I just accept this default folder here oh Go down here documents, right? Let me Can you see this? Is it possible? So so then I get a Error message cannot read history file palm palm palm our history That's obvious because in this document folder. There is no file about the our history, right? But if you always use the same directory, then you do not get this. It's basically not an error message It's a warning, right? Okay, and you see that already your screen has changed to some console like Format where you just type what I wrote in the hand out So you create a file where you later save your our code so file Create so let me call this as in the example test our Okay, bum true means I Got it. So then You have to plot a function and the function for the function you have to specify the x coordinate and the y coordinate, right? For the x coordinate in this example, we assume that you have to plot it in an interval between minus five and five so so that means I That means I assigned to the x a A sequence of values from minus five to five So and now I have to give the raster so by equals 0.1, okay So what about the why the why should be the normal distribution, right? Our function So we hope that someone was already nice enough to implement this, right? so so We So they we gave you the name of the normal function. It's called D norm of X Right so Okay, and now we have to tell To plot So plot is plot and then what do we have to plot x y and then I try types So elm means line if you don't Specify the time you get dots you can try with this. Yeah, and there are two other things so cx Lap equals two Effects the size of the x and the y if you want to see what this is doing You just change the value to five or one or something and then you see what the effect on the graph, right? so and cx Access gives you the size of the axis Description of the numbers To bomb so that was it, right? You type these three lines and now it's 12 so and we are done You type these three lines and then you get this plot here You can try with the different parameters. You save this plot It's also written on the slide how to save it and then you include it in your presentation That's that. Yeah, so we made it that simple in order to give you a quick success Did you see it's very easy to? Learn about this. So that's the reason I can go back here just one second So that's the success. Yeah, the picture you have just seen Okay, and these are the questions you should answer at home in order to Recapsulate what I said in the course Next week we are going to get more To the detail so where the meat is so we start by describing data sets and by treating them in a Mathematical and statistical way so next week. I'm not having all these slides with the bullet points I'm having more equations. I'm having more pictures there And I hope that many of you will join us next week Yeah, and in the meantime, you just install our and Then you use e-max that would be my recommendation if you use our studio There's nothing wrong with doing that. That's a graphical interface. It's also a standard interface But there's this other one was my recommendation. Okay. Thank you very much and see you next week